Molecular Kinesis in Cellular Function and Plasticity (NAS Colloquium)

About this PDF file: This new digital representation of the original work has been recomposed from XML files created fr...

Author: Proceedings of the National Academy of Sciences

56 downloads 1187 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution.

i

Colloquium on Molecular Kinesis in Cellular Function and Plasticity

National Academy of Sciences Washington, D.C.

2000


NATIONAL ACADEMY OF SCIENCES

ii

National Academy of Sciences

In 1991, the National Academy of Sciences inaugurated a series of scientific colloquia, five or six of which are scheduled each year under the guidance of the NAS Council’s Committee on Scientific Programs. Each colloquium addresses a scientific topic of broad and topical interest, cutting across two or more of the traditional disciplines. Typically two days long, colloquia are international in scope and bring together leading scientists in the field. Papers from colloquia are published in the Proceedings of the National Academy of Sciences (PNAS).


CONTENTS

iii

PNAS Proceedings of the National Academy of Sciences of the United States of America

Contents

Papers from the National Academy of Sciences Colloquium on Molecular Kinesis in Cellular Function and Plasticity INTRODUCTION Molecular kinesis in cellular function and plasticity Henri Tiedge, Floyd E.Bloom, and Dietmar Richter COLLOQUIUM PAPERS Kinesin molecular motors: Transport pathways, receptors, and human disease Lawrence S.B.Goldstein All kinesin superfamily protein, KIF, genes in mouse and human Harukata Miki, Mitsutoshi Setou, Kiyofumi Kaneshiro, and Nobutaka Hirokawa Assembly and transport of a premessenger RNP particle Bertil Daneholt Ribonucleoprotein infrastructure regulating the flow of genetic information between the genome and the proteome Jack D.Keene Spatial and temporal control of RNA stability Arash Bashirullah, Ramona L.Cooperstock, and Howard D.Lipshitz Molecular mechanisms of translation initiation in eukaryotes Tatyana V.Pestova, Victoria G.Kolupaeva, Ivan B.Lomakin, Evgeny V.Pilipenko, Ivan N.Shatsky, Vadim I.Agol, and Christopher U.T.Hellen The target of rapamycin (TOR) proteins Brian Raught, Anne-Claude Gingras, and Nahum Sonenberg The physiological significance of ß-actin mRNA localization in determining cell polarity and directional motility Elena A.Shestakova, Robert H.Singer, and John Condeelis Sorting and directed transport of membrane proteins during development of hippocampal neurons in culture M.A.Silverman, S.Kaech, M.Jareb, M.A.Burack, L.Vogt, P.Sonderegger, and G.Banker Molecular organization of the postsynaptic specialization Morgan Sheng A cellular mechanism for targeting newly synthesized mRNAs to synaptic sites on dendrites Oswald Steward and Paul F.Worley Think globally, translate locally: What mitotic spindles and neuronal synapses have in common Joel D.Richter Vasopressin mRNA localization in nerve cells: Characterization of cis-acting elements and trans-acting factors Evita Mohr, Nilima Prakash, Kerstin Vieluf, Carola Fuhrmann, Friedrich Buck, and Dietmar Richter Local translation of classes of mRNAs that are targeted to neuronal dendrites James Eberwine, Kevin Miyashiro, Janet Estee Kacharmina, and Christy Job Cytoskeletal microdifferentiation: A mechanism for organizing morphological plasticity in dendrites Stefanie Kaech, Hema Parmar, Martijn Roelandse, Caroline Bornmann, and Andrew Matus Tracking the estrogen receptor in neurons: Implications for estrogen-induced synapse formation Bruce McEwen, Keith Akama, Stephen Alves, Wayne G.Brake, Karen Bulloch, Susan Lee, Chenjian Li, Genevieve Yuen, and Teresa A.Milner Synaptic regulation of protein synthesis and the fragile X protein William T.Greenough, Anna Y.Klintsova, Scott A.Irwin, Roberto Galvez, Kathy E.Bates, and Ivan Jeanne Weiler

6997

6999 7004 7012 7018 7025 7029 7037 7045 7051 7058 7062 7069 7072 7080 7086 7093 7101


CONTENTS iv


MOLECULAR KINESIS IN CELLULAR FUNCTION AND PLASTICITY

6997

Colloquium Molecular kinesis in cellular function and plasticity

Henri Tiedge* †, Floyd E.Bloom‡, and Dietmar Richter§ of Physiology and Pharmacology, and Department of Neurology, State University of New York, Health Science Center, Brooklyn, NY 11203; ‡Department of Neuropharmacology, Scripps Research Institute, La Jolla, CA 92037; and §Institut für Zellbiochemie und klinische Neurobiologie, Universität Hamburg, D-20246 Hamburg, Germany Intracellular transport and localization of cellular components are essential for the functional organization and plasticity of eukaryotic cells. Although the elucidation of protein transport mechanisms has made impressive progress in recent years, intracellular transport of RNA remains less well understood. The National Academy of Sciences Colloquium on Molecular Kinesis in Cellular Function and Plasticity therefore was devised as an interdisciplinary platform for participants to discuss intracellular molecular transport from a variety of different perspectives. Topics covered at the meeting included RNA metabolism and transport, mechanisms of protein synthesis and localization, the formation of complex interactive protein ensembles, and the relevance of such mechanisms for activity-dependent regulation and synaptic plasticity in neurons. It was the overall objective of the colloquium to generate momentum and cohesion for the emerging research field of molecular kinesis. The meeting-bound researcher, approaching one of our cities by air, cannot help but muse about the similarities in the way cities and cells appear to be organized. As the urban arteries come into view—streets, highways, railroad tracks—one can observe traffic in diverse forms, cars, trucks and buses traveling to their various destinations, trains proceeding along their tracks. The underlying rationale for every single movement may not be apparent to our airborne observer, but it is obvious that for the city to operate urban transportation is a prerequisite. Conversely, occasional congestion or traffic jams would indicate a breakdown of local traffic flows even if the cause of any such breakdown may not be immediately obvious from a bird’s-eye perspective. Upon such reflections on urban traffic, and on the parallels with cellular transportation, our biologist may further ponder on purpose, underlying principles and mechanisms of the latter. Like cities, cells have developed diverse transport systems to ensure that the right components are delivered to, or manufactured at, the right location at the right time. What are these transport systems? What are the intracellular roads or tracks, what are the engines and motors, and how do they operate? How are the various types of cargo shipped, and how is such transport tailored to demand? What determines whether it is the finished product that is shipped, or rather smaller parts or subunits for local on-site assembly? How are such mechanisms regulated to maintain cellular function, react to physiological stimuli, and ensure flexible adaptation to changing environments? These were some of the more basic questions that were addressed at the National Academy of Sciences Colloquium on Molecular Kinesis in Cellular Function and Plasticity, held at the Arnold and Mabel Beckman Center in Irvine, California, December 7–9, 2000. This colloquium was conceived as interdisciplinary in nature, bringing together researchers who examine principles of intracellular molecular motion from a diverse range of viewpoints. It has become apparent over the last few years that intracellular transport and localization of both proteins and RNAs play important roles in the development and function of eukaryotic cells as diverse as yeast and neurons. However, although both mechanisms have been implicated in the establishment and maintenance of cellular polarity and plasticity, the two fields have developed essentially in parallel, with little interdisciplinary contact. Mechanisms of intracellular organelle transport have by now been sufficiently well established, as have the modes of action of various motor proteins that are underlying such mechanisms. In contrast, proteins responsible for RNA localization are only now being identified, and RNA-transporting molecular motors have remained elusive. Cross-disciplinary interactions between the areas of protein kinesis and RNA kinesis have been informal and sporadic. It was therefore the explicit intent of the National Academy Colloquium to overcome this fragmentation by providing a formal joint forum for scientific exchange between these disciplines. *Department

MOLECULAR MOTORS Motor proteins such as myosins, dyneins, and kinesins are the engines of intracellular molecular transport. Kinesins in particular are seen as major movers in neurons as they have been implicated in microtubule-based transport in both axons and dendrites (1, 2). Kinesins form a rather large super family, and the individual superfamily proteins operate as motor molecules in various cell types with diverse cargoes. Given that transportation requirements are particularly demanding and complex in neurons, it does not come as a surprise that the highest diversity of kinesins is found in brain. In neurons, kinesin and dynein motor molecules have not only been implicated in intracellular axonal and dendritic transport, but also in neuronal pathfinding and migration (1). Given the various fundamental cellular functions they subserve in neurons, such mechanisms, should they become defective, also can be expected to contribute to onset or progression of neurological disorders. TRANSLATION INITIATION In eukaryotic cells, the flow of information originates in the nucleus. Subsequent to its export into the cytoplasm, an mRNA may be translated in the perikaryal somatic region, or it may continue its travel to distant extrasomatic destinations for local on-site translation. These mechanisms may not be mutually exclusive for any given mRNA, but it is assumed that while en route, mRNAs are not actively translated. Many mRNAs, including those that are transported to and translated at extrasomatic target sites, are likely to be subject to specific translational control. Significant progress has been made in recent years in the functional dissection of translation initiation complexes and pathways (3–5), and it appears plausible, in view of such work, that translation initiation mechanisms play important roles in the

This paper is the introduction to the following papers, which were presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. †To whom reprint requests should be addressed at: Department of Physiology and Pharmacology, State University of New York, Health Science Center, 450 Clarkson Avenue, Brooklyn, NY 11203. E-mail: [email protected].


MOLECULAR KINESIS IN CELLULAR FUNCTION AND PLASTICITY

6998

regulation of protein synthesis both at perykaryal somatic and at distant extrasomatic sites. LOCALIZED RNAS The analysis of RNA transport and localization has in recent years matured into a novel discipline in cell biology and neuroscience. In traditional cell biology, proteins are manufactured in the perikaryal soma and subsequently delivered to their respective sites of function. Although this may often be so, it is now accepted that this scenario does not necessarily represent the whole story. In diverse cell types, RNAs have been identified that are targeted to specific subcellular locations for on-site translation (6). In 1982, the first such localized mRNA, encoding myelin basic protein, was identified in oligodendrocytes (7). Subsequently, RNA localization was documented also in Xenopus oocytes, Drosophila embryos, and a variety of somatic eukaryotic cell types ranging from fibroblasts to neurons (8–12). In neurons, localized RNAs were discovered rather late, and only after the presence of polyribosomes in postsynaptic dendritic microdomains (13, 14) had already been documented for a while. The first three RNAs identified in dendrites were the mRNAs encoding MAP2 (15) and CaMKIIα (16) as well as BC1 RNA, a noncoding RNA polymerase III transcript (17). These were joined by neuropeptideencoding transcripts in the axonal domain (18). Today, research is focused on the mechanism of RNA transport in neuronal processes and on the elucidation of the signals involved—both at the level of RNA (cis-acting elements) and proteins (trans-acting factors). This work eventually will shed light on how a neuron administers translation of a distinct mRNA at or near a synapse in an input-specific and activitydependent manner (11, 19). NEURONAL PLASTICITY In terms of subcellular location, the ultimate and critical determinant of cellular function is of course a correct spatio-temporal expression pattern of the protein repertoire, regardless of whether any given protein is delivered from the perikaryon or synthesized locally on site. Consequently, given the paramount importance of subcellular “location” in particular in neurons, protein targeting and anchoring mechanisms will directly impact long-term neuronal plasticity and are likely to figure prominently in the development of neurological disorders. In this respect, the discovery of novel scaffolding multidomain proteins that are involved in the functional organization of the postsynaptic density has significantly furthered our understanding of how signal transduction pathways might be regulated at the synapse. Activity-dependent modification of protein structure, location, and/or interaction may be essential for the molecular reorganization of postsynaptic functional architecture (20, 21). In addition, local translation of mRNA(s) encoding one or several of the scaffolding proteins also may contribute to the dynamic plasticity at a postsynaptic specialization after stimulation (21). It then appears that neurons, being among the spatially most extended and functionally most complex of all eukaryotic cells, have to cope with organizational tasks that are indeed reminiscent of those associated with the maintenance and development of large metropolitan areas. And it thus holds true for cities and cells alike that the larger and more complex they are, the more relevant becomes an old New Yorker real estate adage that of all determinants of functional value, none are more important than the following three: location, location, location. We thank the National Academy of Sciences for encouragement in planning this colloquium and for generous financial and administrative support. We also thank Mr. E.Patte of the National Academy of Sciences Executive Office and Ms. M.Gray-Kadar of the Beckman Center for their help in organizing the meeting and the National Academy for providing the excellent resources and facilities of the Arnold and Mabel Beckman Center in Irvine. 1. Goldstein, L.S. & Yang, Z. (2000) Annu. Rev. Neurosci. 23, 39–71. 2. Hirokawa, N. (1998) Science 279, 519–526. 3. Sachs, A.B., Sarnow, P. & Hentze, M.W. (1997) Cell 89, 831–838. 4. Gingras, A. C, Raught, B. & Sonenberg, N. (1999) Annu. Rev. Biochem. 68, 913–963. 5. Pestova, T.V. & Hellen, C.U.T. (1999) Trends Biochem. Sci. 24, 85–87. 6. Bassell, G.J., Oleynikov, Y. & Singer, R.H. (1999) FASEB J. 13, 447–454. 7. Colman, D.R., Kreibich, G., Frey, A.B. & Sabatini, D.D. (1982) J. Cell Biol. 95, 598–608. 8. Singer, R.H. (1992) Curr. Opin. Cell Biol. 4, 15–19. 9. St Johnston, D. (1995) Cell 81, 161–170. 10. Steward, O. (1997) Neuron 18, 9–12. 11. Tiedge, H., Bloom, F.E. & Richter, D. (1999) Science 283, 186–187. 12. Richter, D., ed. (2001) Cell Polarity and Subcellular RNA Localization (Springer, Berlin). 13. Steward, O. & Levy, W.B. (1982) J.Neurosci. 2, 284–291. 14. Steward, O. & Reeves, T.M. (1988) J.Neurosci. 8, 176–184. 15. Garner, C.C., Tucker, R.P. & Matus, A. (1988) Nature (London) 336, 674–677. 16. Burgin, K.E., Waxham, M.N., Rickling, S., Westgate, S.A., Mobley, W.C. & Kelly, P.T. (1990) J.Neurosci. 10, 1788–1798. 17. Tiedge, H., Fremeau, R.T., Jr., Weinstock, P.H., Arancio, O. & Brosius, J. (1991) Proc. Natl Acad. Sci. USA 88, 2093–2097. 18. Mohr, E., Fehr, S. & Richter, D. (1991) EMBO J. 10, 2419–2424. 19. Kiebler, M.A. & DesGroseillers, L. (2000) Neuron 25, 19–28. 20. Husi, H., Ward, M.A., Choudhary, J.S., Blackstock, W.P. & Grant, S.G. (2000) Nat. Neurosci. 3, 661–669. 21. Sheng, M. & Kim, E. (2000) J.Cell Sci. 113, 1851–1856.


KINESIN MOLECULAR MOTORS: TRANSPORT PATHWAYS, RECEPTORS, AND HUMAN DISEASE

6999

Colloquium Kinesin molecular motors: Transport pathways, receptors, and human disease Lawrence S.B.Goldstein* Howard Hughes Medical Institute, Department of Cellular and Molecular Medicine, University of California at San Diego School of Medicine, 9500 Gilman Drive, La Jolla, CA 92093–0683 Kinesin molecular motor proteins are responsible for many of the major microtubule-dependent transport pathways in neuronal and non-neuronal cells. Elucidating the transport pathways mediated by kinesins, the identity of the cargoes moved, and the nature of the proteins that link kinesin motors to cargoes are areas of intense investigation. Kinesin-II recently was found to be required for transport in motile and nonmotile cilia and flagella where it is essential for proper left-right determination in mammalian development, sensory function in ciliated neurons, and opsin transport and viability in photoreceptors. Thus, these pathways and proteins may be prominent contributors to several human diseases including ciliary dyskinesias, situs inversus, and retinitis pigmentosa. Kinesin-I is needed to move many different types of cargoes in neuronal axons. Two candidates for receptor proteins that attach kinesin-I to vesicular cargoes were recently found. One candidate, Sunday driver, is proposed to both link kinesin-I to an unknown vesicular cargo and to bind and organize the mitogen-activated protein kinase components of a c-Jun Nterminal kinase signaling module. A second candidate, amyloid precursor protein, is proposed to link kinesin-I to a different, also unknown, class of axonal vesicles. The finding of a possible functional interaction between kinesin-I and amyloid precursor protein may implicate kinesin-I based transport in the development of Alzheimer’s disease. The large size and extreme polarity of neurons presents these cells with an unusual and substantial transport challenge. Materials synthesized in the cell body must be transported down long axons to presynaptic sites of utilization. These distances can reach 1 m or more in the case of humans and larger animals, and axonal volumes can exceed the volume of the cell body by 1,000-fold or more. In addition, axons and dendrites can be highly branched, and in some cases have very small diameters, which can limit transport rate and volume. The polarity of neurons presents analogous problems. Structural and signaling components destined for the axon must somehow be sorted from components needed in dendrites; the transport system appears to play a critical role in these processes (1). The combination of the substantial pressure of distance and volume, coupled to the enormous branching and narrow caliber of many neuronal processes, suggests that the intracellular transport system could be the “Achilles heel” of these large, complex cells—easily disturbed by environmental insult, mutation, or other trauma to cause neurodegenerative disease. This possibility has been suggested repeatedly over the past decades, but without a great deal of supporting evidence (e.g., refs. 2 and 3). This article revisits these themes and discusses data that suggest a possible interplay of kinesin molecular motor-based neuronal transport pathways and human disease. LESSONS FROM GREEN ALGAE: POSSIBLE LINKS OF INTRAFLAGELLAR TRANSPORT TO HUMAN DISEASE A non-neuronal transport system that has the potential to teach us a great deal about neuronal transport recently was discovered in the green alga, Chlamydomonas reinhardii (reviewed in ref. 4). These small, free-living, unicellular organisms have long flagella that are used to swim. Flagellar assembly appears to occur at the site most distant from the cell body, and there is strong evidence that a kinesin-based transport pathway is responsible for moving key membrane and flagellar components from sites of synthesis in the cell body to sites of assembly. This system uses an evolutionarily conserved kinesin called kinesin-II to power the movement of proteinaceous “rafts.” These rafts are closely apposed to the flagellar plasma membrane as they move along the outer surface of flagellar microtubules. Kinesin-II is composed of two related motor polypeptides, KIF3A and KIF3B in mammals, coupled to a nonmotor kinase-associated protein subunit (reviewed in ref. 5). Several raft complex proteins also have been identified and found to be highly conserved from algae to mammals (6, 7). Mutations in the gene encoding the KIF3A or KIF3B subunits in mice cause an embryonic lethal phenotype (8–10). Strikingly, the cilia normally present on cells of the embryonic node fail to form in these mutants, confirming the broad evolutionary requirement for a kinesinII based transport pathway for flagellar assembly. In addition to missing nodal cilia, embryos lacking KIF3A and KIF3B exhibit defective left-right body axis determination, providing strong experimental support for the long-standing hypothesis that cilia are crucial to left-right body axis determination in mammals. Similarly, mouse mutants lacking a homologue of a raft complex protein also fail to form embryonic nodal cilia and have defective left-right body axis determination (7, 11). It is noteworthy that a complex of heterogeneous human diseases called Kartagener’s triad or primary ciliary dyskinesia have been known for some time and appear to result from defects in bronchial cilia and sperm flagella. Thus, these diseases generally present with male infertility (sperm motility defects), bronchial abnormalities (bronchial ciliary defects), and situs inversus (defects in left-right body axis determination, causes previously unknown). These syndromes were previously suggested, with little supporting evidence, to alter embryonic cilia in human embryos (12). An intriguing possibility is that the components of the flagellar transport pathways may identify susceptibility loci for this class of human diseases. In addition to typical, usually motile, cilia, eukaryotes have an array of cells that bear modified nonmotile cilia, often to serve sensory functions. Among these are so-called primary cilia whose functions are unknown (reviewed in ref. 13). Recently, a mouse homologue of a raft complex protein surfaced as a gene that when mutant causes polycystic kidney disease and leads to

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: KHC, kinesin heavy chain; KLC, kinesin light chain; APR, amyloid precursor protein. *E-mail: [email protected].



7000

shorter primary cilia in the kidney (7, 11). It was suggested that these cilia function in the kidney to sense ionic concentrations, disturbance of which leads to disease. Similar mutants lacking kinesin-II motor or raft complex homologues in Caenorhabditis elegans disturb the structure and function of nonmotile chemosensory cilia (6).

Fig. 1. Schematic diagram of mammalian photoreceptor. Microtubule organization and location of major cellular organelles are shown. In the inner segment, microtubules have their minus ends located near the basal bodies; connecting cilium microtubules have their minus ends at the basal body as well. ER, endoplasmic reticulum. Perhaps the most distinctive use of nonmotile cilia is presented by the vertebrate photoreceptor. This neuronal cell has an axon, but in place of a typical dendritic arbor it has a cellular compartment called the inner segment in which most biosynthesis takes place. Components such as opsin that are needed to sense light then are transported to sites of utilization in the disks of the outer segment (Fig. 1). Transport appears to occur through a narrow isthmus or connecting cilium, which is structurally a typical nonmotile cilium. A substantial amount of material must be moved through the connecting cilium because the photoreceptor turns over ca. 10% of its mass daily. Thus, it is perhaps not surprising that kinesin-II has been reported by a number of groups to be localized in the connecting cilium of the photoreceptor (14–16). These observations suggested that the transport system found in more typical cilia and flagella might be harnessed to move opsin, and perhaps other photoreceptor components, from the inner segment to the outer segment through the connecting cilium. Recently, specific removal of kinesin-II from photoreceptors using the lox-cre system was found to cause a substantial accumulation of opsin and arrestin in the inner segment accompanied by apoptosis. It was suggested that this phenotype was caused by a defect in transport of opsin and arrestin from the inner segment to the outer segment (17). Similar phenotypes have been seen in a particular class of opsin mutants that cause retinitis pigmentosa in humans. These mutants have been suggested to interfere with opsin transport and cause opsin accumulation in the inner segment and apoptosis (18–20). The region of opsin to which these mutants map also appears to interact with the dynein molecular motor (21), further suggesting a role for transport dysfunction in the development of degenerative retinal diseases such as retinitis pigmentosa. As with primary ciliary dyskinesia, it is tempting to speculate that the collection of genes encoding the components required for transport from the inner segment to the outer segment, and in particular for opsin transport, may represent susceptibility loci for retinitis pigmentosa and other diseases where photoreceptor degeneration is a central feature. Indeed, it is intriguing that myosin VIIA, which when mutant can cause retinitis pigmentosa in humans (but curiously not mice), has been suggested to play a minor role in opsin transport and to be localized in the connecting cilium of the photoreceptor in addition to the retinal pigment epithelium (22, 23).

Fig. 2. Organization of kinesin-I. Two heavy chain components (KHC) and two light chain components (KLC) form the native heterotetramer. Proposed TPR domains are thought to mediate cargo binding via protein-protein interactions. Finally, in thinking about neuronal transport pathways, it is striking that kinesin-II has been found in many typical neurons that lack cilia (24–27). In Drosophila, mutants lacking a kinesin-II subunit exhibit defects in axonal transport of choline acetyltransferase, a possibly cytosolic enzyme (28). In mammals, antibody inhibition, two-hybrid and biochemical experiments suggest a direct functional linkage between kinesin-II and non-erythroid spectrin (fodrin) in neurons (29). Perhaps nonciliated neurons also use a raft-based kinesin-II transport system to move cytoplasmic proteins in association with membrane-associated rafts or vesicles. An intriguing possibility is that kinesin-II and associated raft complexes might play an important role in the movement of cytosolic proteins by the slow axonal transport system. Further experimental work is needed to test this idea. LESSONS FROM FRUIT FLIES: ANTEROGRADE AXONAL TRANSPORT AND MITOGEN-ACTIVATED PROTEIN KINASE SIGNALING Conventional kinesin, kinesin-I, was first discovered in a squid fast axoplasmic transport system, prompting early suggestions that kinesin-I would be an important motor protein to power fast anterograde axonal transport. This suggestion has been amply supported by a large number of antibody, antisense, and genetic experiments that support a general role of kinesin-I in axonal transport, but have not clearly linked this motor protein to a particular type of vesicular cargo (reviewed in ref. 30). It is thus not surprising that a “receptor” that mediates the attachment of kinesin-I to vesicular cargoes and other organelles has been elusive. In addition, whether it is the kinesin heavy chain (KHC) or the kinesin light chain (KLC) subunit of kinesin-I (Fig. 2) that binds to cargo has been unclear. Although a protein called kinectin has been suggested to play a role in linking kinesin-I to vesicles in non-neuronal cells (31, 32), its apparent absence in mammalian axons, Drosophila, and Caenorhabditis (33–35) has motivated additional searches for kinesin-I cargo receptors. Two serious candidates recently have emerged. One called Sunday driver was found in a genetic screen for axonal transport mutants in Drosophila (36). The other called amyloid precursor protein (APP) was identified initially in biochemical experiments (37). The genetic screen that identified syd was based on work in Drosophila that revealed a constellation of phenotypes common to mutants defective in components of the anterograde or



7001

retrograde axonal transport systems (3, 38, 39). This phenotype includes a relatively late larval lethality coupled to asymmetric paralysis of the animal. This paralysis manifests as either an upward tail “flip” during larval crawling or frank posterior paralysis of the motile larva. The underlying cellular phenotype is an accumulation of vesicles and organelles in apparent “traffic jams” or “clogs” in long narrow caliber axons. The first such example was presented by mutants lacking KHC, followed by mutants lacking KLC, dynein, and dynactin components, all components of the motor proteins themselves. The first nonmotor protein subunit found to cause this phenotype when missing is encoded by the Sunday driver gene, syd, which was found in a screen for mutants with the axonal transport phenotypic constellation (36). The syd gene was found to encode an evolutionarily highly conserved protein predicted to be a type II transmembrane protein. Similar proteins are found in Caenorhabditis and mammals, which have two related genes encoding syd homologues. Because antibodies specific for syd are thus far of poor quality, localization of syd could not be accomplished, but transaction experiments with green fluorescent protein-tagged mouse syd in cultured mammalian COS cells revealed that syd could target to tubulovesicular organelles and small vesicles. These structures costain with antibodies recognizing both a marker of the secretory pathway and KLC, but not with probes for mitochondria or the endoplasmic reticulum-Golgi intermediate compartment. Two-hybrid coimmunoprecipitation and direct binding analyses demonstrated that the syd protein can bind directly to the KLC subunit of kinesin-I. Strikingly the interaction appeared to be with the predicted TPR repeat domains of KLC, which have previously been implicated in kinesin-I attachment to vesicular cargoes in axons (40). The combination of the axonal transport defective phenotype of syd mutants, the tubulovesicular localization of the protein in transfected cells, and direct binding of syd to KLC lead to the proposal that syd has a function as a kinesin-I receptor for at least one class of vesicles transported in the axon. In a surprising development, it turns out that mammalian syd is identical to a previously discovered gene called JIP3 or JSAP1, which was found to encode a protein having a protein kinase scaffold function that can bind and organize the mitogenactivated protein kinase components of a c-Jun N-terminal kinase signaling module (41, 42). Although not previously recognized as a membraneassociated protein the data suggest some role of syd (JIP3/JSAP1) in signaling networks in addition to kinesin-I attachment. Although a simple possibility is that a signaling protein has a dual function as a kinesin-I motor receptor, it is also possible that syd serves to integrate mitogenactivated protein kinase signaling with the regulation of some kinesin-I transport pathways. LESSONS FROM HUMANS: FROM ALZHEIMER’S DISEASE TO KINESIN-I RECEPTORS APP was identified because of its possible role in the initiation or progression of Alzheimer’s disease (reviewed in refs. 43 and 44). APP is a type I transmembrane protein whose normal cellular function is poorly understood. Null mutants in both Drosophila and mice are viable and have relatively minor neuronal phenotypes (45, 46). However, proteolytic fragments of the APP are an abundant component of the plaques found throughout the brains of people afflicted with Alzheimer’s disease. Missense mutants in the gene encoding APP cause some forms of familial Alzheimer’s disease whereas mutants in presenilin genes cause others. Both types of mutants appear to increase the number of plaques and the abundance of toxic proteolytic fragments of APP. The presenilin genes may encode one of the key proteases, thus accounting for their role in disease. There also have been suggestions that axonal transport dysfunction or aberrant trafficking of APP might be an important element in causing disease. The possibility that APP might have a kinesin-I receptor function was suggested initially by coimmunoprecipitation studies (37). Subsequent analyses of velocity gradient sedimentation, microtubule-binding, and direct binding analyses with expressed proteins confirmed this interaction and revealed that like syd, APP binds directly and tightly to the tetratrico peptide repeat region of KLC (37). In addition, although previous antisense experiments revealed that kinesin-I was needed for APP transport in neurons, whether this was direct or indirect, or a reflection of axonal versus preaxonal events was unclear (47–49). Analysis of mice lacking the neuron-enriched form of KLC, KLC1, revealed that APP transport in sciatic nerve axons strongly depended on KLC1, showing dramatic reduction in its absence. Thus, based on the tight binding of APP to KLC, and the strong dependence of APP axonal transport on KLC1, it was proposed that APP has a function as a kinesin-I receptor for a class of vesicular cargoes in the axon, perhaps distinct from vesicles whose transport is mediated by syd. KINESIN RECEPTORS AND KINESIN REGULATION Taken together, these data on potential kinesin receptors suggest that proteins that have other roles in the cell may function to attach kinesin-I to cellular vesicles. This suggestion fits nicely with recent findings that other previously recognized proteins with nontransport functions may have dual roles as receptors and adaptors for motor proteins (reviewed in ref. 50). In fact, an intriguing possibility is that there are many cellular proteins that interact directly with the transport machinery to mediate movement. This view is an interesting alternative to the possibility that there are only a few proteins that interact directly with motor proteins, and that many proteins depend on these few “motor receptor” proteins for their transport. Further work is needed to evaluate these ideas. What then are the relative roles of KLC and the KHC tail in binding cargo and regulating motor activity? Formulation of a compelling model is complicated by the apparent contradictions in the experimental literature to date (reviewed in ref. 37). In brief, the tail domain of KHC has been reported both to repress the KHC motor activity and to bind membranes and perhaps cargoes in the absence of KLC. Some fungi also appear to have KHC but not KLC. Yet, mutants that lack KLC in flies and mice have significant phenotypes, and antibodies that bind KLC can block membrane binding of kinesin-I. KLC also has been suggested to have a function as either an activator or represser of KHC motor activity (51, 52). Although it is possible that one or more of these observations is incorrect, a model that accounts for most of the data has been proposed (37). In this model, both KHC and KLC are suggested to repress the KHC motor activity and both KHC and KLC have membrane binding activity. Binding to both is suggested to derepress the motor and initiate transport. Thus, in the absence of KLC in organisms that ordinarily have it, KHC cannot initiate transport of classes of cargo that require KLC for attachment and derepression. Organisms that ordinarily lack KLC naturally may rely solely on the KHC tail for membrane binding and repression. A SPECULATIVE PROPOSAL FOR THE RELATIONSHIP OF AXONAL TRANSPORT TO THE INITIATION OF ALZHEIMER’S DISEASE At present, most workers accept the hypothesis that inappropriate proteolytic processing of APP to generate aggregates of Aß is an important early event in the pathogenesis of Alzheimer’s disease. Sorely lacking, however, is an understanding of whether inappropriate processing of APP is the initiating event in disease, and if so, why it occurs. Additional important holes in our understanding of disease include knowledge of where in the neuron this inappropriate processing takes place, i.e., in the



7002

axonal, dendritic, synaptic, or cell body compartment, and why Aß is neurotoxic. Several groups have suggested that axonal transport defects may occur later in the pathogenesis of the disease (e.g., ref. 53), but have left unanswered the question of whether it might be an initiating event as well. Could the transport function of APP be related to the initiation or progression of Alzheimer’s disease? Several observations suggest that the answer to this question could be yes. First, as just discussed, normal transport of the APP protein in mammalian axons appears to depend on a direct interaction with the KLC subunit of the kinesin-I molecular motor protein (37). Thus, the disease causing protein may have a kinesin-I receptor function and thus be in close apposition to the transport machinery. Second, overexpression of the Drosophila homologue of APP called APPL in Drosophila (54) causes axonal clogs analogous to those found in syd and many other mutants with defective axonal transport. Perhaps overexpression of a protein such as APP with a motor receptor function either titrates out needed kinesin motor function in the axon or unbalances traffic in these narrow caliber axons leading to transport dysfunction, clogging, and other abnormalities. Third, humans bearing trisomy 21 suffer from premature onset of the symptoms characteristic of Alzheimer’s disease, perhaps because of overproduction of Aß (55). Although other genes are clearly present in excess in trisomy 21, it is striking that the gene encoding APP is located on chromosome 21 and thus is certainly one of the genes overexpressed in these people. Experiments in mouse models that overexpress APP have given equivocal results, but in some cases similar phenotypes have been reported (reviewed in ref. 56). Fourth, one of the early phenotypes of mouse models of Alzheimer’s disease, and perhaps human Alzheimer’s disease, is “dystrophic neurites,” whose morphology includes organelle and vesicle accumulations in axons (e.g., ref. 57). This phenotype is strikingly reminiscent of the axonal clogging observed after disturbance of axonal transport in Drosophila. Fifth, while it is unclear where in neurons Aß production is prominent, it is striking that even though APP is widely expressed, Alzheimer’s disease is primarily a neuronal disease. One feature that sets neurons apart from other cells is long axonal and dendritic processes. That a critical function of these processes is movement and transport of vesicular cargoes may be important to the development of disease. How could alterations in axonal transport of APP lead to the generation of excess Aß and Alzheimer’s disease? Perhaps enhanced proteolysis of APP caused by axonal damage, presenilin or APP mutations, or elevated APP levels, cause impairments of APP transport efficiency in axons and increase the time spent by APP and proteases in a common axonal vesicular transport compartment. Timedependent or damage-induced generation and accumulation of Aß by proteolysis of APP in this compartment might lead to aggregates of Aß that impair or block axonal transport and further stimulate Aß production in an autocatalytic spiral. Such a process could lead to neuronal dysfunction and progressive, age-related neurodegeneration and disease. In fact, although not measured directly by any researcher, it is possible that the populations of neurons affected first in Alzheimer’s disease could be those that combine the narrowest caliber with a higher than usual transport burden. Such features might predispose these axons to aggregation, or reduction in velocity, of their axonal transport cargoes, analogous to what has been observed in genetic models of axonal transport disturbance in Drosophila. Ultimately, neurotrophic signaling in neurons could be blocked by formation of axonal clogs, leading to apoptotic neuronal cell death. This proposal also may explain why some people appear to be more susceptible to Alzheimer’s disease than others. Perhaps the degree of axonal branching and caliber and perhaps allelic state at crucial molecular motor subunit genes will be found to be important once explored. If correct, this view also can account for the observation that it generally takes decades for Alzheimer’s disease to develop. Slight decrements in transport rate or efficiency could lead to slightly enhanced proteolysis rates that will in turn eventually lead to Alzheimer’s disease. Clearly, further work to test these ideas is needed. CONCLUDING REMARKS We may be at the beginning of an era in which neuronal transport is recognized as a major cellular target for the development of neurodegenerative disease. Although ciliary dyskinesias, retinitis pigmentosa, and Alzheimer’s disease are the major examples discussed above, there are also suggestions that amyotrophic lateral sclerosis may be caused or complicated by transport defects in motor neurons (58, 59). Similarly, it may be more than a coincidence that huntingtin, tau, and ApoE4, all of which are implicated in causation or susceptibility to neurodegenerative disease, all have been suggested to modulate transport when experimentally manipulated or to interact directly with the transport machinery (60–65). Finally, it is possible that for those neurodegenerative diseases in which formation of aggregates is an important feature, inhibition of axonal transport by the aggregates may be an important element in disease progression. In this regard, a recent report that axonal blockages and possible impairment of axonal transport may be present in CreutzfeldtJacob disease is intriguing (66). I am an Investigator of the Howard Hughes Medical Institute. Some of the work discussed in this article was supported by National Institutes of Health Grant GM35252. 1. Burack, M.A., Silverman, M.A. & Banker, G. (2000) Neuron 26, 465–472. 2. Griffin, J.W. & Watson, D.F. (1988) Ann. Neurol. 23, 3–13. 3. Hurd, D.D. & Saxton, W.M. (1996) Genetics 144, 1075–1085. 4. Rosenbaum, J.L., Cole, D.G. & Diener, D.R. (1999) J. Cell Biol. 144, 385–388. 5. Marszalek, J.R. & Goldstein, L.S. (2000) Biochim. Biophys. Acta 1496, 142–150. 6. Cole, D.G., Diener, D.R., Himelblau, A.L., Beech, P.L., Fuster, J.C. & Rosenbaum, J.L. (1998) J. Cell Biol. 141, 993–1008. 7. Pazour, G.J., Dickert, B.L., Vucica, Y., Seeley, E.S., Rosenbaum, J.L., Witman, G.B. & Cole, D.G. (2000) J. Cell Biol. 151, 709–718. 8. Nonaka, S., Tanaka, Y., Okada, Y., Takeda, S., Harada, A., Kanai, Y., Kido, M. & Hirokawa, N. (1998) Cell 95, 829–837. 9. Marszalek, J.R., Ruiz-Lozano, P., Roberts, E., Chien, K.R. & Goldstein, L.S. (1999) Proc. Natl. Acad. Sci. USA 96, 5043–5048. 10. Takeda, S., Yonekawa, Y., Tanaka, Y., Okada, Y., Nonaka, S. & Hirokawa, N. (1999) J.Cell Biol. 145, 825–836. 11. Murcia, N.S., Richards, W.G., Yoder, B.K., Mucenski, M.L., Dunlap, J.R. & Woychik, R.P. (2000) Development (Cambridge, U.K.) 127, 2347–2355. 12. Afzelius, B.A. (1985) CRC Crit. Rev. Biochem. 19, 63–87. 13. Wheatley, D.N., Wang, A.M. & Strugnell, G.E. (1996) Cell Biol. Int. 20, 73–81. 14. Beech, P.L., Pagh-Roehl, K., Noda, Y., Hirokawa, N., Burnside, B. & Rosenbaum, J.L. (1996) J. Cell Sci. 109, 889–897. 15. Muresan, V., Bendala-Tufanisco, E., Hollander, B.A. & Besharse, J.C. (1997) Exp. Eye Res. 64, 895–903. 16. Muresan, V., Lyass, A. & Schnapp, B.J. (1999) J. Neurosci. 19, 1027–1037. 17. Marszalek, J.R., Liu, X., Roberts, E.A., Chui, D., Marth, J.D., Williams, D.S. & Goldstein, L.S. (2000) Cell 102, 175–187. 18. Sung, C.H., Makino, C., Baylor, D. & Nathans, J. (1994) J. Neurosci. 14, 5818–5833. 19. Portera-Cailliau, C., Sung, C.H., Nathans, J. & Adler, R. (1994) Proc. Natl. Acad. Sci. USA 91, 974–978. 20. Sung, C.H., Davenport, C.M. & Nathans, J. (1993) J. Biol. Chem. 268, 26645–26649. 21. Tai, A.W., Chuang, J.Z., Bode, C., Wolfrum, U. & Sung, C.H. (1999) Cell 97, 877–887. 22. Liu, X., Vansant, G., Udovichenko, I.P., Wolfrum, U. & Williams, D.S. (1997) Cell Motil. Cytoskeleton 37, 240–252. 23. Liu, X., Udovichenko, I.P., Brown, S.D., Steel, K.P. & Williams, D.S. (1999) J. Neurosci. 19, 6267–6274.



7003

24. Kondo, S., Sato-Yoshitake, R., Noda, Y., Aizawa, H., Nakata, T., Matsuura, Y. & Hirokawa, N. (1994) J. Cell Biol. 125, 1095–1107. 25. Muresan, V., Abramson, T., Lyass, A., Winter, D., Porro, E., Hong, F., Chamberlin, N.L. & Schnapp, B.J. (1998) Mol. Biol. Cell 9, 637–652. 26. Yamazaki, H., Nakata, T., Okada, Y. & Hirokawa, N. (1995) J. Cell Biol. 130, 1387–1399. 27. Yang, Z. & Goldstein, L.S. (1998) Mol. Biol. Cell 9, 249–261. 28. Ray, K., Perez, S.E., Yang, Z., Xu, J., Ritchings, B.W., Steller, H. & Goldstein, L.S. (1999) J. Cell Biol. 147, 507–518. 29. Takeda, S., Yamazaki, H., Seog, D.H., Kanai, Y., Terada, S. & Hirokawa, N. (2000) J. Cell Biol. 148, 1255–1265. 30. Goldstein, L.S. & Philp, A.V. (1999) Annu. Rev. Cell Dev. Biol. 15, 141–183. 31. Yu, H., Nicchitta, C.V., Kumar, J., Becker, M., Toyoshima, I. & Sheetz, M.P. (1995) Mol. Biol. Cell 6, 171–183. 32. Toyoshima, I., Yu, H., Steuer, E.R. & Sheetz, M.P. (1992) J. Cell Biol. 118, 1121–1131. 33. Toyoshima, I. & Sheetz, M.P. (1996) Neurosci. Lett. 211, 171–174. 34. Goldstein, L.S. & Gunawardena, S. (2000) J.Cell Biol. 150, F63-F68. 35. Rubin, G.M., Yandell, M.D., Wortman, J.R., Gabor Miklos, G. L, Nelson, C.R., Hariharan, I.K., Fortini, M.E., Li, P.W., Apweiler, R., Fleischmann, W., et al. (2000) Science 287, 2204–2215. 36. Bowman, A.B., Kamal, A., Ritchings, B.W., Philp, A.V., McGrail, M., Gindhart, J.G. & Goldstein, L.S. (2000) Cell 103, 583–594. 37. Kamal, A., Stokin, G.B., Yang, Z., Xia, C. & Goldstein, L.S. (2000) Neuron 28, 449–459. 38. Bowman, A.B., Patel-King, R.S., Benashski, S.E., McCaffery, J.M., Goldstein, L.S. & King, S.M. (1999) J. Cell Biol. 146, 165–180. 39. Martin, M., Iyadurai, S.J., Gassman, A., Gindhart, J.G., Jr., Hays, T.S. & Saxton, W.M. (1999) Mol. Biol. Cell 10, 3717–3728. 40. Stenoien, D.L. & Brady, S.T. (1997) Mol. Biol. Cell 8, 675–689. 41. Kelkar, N., Gupta, S., Dickens, M. & Davis, R.J. (2000) Mol. Cell. Biol. 20, 1030–1043. 42. Ito, M., Yoshioka, K., Akechi, M., Yamashita, S., Takamatsu, N., Sugiyama, K., Hibi, M., Nakabeppu, Y., Shiba, T. & Yamamoto, K.I. (1999) Mol. Cell. Biol. 19, 7539–7548. 43. De Strooper, B. & Annaert, W. (2000) J. Cell Sci. 113, 1857–1870. 44. Price, D.L. & Sisodia, S.S. (1998) Annu. Rev. Neurosci. 21, 479–505. 45. Zheng, H., Jiang, M., Trumbauer, M.E., Hopkins, R., Sirinathsinghji, D.J., Stevens, K.A., Conner, M.W., Slunt, H.H., Sisodia, S.S., Chen, H.Y. & Van der Ploeg, L.H. (1996) Ann. N.Y. Acad. Sci. 777, 421–426. 46. Luo, L., Tully, T. & White, K. (1992) Neuron 9, 595–605. 47. Feiguin, F., Ferreira, A., Kosik, K.S. & Caceres, A. (1994) J. Cell Biol. 127, 1021–1039. 48. Ferreira, A., Caceres, A. & Kosik, K.S. (1993) J. Neurosci. 13, 3112–3123. 49. Ferreira, A., Niclas, J., Vale, R.D, Banker, G. & Kosik, K.S. (1992) J. Cell Biol. 117, 595–606. 50. Klopfenstein, D.R., Vale, R.D. & Rogers, S.L. (2000) Cell 103, 537–540. 51. Rahman, A., Kamal, A., Roberts, E.A. & Goldstein, L.S. (1999) J. Cell Biol. 146, 1277–1288. 52. Verhey, K.J., Lizotte, D.L., Abramson, T., Barenboim, L., Schnapp, B.J. & Rapoport, T.A. (1998) J. Cell Biol. 143, 1053–1066. 53. Kasa, P., Papp, H., Kovacs, I., Forgon, M., Penke, B. & Yamaguchi, H. (2000) Neurosci. Lett. 278, 117–119. 54. Torroja, L., Chu, H., Kotovsky, I. & White, K. (1999) Curr. Biol. 9, 489–492. 55. Takashima, S. (1997) Curr. Opin. Neurol. 10, 148–152. 56. Price, D.L., Tanzi, R.E., Borchelt, D.R. & Sisodia, S.S. (1998) Annu. Rev. Genet. 32, 461–493. 57. Masliah, E., Sisk, A., Mallory, M., Mucke, L., Schenk, D. & Games, D. (1996) J. Neurosci. 16, 5795–5811. 58. Williamson, T.L. & Cleveland, D.W. (1999) Nat. Neurosci. 2, 50–56. 59. Xu, Z., Cork, L.C., Griffin, J.W. & Cleveland, D.W. (1993) Cell 73, 23–33. 60. Li, S.H., Gutekunst, C.A., Hersch, S.M. & Li, X.J. (1998) J. Neurosci. 18, 1261–1269. 61. Block-Galarza, J., Chase, K.O., Sapp, E., Vaughn, K.T., Vallee, R.B., DiFiglia, M. & Aronin, N. (1997) NeuroReport 8, 2247–2251. 62. Engelender, S., Sharp, A.H., Colomer, V., Tokito, M.K., Lanahan, A., Worley, P., Holzbaur, E.L. & Ross, C.A. (1997) Hum. Mol. Genet. 6, 2205– 2212. 63. Ebneth, A., Godemann, R., Stamer, K., Illenberger, S., Trinczek, B. & Mandelkow, E. (1998) J. Cell Biol. 143, 777–794. 64. Tesseur, I., Van Dorpe, J., Bruynseels, K., Bronfman, F., Sciot, R., Van Lommel, A. & Van Leuven, F. (2000) Am.J.Pathol. 157, 1495–1510. 65. DiFiglia, M., Sapp, E., Chase, K.O., Davies, S.W., Bates, G.P., Vonsattel, J.P. & Aronin, N. (1997) Science 277, 1990–1993. 66. Liberski, P.P. & Budka, H. (1999) Acta Neuropathol. 97, 329–334.


ALL KINESIN SUPERFAMILY PROTEIN, KIF, GENES IN MOUSE AND HUMAN

7004

Colloquium All kinesin superfamily protein, KIF, genes in mouse and human

Harukata Miki, Mitsutoshi Setou, Kiyofumi Kaneshiro, and Nobutaka Hirokawa* Department of Cell Biology, Graduate School of Medicine, University of Tokyo, 7–3-1 Hongo, Bunkyo, Tokyo 113–0033, Japan Intracellular transport is essential for morphogenesis and functioning of the cell. The kinesin superfamily proteins (KIFs) have been shown to transport membranous organelles and protein complexes in a microtubule- and ATP-dependent manner. More than 30 KIFs have been reported in mice. However, the nomenclature of KIFs has not been clearly established, resulting in various designations and redundant names for a single KIF. Here, we report the identification and classification of all KIFs in mouse and human genome transcripts. Previously unidentified murine KIFs were found by a PCR-based search. The identification of all KIFs was confirmed by a database search of the total human genome. As a result, there are a total of 45 KIFs. The nomenclature of all KIFs is presented. To understand the function of KIFs in intracellular transport in a single tissue, we focused on the brain. The expression of 38 KIFs was detected in brain tissue by Northern blotting or PCR using cDNA. The brain, mainly composed of highly differentiated and polarized cells such as neurons and glia, requires a highly complex intracellular transport system as indicated by the increased number of KIFs for their sophisticated functions. It is becoming increasingly clear that the cell uses a number of KIFs and tightly controls the direction, destination, and velocity of transportation of various important functional molecules, including mRNA. This report will set the foundation of KIF and intracellular transport research. Intracellular transport is essential for morphogenesis and functioning of the cell. After synthesis, proteins and lipids are sorted and transported to specific destinations within the cell as membranous organelles or protein complexes. The trafficking of proteins is tightly regulated and various different types of proteins are known to be involved. The kinesin super-family proteins (KIFs) have been shown to transport organelles, protein complexes, and mRNAs to specific destinations in a microtubule- and ATP-dependent manner (1–3). KIFs are not only involved in the transport of organelles, protein complexes, and mRNAs, but also participate in chromosomal and spindle movements during mitosis and meiosis (4–6). KIFs contain amino acid sequences that are highly conserved among all eukaryotic phyla studied thus far. Within the motor domain, there are two conserved sequences that are proximal to a Walker A ATP binding motif and a microtubule binding domain (2, 5, 7). Outside the motor domain, KIFs show few similarities. Interactions with cargo molecules have been shown to occur in regions outside the motor domain. Recently, it has been clearly shown that several KIFs attach to specific cargoes through interactions with adaptor proteins in these regions (8, 9). Here, we report the identification of all KIFs in the mouse and human genomes. There are 45 members in total. Additional KIFs were identified by PCR cloning. The total number of KIFs was confirmed by a BLAST search of proteins in public and private genome databases. A unified nomenclature and phylogenic analysis also are presented to help categorize and understand functions of KIFs. This will set the foundation of KIF and intracellular transport research. MATERIALS AND METHODS Identification of Additional KIFs by PCR Cloning. To obtain sequences of mouse KIFs, PCR was conducted by using mouse cDNA and degenerate primers. Upstream primer sequences were derived from a putative ATP-binding motif and downstream primers from a conserved region 5 to the second microtubule binding site (see Table 2, which is published as supplemental data on the PNAS web site, www.pnas.org). mRNA was isolated from 6- or 2-week-old or embryonic ICR mice (Oriental Yeast, Tokyo) tissue by the method of Okayama et al. (10) for reverse transcription (RT). RT was conducted by using the Choice cDNA synthesis system (Life Technologies, Rockville, MD). PCR was conducted for 40 cycles at 96°C for 30 sec, 55°C for 90 sec, and 72°C for 60 sec in a GeneAmp PCR system 9700 Thermal cycler (Perkin-Elmer). PCR products were blunted and subcloned into an EcoRV-digested pBluescript KS(+) vector (Toyobo, Osaka). Sequencing was performed by using the Dyenamic ET primer and Deza sequencing kit (Amersham Pharmacia) and an Applied Biosystems 377 DNA sequencer (Perkin-Elmer). Northern Blotting. Obtained mRNAs also were used for Northern blotting whose results are shown in Fig. 1, KIF2B and KIF18A. Two micrograms of mRNA was run on a 1% formaldehyde agarose gel and transferred to Duralon UV uncharged nylon membranes (Stratagene). Semidried membranes were UV-crosslinked by using a Spectrolinker XL-1000 (Spectronic, Rochester, NY). The Northern blotting sheet used in Fig. 1 for KIF24 was purchased from CLONTECH. The sheet for KIF18B was purchased from Origene Technologies (Rockville, MD). KIF16B and KIF19A were analyzed by using Northern blotting sheets purchased from Ambion (Austin, TX). Random primed 32P-labeled probes were prepared by using the T7 Quick prime kit (Amersham Pharmacia). Hybridization was performed as described (11, 12). Radioactivity was visualized by using the Fuji Biological Analysis System BAS-2000. Membranes were exposed to a BAS-MS imaging plate, and the plate was processed through a BAS-2000. Database Homology Search and Phylogenic Analysis. Full-length and partial amino acid sequences of KIFs obtained as described above were used for a database homology search through GenBank and the Celera Discovery System and Celera Genomics’ associated databases (Celera Genomics, Rockville MD). An

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: KIF, kinesin superfamily protein; KHC, kinesin heavy chain. Data deposition: The sequences reported in this paper have been deposited in the GenBank database [accession nos. AB054029 (KIF2C), AB054024 (KIF18A), AB054025 (KIF18B), AB054026 (KIF19A), AB054030 (KIF20B), AB054027 (KIF23), AB054028 (KIF24), AB054031 (KIF26A), and AB053955 (KIF26B)]. *To whom reprint requests should be addressed. E-mail: [email protected].



7005

exhaustive TBLASTN search was conducted to detect KIF transcripts in the complete human genome. After obtaining all KIFs, the amino acid sequences between IFAY and LAGSE motifs were subjected to phylogenic analysis. In Fig. 2, human and mouse homologs were aligned with CLUSTAL W (13) software by the neighbor-joining method (14). The phylogenic tree was drawn with MACVECTOR software (Oxford Molecular, Cambridge, U.K.). For Fig. 3, maximum parsimony was calculated (15), and the phylogenic tree was drawn by using TREEVIEWPPC (16). Bootstrap values were assessed by 10,000 random samplings. Classification of all KIFs were carried out as described (2). Sequences used in this study can be obtained from our supplemental data or through the Celera database public segment (found on the web at www.celera.com).

Fig. 1. Northern blotting of additional KIFs. KIF2B is expressed ubiquitously in 2-week-old mouse tissue. KIF16B mRNA is detected in testis as a 4.1-kb band and a 3.3-kb band in brain. KIF18A was found in adult brain and embryonic head. KIF18B expression is dominant in testis. KIF19A is detected in testis, lung, and brain. KIF24 bands are seen in testis and spleen lanes. Ts, testis; Si, small intestine; Kd, kidney; Ht, heart; Br, brain; Sp, spleen; Sc, spinal cord; Lv, liver; Lu, lung; Pa, pancreas; Eh, embryo head; E, embryo; Sm, skeletal muscle; St, stomach; Ty, thymus; Ov, ovary. The KIFs presented here were identified by the following criteria: conservation of upstream Walker A ATP-binding motifs and a LAGSE or similar sequence 150–200 aa residues downstream, a YXXXXXDLL motif where X is any amino acid, and a SSRSH motif located between the Walker A and LAGSE sequences. Two predicted transcripts contained only the LAGSE consensus, and two transcripts had only IFAY sequences. In other organisms, a gene encoding only the LAGSE region and another encoding only IFAY were found in Drosophila. The prediction of KIFs using conventional software will automatically predict LAGSE-containing proteins to be KIFs. However, in a previous study it has been implied that the motor domain cannot be separated into modules (17). This indicates that IFAY and LAGSE sequences must be present at an appropriate order and spacing. Therefore, sequences lacking conserved motifs may not function as molecular motors and were excluded from this study. Additionally, genes from the same locus were considered to be splice variants and were omitted. RESULTS AND DISCUSSION KIFs Previously Identified in Our Laboratory. Previously, we have identified 25 KIFs in mice (11, 18–22). Most were found by using molecular biological approaches. This study presents all KIFs in mouse and human and concludes the search for further unidentified KIFs. Identification of 13 Additional KIFs. We report 10 previously unidentified KIFs. KIF18B, KIF19A, KIF23, and KIF24 were identified in adult mouse brain, spinal cord, and small intestine cDNA by PCR. KIF23 has been reported in humans (23), but we have isolated it from mice by using PCR. KIF2B and KIF18A were found in embryonic cDNA by PCR. KIF4A and KIF4B, and KIF19A and KIF19B have a highly homologous motor domain. Therefore, it was difficult to discriminate the differences between them by PCR. These paralogs were found by using KIF4A and KIF19A amino acid sequences, respectively, as a template for BLAST searches. KIF26A and KIF26B also were discovered by BLAST searches using ScSMY1 as a template. SMY1 has amino acid motifs similar to KIF motor domains, although it may not be functional (24). With the exception of KIFs with motor domain sequences similar to that of SMY and therefore having low conservation in amino acid motifs, we were able to identify all KIFs by PCR. KIF2C (25), KIF20B (26), and KIF25 (27) have been reported in humans and not in mice and were found in mice by our database search. KIFs presented in this paper and previously identified KIFs were found by cross-hybridization methods or PCR using degenerate primers (11, 18, 28). These methods are laborious, hazardous, and time consuming. With this report there is no further need to search for new KIFs. However, the actual number of functional KIFs can only be determined after completing endogenous protein purification, peptide sequencing, and motility assays. Tissue Distribution of Additional KIFs. As part of our preliminary results concerning novel KIFs, Northern blotting results of KIF2B, KIF16B, KIF18A, KIF18B, KIF19A, and KIF24 are shown in Fig. 1. KIF2B is ubiquitously expressed in 2-week-old mice at 2.8 kb. KIF16B displays an intense 4-kb band in the testis lane and a 3.3-kb band in the brain lane. KIF18A is expressed in adult lung and embryonic head. The band migrates at 4.6 kb. The 3.9-kb band corresponding to KIF18B is most intense in adult testis. It is also highly expressed in the spleen and thymus and weakly in kidney, liver, lung, skin, small intestine, and stomach. No signal is detected in adult brain, heart, and muscle. KIF19A transcripts are found in adult testis, lung, and brain. A strong doublet band can be seen in the ovary lane. There are also bands in the embryo and spleen lanes. The higher band in ovary corresponds to a protein of 4.6 kb, the lower band to a protein



7006

of 3.8 kb. The lower band is found in testis and lung. The brain, embryo, and spleen lanes display a slightly lower band. KIF24 bands corresponding to proteins of 7.4 kb are found in the testis and spleen lanes. Mouse and Human Similarity. After obtaining all KIFs as described above, we next compared mouse and human KIFs. A phylogenic tree showing the analogy of murine and human KIFs is presented in Fig. 2. Forty-four of 45 mouse KIFs have orthologs in humans. The respective amino acid sequence conservation among murine and human KIFs exceeds 90%, indicating the significance of mice as a model of humans in KIF research. Detection of KIFs in Mouse Brain. There are eight KIFs reported to be expressed specifically or dominantly in mouse brain as demonstrated by Northern blotting, namely, KIF5A (18), KIF1A (18, 29), KIFC2 (21, 30), KIF3C (28), KIF5C (11), KIF21A and KIF21B (31), and KIF17 (8). Nineteen KIFs have been detected in the adult brain as intense bands by Northern blotting. PCR and KIF detection at various developmental stages reveal 11 KIFs expressed in brain. In total, the number of KIFs that have been detected at all stages of murine brain is 38. This number is much larger than the six KIFs reported in the single cell organism of Saccharomyces cerevisiae. This large number could mainly represent the necessity of delivering various functional molecules in highly polarized axons and dendrites for achieving complex functions of neurons. Classification, Phylogenic Analysis, and Nomenclature of KIFs in Mouse and Human. All KIFs shown or predicted to be transcribed in the human and mouse genomes are presented as a phylogenic tree along with all KIFs in S. cerevisiae, Drosophila melanogaster, and Caenorhabditis elegans (Fig. 3). Sequences used are available as supplemental data. Using this occasion, a unified nomenclature is proposed (Table 1), which will abolish redundant designations that confuse researchers inside and outside this field of science. The number of KIFs is in accordance with the total number of genes in comparison with other phyla. The entire human genome is predicted to have 30,000 to 35,000 genes. The number of KIFs found in Drosophila and C. elegans are 23 and 21, respectively (Fig. 3). This is approximately half that of humans. The predicted numbers of genes in these two organisms are 13,601 and 18,424, respectively, likewise roughly half that of humans. S. cerevisiae has 6,241 proteins, of which six are KIFs. The total number of proteins is one-fifth of humans but there are less than a seventh of KIFs. This may be another example of the increased necessity of KIFs for higher cell function. Three major types of KIFs have been identified according to the position of the motor domain: the NH2-terminal motor domain type (32, 33), middle motor domain type (18, 34, 35), and COOH-terminal motor domain type (21, 30, 36, 37) (referred to below as N-kinesins, M-kinesins, and C-kinesins, respectively). This study unexpectedly revealed abundant N-kinesins and few M- and C-kinesins (Fig. 3). Of the 45 KIFs, there are only three M-kinesins and C-kinesins each, leaving 39 N-kinesins. Of the 39 N-kinesins, two are monomeric and 37 seem to be multimeric. There are 14 classes in total. C-kinesins are classified into two classes, C-1 kinesin and C-2 kinesin. M-kinesins make one class. Nkinesins are classified into 11 classes, comprising 16 families. Most classes consist of one family with the exception of N-3, N-4, and N-8 classes. N-3 kinesins consist of Unc104/KIF1, KIF13, and KIF16 families. Members of the Unc104/KIF1 family are mostly monomeric (29, 19), and amino acid sequences imply that those of the KIF16 family are multimeric (M.S., unpublished data). Members of the KIF13 family also have different characteristics (9). Thus, these families form subgroups within N-3 kinesins. N-4 kinesins consist of the KIF3 family and Osm3/ KIF17 family. KIF3s are heterotrimeric, and Osm3/KIF17 form homodimers, indicating that these two families are distinct within this class. N-8 kinesins consist of the KIF18 family and Kid/KIF22 family. Due to the lack of similarity in the functional Kid domains, the KIF18 family was separated (38). This separation is supported by data obtained by using the neighbor-joining method (Fig. 2).

Fig. 2. Phylogenic analysis of mouse and human orthologs. Forty-four of 45 murine KIFs have orthologs in humans. Sequences were analyzed by the neighbor-joining method. Most of the KIFs of other species including plants could be categorized into these 14 classes (data not shown). Below, we present a brief summary of the characterization of each kinesin class. N-1 Kinesins. This class consists of the kinesin heavy chain (KHC) family. KHC, the first KIF reported (39, 40), forms a heterotetramer made of two KHCs and two kinesin light chains (41). KHCs form a highly related family (KIF5A, KIF5B, and KIF5C). KIF5B is expressed ubiquitously in many tissues (42), whereas KIF5A and KIF5C are specific to nerve tissue (18, 22). Kinesin initially was characterized as a motor transporting membranous organelles anterogradely toward the plus end of microtubules and forming a crossbridge between membranous organelle and microtubules in nerve axons (32, 39, 40, 43, 44). However, recent studies have revealed various functions. In a wide variety of cells, kinesin works as a motor for transport of mitochondria, lyso



7007

somes, tubulin oligomer, and mRNA complex toward the plus end of microtubules (3, 45–48). The light chains of KIF5 have a role in binding these cargoes (49, 50). However, in the Ascomycete fungus, Neurospora crassa, kinesin light chains are lacking, implying that KHCs alone are sufficient by themselves for binding to some cargoes (51).

Fig. 3. Phylogenic analysis of all KIFs expressed in mouse/humans, D. melanogaster, C. elegans, and S. cerevisiae. Amino acid sequences were aligned by using maximum parsimony. Sequences used for alignment are available as supplemental data or in the public segment of the Celera database (www.celera.com). N-2 Kinesins. This class consists of the BimC/Eg5/KIF11 family. This family contains KIF11 (Eg5), first found in Xenopus (52), a homotetrameric KIF (53) for bipolar spindle formation (54). Cell cycle-specific phosphorylation also has been reported in human (55). BimC is a well-characterized KIF highly related to Eg5, functioning in cell division (56). N-3 Kinesins. This class is composed of three families, the Unc104/KIF1 family, the KIF13 family, and the KIF14 family. The Unc104/KIF1 family. There are three KIFs in this family: KIF1A, KIF1B, and KIF1C. C. elegans Unc104 is a homolog of mouse KIF1A (57). KIF1A is an anterograde motor transporting a subset of synaptic vesicle precursors and plays an important role in neuronal function and survival (29, 58). KIF1B is thought to convey mitochondria, sharing the role with KIF5B (19, 46). Interestingly, KIF1A and KIF1B are monomeric (19, 29). KIF1C is dimeric in vivo (59) and reported to have functions in endoplasmic reticulum-Golgi transport (60). The KIF13 family. In mice, this family consists of two proteins, KIF13A and KIF13B. KIF13A transports a cargo containing M6PR through direct interaction with the AP-1 complex (9). KIF13B (GAKIN) is reported to interact with hDLG and PSD95 in vitro with its proximal tail, which is highly conserved from C. elegans (61). The KIF16 family. KIF14, KIF16A, and KIF16B constitute a family with DmKlp98a. As the tail domains along with the expression patterns of KIF14, KIF16A, and KIF16B are different, these KIFs may have separate functions. N-4 Kinesins. This class consists of KIF3 and Osm3/KIF17 families. The KIF3 family. The KIF3 family is composed of KIF3A, KIF3B, and KIF3C in mice (18, 20, 62). A KIF3A-KIF3B heterodimer (KIF3A/3B) assembles with KAP3, forming a heterotrimeric complex (63, 64). The motor is expressed ubiquitously and is used for anterograde transport of membranous organelles containing fodrin in neurons (65). The KIF3 complex and its homolog were shown to transport protein complexes to form cilia (66–69). Gene-targeting studies showed that the nodal cilia, in which KIF3A/B are localized, rotate to generate a unidirectional flow of extraembryonic fluid (nodal flow), which could fundamentally control left-right determination (70, 71). Without KIF3, there is no nodal flow (70, 71). Thus, KIF3 is essential for development of the left-right axis determination in embryos (70– 72). The Osm3/KIF17 family. Another molecule closely related to the KIF3 family is Osm3 in C. elegans (73). Mutations in Osm3



7008

cause defects in chemosensory responses (74). Osm3 is necessary for sensory cilia growth in the dendrites of sensory neurons. Mutations in Osm-3 and Lin-10 result in a similar phenotype in osmotic avoidance, and Lin-10 has defect in glutamate receptor localization (75). Interestingly, KIF17, a member of this family, binds to mouse homolog of Lin-10 and transports vesicles containing a N-methyl-D-aspartate receptor subunit, NR2B, through the following interactions: KIF17-mLin10-mLin2mLin7-NR2B (8). Human KIF17 also is reported to be highly expressed only in central nervous system, possessing a highly conserved Lin-10 binding domain (Kazusa DNA bank). Table 1. Proposed and previous nomenclature of all KIFs Proposed H. sapiens D. melanogaster KIF 1A ATSV KIF 1B CG8566 KIF 1C KIF 2A KIF 2 CG1453 KIF 2B CG3219 KIF 2C CAKin/KNSL6 KIF 3A KIF 3B KIF 3C KIF 4A KIF 4B KIF 5A KIF 5B KIF 5C KIF 6 KIF 7 KIF 8 KIF 11 KIF 9 KIF 10 KIF 12 KIF 13A KIF 13B KIF 14 KIF 15 KIF 16A KIF 16B KIF 17 KIF 18A KIF 18B KIF 19A KIF 19B KIF 20A KIF 20B KIF 21A KIF 21B KIF 22 KIF 23 KIF 24 KIF 25 KIF 26A KIF 26B KIF C1 KIF C2 KIF C3

KIF4 nKHC uKHC xKHC

C. elegans unc104

GAKIN HUMORFW Hklp2

Rab 6 Kin KlpMPP1 Kid/KNSL4 MKLP1/ KNSL5 KNSL3 HSET/KNSL2

Others

(–) (–) K11D9.1

(–)

RnKIF1D XIXKIF2 CgMCAK XIKCM1 SpKRP85 SpKRP95

Klp 64D Klp68D CG 17461 Klp3A

KRP85 KRP95

(–)

Y43F4B.6

(–)

GgChrkin

KHC

Unc116

(–)

RnnKHC

(–) (–) Klp61F

(–) (–) BimC

(–) Cmet CP15516 CG 15844 Kin 73

(–) (–)

(–) (–) Kip1 Cin 8 (–) Kip 2

(–) Klp 4

(–) (–) (–) (–)

Klp98A

Klp 6 C06G3.2 C33H5.4 (–)

(–) Klp67A

Osm3 (–)

(–) Kip3

CG9913

(–)

CG 12298

(–)

(–)

CG5300

T01G1.1

(–)

Nod Pav CP38609 (–) (–)

(–) Zen4A,B (–) (–) Vab8

(–) (–) (–) (–) (–)

Ncd

C41G7.2 M01E11.6 W02B12.7 Klp3

Kar3

CgCHO2

(–)

XICTK1

Eg5/KNSL1 CENP-E

S. cerevisiae

Neb (–)

(–)

(–)

AnBimC CrKlp1

Reference 18, GB 19, 57 102 18, 103, 104 0, GB, WB 35 25, 105 18, 28, 106, 107 11, 23, GB 18, 80 0, 78, WB 18, 22, 108 109, 110 11 11 11 11, 52, 54 56, 88, 1 1 1 11, 98 11, 55 86, 88, GB

11 9, 11, 112, 113 11, 60 11, 114, *, GB 11, 91, WB 11, 112 11 8, 11, 73 0, 112 0, 115, GB MmKlp174

CgCHO1

83 26, GB 31, GB, WB 38, 116 23, 81, 85, 117 0, GB 27 0, 94 36, 37 118, 119 120, WB 21, 30 28, 112

GB, GenBank direct submission; WB, Worm Base; O, this paper, *Siddiqui, S.S., Hori, H., Mohammed, A.S. &Ali, M.Y., International C. elegans meeting, June 2–6, 1999, Madison, WI.

N-5 Kinesins. This class is composed of the KIF4 family. A member of this family is KIF4A. KIF4A mRNA is expressed abundantly in juvenile tissues, including differentiated young neurons (76). KIF4A is a microtubule plus end-directed anterograde motor. Evidence has been shown of KIF4A transporting membranous organelles containing L1 in juvenile neurons (77). In this study, KIF4B, a transcript highly homologous to KIF4A was identified. Both genes have been assigned to respective loci (78). Chromokinesin, the chicken isolog of KIF4, is associated with chromosome arms and functions as a mitotic motor with DNA-as its cargo (79). The other members of this family,



7009

KIF21A and KIF21B, are assumed to have some role in neurons (31). DmKlp3a is a critical component in the establishment or stabilization of the central spindle (80). Thus, members of this family may have multiple functions, including membrane trafficking and cell division. N-6 Kinesins. This class consists of the CHO1/KIF23 and KIF20/ Rab6 kinesin families. The CHO1/KIF23 family. KIF23 (CHO1) is a KIF involved in mitosis, originally identified by a mAb raised against mitotic spindle components (81). It has been shown to be expressed in cultured neurons (82). The KIF20/Rab6 kinesin family. KIF20A (Rab6-KIF) is reported to have a fundamental role in Golgi-derived vesicle transport (83) and/or cell division (84). KIF20B (KlpMPP1) was isolated by using an antibody specific for M-phase phosphorylated proteins (26). The Drosophila homolog of KIF23, Pavarotti (Pav), is required for central spindle pole organization and cytokinesis (85). N-7 Kinesins. The CENP-E/KIF10 family forms N-7 kinesins. KIF10 (CENP-E) was first identified as a centromereassociated protein (86). It functions in chromosome segregation (87). Cenp-meta has a similar function in Drosophila (55). S. cerevisiae Kip2p, homolog of KIF10, functions in spindle positioning (88, 89). N-8 Kinesins. The KIF18 and Kid/KIF22 families constitute N-8 kinesins. The KIF18 family. This most recently discovered family of KIFs has not yet been well characterized. The tissue expression patterns of KIF18A and KIF18B are shown in Fig. 1. This family has a counterpart in Drosophila, but not in C. elegans. The Kid/KIF22 family. This family contains KIF22 (Kid), a KIF that colocalizes with mitotic chromosomes and may bind DNA (38). A highly homologous KIF, Nod is found in Drosophila (90). We report additional members of this family, KIF19A and KIF19B. KIF19A and KIF19B are highly homologous to each other, but have two loci (see Table 3, which is published as supplemental data, for accession numbers). We also have detected a splice variant of KIF19A. In the Northern blotting for KIF19A (Fig. 1), we detected bands at three different heights, in good agreement with the existence of three transcripts. However, only one loci is found in the Celera Discovery System. Further studies are necessary to clarify this inconsistency. N-9 Kinesins. The KIF12 family form this class. KIF12 has been an orphan KIF, not affiliated with any family previously reported (2). Here, it composes a family with its Dm counterpart, DmCG15844. KIF12 is highly expressed in kidney and may have a significant role in kidney cells (11). N-10 Kinesins. The KIF15 family forms this class. KIF15 was reported to be dominantly expressed in spleen and testis (11). Human KIF15 (Hklp2) is reported to associate with a cell proliferation marker protein (91). The two C. elegans proteins in this family, C06G3.2 and C33H5.4, seem to be paralogs (Worm Base). N-11 Kinesins. N-11 kinesins contain the KIF26 family. This family is related to DmCos2 and CeVab-8. Some members of this family have relatively low consensus motif conservation in comparison to other KIFs (see supplemental data). Localization of ScSMY1 is not affected by microtubule abolition (24). Costal2 (Cos2) has been shown to be part of the hedgehog signaling cascade with microtubule binding abilities (92, 93). Vab-8 is implicated in the regulation of cell and axon growth cone migration (94). KIF25 (KNSL3) has four splice variants and is expressed ubiquitously (27). Three additional KIFs, KIF24, KIF26A, and KIF26B, join this family. Their functions are yet to be determined. M-Kinesins. The KIF2 family forms M-kinesins. M-kinesins have a motor domain in the center of the molecule (18). KIF2A (formerly KIF2) forms a homodimer. KIF2A is a microtubule plus end-directed motor and is expressed ubiquitously (34). The cargo of KIF2A includes ßgc, a ß-subunit of the insulin-like growth factor-1 receptor (95). A splice variant of KIF2A has been reported (96). A Xenopus homolog of KIF2A, XKIF2, is reported to destabilize microtubules in vitro (104). KIF2C (MCAK) was identified as a mitotic-centromereassociated kinesin (35). KIF2B is a novel member of this family. C-1 Kinesins. This class contains the Ncd/Kar3/KIFC1 family. Several C-type motors, such as Ncd in Drosophila and Kar3 in S. cerevisiae, are motors for meiosis, mitosis, and karyogamy. These family members exhibit a microtubule minus end-directed motility (36, 37). Mammals have a counterpart, KIFC1 (21). It is noteworthy that C. elegans has developed many KIFC1 homologs. A highly related KIF also has been reported (97). C-2 Kinesins. The KIFC2/C3 family constitutes C-2 kinesins. Three C-type KIFs have been identified in mouse brain. KIFC2 forms a homodimer without associated polypeptides (21). It is a unique C-type motor that mainly functions in the dendritic transport of multivesicular body-like membranous organelles (21). KIFC3 is ubiquitously expressed (11). A large number of C-kinesins were expected to function in the complex dendritic transport. Recently, it has been revealed that Nkinesins, plus end-directed motors, play important roles not only in axons but also in dendrites (8, 31). Indeed, the distal portions of developed dendrites have microtubule plus ends directed to the tips, having the same polarity as axons. The “mixed polarity” of microtubules exists only in proximal portions of dendrites. Future work should reveal the roles of KIFs in dendrites, which requires an accurate delivery regulation mechanism in comparison to axonal transport. Additionally, if the same motor for axonal transport is used for dendritic transport, how cargoes orient KIFs toward specific destinations will be a key question that needs to be answered in future studies. Orphans. These KIFs have no counterpart in Drosophila or C. elegans. KIF6 and KIF9 are localized near the BimC family. The fulllength sequences of KIF6 and KIF9 are not similar with BimC. The phylogenic distance between KIF9 and KIF6 is large, implying that they do not form a family. These molecules may have evolved to attain other functions in mammals. KIF9 has a homolog in Chlamydomonas, Klp1 (98). KIF7 has no evident homolog in Drosophila, C. elegans, or S. cerevisiae. Originally identified from murine brain by using PCR, it was found to be dominantly expressed in the testis by Northern blotting (11). There have been reports of many proteins, RNA, and membranous organelles transported in a microtubule-dependent manner (99). Forty-five KIFs are insufficient to transport all organelles and vesicles. There must be a way to transport various cargoes using a limited number of KIFs. Other KIF members may arise through improved transcription algorithms. Alternative splicing also may increase the number of KIFs contributing to intracellular transport, and adaptor proteins also may contribute. The search for adaptor proteins other than kinesin light chains and kinectin (100) has begun (8, 9). The trend has extended to myosin research (101). To elucidate the mechanism of intracellular transport, the regulation of cargo binding is also an important problem. Currently, there are only a few reports clarifying this essential topic (102). Another question is how cargo and KIF dissociate. To enable cargo proteins to function properly, this dissociation is indispensable. Concerning how each KIF recognizes and binds to their specific cargo molecules, one significant method is the formation of a receptor-adaptor (scaffold/scaffolding protein)-motor complex as in the case of KIF3, KIF17, and KIF13A (8, 9, 63). Alternatively, KIFs can bind to membrane proteins through light chains (49, 50). These adaptor proteins may contribute in increasing the variety of cargoes a KIF can convey. Thus, it is rapidly becoming clear that the cell uses a number of KIFs and tightly controls the direction, destination, and velocity of transports for various important functional molecules.



7010

We are grateful to members of the Hirokawa laboratory, especially Dr. T.Nakagawa. This work was supported by a grant for Center of Excellence (COE) from the Ministry of Education, Science, and Culture (to N.H). 1. Hirokawa, N. (1996) Trends Cell Biol. 6, 135–141. 2. Hirokawa, N. (1998) Science 279, 519–526. 3. Brendza, R.P., Serbus, L.R., Duffy, J.B. & Saxton, W.M. (2000) Science 289, 2120–2122. 4. Hirokawa, N., Noda, Y. & Okada, Y. (1998) Curr. Opin. Cell Biol. 10, 60–73. 5. Vale, R.D. & Fletterick, R.J. (1997) Annu. Rev. Cell Dev. Biol. 13, 745–777. 6. Sharp, D.J., Rogers, G.C. & Scholey, J.M. (2000) Nature (London) 407, 41–47. 7. Kim, A.J. & Endow, S.A. (2000) J. Cell Sci. 113, 3681–3682. 8. Setou, M., Nakagawa, T., Seog, D.H. & Hirokawa, N. (2000) Science 288, 1796–1802. 9. Nakagawa, T., Setou, M., Seog, D., Ogasawara, K., Dohmae, N., Takio, K. & Hirokawa, N. (2000) Cell 103, 569–581. 10. Okayama, H., Kawaichi, M., Brownstein, M., Lee, F., Yokota, T. & Arai, K. (1987) Methods Enzymol. 154, 3–28. 11. Nakagawa, T., Tanaka, Y., Matsuoka, E., Kondo, S., Okada, Y., Noda, Y., Kanai, Y. & Hirokawa, N. (1997) Proc. Natl. Acad. Sci. USA 94, 9654– 9659. 12. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press, Plainview, NY), 2nd Ed. 13. Thompson, J.D., Higgins, D.G. & Gibson, T.J. (1994) Nucleic Acids Res. 22, 4673–4680. 14. Saitou, N. & Nei, M. (1987) Mol. Biol. Evol. 4, 406–425. 15. Tanaka, Y., Zhang, Z. & Hirokawa, N. (1995) J. Cell Sci. 108, 1883–1893. 16. Page, R.D. (1996) Comp. Appl. Biosci. 12, 357–358. 17. Kikkawa, M., Okada, Y. & Hirokawa, N. (2000) Cell 100, 241–252. 18. Aizawa, H., Sekine, Y., Takemura, R., Zhang, Z., Nangaku, M. & Hirokawa, N. (1992) J. Cell Biol. 119, 1287–1296. 19. Nangaku, M., Sato-Yoshitake, R., Okada, Y., Noda, Y., Takemura, R., Yamazaki, H. & Hirokawa, N. (1994) Cell 79, 1209–1220. 20. Yamazaki, H., Nakata, T., Okada, Y. & Hirokawa, N. (1995) J. Cell Biol. 130, 1387–1399. 21. Saito, N., Okada, Y., Noda, Y., Kinoshita, Y., Kondo, S. & Hirokawa, N. (1997) Neuron 18, 425–438. 22. Kanai, Y., Okada, Y., Tanaka, Y., Harada, A., Terada, S. & Hirokawa, N. (2000) J. Neurosci. 20, 6374–6384. 23. Kuriyama, R., Dragas-Granoic, S., Maekawa, T., Vassilev, A., Khodjakov, A. & Kobayashi, H. (1994) J. Cell Sci. 107, 3485–3499. 24. Lillie, S.H. & Brown, S.S. (1998) J. Cell Biol. 140, 873–883. 25. Kim, I.G., Jun, D.Y., Sohn, U. & Kim, Y.H. (1997) Biochim. Biophys. Acta 1359, 181–186. 26. Westendorf, J.M., Rao, P.N. & Gerace, L. (1994) Proc. Natl. Acad. Sci. USA 91, 714–718. 27. Okamoto, S., Matsushima, M. & Nakamura, Y. (1998) Cytogenet. Cell Genet. 83, 25–29. 28. Yang, Z., Hanlon, D.W., Marszalek, J.R. & Goldstein, L.S. (1997) Genomics 45, 123–131. 29. Okada, Y., Yamazaki, H., Sekine-Aizawa, Y. & Hirokawa, N. (1995) Cell 81, 769–780. 30. Hanlon, D.W., Yang, Z. & Goldstein, L.S. (1997) Neuron 18, 439–451. 31. Marszalek, J.R., Weiner, J.A., Farlow, S.J., Chun, J. & Goldstein, L.S. (1999) J. Cell Biol. 145, 469–479. 32. Hirokawa, N., Pfister, K.K., Yorifuji, H., Wagner, M.C., Brady, S.T. & Bloom, G.S. (1989) Cell 56, 867–878. 33. Yang, J.T., Laymon, R.A. & Goldstein, L.S. (1989) Cell 56, 879–889. 34. Noda, Y., Sato-Yoshitake, R., Kondo, S., Nangaku, M. & Hirokawa, N. (1995) J. Cell Biol. 129, 157–167. 35. Wordeman, L. & Mitchison, T.J. (1995) J. Cell Biol. 128, 95–104. 36. Meluh, P.B. & Rose, M.D. (1990) Cell 60, 1029–1041. 37. Endow, S.A., Henikoff, S. & Soler-Niedziela, L. (1990) Nature (London) 345, 81–83. 38. Tokai, N., Fujimoto-Nishiyama, A., Toyoshima, Y., Yonemura, S., Tsukita, S., Inoue, J. & Yamamota, T. (1996) EMBO J. 15, 457–467. 39. Vale, R.D., Reese, T.S. & Sheetz, M.P. (1985) Cell 42, 39–50. 40. Brady, S.T. (1985) Nature (London) 317, 73–75. 41. Bloom, G.S., Wagner, M.C., Pfister, K.K. & Brady, S.T. (1988) Biochemistry 27, 3409–3416. 42. Gudkov, A.V., Kazarov, A.R., Thimmapaya, R., Axenovich, S.A., Mazo, I.A. & Roninson, I.B. (1994) Proc. Natl. Acad. Sci. USA 91, 3744–3748. 43. Hirokawa, N., Sato-Yoshitake, R., Kobayashi, N., Pfister, K.K., Bloom, G.S. & Brady, S.T. (1991) J. Cell Biol. 114, 295–302. 44. Hirokawa, N. (1982) J. Cell Biol. 94, 129–142. 45. Hollenbeck, P.J. & Swanson, J.A. (1990) Nature (London) 346, 864–866. 46. Tanaka, Y., Kanai, Y., Okada, Y., Nonaka, S., Takeda, S., Harada, A. & Hirokawa, N. (1998) Cell 93, 1147–1158. 47. Nakata, T. & Hirokawa, N. (1995) J. Cell Biol. 131, 1039–1053. 48. Terada, S., Kinjo, M. & Hirokawa, N. (2000) Cell 103, 141–155. 49. Kamal, A., Stokin, G.B, Yang, Z., Xia, C.H. & Goldstein, L.S. (2000) Neuron 28, 449–459. 50. Bowman, A.B., Kamal, A., Ritchings, B.W., Philp, A.V., McGrail, M., Gindhart, J.G. & Goldstein, L.S. (2000) Cell 103, 583–594. 51. Steinberg, G. & Schliwa, M. (1995) Mol. Biol. Cell 6, 1605–1618. 52. Le Guellec, R., Paris, J., Couturier, A., Roghi, C. & Philippe, M. (1991) Mol. Cell. Biol. 11, 3395–3398. 53. Cole, D.G., Saxton, W.M., Sheehan, K.B. & Scholey, J.M. (1994) J. Biol. Chem. 269, 22913–22916. 54. Blangy, A., Lane, H.A., d’Herin, P., Harper, M., Kress, M. & Nigg, E.A. (1995) Cell 83, 1159–1169. 55. Yucel, J.K., Marszalek, J.D., McIntosh, J.R., Goldstein, L.S., Cleveland, D.W. & Philp, A.V. (2000) J. Cell Biol. 150, 1–11. 56. Enos, A.P. & Morris, N.R. (1990) Cell 60, 1019–1027. 57. Otsuka, A.J., Jeyaprakash, A., Garcia-Anoveros, J., Tang, L.Z., Fisk, G., Hartshorne, T., Franco, R. & Born, T. (1991) Neuron 6, 113–122. 58. Yonekawa, Y., Harada, A., Okada, Y., Funakoshi, T., Kanai, Y., Takei, Y., Terada, S., Noda, T. & Hirokawa, N. (1998) J. Cell Biol. 141, 431–441. 59. Dorner, C., Ullrich, A., Haring, H.U. & Lammers, R. (1999) J. Biol. Chem. 274, 33654–33660. 60. Dorner, C., Ciossek, T., Muller, S., Moller, P.H., Ullrich, A. & Lammers, R. (1998) J. Biol. Chem. 273, 20267–20275. 61. Hanada, T., Lin, L., Tibaldi, E.V., Reinherz, E.L. & Chishti, A.H. (2000) J. Biol. Chem. 275, 28774–28784. 62. Kondo, S., Sato-Yoshitake, R., Noda, Y., Aizawa, H., Nakata, T., Matsuura, Y. & Hirokawa, N. (1994) J. Cell Biol. 125, 1095–1107. 63. Yamazaki, H., Nakata, T., Okada, Y. & Hirokawa, N. (1996) Proc. Natl. Acad. Sci. USA 93, 8443–8448. 64. Wedaman, K.P., Meyer, D.W., Rashid, D.J., Cole, D.G. & Scholey, J.M. (1996) J. Cell Biol. 132, 371–380. 65. Takeda, S., Yamazaki, H., Seog, D.H., Kanai, Y., Terada, S. & Hirokawa, N. (2000) J. Cell Biol. 148, 1255–1265. 66. Cole, D.G., Cande, W.Z., Baskin, R.J., Skoufias, D.A., Hogan, C.J. & Scholey, J.M. (1992) J. Cell Sci. 101, 291–301. 67. Cole, D.G., Chinn, S.W., Wedaman, K.P., Hall, K., Vuong, T. & Scholey, J.M. (1993) Nature (London) 366, 268–270. 68. Morris, R.L. & Scholey, J.M. (1997) J. Cell Biol. 138, 1009–1022. 69. Cole, D.G., Diener, D.R., Himelblau, A.L., Beech, P.L., Fuster, J.C. & Rosenbaum, J.L. (1998) J. Cell Biol. 141, 993–1008. 70. Nonaka, S., Tanaka, Y., Okada, Y., Takeda, S., Harada, A., Kanai, Y., Kido, M. & Hirokawa, N. (1998) Cell 95, 829–837. 71. Takeda, S., Yonekawa, Y., Tanaka, Y., Okada, Y., Nonaka, S. & Hirokawa, N. (1999) J. Cell Biol. 145, 825–836. 72. Marszalek, J.R., Ruiz-Lozano, P., Roberts, E., Chien, K.R. & Goldstein, L.S. (1999) Proc. Natl. Acad. Sci. USA 96, 5043–5048. 73. Shakir, M.A., Fukushige, T., Yasuda, H., Miwa, J. & Siddiqui, S.S. (1993) NeuroReport 4, 891–894. 74. Tabish, M., Siddiqui, Z.K., Nishikawa, K. & Siddiqui, S.S. (1995) J. Mol. Biol. 247, 377–389. 75. Rongo, C., Whitfield, C.W., Rodal, A., Kim, S.K. & Kaplan, J.M. (1998) Cell 94, 751–759. 76. Sekine, Y., Okada, Y., Noda, Y., Kondo, S., Aizawa, H., Takemura, R. & Hirokawa, N. (1994) J. Cell Biol. 127, 187–201. 77. Peretti, D., Peris, L., Rosso, S., Quiroga, S. & Caceres, A. (2000) J. Cell Biol. 149, 141–152. 78. Ha, M.J., Yoon, J., Moon, E., Lee, Y.M., Kim, H.J. & Kim, W. (2000) Cytogenet. Cell Genet. 88, 41–42.



7011

79. Wang, S.Z. & Adler, R. (1995) J. Cell Biol. 128, 761–768. 80. Williams, B.C., Riedy, M.F., Williams, E.V., Gatti, M. & Goldberg, M.L. (1995) J. Cell Biol. 129, 709–723. 81. Nislow, C., Sellitto, C., Kuriyama, R. & McIntosh, J.R. (1990) J. Cell Biol. 111, 511–522. 82. Ferhat, L., Kuriyama, R., Lyons, G.E., Micales, B. & Baas, P.W. (1998) Eur. J. Neurosci. 10, 1383–1393. 83. Echard, A., Jollivet, F., Martinez, O., Lacapere, J.J., Rousselet, A., Janoueix-Lerosey, I. & Goud, B. (1998) Science 279, 580–585. 84. Hill, E., Clarke, M. & Barr, F.A. (2000) EMBO J. 19, 5711–5719. 85. Adams, R.R., Tavares, A.A., Salzberg, A., Bellen, H.J. & Glover, D.M. (1998) Gene. Dev. 12, 1483–1494. 86. Yen, T.J., Compton, D.A., Wise, D., Zinkowski, R.P., Brinkley, B.R., Earnshaw, W.C. & Cleveland, D.W. (1991) EMBO J. 10, 1245–1254. 87. Yen, T.J., Li, G., Schaar, B.T., Szilak, I. & Cleveland, D.W. (1992) Nature (London) 359, 536–539. 88. Roof, D.M., Meluh, P.B. & Rose, M.D. (1992) J. Cell Biol. 118, 95–108. 89. Cottingham, F.R. & Hoyt, M.A. (1997) J. Cell Biol. 138, 1041–1053. 90. Zhang, P., Knowles, B.A, Goldstein, L.S. & Hawley, R.S. (1990) Cell 62, 1053–1062. 91. Sueishi, M., Takagi, M. & Yoneda, Y. (2000) J. Biol. Chem. 275, 28888–28892. 92. Sisson, J.C, Ho, K.S., Suyama, K. & Scott, M.P. (1997) Cell 90, 235–245. 93. Robbins, D.J., Nybakken, K.E., Kobayashi, R., Sisson, J.C., Bishop, J.M. & Therond, P.P. (1997) Cell 90, 225–234. 94. Wolf, F.W., Hung, M.S., Wightman, B., Way, J. & Garriga, G. (1998) Neuron 20, 655–666. 95. Morfini, G., Quiroga, S., Rosa, A., Kosik, K. & Caceres, A. (1997) J. Cell Biol. 138, 657–669. 96. Santama, N., Krijnse-Locker, J., Griffiths, G., Noda, Y., Hirokawa, N. & Dotti, C.G. (1998) EMBO J. 17, 5855–5867. 97. Navolanic, P.M. & Sperry, A.O. (2000) Biol. Reprod. 62, 1360–1369. 98. Bernstein, M., Beech, P.L., Katz, S.G. & Rosenbaum, J.L. (1994) J. Cell Biol. 125, 1313–1326. 99. Terada, S. & Hirokawa, N. (2000) Curr. Opin. Neurobiol. 10, 566–573. 100. Toyoshima, I., Yu, H., Steuer, E.R. & Sheetz, M.P. (1992) J. Cell Biol. 118, 1121–1131. 101. Kussel-Andermann, P., El-Amraoui, A., Safieddine, S., Nouaille, S., Perfettini, I., Lecuit, M., Cossart, P., Wolfrum, U. & Petit, C. (2000) EMBO J. 19, 6020–6029. 102. Sato-Yoshitake, R., Yorifuji, H., Inagaki, M. & Hirokawa, N. (1992) J. Biol. Chem. 267, 23930–23936. 103. Rogers, S.L., Tint, I.S., Fanapour, P.C. & Gelfand, V.I. (1997) Proc. Natl. Acad. Sci. USA 94, 3720–3725. 104. Desai, A., Verma, S., Mitchison, T.J. & Walczak, C.E. (1999) Cell 96, 69–78. 105. Walczak, C.E, Mitchison, T.J. & Desai, A. (1996) Cell 84, 37–47. 106. Rashid, D.J., Wedaman, K.P. & Scholey, J.M. (1995) J. Mol. Biol. 252, 157–162. 107. Cole, D.G., Diener, D.R., Himelblau, A.L., Beech, P.L., Fuster, J.C. & Rosenbaum, J.L. (1998) J. Cell Biol. 141, 993–1008. 108. Patel, N., Thierry-Mieg, D. & Mancillas, J.R. (1993) Proc. Natl. Acad. Sci. USA 90, 9181–9185. 109. Niclas, J., Navone, F., Hom-Booher, N. & Vale, R.D. (1994) Neuron 12, 1059–1072. 110. Saxton, W.M., Hicks, J., Goldstein, L.S. & Raff, E.C. (1991) Cell 64, 1093–1102. 111. Hoyt, M.A., He, L., Loo, K.K. & Saunders, W.S. (1992) J. Cell Biol. 118, 109–120. 112. Stewart, R.J., Pesavento, P.A., Woerpel, D.N. & Goldstein, L.S. (1991) Proc. Natl. Acad. Sci. USA 88, 8470–8474. 113. Li, H.P., Liu, Z.M. & Nirenberg, M. (1997) Proc. Natl. Acad. Sci. USA 94, 1086–1091. 114. Molina, I., Baars, S., Brill, J.A., Hales, K.G., Fuller, M.T. & Ripoll, P. (1997) J. Cell Biol. 139, 1361–1371. 115. DeZwaan, T.M., Ellingson, E., Pellman, D. & Roof, D.M. (1997) J. Cell Biol. 138, 1023–1040. 116. Funabiki, H. & Murray, A.W. (2000) Cell 102, 411–424. 117. Raich, W.B., Moran, A.N., Rothman, J.H. & Hardin, J. (1998) Mol. Biol. Cell 9, 2037–2049. 118. Kuriyama, R., Kofron, M., Essner, R., Kato, T., Dragas-Granoic, S., Omoto, C.K. & Khodjakov, A. (1995) J. Cell Biol. 129, 1049–1059. 119. Ando, A., Kikuti, Y.Y., Kawata, H., Okamoto, N., Imai, T., Eki, T., Yokoyama, K., Soeda, E., Ikemura, T., Abe, K., et al. (1994) Immunogenetics 39, 194–200. 120. McDonald, H.B. & Goldstein, L.S. (1990) Cell 61, 991–1000.


ASSEMBLY AND TRANSPORT OF A PREMESSENGER RNP PARTICLE

7012

Colloquium Assembly and transport of a premessenger RNP particle

Bertil Daneholt* Department of Cell and Molecular Biology, Medical Nobel Institute, Karolinska Institutet, Box 285, SE-17177 Stockholm, Sweden Salivary gland cells in the larvae of the dipteran Chironomus tentans offer unique possibilities to visualize the assembly and nucleocytoplasmic transport of a specific transcription product. Each nucleus harbors four giant polytene chromosomes, whose transcription sites are expanded, or puffed. On chromosome IV, there are two puffs of exceptional size, Balbiani ring (BR) 1 and BR 2. A BR gene is 35–40 kb, contains four short introns, and encodes a 1-MDa salivary polypeptide. The BR transcript is packed with proteins into a ribonucleoprotein (RNP) fibril that is folded into a compact ring-like structure. The completed RNP particle is released into the nucleoplasm and transported to the nuclear pore, where the RNP fibril is gradually unfolded and passes through the pore. On the cytoplasmic side, the exiting extended RNP fibril becomes engaged in protein synthesis and the ensuing polysome is anchored to the endoplasmic reticulum. Several of the BR particle proteins have been characterized, and their fate during the assembly and transport of the BR particle has been elucidated. The proteins studied are all added cotranscriptionally to the premRNA molecule. The various proteins behave differently during RNA transport, and the flow pattern of each protein is related to the particular function of the protein. Because the cotranscriptional assembly of the pre-mRNP particle involves proteins functioning in the nucleus as well as proteins functioning in the cytoplasm, it is concluded that the fate of the mRNA molecule is determined to a considerable extent already at the gene level. The organization of chromatin in a diploid cell nucleus is complex and dynamic. The chromosomes form chromosomal territories, each consisting of several more-or-less condensed and variable domains (1, 2). The individual territories are separated by a delicate network of thin channels, the interchromosomal space (2–4). The active genes are usually situated in the periphery of the domains and deliver the transcription products into the channel system (5, 6). The products move toward the periphery of the nucleus and leave the nucleus through the nuclear pores in the nuclear envelope (7, 8). In the ordinary diploid nucleus, it has proven difficult to follow the flow of specific transcription products from the gene to the nuclear pores. At the light microscopy level, specific genes and their growing transcripts can be located by in situ hybridization (e.g., ref. 9), but the completed and released transcripts are usually too scarce in the nucleoplasm to be detected and traced in the channel system. In the electron microscope, it is difficult to identify specific active genes as well as the corresponding transcription products in transit from the gene to the periphery of the nucleus. However, in the polytene nuclei of dipteran insects, it is feasible, in exceptional cases, to visualize both the transcription process and the transport of the transcription product from the gene to the nuclear pores. The most extensively studied system in this respect is the Balbiani rings (BRs) on the polytene chromosomes in the larval salivary glands of the midge Chironomus tentans (8). Polytene chromosomes consist of thousands of identical chromatids perfectly arranged side by side into well-defined cable-like structures (for review, see ref. 10). The transversely banded chromosomes allow specific chromosomal regions to be identified and synthetic events along the chromosomes to be studied. The transcriptionally active regions are blown-up, or puffed. In the salivary glands of C. tentans, there are three exceptionally large puffs, designated BR1, BR2, and BR3, which are all located on the short chromosome IV (Fig. 1). In the two largest BRs, BR1 and BR2, the transcriptionally active genes are 35–40 kb in size and contain four introns, three close to the 5 end of the gene and one close to the 3 end (11, 12). The introns are very short, and the BR1 and BR2 transcripts are, therefore, only minimally reduced in size during processing. The transcripts encode giant salivary polypeptides (about 1 MDa) that are secreted and form a proteinaceous tube in which the larva lives (13). As the BR transcripts are made large, remain large, and are abundant both on the gene and in the nucleoplasm, the BR transcription products are optimal for visualization of the assembly and transport of these transcription products; in fact, it has been possible to follow the formation of the product during transcription as well as the transport to and through the nuclear pores and finally the exit of the transcript and the formation of polysomes on the cytoplasmic side of the pore (8). VISUALIZATION OF ASSEMBLY AND TRANSPORT OF BR PARTICLES The active BR genes have been studied both when spread on the surface of an electron microscopic grid (14) and when present within the cell (8, 15). The genes are heavily loaded with RNA polymerases and resemble in the electron microscope the wellknown “Christmastree”-like ribosomal genes (16). The transcripts increase in size along the gene, and proteins associate with the growing RNAs to form thin ribonucleoprotein (RNP) fibrils. In spread preparations, the RNP fibrils are more or less extended because of the low salt conditions used. In situ, however, the packing of the RNP fibril into higher-order structure can be followed. At low resolution, an RNP fiber is first recognized, which is later on packed into a globular structure (Fig. 2D). At higher resolution, it can be seen how the thin RNP fibril is initially loosely coiled (forming the RNP fiber) and is subsequently tightly folded into a short ribbon, which is bent into a partial ring (the globule) (Fig. 3). When the particle is released from the gene, the RNP fiber is retracted into the globular portion, and the particle attains an almost ring-like conformation. The particle moves randomly in the interchromosomal space (17), although it can transiently bind to a fibrous network (18). When the particle gets to the nuclear pore complex and passes through the pore, the bent ribbon becomes straightened out, the RNP fibril unfolds and emerges extended on the

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: BR, Balbiani ring; RNP, ribonucleoprotein; snRNP, small nuclear RNP; hnRNP, heterogeneous nuclear RNP; CBP, capbinding protein; RBD, RNA-binding domain. *E-mail: [email protected].



7013

cytoplasmic side, and protein synthesis is initiated (Fig. 3) (19). The translocation process has been studied in detail, and several discrete steps have been elucidated: binding of the BR RNP particle to the nucleoplasmic fibers of the nuclear pore complex, docking of the particle in front of the central channel of the pore complex, unrolling of the ribbon and translocation of the RNP complex with its 5 end in the lead through the channel, exit of the unfolded RNP fibril into the cytoplasm, and formation of a polysome just outside the pore (8). Thus, the translocation of the BR RNP particle appears to be an ordered process with several well-defined stages. Furthermore, the spectacular conformational changes of the BR particle indicate that the process is quite dynamic, which is further supported by the observation that during translocation the BR particle loses proteins while others are presumably added (see below).

Fig. 1. Electron micrograph showing chromosome IV with its three giant puffs (BRs) in a salivary gland cell from C. tentans. The three BRs (BR1, BR2, and BR3) are indicated as well as the nucleoplasm (Npl) and cytoplasm (Cpl). The arrows mark a few prominent transcription loops (cf. Fig. 2D). (Bar equals 2 µm.) APPROACH TO STUDY BR RNA-BINDING PROTEINS Evidently proteins become associated with the RNA concomitant with transcription. In fact, the proteins seem to bind to the growing RNA molecule in the immediate vicinity of the RNA polymerase. Several questions are close at hand: What proteins are associated with the RNP particle? Are the proteins simply packaging proteins, or do they also play other functional roles? It has been estimated that there are 400–500 average-sized protein molecules in a BR particle (20). It is well established that pre-mRNA is associated with many different proteins, usually designated hnRNP proteins (heterogeneous nuclear RNP proteins) (21). For example, in humans there are 30 major hnRNP proteins and a large number of minor ones (22). As a rule, the proteins can bind to a broad range of different sequences, some with higher affinity, others with lower affinity (21). Thus, as the hnRNP proteins show sequence preference in their interaction with RNA, they are likely to be nonrandomly bound to pre-mRNA. It has been directly shown in reconstitution experiments that each different RNA species is associated with a unique combination of hnRNP proteins (23). These studies were performed under conditions for binding sites and, therefore, resemble the in vivo situation in the cell nucleus. Furthermore, the hnRNP protein compositions at various puffs on polytene chromosomes in Drosophila (24) and Chironomus (25) differ quantitatively but also qualitatively, suggesting that each type of transcript binds a specific subset of hnRNP proteins. It is, therefore, an interesting possibility that the hnRNP proteins are not only unspecific RNA packaging proteins but also capable of exerting specific, transcript-related functions. To test such a hypothesis, it is attractive to study the protein set-up of individual specific transcripts and relate the individual proteins to the fate of the transcript.

Fig. 2. Intracellular distribution of the cap-binding protein CBP20 in C. tentans salivary gland cells studied by immunoelectron microscopy. The assembly of the BR RNP particle is shown in A-D: proximal portions of the BR gene are displayed in A distal portions in B and C, and a schematic drawing of the BR gene in D (p, proximal; m, middle; d, distal portions of the gene). The fate of the released BR particles is shown in E-H: BR particles are present in the nucleoplasm (E), at the pore (F), and in an unfolded conformation when passing through the pore (G and H). Gold particles are marked by arrows and indicate the position of CBP20. It should be noted that gold particles are at the leading 5 end of the BR particle when it passes through the nuclear pore. (Bar equals 100 nm.) Modified from ref. 27; produced by permission of The Rockefeller University Press. It would have been most satisfactory if the proteins in the BR particles could have been studied by a direct approach. It is true that the BR particles can be isolated as a 300S fraction (20), but the quantities are not sufficient to allow a direct biochemical characterization. Instead, we adopted an indirect approach devised by Dreyfuss and coworkers (26). Nuclear RNA-binding proteins were isolated from C. tentans cultured cells by single-stranded DNA-Sepharose affinity chromatography and were used to raise monoclonal antibodies in mice. A collection of such antibodies was obtained (25). Antibodies that showed high specificity in Western blot experiments and bound to the BRs in immunocytochemical experiments were selected for further experiments. The antibodies were used to characterize the corresponding proteins by cDNA cloning and to study the fate of the proteins during the assembly and transport of the BR particle by using immunocytochemical and immunoelectron microscopy experiments.



7014

Fig. 3. Assembly and transport of the BR RNP particle and its relation to a number of BR RNA-associated proteins. The BR particle is assembled on the gene (left), passes through the nucleoplasm, unfolds, and translocates through the nuclear pore (middle). On the cytoplasmic side, the BR RNP fibril becomes engaged in protein synthesis and the polysomes anchor at the endoplasmic reticulum (right). The tripartite nuclear pore complex with its central channel is seen in black and its nuclear and cytoplasmic fibers are presented in pink. The BR gene with its five exons is displayed above the BR particle scheme, and the flow patterns of the BR RNA-associated proteins are outlined below. snRNP, small nuclear RNP. Modified from ref. 8; printed with permission from Elsevier Science. As an example of a protein flow analysis, I have chosen the immunoelectron microscopic analysis of a cap-binding protein, CBP20 (27). CBP20 is known to bind to the 5 end of the transcript in a cap-binding complex (CBC) together with another protein, CBP80 (28). An antibody raised against the human CBP20 was applied in the study of the BR particle. Cryosections through salivary gland cells were prepared and challenged with the anti-CBP20 antibody and subsequently with a secondary antibody coupled to gold. As shown in Fig. 2, the gold particles are present in the proximal portions of the active BR gene (Fig. 2A) as well as in the distal portions (Fig. 2 B and C). In the almost finished BR particles it can be seen that the gold is at the 5 end of the particle (Fig. 2B; cf. schematic drawing in Fig. 3)—i.e., the position of the cap structure. Furthermore, it was noted that there is no increase in binding during the course of transcription, suggesting that the protein is added to the cap structure almost immediately upon initiation of transcription. BR particles released into the nucleoplasm are also labeled with gold (Fig. 2 E and F). Finally, during translocation through the nuclear pore, the leading 5 end of the BR particle is labeled and the gold can also be seen on the cytoplasmic side of the nuclear pore complex (Fig. 2 G and H). Further out in the cytoplasm, there are no gold particles. We conclude that CBP20 is added cotranscriptionally and remains associated with the particle to and through the nuclear pore. On the cytoplasmic side, it is released from the particle and probably returns to the nucleus. These data are in good agreement with the observation that CBPs are shuttling proteins (28). During the last couple of years a number of various RNA-binding proteins have been studied, and our results are summed up in Fig. 3. The flow patterns of the proteins are presented below the morphological description of the assembly and transport of the BR particle; the exon-intron organization of the BR gene is shown above. It is evident that the various proteins show quite different behavior during gene expression. Thus, not only the particle’s morphology but also its protein composition during the transport from the gene to the cytoplasm is drastically changed. SPLICEOSOME ASSEMBLY AND DISASSEMBLY As a marker for Spliceosome components we chose the snRNP proteins and used a monoclonal anti-snRNP antibody (Y12) to perform immunoelectron microscopy experiments (29). When the growing BR RNP products were studied in situ, it was noted that the snRNP proteins were present mainly in the proximal portion and only to a minor extent in the middle and distal portions of the active gene. Furthermore, nucleoplasmic BR particles, isolated, unfolded, and spread on a grid surface, showed labeling only at one end of the transcript, presumably the 3 end. Thus, the snRNPs do not associate along the whole pre-mRNP fibril but rather bind to the 5 and 3 ends—i.e., the regions containing introns. These results nicely agree with an earlier analysis carried out at the RNA level, showing that the three 5 end introns are spliced concomitantly with transcription in the promoter-proximal third of the gene, whereas the 3 intron is spliced mainly posttranscriptionally (30). We conclude that the observed discontinuous distribution of snRNP proteins along the pre-mRNP fibril implies that spliceosomes both assemble and disassemble rapidly on the RNP fibril. PROTEINS CONFINED TO THE NUCLEUS Two of the studied proteins, hrp45 (31) and hrp23 (32), proved to be confined to the cell nucleus. The hrp45 protein contains two amino-terminal RNP-consensus RNA-binding domains (RBDs) and a carboxyl-terminal region rich in arginine-serine dipeptide repeats (RS domain), an organization characteristic of



7015

the SR family of RNA splicing factors (for review, see ref. 33). The hrp45 protein shows a high sequence homology to the human ASF-SF2 protein (34, 35) and the Drosophila SRp55 protein (36, 37), which are both known to be essential splicing factors (33). The hrp23 protein contains a single amino-terminal RED and a carboxyl-terminal auxiliary region rich in glycine, arginine, and serine. It resembles the RBDGly type of hnRNP proteins (e.g., hnRNP A1), which contain one or two RBDs and a glycine-rich auxiliary domain. However, hrp23 share features with the SR proteins (e.g., several SR/RS dipeptides in the auxiliary domain), suggesting that hrp23 represents a group of proteins intermediate in structure between these two major groups of pre-mRNA-binding proteins. The hrp23 protein has a homologue in Drosophila, ROX21 (38), which has recently been shown to be a splicing represser and, therefore, renamed RSF1 (represser splicing factor 1) (39). Thus, the two BR particle proteins hrp45 and hrp23 are likely to be splicing factors. Both hrp45 and hrp23 are added to the growing BR transcript along the large exon, and most likely along the entire transcript. Furthermore, they are both present in the nucleoplasmic BR particles, most of which contain fully spliced RNA (30). It should be stressed that neither of these putative splicing factors seems to behave as a genuine spliceosome component—i.e., a component that appears transiently on the pre-mRNP fibril and only at intron regions (compare the asymmetric distribution of snRNP proteins described above). Instead, they appear evenly along the transcript and remain with the fully spliced transcript in the nucleoplasm. It seems likely that the two proteins play important roles in the structural organization of the pre-mRNP particle, setting the stage for splicing rather than directly participating in the splicing process. The two proteins are not released at the same time in conjunction with the translocation of the BR particle through the nuclear pore: whereas hrp23 is shed just before or at the binding of the particle to the pore, hrp45 is released when the particle enters the central channel. Thus, it seems likely that there is not a single protein-removal step at nucleocytoplasmic transport but rather a series of preparatory steps before the actual translocation of the RNP particle through the pore. It could be speculated that the shedding of hrp23 is required for binding of the particle to the nuclear pore complex, whereas the removal of hrp45 is closely connected to the translocation of the particle through the central channel. It is interesting to note that some mammalian hnRNP proteins—e.g., hnRNP C—contain a nuclear retention signal in the auxiliary domain (40). This signal can override nuclear export signals in the shuttling hnRNP proteins and, therefore, the nonshuttling proteins have to be actively displaced from the hnRNP complex before the nucleocytoplasmic translocation. We conclude that the hrp23 and hrp45 proteins are removed in a consecutive fashion beginning before or at the binding of the RNP particle to the nuclear pore complex. The fact that the proteins behave differently during nucleocytoplasmic translocation could imply that each of them plays a specific role during export of mRNA from the nucleus to the cytoplasm. PROTEINS ACCOMPANYING THE MRNA INTO THE CYTOPLASM As discussed above, the CBP20 protein is bound to the cap structure early during transcription and accompanies the particle to and through the pore but is immediately dismissed just outside the pore. The rapid association of CBP20 with nascent RNA transcripts is consistent with the proposed role of the capbinding complex (CBC) in splicing and 3 end formation (28). Furthermore, the retention of the CBC on the RNP during translocation through the nuclear pore suggests that the CBC could also have a function at the recognition of the particle at the pore complex and/or in the translocation process itself when the 5 end of the RNA is in the lead. Such a view is supported by the observation that the transport of snRNP particles is dependent on the cap structure and CBPs (41). However, although the cap structure facilitates transport of mRNA, it does not seem to be necessary (42, 43). Because the exiting 5 end of the transcript is immediately engaged in protein synthesis, it is evident that the proteins bound to the cap structure are rapidly exchanged, CBPs being shed from the cap and translation initiation factors being recruited to the cap (28). Three proteins, hrp36 (44), actin (45), and hrp84 (J.Zhao, D. Nashchekin, N.Visa, and B.D., unpublished data), have been found accompanying the BR RNA all the way from the gene via the nuclear pore into polysomes in the cytoplasm. The hrp36 protein is a 2xRBD-Gly protein and resembles the human hnRNP A1 protein and the Drosophila hrp40 protein (21). The hnRNP A1 protein is known to be a shuttling protein (46) and contain a nuclear export signal (NES) (47). It was early proposed that hnRNP A1 functions as a transport mediator for mRNA (46). The observation that hrp36 is associated with BR RNA during its translocation through the nuclear pore is in good agreement with such a concept (44). However, it is also remarkable that hrp36 stays with the mRNA also during protein synthesis and remains distributed along the messenger molecule. The role of hrp36 in polysomes is still only a matter of speculation, but the appearance of hrp36 along the entire message suggests a global role. One possibility could be that it keeps the RNA extended, thereby facilitating proteinRNA interactions and the translation process. Another possibility would be that it is available to package the RNA when not translated (compare DNA and nucleosomes). A third possibility would be that hrp36, like other hnRNP proteins, favors cap-dependent initiation of translation by preventing aberrant initiations along the message (48). In our search for an export receptor binding to hrp36, we observed that actin forms a complex with hrp36 (45). It was shown first by immunoelectron microscopy that actin appears in the BR particle cotranscriptionally and remains attached to the particle in the nucleoplasm. Using DNase I affinity chromatography, we could demonstrate that actin is bound to hrp36 in nuclear as well as cytoplasmic extracts from C. tentans culture cells. The interaction is direct, as purified actin binds to recombinant hrp36 in an in vitro reconstitution experiment. Furthermore, the interaction between hrp36 and actin takes place in vivo as demonstrated by cross-linking. Thus, there is an hrp36-actin complex in the BR particle in the cell nucleus. This complex is also detected in the cytoplasm. As hrp36 enters polysomes, it seems likely that the complex is also present in the polysomes. A central issue is whether the actin is monomeric or polymeric. In the fixed cells studied we found no evidence for actin filaments in the salivary gland cell nucleus. Most remarkably, many of the actin-containing BR particles do not seem to be associated with any fibers. Furthermore, no phalloidin staining was detected in the nucleus, although the brush border of the salivary gland cells, known to contain Factin, was heavily stained. Finally, the anti-actin antibody used is known to have a strong preference for monomeric or short oligomeric actin. We conclude that in the fixed cells actin bound to hrp36 in the cell nucleus is likely to be in a monomeric or short oligomeric form. However, it has to be recalled that microfilaments can be extremely sensitive to fixation and could have disassembled during fixation. In fact, early microdissection experiments with C. tentans salivary gland cells showed that the polytene chromosomes are embedded in a labile gel (49), which has properties like the actin gel in amphibian oocyte nuclei (50). Thus, presently, the issue of the state of actin in the nuclear actin-hrp36 complex has to be left open. The state of actin in the cytoplasmic actin-hrp36 complex is also unclear, as it has not been possible to decide to what extent



7016

the immunolabeled actin in the cytoplasm reflects the distribution of the actin-hrp36 complex. It can be speculated that the hrp36-actin complex is important for packing the RNA into a BR RNP fibril and further into well-defined higher-order structures (51). Other possibilities would be that actin promotes interaction of the BR particle with a fibrous network in the nucleoplasm, allows binding to export receptors (cf. ref. 52), or is involved in the dramatic conformational change of the particle upon translocation through the nuclear pore. Because actin remains bound to hrp36 in the cytoplasm, it is important to recall that hnRNP proteins have been found to affect translation efficiency, mRNA stability, and RNA location within cytoplasm (22). Evidently, a wide range of functional options have to be considered for the actin-hrp36 complex. The third protein that is added cotranscriptionally to the BR transcript and accompanies the RNA through the nuclear pores and enters polysomes is hrp84, which is a putative RNA helicase. It belongs to the PL10 family of DEAD box proteins, which comprises, e.g., the human DBX (53), the mouse PL10 (54), the Xenopus An3 (55), and the yeast Ded1 proteins (56). The Ded1 protein is known to be important for initiation of translation (57). It is interesting to note that the mouse PL10 protein (57) and the human DBX (53) are probably also involved in the initiation of translation, as they can complement a deletion of the yeast gene DED1. Thus, it seems likely that hrp84 exerts its function in polysomes and presumably during the initiation of translation. We conclude that hrp84 represents a protein that functions in the polysomes in cytoplasm but is added to the transcript already in the nucleus. THE COTRANSCRIPTIONAL LOADING STAGE The general picture that emerges from our studies of the proteins in the BR particle is that during the assembly of the particle the premRNA molecule is loaded with proteins functioning early in the cell nucleus and with proteins functioning late in the cytoplasm. Thus, both the nuclear fate and the cytoplasmic fate of the mRNA are influenced by the proteins that are carried along with the RNA. This conclusion is supported by studies of gene expression in other species. The human hnRNP proteins are located predominantly in the cell nucleus, but many of them, including hnRNP A1, A2, D, E, and K, are shuttling between nucleus and cytoplasm (21, 46). In addition, more and more information accumulates showing that hnRNP proteins affect the fate of the mRNA in the cytoplasm—e.g., the transport of mRNA within the cytoplasm, the translational efficiency, and the mRNA turnover (for review, see ref. 22). It seems reasonable to assume that also these proteins travel with the mRNA from the nucleus to the cytoplasm, like hrp36 and hrp84 with BR RNA in C. tentans. Recently, it was shown in Drosophila by microinjection experiments that the proper cytoplasmic localization of fushi tarazu transcripts requires that the transcript enters the cytoplasm associated with the hnRNP protein hrp40 (58). We conclude that proteins loaded cotranscriptionally on pre-mRNA determine to a large extent the fate of the mRNA in both the nucleus and the cytoplasm. The assembly of proteins along the pre-mRNA molecule is likely to be a complex process (Fig. 4). Some of the proteins, such as the cap-binding protein CBP20, bind to a specific sequence with high affinity, whereas most of the major RNA-binding proteins, such as the 2xRBD-Gly protein hrp36, bind with lower affinity at many sites along the pre-mRNA molecule. The presence of many RNA-binding proteins with limited sequence specificity will result in competition for available binding sites. The outcome of the assembly will, therefore, depend on the particular proteins present for binding and their relative abundance in the vicinity of the gene. It should be emphasized that the composition of nuclear hnRNP proteins is known to vary considerably between tissues and developmental stages (59). Thus, the set of proteins associated with a given transcript is not likely to be fixed but rather dependent on the cell type studied, physiological conditions, etc.

Fig. 4. Cotranscriptional loading of proteins onto growing BR pre-mRNA molecules. Some proteins bind to the pre-mRNA with high sequence specificity (e.g., CBP20), whereas others bind with lower specificity along the entire RNA molecule (e.g., the SR protein hrp45 and the 2xRBD-Gly protein hrp36). The RNP fibril formed serves as the substrate for trans-acting factors, and its structure affects a number of mRNA-related processes, including splicing, transport, and translation. As most proteins bound to the pre-mRNA not only are packaging proteins but also exert more specific functions during the gene expression process, a modulated loading of the transcript with proteins will have functional implications. For example, it has been shown that the relative amounts of the two antagonistically acting RNA-binding proteins hnRNP A1 and ASF/SF2 decide the outcome of alternative splicing (60). The primary transcription product, the pre-mRNP fibril, should therefore be looked upon as a variable substrate for transacting factors, and the molecular organization of the fibril will influence not only splicing but also processes such as transport, translation, and mRNA degradation. Unfortunately, today we have only limited information on how the RNP fibril is organized at the molecular level, and we know even less about the rules that govern the assembly of the RNP fibril. CONCLUSIONS A specific transcription product, the BR RNP particle, has been studied during assembly on the gene and transport through the nucleoplasm to and through the nuclear pores. On the cytoplasmic side, the BR RNP particle appears as an extended RNP fibril that immediately engages in protein synthesis. A number of BR RNA-associated proteins have been identified, and their flow patterns have been studied in relation to the assembly and transport of the BR particle. The following major conclusions have been drawn: (i) The BR RNP particle carries a specific subset of hnRNP proteins. (ii) The proteins are added to the pre-mRNA cotranscriptionally.



7017

(iii) The various proteins behave differently during RNA transport: some leave the transcript in the nucleoplasm or at the nuclear pore, others are shed subsequent to the translocation of the particle through the nuclear pore, whereas still others accompany the mRNA into polysomes. (iv) The flow pattern of a protein seems related to the function of the protein. (v) The particle proteins exert specific mRNA-related functions rather than merely serving as RNA-packaging devices. (vi) The cotranscriptional assembly process sets the stage for both the nuclear and the cytoplasmic fate of the mRNA sequence. I thank Sergej Masich, Birgitta Björkroth, and Birgitta Ivarsson for preparing the figures. The research was supported by the Swedish Natural Science Research Council, the Human Frontier Science Program Organization, the Knut and Alice Wallenberg Foundation, the Marianne and Marcus Wallenberg Foundation, and the Gunvor and Josef Anér Foundation. 1. Cremer, T., Kurz, A., Zirbel, R., Dietzel, S., Rinke, B., Schröck, E., Speicher, M.R., Mathieu, U., Jauch, A., Emmerich, P., et al. (1973) Cold Spring Harbor Symp. Quant. Biol. 58, 777–792. 2. van Driel, R., Wansink, D.G., van Steensel, B., Grande, M.A., Schul, W. & de Jong, L. (1995) Int. Rev. Cytol. 162A, 151–188. 3. Specter, D.L. (1990) Proc. Natl. Acad. Sci. USA 87, 147–151. 4. Zachar, Z., Kramer, J., Mims, I.P. & Bingham, P.M. (1993) J. Cell Biol. 121, 729–742. 5. Fakan, S., Puvion, E. & Spohr, G. (1976) Exp. Cell Res. 99, 155–164. 6. Kurz, A., Lampel, S., Nickolenko, J.E., Bradl, J., Benner, A., Zirbel, R.M., Cremer, T. & Lichter, P. (1996) J. Cell Biol. 135, 1195–1205. 7. Dworetzky, S.I. & Feldherr, C.M. (1988) J. Cell Biol. 106, 5J5–584. 8. Daneholt, B. (1997) Cell 88, 585–588. 9. Xing, Y., Johnson, C.V., Dobner, P.R. & Lawrence, J.B. (1993) Science 259, 1326–1330. 10. Case, S.T. & Daneholt, B. (1977) in Biochemistry of Cell Differentiation II, ed. Paul, J. (University Park Press, Baltimore), pp. 45–77. 11. Wieslander, L. & Paulsson, G. (1992) Proc. Natl. Acad. Sci. USA 89, 4578–4582. 12. Wieslander, L. (1994) Prog. Nucleic Acids Res. Mol. Biol. 48, 275–313. 13. Case. S.T. & Wieslander, L. (1992) in Biopolymers, Results and Problems in Cell Differentiation, ed. Case, S.T. (Springer, Berlin), Vol. 19, pp. 187– 226. 14. Lamb, M.M. & Daneholt, B. (1979) Cell 17, 835–848. 15. Skoglund, U., Andersson, K., Björkroth, B., Lamb, M.M. & Daneholt, B. (1983) Cell 34, 847–855. 16. Miller, O.L., Jr., & Beatty, B.R. (1969) Science 164, 955–957. 17. Singh, O.P., Björkroth, B., Masich, S., Wieslander.L. & Daneholt, B. (1999) Exp. Cell Res. 251, 135–146. 18. Miralles, F., Öfverstedt, L.-G., Sabri, N., Aissouni, Y., Hellman, U.Skoglund, U. & Visa, N. (2000) J. Cell Biol. 148, 271–282. 19. Mehlin, H., Daneholt, B. & Skoglund, U. (1992) Cell 69, 605–613. 20. Wurtz, T., Lönnroth, A., Ovchinnikov, L., Skoglund, U. & Daneholt, B. (1990) Proc. Natl. Acad. Sci. USA 87, 831–835. 21. Dreyfuss, G., Matunis, M.J., Pinol-Roma, S. & Burd, C.G. (1993) Annu. Rev. Biochem. 62, 289–321. 22. Krecic, A.M. & Swanson, M.S. (1999) Curr. Opin. Cell Biol. 11, 363–371. 23. Bennett, M., Pinol-Roma, S., Staknis, D., Dreyfuss, G. & Reed, R. (1992) Mol. Cell. Biol. 12, 3165–3175. 24. Matunis, E.L., Matunis, M.J. & Dreyfuss, G. (1993) J. Cell Biol. 121, 219–228. 25. Wurtz, T., Kiseleva, E., Nacheva, G., Alzhanova-Ericsson, A., Rosén, A. & Daneholt, B. (1996) Mol. Cell. Biol. 16, 1425–1435. 26. Pinol-Roma, S., Choi, Y.D., Matunis, M.M. & Dreyfuss, G. (1988) Genes Dev. 2, 215–227. 27. Visa, N., Izaurralde, E., Ferreira, J., Daneholt, B. & Mattaj, I.W. (1996) J. Cell Biol. 133, 5–14. 28. Lewis, J.D. & Izaurralde, E. (1997) Eur. J. Biochem. 247, 461–469. 29. Kiseleva, E., Wurtz, T., Visa, N. & Daneholt, B. (1994) EMBO J. 13, 6052–6061. 30. Baurén, G. & Wieslander, L. (1994) Cell 76, 183–192. 31. Alzhanova-Ericsson, A., Sun, X., Visa, N., Kiseleva, E., Wurtz, T. & Daneholt, B. (1996) Genes Dev. 10, 2881–2893. 32. Sun, X., Alzhanova-Ericsson, A., Visa, N., Aissouni, Y., Zhao, J. & Daneholt, B. (1998) J. Cell Biol. 142, 116J–1180. 33. Fu, X.-D. (1995) RNA 1, 663–680. 34. Ge, H., Zuo, P. & Manley, J. L. (1991) Cell 66, 373–382. 35. Krainer, A.R., Mayeda, A., Kozak, D. & Binns, G. (1991) Cell 66, 383–394. 36. Champlin, D.T., Frasch, M., Saumweber, H. & Lis, J.T. (1991) Genes Dev. 5, 1611–1621. 37. Roth, M.B., Zahler, A.M. & Stolk, J.A. (1991) J. Cell Biol. 115, 587–596. 38. Brand, S.F., Pichoff, S., Noselli, S. & Bourbon, H.-M. (1995) Genes 154, 187–192. 39. Labourier, E., Bourbon, H.-M., Gallouzi, I., Fostier, M., Allemand, E. & Tazi, J. (1999) Genes Dev. 13, 740–753. 40. Nakielny, S. & Dreyfuss, G. (1996) J. Cell Biol. 134, 1365–1373. 41. Izaurralde, E., Lewis, J., Gamberi, C., Jarmolowski, A., McGuigan, C. & Mattaj, I.W. (1995) Nature (London) 376, 709–712. 42. Hamm, J. & Mattaj, I.W. (1990) Cell 63, 109–118. 43. Jarmolowski, A., Boelens, W.C, Izaurralde, E. & Mattaj, I.W. (1994) J. Cell Biol. 124, 627–635. 44. Visa, N., Alzhanova-Ericsson, Sun, S., Kiseleva, E., Björkroth, B., Wurtz, T. & Daneholt, B. (1996) Cell 84, 253–264. 45. Percipalle, P., Zhao, J., Pope, B., Weeds, A., Lindberg, U. & Daneholt, B. (2001) J. Cell Biol. 153, 229–236. 46. Pinol-Roma, S. & Dreyfuss, G. (1992) Nature 355, 730–732. 47. Michael, W.M., Choi, M. & Dreyfuss, G. (1995) Cell 83, 415–422. 48. Svitkin, Y.V., Ovchinnikov, L.P., Dreyfuss, G. & Sonenburg, N. (1996) EMBO J. 15, J147–7155. 49. D’Angelo, E.G. (1946) Biol. Bull. 90, 71–87. 50. Clark, T.G. & Merriam, R.W. (1977) Cell 12, 883–891. 51. Skoglund, U., Andersson, K., Strandberg, B. & Daneholt, B. (1986) Nature (London) 319, 560–564. 52. Wada, A., Fukuda, M., Mishima, M. & Nishida, E. (1998) EMBO J. 17, 1635–1641. 53. Mamiya, N. & Worman, H.J. (1999) J. Biol. Chem. 214, 15751–15756. 54. Leroy, P., Alzari, P., Sassoon, D., Wolgemuth, D. & Fellous, M. (1989) Cell 57, 549–559. 55. Gururajan, R., Perry-O’Keefe, H., Melton, D.A. & Weeks, D.L. (1991) Nature (London) 349, 717–719. 56. Jamieson, D.J., Rahe, B., Pringle, J. & Beggs, J.D. (1991) Nature (London) 349, 715–717. 57. Chuang, R.Y, Weaver, P.L., Liu, Z. & Chang, T.H. (1997) Science 275, 1468–1471. 58. Lall, S., Francis-Lang, H., Flament, A., Norvell, A., Schüpbach, T. & Ish-Horowicz, D. (1999) Cell 98, 171–180. 59. Kamma, H., Portman, D.S. & Dreyfuss, G. (1995) Exp. Cell Res. 221, 187–196. 60. Mayeda, A. & Krainer, A. R. (1992) Cell 68, 365–375.


RIBONUCLEOPROTEIN INFRASTRUCTURE REGULATING THE FLOW OF GENETIC INFORMATION BETWEEN THE GENOME AND THE PROTEOME

7018

Colloquium Ribonucleoprotein infrastructure regulating the flow of genetic information between the genome and the proteome Jack D.Keene* Department of Microbiology, Duke University Medical Center, Durham, NC 27710 Following transcription and splicing, each mRNA of a mammalian cell passes into the cytoplasm where its fate is in the hands of a complex network of ribonucleoproteins (mRNPs). The success or failure of a gene to be expressed depends on the performance of this mRNP infrastructure. The entry, gating, processing, and transit of each mRNA through an mRNP network helps determine the composition of a cell’s proteome. The machinery that regulates storage, turnover, and translational activation of mRNAs is not well understood, in part, because of the heterogeneous nature of mRNPs. Recently, subsets of cellular mRNAs clustered as members of mRNP complexes have been identified by using antibodies reactive with RNA-binding proteins, including ELAV/Hu, elF-4E, and poly(A)-binding proteins. Cytoplasmic ELAV/Hu proteins are involved in the stability and translation of early response gene (ERG) transcripts and are expressed predominately in neurons. mRNAs recovered from ELAV/Hu mRNP complexes were found to have similar sequence elements, suggesting a common structural linkage among them. This approach opens the possibility of identifying transcripts physically clustered in vivo that may have similar fates or functions. Moreover, the proteins encoded by physically organized mRNAs may participate in the same biological process or structural outcome, not unlike operons and their polycistronic mRNAs do in prokaryotic organisms. Our goal is to understand the organization and flow of genetic information on an integrative systems level by analyzing the collective properties of proteins and mRNAs associated with mRNPs in vivo. Understanding the physical organization of gene transcripts in mammalian cells has presented significant difficulties for several reasons: (i) mRNA is generally unstable, (ii) each mRNA is relatively inabundant, and (iii) the association of each mRNA with proteins results in the formation of heterogeneous complexes with diverse biophysical properties. These qualities, together with the lack of suitable technologies to purify ribonucleoprotein (mRNP) complexes, have precluded molecular dissection of their component parts. Recently, immunological and biochemical techniques, combined with genomic methods, have allowed the elucidation of mRNAs associated with ELAV/Hu and other mRNP complexes, and the subsequent analysis of their structural and functional properties. This approach to understanding the collective properties of the mRNP infrastructure, and the organized networks of transcripts that are regulated posttranscriptionally has been termed ribonomics (1). ELAV/Hu RNA Recognition Motif (RRM) Proteins Associate with a Distinct Subset of mRNAs. ELAV/Hu proteins were useful for developing ribonomic methods because they bind in vitro to a class of messenger RNA containing AU-rich sequences (2–6). Although many early response gene (ERG) mRNAs that contain AU-rich elements (ARE) in their 3 UTRs tend to be unstable in vivo, many other mRNAs that are not considered unstable also contain AU-rich regions. Several different proteins have been reported to bind AU-rich elements in mRNA, but the ELAV/Hu proteins are unique in that they have been shown to stabilize and/or activate translation of target mRNAs (reviewed in refs. 7 and 8). Three of the four known types of ELAV/Hu protein (HuB, HuC, and HuD) are expressed specifically in neurons or gonads and are predominately cytoplasmic, which is consistent with a role in mRNA stability and translation (reviewed in ref. 9). As shown in Fig. 1A, ELAV/Hu proteins reside in cytoplasmic granules that extend out of the cell body and along dendrites (10, 11). It is presumed that these granules represent mRNPs containing ELAV/Hu proteins bound to mRNAs. An early clue that suggested a role for ELAV/Hu proteins in translational control was the altered distribution of the protein in cortical neurons following treatment with puromycin, an antibiotic that blocks the elongation of mRNAs on polysomes (Fig. 1B; ref. 11). The presence of ELAV/Hu proteins in dendritic granules is consistent with their playing a localized role in translation. Ectopic expression of HuB in 3T3L1 preadipocytes and in hNT2 preneuronal teratocarcinoma cells resulted in translational activation of target mRNAs encoding glucose transporter-1 protein (12) and neurofilament M protein (NF-M) (13), respectively. Moreover, in the hNT2 preneuronal cells (13), and in chicken neural crest cells (14), forced expression of Hu proteins resulted in the spontaneous development of neurites. In addition to having a profound biological effect on cell morphology, stability, and translation of specific target mRNAs, ELAV/Hu proteins appear to be multitargeted toward a broad range of AU-rich and ERG-type mRNAs. Based on in vitro selection of an AU-rich consensus sequence, Levine et al. (2) tested the binding of HuB in vitro to transcripts representing c-myc, c-fos, and GM-CSF and found high affinity binding. To define the larger mRNA-binding population, methods were subsequently developed to select mRNAs from cDNA libraries by using HuB (3). This resulted in the identification of at least one hundred putative mRNA targets for the HuB protein. In nearly every case, these mRNAs represented members of a subset of cellular growth regulatory proteins containing AREs. This result opened the intriguing possibility that dozens of ELAV/Hu targeted mRNAs containing AREs could be stabilized and/or translationally activated as a group in response to ELAV/Hu protein. mRNAs shown to be affected following overexpression of Hu proteins include glucose trans

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: mRNP, ribonucleoprotein; ERG, early response gene; ARE, AU-rich elements; NF-M, neurofilament M protein; RRM, RNA recognition motif (RRM). *To whom reprint requests should be addressed at: Department of Microbiology, Box 3020, Duke University Medical Center, Durham, NC 27710. E-mail: [email protected].



7019

porter 1 (12), NF-M (13), GAP43 (15), VEGF (16), c-fos (17–19), c-myc (unpublished results), TNF-α (19), GM-CSF (19), and tau (20). With the exception of NF-M mRNA (13), the binding of these mRNA targets to ELAV/Hu proteins has only been demonstrated when using in vitro methods.

Fig. 1. ELAV/Hu RRM proteins form distinct granules in cell body and dendrites of neurons. Rabbit polyclonal serum prepared against recombinant HuB was used to visualize ELAV/Hu proteins in isolated rat embryonal cortical neurons by using confocal microscopy (reprinted from ref. 11). Prebleed serum showed no appreciable fluorescence of any neuronal samples (10, 11). The granules containing Hu proteins (A) coalesced following treatment with puromycin (B) to disrupt translation. [Reproduced with permission from ref. 11 (Copyright 1998, J. Cell Sci.).] Messenger RNAs Are Generally Inabundant and Unstable. The average number of any particular mRNA species present in a mammalian cell varies over a range from less than one to as many as 1,000. This is in contrast to the U1 snRNA that is present in approximately 1 million copies per cell. In human cells, an average of about six copies of each mRNA per cell has been approximated with very few genes having at steady state as high as 50 to 100 copies per cell (21). In yeast, this number is approximately an order of magnitude lower. It is striking that so few copies of each mRNA are maintained in the steady state, and this suggests that mRNAs are continuously supplied and destroyed during normal cell metabolism. It is likely that a constant flux of mRNA through the mRNP infrastructure provides agility to the gene expression program. In profiling the expression of mRNAs by using techniques like microarray analysis or Serial Analysis of Gene Expression (SAGE), the steady-state level of each mRNA can be quantitated (22, 23). However, these procedures do not distinguish translationally active messages from inactive messages, and the relative turnover rate of each message can significantly affect protein output (23). Furthermore, the organization of mRNAs into functional complexes may influence their state of expression. The instability of many mRNAs in comparison to ribosomal RNAs, transfer RNAs, and small nuclear RNAs, as well as their inabundance, has made analysis of their in vivo-associated protein interactions particularly difficult. As a result, most of what is currently known about mRNA-protein interactions has been derived from in vitro binding experiments. Nonetheless, the stability of endogenous mRNAs has been studied by using a variety of analytical tools (24–26). The relative stability of mRNAs involved in various biological processes can be depicted on a time line along which the half lives of ERG mRNAs (such as protooncogene and cytokine transcripts) is as short as a few minutes, and housekeeping proteins like cytoskeletal components and histones have half-lives equivalent to one full cell cycle (Fig. 2). It is generally true that whereas mRNAs that encode highly abundant and stable housekeeping proteins appear to be stable themselves, mRNAs encoding many growth regulatory proteins are very unstable (25). This instability is presumably due to the powerful and possibly undesirable effects on normal cell growth and differentiation that these gene products can have. The necessity to retain tight control over growth stimulatory proteins begins at the level of transcription, but is usually maintained also at the posttranscriptional level. Short half-lives for mRNAs encoding growth factors such as c-fos or c-myc allow cells to retain tighter control at the level of transcription, and therefore, the final production of the protein can be regulated with greater precision (24–26). In keeping with this line of reasoning, the ERG (also known as the immediate early gene) products encode growth regulatory proteins, and include mRNAs with short half-lives. The ability of the ELAV/Hu proteins to bind certain AU-rich-region-containing ERG mRNAs suggested to us that a large target set of AUrichregion-containing mRNAs might be captured by using ELAV/Hu proteins to identify en masse a unique subset of the total cell mRNA population (3). More recently, a direct in vivo approach has been possible by isolating mRNP complexes and identifying the mRNA subsets by using nucleic acid hybridization (1).

Fig. 2. The relative stability of some diverse cellular mRNAs. The Heterogeneous Nature of mRNPs. Heterogeneous nuclear RNA (hnRNA) and the correspondingly diverse hnRNPs have been recognized for many years (reviewed in refs. 27–30). Analysis on density and velocity gradients has revealed that whole-cell mRNA, and mRNA-binding proteins, often spread across a gradient making it difficult to discern specific proteins or mRNAs that might associate with one another (Fig. 3). This heterogeneity has caused the field to rely heavily on in vitro binding methods for study of the interactions between individual proteins and the sequence elements found in mRNAs (30, 31). It has been possible to analyze the migration of individual mRNAs on sucrose velocity gradients by using Northern blotting of gradient fractions. Indeed, several studies have used gradient analysis to localize translationally engaged mRNAs in fractions containing active polysomes (12, 13, 32, 33). However, mRNPs that are not associated with the assembled translation apparatus often remain widely distributed between the free mRNA and the assembled polysomes as exemplified with ELAV/Hu proteins (Fig. 3A). It has been assumed that these widely distributed mRNPs represent complexes containing mRNAs that are competent for, but not engaged in, translation. However, as shown in Fig. 3, treatment of cell extracts with EDTA can release the ELAV/Hu proteins from the region of active translation (ß complexes) and shift them to an intermediate position (a complexes). Similar results were evident when cells were treated with puromycin to inhibit translation without disrupting polysomes (Fig. 3C). As described by Tenenbaum et al. (1), mRNP complexes that are recovered by immunoprecipitation of tagged HuB from transfected P19 cells following treatment with EDTA



7020

contained a unique subset of mRNAs possessing AU-rich sequences. This offers a useful convention for discerning aspects of the physical organization of mRNAs in mRNP complexes.

Fig. 3. Distribution of ELAV/Hu proteins following separation on Accudenz density gradients. (A) ELAV/Hu proteins that migrate in the upper (Left) and in the lower (Right) regions of the gradient represent free protein and polysome-associated protein, respectively. In the region between these populations (α-complexes) are ELAV/Hu mRNPs. Treatment with EDTA (B) or puromycin (C) to disrupt the translation apparatus results in the appearance of ELAV/Hu protein in the a-complex region. The release of ELAV/Hu protein from ß-complexes following treatment with EDTA (B) was monitored by tracking the ribosomal protein L22. Puromycin treatment (C) also resulted in the appearance of α-complexes with a concomitant loss of ß-complexes, but as expected, did not disrupt polysomes as shown by the migration of L22. [Reproduced with permission from ref. 11 (Copyright 1998, J. Cell Sci.).]. Combinatorial Interactions in mRNPs. The composition of mRNP complexes includes the mRNA(s) and the proteins that assemble onto the mRNAs in various combinations. Not unlike transcription, splicing, and polyadenylation, translation also involves combinatorial interactions of RNAs and proteins (34, 35). For example, in the case of translation factors, the circularization of the engaged mRNA is mediated by interactions between eIF-4E and eIF-4G. Whereas some proteins like the ELAV/Hu are bound directly to AU-rich-regioncontaining mRNAs, other factors are not directly bound, but participate in forming RNP complexes through protein-protein interactions with RNA-binding proteins. The availability of a variety of combinations of assembled factors is thought to allow specificity of recognition, and possibly specificity of localization of transcripts. Although it has yet to be demonstrated, thousands of different combinations of transcripts may be organized into distinct mRNP classes by using different combinations of proteins. Via this process, it would not be necessary to access thousands of different RNA binding proteins to govern the organization of thousands of different mRNAs. However, it has not yet been determined for any mRNP complex whether multiple mRNAs are clustered together into the same physical particle. Nor is it known whether the regulation of expression of clustered mRNAs is coordinated. Although the physical organization of mRNAs within the mRNP infrastructure involves both direct and indirect RNA binding, the factors that regulate each mRNA are yet to be determined. Regardless of the exact structure of ELAV/Hu mRNPs, the ERG mRNAs that successfully exit the nucleus are hypothesized to pass into a cytoplasmic infrastructure where their fate is determined by the organizational properties of mRNA-binding proteins and mRNP-associated factors (reviewed in ref. 7).

Fig. 4. ELAV/Hu protein colocalizes with poly(A)+ mRNA in distinct cytoplasmic granules. Total poly(A)+ mRNA in puromycin-treated human medulloblastoma cells was stained with an antibody to a digoxigenin-labeled oligo dT probe (A) and ELAV/Hu protein was stained (B) with the same recombinant HuB antibody used in Fig. 1 (reprinted from ref. 11). Large arrowheads show the coalesced ELAV/Hu protein granules that overlap a region stained for the total poly(A)+ mRNA population (C), whereas arrows show regions stained for poly(A)+ mRNA but without detectable ELAV/Hu protein. Recovery of ELAV/Hu complexes is expected to reveal those mRNAs within the population of total cell mRNA that are specifically associated with ELAV/Hu mRNPs. [Reproduced with permission from ref. 11 (Copyright 1998, J. Cell Sci.).] As shown in Fig. 4, ELAV/Hu mRNP complexes visually overlap with polyadenylated mRNA in the cytoplasm of medulloblastoma cells (11). As expected, only a fraction of poly(A)+ mRNA colocalizes with ELAV/Hu protein. For example, Fig. 4C shows the overlap of both poly(A)+ mRNA and the ELAV/Hu proteins as yellow granules following treatment with puromycin. These data are consistent with the demonstrated role of HuB in activating the translation and stability of target mRNAs (7), but also illustrate the clusters of mRNP complexes containing both ELAV/Hu proteins and polyadenylated mRNAs. These findings led to experiments designed to isolate ELAV/Hu mRNP complexes following their release from polysomes to characterize the associated mRNAs (1). It is presumed that these treatments release mRNPs similar to the α complexes shown in Fig. 3. Therefore, the limitations imposed by in vitro binding and selection (3) can be overcome by the isolation of endogenous mRNPs from cell extracts and identification of the mRNAs contained in mRNPs by directly identifying the mRNA subset on microarrays (1). Ribonomics provides a set of biochemical conventions for isolating mRNPs that can be applied systematically to determine the clustering of mRNAs that have



7021

structural commonality, and potentially functional relationships among their gene products. Is it possible that multicellular organisms derive genetic complexity by organizing various combinations of mRNAs as mRNPs rather than using polycistronic transcripts from operons? Posttranscriptional Regulation of Gene Expression Involving Hu mRNPs. Why is it valuable to identify mRNA subsets in messenger RNP complexes? One reason is that the expression of gene products encoded by mRNAs may need to be temporally controlled, whether sequentially or simultaneously. For example, during neuronal differentiation new proteins including neurofilaments, MAP proteins, and tau proteins are expressed sequentially following addition of retinoic acid to embryonic carcinoma cells (36). Although the transcription program is essential to neuronal differentiation, posttranscriptional regulation also has been documented to be important during growth and differentiation in neuronal and other systems (37–42). As noted above, mRNAs encoding neurofilament M (13) and tau (22) can bind in their noncoding regions to ELAV/Hu proteins that are, in turn, induced by retinoic acid before the appearance of NF-M or tau. In the case of NF-M, it has been shown that the HuB protein recruits the mRNA to active polysomes where protein production is up-regulated (13). Other examples of posttranscriptional regulation of mRNA expression by the ELAV/Hu proteins have been demonstrated with other early response gene (ERG) transcripts (7). It appears that many neuron-specific mRNAs are contained in mRNP complexes with the ELAV/Hu proteins and other RNA-binding proteins, and that their expression is activated during differentiation (1). The clustering of mRNAs as subsets that are expressed during neuronal differentiation and captured in these mRNP complexes may indicate that they are posttranscriptionally regulated in parallel. It is hypothesized that ELAV/Hu mRNPs represent a critical node in a pathway of posttranscriptional gene regulation in which decisions to stabilize, degrade, or translate multiple members of a subset of mRNAs can affect neuronal differentiation (refs. 2 and 3; reviewed in ref. 7). Because many of the ARE-containing mRNAs shown to bind ELAV/Hu proteins encode transcription factors such as fos, myc, Id, and CREB (2–9, 43), there is potential for posttranscriptional events to feedback and alter the transcriptional program during differentiation. Likewise, secreted cytokines whose mRNAs also contain AREs have the potential to affect the growth and activation of T cells in an autocrine or paracrine manner (44). Trans-acting mRNA-binding proteins that affect the expression of cytokine mRNAs as a distinct subset have not been identified, but are thought to exist. The ubiquitously expressed ELAV/Hu protein, HuA (HuR), can bind to cytokine mRNAs in T cells (ref. 45; unpublished results), but has not been shown to have a direct regulatory role on cytokine expression. It is likely that specialized ARE-binding proteins regulate subclasses of cytokine mRNAs because different cytokines are regulated independently of one another at the posttranscriptional level (44, 45). It is unlikely that the ARE represents the only cis-acting sequence among cellular transcripts that defines a structurally or functionally related subset of mRNAs. By using RNA-binding or RNP-associated proteins to isolate mRNPs, it should be possible to discover such relationships. Our laboratory has isolated mRNPs containing cap-binding protein, poly(A)-binding protein, and other mRNP proteins, and detected mRNAs, which appear to have relationships with one another (ref. 1; unpublished results). For example, we have used antibodies reactive with the p62 RRM/KH protein, which is a member of the insulin-like growth factor mRNA-binding protein (IMP) family (46), and detected a distinct set of mRNAs (S.Tenenbaum, C. Carson, P.Lager, E.Tan, and J.D.K., unpublished results). Among the subset were three mRNAs reported to bind members of the IMP RNA-binding protein family: c-myc, ß-actin, and insulin-like growth factor mRNAs. One implication of identifying mRNA subsets that encode functionally linked proteins is that they may be involved in the same biochemical pathway or form the same macromolecular structure. Thus, coordinated expression may be regulated at the posttranscriptional level much like operons are regulated at the transcriptional level in prokaryotic systems. Gene Expression May Be Regulated Posttranscriptionally in Dendrites. One intriguing hypothesis regarding the organization of mRNAs in neurons is that posttranscriptional events in the cytoplasm may affect transcriptional events in the nucleus. For example, mRNAs encoding transcription factors appear to be packaged in the cytoplasm at distances far from the nucleus, and their localized expression in response to external stimuli may influence cellular mechanisms in the nucleus (39, 40, 47). As noted above, many of the mRNAs to which the ELAV/Hu proteins bind encode transcription factors, including CREB, ERG-1, fos, myc, and Id (1–9). Eberwine and colleagues (39) have suggested that “nuclear imprinting” is a phenomenon in which the production of transcription factors is regulated posttranscriptionally in dendrites. The expression of these factors is activated locally following stimulation of neurons, thus leading to secondary activation of nuclear genes when the transcription factors are transported back to the nucleus (39, 47). The advantages of such a regulatory pathway may include direct activation of specific genes (e.g., ERG) without the potential complications involved in activating multiple signal transduction cascades intended to activate multiple downstream functions. We have proposed that the ELAV/Hu proteins could be involved in multifunctional activation in neurons by regulating not only transcription factor mRNAs, but also other ERG-type mRNAs that participate in intracellular signaling, cytoskeletal assembly, and membrane activity (reviewed in refs. 7 and 9). Parallel Analysis of mRNPs Implicated in Posttranscriptional Gene Expression: A Ribonomic Approach. It is possible to classify mRNA-binding proteins into three groups: those that are global and bind nearly all mRNAs without distinguishing unique sequences, the group-specific mRNA-binding proteins that associate with subsets of the global mRNA population, and those that are type-specific because they recognize a highly unique mRNA sequence, perhaps present in only one mRNA, with high specificity. We suggest that in some cases, the group-specific mRNA-binding proteins associate with multiple mRNAs that are structurally, and/or perhaps, functionally related. The functional relationships may concern RNA stability or instability, translational activation, transport, or the mRNA subset may encode a group of proteins involved in a common pathway or phenotypic outcome. Whereas the ARE-containing mRNAs represent an example of a group-specific subset of mRNAs that are regulated at the level of stability and translation, the iron response element (IRE)-binding protein (48) and the histone mRNA stem-loop-binding protein (49, 50) represent type-specific mRNA-binding proteins that are also involved in RNA stability and translation. The type-specific proteins recognize sequence elements that tend to encode protein products needed in large amounts within short time intervals during biological processes such as the cell cycle (50). As suggested above, it is not likely that every mRNA transcript has its own unique binding protein because tens of thousands of cell proteins would have to be dedicated to controlling posttranscriptional gene expression. Because mRNA-protein interactions in vivo are likely to be combinatorial, it is reasonable to predict that most mRNAs can be grouped into structurally and/or functionally related subsets that associate with a limited



7022

set of protein components. Gaining access to these putative mRNA subsets requires an ability to isolate mRNPs by using biochemical procedures. A major goal of our ribonomic approaches is to understand the dynamics of these mRNA subsets and their structural and/or functional clustering during growth and development. Structural and Functional Linkages Among mRNAs Clustered in mRNP Complexes. mRNAs that share common sequence elements in their untranslated or coding regions have the potential to interact with the same RNA-binding proteins at those sites. With the exception of the ARE, few sequence elements common to a collection of mRNAs have been identified. Searching for homologous sequence elements by computer may reveal common features among multiple mRNAs, but a more concrete approach would be to design methods that allow RNA-binding proteins to find such elements. The method of Gao et al. (3) provided an in vitro approach to partitioning mRNA populations with related sequence elements. More recently, the ribonomic approach of Tenenbaum et al. (1) has opened the possibility of identifying mRNAs with related protein binding elements by immunoprecipitation of proteins present in mRNP complexes followed by microarray analysis. Via this approach, ARE-binding proteins of the Hu family have made it possible to identify multiple mRNAs with common Hu protein binding elements. Broader applications using many other mRNP proteins are expected to identify additional structural linkages among subsets of mRNAs. The most important feature in common among a group of physically clustered mRNAs associated with mRNPs would be a functional relationship among their encoded proteins. If mRNAs that are isolated by purifying mRNP complexes encode proteins that function in a common pathway, it is logical to conclude there is a functional linkage. On the other hand, if functional relationships among the protein products encoded by an mRNA subset were not readily apparent from the literature, it would be necessary to investigate their individual functions. This raises the question of what properties define functional relationships among gene products. Our hypothesis is that mRNP complexes contain mRNAs that encode proteins that work together in a biological process or form a biological structure such as a ribosome or a spliceosome. In many cases, the encoded products may appear to be of diverse function, perhaps because they regulate complex biological outcomes. In some cases, these diverse gene products may be required for biological remodeling during growth and differentiation. For example, the ELAV/Hu proteins are associated with mRNAs that encode early response gene products (often transcription factors), signaling proteins that can activate downstream pathways, cytoskeletal proteins, and glucose transporters that can mobilize cellular energetics. These, and other activated functions, can provide gene products needed to construct new cell structures such as neuronal processes or dendritic spines (13–15). Therefore, if functional linkage is defined as the mobilization of a variety of gene products needed to remodel cell structure or behavior, it is expected that mRNA subsets clustered in mRNP complexes would encode proteins with seemingly diverse properties. How Many Different RNA-Binding Proteins Exist in Model Organisms? The complexity of mRNPs and their potential roles in posttranscriptional regulation can be approximated by considering the number of RNA-binding proteins available for interactions with RNA. As one indication of their abundance, the RRM RNA-binding proteins are one of the largest families of proteins found in the genomic databases (51, 52). Moreover, based on the number of annotated RNA-binding proteins identified in more exhaustively studied model systems such as yeast, nematode, and fly, one can estimate the number of RNA-binding proteins in the human genome. For example, S. cerevisiae has over 6,700 genes with 471 annotated RNA-binding proteins in the databases. However, only 312 of the 471 annotated proteins have been designated to have a role in RNA processing, modification, splicing, or turnover (Yeast Proteome Database at www.proteome.com). These numbers suggest that an estimated range of 5–8% of the genes encode proteins involved in RNA processing. On the other hand, C. elegans and D. melanogaster have fewer annotated RNA-binding proteins in the databases than yeast, but the values range from 2–3% of the total number of known genes. Because the number of human genes is estimated to be 32,000, the estimated number of RNA-binding proteins encoded in the human genome would be as low as 640 (2% of 32,000) or as high as 2,560 (8% of 32,000). The most probable and conservative estimates would place the number of human RNA-binding proteins at 1,500, but their distribution across the global, group-specific, and type-specific classes is unknown. It is also not known which of these proteins might interact with small RNAs, ribosomal RNAs, or mRNAs. Although the significance of this estimated number of human RNA-binding proteins is not clear at the present time, it does suggest that ribonomic analyses have the potential to elucidate many structurally and/or functionally linked mRNA subsets. Global mRNA-binding proteins such as poly (A)-binding protein recognize a large set of mRNAs, whereas the group-specific and type-specific mRNAs encompass subsets or even subsets of subsets of mRNAs. If the technology was available to isolate every cellular mRNP, one should be able to account quantitatively for every mRNA in the cell. The ability to define a ribonome by categorizing all mRNAs into overlapping subsets of mRNP complexes is expected to reveal a structural network for organizing genetic information (1). Furthermore, alterations of the ribonomic network may be characteristic of particular diseases. Likewise, the effects of drugs, chemicals, or toxins, as well as states of differentiation or aging should be reflected in the ribonomic analysis of a cell or tissue. For example, Tenenbaum et al. (1) demonstrated that upon treatment of ELAV/Hu-transfected P19 cells with retinoic acid to induce neuronal differentiation, new mRNAs entered into the ELAV/Hu mRNP complexes. By quantitative analysis, it was evident that these mRNAs were uniquely compartmentalized in ELAV/Hu mRNP complexes and would not all have been identified by using standard transcriptomics. Cell Type-Specific Gene Expression Profiling. One of the potential applications of ribonomics is the recovery of mRNPs that are cell type-specific (1). The ability to recover mRNPs from whole tissue extracts that contain certain RNA-binding proteins only in a single cell type within the tissue should allow recovery of the mRNAs from that single cell type. For example, because ELAV/Hu proteins are expressed in neurons, but are not expressed in glial cells (10), one would expect to find only the neuronal mRNAs in Hu mRNP immunoprecipitates from whole brain extracts. Moreover, ELAV/Hu proteins are ectopically expressed in small cell lung tumor cells and in medulloblastoma cells (10, 11), making it possible to recover mRNAs that are present in the ELAV/Hu mRNPs of the tumor cells by using whole tumor extracts. Therefore, the potential to perform gene expression profiling of single cell types within complex tissues or tumors by using ribonomic approaches could shed light on how cell-cell communication affects the gene expression of neighboring cells. This could provide a means for understanding the crosstalk among cells within tumors, and the effects of antiangiogenesis factors on endothelial cell versus tumor cell gene expression (53). Multiplexing RNA Processing During Growth and Differentiation. The genomics era has brought powerful tools to analyze expressed genes by using a variety of techniques, including sequencing by



7023

hybridization and computational analysis of gene sequences (53, 54). However, it is clear that the protein output of cells and tissues does not always correlate precisely with the mRNA content for a variety of reasons (21, 23, 30, 33, 54, 55). In particular, posttranscriptional regulation and posttranslational modifications can significantly affect the quality and quantity of protein that a given gene will generate. To account for post-transcriptional effects, methods are needed to assess the organization of mRNAs in mRNP complexes and their corresponding functional relationships. Investigations generally examine single gene products for RNA splicing, transport, stability, or translation, but could benefit from the ability to capture mRNP complexes en masse so that the functional complexity and dynamics of posttranscriptional control can be studied. Therefore, the ability to isolate mRNP complexes from cell extracts may provide a means by which to multiplex the analysis of mRNAs as subsets based on structural and/or functional relatedness. Examples of parallel analysis of mRNA targets for RNA-binding proteins have included cDNA libraries prepared from mRNA subsets isolated by iterative in vitro selection using ELAV/Hu RRM proteins (3); and immunoprecipitation of various mRNP complexes followed by mRNA identification using microarray analysis (1). These approaches have wide applicability to other RNA-binding proteins or to other mRNP-associated proteins involved in the processing and localization of mRNA.

Fig. 5. Depiction of the possible organization of mRNAs in mRNP complexes in the cell cytoplasm. Messenger RNP complexes may contain single mRNAs or multiple mRNAs in association with proteins that are either directly or indirectly associated. The Upper Insets depict the total mRNA expressed in the cell (transcriptome) as a microarray and the proteome of the cell is depicted as a two-dimensional gel. The microarrays below the mRNP complexes (Ribonome) are labeled mRNP-1 through mRNP-X and depict multiple mRNAs found in mRNP complexes isolated by using antibodies reactive with mRNA-associated proteins. Microarrays representing nuclear run-on experiments (Left) can be derived by transcription using isolated nuclei (unpublished results) and analysis on Atlas arrays (CLONTECH). As opposed to transcriptomic profiles that are the result of both transcriptional and posttranscriptional contributions and represent accumulated steady-state levels of mRNA, the mRNAs detected by nuclear run-on represent only the transcriptional contribution of genes before the influence of posttranscriptional events in the ribonome. Mammalian cells need a robust and dynamic infrastructure to convey, channel, and rout messenger RNA transcripts with precision. Whether the information in a message is translated immediately, stored for later use, or routed to other locations, the inherent signals in the mRNAs or the corresponding RNA-binding proteins must be elucidated before the mechanisms of information transfer can be understood. Messenger RNP complexes consisting of RNA-binding proteins and/or structurally related mRNAs likely represent nodes of information transfer and accumulation. Decisions as to routing, activation, or disposal of individual transcripts involves the recognition of signals in coding or noncoding portions of each mRNA (24–31). The finding that multiple mRNAs can be identified in mRNPs containing specific proteins such as the ELAV/Hu family (1, 3) suggests that precise routing and information flow, whether individually or in clusters, represents organizational nodes of information transfer. Understanding the network of interacting RNAs and proteins that form a ribonome will require parallel multiplexing of global, group-specific, and type-specific mRNA-binding proteins because current methods of evaluating individual protein-RNA interactions are too limiting. Therefore, the dynamic interactions between RNAs and proteins involved in splicing, transport, and translation need to be evaluated in parallel and en masse so that regulatory loops and feedback mechanisms can be better understood. How might one distinguish transcriptional from posttranscriptional regulation of multiple genes simultaneously? Fig. 5 illustrates our approach to multiplexing posttranscriptional events by categorizing mRNAs associated with mRNP complexes (1) and comparing their levels with transcriptional output. The conventional approach to distinguishing transcriptional contributions from posttranscriptional regulation of single gene transcripts often involves nuclear run-on experiments. Our laboratory has used nuclear run-on analysis en masse, employing P19 embryonic carcinoma cells, HeLa cells, and EL4 thymoma cells (C.Carson and J.D.K., unpublished results). By radiolabeling elongating transcripts in nuclear extracts, the transcriptionally active genes were identified on microarrays and compared with global mRNAs detected in the whole cell population. This approach allows a large number of candidate genes to be analyzed in parallel for



7024

posttranscriptional effects such as mRNA stability by distinguishing those genes exclusively regulated at the level of transcription. Similar approaches for multiplexing splicing reactions and translation are imaginable, and will be necessary to fully understand the posttranscriptional network operating system. One can imagine databases in which the functional linkages between multiple mRNAs can be accessed based on their membership in one or more mRNP complexes. For example, it should be possible to account for 100% of any given mRNA within a cell whether it is a member of a structurally or functionally related group of mRNAs, or a member of a small subset of a larger set of physically clustered mRNAs. In the future, when all of the RNA-binding proteins associated with every mRNA are known, it should be possible to describe, and ultimately simulate the organization and flow of genetic information within cells. Thus, by identifying mRNAs that are members of a physically clustered mRNP subset, the functions of proteins encoded by the mRNAs in the subset may become readily apparent through “guilt by association.” As a specific case in point, growth regulatory proteins like those encoded by the mRNAs associated with ELAV/Hu proteins are believed to have related functional properties (1). In addition to the functions of the encoded proteins, mRNAs may be clustered in vivo to optimize regulatory control of their expression, including mRNA stability, translation, and localization (1–3). Ribonomic databases may be constructed based on physical clustering of mRNAs and the functional relationships among their protein products. Such databases would allow tracking of mRNAs although their unique nodes of information management and transfer. Therefore, being able to organize each mRNP cluster into a relational database that accounts for the functional networking among its mRNAs and their protein products may offer insights into functional genomics. A challenge for ribonomics will be to account for a full set of cellular transcripts, and to assess the dynamics of activation, repression, and product feedback that are inherent in an mRNP network. Functional perturbations by mutation, antisense expression, RNAi, or small molecules would be expected to alter the mRNP ribonomic network with a discernable outcome in the composition of the proteome. Like genomics and proteomics, ribonomics will require sophisticated computational systems to simulate the cellular dynamics of the posttranscriptional infrastructure during development. Indeed, this is a problem suited for the complexity sciences. Many thanks to Craig Carson and Scott Tenenbaum for intellectual input and help in the preparation of figures. 1. Tenenbaum, S.A., Carson, C.C, Lager, P.J. & Keene, J.D. (2000) Proc. Natl. Acad. Sci. USA 97, 14085–14090. 2. Levine, T.D., Gao, F., King, P.H., Andrews, L.G. & Keene, J.D. (1993) Mol. Cell. Biol. 13, 3494–3504. 3. Gao, F.B., Carson, C.C., Levine, T. & Keene, J.D. (1994) Proc. Natl. Acad. Sci. USA 91, 11207–11211. 4. Liu, J., Dalmau, J., Szabo, A., Rosenfeld, M., Huber, J. & Furneaux, H. (1995) Neurology 45, 544–550. 5. Myer, V.E., Fan, X.C. & Steitz, J.A. (1997) EMBO J. 16, 2130–2139. 6. Fan, X.C., Myer, V.E. & Steitz, J.A. (1997) Genes Dev. 11, 2557–2568. 7. Keene, J.D. (1999) Proc. Natl. Acad. Sci. USA 96, 5–7. 8. Brennan, C.M. & Steitz, J.A. (2000) Cell Mol. Life Sci., in press. 9. Antic, D. & Keene, J.D. (1997) Am. J.Hum. Genet. 61, 273–278. 10. Gao, F.B. & Keene, J.D. (1996) J. Cell Sci. 109, 579–589. 11. Antic, D. & Keene, J.D. (1998) J. Cell Sci. 111, 183–19J. 12. Jain, R.G., Andrews, L.G., McGowan, K.M., Pekala, P.H. & Keene, J.D. (1997) Mol. Cell. Biol. 17, 954–962. 13. Antic, D., Lu, N. & Keene, J.D. (1999) Genes Dev. 13, 449–461. 14. Wakamatsu, Y. & Weston, J.A. (1997) Development 124, 3449–3460. 15. Chung, S., Eckrich, M., Perrone-Bizzozero, N., Kohn, D.T. & Furneaux, H. (1997) J. Biol. Chem. 272, 6593–6598. 16. Levy, N.S., Chung, S., Furneaux, H. & Levy, A.P. (1998) J. Biol. Chem. 273, 6417–6423. 17. Peng, S.S., Chen, C.Y., Xu, N. & Shyu, A.B. (1998) EMBO J. 17, 3461–3470. 18. Fan, X.C. & Steitz, J.A. (1998) EMBO J. 17, 3448–3460. 19. Ford, L.P., Watson, J., Keene, J.D. & Wilusz, J. (1999) Genes Dev. 13, 188–201. 20. Aranda-Abreu, G.E., Behar, L., Chung, S., Furneaux, H. & Ginzburg, I. (1999) J. Neurosci. 19, 6907–6917. 21. Lockhart, D.J. & Winzeler, E.A. (2000) Nature (London) 405, 827–836. 22. Velculescu, V.E., Zhang, L., Zhou, W., Vogelstein, J. & Kinzler, K.W. (1997) Cell 88, 243–251. 23. Eisen, M.B., Spellman, P.T., Brown, P.O. & Botstein, D. (1998) Proc. Natl. Acad. Sci. USA 95, 14863–14868. 24. Schiavi, S.C., Belasco, J.G. & Greenberg, M.E. (1992) Biochim. Biophys. Acta 1114, 95–106. 25. Ross, J. (1995) Microbiol. Rev. 59, 423–450. 26. Chen, C.Y. & Shyu, A.B. (1995) Trends Biochem. Sci. 20, 465–470. 27. Pinol-Roma, S. & Dreyfuss, G. (1993) Trends Cell Biol. 3, 151–155. 28. St. Johnston, D. (1995) Cell 81, 161–170. 29. Dreyfuss, G., Matunis, M.J., Pinol-Roma, S. & Burd, C.G. (1993) Annu. Rev. Biochem. 62, 289–321. 30. Richter, J.D. (1997) mRNA Formation and Function (Academic, New York). 31. Haynes, S.R. (1999) RNA-Protein Interactions Protocols (Humana, Totowa, NJ). 32. Savant-Bhonsale, S. & Cleveland, D.W. (1992) Genes Dev. 6, 1927–1939. 33. Zong, Q., Schummer, M., Hood, L. & Morris, D.R. (1999) Proc. Natl. Acad. Sci. USA 96, 10632–10636. 34. Gingras, A.-C., Raught, B. & Sonenberg, N. (1999) Annu. Rev. Biochem. 68, 913–963. 35. Gale, M., Tan, S.-L. & Katze, M.G. (2000) Micro. Mol. Biol. Rev. 64, 239–280. 36. Pleasure, S.J. & Lee, V.M. (1993) J. Neurosci. Res. 35, 585–602. 37. Steward, O., Wallace, C.S., Lyford, G.L. & Worley, P.F. (1998) Neuron 21, 741–751. 38. Steward, O. & Banker, G.A. (1992) Trends Neurosci. 15, 180–186. 39. Crino, P., Khodakhah, K., Becker, K., Ginsberg, S., Hemby, S. & Eberwine, J. (1998) Proc. Natl. Acad. Sci. USA 95, 2313–2318. 40. Mayford, M., Baranes, D., Podsypanina, K. & Kandel, E.R. (1996) Proc. Natl. Acad. Sci. USA 93, 13250–13255. 41. Schuman, E. (1999) Neuron 23, 645–648. 42. Bassell, G.J., Oleynikov, Y. & Singer, R.H. (1999) FASEB J. 13, 447–454. 43. King, P.H., Levine, T.D., Fremeau, R.T. & Keene, J.D. (1994) J. Neurosci. 14, 1943–1952. 44. Lindstren, T., June, C.H., Ledbetter, J.A., Stella, G. & Thompson, C.R. (1989) Science 244, 339–343. 45. Atasoy, U., Watson, J., Patel, D. & Keene, J.D. (1998) J. Cell Sci. 111, 3145–3156. 46. Zhuang, J.-Y., Chan, E.K.E., Peng, X.-X. & Tan, E.M. (1999) J. Exp. Med. 189, 1101–1110. 47. Albright, T.D., Jessell, T.M., Kandel, E.R. & Posner, M.I. (2000) Cell 100, S1-S55. 48. Rouault, T.A. & Klausner, R.D. (1997) Curr. Top. Cell. Regul. 35, 1–19. 49. Wang, Z.F., Ingledue, T.C., Dominski, Z., Sanchez, R. & Marzluff, W.F. (1999) Mol. Cell. Biol. 19, 835–845. 50. Whitfield, M.L., Zheng, L.X., Baldwin, A., Ohta, T., Hurt, M.M. & Marzluff, W.F. (2000) Mol. Cell. Biol. 20, 4188–4198. 51. Kenan, D.J., Query, C.C. & Keene, J.D. (1991) Trends Biochem. Sci. 16, 214–220. 52. Burd, C.G. & Dreyfuss, G. (1994) Science 265, 615–621. 53. Hanahan, D. & Weinberg, R.A. (2000) Cell 100, 57–70. 54. Lander, E.S. (1999) Nat. Genet. 21, 3–4. 55. Brown, P.O. & Botstein, D. (1999) Nat. Genet. 21, 38–41.


SPATIAL AND TEMPORAL CONTROL OF RNA STABILITY

7025

Colloquium Spatial and temporal control of RNA stability

Arash Bashirullah*†, Ramona L.Cooperstock*‡, and Howard D.Lipshitz*‡§ in Developmental Biology, Research Institute, The Hospital for Sick Children, and ‡Department of Molecular and Medical Genetics, University of Toronto, 555 University Avenue, Toronto, ON M5G 1×8, Canada Maternally encoded RNAs and proteins program the early development of all animals. A subset of the maternal transcripts is eliminated from the embryo before the midblastula transition. In certain cases, transcripts are protected from degradation in a subregion of the embryonic cytoplasm, thus resulting in transcript localization. Maternal factors are sufficient for both the degradation and protection components of transcript localization. Cisacting elements in the RNAs convert transcripts progressively (i) from inherently stable to unstable and (ii) from uniformly degraded to locally protected. Similar mechanisms are likely to act later in development to restrict certain classes of transcripts to particular cell types within somatic cell lineages. Functions of transcript degradation and protection are discussed. The early development of all animals is programmed by maternally synthesized RNAs and proteins that are loaded into the developing oocyte by the mother (reviewed in ref. 1). The volumes of mature oocytes from mammals, amphibians, insects, and sea urchins range over several orders of magnitude, as do their total mass of RNA. However, studies several decades ago showed that the complexity of maternal RNA populations in the oocytes of different species—a measure of the number of different classes of transcripts present—varies at most a few fold. For example, the RNA complexity in the mature Drosophila oocyte is 1.2×107 nt (2), whereas that in sea urchin or Xenopus oocytes is 4×107 nt (3, 4). In Drosophila, the measured RNA complexity represents 5,000 different classes of transcripts of average length 2.5 kb. With the completion of the Drosophila genome sequence (5–8), we now know that this complexity represents the products of more than a third of all of the genes in the fly. At the midblastula transition (MBT), control of development passes from maternally encoded molecules to proteins synthesized from zygotically transcribed mRNAs. The maternal transcripts that are present in the early embryo can be subdivided into two classes according to whether they are destroyed before the MBT or are stable through this transition (reviewed in ref. 1). It is widely presumed—although there is little evidence that addresses this presumption—that degradation of certain maternal mRNAs is necessary so that the zygotically synthesized transcripts and proteins can take control of development at the MBT. To date, quantitative analyses of individual transcripts have been very limited. In Drosophila, for example, the ribosomal protein-encoding mRNA, rpA1, is stable through the MBT whereas nanos, string, Pgc, and Hsp83 transcripts are unstable (9). Transcript stability is regulated in space as well as in time. In the early embryo of Drosophila, spatial control of transcript stability functions as a novel RNA localization mechanism (9, 10). For example, string transcripts are degraded throughout the embryo before the MBT. In contrast nanos, Hsp83, and Pgc transcripts are degraded everywhere except in the posterior polar plasm and the pole cells (9, 11, 12). Here the focus is on the spatial and temporal control of RNA stability in Drosophila. The default state of maternal transcripts in the early embryo is stability; specific cis-acting sequences tag certain classes of transcripts for degradation. Further, if a transcript is targeted for degradation, then the default is generalized degradation throughout the embryo. Localization of a subset of the unstable classes of transcripts is achieved through cis-acting elements that allow protection of these RNAs from degradation in particular cytoplasmic domains. The genetic requirements for transcript degradation and protection are discussed. Preliminary evidence is presented that transcript localization to mother versus daughter cells in the neuronal cell lineage also is accomplished through a degradation-protection mechanism. The mechanisms that regulate transcript stability are likely to be evolutionarily conserved in all metazoa. Possible functions of temporal and spatial control of transcript stability are considered. *Program

TEMPORAL CONTROL OF MATERNAL RNA STABILITY IN DROSOPHILA Two transcript degradation pathways function together to eliminate maternal transcripts from the early Drosophila embryo (9). One of these pathways begins to function at or shortly after egg activation, independent of fertilization (Fig. 1) (9). This “maternal” pathway is active in unfertilized eggs and thus must be exclusively maternally encoded because there is no “zygotic” transcription in this situation. For example, degradation of maternal Hsp83, string, nanos, and Pgc transcripts occurs in activated unfertilized eggs (in contrast to rpA1 transcripts, which are stable; Fig. 1) (9). A second transcript degradation pathway, which has been termed the zygotic pathway, requires fertilization and becomes active 2 h after this event (9). At the present time there is no definitive evidence that addresses whether the zygotic pathway in fact requires zygotic transcription either to produce the degradation machinery or to activate it. Reinterpretation of drug inhibition experiments carried out before discovery of the two pathways suggests that zygotic transcription is indeed required; for example, inhibition of zygotic transcription using α-amanitin before 1 h after fertilization partially stabilizes transcripts such as string (9, 13). The time course of degradation is thus biphasic: the degradation rate is significantly slower before 2 h after fertilization (when only the maternal pathway is active; during this period the half-life of transcripts such as Hsp83 is 75 min) than after 2 h (when both the maternal and zygotic pathways function; the half-life of Hsp83 transcripts is reduced to 25 min) (9). Stability is the default state of transcripts in the early embryo of Drosophila. Unstable transcripts contain cis-acting sequences that target them for degradation. For example, it has been possible to define small elements (100 nt) in the 3 untrans

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: MBT, midblastula transition; UTR, untranslated region; HDE, Hsp83 degradation element. †Present address: Department of Human Genetics, University of Utah, 15 North 2030 East, Salt Lake City, UT 84112–5331. §To whom reprint requests should be addressed. E-mail: [email protected].



7026

lated region (UTR) of Hsp83 and nanos that, when deleted, result in substantial stabilization of corresponding transgenic transcripts in both unfertilized and fertilized eggs (Fig. 2) (9, 14-16). It was comparison of the stability of the cis element-deleted transcripts in unfertilized versus fertilized eggs that led to the discovery of the zygotic degradation pathway (9): transcripts deleted for an element required for maternal degradation are fully stabilized in unfertilized eggs (Fig. 2 A) but are destabilized starting 2 h after fertilization in developing embryos (Fig. 2 B). Thus, there must be additional cis-acting elements that are still present in these transcripts and that mediate zygotic degradation.

Fig. 1. Time course of maternal transcript degradation in activated, unfertilized eggs. (A) The same Northern blot probed for rpA1 (stable) and string, Hsp83, or nanos (unstable) transcripts. (B) Quantitative analysis of the time course of Hsp83 transcript degradation. The points represent the ratio of Hsp83 transcripts to stable rpA1 transcripts relative to the initial (0-0.5 h) concentration. It can be seen that more than 95% of the Hsp83 transcripts have disappeared by 3.0 to 3.5 h after egg activation. Data from two independent experiments are presented. Half-hour time windows are shown. See ref. 9 for details. Spatial Control of Maternal RNA Stability in Drosophila Although unstable maternal transcripts such as string and Hsp70 are eliminated throughout the egg or early embryo (Fig. 3 A and B), unstable transcripts such as Pgc, Hsp83, and nanos are eliminated from the bulk cytoplasm of the egg or embryo but remain stable at the posterior (Fig. 3 C-F) (9). Uniform instability is the default state for the unstable classes of transcripts: if the Hsp83 3 UTR is replaced with a 3 UTR from a uniformly degraded transcript (e.g., Hsp70) or if a cisacting "protection" element is deleted (see below), then the resulting transgenic Hsp83 transcripts are degraded throughout the embryo (see Fig. 5 A and B) (9). This experiment proves that the endogenous Hsp83, nanos, and Pgc transcripts that remain at the posterior of the egg and early embryo are protected from degradation in that region of the cytoplasm (i.e., the degradation machinery is

Fig. 2. Removal of a maternal Hsp83 degradation element stabilizes transgenic transcripts in unfertilized (A) but not fertilized (B) eggs. Northern blots are shown that were simultaneously probed for (i) endogenous Hsp83 transcripts; (ii) transgenic reporter transcripts carrying the Hsp83 5 UTR, the first 111 codons of the Hsp83 ORF, an Escherichia coli ß-galactosidase RNA tag, and the Hsp83 3' UTR deleted for a 97-nt element referred to as the Hsp83 degradation element (HDE) (for a detailed description of this transgene see ref. 9); (iii) endogenous rpA1 transcripts. On both blots it can be seen that endogenous Hsp83 transcripts are unstable and endogenous rpA1 transcripts are stable. However, although transgenic ∆HDE transcripts are stable in unfertilized eggs (1A), they are degraded commencing 2 h after fertilization in developing embryos (B). Half-hour time windows after egg activation or fertilization are shown. See ref. 9 for details.

Fig. 3. Certain classes of maternal transcripts are degraded throughout the cytoplasm of activated, unfertilized eggs whereas others are protected from degradation in the posterior polar plasm, string transcripts are initially present throughout the egg (A) and are subsequently degraded (B). In contrast, whereas Hsp83 (C) and nanos (E) transcripts are initially present in both the posterior polar plasm and the presumptive somatic region (C and E), degradation is limited to the somatic region whereas transcripts are protected from degradation in the posterior polar plasm (D and F). (A, C, and E) One to 2 h after egg activation; (B, D, and F) 3-4 h after egg activation. Whole-mount RNA in situ hybridizations are shown, with anterior to the left and dorsal toward the top of the page. See ref. 9 for details.



7027

present in the posterior but the transcripts are masked from the machinery).

Fig. 4. The posterior polar plasm is necessary and sufficient for transcript protection. (A) Posterior protection of maternal Hsp83 transcripts fails in an embryo from a cappuccino mutant female because posterior polar plasm is not assembled. The anterior expression of Hsp83 is zygotic and serves as an internal control for the in situ hybridization, (B) Protection of Hsp83 transcripts occurs at both poles of an unfertilized egg derived from a female carrying an osk-bcd 3 UTR transgene. Posterior polar plasm is ectopically assembled at the anterior pole of such embryos and is sufficient for transcript protection. (A) Stage 5 embryo, about 2.5 h after fertilization; (B) unfertilized egg 3–4 h after egg activation. Whole-mount RNA in situ hybridizations are shown, with anterior to the left and dorsal toward the top of the page. See ref. 18 for details. Cis-acting protection elements can be mapped within the 3 UTRs of these localized transcripts; deletion of such an element converts an RNA from one localized by degradation/protection into one that is eliminated from the entire embryo (9). These mapping experiments also demonstrate that degradation elements and protection elements map to distinct regions within the 3 UTR. The fact that transcript protection occurs in activated, unfertilized eggs (9), together with the absence of zygotic transcription in the pole plasm and pole cells of the early embryo (17), indicates that transcript protection is carried out exclusively by maternally encoded molecules. GENETIC CONTROL OF MATERNAL RNA STABILITY IN DROSOPHILA Given the fact that maternally encoded molecules are sufficient for both degradation and protection of transcripts in the early embryo, genetic screens initially focused on identification of maternal effect degradation or protection mutants (9). To date, several degradation mutants have been identified, each of which is defective in aspects of egg activation per se. These mutants confirm the importance of egg activation as a prerequisite for transcript degradation (i.e., if the egg is not activated normally, transcript degradation is not triggered). However, they also emphasize the need for larger scale maternal effect screens as well as screens that do not focus exclusively on maternal effect mutants, if mutations in the degradation machinery itself are to be obtained. Transcript protection at the posterior requires assembly of the posterior polar plasm and its constituent polar granules, which are a crucial component of this specialized cytoplasmic domain (9, 18). Any mutant that eliminates the polar granules results in failure of transcript protection at the posterior (Fig. 4A) (18). Similarly, ectopic assembly of polar plasm and polar granules at the anterior of the egg or early embryo results in ectopic protection of transcripts at the anterior (Fig. 4B) (18). Whether RNA-binding proteins that are known to reside in, and be required for assembly of, the posterior polar plasm (e.g., Staufen, Vasa) interact directly with protection elements in the localized RNAs is not yet known. It is, however, clear that certain RNA-binding proteins that reside in the polar plasm but are not required for polar granule assembly (e.g., Nanos) also are not required for protection, because protection occurs normally in mutants that eliminate these proteins (18). CONTROL OF RNA STABILITY IN A SOMATIC CELL LINEAGE Evidence is beginning to accumulate that transcript localization mechanisms are conserved among different cell types. For example, it has been shown that, in Drosophila, the Staufen RNA-binding protein functions in transcript localization in the egg as well as in neuroblasts (reviewed in ref. 10). Of interest here is the possibility that the degradation-protection mechanisms that operate to localize certain maternally encoded transcripts to the germ plasm and germ cells of the early embryo also act at other stages and in other cell types. It was therefore particularly tantalizing to find that zygotically synthesized Hsp83 transcripts accumulate at high levels in the embryonic neuroblasts, stem cells that divide asymmetrically to give rise to the central nervous system (Fig. 5). Strikingly, Hsp83 transcripts are absent from the neuroblasts’ small daughter cells, the ganglion mother cells (Fig. 5 C). This finding suggested that zygotic Hsp83 transcripts might be protected from degradation in the neuroblasts but not in the ganglion mother cells.

Fig. 5. Maternally synthesized Hsp83 transcripts are protected from degradation in the pole cells (A and B) whereas zygotically synthesized Hsp83 transcripts are protected from degradation in neuroblasts (NB) but not their daughter cells, the ganglion mother cells (GMC) (C and D). (A and C) Endogenous Hsp83 transcripts accumulate in the pole cells but are degraded in the somatic cells of a stage 5 embryo (A, maternal transcripts) and in the NBs but not their daughter cells, the GMCs, of a stage 10 embryo (C, zygotic transcripts; the arrowhead points to a GMC). (D) Transgenic transcripts carrying a degradation element but not a protection element are degraded in both the somatic cells and the pole cells of a stage 5 embryo (B, maternal transcripts). Such transcripts are expressed in the NBs of a stage 10 embryo but are degraded in both the NBs and GMCs. The transgenic transcripts comprise the Hsp83 5' UTR, the first 111 codons of the Hsp83 ORF, an E.coli ß-galactosidase RNA tag, and the Hsp70 3 UTR (for a detailed description of this transgene see ref. 9). The Hsp70 3 UTR lacks a protection element but carries a degradation element. Whole-mount RNA in situ hybridizations are shown. To address this possibility, zygotically synthesized transgenic transcripts with a 3 UTR carrying a degradation element but not a protection element were examined in the neuroblast lineage. Transcripts lacking the protection element are expressed in neuroblasts but rapidly disappear, suggesting that removal of this element results in transcript degradation in both neuroblasts and ganglion mother cells rather than only in the ganglion mother cells (Fig. 5 D). This finding suggests that the same degradation-protection machinery acts on maternal transcripts in the early embryo as well as on zygotically synthesized transcripts later in development. In each case, the result is to restrict transcripts to one cell type: in the early embryo, to primordial germ cells but not somatic cells; in the neuronal lineage, to stem cells (neuroblasts) but not their daughter cells (ganglion mother cells).



7028

Fig. 6. Transcript degradation and protection are evolutionary conserved processes. Localization of maternal Hsp83 transcripts to the pole cells of D. virilis occurs by degradation and protection as in D. melanogaster. (A) Maternal Hsp83 transcripts are initially uniformly distributed throughout a syncytial stage D. virilis embryo. (B) Subsequently, these transcripts are degraded in the somatic region but are protected from degradation in the pole cells that bud from the posterior (arrowhead). Note that Bicoiddependent zygotic expression of Hsp83 that occurs in the anterior of D. melanogaster embryos (see Fig. 4 A and ref. 18) does not occur in D. virilis. (C) In vitro-transcribed D. melanogaster Hsp83 3 UTR transcripts that carry the HDE(+HDE) are highly unstable when injected into X. laevis stage 6 oocytes (the time points are hours after injection). (D) In contrast, Hsp83 3 UTR transcripts lacking the HDE (∆HDE) are stable for at least 24 h after injection. (A and B) Whole-mount RNA in situ hybridizations are shown, with anterior to the left and dorsal toward the top of the page. (C and D) Blots are shown of digoxigenin-labeled transcripts recovered the specified number of hr after injection into Xenopus oocytes. See ref. 9 for details. EVOLUTIONARY CONSERVATION OF DEGRADATION-PROTECTION MECHANISMS To determine whether transcript localization by degradationprotection is conserved between distant Drosophila species, Hsp83 transcripts were examined in early embryos of Drosophila virilis, which has diverged 60 million years from D. melanogaster (19). Both components of the localization mechanism—degradation and protection—are conserved (Fig. 6 A and B). When in vitro-synthesized transcripts comprising the D. melanogaster Hsp83 3 UTR (i.e., carrying the degradation element) are injected into Xenopus laevis stage 6 oocytes or early embryos, the transcripts are unstable (Fig. 6 C and D) (9). Strikingly, deletion of the degradation element increases the half-life of injected transcripts approximately an order of magnitude (9). Thus, Drosophila cis-acting sequences can be recognized by the Xenopus trans-acting machinery, which suggests that the fundamental maternal transcript degradation machinery is conserved throughout the metazoa. FUNCTIONS OF TRANSCRIPT DEGRADATION AND PROTECTION As discussed above, maternal transcript degradation in the early embryo has been presumed to be necessary for the passage of developmental control to the zygotic genome. However, there have been few experiments that address whether this is in fact the case. Perhaps the best data derives from dosage analyses in which the concentration of maternal string transcripts was either halved or doubled (13) (string encodes the Drosophila Cdc25 cell cycle regulator). The transition from uniform maternally regulated to spatially patterned zygotically regulated cell divisions was delayed (double dose) or induced prematurely (half dose). Now that the pathways of transcript instability in the early Drosophila embryo have been defined, it should be possible to delete instability elements from string transcripts to ask whether stabilization of maternal string mRNA results in delay of the cell cycle transition at the MBT. In contrast to the poorly defined role of maternal RNA degradation in the early embryo, the functions of posterior protection are better understood. For example, it is known that posterior protection of nanos and Pgc transcripts is crucial for the biological roles of these RNAs in germ-cell differentiation; elimination of posterior localization of these transcripts results in abnormal differentiation of the germ cells (12, 20, 21). The role of posterior protection of Hsp83 transcripts has not yet been defined. CONCLUSIONS The analyses summarized above have shown that transcript stability is exquisitely regulated both in time and in space in the early Drosophila embryo as well as at other developmental stages and in other cell types. Transcript localization by degradation-protection represents a distinct mechanism from that involving directed cytoplasmic transport via cytoskeletal motors, although these two mechanisms are not mutually exclusive (reviewed in refs. 10 and 22). Further analyses of degradation and protection using the combination of genetic and biochemical methods available in Drosophila are expected to lead to general insights into the mechanisms and functions of these processes during animal development. R.L.C. has been supported in part by a Medical Research Council of Canada Graduate Scholarship and a University of Toronto Open Scholarship. Our research on RNA localization mechanisms is funded by an operating grant to H.D.L. from the Canadian Institutes of Health Research (formerly the Medical Research Council of Canada). 1. Davidson, E.H. (1986) Gene Activity in Early Development (Academic, Orlando, FL). 2. Hough-Evans, B.R., Jacobs-Lorena, M., Cummings, M.R., Britten, R.J. & Davidson, E.H. (1980) Genetics 95, 81–94. 3. Davidson, E.H. & Hough, B.R. (1971) J. Mol. Biol. 56, 491–506. 4. Hough-Evans, B.R., Wold, B.J., Ernst, S.G., Britten, R.J. & Davidson, E.H. (1977) Dev. Biol. 60, 258–277. 5. Rubin, G.M., Hong, L., Brokstein, P., Evans-Holm, M., Frise, E., Stapleton, M. & Harvey, D.A. (2000) Science 287, 2222–2224. 6. Adams, M.D., Celniker, S.E., Holt, R.A., Evans, C.A., Gocayne, J.D., Amanatides, P.G., Scherer, S.E., Li, P.W., Hoskins, R.A., Galle, R.F., et al. (2000) Science 287, 2185–2195. 7. Myers, E.W., Sutton, G.G., Delcher, A.L., Dew, I.M., Fasulo, D.P., Flanigan, M.J., Kravitz, S.A., Mobarry, C.M., Reinert, K.H., Remington, K.A., et al. (2000) Science 287, 2196–2204. 8. Rubin, G.M., Yandell, M.D., Wortman, J.R., Gabor Miklos, G.L., Nelson, C.R. , Hariharan, I.K., Fortini, M.E., Li, P.W., Apweiler, R., Fleischmann, W., et al. (2000) Science 287, 2204–2215. 9. Bashirullah, A., Halsell, S.R., Cooperstock, R.L., Kloc, M., Karaiskakis, A., Fisher, W.W., Fu, W., Hamilton, J.K., Etkin, L.D. & Lipshitz, H.D. (1999) EMBO J. 18, 2610–2620. 10. Lipshitz, H.D. & Smibert, C.A. (2000) Curr. Opin. Genet. Dev. 10, 476–488. 11. Bergsten, S.E. & Gavis, E.R. (1999) Development (Cambridge, U.K.) 126, 659–669. 12. Nakamura, A., Amikura, R., Mukai, M., Kobayashi, S. & Lasko, P.F. (1996) Science 274, 2075–2079. 13. Edgar, B.A. & Datar, S.A. (1996) Genes Dev. 10, 1966–1977. 14. Gavis, E.R., Lunsford, L., Bergsten, S.E. & Lehmann, R. (1996) Development (Cambridge, U.K.) 122, 2791–2800. 15. Dahanukar, A. & Wharton, R.P. (1996) Genes Dev. 10, 2610–2620. 16. Smibert, C.A., Wilson, J.E., Kerr, K. & Macdonald, P.M. (1996) Genes Dev. 10, 2600–2609. 17. Van Doren, M., Williamson, A.L. & Lehmann, R. (1998) Curr. Biol. 8, 243–246. 18. Ding, D, Parkhurst, S.M., Halsell, S.R. & Lipshitz, H.D. (1993) Mol. Cell. Biol. 13, 3773–3781. 19. Beverley, S.M. & Wilson, A.C. (1984) J. Mol. Evol 21, 1–13. 20. Forbes, A. & Lehmann, R. (1998) Development (Cambridge, U.K.) 125, 679–690. 21. Kobayashi, S., Yamada, M., Asaoka, M. & Kitamura, T. (1996) Nature (London) 380, 708–711. 22. Bashirullah, A., Cooperstock, R.L. & Lipshitz, H.D. (1998) Annu. Rev. Biochem. 67, 335–394.


MOLECULAR MECHANISMS OF TRANSLATION INITIATION IN EUKARYOTES

7029

Colloquium Molecular mechanisms of translation initiation in eukaryotes

Tatyana V.Pestova*†, Victoria G.Kolupaeva*, Ivan B.Lomakin*, Evgeny V.Pilipenko*‡§, Ivan N.Shatsky†, Vadim I.Agol†‡, and Christopher U.T.Hellen*¶ *Department of Microbiology and Immunology, State University of New York Health Science Center at Brooklyn, Brooklyn, NY 11203; †A.N.Belozersky Institute of Physico-chemical Biology, Moscow State University, Moscow 119899, Russia; and ‡Institute of Poliomyelitis and Viral Encephalitides, Russian Academy of Medical Sciences, Moscow Region 142782, Russia Translation initiation is a complex process in which initiator tRNA, 40S, and 605 ribosomal subunits are assembled by eukaryotic initiation factors (eIFs) into an 80S ribosome at the initiation codon of mRNA. The cap-binding complex eIF4F and the factors eIF4A and eIF4B are required for binding of 43S complexes (comprising a 40S subunit, eIF2/GTP/Met-tRNAi and eIF3) to the 5 end of capped mRNA but are not sufficient to promote ribosomal scanning to the initiation codon. eIF1A enhances the ability of eIF1 to dissociate aberrantly assembled complexes from mRNA, and these factors synergistically mediate 48S complex assembly at the initiation codon. Joining of 48S complexes to 60S subunits to form 80S ribosomes requires eIF5B, which has an essential ribosome-dependent GTPase activity and hydrolysis of eIF2-bound GTP induced by eIF5. Initiation on a few mRNAs is capindependent and occurs instead by internal ribosomal entry. Encephalomyocarditis virus (EMCV) and hepatitis C virus epitomize distinct mechanisms of internal ribosomal entry site (IRES)-mediated initiation. The eIF4A and eIF4G subunits of eIF4F bind immediately upstream of the EMCV initiation codon and promote binding of 43S complexes. EMCV initiation does not involve scanning and does not require eIF1, eIF1A, and the eIF4E subunit of eIF4F. Initiation on some EMCV-like IRESs requires additional noncanonical initiation factors, which alter IRES conformation and promote binding of eIF4A/4G. Initiation on the hepatitis C virus IRES is even simpler: 43S complexes containing only eIF2 and eIF3 bind directly to the initiation codon as a result of specific interaction of the IRES and the 40S subunit. Translation of mRNA into protein begins after assembly of initiator tRNA (Met-tRNAi), mRNA, and separated 40S and 60S ribosomal subunits into an 80S ribosome in which MettRNAi is positioned in the ribosomal P site at the initiation codon. The complex initiation process that leads to 80S ribosome formation consists of several linked stages that are mediated by eukaryotic initiation factors. These stages are: (i) Selection of initiator tRNA from the pool of elongator tRNAs by eukaryotic initiation factor (eIF)2 and binding of an eIF2/ GTP/Met-tRNAi ternary complex and other eIFs to the 40S subunit to form a 43S preinitiation complex. (ii) Binding of the 43S complex to mRNA, which in most instances occurs by a mechanism that involves initial recognition of the m7G cap at the mRNA 5-terminus by the eIF4E (cap-binding) subunit of eIF4F. Ribosomes bind to a subset of cellular and viral mRNAs as a result of cap- and end-independent internal ribosomal entry. (iii) Movement of the mRNA-bound ribosomal complex along the 5 nontranslated region (5NTR) from its initial binding site to the initiation codon to form a 48S initiation complex in which the initiation codon is base paired to the anticodon of initiator tRNA. (iv) Displacement of factors from the 48S complex and joining of the 60S subunit to form an 80S ribosome, leaving Met-tRNAi in the ribosomal P site. Research in our laboratory has addressed the molecular mechanisms of these different stages in translation initiation and the means by which they are bypassed during initiation by internal ribosomal entry. We have reconstituted each of these stages in vitro using purified translation components to identify the minimum set of eIFs that is required for each stage and to provide a framework for more detailed mechanistic analysis. Factor Requirements for Ribosomal Attachment and Scanning of 43S Ribosomal Complexes on ß-Globin mRNA. The initiation codon of a eukaryotic mRNA is normally the first AUG triplet downstream of the 5-terminal cap and is usually separated from it by 50–100 nt. After cap-mediated attachment to mRNA, a 43S complex is thought to scan downstream from the 5-end until it encounters the initiation codon. We used native capped ß-globin mRNA as a model in in vitro reconstitution experiments to address three basic questions. (i) Which eIFs are required for a 43S complex to bind capped mRNA? (ii) Which eIFs are required for the bound complex to move downstream to the initiation codon? (iii) How does the scanning 43S complex recognize and reject mismatched interactions between the Met-tRNAi anticodon, and triplets in the 5 NTR until the correct initiation codon is reached and recognized? In these experiments, the position of the leading edge of bound ribosomal complexes on mRNA was mapped by primer extension inhibition (“toeprinting”). The estimated length of the mRNA-binding cleft in 40S subunits is 30nt, and 48S complexes usually yield toeprints at positions +15—+17 downstream of the A of the initiation codon. Ribosomal binding at the 5-end of the mRNA required eIF3, the eIF2/GTP/Met-tRNAi complex, ATP, and the eIF4F cap-binding complex, and was enhanced by eIF4B (1). eIF4F is a heterotrimeric factor, and its eIF4A (ATP-dependent RNA helicase) and eIF4E subunits and the eIF4G550–1090 fragment of its 1,560-amino acid eIF4G subunit constitute the core of eIF4F

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: eIF, eukaryotic initiation factor; 5 NTR, 5 nontranslated region; IRES, internal ribosomal entry site; EMCV, encephalomyocarditis virus; FMDV, foot-and-mouth disease virus; TMEV, Theiler’s murine encephalitis virus; BVDV, bovine viral diarrhea virus; CSFV, classical swine fever virus; HCV, hepatitis C virus; ITAF, IRES transactivating factors; PTB, pyrimidine tractbinding protein; RRL, rabbit reticulocyte lysate; GMP-PNP, guanosine 5-[ß, γ-imido] triphosphate. §Present address: Department of Neurology, University of Chicago Medical Center, Chicago, IL 60637. ¶To whom reprint requests should be addressed at: Department of Microbiology and Immunology, State University of New York Health Science Center at Brooklyn, 450 Clarkson Avenue, Box 44, Brooklyn, NY 11203. E-mail: [email protected].



7030

sufficient for efficient ribosomal attachment to capped mRNA (2). This fragment of eIF4G binds both eIF4E and eIF4A and probably coordinates their activities so that a cap-proximal region of mRNA is unwound and is thus rendered accessible to an incoming 43S complex so it can bind productively. The molecular interactions that enable the incoming 43S complex to bind this “prepared” template are not known but are thought to involve interaction of the eIF3 component of 43S complexes with cap-associated eIF4G. The bound ribosomal “complex I” was arrested in a cap-proximal position and did not reach the initiation codon (Fig. 1 A).

Fig. 1. The mechanism of action of eIF1 and eIF1A in promoting assembly of 48S ribosomal complexes at the authentic initiation codon of a conventional capped mRNA. The 5 terminal m7G residue is shown as a filled black circle, the 5 NTR as a black line, and the ORF downstream of the AUG initiation codon as a black rectangle. (A) In the presence of eIFs 2, 3, 4A, 4B, and 4F, an aberrant ribosomal complex (“complex I”) assembles at a cap-proximal position but is unable to scan downstream to the initiation codon. (B) In the presence of eIFs 1, 1A, 2, 3, 4A, 4B, and 4F, 48S ribosomal complexes assemble exclusively at the authentic initiation codon. (C) Addition of eIF1 and eIF1A to complex I promotes its complete conversion to correctly assembled 48S complexes after dissociation of complex I and rebinding of 43S ribosomal complexes in a scanning-competent form. Two additional activities present in rabbit reticulocyte lysate (RRL) enabled 43S complexes to reach the initiation codon, forming “complex II” without being arrested at the initial binding site (Fig. 1 B). These small factors were purified and identified by sequencing as eIF1 (13.5 kDa) and eIF1A (19 kDa) and could be functionally replaced by corresponding recombinant polypeptides. These two factors acted synergistically; eIF1 A without eIF1 enhanced eIF4F-mediated binding of 43S complexes to mRNA but did not enable these complexes to reach the initiation codon, whereas eIF1 without eIF1A reduced the prominence of the cap-proximal complex I and promoted formation of low levels of 48S complexes. The interaction with mRNA of 48S complexes assembled in the absence of eIF1A differed subtly from complexes formed in their presence, in that only two (+16–+17) rather than three toeprints (+15–+17) were apparent. eIF1A therefore increases the competence of 43S complexes to bind mRNA and the processivity of scanning 43S/eIF1/mRNA complexes. eIF1A also stabilizes binding of the ternary complex to 40S subunits in the absence of mRNA (3, 4), presumably by an allosteric mechanism, because it is not known to interact directly with any component of the ternary complex. This stabilization by eIF1A is weak but might be indicative of a role for eIF1A in ensuring that initiator tRNA and mRNA adopt the correct relative orientation on the scanning ribosomal complex. eIF1A comprises an oligonucleotide-binding (OB) ß-barrel fold that closely resembles prokaryotic initiation factor IF1 (and corresponds to the region of sequence homology between them) and an additional C-terminal domain (4). The experimentally determined RNA-binding surface of eIF1A is large, extending over the OB fold and the adjacent groove leading to the second domain. Mutations at multiple positions on this surface resulted in a reduced ability of eIF1A to promote assembly of 48S initiation complexes at the initiation codon. The RNA ligand for eIF1A is not known, but by analogy with IF1 (5), eIF1A might bind 18S ribosomal RNA in the ribosomal A site. In the absence of eIF1 and eIF1A, the mRNA-binding cleft on 40S subunits appears to be open, because they can bind mRNA in an end-independent manner during initiation by internal ribosomal entry (see below). eIF1 and eIF1 A may contribute to the correct interactions of components of the 43S complex with mRNA that enable it to enter a processive mode, for example by closing this cleft directly or indirectly and possibly even by forming part of the channel on the 40S subunit through which mRNA moves during ribosomal scanning. Experiments done by using competitor mRNAs indicated that complex I cannot be “chased” directly into complex II and is therefore not its immediate precursor. Complex I is aberrantly assembled (because it is arrested at a non-AUG triplet and is unable to scan to the initiation codon) and is intrinsically unstable. eIF1 and eIF1A together (but not individually) promote dissociation of complex I and enable the released 43S complex to rebind mRNA in a competent state to scan to the initiation codon (Fig. 1 C). eIF1 alone is able to recognize and destabilize ribosomal complexes incorrectly assembled by internal ribosomal entry (see below). Identification of this activity of mammalian eIF1 is consistent with characterization of its yeast homologue Sui1 as a monitor of translation accuracy. Mutations in Sui1 allow aberrant initiation in vivo at non-AUG codons by mismatch base pairing with Met-tRNAi (e.g., ref. 6). Determination of the solution structure of eIF1 by NMR (7) has revealed that these mutated residues form part of a surface that is almost perfectly conserved among all eIF1 homologues and that is likely directly involved in initiation codon selection by eIF1. In summary, we have determined the set of factors required for binding of a 43S complex to a model native capped mRNA and for it to scan to the initiation codon. These experiments were done by using ß-globin mRNA, and it is possible that ribosomal scanning on longer or more highly structured 5 NTRs may require additional as-yet-unidentified factors, for example to enhance processivity or to promote unwinding of stable secondary structures. Almost all aspects of the mechanism of ribosomal scanning remain uncharacterized (8). For example, scanning is an ATP-dependent process, but it is not known whether ribosomal movement itself involves hydrolysis of ATP or whether chemical energy is required only to unwind secondary structure in the 5 NTR to permit ribosomal movement by one-dimensional diffusion from its initial 5-terminal attachment site. The ability to reconstitute this process in vitro will enable this and other outstanding questions to be addressed. Factor Displacement from the 48S Complex and Joining to a 60S Subunit to Form Active 80S Ribosomes. The 48S complex assembled at the initiation codon of ß-globin mRNA is bound by factors that must be displaced before the 40S subunit/mRNA/Met-



7031

tRNAi complex can join with a 60S subunit. Substitution of GTP by guanosine 5-[ß,γ-imido] triphosphate (GMP-PNP) (a nonhydrolyzable analogue) arrests initiation at the stage of 48S complex formation, indicating that displacement of factors and subunit joining both require hydrolysis of GTP bound to eIF2 in 48S complexes. GTP hydrolysis by eIF2 is activated by eIF5, a 49-kDa polypeptide that interacts specifically with eIF2 and eIF3 (9, 10). Recent data suggest that eIF5 is a component of multifactor complex comprising eIF1, eIF3, eIF5, and the eIF2/GTP/Met-tRNAi ternary complex that can exist free of the ribosome and probably binds to it as a whole rather than sequentially (10). eIF5 binds strongly to eIF2 but induces its GTPase activity only when eIF2 is associated with the 40S subunit. GTP hydrolysis, which leads to dissociation of eIF2-GDP, is thought to be induced in response to base pairing between the initiation codon and the anticodon of Met-tRNAi, thereby ensuring stringent selection of the initiation codon during the scanning process (11). Until recently, the hydrolysis of eIF2-bound GTP was considered the only requirement for the joining of a 60S subunit to the 48S complex (see ref. 12 for a review). However, we found that addition of 60S subunits and recombinant eIF5 to 48S complexes assembled on globin mRNA did not lead to formation of 80S ribosomes (13). A partially purified ribosomal salt wash fraction from mouse ascites cells was active in promoting 80S ribosome assembly and was therefore used as a source for purification of additional factors. We purified two proteins to apparent homogeneity, which together but not separately were able to mediate assembly of 48S complexes and 60S subunits into 80S ribosomes. The smaller (49 kDa) protein could be functionally replaced by recombinant eIF5. The second protein had an apparent molecular mass of 175 kDa, and its N-terminal sequence identified it as a mouse homologue of prokaryotic initiation factor IF2 (Fig. 2). A role for a eukaryotic homologue of IF2 was first revealed by studies in yeast (14). Analysis of polyribosome profiles showed that deletion of the yeast IF2 homologue led to a reduction in formation of larger polysomes and an accumulation of inactive 80S ribosomal particles, and in vitro translation assays confirmed that this deletion led to a defect in translation initiation on the majority, if not all, cellular mRNAs. This defect could be rescued by adding back purified recombinant protein (14). These results indicated that this protein is a general translation factor in yeast. Human, Drosophila, and archaeal homologues have also been identified (15, 16). In light of its function in subunit joining, we named this factor eIF5B (13). Recombinant human eIF5B587–1220 lacking amino acids 1–586 could substitute for yeast eIF5B in vivo (15) and for native mammalian eIF5B in subunit joining in our in vitro reconstitution experiments (13). It is almost certain that eIF5B is a protein that was previously implicated in subunit joining but subsequently erroneously discounted as an inactive contaminant of eIF5 (12). Puromycin resembles the 3-end of aminoacylated tRNA and can bind to the ribosomal A site to react with Met-tRNAi in the P site to form methionylpuromycin. This reaction mimics formation of the first peptide bond, and we therefore used it to confirm that 80S ribosomes assembled by using eIF5 and eIF5B were active. Assembly of 48S complexes on AUG triplets is much simpler than on native mRNA, because it involves neither 5-end-dependent attachment nor scanning. 48S complex formation on AUG triplets requires only a 40S subunit and the eIF2/GTP/Met-tRNAi complex, which enabled us to investigate the influence of other factors on the requirements for subunit joining (13). A requirement for both eIF5 and eIF5B for 80S assembly was apparent only when 48S complexes were assembled by using eIF1, eIF1A, eIF2, and eIF3. These four factors are all normally associated with a 48S complex at the initiation codon. Individually, eIF5 and eIF5B were equally active in subunit joining in reactions lacking eIF1 and eIF3, but inclusion of eIF1 and eIF3 together reduced the individual activities of both eIF5 and eIF5B. The requirement for both eIF5 and eIF5B in these circumstances indicates that they have complementary functions.

Fig. 2. Sequence and structural conservation of eukaryotic eIF5B proteins from Homo sapiens (15), Drosophila melanogaster (16), and Saccharomyces cerevisiae (14), archaeal IF2 from Methanococcus jannaschii and prokaryotic IF2 from Escherichia coli (12). The percentages of amino acid identities to human eIF5B in the N-terminal region of the protein, the GTP-binding domain, and the C-terminal region of the protein are shown. The black rectangle in the schematic representation identifies the position of the GTP-binding domains in these proteins with the indicated GTP-binding protein consensus sequence motifs G1-G5 aligned with sequence motifs G1-G5 of E. coli IF2 and human eIF5B. Numbers above the domains of eIF5B/IF2 proteins refer to the amino acid residues in each protein; numbers below the aligned sequences refer to the amino acid residues in G1-G5 motifs of human eIF5B. Hydrolysis of GTP bound to 48S complexes is a prerequisite for subunit joining and was therefore also compared in the presence and absence of eIF1 and eIF3 (13). eIF5 and eIF5B stimulated GTP hydrolysis by eIF2 equally when 48S complexes contained only eIF1A and eIF2, but inclusion of eIF1 and eIF3 inhibited the stimulatory activity of eIF5B without affecting that of eIF5. This effect can account for the reduced ability of eIF5B to promote methionylpuromycin synthesis in the presence of eIF1 and eIF3. We conclude that, although eIF5 is active in inducing GTP hydrolysis on 48S complexes in the presence of a full set of factors (including eIF1 and eIF3), this is insufficient for subunit joining. Under these circumstances (when all factors associated with 48S complexes are present, which corresponds to the normal situation for initiation on capped mRNAs), eIF5B is also required. The central domain of eIF5B contains sequence motifs characteristic of GTP-binding proteins (Fig. 2). By UV crosslinking, we found that [32P]GTP bound directly to eIF5B independently of ribosomal subunits, and that bound [32P]GTP exchanged readily with unlabeled GTP, GMP-PNP, or GDP. eIF5B had no detectable intrinsic GTPase activity, but its ability to hydrolyze GTP was activated by 60S subunits and considerably more by 40S and 60S subunits together. Interestingly, prokaryotic IF2 is also



7032

a GTPase that is specifically activated by large and small ribosomal subunits together (17). This similarity between the homologous factors eIF5B and IF2 suggests that ribosomal activation of their GTPase activity may occur by a common mechanism. Binding of GTP to eIF5B may be required for it to adopt an active conformation. To test this hypothesis, 48S complexes were assembled with GTP, separated from unincorporated GTP by gel filtration, and then incubated with eIF5, 60S subunits, different nucleotides, and either full-length native eIF5B or recombinant eIF5B587–1220. The degree of dependence of eIF5B’s activity in 80S assembly on binding GTP was determined by the integrity of the protein. eIF5B587–1220 was completely GTP-dependent, whereas native eIF5B retained low activity in the absence of GTP but was nevertheless stimulated 3-fold by GTP (T.V.P., unpublished work). This result suggests that eIF5B adopts the active conformation required for subunit joining when it binds GTP. eIF5B acts catalytically in the presence of GTP, promoting multiple rounds of subunit joining. 80S complexes were also formed by eIF5B bound to GMP-PNP, but eIF5B-GMPPNP acted stoichiometrically rather than catalytically. This defect in the activity of eIF5B in the presence of GMP-PNP could be because hydrolysis of GTP bound to eIF5B is required for the release of eIF5B from assembled 80S ribosomes, for the release of other factors, or for both. The proportion of Met-tRNAi in 80S ribosomes assembled in the presence of GTP (60%) that reacted with puromycin was significantly greater than in complexes assembled by using GMP-PNP (8%). Methionylpuromycin synthesis by purified 80S ribosomes assembled in the presence of GTP was completely inhibited by addition of eIF5B587–1220 with GMP-PNP but not by either eIF5B587–1220 or GMP-PNP alone. This result indicates that eIF5BGMP-PNP can interact with preassembled 80S complexes, blocking their ability to react with puromycin (13). The specific inhibition of this reaction suggests that eIF5B binds to the ribosomal A site. When ribosomal complexes assembled by using GTP were resolved on sucrose density gradients, no eIF5B587–1220 was detected on 40S, 48S, 60S, or 80S complexes. However, a large amount of eIF5B587–1220 was bound to 80S complexes assembled in the presence of GMP-PNP. The inability of eIF5B 587–1220 to hydrolyze GMP-PNP therefore locks the factor on 80S complexes and renders them inactive in methionylpuromycin synthesis. eIF1, eIF2, and eIF3 were detected in 48S complexes but not in 80S complexes assembled with GTP or GMP-PNP. GTP hydrolysis by eIF5B is therefore not required for the release of these factors during subunit joining but is needed for release of eIF5B itself. The inability of eIF5B587–1220/GMP-PNP to dissociate from 80S ribosomes explains the requirement for stoichiometric rather than catalytic amounts of this factor in assembly reactions in the presence of GMP-PNP. Neither the stage during the initiation process at which eIF1, eIF1 A, eIF2, eIF3, and eIF5 are released nor the mechanism by which release occurs during initiation on native mRNAs has yet been established. Ribosomal subunit joining to form active 80S ribosomes that are competent to begin elongation therefore involves two successive GTP hydrolysis events: activation by eIF5 of hydrolysis of eIF2-bound GTP and ribosome-activated hydrolysis of eIF5B-bound GTP. Remarkably, eIF5B is a homologue of prokaryotic IF2, which also mediates a similar subunit-joining step that also involves ribosomeactivated hydrolysis of factor-bound GTP. Initiation of Picornavirus Translation by Internal Ribosomal Entry: The Role of Canonical Initiation Factors. Picornavirus RNA genomes are uncapped and have highly structured 5 NTRs that are barriers to scanning ribosomes. Initiation on these mRNAs is endindependent and is instead mediated by a 400-nt internal ribosomal entry site (IRES) in the 5 NTR (18, 19). The activity of an IRES depends on its structural integrity, and even point mutations can cause general or cell type-specific loss of function. Picornavirus IRESs are divided into two major groups on the basis of sequence and structural similarities (20, 21). One group contains poliovirus and rhinovirus, and the other group contains encephalomyocarditis virus (EMCV), Theiler’s murine encephalomyelitis virus (TMEV), and foot-and-mouth disease virus (FMDV). The EMCV and TMEV initiation codons are located at the 3 border of the IRES, and ribosomes bind directly to them without scanning (22, 23). In poliovirus, the initiation codon is 160 nt from the 3 border of the IRES, and it is possible that the ribosome reaches it either by scanning or by discontinuous transfer (“shunting”) after initial attachment to the IRES (24). Picornavirus infection often leads to shutoff of cap-mediated translation initiation, for example by rhinovirus protease cleavage of eIF4G at R641/G642, such that the N-terminal domain of eIF4G that binds eIF4E and the poly (A)-binding protein is separated from the Cterminal domain that binds eIF3 and eIF4A (25). This cleavage impairs eIF4F’s function in initiation on capped mRNAs. However, as described below, this cleavage yields a fragment of eIF4F that retains functions necessary for picornavirus IRES-mediated initiation. We reconstituted initiation in vitro on the EMCV IRES and found that it is ATP-dependent and requires only eIFs 2, 3, and either eIF4F or eIF4A and the central third of eIF4G to which eIF4A binds (26, 27). The requirement for eIF4A and the cognate domain of eIF4G is consistent with the profound inhibition of EMCV translation caused by dominant negative eIF4A mutant polypeptides (28). The inhibition caused by these mutants is thought to be because of their failure to exit the eIF4F complex and recycle efficiently, thereby trapping it in an inactive form. In this model, eIF4A therefore plays its role in initiation as part of a complex with eIF4G rather than as a singular polypeptide. 48S complex formation was enhanced 4-fold by eIF4B and less than 2-fold by the pyrimidine tract-binding protein (PTB), a noncanonical mRNA-specific initiation factor (see below). Together, these factors promoted 48S complex formation equally at AUG834 (the authentic initiation codon) and at AUG826 (which is virtually unused in vivo). Remarkably, inclusion of eIF1 in assembly reactions or even addition of eIF1 to preformed complexes led to dissociation of the ribosomal complex at AUG826 (13). This observation is consistent with the previously noted function of eIF1 in enhancing the fidelity of initiation codon selection. The principal difference between the factor requirements for initiation on ß-globin mRNA and on the EMCV IRES is that the latter has no requirement for eIF4E or the fragment of eIF4G to which it binds and is therefore not impaired by cleavage of eIF4G by viral proteases. eIF4E is a major focus of mechanisms that regulate initiation of translation in vivo (29). The EMCV IRES and other IRESs that do not require eIF4E are therefore active in circumstances that lead to inhibition of cap-mediated initiation by impairment of eIF4E function. A significant insight to the mechanism of initiation on the EMCV IRES came from the observations that eIF4F bound to the J-K domain of the EMCV IRES a little upstream of the initiation codon (Fig. 3), and that this interaction is essential for initiation (26). The essential central eIF4G722–949 domain binds specifically to the IRES, and its binding is strongly enhanced by eIF4A (33). The interaction of eIF4G with this IRES may play a role analogous to that of eIF4E on capped mRNAs, that is, to recruit the eIF4F complex and associated factors and to promote ribosomal attachment at a defined location on an mRNA. The Role of IRES Transacting Factors in Initiation of Picornavirus Translation. The activity of a number of picornavirus IRESs is subject to cell-type-specific restriction: for example, the atten



7033

uation of poliovirus vaccine strains is in part because of a defect in translation in neuronal cells. Poliovirus and rhinovirus IRESs mediate initiation of translation efficiently in HeLa and Krebs 2 cells, and their restricted activity in RRL in vitro can be alleviated by deletion of the IRES or by supplementation of RRL by ribosomal salt wash fractions from HeLa or Krebs cells (e.g., refs. 34, 35). Translation mediated by poliovirus and rhinovirus IRESs thus depends on the interaction with these IRESs of noncanonical IRES transacting factors (ITAFs) that are either absent from RRL or significantly less abundant in RRL than in permissive cells. In early experiments to identify ITAFs required for picornavirus IRES function, we and others identified a 57-kDa protein that bound specifically to all picornavirus IRESs as the PTB, a cellular polypeptide that contains four RNA-recognition motif -like domains (36–39).

Fig. 3. Schematic representation of EMCV, FMDV, and TMEV IRES domains H-L, showing binding sites for PTB (thick gray line) and ITAF45 (black icosahedron), as determined by footprinting (30, 31), and for the eIF4G/eIF4A complex, as determined by footprinting (31, 32) and toeprinting (26, 27, 33). The interaction of eIF4G/eIF4A with the J-K domain, which is essential for recruitment of the 43S complex to the initiation codon, is enhanced by PTB and ITAF45 in an IRES-specific manner, as discussed in the text. The PTB dependence of initiation on the wild-type EMCV IRES is small (27, 40) but was significantly enhanced by a single nucleotide insertion in the eIF4G-binding site or by alteration of the sequence downstream of the initiation codon (40). Foot-printing analysis indicated that PTB binds multiple noncontiguous sites on the EMCV IRES (ref. 30; Fig. 3). Taken together, these observations suggested a model in which binding of ITAFs such as PTB stabilizes an IRES in an optimal conformation for binding of essential factors and the 43S complex. Our analysis of initiation on the related TMEV and FMDV IRESs provided strong support for this hypothesis (31). TMEV (GDVII strain) and FMDV (type 01K) are neurotropic and epitheliotropic picornaviruses, respectively. Their IRESs are 40% identical and are closely related to the EMCV IRES. Substitution of the IRES in an infectious genomic TMEV clone by that of FMDV strongly attenuated the neurovirulence of the resulting chimeric virus without significantly affecting its ability to replicate in cultured BHK-21 cells or to be translated in vitro in RRL (31). We used biochemical reconstitution of the initiation process on these mRNAs to define the molecular basis for this cell type-specific difference in the function of these IRESs. Initiation on the FMDV and TMEV IRESs had identical requirements for canonical initiation factors to those described above for EMCV. However, whereas initiation on the EMCV IRES was only weakly stimulated by PTB, initiation on the TMEV IRES depended strongly on PTB, and initiation on the FMDV IRES required both PTB and a 45kDa ITAF (ITAF45). We purified and sequenced ITAF45 and found that it is identical to a previously identified proliferation-dependent protein that is not expressed in nonproliferating cells such as neurons (31). The absence of this factor may account for the inability of the chimeric TMEV virus to replicate in neurons. The activities of PTB and ITAF45 in promoting 48S complex formation on the FMDV IRES were strongly synergistic. These ITAFs bound to nonoverlapping sites on the IRES (Fig. 3) and together caused localized conformational changes in it, specifically in regions adjacent to the binding site for the eIF4G/eIF4A complex. The interaction of the IRES with the eIF4G/eIF4A complex is essential for initiation and, significantly, this interaction was specifically enhanced by these two ITAFs. EMCV, FMDV, and TMEV IRESs all bind to PTB and ITAF45, so it is the requirement for them rather than their ability to interact that differs as a consequence of sequence differences between these IRESs. Similar observations have also been made for the second group of picornavirus IRESs, members of which are also closely related to each other yet also appear to have different ITAF requirements. For example, poliovirus and rhinovirus IRESs both bind to PTB, to the poly(rC)-binding protein 2 (PCBP2), and to unr (35, 41). PTB contains four RNA-recognition motiflike domains, PCBP2 has three KH domains that likely constitute its RNA-binding surface, and unr contains five cold-shock RNA-binding domains. These polypeptides therefore all have the potential to make multipoint interactions with these IRESs and to stabilize their folding in an active conformation. However, whereas initiation on the rhinovirus IRES depends on unr, strongly enhanced by PTB and less responsive to PCBP2, the poliovirus IRES depends on PTB and PCBP2 and does not respond to unr (35). Our analysis of initiation on EMCV-like IRESs suggests a model that will likely be applicable to poliovirus-like IRESs and possibly to some other viral and cellular IRESs. Specific binding of eIF4F (or its eIF4G and eIF4A subunits) to the IRES is required to mediate internal ribosomal entry and itself depends on the eIF4F and ribosomal binding sites having the correct conformation. The role of ITAFs is to bind an IRES to enable it to attain or maintain this conformation, for binding both of eIF4G/eIF4A and possibly of other components of the 43S complex. The diversity of IRES sequences and structures leads to the requirement for a variety of ITAFs. Internal Initiation by Factor-Independent Binding of Ribosomes to the Initiation Codon. The 5 NTRs of HCV and of the related classical swine fever virus (CSFV) and bovine viral diarrhea virus (BVDV) also promote translation by cap-independent internal ribosomal entry (e.g., ref. 42). IRESs are defined solely by functional criteria and cannot yet be predicted by the presence of characteristic RNA sequence or structural motifs. As a rule, there are no significant similarities between individual IRESs (unless they are from related viruses). The related BVDV, CSFV, and HCV IRESs are the best characterized members of an IRES group that is wholly distinct from both the EMCV-like and poliovirus-like groups of IRESs with regard to length, sequence, and structure. We investigated initiation on BVDV, CSFV, and HCV IRESs to determine whether all IRESs mediate internal initiation of translation by a single common mechanism irrespective of sequence variation and, if not, to characterize unique aspects of initiation on this group of HCV-like IRESs. The BVDV, CSFV, and HCV 5 NTRs are 385, 372, and 342 nt long, respectively, and although they differ from each other at 35–50% of base positions, many of these nucleotide differences are covariant substitutions, indicative of conserved higher order structure. Even minor mutations in structural elements substantially reduced IRES activity, but this could in most instances be restored by compensatory second site mutations that restored



7034

secondary structure (43–46). The most highly conserved residues are often unpaired and may thus be able to interact with components of the translation apparatus. These results and observations have led to a model for IRES function in which structural elements in the IRES act as a scaffold that orients these potential binding sites in such a way that their interaction with factors and ribosomes leads to assembly of functional ribosomal initiation complexes.

Fig. 4. Schematic secondary structure of domains II, III, and IV of HCV-like IRESs, showing sites of interaction with eIF3 (thick black lines) (47) and with 40S subunits (thick gray lines) (48, 49). The toeprints detected at the leading edge of bound eIF3 (46, 47, 49, 50) are indicated by an arrow. The toeprints at the leading edge of 40S subunits in binary IRES:40S subunit complexes are indicated by open circles and in 48S complexes formed on inclusion of eIF2/ GTP/Met-tRNAi with 40S subunits by filled circles (46, 50). Toeprints caused by ribosomal contact with the pseudoknot are not shown. Sequences flanking the initiation codon are base paired to form domain IV in HCV but not in BVDV and CSFV. BVDV and CSFV contain two hairpins (IIId1 and IIId2) at an analogous position to HCV IIId. The nomenclature of helices in the pseudoknot and of domains is as in ref. 48. These HCV-like 5 NTRs consist of four major structural domains (I–IV) and a complex pseudoknotted structure between domains II, III, and IV (Fig. 4). HCV domain IV is base-paired, whereas equivalent residues in BVDV and CSFV are not. The boundaries of these IRESs extend from the 3 border of the basal helix of domain II to the initiation codon, and IRES activity is affected by the coding sequence downstream of the initiation codon. Minor mutations in domain II, domain III, and the pseudoknot can cause substantial reductions in IRES activity. We determined the minimum set of factors required for assembly of 48S complexes on these IRESs by in vitro reconstitution by using purified translation components (46, 50). The most remarkable aspect of initiation on these IRESs is that they bind 40S subunits specifically and stably, in the absence of initiation factors, so that the ribosomal P site is placed in the immediate proximity of the initiation codon. Addition of eIF2/GTP/Met-tRNAi is sufficient for the bound 40S subunit to lock onto the initiation codon. The direct attachment of the 43S complex to the initiation codon is consistent with earlier reports that translation initiation on HCV and CSFV IRESs in RRL does not involve ribosomal scanning after initial attachment (44, 45). Although eIF3 is not required for assembly of this minimal 48S complex, eIF3 has been reported to be associated with free 40S subunits in the cytoplasm, and it is therefore likely that, in vivo, it is also a constituent of 48S complexes on these IRESs. eIF3 itself also binds specifically and stably to the IRES; the independent interaction of two different components of the 43S complex with the IRES may enhance the affinity and specificity of binding. The binding site for eIF3 has been mapped by toeprinting and chemical/enzymatic footprinting to the apical half of domain III (ref. 47; Fig. 4) and includes subdomain IIIb and junction IIIabc. Notably, 48S complex formation on HCV-like IRESs has no requirement for eIF4A, 4B, 4E, or 4G, nor any requirement for ATP hydrolysis. Translation mediated by these IRESs is also not inhibited by dominant negative eIF4A mutants (46, 50). It thus differs fundamentally from cap-mediated initiation and initiation mediated by picornavirus IRESs of both the EMCV and poliovirus-like groups. The initiation factors eIF4A, 4B, 4E, and 4G do not influence initiation complex formation on HCV-like IRESs, and indeed these factors are probably unable to gain access to and unwind the region of the IRES immediately upstream and downstream of the initiation codon that enters the mRNA-binding cleft of the 40S subunit. For example, translation efficiency is strongly reduced by mutations that increase base pairing in HCV domain IV (which contains the initiation codon) and thus stabilize it (51) and by introduction of a hairpin immediately downstream of the CSFV initiation codon (52). Secondary structures of equivalent stability are readily unwound during cap-mediated initiation. Initiation on prokaryotic mRNAs involves factor-independent binding of small (30S) ribosomal subunits as a result of base pairing between the linear Shine-Dalgarno (SD) sequence in mRNAs and complementary linear anti-SD sequences in the ribosomal 16S rRNA (52). Although there are striking analogies between this mechanism and the factor-independent binding of 40S subunits to HCV-like IRESs, it is evident that binding of 40S subunits is determined by multiple noncontiguous sequences in the IRES rather than by a single linear sequence. We do not yet know whether binding of an IRES to the 40S subunit involves RNA-RNA base pairing with 18S rRNA. The only contact identified so far is with a ribosomal protein, but this interaction likely is not a primary determinant of the IRES/40S subunit interaction (46, 48). Toeprinting and deletion analyses indicated that a 40S subunit interacts with the IRES at multiple sites; primer extension is arrested by bound 40S subunits in the pseudoknot and downstream of the initiation codon (Fig. 4). We used enzymatic footprinting to identify the principal sites in HCV and CSFV IRESs that are protected from cleavage by bound 40S subunits (48, 49). Similar interaction sites were identified in these two IRESs, and they are located in regions of high sequence conservation in HCV-like IRESs. These sites include the apex of HCV subdomain IIId and the equivalent CSFV subdomain IIId1, the pseudoknot, and nucleotides flanking the initiation codon. The number of protected residues in the last of these sites corresponds closely to the length estimated for the mRNA-binding cleft in 40S subunits, and it is therefore likely that additional upstream contacts between the IRES and the 40S subunit involve regions of the 40S subunit outside this cleft. The ribosomal binding surface of the IRES is therefore extensive; these footprinting and mutational analyses (see below) suggest that it does not overlap the eIF3-binding site except in subdomain IIIa. The importance of these sites of interaction with the 40S subunit for IRES function is supported by the results of mutational analysis. The apical residues GGG266–268 in HCV IIId and analogous residues (GGG268–270) in CSFV IIId1 are essential for ribosomal binding and IRES function (49, 53). The pseudoknot has long been known to be functionally important (43, 45, 46). We found that substitutions in its 5 helical segment abrogate ribosomal binding, whereas substitutions in its 3 helix do not prevent ribosomal attachment to the IRES but impair binding of sequences flanking the initiation codon to the ribosomal mRNA-binding cleft (46, 49). Consistent with this conclusion, we found that residues flanking the initiation codon are also not required for ribosomal



7035

attachment to the IRES to form a stable binary complex, even though they constitute a major site of interaction between these IRESs and the 40S subunit. Similarly, deletion of domain II or mutations in domain IIIa also impaired binding of the initiation codon region to the ribosomal mRNA-binding cleft but did not prevent binary complex formation. These parts of the IRES therefore do not contain primary determinants of ribosomal binding (48, 49). We conclude that HCV-like IRESs contain one set of determinants that is required for initial ribosomal attachment (including subdomain IIId/IIId1, adjacent regions of domain III, and the 5 helical segment of the pseudoknot) and a second set of determinants (including domain II, the 3 helical segment of the pseudoknot, and downstream sequences) that is required for, or at least promotes, subsequent accurate placement of the initiation codon in the ribosomal P site. The IRES is not a static structure, and it is likely that it undergoes structural transitions during these two stages in ribosomal binding and subsequently during subunit joining. The mechanism of initiation on HCV-like IRESs is therefore distinct from both cap-mediated and EMCV IRES-mediated initiation in having no requirement for ATP or for any member of the eIF4F group of factors. HCV-like IRESs bypass the requirement for these factors and for eIF1 and eIF1A by virtue of their ability to recruit 43S complexes directly to the initiation codon by binding specifically to eIF3 and to the 40S subunit. The importance of the integrity of the structure of these IRESs for this mode of translation initiation suggests that these IRESs constitute valid targets for potential chemotherapeutic agents such as antibiotics that could bind the IRES and distort the structure of binding sites for these components of the translation apparatus. Perspectives. We have characterized the outlines of three different mechanisms of translation initiation by using biochemical reconstitution to determine the minimum set of factors required for assembly of 48S and 80S ribosomal complexes on three distinct types of eukaryotic mRNA and by using toeprinting and footprinting to map the location of translation components on these mRNAs. The findings reported here raise both general and specific questions about translation initiation. The finding that internal ribosomal entry on the two types of IRES that we have examined occurs by very different mechanisms indicates there is no single mode of internal ribosomal entry. Indeed, the implicit possibility that there are yet other mechanisms for initiation directed by an IRES has been borne out by recent analysis of initiation on the intergenic IRES of cricket paralysis virus (CrPV), which remarkably requires neither initiator tRNA nor initiation factors (54). Like HCV and related IRESs, this CrPV IRES also binds directly to 40S subunits but in a significantly different manner, such that the P site is apparently filled by a pseudoknot and is inaccessible to the eIF2/GTP/Met-tRNAi complex. Because the number of known cellular and viral IRESs is constantly growing, we cannot rule out that additional mechanisms of internal ribosomal entry exist that are distinct from those used by EMCV, HCV, or CrPV-like IRESs. It seems probable that even those IRESs on which initiation occurs by a mechanism fundamentally similar to one of these three groups of IRESs will nevertheless require ITAFs different from those identified to date. It will be interesting to see whether the “induced active conformation” model for ITAF function described for the FMDV IRES (31) will be more generally applicable. Just as it is unlikely that initiation on all IRESs will be described by one of the three models described above, so it would be premature to assume that initiation on all capped mRNAs occurs by the mechanism that we have described for ß-globin mRNA. More specifically, our knowledge of the scanning process is very rudimentary, and a number of open questions need to be addressed in the near future. These questions include: (i) What are the molecular interactions and conformational changes that lead to binding of a 43S complex to the capped eIF4F-bound 5 end of an mRNA? (ii) How and when are interactions between cap-bound factors and the 43S complex dissociated as this complex begins to scan from the cap-proximal region of an mRNA? (iii) Is ribosomal movement on the 5 NTR obligatorily linked to “melting” secondary structure in the 5 NTR, and, if these processes can be uncoupled, is the 43S complex intrinsically capable of movement on mRNA without concomitant ATP hydrolysis? (iv) Which factors influence the processivity of scanning? (v) How does the local sequence context of an initiation codon influence the efficiency of initiation at that codon? (vi) How does recognition of the initiation codon trigger all of the events associated with subunit joining? Answers to these questions not only will lead to a more detailed understanding of the molecular mechanism of the initiation process but also will offer insights into how structural differences between different mRNAs determine when and how efficiently they are translated. Research done in our laboratories was supported Grants AI44108–01 and GM59660 from the National Institutes of Health (to C.U.T.H. and T.V.P.), by Grant MCB-9726958 from the National Science Foundation (to C.U.T.H.), and by grants from the Council for Tobacco Research Council (to C.U.T.H.), the Howard Hughes Medical Institute (to I.N.S. and C.U.T.H.), and the Russian Foundation of Basic Research (to V.I.A. and I.N.S.). 1. Pestova, T.V., Borukhov, S.I. & Hellen, C.U.T. (1998) Nature (London) 394, 854–859. 2. Morino, S., Imataka, H., Svitkin, Y.V., Pestova, T.V. & Sonenberg, N. (2000) Mol. Cell Biol. 20, 468–477. 3. Chaudhuri, J., Chowdhury, D. & Maitra, U. (1999) J. Biol. Chem. 274, 17975–17980. 4. Battiste, J.L., Pestova, T.V., Hellen, C.U.T. & Wagner, G. (2000) Mol. Cell 5, 109–119. 5. Dahlquist, K.D. & Puglisi, J.D. (2000) J. Mol. Biol. 299, 1–15. 6. Yoon, H. & Donahue, T.F. (1992) Mol. Cell Biol. 12, 248–260. 7. Fletcher, C.M., Pestova, T.V., Hellen, C.U.T. & Wagner, G. (1999) EMBO J. 18, 2631–2637. 8. Pestova, T.V. & Hellen, C.U.T. (1999) Trends Biochem. Sci. 24, 85–87. 9. Chakrabarti, A. & Maitra, U. (1991) J. Biol. Chem. 266, 14039–14045. 10. Asano, K., Clayton, J., Shalev, A. & Hinnebusch, A.G. (2000) Genes Dev. 14, 2534–2546. 11. Huang, H.K., Yoon, H., Hannig, E.M. & Donahue, T.F. (1997) Genes Dev. 11, 2396–2413. 12. Pestova, T.V., Hellen, C.U.T. & Dever, T.E. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Mathews, M.B. & Hershey, J.W.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 425–445. 13. Pestova, T.V., Lomakin, I.B., Lee, J.H., Choi, S.K., Dever, T.E. & Hellen, C.U.T. (2000) Nature (London) 403, 332–335. 14. Choi, S.K., Lee, J.H., Zoll, W.L., Merrick, W.C. & Dever, T.E. (1998) Science 280, 1757–1760. 15. Lee, J.H., Choi, S.K., Roll-Mecak, A., Burley, S.K. & Dever, T.E. (1999) Proc. Natl. Acad. Sci. USA 96, 1066–1070. 16. Carrera P., Johnstone, O., Nakamura, A., Casanova, J., Jackle, H. & Lasko P. (2000) Mol. Cell 5, 181–187. 17. Kolakofsky, D., Ohta, T. & Thach, R.E. (1968) Nature (London) 220, 244–247. 18. Jang, S.-K., Kräusslich, H.-G., Nicklin, M.J.H., Duke, G.M., Palmenberg, A.C. & Wimmer, E. (1988) J. Virol. 62, 2636–2643. 19. Pelletier, J. & Sonenberg, N. (1988) Nature (London) 334, 320–325. 20. Pilipenko, E.V., Blinov, V.M., Chernov, B.K., Dmitrieva, T.M. & Agol, V.I. (1989) Nucleic Acids Res. 17, 5701–5711. 21. Pilipenko, E.V., Blinov, V.M., Romanova, L.I., Sinyakov, A.N., Maslova, S.V. & Agol, V.I. (1989) Virology 168, 201–209. 22. Kaminski, A., Howell, M.T. & Jackson, R.J. (1990) EMBO J. 9, 3753–3759. 23. Pilipenko, E.V., Gmyl, A.P., Maslova, S.V., Belov, G.A., Sinyakov, A.N., Huang, M., Brown, T.D.K. & Agol, V.I. (1994) J. Mol. Biol. 241, 398– 414. 24. Hellen, C.U.T., Pestova, T.V. & Wimmer, E. (1994) J. Virol 68, 6312–6322.



7036

25. Lamphear, B.J., Kirchenweger, R., Skern, T. & Rhoads, R.E. (1995) J. Biol. Chem. 270, 21975–21982. 26. Pestova, T.V., Hellen, C.U.T. & Shatsky, I.N. (1996) Mol. Cell. Biol. 16, 6859–6869. 27. Pestova, T.V., Shatsky, I.N. & Hellen, C.U.T. (1996) Mol. Cell. Biol. 16, 6870–6878. 28. Pause, A., Methot, N., Svitkin, Y., Merrick, W.C. & Sonenberg, N. (1994) EMBO J. 13, 1205–1215. 29. Gingras, A.C., Raught, B. & Sonenberg, N. (1999) Annu. Rev. Biochem. 68, 913–963. 30. Kolupaeva, V.G., Hellen, C.U.T. & Shatsky, I.N. (1996) RNA 2, 1199–1212. 31. Pilipenko, E.V., Pestova, T.V., Kolupaeva, V.G., Khitrina, E.V., Poperechnaya, A.N., Agol, V.I. & Hellen, C.U.T. (2000) Genes Dev. 14, 2028–2045. 32. Kolupaeva, V.G., Pestova, T.V, Hellen, C.U.T. & Shatsky, I.N. (1998) J. Biol. Chem. 273, 18599–18604. 33. Lomakin, I.B., Hellen, C.U.T. & Pestova, T.V. (2000) Mol. Cell. Biol. 20, 6019–6029. 34. Svitkin, Y.V., Pestova, T.V., Maslova, S.V. & Agol, V.I. (1988) Virology 166, 394–404. 35. Hunt, S.L., Hsuan, J.J., Totty, N. & Jackson, R.J. (1999) Genes Dev. 13, 437–448. 36. Borovjagin, A.V., Evstafieva, A.G., Ugarova, T.Y. & Shatsky, I.N. (1990) FEBS Lett. 261, 237–240. 37. Jang, S.K. & Wimmer, E. (1990) Genes Dev. 4, 1560–1572. 38. Hellen, C.U.T., Witherell, G.W., Schmid, M., Shin, S.H., Pestova, T.V., Gil, A. & Wimmer, E. (1993) Proc. Natl. Acad. Sci. USA 90, 7642–7646. 39. Borman, A., Howell, M.T., Patton, J. & Jackson, R.J. (1993) J. Gen. Virol. 74, 1775–1788. 40. Kaminski, A. & Jackson, R.J. (1998) RNA 4, 626–638. 41. Walter, B.L., Nguyen, J.H., Ehrenfeld, E. & Semler, B.L. (1999) RNA 5, 1570–1585. 42. Tsukiyama-Kohara, K., lizuka, N., Kohara, M. & Nomoto, A. (1992) J.Virol. 66, 1476–1483. 43. Wang, C., Le, S.Y., Ali, N. & Siddiqui, A. (1995) RNA 1, 526–537. 44. Reynolds, J.E., Kaminski, A., Carroll, A.R., Clarke, B.E., Rowlands, D.J. & Jackson, R.J. (1996) RNA 2, 867–878. 45. Rijnbrand, R., van der Straaten, T., van Rijn, P.A., Spaan, W.J.M. & Bredenbeek, P.J. (1997) J.Virol. 71, 451–457. 46. Pestova, T.V., Shatsky, I.N., Fletcher, S.P., Jackson, R.J. & Hellen, C.U.T. (1998) Genes Dev. 12, 67–83. 47. Sizova, D.V., Kolupaeva, V.G., Pestova, T.V., Shatsky, I.N. & Hellen, C.U.T. (1998) J.Virol 72, 4775–4782. 48. Kolupaeva, V.G., Pestova, T.V. & Hellen, C.U. T. (2000) J. Virol 74, 6242–6250. 49. Kolupaeva, V.G., Pestova, T.V. & Hellen, C.U.T. (2000) RNA 6, 1791–1807. 50. Pestova, T.V. & Hellen, C.U.T. (1999) Virology 258, 249–256. 51. Honda, M., Brown, E.A. & Lemon, S.M. (1996) RNA 2, 955–968. 52. Jackson, R.J. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Mathews, M.B. & Hershey, J.W.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 127–183. 53. Jubin, R., Vantuno, N.E., Kieft, J.S., Murray, M.G., Doudna, J.A., Lau, J.Y. & Baroudy, B.M. (2000) J. Virol 74, 10430–10437. 54. Wilson, J.E., Pestova, T.V., Hellen, C.U.T. & Sarnow, P. (2000) Cell 102, 511–520.


THE TARGET OF RAPAMYCIN (TOR) PROTEINS

7037

Colloquium The target of rapamycin (TOR) proteins

Brian Raught*, Anne-Claude Gingras*, and Nahum Sonenberg† Department of Biochemistry and McGill Cancer Centre, McGill University, 3655 Promenade Sir-William-Osler, Montréal, QC H3G 1Y6 Canada Rapamycin potently inhibits downstream signaling from the target of rapamycin (TOR) proteins. These evolutionary conserved protein kinases coordinate the balance between protein synthesis and protein degradation in response to nutrient quality and quantity. The TOR proteins regulate (i) the initiation and elongation phases of translation, (ii) ribosome biosynthesis, (iii) amino acid import, (iv) the transcription of numerous enzymes involved in multiple metabolic pathways, and (v) autophagy. Intriguingly, recent studies have also suggested that TOR signaling plays a critical role in brain development, learning, and memory formation. RAPAMYCIN INHIBITS LONG-TERM FACILITATION Synaptic plasticity, the capacity of neurons to modulate the strength of synaptic connections, is believed to be critical for learning and memory formation. Long-term synaptic plasticity (necessary for the formation of long-term memory) requires alterations in gene expression and the establishment of new synaptic connections (1–3). These findings presented an interesting dilemma: That is, how can changes in gene expression in the cell body alter the strength of individual synaptic connections? Recent data suggest that stimulated synapses are “tagged” to capture mRNAs produced in the soma and exported throughout the cell (4). Synaptic tagging thus results in localization of mRNAs only to those synapses marked by previous activity. This model also presupposes that long-term plasticity depends on local translation of the localized mRNAs. Indeed, ribosomes, tRNAs, translation initiation factors, and translation elongation factors are found in dendrites (5, 6), and protein synthesis has been demonstrated to occur in isolated synaptic bodies (7, 8). Functional studies have demonstrated that protein synthesis is required for potentiation of synaptic transmission elicited by neurotrophic factors in hippocampal slices (9), and for the establishment of long-term facilitation in Aplysia neurons (10). Kandel and coworkers implicated a specific intracellular signaling pathway in this process by demonstrating that serotonin-stimulated synaptic protein synthesis can be blocked with rapamycin, an inhibitor of the target of rapamycin (TOR) proteins (11). The aim of this review is to outline a current model regarding the intracellular signaling pathway inhibited by rapamycin, to detail known downstream targets of this signaling module, and to discuss putative links between TOR signaling and localized protein synthesis in neurons. RAPAMYCIN AND TOR Rapamycin is a lipophilic macrolide, isolated from a strain of Streptomyces hygroscopicus indigenous to Easter Island (known as Rapa Nui to the inhabitants; ref. 12). The intracellular rapamycin receptor in all eukaryotes is a small, ubiquitous protein termed FKBP12 (FK506-binding protein, molecular mass of 12 kDa; refs. 13, 14, 15). A rapamycin-FKBP12 “gain-of-function” complex interacts specifically with the evolutionarily conserved TOR proteins, to potently inhibit signaling to downstream targets. Two Saccharomyces cerevisiae TOR genes code for two large molecules (>280 kDa) sharing 67% identity at the amino acid level (16–19). Two Tor orthologs have also been isolated from Schizosaccharomyces pombe (20). Metazoans appear to possess only one TOR protein. A single Drosophila melanogaster ortholog, dTOR, is present in the completed fly genome, and shares 38% identity with the S. cerevisiae Tor2 protein (21, 22). A single mammalian TOR protein has been cloned from several species, and alternatively termed mTOR, FRAP (FKBP12 and rapamycin associated protein), RAFT (rapamycin and FKBP12 target), SEP (sirolimus effector protein), or RAPT (rapamycin target; refs. 23–27). Here, we refer to the mammalian protein as mTOR. mTOR is 289 kDa and shares 45% identity with the S. cerevisiae Tor1 and -2 proteins, and 56% identity with dTOR (21–23, 26, 27). The human, rat, and mouse mTOR proteins share >95% identity at the amino acid level (reviewed in ref. 28). TOR SIGNALING The TOR proteins have been assigned to a protein family termed the phosphatidylinositol kinase-related kinases (or PIKKs), a large group of signaling molecules that also includes the ataxiatelangiectasia mutated (ATM) protein, ATR/FRP (ataxia-telangiectasia- and rad3related/FRAP related protein), and DNA-dependent protein kinase (DNA-PK; e.g., ref. 29). Despite significant homology to lipid kinases, the TOR proteins (as well as the other PIKKs) function as Ser/Thr protein kinases (reviewed in refs. 30 and 31). How Does Rapamycin Inhibit TOR Signaling? The rapamycin-FKBP12 gain-of-function complex inhibits downstream signaling from the TOR proteins in vivo. However, whether this complex directly inhibits the kinase activity of the TOR proteins is an unresolved issue. Rapamycin was reported to inhibit a moderate stimulation of mTOR kinase activity (measured in vitro, using an mTOR immunoprecipitate) in response to insulin treatment (32), and rapamycin-FKBP12 can inhibit mTOR autokinase activity in vitro. However, it appears that a much higher concentration of rapamycin than is required in vivo is necessary to elicit this effect (ref. 33; and references therein). Furthermore, only very modest differences, or no change at all, in the kinase activity of TOR immunoprecipitates have been reported after mitogenic stimulation, amino acid withdrawal, or rapamycin treatment (refs. 22 and 33; and references therein). Rapamycin treatment of cells in culture does not inhibit autophosphorylation at S2481, as determined with a phosphospecific antibody directed against this site (33). Finally, in S. cerevisiae, a mutation in the kinase domain of the Tor2 protein is lethal, yet rapamycin treatment of yeast leads only to G1 arrest. If rapamycin were to inhibit Tor2p kinase activity, mutation of the Tor2p kinase

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: TOR, target of rapamycin; FKBP12, FK506-binding protein, molecular mass of 12 kDa; P13K, phosphoinositide 3kinase; 5’TOP, 5 terminal oligopyrimidine tract; NMDA, N-methyl-D-aspartate. *B.R. and A.-C.G. contributed equally to this report. †To whom reprint requests should be addressed. E-mail: [email protected].



7038

region and rapamycin treatment should elicit identical phenotypes. Thus, whereas it is clear that rapamycin functions through an inhibition of downstream signaling from the TOR proteins, this repression may involve mechanisms other than a direct suppression of TOR kinase activity.

Fig. 1. The Tor proteins regulate the balance between protein synthesis and protein degradation. TOR signaling is active in the presence of sufficient nutrients to fuel protein synthesis. The TOR signal allows for the translation of mRNAs coding for components of the translation machinery, ribosome biosynthesis, and the stabilization of high affinity amino acid permeases. At the same time, TOR signaling destabilizes general amino acid permeases, inhibits autophagy, and represses the transcription of a subset of genes required for amino acid biosynthesis. What Signals to TOR? The TOR proteins do not appear to function as components of a conventional linear signaling pathway. Rather, several lines of evidence suggest that the TOR proteins function in a nutrient-sensing checkpoint control capacity. As discussed further below, both TOR and phosphoinositide 3-kinase (PI3K) signaling are required for the activation (or inactivation) of several downstream effector proteins. However, whether TOR activity is regulated by PI3K, or whether the two signaling pathways function independently, is unknown. Over-expression of a membrane-targeted Akt/PKB protein (a downstream effector of PI3K) in mammalian cells leads only to a modest increase (or no change) in mTOR kinase activity (as assayed in vitro), and moderately increases mTOR autophosphorylation in vivo, as assessed with the S2481 phosphospecific antibody (32–34). A putative Akt consensus phosphorylation site, S2448, was observed to be phosphorylated on mTOR in vivo, as determined with a phosphospecific antibody. Addition of insulin or IL-3 engenders an increase in S2448 phosphorylation in a PI3K- and Akt-dependent manner (34, 35). However, an mTOR mutant protein possessing an alanine substitution at this site retains the ability to activate S6K1 (a downstream effector of mTOR, see below) after growth factor stimulation (34). Thus, the role of this phosphorylation event in the regulation of mTOR activity is not clear. Inactivation of the TOR proteins, or rapamycin treatment, mimics nutrient deprivation in yeast, Drosophila, and mammalian cells (21, 36–38). Thus, a current working model for TOR signaling proposes that these kinases relay a permissive signal to downstream targets only in the presence of sufficient nutrients to fuel protein synthesis (Fig. 1). In some cases, the TOR proteins appear to function in a coregulatory capacity with other conventional, linear signaling pathways (such as the PI3K pathway; see below). In this way, a passive nutrient sufficiency signal may be combined with stimulatory signaling from a second pathway to coordinate cellular processes that require the uptake of nutrients. The absence of either signal is predicted to prohibit activation of downstream targets. A Model for TOR Signaling. How does TOR signal to downstream effectors? TOR signaling is thought to be effected through a combination of direct phosphorylation of downstream targets, and repression of phosphatase activity (Fig. 2). Genetic screening in S. cerevisiae has identified the PP2A-like phosphatase Sit4p, two PP2A regulatory subunits (CDC55 and TPD3), and a phosphatase-associated protein (Tap42p), as components of a rapamycin-sensitive signaling pathway (38, 39). Tap42p interacts directly with the catalytic subunits of PP2A and Sit4p. S. cerevisiae expressing a temperature-sensitive Tap42 mutant protein exhibit a dramatic defect in translation initiation at the nonpermissive temperature (39). Thus, Tap42p is thought to repress PP2A (or Sit4p) activity (also see refs. 40 and 41). Phosphorylation of Tap42p regulates its interaction with phosphatases. Whereas phosphorylated Tap42p competes with the phosphatase adapter (A) subunit for binding to the catalytic subunit, dephosphorylated Tap42p does not efficiently compete for binding (42). Tap42p phosphorylation is modulated by Tor signaling. The Tap42p-PP2A association in vivo is disrupted by nutrient deprivation or rapamycin treatment (39, 42). Further, a yeast Tor2p immunoprecipitate can phosphorylate Tap42p in vitro (42), and Tap42p phosphorylation is rendered rapamycin resistant in yeast strains expressing a rapamycin-resistant Tor1 protein (42). Tap42 orthologs are found in Arabidopsis (43), Drosophila, (GenBank accession number AAF53289), and mammalian cells (44, 45). The B cell receptor binding protein α4 (a.k.a Ig binding protein 1, IGBP1) is the mammalian ortholog of Tap42p (44, 45). The ability of this protein to interact with PP2A-like phosphatases is conserved in mammals, as a4 binds directly to the catalytic subunits of PP2A (46, 47), PP4, and PP6 (48, 49). Like Tap42p, α4 is also a phosphoprotein, and the α4-PP2A interaction was reported to be abrogated by rapamycin treatment (although this finding remains somewhat controversial; refs. 46 and 47). These observations suggest that Tap42p/a4 phosphorylation, and PP2A binding, are regulated by TOR signaling, and that an inhibition in TOR signaling leads to Tap42p/α4 dephosphorylation, dissociation of the Tap42p/α4-phosphatase complex, and phosphatase derepression. Interestingly, mTOR was reported to undergo nucleocytoplasmic shuttling (50). Abrogation of shuttling (by treatment with leptomycin B, a specific inhibitor of the nuclear export receptor Crml, or by transfection of mTOR tagged with exogenous nuclear export or import signals) was demonstrated to inhibit signaling to S6K1 and 4E-BP1 (50). Why mTOR shuttling may be important for 4E-BP1 and S6K1 activity is unknown (50). TOR SIGNALING MODULATES THE PHOSPHORYLATION STATE OF PROTEINS INVOLVED IN TRANSLATIONAL CONTROL Tor and Translation in S. cerevisiae. Inhibition of Tor activity in yeast potently represses translation initiation, concomitant with polysome disaggregation and cell cycle arrest in G1 (36). The mechanism for this translational repression is not understood, but could be due, at least in some strains, to the degradation of the initiation factor eIF4G (51, 52). A putative regulator of yeast eIF4E function, termed Eap1p (eIF4E-associated protein 1), may also be involved in this process, as disruption of the EAP1 gene results in partial rapamycin resistance (53). The G1 arrest in response to Tor inactivation was suggested to be due to the inhibition of translation of an mRNA coding for a cyclin involved in G1 to S progression, CLN3, because the cell cycle block can be overcome by forced expression of CLN3 (54–56). TOR and Translation in Mammalian Cells. TOR activity also regulates translation in mammalian cells (reviewed in refs. 57, 58, and



7039

59). However, whereas rapamycin treatment of S. cerevisiae leads to a precipitous disaggregation of polysomes, rapamycin treatment of mammalian cells specifically inhibits only the translation of certain classes of mRNAs. As detailed below, mTOR is thought to modulate translation of these mRNAs via the regulation of the phosphorylation state of several different translation effector proteins (Fig. 2).

Fig. 2. Signaling to eukaryotic translation initiation and elongation factors. mTOR signaling, in combination with the PI3K pathway, activates the translation of rapamycin-sensitive mRNAs. In the presence of sufficient nutrients to fuel protein synthesis, mTOR and PI3K signaling activate the S6Ks, and one or more unknown kinases, to effect phosphorylation of the ribosomal S6 protein, eIF4B, eIF4GI, and the4E-BPs. In response to agents that raise intracellular Ca2+ (such as glutamate or NMDA), a specific Ca2+/CaM-dependent kinase effects the phosphorylation of eEF2 to inhibit elongation. mTOR signaling has been reported to inhibit eEF2 phosphorylation (possibly via inhibition of the eEF2 kinase), and thus, to increase elongation rates. Phosphatases have been implicated in the dephosphorylation of several translation effectors, but are not depicted in this figure. The S6Ks. The ribosomal S6 kinases (S6K1 and S6K2) regulate the translation of a group of mRNAs possessing a 5 terminal oligopyrimidine tract (5’TOP), a stretch of 4–14 pyrimidines found at the extreme 5 terminus of ribosomal protein mRNAs, and mRNAs coding for other components of the translation machinery (reviewed in ref. 60). When nutrient levels are low, the translation of 5TOPcontaining mRNAs is repressed. Even in the presence of sufficient nutrients, translation of 5TOP-containing mRNAs is inhibited by rapamycin treatment (reviewed in ref. 59). 5TOP-containing mRNAs are present in mammalian and Drosophila cells (61), and comprise a significant amount of the total mRNA. The mechanism for 5TOP regulation is not understood; however, two S6K substrates that could play a role in the modulation of 5 TOP translation are the ribosomal S6 protein and the translation initiation factor eIF4B (see below; reviewed in refs. 59 and 62). S6K activity is inhibited by both PI3K inhibitors and rapamycin, indicating that both PI3K and mTOR signaling are required for S6K activation (63). A key finding in the understanding of this signaling module is that the PI3K and mTOR inputs to S6K1 can be separated. Deletion of an N-terminal S6K1 fragment confers rapamycin resistance to the S6K1 protein, yet this truncation mutant remains sensitive to treatment with PI3K inhibitors (64, 65). These data thus argue against a linear pathway to S6K1 comprised of PI3K and mTOR, but instead suggest that two separate inputs are required for S6K activation. 4E-BPs. Protein synthesis is regulated in many instances at the initiation phase (Fig. 3), the stage during which a ribosome is recruited to the 5 end of an mRNA, and positioned at a start codon (reviewed in ref. 66). The eukaryotic ribosome does not have the ability to locate and bind to the 5 end of an mRNA; it must rely on a number of translation initiation factors to guide it there. The mRNA 5 end is distinguished by the presence of a “cap” (the structure m7GpppN, in which m is a methyl group and N any nucleotide), which is specifically recognized by the eukaryotic translation initiation factor 4E (eIF4E). eIF4E, via an interaction with one of two large scaffolding proteins, termed eIF4GI and eIF4GII, directs the translation machinery to the 5 end of the mRNA (reviewed in refs. 57, 58, and 62). The interaction between eIF4E and eIF4G is regulated in mammalian and Drosophila cells by a family of translation repressor peptides, the eIF4E-binding proteins (4E-BPs; refs. 67 and 68–71). The 4E-BPs compete with the eIF4G proteins for an overlapping binding site on eIF4E, such that binding of a 4E-BP or an eIF4G protein to eIF4E is mutually exclusive (72–74). Binding of the mammalian and Drosophila 4E-BPs to eIF4E is regulated by phosphorylation (67, 68, 70). Whereas hypophosphorylated 4E-BPs bind with high affinity to eIF4E, 4E-BP hyperphosphorylation abrogates this interaction. As is the case with S6K1, the PI3K and TOR signaling pathways modulate 4E-BP phosphorylation. Immunoprecipitates of mTOR phosphorylate two “priming” sites in the mammalian 4E-BP1 protein in vitro (75–77). This phosphorylation event is thought to be required for subsequent PI3K-dependent phosphorylation of other S/T residues, resulting in release from eIF4E (refs. 77, 79, and 80; A.-C.G., B.R., S.P. Gygi, A.Niedzwiecka, M.Miron, S. K.Burley, R.D.Polakiewicz, A.Wyslouch-Cieszynska, and R. Aebersold, unpublished observations). Using a panel of pharmacological inhibitors, the D. melanogaster 4E-BP ortholog was also demonstrated to lie downstream of dTOR and dPI3K (70). eIF4GI. Two eIF4G homologs have been identified in mammalian cells (81, 82). Both eIF4GI and eIF4GII are phosphoproteins (83). Whereas the intracellular signaling pathways that modulate the phosphorylation of eIF4GII have not been elucidated, phosphorylation of the eIF4GI isoform is regulated by mTOR and PI3K signaling. Three phosphorylation sites (S1108, S1148, and S1 192) were demarcated in a C-terminal eIF4GI “hinge” region. Phosphorylation of the hinge residues is elevated by serum or insulin treatment, and is inhibited by rapamycin or PI3K inhibitors (83). However, neither mTOR nor S6Ks can directly phosphorylate the hinge residues in vitro. Interestingly, eIF4GI proteins truncated at their N termini are constitutively phosphorylated on the hinge residues, even in the presence of PI3K or mTOR inhibitors (83). Thus, rapamycin-insensitive kinases appear to phosphorylate these residues, but an amino-terminal domain could regulate the accessibility of the hinge phosphorylation sites to these kinases in a rapamycin-sensitive manner. The function of these phosphorylation events is unclear.



7040

The hinge region residues do not overlap with binding sites for any known eIF4GI interacting protein, and no differences in the interaction of eIF4GI with several known binding partners were observed for eIF4GI isolated from serum-starved vs. serumstimulated cells. It was thus suggested that phosphorylation could effect changes in eIF4GI structure, to increase eIF4GI activity toward rapamycin-sensitive mRNAs (83).

Fig. 3. The initiation and elongation phases of translation in eukaryotes. In starved or stressed cells, the cap binding protein eIF4E is sequestered by hypophosphorylated 4E-BPs. In growing or stimulated cells, the 4E-BPs are hyperphosphorylated to release eIF4E, such that it can interact with the scaffolding protein, eIF4G. In conjunction with the RNA helicase eIF4A and the cofactor eIF4B, 5 secondary structure is melted, and a small ribosomal subunit is recruited to a single-stranded, cap-proximal region of an mRNA via an interaction between eIF4G and the ribosome-associated factor eIF3. The small ribosomal subunit, along with a ternary complex composed of eIF2, GTP, and Met-tRNAi, then scans the mRNA in a 5 to 3 direction until an AUG start codon in the proper sequence context is encountered. At this point, initiation factors are released, and the large ribosomal subunit is recruited. The elongation factors catalyze aminoacyl-tRNA binding to ribosomes, and the translocation of the mRNA from the ribosomal A site to the P site. eIF4B. eIF4B is a ubiquitous protein that dramatically stimulates the activity of eIF4A, an RNA helicase thought to unwind mRNA 5 secondary structure (84). Mammalian eIF4B is a phosphoprotein (85), and treatment of cells with serum, insulin, or phorbol esters results in eIF4B hyperphosphorylation (62, 86). eIF4B can be phosphorylated in vitro with several different kinases, including S6K1 (refs. 87 and 88; F.Peiretti and J.W.B. Hershey, personal communication). Two-dimensional tryptic phosphopeptide mapping has revealed that eIF4B possesses at least one serum-stimulated phosphorylation site that is sensitive to rapamycin and inhibitors of PI3K (B.R., F.Peiretti, A.-C.G., and J.W.B.Hershey, unpublished observations). Thus, PI3K and mTOR also appear to signal to eIF4B. Unlike the 4E-BPs and eIF4GI, however, eIF4B appears to be a direct target of the S6Ks. eEF2. Another level at which translation may be modulated in eukaryotes is the elongation phase (Fig. 3). The eukaryotic elongation factors (eEFs) 1 and 2 regulate this process (reviewed in ref. 89). eEF1 promotes aminoacyl-tRNA binding to ribosomes, whereas eEF2 promotes the translocation of the mRNA from the ribosomal A site to the P site (90). Phosphorylation of eEF2 by a specific Ca2+/CaMdependent kinase inhibits eEF2 activity (reviewed in ref. 89). Amino acid withdrawal from cultured mammalian cells results in a marked increase in eEF2 phosphorylation, accompanied by a decrease in elongation rates (e.g., ref. 91). Many agents that raise intracellular Ca2+ concentrations also bring about eEF2 phosphorylation, including histamine treatment of epithelial cells (92, 93), or glutamate or N-methylD-aspartate (NMDA) treatment of neurons (see below). Conversely, insulin stimulation leads to eEF2 dephosphorylation, resulting in a decrease in ribosomal transit time, and an increase in elongation rates (94–96). Rapamycin treatment inhibits the insulin-stimulated dephosphorylation of eEF2 (91, 96). Thus, eEF2 phosphorylation also appears to be modulated by mTOR signaling. How mTOR signaling regulates eEF2 activity is unknown; mTOR has been proposed to regulate the activity of the eEF2 kinase and/or to modulate the dephosphorylation of eEF2 via regulation of PP2A activity (89, 97). As discussed further below, eEF2 activity has been implicated in the control of protein synthesis in neurons. In sum, mTOR signaling regulates the phosphorylation state of many proteins involved in translation control, including the S6 kinases, the translation initiation factors eIF4B and eIF4GI, the translation elongation factor eEF2, and a family of translation inhibitory proteins, the 4E-BPs. Many other translation factors are also known to be phosphoproteins (62), but the pathways modulating the phosphorylation state of these factors have not been studied. Additional proteins involved in translation control may thus also be downstream of mTOR. TOR REGULATES THE ABUNDANCE OF THE TRANSLATION MACHINERY In addition to its effect on the phosphorylation state of proteins involved in translational control, TOR signaling regulates the abundance of the components of the translation machinery (Fig.1), at both the transcriptional and translational levels. The



7041

number of ribosomes in a given cell can vary dramatically, according to growth conditions (reviewed in ref. 98). Actively growing cells require numerous ribosomes (e.g., logarithmically dividing yeast cells produce 2000 ribosomes/minute), and ribosome synthesis represents a major energy expenditure for the cell (98). In S. cerevisiae, ribosome biosynthesis requires the transcription of over 100 different genes, involving all three RNA polymerases (98). In response to nutrient availability, TOR signaling in S. cerevisiae regulates the transcription of rRNA by Pol I and Pol III (52, 99), and the transcription of ribosomal protein mRNAs by Pol II (41, 52, 100, 101). TOR signaling has also been implicated in the processing of the ribosomal 35S precursor rRNA (52). When nutrients are limiting, ribosome production is curtailed (or a cell may even begin to degrade ribosomes, in a scavenging process termed autophagy; see below). The abundance of several yeast translation factors was also demonstrated to be regulated by the TOR pathway (52). Transcriptional modulation in S. cerevisiae is responsible for a decrease in the mRNA levels of initiation and elongation factors after rapamycin treatment, although the extent of this transcriptional inhibition is less than that observed for the ribosomal proteins (52). Through the S6Ks, mTOR signaling regulates the translation of ribosomal protein mRNAs in mammalian cells (102, 103). In Drosophila and mammalian cells, translation of the elongation factor mRNAs, and mRNAs coding for other proteins involved in translation, such as the poly(A) binding protein, is also regulated by the presence of the 5TOP element (reviewed in ref. 60). Thus, the TOR pathway simultaneously regulates the abundance and activity of the translation machinery in both unicellular and multicellular organisms. TOR AS A MASTER SWITCH FOR CATABOLISM VS. ANABOLISM In yeast, TOR signaling has been demonstrated to coordinate the activity of various metabolic pathways in response to nutrient quality (Fig. 1). In particular, TOR signaling modulates the transcription of genes involved in amino acid biosynthesis, and regulates the activity of amino acid permeases. In both yeast and mammalian cells, TOR signaling regulates autophagy. Nutrient-Sensitive Transcriptional Regulation. Switching yeast cells to a poor carbon or nitrogen source induces a state of quiescence (G0). Whereas the transcription of many genes is inhibited after a switch from a rich to a poor nitrogen or carbon source (or after rapamycin treatment), global mRNA profiling has revealed that the transcription of mRNAs coding for proteins involved in nutrient utilization, respiration, and protein degradation is actually augmented (41, 100, 101, 104). Tor signaling modulates gene expression via cytoplasmic sequestration of several nutrient-responsive transcription factors. For example, the GATA transcription factor Gln3p is retained in the cytoplasm through an interaction with the Ure2 protein, whereas the zinc-fingercontaining transcription factors Msn2p and Msn4p are sequestered in the cytoplasm via an interaction with the 14–3-3 protein Bmh2p (reviewed in ref. 38). Starvation abrogates Tor signaling and results in a loss of cytoplasmic retention of Gln3p, Msn2p, and Msn4p, followed by nuclear translocation and transcription of various target genes (38, 41). Tor signals to several specific effectors (Tap42, Mks1p, Ure2p, Gln3p, and Gat1p) to elicit changes in the expression levels of enzymes involved in several different metabolic pathways (104, 105). How TOR signaling may affect the transcription rates of metabolic enzymes in multicellular organisms has not yet been elucidated. Amino Acid Permeases. Permeases are necessary for nutrient uptake, and may be divided into two functional classes. One class is regulated in response to the available nitrogen source (e.g., the general amino acid permease Gap1p), and members of this class transport amino acids to be used as a nitrogen source. The second class mainly consists of high affinity permeases, which specifically transport one or a small group of related amino acids to be used as building blocks for protein synthesis. In starved yeast cells, or in cells treated with rapamycin, ubiquitination and degradation of the high affinity tryptophan permease Tat2p is induced, leading to a decrease in tryptophan import (40, 106). This phenomenon is not unique to Tat2p, as a histidine permease (Hip1p) is also degraded upon nutrient deprivation or rapamycin treatment (106). In contrast, rapamycin treatment increases the abundance of the general permease Gap1p, indicating that TOR signaling inversely regulates the two classes of permeases (106). TOR regulation of permeases is mediated through the serine/ threonine kinase Npr1p, whose phosphorylation is regulated by the Tor proteins and Tap42p, in a manner similar to the regulation of S6Ks and 4EBPs in mammalian cells (40). Autophagy. When nutrient levels are low, eukaryotic cells degrade cytoplasmic proteins and organelles to scavenge amino acids, in a process termed autophagy (107–109). Autophagy involves the sequestration of a portion of cytoplasm by a double (or multi) layered membrane structure termed the autophagosome or autophagic vacuole. This structure fuses with lysosomal or endosomal membranes, resulting in the degradation of cytoplasmic components. The TOR proteins regulate autophagy. Rapamycin addition to yeast cultures or to mammalian cells in culture induces autophagy, even in a nutrient-rich medium (110, 111). Shifting a temperature-sensitive TOR2 yeast mutant to the nonpermissive temperature also induces autophagy (110). In mammalian cells, autophagy is inhibited by amino acids and insulin. Activation of S6K is associated with inhibition of autophagy in rat hepatocytes, and the inhibition of autophagy by amino acids could be partially prevented by rapamycin treatment (111, 112). In sum, the TOR proteins appear to act as master regulators of the balance between protein synthesis and degradation. In the presence of sufficient nutrients to fuel protein synthesis, TOR provides a permissive signal to translation, ribosome biosynthesis, and high affinity amino acid permeases, while repressing autophagy and the general amino acid permeases. In the absence of TOR signaling, the translation of mRNAs coding for components of the translation machinery is specifically inhibited, ribosome biosynthesis is blocked, and autophagy is activated. HOW MIGHT TOR SIGNALING BE INVOLVED IN LEARNING AND MEMORY? The observation that rapamycin can inhibit long-term facilitation in Aplysia neurons has implicated TOR signaling in the control of neuronal protein synthesis (11). How might a kinase involved in the regulation of protein metabolism also be involved in learning and memory? In fact, several putative links have been established between TOR and neuronal function. Several types of neurotransmitters were described to affect the activity of the rapamycin-sensitive pathway leading to S6K1 and 4EBP1 phosphorylation. Serotonin (5-HT) addition to Aplysia neurons or Chinese hamster ovary (CHO) cells expressing the 5-HT1B receptor increases phosphorylation of S6K1 in a rapamycin-dependent manner (113, 114). Dopamine addition to CHO cells also activates S6K1 in a rapamycin-dependent manner (115). Finally, both S6K1 and 4E-BP1 phosphorylation is induced by stimulation of the µ-opioid receptors (which mediate the analgesic and addictive properties of morphine) by the agonist [D-Ala2, N-MePhe4,Gly5-ol] enkephalin (DAMGO; ref. 116). mTOR interacts with gephyrin, a tubulin-binding protein involved in neuronal γ-aminobutyric acid type A (GABAA) and glycine receptor clustering (117–120). Gephyrin binding was reported to be required for signaling to S6K1 and 4E-BP1, and, consistent with a role in localized protein synthesis, a fraction



7042

ation experiment demonstrated that mTOR and gephyrin were enriched in the synaptosomal fraction, but not the postsynaptic density fraction (117). Another possible connection between mTOR signaling and localized translation is via the modulation of eEF2 phosphorylation. Several studies have noted an increase in eEF2 phosphorylation in response to various neurotransmitters. For example, glutamate or NMDA treatment of cortical neurons in culture leads to a rapid and pronounced increase in eEF2 phosphorylation, and a decrease in translation rates in cell bodies and proximal (but not distal) cell processes (121). Activation of the NMDA receptor also leads to eEF2 phosphorylation, in tadpole tecta (122). It is tempting to speculate that mTOR could inhibit eEF2 phosphorylation in active synapses to locally derepress translation. It has also been suggested that eEF2 phosphorylation could actually enhance the translation of specific mRNAs localized to dendrites by driving these mRNAs from untranslated ribonucleotide particles or small polysomes into larger polysomes (122–125). Another possible link between TOR and neuronal function is the regulation of autophagy. In addition to nutrient scavenging during starvation, autophagy has been demonstrated to play an important role in developmental processes that involve cellular remodeling, such as insect metamorphosis (126) or luteal regression (127). Whereas neuronal death certainly involves apoptosis (128), several reports have suggested that an alternative form of cell death may occur in some nerve cells. For example, nerve growth factor (NGF)-deprivation of sympathetic neurons was reported to induce a rapid, 30-fold increase in autophagic particles, before any signs of DNA fragmentation (a hallmark of apoptosis) were observed. Treatment of these cells with the anti-autophagic drug 3-methyladenine delayed cell death (129). In another study, autophagic vacuoles were observed in PC12 cells 3 h after serum starvation, whereas chromatin condensation did not occur until 6 h poststarvation (130). Finally, the removal of specific spinal cord neurons in Xenopus tadpoles (a normal developmental process during metamorphosis) was also suggested to occur through autophagy-directed cell death (131). Intriguingly, elevated levels of autophagy have been reported to be associated with neurodegenerative disorders such as Parkinson’s disease (132). TOR ACTIVITY IS REQUIRED FOR MURINE FOREBRAIN DEVELOPMENT A recently described mouse mutant suggests that mTOR plays a critical role in embryonic brain development (133; K.Hentges and A.Peterson, personal communication). The murine flat top mutation was isolated in a chemical mutagenesis screen designed to identify genes involved in embryonic telencephalic development (133). Flat top defects include a failure of the embryo to up-regulate proliferation in the telencephalic primordium, and a failure to establish dorsal and ventral domains of gene expression in the developing telencephalon. Homozygous mutant embryos fail to rotate around the body axis, and die in utero (78). The flat top mutation was mapped to a single nucleotide change in an mTOR intron, which leads to aberrant splicing. The protein products derived from these abnormally spliced mRNAs appear to be inactive (or much less active), because of the presence of a 3-aa insertion or 3-aa deletion at the intron-exon junction. Transgenic rescue experiments confirmed that mTOR is the affected gene in this animal, and a rapamycin injection regimen during pregnancy yields embryos with an identical phenotype (K. Hentges and A.Peterson, personal communication). Whether the brain defect is the result of a failure to inhibit autophagy, or is elicited through some other function of mTOR is unknown. S6K1 activity was demonstrated to be significantly lower (17% of wild-type levels) in flat-top embryos, but effects on other translation factors have not yet been determined. The flat top mouse should provide a very valuable tool for the study of TOR function in mammalian cells. SUMMARY AND FUTURE PROSPECTS The TORs are evolutionarily conserved protein kinases that regulate the balance between protein synthesis and degradation in unicellular and multicellular organisms. This complex balance is maintained via the regulation of translation initiation and elongation factor activity, the modulation of ribosome biosynthesis at both the transcriptional and translational levels, the control of amino acid permease activity, the coordination of the transcription of many enzymes involved in various metabolic pathways, and the control of autophagy. An interesting and unexpected finding was that mTOR also appears to play a critical role in embryonic brain development, learning, and memory formation. There is still much to be learned. For instance, how the TOR proteins sense the quality or quantity of nutrients is unknown. The mammalian GCN2 kinase, which senses intracellular amino acid levels by binding to deacylated tRNAs, does not appear to play a role in this process, because amino acid withdrawal leads to S6K1 and 4E-BP1 dephosphorylation even in GCN2 null cells (C.Jousse and D.Ron, personal communication). Further, whereas the role of the TOR proteins in the control of metabolic enzymes and amino acid permeases in S. cerevisiae is now well documented, similar studies have not been conducted for the mammalian and Drosophila systems. The recent description of the Drosophila TOR homolog (dTOR; refs. 21 and 22) and the isolation of the murine flat top mTOR mutant (133) should provide invaluable tools for further dissection of the TOR signaling module in multicellular organisms. We thank A.Peterson, K.Hentges, C.Jousse, D.Ron, F.Peiretti, and J.W.B.Hershey for sharing unpublished data, and W.S.Sossin, F. Poulin, P.F.Cho-Park, and M.Miron for critical reading of the manuscript. Work in the authors’ laboratory is supported by grants from the Canadian Institutes of Health Research, the National Cancer Institute of Canada, the Howard Hughes Medical Institute (HHMI), and the Human Frontier Science Program. B.R. is supported by a Medical Research Council (MRC) of Canada postdoctoral fellowship. A.-C.G. is supported by an MRC of Canada doctoral fellowship. N.S. is an MRC of Canada Distinguished Scientist and an HHMI International Scholar. 1. Goelet, P., Castellucci, V.F., Schacher, S. & Kandel, E.R. (1986) Nature (London) 322, 419–422. 2. Bailey, C.H., Bartsch, D. & Kandel, E.R. (1996) Proc. Natl. Acad. Sci. USA 93, 13445–13452. 3. Sossin, W. S. (1996) Trends Neurosci. 19, 215–218. 4. Frey, U. & Morris, R.G. (1998) Trends Neurosci. 21, 181–188. 5. Steward, O. & Levy, W.B. (1982) J. Neurosci. 2, 284–291. 6. Tiedge, H. & Brosius, J. (1996) J. Neurosci. 16, 7171–7181. 7. Torre, E.R. & Steward, O. (1992) J. Neurosci. 12, 762–772. 8. Weiler, I.J. & Greenough, W.T. (1993) Proc. Natl. Acad. Sci. USA 90, 7168–7171. 9. Kang, H. & Schuman, E.M. (1996) Science 273, 1402–1406. 10. Martin, K. C., Casadio, A., Zhu, H., E, Y., Rose, J.C, Chen, M., Bailey, C.H. & Kandel, E.R. (1997) Cell 91, 927–938. 11. Casadio, A., Martin, K.C., Giustetto, M., Zhu, H., Chen, M., Bartsch, D., Bailey, C.H. & Kandel, E.R. (1999) Cell 99, 221–237. 12. Vezina, C., Kudelski, A. & Sehgal, S.N. (1975) J. Antibiot. (Tokyo) 28, 721–726. 13. Harding, M.W., Galat, A., Uehling, D.E. & Schreiber, S.L. (1989) Nature (London) 341, 758–760. 14. Siekierka, J.J., Hung, S.H., Poe, M., Lin, C.S. & Sigal, N.H. (1989) Nature (London) 341, 755–757. 15. Siekierka, J.J., Wiederrecht, G., Greulich, H., Boulton, D., Hung, S.H., Cryan, J., Hodges, P.J. & Sigal, N.H. (1990) J. Biol. Chem. 265, 21011– 21015. 16. Cafferkey, R., Young, P.R., McLaughlin, M.M., Bergsma, D.J., Koltin, Y., Sathe, G.M., Faucette, L., Eng, W.K., Johnson, R.K. & Livi, G.P. (1993) Mol. Cell. Biol. 13, 6012–6023. 17. Heitman, J., Movva, N.R. & Hall, M.N. (1991) Science 253, 905–909.



7043

18. Helliwell, S.B., Wagner, P., Kunz, J., Deuter-Reinhard, M., Henriquez, R. & Hall, M.N. (1994) Mol. Biol. Cell 5, 105–118. 19. Kunz, J., Henriquez, R., Schneider, U., Deuter-Reinhard, M., Movva, N.R. & Hall, M.N. (1993) Cell 73, 585–596. 20. Weisman, R. & Choder, M. (2001) J. Biol. Chem. 276, 7027–7032. 21. Zhang, H., Stallock, J.P., Ng, J.C, Reinhard, C. & Neufeld, T.P. (2000) Genes Dev. 14, 2712–2724. 22. Oldham, S., Montagne, J., Radimerski, T., Thomas, G. & Hafen, E. (2000) Genes Dev. 14, 2689–2694. 23. Brown, E. J., Albers, M.W., Shin, T.B., Ichikawa, K., Keith, C.T., Lane, W.S. & Schreiber, S.L. (1994) Nature (London) 369, 756–758. 24. Chen, Y., Chen, H., Rhoad, A.E., Warner, L., Caggiano, T.J., Failli, A., Zhang, H., Hsiao, C.L., Nakanishi, K. & Molnar-Kimber, K.L. (1994) Biochem. Biophys. Res. Commun. 203, 1–7. 25. Chiu, M.I., Katz, H. & Berlin, V. (1994) Proc. Natl. Acad. Sci. USA 91, 12574–12578. 26. Sabatini, D.M., Erdjument-Bromage, H., Lui, M., Tempst, P. & Snyder, S.H. (1994) Cell 78, 35–43. 27. Sabers, C.J., Martin, M.M., Brunn, G.J., Williams, J.M., Dumont, F.J., Wiederrecht, G. & Abraham, R.T. (1995) J. Biol. Chem. 270, 815–822. 28. Abraham, R.T. & Wiederrecht, G.J. (1996) Annu. Rev. Immunol. 14, 483–510. 29. Keith, C.T. & Schreiber, S.L. (1995) Science 270, 50–51. 30. Hunter, T. (1995) Cell 83, 1–4. 31. Hoekstra, M.F. (1997) Curr. Opin. Genet. Dev. 7, 170–175. 32. Scott, P.H., Brunn, G.J., Kohn, A.D., Roth, R.A. & Lawrence, J.C., Jr. (1998) Proc. Natl. Acad. Sci. USA 95, 7772–7777. 33. Peterson, R.T., Beal, P.A., Comb, M.J. & Schreiber, S.L. (2000) J. Biol. Chem. 275, 7416–7423. 34. Sekulic, A., Hudson, C.C., Homme, J.L., Yin, P., Otterness, D.M., Karnitz, L.M. & Abraham, R.T. (2000) Cancer Res. 60, 3504–3513. 35. Navé, B.T., Ouwens, M., Withers, D.J., Alessi, D.R. & Shepherd, P.R. (1999) Biochem. J. 344, 427–431. 36. Barbet, N.C., Schneider, U., Helliwell, S.B., Stansfield, I., Tuite, M.F. & Hall, M.N. (1996) Mol. Biol. Cell 7, 25–42. 37. Dennis, P.B., Fumagalli, S. & Thomas, G. (1999) Curr. Opin. Genet. Dev. 9, 49–54. 38. Schmelzle, T. & Hall, M.N. (2000) Cell 103, 253–262. 39. Di Como, C.J. & Arndt, K.T. (1996) Genes Dev. 10, 1904–1916. 40. Schmidt, A., Beck, T., Koller, A., Kunz, J. & Hall, M.N. (1998) EMBO J. 17, 6924–6931. 41. Beck, T. & Hall, M.N. (1999) Nature (London) 402, 689–692. 42. Jiang, Y. & Broach, J.R. (1999) EMBO J. 18, 2782–2792. 43. Harris, D.M., Myrick, T.L. & Rundle, S.J. (1999) Plant Physiol. 121, 609–617. 44. Inui, S., Kuwahara, K., Mizutani, J., Maeda, K., Kawai, T., Nakayasu, H. & Sakaguchi, N. (1995) J. Immunol. 154, 2714–2723. 45. Onda, M., Inui, S., Maeda, K., Suzuki, M., Takahashi, E. & Sakaguchi, N. (1997) Genomics 46, 373–378. 46. Inui, S., Sanjo, H., Maeda, K., Yamamoto, H., Miyamoto, E. & Sakaguchi, N. (1998) Blood 92, 539–546. 47. Murata, K., Wu, J. & Brautigan, D.L. (1997) Proc. Natl. Acad. Sci. USA 94, 10624–10629. 48. Chen, J., Peterson, R.T. & Schreiber, S.L. (1998) Biochem. Biophys. Res. Commun. 247, 827–832. 49. Nanahoshi, M., Tsujishita, Y., Tokunaga, C., Inui, S., Sakaguchi, N., Hara, K. & Yonezawa, K. (1999) FEBS Lett. 446, 108–112. 50. Kim, J.E. & Chen, J. (2000) Proc. Natl. Acad. Sci. USA 97, 14340–14345. (First Published December 12, 2000; 10.1073/pnas.011511898) 51. Berset, C., Trachsel, H. & Altmann, M. (1998) Proc. Natl. Acad. Sci. USA 95, 4264–4269. 52. Powers, T. & Walter, P. (1999) Mol. Biol. Cell 10, 987–1000. 53. Cosentino, G.P., Schmelzle, T., Haghighat, A., Helliwell, S.B., Hall, M.N. & Sonenberg, N. (2000) Mol. Cell. Biol. 20, 4604–4613. 54. Polymenis, M. & Schmidt, E.V. (1997) Genes Dev. 11, 2522–2531. 55. Gallego, C., Gari, E., Colomina, N., Herrero, E. & Aldea, M. (1997) EMBO J. 16, 7196–7206. 56. Danaie, P., Altmann, M., Hall, M.N., Trachsel, H. & Helliwell, S.B. (1999) Biochem. J. 340, 135–141. 57. Gingras, A.-C., Raught, B. & Sonenberg, N. (1999) Annu. Rev. Biochem. 68, 913–963. 58. Raught, B., Gingras, A.-C. & Sonenberg, N. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 245–294. 59. Fumagalli, S. & Thomas, G. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 695–718. 60. Meyuhas, O. & Hornstein, E. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 671–694. 61. Meyuhas, O., Avni, D. & Shama, S. (1996) in Translational Control, eds. Hershey, J.W.B., Mathews, M. & Sonenberg, N. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 363–388. 62. Hershey, J.W.B. & Merrick, W.C. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 33–88. 63. Thomas, G. & Hall, M.N. (1997) Curr. Opin. Cell Biol. 9, 782–787. 64. Mahalingam, M. & Templeton, D. (1996) Mol. Cell. Biol. 16, 405–413. 65. Dennis, P.B., Pullen, N., Kozma, S.C. & Thomas, G. (1996) Mol. Cell. Biol. 16, 6242–6251. 66. Mathews, M.B., Sonenberg, N. & Hershey, J.W.B. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), Vol. 1, pp. 1–31. 67. Pause, A., Belsham, G.J., Gingras, A.-C., Donzé, O., Lin, T.A., Lawrence, J.C., Jr., & Sonenberg, N. (1994) Nature (London) 371, 762–767. 68. Lin, T.A., Kong, X., Haystead, T.A., Pause, A., Belsham, G., Sonenberg, N. & Lawrence, J. C., Jr. (1994) Science 266, 653–656. 69. Bernal, A. & Kimbrell, D.A. (2000) Proc. Natl. Acad. Sci. USA 97, 6019–6024. (First Published May 16, 2000; 10.1073/pnas.100391597) 70. Miron, M., Verdœ, J., Lachance, P.E.D., Birnbaum, M.J., Lasko, P.F. & Sonenberg, N. (2001) Nat. Cell Biol., in press. 71. Poulin, F., Gingras, A.-C., Olsen, H., Chevalier, S. & Sonenberg, N. (1998) J. Biol. Chem. 273, 14002–14007. 72. Mader, S., Lee, H., Pause, A. & Sonenberg, N. (1995) Mol. Cell. Biol. 15, 4990–4997. 73. Haghighat, A., Mader, S., Pause, A. & Sonenberg, N. (1995) EMBO J. 14, 5701–5709. 74. Marcotrigiano, J., Gingras, A.-C., Sonenberg, N. & Burley, S.K. (1999) Mol. Cell 3, 707–716. 75. Brunn, G.J., Hudson, C.C., Sekulic, A., Williams, J.M., Hosoi, H., Houghton, P.J., Lawrence, J.C, Jr., & Abraham, R.T. (1997) Science 277, 99–101. 76. Burnett, P.E., Barrow, R.K., Cohen, N.A., Snyder, S.H. & Sabatini, D.M. (1998) Proc. Natl. Acad. Sci. USA 95, 1432–1437. 77. Gingras, A.-C., Gygi, S.P., Raught, B., Polakiewicz, R.D., Abraham, R.T., Hoekstra, M.F., Aebersold, R. & Sonenberg, N. (1999) Genes Dev. 13, 1422–1437. 78. Hentges, K., Thompson, K. & Peterson, A. (1999) Development 126, 1601– 1609. 79. Heesom, K.J. & Denton, R.M. (1999) FEBS Lett. 457, 489–493. 80. Mothe-Satney, I., Yang, D., Fadden, P., Haystead, T.A. & Lawrence, J.C., Jr. (2000) Mol. Cell. Biol. 20, 3558–3567. 81. Yan, R., Rychlik, W., Etchison, D. & Rhoads, R.E. (1992) J. Biol. Chem. 267, 23226–23231. 82. Gradi, A., Imataka, H., Svitkin, Y.V., Rom, E., Raught, B., Morino, S. & Sonenberg, N. (1998) Mol. Cell. Biol. 18, 334–342. 83. Raught, B., Gingras, A.-C., Gygi, S.P., Imataka, H., Morino, S., Gradi, A., Aebersold, R. & Sonenberg, N. (2000) EMBO J. 19, 434–444. 84. Rozen, F., Edery, I., Meerovitch, K., Dever, T.E., Merrick, W.C. & Sonenberg, N. (1990) Mol. Cell. Biol. 10, 1134–1144. 85. Benne, R., Edman, J., Traut, R.R. & Hershey, J.W.B. (1978) Proc. Natl. Acad. Sci. USA 75, 108–112. 86. Duncan, R. & Hershey, J.W. (1985) J. Biol. Chem. 260, 5493–5497. 87. Morley, S.J. & Traugh, J.A. (1990) J. Biol. Chem. 265, 10611–10616. 88. Morley, S.J. & Traugh, J.A. (1989) J. Biol. Chem. 264, 2401–2404. 89. Proud, C.G. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 719–739. 90. Merrick, W.C. & Nyborg, J. (2000) in Translational Control of Gene Expression, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 89–125. 91. Wang, X., Campbell, L.E., Miller, C.M. & Proud, C.G. (1998) Biochem. J. 334, 261–267. 92. Mackie, K.P., Nairn, A.C., Hampel, G., Lam, G. & Jaffe, E.A. (1989) J. Biol. Chem. 264, 1748–1753.



7044

93. Marino, M.W., Dunbar, J.D., Wu, L.W., Ngaiza, J.R, Han, H.M., Guo, D., Matsushita, M., Nairn, A.C., Zhang, Y., Kolesnick, R., et al. (1996) J. Biol. Chem. 271, 28624–28629. 94. Nielsen, P.J. & McConkey, E.H. (1980) J. Cell. Physiol. 104, 269–281. 95. Chang, Y.W. & Traugh, J.A. (1997) J. Biol. Chem. 272, 28252–28257. 96. Redpath, N.T., Foulstone, E.J. & Proud, C.G. (1996) EMBO J. 15, 2291–2297. 97. Wang, L., Wang, X. & Proud, C.G. (2000) Am. J. Physiol. Heart Circ. Physiol. 278, H1056-H1068. 98. Warner, J.R. (1999) Trends Biochem. Sci. 24, 437–440. 99. Zaragoza, D., Ghavidel, A., Heitman, J. & Schultz, M.C. (1998) Mol. Cell. Biol. 18, 4463–4470. 100. Hardwick, J. S., Kuruvilla, F.G., Tong, J.K., Shamji, A.F. & Schreiber, S.L. (1999) Proc. Natl. Acad. Sci. USA 96, 14866–14870. 101. Cardenas, M.E., Cutler, N.S., Lorenz, M.C., Di Como, C.J. & Heitman, J. (1999) Genes Dev. 13, 3271–3279. 102. Leicht, M., Simm, A., Bertsch, G. & Hoppe, J. (1996) Cell Growth Differ. 7, 1199–1209. 103. Mahajan, P.B. (1994) Int. J. Immunopharmacol. 16, 711–721. 104. Shamji, A.F., Kuruvilla, F.G. & Schreiber, S.L. (2000) Curr. Biol. 10, 1574–1581. 105. Chan, T.-F., Carvalho, J., Riles, L. & Zheng, X.F.S. (2000) Proc. Natl. Acad. Sci. USA 97, 13227–13232. (First Published November 14, 2000; 10.1073/ pnas.240444197) 106. Beck, T., Schmidt, A. & Hall, M.N. (1999) J. Cell Biol. 146, 1227–1238. 107. Klionsky, D.J. & Emr, S.D. (2000) Science 290, 1717–1721. 108. Kim, J. & Klionsky, D.J. (2000) Annu. Rev. Biochem. 69, 303–342. 109. Klionsky, D.J. & Ohsumi, Y. (1999) Annu. Rev. Cell Dev. Biol. 15, 1–32. 110. Noda, T. & Ohsumi, Y. (1998) J. Biol. Chem. 273, 3963–3966. 111. Blommaart, E.F., Luiken, J.J., Blommaart, P.J., van Woerkom, G.M. & Meijer, A.J. (1995) J. Biol. Chem. 270, 2320–2326. 112. Shigemitsu, K., Tsujishita, Y., Kara, K., Nanahoshi, M., Avruch, J. & Yonezawa, K. (1999) J. Biol. Chem. 274, 1058–1065. 113. Khan, A., Pepio, A. & Sossin, W. (2001) J. Neurosci. 21, 382–391. 114. Pullarkat, S.R., Mysels, D.J., Tan, M. & Cowen, D.S. (1998) J. Neurochem. 71, 1059–1067. 115. Welsh, G.I., Hall, D.A., Warnes, A., Strange, P.G. & Proud, C.G. (1998) J. Neurochem. 70, 2139–2146. 116. Polakiewicz, R.D., Schieferl, S.M., Gingras, A.C., Sonenberg, N. & Comb, M.J. (1998) J. Biol. Chem. 273, 23534–23541. 117. Sabatini, D.M., Barrow, R.K., Blackshaw, S., Burnett, P.E., Lai, M.M., Field, M.E., Bahr, B.A., Kirsch, J., Betz, H. & Snyder, S.H. (1999) Science 284, 1161–1164. 118. Scheper, G.C., Voorma, H.O. & Thomas, A.A. (1992) J. Biol. Chem. 267, 7269–7274. 119. Essrich, C., Lorez, M., Benson, J.A., Fritschy, J.M. & Luscher, B. (1998) Nat. Neurosci. 1, 563–571. 120. Kirsch, J., Wolters, I., Triller, A. & Betz, H. (1993) Nature (London) 366, 745–748. 121. Marin, P., Nastiuk, K.L., Daniel, N., Girault, J.A., Czernik, A.J., Glowinski, J., Nairn, A.C. & Premont, J. (1997) J. Neurosci. 17, 3445–3454. 122. Scheetz, A.J., Nairn, A.C. & Constantine-Paton, M. (1997) Proc. Natl. Acad. Sci. USA 94, 14770–14775. 123. Scheetz, A.J., Nairn, A.C. & Constantine-Paton, M. (2000) Nat. Neurosci. 3, 211–216. 124. Walden, W.E. & Thach, R.E. (1986) Biochemistry 25, 2033–2041. 125. Ray, A., Walden, W.E, Brendler, T., Zenger, V.E. & Thach, R.E. (1985) Biochemistry 24, 7525–7532. 126. Schwartz, L.M., Smith, S.W., Jones, M.E. & Osborne, B.A. (1993) Proc. Natl. Acad. Sci. USA 90, 980–984. 127. Paavola, L.G. (1978) J. Cell Biol. 79, 59–73. 128. Dudek, H., Datta, S.R., Franke, T.F., Birnbaum, M.J., Yao, R., Cooper, G.M., Segal, R.A., Kaplan, D.R. & Greenberg, M.E. (1997) Science 275, 661–665. 129. Xue, L., Fletcher, G.C. & Tolkovsky, A.M. (1999) Mol. Cell. Neurosci. 14, 180–198. 130. Isahara, K., Ohsawa, Y., Kanamori, S., Shibata, M., Waguri, S., Sato, N., Gotow, T., Watanabe, T., Momoi, T., Urase, K., Kominami, E. & Uchiyama, Y. (1999) Neuroscience 91, 233–249. 131. Lamborghini, J.E. (1987) J. Comp. Neurol. 264, 47–55. 132. Anglade, P., Vyas, S., Javoy-Agid, F., Herrero, M.T., Michel, P.P., Marquez, J., Mouatt-Prigent, A., Ruberg, M., Hirsch, E.C. & Agid, Y. (1997) Histol. Histopathol. 12, 25–31.


THE PHYSIOLOGICAL SIGNIFICANCE OF ß-ACTIN MRNA LOCALIZATION IN DETERMINING CELL POLARITY AND DIRECTIONAL MOTILITY

7045

Colloquium The physiological significance of ß-actin mRNA localization in determining cell polarity and directional motility Elena A.Shestakova, Robert H.Singer*, and John Condeelis Department of Anatomy and Structural Biology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461 ß-actin mRNA is localized near the leading edge in several cell types, where actin polymerization is actively promoting forward protrusion. The localization of the ß-actin mRNA near the leading edge is facilitated by a short sequence in the 3 untranslated region, the “zip code.” Localization of the mRNA at this region is important physiologically. Treatment of chicken embryo fibroblasts with antisense oligonucleotides complementary to the localization sequence (zip code) in the 3 untranslated region leads to delocalization of ß-actin mRNA, alteration of cell phenotype, and a decrease in cell motility. To determine the components of this process responsible for the change in cell behavior after ß-actin mRNA delocalization, the Dynamic Image Analysis System was used to quantify movement of cells in the presence of sense and antisense oligonucleotides to the zip code. It was found that net path length and average speed of antisense-treated cells were significantly lower than in sense-treated cells. Total path length and the velocity of protrusion of antisense-treated cells were not affected compared with those of control cells. These results suggest that a decrease in persistence of direction of movement and not in velocity results from treatment of cells with zip code-directed antisense oligonucleotides. To test this, direct analysis of directionality was performed on antisense-treated cells and showed a decrease in directionality (net path/total path) and persistence of movement. Less directional movement of antisensetreated cells correlated with a unpolarized and discontinuous distribution of free barbed ends of actin filaments and of ß-actin protein. These results indicate that delocalization of ß-actin mRNA results in delocalization of nucleation sites and ß-actin protein from the leading edge followed by loss of cell polarity and directional movement. Dynamic Image Analysis System | directionality | antisense oligonucleotides Beta-actin mRNA is localized to the leading lamella in chicken embryo fibroblasts (CEFs) and several other cell types, just proximal to the lamellipodia (1–4). Localization of ß-actin mRNA depends on an intact actin cytoskeleton in CEFs (5). The nucleotide sequence that determines the localization of ß-actin mRNA was found in the 3 untranslated region (UTR) and is composed of 54 nt 3 of the stop codon (the “zip code,” ref. 6). A protein of 68 kDa (zip code binding protein 1, ZBP1) binds the zip code in ß-actin mRNA (7). Binding of ZBP1 to the zip code correlated with localization of ß-actin mRNA; a mutated zip code unable to localize was unable to bind ZBP1. Delocalization of ß-actin mRNA with antisense oligonucleotides complementary to the zip code (zip code antisense) suppresses cell polarity (6) and motility (2). Likewise, inhibition of protein synthesis also slowed cell motility (2). These results suggested that there was some aspect of cell motility that was enhanced by the synthesis of ß-actin near the leading edge. In this work, we elucidate the significance of this localization of ß-actin mRNA and show that it plays a role in determining the polarity of nucleation sites for actin polymerization. There could be several reasons for suppression of cell motility upon delocalization of ß-actin mRNA. Cell motility requires actin polymerization in the leading edge (8–10). Cells with delocalized ß-actin mRNA may not polymerize actin filaments at the same rate if ßactin is not synthesized at sites of polymerization. As a result the cells would have a lower velocity of protrusion, which is driven by actin polymerization. Alternatively, the rate of actin polymerization may be unaffected by actin synthesis. Instead, the site of actin synthesis may affect the location of nucleation of actin polymerization that would define the direction of protrusion and, therefore, polarity of movement. To test which of these hypotheses was likely to be correct, several parameters of movement were measured in cells treated with zip code antisense oligonucleotides to delocalize ß-actin mRNA. The measurements were correlated with the sites of actin polymerization to determine how delocalization of ß-actin mRNA affected the location of protrusive activity. Our results indicate that the delocalization of the mRNA does not substantially change the rate of protrusion, but rather it significantly alters the sites where this protrusion occurs. MATERIALS AND METHODS Cell Culture. Primary CEFs were prepared as described (11), cultured 72–96 h in alpha-modified MEM (GIBCO) containing 10% FBS and antibiotics (penicillin, streptomycin). For further experiments, cells were replated on 0.5% gelatin-coated 12-mm gridded coverslips (Eppendorf). Cells were used for motility analysis after plating on coverslips and oligonucleotide treatment for 12 h. For in situ hybridization, cells were fixed at 37°C in 4% paraformaldehyde in PBS (1 mM KH2PO4, 10 mM N2HPO4, 137 mM NaCl, 2.7 mM KCl, pH 7.0), washed in PBS, and dehydrated in 70% ethanol at 4°C overnight. Oligodeoxynucleotide (ODN) Treatment of Cells. Phosphorothioate-modified ODNs comprising the antisense and sense sequence to 18 nt of the zip code (6) were synthesized on an Applied Biosystems 394 DNA/RNA Synthesizer and purified by electophoresis through polyacrylamide gel, eluted, lyophilized, resuspended in water, and additionally purified by gel filtration on Sephadex G50. Purified ODNs were lyophilized and resus

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: UTR, untranslated region; CEF, chicken embryo fibroblast; ODN, oligodeoxynucleotide. *To whom reprint requests should be addressed. E-mail: [email protected].



7046

pended in diethyl-pyrocarbonate-treated water. ODNs (8 µM) were added to a cell medium three times at 4-h intervals (6). Rhodamine-Actin-Based Detection of Barbed Ends of Actin Filaments. Stock rhodamine-labeled actin was thawed and diluted with 1 mM Hepes (pH 7.5), 0.2 mM MgCl2, and 0.2 mM ATP, sonicated, and clarified in Beckman centifuge at 95 krpm, for 20 min. Cells were permeabilized with 20 mM Hepes (pH 7.5), 138 mM KCl, 4 mM MgCl2, 9 mM EGTA, 0.25 mg/ml saponin, 1 mM ATP, 1% BSA containing 0.45 µm rhodamine-actin that was added to the lysis buffer just before application to cells. One to three minutes after incubation, cells were fixed for 5 min with 3.7% formaldehyde in PBS, incubated with 0.1 M glycine in PBS for 10 min, and washed with PBS. Cells were stained with 1 µM fluorescein phalloidin in buffer for 40 min in humidified chamber, washed, and mounted on 0.1 M N-propylgallate in 50% glycerol in PBS, pH 7.0. Immunofluorescence. Cells were plated on coverslips, fixed in 3.7% formaldehyde, permeabilized with 0.5% Triton in PBS, and incubated with primary antibodies to ß-actin (a gift of Ira Herman, Tufts Medical School, Boston) and secondary fluorescein-labeled antibodies to rabbit IgG for 1 h and mounted as described (12). In Situ Hybridization. Chicken ß-actin-specific 3 UTR probes (five probes of 50 nt each, with five amino linkers per probe spaced 10 nt apart) were synthesized on an Applied Biosystems 394 DNA/RNA Synthesizer. Chicken ß-actin probes were labeled with CY3. To detect ß-actin mRNAs, coverslips were rehydrated in PBS, permealized with 0.5% Triton in PBS for 10 min, and then hybridized for 3 h at 37°C with 5 ng of the mixture of five oligonucleotides. Each oligonucleotide can hybridize independently with ß-actin mRNA so as to increase the signal to noise when all five have hybridized to a single molecule (25 fluorochromes total; ref. 13). Coverslips were washed twice with 50% formamide in 2×SSC (300 mM NaCl, 30 mM sodium citrate, pH 7.0), then in 2×SSC, 1×SSC, and mounted. High-Resolution Microscopy. An Olympus BX60 microscope was used with a×60 planapo objective numerical aperture 1.4. Digital images were captured by using a Photometries camera and CELLSCAN software. Computer-Assisted Analysis of Cell Behavior. Cells were recorded with an Olympus microscope equipped with a charge-coupled device camera through a×10 objective with a 1-min time interval between image frames over 60 min. Images were processed with DIAS (Dynamic Image Analysis System) software (14). Cell motility data were displayed as an overlay of cell perimeters, i.e., as a stack of every fifth video frame (cell perimeter plot) and as a centroid plot showing the location of the geometrical center of the cell as a function of time. RESULTS Antisense Treatment of Cells. It was shown previously that cisacting elements in the 3 UTR of chicken ß-actin mRNA were responsible for the localization of this mRNA. The 54 nt 3 of the stop codon were most potent in localizing ß-actin mRNA. This region is called the zip code and can be divided into A, B, and C regions (6). In this study, an antisense ODN, complementary to the 3 18 nt of the zip code was used (C–). For a control, the sense strand (C+) was used. Effects of Antisense ODNs on Cell Motility. It was reported previously that the velocity of cell locomotion is reduced by treatment of cells with antisense-oligos directed against the zip code of ß-actin mRNA (2). To confirm and extend this observation we repeated these experiments. CEFs were treated with zip code antisense (C–) or sense (C+) ODNs. The distribution of ß-actin mRNA was determined in each population by using fluorescence in situ hybridization. Cells treated with antisense showed decreased localization of ß-actin mRNA to the leading edge, whereas cells treated with sense ODNs (control) localized the mRNA to the leading edge to an extent that was statistically indistinguishable from untreated cells (Table 1, ß-actin mRNA). Table 1. Percent of cells with ß-actin mRNA, ß-actin protein, and nucleation (barbed ends) localized to the leading edge as a function of zip code antisense (C–) or sense (C+) oligonucleotide treatments C– % localization C+ % localization Leading edge Diffuse Leading edge Diffuse ß-actin 33 67 58 42 mRNA (186) (375) (392) (288) (n) Barbed 30 70 70 30 ends (130) (303) (210) (90) (n) 32 68 68 32 ß-actin (70) (149) (128) (60) protein (n) The average path length migrated by antisense-treated and control cells was measured as a change in the nuclear position during a 60min observation period. In Table 2, the average path lengths for 100 antisense-treated and 64 control cells are presented. The antisensetreated cells migrated shorter distances, and this difference was statistically significant. This result is similar to that reported previously by Kislauskis et al. (2). However, the underlying mechanism for this observation has not been investigated. Therefore, we subjected the cells to a more rigorous analysis of their motility to ascertain which of the components of cell motility was most affected. To do this we correlated various aspects of cell motility with ß-actin mRNA localization. Treated and control cells were monitored by using an inverted microscope supplemented with a heating chamber. Time-lapse movies were obtained over 60 min with 1-min intervals between frames. Fig. 1 demonstrates the difference in behavior between these two cell populations. In the presence of the zip code antisense, cells did not translocate appreciably, whereas in the presence of sense ODNs, the cells continued to migrate. The movies obtained in this way were analyzed by using the Dynamic Image Analysis System (Materials and Methods). Several cell motility parameters were determined: net path length, average speed, average instantaneous speed (protrusion velocity), directionality, and persistence (Table 3). Delocalization of ßactin mRNA in CEFs correlated with a significant decrease in net path length and average speed. Total path length and average protrusive velocity were not statistically different from control cells (Table 3). These results are explained by a decrease in the directionality and persistence of movement Table 2. Average path length migrated by CEFs (measured as a change in the nuclear position) during 60 min as a function of zip code antisense (C–) or sense (C+) oligonucleotide treatment C– (n=100) C+ (n=64) Average net path length, µm 12.67 16.32 SE 0.87 1.56 t test 4.39%



7047

Fig. 1. Movement of CEFs in the presence of (Upper) zip code sense (C–) and (Lower) antisense oligonucleotides(C+). Pictures depict frames 5 min apart from the video analysis of two fields of cells. Arrows indicate direction of movement of each cell over the subsequent frames. Note that cells move in the presence of sense, but not antisense, oligonucleotides. without a decrease in rate of locomotion. Consistent with this possibility is the comparison of centroid plots of antisense-treated and control cells. An example of one of these analyses is represented in Fig. 2, which shows that cells with delocalized ß-actin mRNA exhibit random directionality of motility and have less persistence in the direction of motility whereas control cells move with fixed polarity and linear directionality and are more persistent in the direction of motility (Table 3). This finding indicates that mRNA localization is not necessary for the ability of the cells to move, but rather for their ability to maintain this movement in one direction. The Distribution of Free Barbed Ends Is Randomized in Antisense-Treated CEFs. The above results indicate that the decrease in apparent cell velocity observed by Kislauskis et al. (2) was due to a decrease in net path traveled over the time of observation. This resulted from a decrease in persistence rather than a decrease in speed of locomotion. This observation is consistent with a conclusion that the rate of protrusion of the leading edge was unaffected by delocalization of ß-actin mRNA. In contrast, it is the persistence of the direction of protrusion that is affected by delocalization of ß-actin mRNA. We speculated that the mechanism behind the loss of polarity of protrusion in antisense-treated cells involved loss of polarized nucleation of actin polymerization. To test this hypothesis, permeabilized CEFs were incubated with a concentration of rhodamine-actin close to the critical concentration of barbed end addition to label the barbed ends of actin filaments. Labeling of barbed ends demonstrates sites of nucleation of actin polymerization (9, 12). The sites of rhodamine-actin incorporation were not polarized in antisense-treated CEFs but rather were around the entire periphery (Fig. 3) whereas in sense-treated CEFs the sites of rhodamine-actin incorporation were polarized and distributed ontinuously along the leading edge of the lamellipod (Fig. 3; Table 1, barbed ends). The peripheral, nonpolarized distribution of the nucleation sites would predict that ß-actin would be distributed likewise peripherally, but not in a polarized distribution. To test whether the change in the distribution of barbed ends resulted in changes in ß-actin distribution, we used isoform-specific antibodies to determine the ß-actin protein location in sense- and antisense-treated cells. Fig. 4 demonstrates the common observation between these two populations; treatment with the zip code antisense results in a peripheral, nonlocalized distribution of ß-actin whereas the sense-treated cells show the characteristic concentration of ß-actin at the leading and retracting poles of the fibroblast. Therefore, in these two populations of cells the distribution of ß-actin protein, which normally is localized to the leading edge, was unaffected by sense treatment but became diffusely distributed in antisense-treated cells (Fig. 4; Table 1, ß-actin protein). This finding indicates that the distribution of barbed ends and the synthesis of ß-actin are likely to be related functionally. This functionality would derive directly from the distribution of the site of synthesis of the ß-actin. DISCUSSION This study was undertaken to elucidate the observation whereby delocalization of ß-actin mRNA can affect cell motility and polarity in fibroblasts. Delocalization of ß-actin mRNA with antisense oligonucleotides has been observed to reduce the motility of CEFs (2) and smooth muscle cells (15). Table 3. Delocalization of ß-actin mRNA with zip code antisense oligonucleotides (C–) but not sense (C+) oligonucleotides causes loss of polarized cell movement but no significant change in protrusion rates Net path Average speed, Total path Protrusion Directionality, net/ Persistence, µm/ length, µm µm/min length, µm velocity total min×deg C– (n=20) 12.72 0.22 84 0.79 0.18 0.12 SE 0.33 0.0056 2.3 0.017 0.0063 0.003 C+ (n=20) 20.92 0.35 85.64 0.88 0.26 0.16 SE 0.53 0.009 1.19 0.011 0.0064 0.002 t test (%) 1 1 89 33 6 4 Explanation of categories (see ref. 14 for details): net path length, the distance traveled between the beginning and ending points of the analysis (60 min); average speed, the average distance traveled divided by the observation time (60 min); total path length, the entire path summed from all movements at each interval (1 min); protrusion velocity, the movement of protrusions during each interval (1 min) and averaged; directionality and persistence are measurements as to how consistently the cells stay on a course (1-min intervals) determined either by length of the net path relative to the total path or number of turns the cell makes in degrees per min, respectively.



7048

Fig. 2. ß-actin mRNA localization in the presence of zip code sense oligonucleotides (C+) or antisense oligonucleotides (C–). Perimeter and centroid (dots) plots are from ODN-treated cells over the time frames of analysis (1 min). These results are representative of the analysis of populations of cells depicted in Fig. 1. (A) Sense-treated cell. (B) Antisense-treated cell. Note that the antisense oligonucleotides cause loss of polarized cell movement defined as a linear centroid track (arrow in A). Our results show that antisense but not control (sense) oligos caused a delocalization not only of ß-actin mRNA, but also of ß-actin protein and barbed ends from the leading edge of fibroblasts and resulted in a random distribution of all three. This was reversible upon removal of the antisense oligonucleotides. By investigating sites of actin filament nucleation, we showed that they were delocalized as a result of disrupting the targeting of ß-actin mRNA. This result reveals a possible mechanism for establishing cell polarity: ß-actin protein, and/or proteins with related zip codes, define the location of nucleation of actin polymerization and consequently, cell polarity and directional motility.

Fig. 3. Sites of rhodamine actin incorporation in zip code sense-treated (C+) and antisense treated (C–) cells. (A–C) Sense treatment. (D–F) Antisense treatment. (A and D) Rhodamine actin incorporation showing the location of free barbed ends on actin filaments. (B and E) FITC-phalloidin-labeling of all actin filaments. (C and F) Phase-contrast image. (Bar, 10µm.) Note that rhodamine actin incorporation sites are unpolarized in antisense-treated cells.

Fig. 4. Localization of ß-actin protein in zip code antisense (A and B) compared with sense-treated cells (C and D). (A and C) Staining with anti-ß-actin antibodies. (B and D) Nomarski optics. (Bar, 10 µm.) Note that the ß-actin staining is not as prominently localized to the leading edge in antisense-treated cells. The molecular mechanism by which polarity of cell crawling is affected by ß-actin mRNA localization could depend on several interdependent events: (i) Localized synthesis of ß-actin from localized mRNA drives protrusion of the lamellipod. (ii) ß-actin isoform specific protein interactions are responsible for the protrusion. (iii) Localization to the leading edge of the mRNAs for other proteins in addition to ß-actin, e.g., the nucleating complex containing Arp 3 mRNA. A discussion of the evidence for each of these is detailed below. Localized Synthesis of ß-Actin from Localized mRNA. Based on the estimated 2,500 ß-actin mRNA molecules per cell, at the established translational rate of 1.5 actins per sec per mRNA molecule, the cell would synthesize 3,900 actin molecules per sec or 2.34×105 per min (2). In moving cells, the polymerization zone uses a minimum of 3.6×106 actin molecules per min (9). Therefore, it is unlikely that a 6.5% contribution of newly synthesized ß-actin will significantly contribute to the rate of actin polymerization at the leading edge. However, if all of the ß-actin is synthesized in a restricted volume, a consequence of localizing the ß-actin mRNA to the leading edge, the local rate of synthesis of ß-actin might significantly impact the actin polymerization events in this restricted volume and, therefore, establish a preferred location for actin polymerization. The above model would be further supported if newly synthesized actin would have a faster rate of polymerization, or a higher affinity for a nucleation complex than “older” actin, which may be posttranslationally modified. For instance, interaction of a chaperone with the ßactin nascent chain (16) could promote assembly of a nucleation complex near the site of synthesis. ß-Actin Isoform-Specific Protein Interactions. The ß isoform of actin may be preferentially stored as the monomer used for polymerization at the leading edge. In this hypothesis, local accumulation of a nonfilamentous form of actin that could be released suddenly upon stimulation of motility would determine the location of actin polymerization. Potential storage particles containing nonfilamentous actin have been identified by comparing the localization patterns of vitamin D-binding protein,



7049

which binds to G-actin with 5 nM kd, and phalloidin, which binds to actin (17). These stores of nonfilamentous actin are found at the leading edge and are located adjacent to sites of actin polymerization and in the region of the cell where the ß-actin mRNA is also present. Possibly these sites could result from islands of ß-actin synthesis. ß-actin is found at the leading edge of crawling cells. ß-actin does not substitute for muscle actin in either the formation of stress fibers (17) or myofibrils in cardiomyocytes (18). In addition, it seems to interact more tightly with certain actin binding proteins that may function at the leading edge of crawling cells. Ezrin (19), profilin (20), thymosin ß 4 (21), and L-plastin (22) bind more strongly to ß-actin than αactin. A capping protein, ß-cap 73, may cap the barbed end in an isoform-specific manner (23). There is growing evidence that the Arp2/3 complex is required for nucleation of actin filaments at the leading edge (12, 24–27). If the Arp2/3 complex is the dominant nucleation activity at the leading edge, a possible preference for the ß-actin isoform by the Arp2/3 complex would require local synthesis of ß-actin to supply the preferred monomer for polymerization. Therefore, the localization of ß-actin synthesis at the leading edge may be functionally important for polarity and motility. Localization to the Leading Edge of Motility-Related mRNAs. The localization of ß-actin mRNA may be representative of the localization of a family of mRNAs with related 3 UTR zip codes, many of which function synergistically at the leading edge. Proteins coded for by these mRNAs therefore might have related functions. We have analyzed the 3 UTRs of mRNAs, which code for proteins believed to have actin binding functions at the leading edge, for the presence of the zip code consensus sequence. This sequence GACUX7–38ACACC is found in ß-actin mRNAs known to target to the leading edge from all vertebrates. Besides ß-actin mRNAs, mRNA for Arp3 and myosin IIB heavy chain contain the consensus sequence and are predicted to be recognized by the localization mechanism that targets ß-actin mRNA to the leading edge. It is known that the ACACCC consensus sequence, when mutated in ß-actin mRNA, results in a failure to localize the mRNA to the leading edge of cells (2, 7), even if the ß-actin coding sequence remains intact and is used as the reporter mRNA. Preliminary results indicate that Arp3 mRNA, like ß-actin mRNA, also localizes to the leading edge (G.Liu, W.Grant, D.Persky, V.L.Lathaur, R.H.S., and J.C., unpublished work). Serum-dependent localization of ß-actin mRNA suggests that signaling mechanisms are involved in the localization of motility-related mRNAs, thereby coordinating their temporal and spatial distribution and expression (28). Furthermore, it is possible that localized synthesis of, for instance, Arp3 could determine the localization of Arp2/3 complex in the leading edge of the cells even if mRNAs coding other components of Arp2/3 complex were more diffusely distributed. Arp2/3 complex and ß-actin, both localized in the leading edge, could determine the nucleation sites for actin polymerization. Newly formed actin filaments could interact with ß-actin isoform-specific binding proteins, thereby stabilizing the cell polarity and consequent directional motility (29). The leading edge of the cell is a complex composite of asymmetrically distributed proteins many of which function in concert to produce the motility response. It is likely that other proteins like ß-actin also are synthesized asymmetrically and therefore would provide not only a differential concentration of these proteins but also an increased likelihood of interactions among relevant proteins in a cellular region where function depends on these interactions. We presume therefore that a panoply of mRNAs comprising a significant complexity of sequences is localized to the lamella to effect the complex events required by motility. It is our expectation that these sequences will contain a common motif and/or structure in the 3 UTR characterizing them as mRNAs for motility-related proteins. It is likely that further investigations will reveal the consensus sequences (see below). The localization of ß-actin mRNA is not restricted to fibroblasts, but seems to be a feature of other localized cells. Neurons localize ßactin mRNA to the growth cone of developing neurites (30, 31). The presence of the mRNA results in the specific translation of ß-actin protein in the growth cone. Like fibroblasts, the delocalization of the mRNA results in growth cone retraction and nondirectionality of growth cone guidance (37). In addition to the neuronal growth cones, embryonic neural crest cells might localize ß-actin mRNA to the front of the cell, in the direction of their migration. Disruption of the Xenopus homolog of ZBP1 appears to inhibit their migration and result in severe embryological defects in forebrain development (J.Yisraeli, personal communication). Furthermore, if the zip code for ß-actin mRNA is transferred to another protein, not normally at the leading edge, in this case vimentin, a distorted morphology results wherein the cell structure at the leading edge is branched and attenuated (32). These results argue that synthesis of the correct protein in the correct place (near the leading edge) is an important requirement for cell structure and polarity. In addition to ß-actin mRNA localization in fibroblasts (1), the field of RNA localization has been advanced by the discovery of a number of systems where mislocalization of the RNA can lead to a significantly altered phenotype or lethality (33–36). In many of these cases mRNA localization is required for normal development and differentiation because the localized mRNA codes for nuclear factors and the resultant cell divisions segregate the mRNAs for these morphogenic determinants. However, the nature of the localization we describe here is important for a different reason: it determines the spatial orientation, morphology, and behavior of these somatic cells. In this second aspect of RNA localization, the complex of proteins involved in cell migration, cellular reaction to the environment and development of cell polarity are organized within the cytoplasm by virtue of the spatial segregation of their cognate mRNAs, and are not in the short term related to transcription of genes. In this way, components of the mechanism controlling cell behavior and structure can rapidly reassemble within the cell. In this model, the proteins involved in forming these multipolypeptide complexes (the nucleation complex, for instance) would be compartmentalized in response to environmental cues and subsequent signal transduction events and then synthesized in proximity to each other where they would interact preferentially because of their higher local concentrations. Possibly these higher concentrations of proteins could autoregulate their own synthesis. In this way, we propose that the localization of ß-actin mRNA represents one mechanism for the spatially compartmentalized assembly of cellular complexes. We thank Michael Cammer in the Einstein Analytical Imaging Facility and Jeff Wyckoff and Shailesh M.Shenoy for technical help with light microscopy and the Dynamic Image Analysis System, Wayne Grant for technical help with experiments, and Steve Braut for synthesis of oligonucleotides. This work was supported by National Institutes of Health grants to R.H.S. and J.C. 1. Lawrence, J.B. & Singer, R.H. (1986) Cell 45, 407–415. 2. Kislauskis, E.H., Zhu, X. & Singer, R.H. (1997) J. Cell Biol. 136, 1263–1270. 3. Hill, M.A., Schedlich, L. & Gunning, P. (1994) J. Cell Biol. 126, 1221–1230. 4. Hoock, T.C, Newcomb, P.M. & Herman, I.M. (1991) J. Cell Biol. 112, 653–664. 5. Sundell, C. & Singer, R.H. (1991) Science 253, 1275–1277. 6. Kislauskis, E.H., Zhu, X. & Singer, R.H. (1994) J. Cell Biol. 127, 441–451. 7. Ross, A.F., Oleynikov, Y., Kislauskis, E.H., Taneja, K.L. & Singer, R.H. (1997) Mol. Cell. Biol. 17, 2158–2165.



7050

8. Bailly, M., Macaluso, F., Cammer, M., Segall, J. & Condeelis, J. (1998) J. Cell Biol. 145, 331–345. 9. Chan, A.Y., Raft, S., Bailly, M., Wyckoff, J.B., Segall, J.E. & Condeelis, J.S. (1998) J. Cell Sci. 111, 199–211. 10. Segall, J.E., Tyerech, S., Boselli, L., Masseling, S., Helft, J., Chan, A., Jones, J. & Condeelis, J. (1996) Clin. Exp. Metastasis 14, 61–72. 11. Kislauskis, E.H., Li, Z., Singer, R.H. & Taneja, K.L. (1993) J. Cell Biol. 123, 165–172. 12. Bailly, M., Yan, L., Whitesides, G., Condeelis, J.S. & Segall, J.E. (1999) Exp. Cell Res. 241, 285–299. 13. Femino, A.N., Fay, F.S., Fogarty, K. & Singer, R.H. (1998) Science 280, 585–590. 14. Soll, D.R. (1995) Int. Rev. Cytol. 163, 43–104. 15. Schedlich, L., Hill, M. & Lockett, T. (1997) Biol. Cell 89, 113–122. 16. Hansen, W.J., Cowan, N.J. & Welch, W.J. (1999) J. Cell Biol. 145, 265–277. 17. Cao, L., Fishkind, D.J. & Wang, Y. (1993) J. Cell Biol. 123, 173–181. 18. von Arx, P., Bantle, S., Soldati, T. & Perriard, J.C. (1995) J. Cell Biol. 131, 1759–1773. 19. Shuster, C. & Herman, I. (1995) J. Cell Biol. 128, 837–848. 20. Segura, M. & Lindberg, U. (1984) J. Biol. Chem. 259, 3949–3954. 21. Weber, A., Nachmias, V.T., Pennise, C.R., Pring, M. & Safer, D. (1992) Biochemistry 31, 6179–6185. 22. Namba, Y., Ito, M., Zu Y., Shigesada, K. & Maruyama, K. (1992) J. Biochem. 112, 503–507. 23. Shuster, C.B., Lin, A.Y., Nayak, R. & Herman, I.M. (1996) Cell Motil. Cytoskeleton 35, 175–187. 24. Machesky, L.M., Reeves, E., Wientjes, F., Mattheyse, F.J., Grogon, A., Totty, N.F., Burlingame, A.L., Hsuan, J.J. & Segal, A.W. (1997) Biochem. J. 328, 105–112. 25. Mullins, R.D., Heuser, J.A. & Pollard, T.D. (1998) Proc. Natl. Acad. Sci. USA 95, 6181–6186. 26. Welch, M.D., Rosenblatt, J., Skoble, J., Portnoy, D.A. & Mitchison, T.J. (1998) Science 281, 105–108. 27. Blanchoin, L., Pollard, T.D. & Mullins, R.D. (2000) Curr. Biol. 10, 1273–1282. 28. Latham, V.M., Jr., Kislauskis, E.H., Singer, R.H. & Ross, A.F. (1994) J. Cell Biol. 126, 1211–1219. 29. Liu, G., Edmonds, B. & Condeelis, J. (1996) Trends Cell Biol. 6, 168–171. 30. Zhang, H.L., Singer, R.H. & Bassell, G.J. (1999) J. Cell Biol. 147, 59–70. 31. Bassell, G. & Singer, R.H. (2001) in Results and Problems in Cell Differentiation, ed. Richter, D. (Springer, Berlin), Vol. 34, pp. 41–56. 32. Morris, E.J., Evason, K., Wiand, C., L’Ecuyer, T.J. & Fulton, A.B. (2000) J. Cell Biol. 113, 2433–2443. 33. St. Johnston, D. (1995) Cell 81, 161–170. 34. Long, R.M., Singer, R.H., Meng, X., Gonzalez, I., Nasmyth, K. & Jansen, R.-P. (1997) Science 277, 383–387. 35. Hazelrigg, T. (1998) Cell 95, 451–460. 36. Carson, J.H., Cui, H., Krueger, W., Schlepchenko, B., Brunwell, C. & Barbarese, E. (2001) in Results and Problems in Cell Differentiation, ed. Richter, D. (Springer, Berlin), Vol. 34, pp. 69–81. 37. Zhang, H.L., Eom, T., Oleynikov, Y., Shenoy, S.M., Liebelt, D.A., Singer, R. H. & Bassell, G.J. (2001) Neuron, in press.


SORTING AND DIRECTED TRANSPORT OF MEMBRANE PROTEINS DURING DEVELOPMENT OF HIPPOCAMPAL NEURONS IN CULTURE

7051

Colloquium Sorting and directed transport of membrane proteins during development of hippocampal neurons in culture M.A.Silverman*, S.Kaech*, M.Jareb†, M.A.Burack*, L.Vogt‡, P.Sonderegger‡, and G.Banker*§ for Research on Occupational and Environmental Toxicology, Oregon Health Sciences University, Portland, OR 97201; †Center for Neurobiology and Behavior, Columbia University, New York, NY 10027; and ‡Institute of Biochemistry, University of Zurich, Zurich, Switzerland CH-8057 Hippocampal neurons in culture develop morphological polarity in a sequential pattern; axons form before dendrites. Molecular differences, particularly those of membrane proteins, underlie the functional polarity of these domains, yet little is known about the temporal relationship between membrane protein polarization and morphological polarization. We took advantage of viral expression systems to determine when during development the polarization of membrane proteins arises. All markers were unpolarized in neurons before axonogenesis. In neurons with a morphologically distinguishable axon, even on the first day in culture, both axonal and dendritic proteins were polarized. The degree of polarization at these early stages was somewhat less than in mature cells and varied from cell to cell. The cellular mechanism responsible for the polarization of the dendritic marker protein transferrin receptor (TfR) in mature cells centers on directed transport to the dendritic domain. To examine the relationship between cell surface polarization and transport, we assessed the selectivity of transport by live cell imaging. TfR-green fluorescent protein-containing vesicles were already preferentially transported into dendrites at 2 days, the earliest time point we could measure. The selectivity of transport also varied somewhat among cells, and the amount of TfR-green fluorescent protein fluorescence on intracellular structures within the axon correlated with the amount of cell surface expression. This observation implies that selective microtubule-based transport is the primary mechanism that underlies the polarization of TfR on the cell surface. By 5 days in culture, the extent of polarization on the cell surface and the selectivity of transport reached mature levels. Neurons are composed of two morphologically and molecularly distinct domains, axons and dendrites. The accurate localization of proteins to these domains is critical for neuronal function. The biosynthetic pathway by which membrane proteins acquire their polarized distribution is thought to begin when proteins destined for different cellular domains are packaged into different populations of carrier vesicles, a step that probably occurs in the trans-Golgi network. Once formed, carrier vesicles are conveyed to the axon or dendrite by microtubule-based transport. In a previous report (1), we demonstrated that neurons utilize two different mechanisms for the targeting of polarized membrane proteins, one based on selective transport, the other based on a selectivity filter that occurs downstream of transport. We found that cargo vesicles containing a dendritic protein, transferrin receptor (TfR), are transported directly to the dendritic domain and excluded from the axon. In contrast, cargo vesicles containing the axonal protein neuron-glia cell adhesion molecule (NgCAM) enter both axons and dendrites, even though NgCAM is polarized to the axonal plasma membrane. When embryonic hippocampal neurons are placed into culture, they acquire their characteristic polarized morphology in a series of well-defined developmental stages (2, 3). Initially, the cells form several short neurites that cannot be distinguished as axons or dendrites (developmental stage 2). After 12–36 h in culture, one of these neurites enters a prolonged period of growth and acquires axonal characteristics, thus defining the cell’s polarity (stage 3). Over the next few days, the remaining neurites acquire dendritic characteristics (stage 4). The molecular events that underlie the development of neuronal polarity are not well understood (for review, see ref. 4). Neurons at stage 2 of development are molecularly and morphologically unpolarized. Previous work has shown that axonal proteins, such as the cell adhesion molecule LI and the synaptic vesicle proteins synaptophysin and synapsin I, become selectively polarized to the axon at stage 3 of development (5–7). The situation for dendritic proteins is less clear. The polarization of some dendritic proteins appears to lag behind the polarization of axonal proteins (8, 9), whereas other dendritic proteins are excluded from axons at developmental stage 3 (10, 11). Bradke and Dotti (4) have hypothesized that the transition from stage 2 to 3 is also marked by a redirection of organelle transport into the nascent axon. One limitation in examining the development of molecular compartmentalization in nerve cells is that many of the relevant endogenous proteins are expressed at very low levels early in development, making it difficult to accurately assess their distribution. In the present study, we have used a different approach: virally mediated expression of axonal and dendritic marker proteins at levels that make it easy to assess their distribution, even early in development. We have also expressed green fluorescent protein (GFP)-tagged versions of these proteins to visualize their transport into axons and dendrites. We show that both dendritic and axonal marker proteins are significantly polarized by developmental stage 3; the selective transport of dendritic proteins is also evident at this stage of development. Both selective transport and the polarized distribution of proteins at the cell surface reach mature levels by 5 days in culture. *Center

METHODS Reagents. We thank the following people who generously provided cDNA, virus, and/or antibodies: James Casanova, Massachusetts General Hospital, Boston, mutant polyimmunoglobulin receptor (plgR) cDNA and plgR sheep antisera (12); Robert Gerard, University of Texas-Southwest Medical Center, Dallas; low-density lipoprotein receptor (LDLR) AdV (13); Joseph Goldstein, Texas-Southwest Medical Center, LDLR rab

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: TfR, transferrin receptor; GFP, green fluorescent protein; LDLR, low-density lipoprotein receptor; pig, polyimmunoglobulin; plgR, pig receptor. §To whom reprint requests should be addressed at: CROET/L-606, Oregon Health Sciences University, 3181 S.W.Sam Jackson Park Road, Portland, OR 97201. E-mail: [email protected].



7052

bit antisera (14); Vance Lemmon, Case Western Reserve University, Cleveland, NgCAM chick-specific monoclonal (15); and Ian Trowbridge, Salk Institute, La Jolla, CA, TfR cDNA (16) and TfR human-specific monoclonal (17). Monoclonal antibodies against LDLR (RPN537) were purchased from Amersham Pharmacia; monoclonal antibodies against TfR (B3/25) were obtained from Boehringer Mannheim. Cell Culture and Viral Infection. Primary cultures of dissociated neurons from embryonic day 18 rat hippocampi were prepared essentially as described (18). Replication-defective herpes simplex viruses and adenoviruses were used to express exogenous proteins (1, 19). Viruses were titered to infect 1–10% of the neurons in culture. Immunostaining. To detect virally expressed proteins present on the cell surface, living cultures were incubated with the primary antibody diluted in culture medium for 5–7 min at 37°C, quickly rinsed in phosphate-buffered saline, and then fixed. Primary antibodies bound to antigen were detected with the appropriate fluorescently labeled secondary antibodies after permeabilization and blocking of nonspecific background. For quantitation of the fluorescence signal, images of labeled cells (specimen images) were acquired by using either a Photometries (Tucson, AZ) CH250 camera (12 bit; 1,315×1,017 pixels) and a Zeiss Axiophot [25×Plan Apo objective; numerical aperture (N.A.) 1.2] or a Princeton Instruments (Monmouth Junction, NJ) Micromax (12 bit, 1,300×1,030 pixels) and a Leica DM-RXA (20×Plan Apo, N.A. 0.5). Infected cells were chosen by examining random fields at approximately 2-mm intervals across the coverslip. A labeled cell whose processes traversed the field was selected for analysis, so long as its processes did not overlap those of other labeled cells; cells with fewer than three identifiable dendrites were excluded. To limit possible photobleaching during the process of cell selection, total exposure time was kept to less than 10 sec. In control experiments, this level of exposure was found to cause less than a 3% reduction in fluorescence intensity. Exposure time was adjusted so that maximum pixel value was at least half saturation. After acquiring the specimen image, a dark current image generated by an equivalent exposure with the camera shutter closed was subtracted, and a shading correction based on an image of a uniformly fluorescent field was applied to compensate for uneven illumination of the field. Finally, a threshold was set to eliminate nonspecific background staining of axons and dendrites of uninfected cells respectively, and the total fluorescence in the axonal and dendritic domain was determined. A process was considered an axon if it was at least twice the length of any of its other processes. The other processes were considered dendrites. Fluorescence in the cell body was excluded from the analysis. Live Imaging. Cells on coverslips were sealed into a heated chamber (Warner Instruments, Hamden, CT) in phenol red-free Hanks’ balanced salt solution buffered with 10 mM Hepes (pH 7.4) and supplemented with 0.6% glucose. Vesicle transport was imaged by capturing frames continuously for 30 sec (600-msec exposures) with a Micromax cooled charge-coupled device camera and a 63×Plan Apo, N.A. 1.32 objective on a Leica DM-RXA. For quantitative analysis, transport events were detected by first extracting difference images of sequential frames followed by analysis by using the kymograph drop-in function of the METAMORPH IMAGING SOFTWARE (Universal Imaging, Downingtown, PA). Briefly, lines were drawn along the axis of individual neurites, and the kymograph function was used to find the brightest pixel along a 10-pixel line perpendicular to the axis of the neurite. These values were then plotted for each frame, with time on the x axis and position along the neurite on the y axis. Thus, moving vesicles appeared as diagonal lines whose slopes were a measure of rate and direction of transport (with positive slope corresponding to anterograde transport). The number of transport events in the axon and at least three of the dendrites were determined for 3–12 cells at each time point. RESULTS Changes in the Polarization of Membrane Proteins During Development. To assess when during development membrane proteins acquire their characteristic polarized distribution, we expressed representative axonal and dendritic membrane proteins at times ranging from 1 to 14 days in culture and assessed their polarization on the cell surface by live-cell immunostaining. We selected the TfR and the LDLR as dendritic markers and NgCAM as an axonal marker. The sorting of these proteins in mature hippocampal neurons has been well characterized (19, 20). As an example of an unsorted protein, we chose a construct of the pIgR whose dendritic sorting signal had been deleted [pIgR665–668 (19, 20)]. We also assessed the polarization of L1, an endogenous axonal protein. We first examined cells at stage 2 of development, before neurites have been specified as axons or dendrites. If the polarization of membrane proteins preceded axonal specification, one might expect axonal markers to be concentrated in a single neurite, whereas dendritic markers might be present in all of the neurites except one. When the distribution of these proteins was assessed in stage 2 neurites, we often found that some neurites exhibited more staining than others, but we never observed a cell with only a single neurite that excluded dendritic markers or that had a high concentration of axonal markers. This lack of polarity was particularly evident when cells were simultaneously infected with viruses expressing axonal and dendritic markers (Fig. 1 a). The staining for LDLR and NgCAM was most intense in the growth cones, with some growth cones staining more brightly than others. However, rather than exhibiting the complementary distribution one would expect for proteins polarized to opposite domains, the two markers tended to have a similar distribution in stage 2 cells: growth cones that were brightly stained with the dendritic marker were often brightly stained with the axonal marker as well. Differences in the intensity of staining among different growth cones may reflect the dynamics of their growth; at this stage of development, neurites undergo alternating periods of extension and retraction (21). At stage 3 of development, both axonal and dendritic markers were polarized, although not to the extent seen in mature neurons. For example, Fig. 1 b illustrates a stage 3 neuron expressing both LDLR and NgCAM. NgCAM was present throughout the cell body and axon, with particularly intense staining in the distal axon. Little staining was present in the dendrites. LDLR was present in the dendrites and proximal axon, but little or no staining was present in the distal axon. On average, we found that 90% of the neuritic cell surface staining for NgCAM was axonal, whereas 81% of the LDLR staining was dendritic. In contrast, staining for the unpolarized protein, pIgR665–668, was about equally divided between dendrites and axon (46% dendritic). Fig. 1 c summarizes the changes in the distribution of axonal and dendritic marker proteins that occur during the first 2 weeks in culture. The polarity of both axonal and dendritic markers increased during the first few days in culture, reaching mature levels by about day 5. Over time, a slightly greater percentage of the unpolarized protein, pIgR665–668, became associated with the axon, presumably reflecting a relative increase in the size of the axonal arbor. The Distribution of Cell Surface and Intracellular TfR-GFP and NgCAM-GFP. During the first 2 days in culture, there were significant differences among stage 3 cells in the extent to which dendritic



7053

markers were polarized. To investigate this finding in more detail, we expressed a GFP construct of TfR that allowed us to compare the distribution of cell surface TfR-GFP (which could be selectively visualized by antibody staining of living cells) and TfR associated with intracellular organelles, including carrier vesicles and endosomes (which, along with cell surface TfR, could be visualized on the basis of GFP fluorescence).

Fig. 1. Changes in the degree of polarization of axonal and dendritic markers during development, (a) In stage 2 neurons, axonal and dendritic markers are not segregated into different neurites. The micrographs illustrate a cell (phase, Left) from a 1-day-old culture 18 h after coinfection with adenoviruses encoding untagged versions of NgCAM (an axonal marker) and LDLR (a dendritic marker). At this stage, labeling of cell surface NgCAM (Center) and LDLR (Right) was primarily observed in the growth cones (arrow). Although the extent of staining varied among different neurites, both the axonal and dendritic markers tended to be concentrated in the same neurites. (Bar, 25 µm.) (b) In stage 3 neurons, axonal and dendritic markers have a complementary distribution, indicating their polarization to different neurites. The micrographs illustrate a stage 3 cell from a 1-day-old culture 18 h after coinfection with NgCAM- and LDLR-encoding adenoviruses. Labeling of cell surface NgCAM (Left) showed a strong polarization to the axon (arrows), including its growth cone, whereas staining of the short dendritic processes (arrowheads) was largely absent. In contrast, cell surface staining of LDLR (Right) was prominent in cell body and dendrites but nearly absent from the axon. (Bar, 25 µm.) (c) As a measure of polarization, we quantified the percentage of staining for each marker protein that was associated with the dendritic arbor. The dendritic proteins TfR and LDLR were already preferentially localized to the dendritic arbor on day 1, and their polarization increased to mature levels by day 5. Likewise, the axonal proteins NgCAM and L1 were preferentially excluded from the dendrites on day 1; their polarization was essentially complete by day 5. A pIgR construct whose sorting signal had been mutated (pIgR665–668) served to illustrate the distribution of an unsorted protein. The percentage of this protein associated with the dendritic membrane decreased slightly during development, paralleling the relative increase in size of the axonal arbor. TfR and pIgR685–668 were expressed with replication defective herpesviruses and LDLR and NgCAM with replication defective adenoviruses. L1 is an endogenous protein. Each point represents data from 10–20 cells examined 12–18 h after infection. In stage 2 cells expressing TfR-GFP, GFP-labeled vesicular structures were found in all processes. In stage 3 cells, vesicular GFPtagged structures could be visualized in the dendrites and the axon, but the number and fluorescence intensity of these organelles was often lower in the axon than in the dendrite. There was a strong correlation between the presence of significant axonal surface labeling and the presence of GFP-tagged intracellular vesicles in the axon. In cells whose surface TfR was highly polarized to the dendrites, GFP-labeled vesicles were absent from the axon (Fig. 2 a). In cells that exhibited significant staining on the cell surface, GFP-labeled organelles were obvious, and the proximodistal distribution of the two was quite similar (Fig. 2 b). In the cell illustrated in Fig. 2 b, for example, the intensity of TfR staining in the distal axon (Fig. 2 b, right, arrows) was similar to that observed on the dendritic cell surface (Fig. 2 b, arrowheads). The staining for TfR and the fluorescence of GFP declined in the proximal axon. We measured the fraction of GFP label and cell surface staining associated with the dendritic compartment for 18 stage 3 cells after 2 days in culture (Fig. 2 c). The two measures were very tightly correlated. These results suggest that when TfR is present in intracellular compartments within the axon, the TfR is delivered to the cell surface. In contrast to the situation for TfR, the polarized distribution of NgCAM in mature neurons depends on a selectivity mechanism downstream of transport (1). NgCAM carrier vesicles are plentiful in dendrites but seem incompetent to fuse with the dendritic plasma membrane. To determine whether a similar mechanism is responsible for the polarized distribution of NgCAM to the axonal cell surface in stage 3 cells, we examined the cell surface and intracellular distribution of a GFP-tagged version of NgCAM. As for untagged NgCAM (Fig. 1 b Left), we found that cell surface NgCAM-GFP was already polarized to the axonal plasma membrane in stage 3 neurons (Fig. 3). However, vesicular structures labeled with NgCAM-GFP were prominent in both the axon and the dendrites, as observed in mature neurons (1). It is thus tempting to speculate that the same mechanisms are at work to polarize axonal and dendritic membrane proteins throughout development. Developmental Changes in the Transport of Carrier Vesicles Labeled with TfR-GFP and NgCAM-GFP. To visualize the transport of carrier vesicles, we made time-lapse recordings of neurons expressing TfR-GFP or NgCAM-GFP at times ranging from 2 days to 1 week in culture. For each recording, high-magnification images were acquired continuously over a recording period of 30 sec (600msec exposures). An example of such a recording from a cell expressing TfR-GFP is illustrated in Fig. 4, along with the method used for data analysis (also see Fig. 6 and Movies 1–4, which are published as supplemental data on the PNAS web site, www.pnas.org). Throughout development, the basic parameters of vesicle transport were essentially the same. The transport of vesicles labeled with either construct was always bidirectional. This observation was true for stage 2 neurites before specification, as well as for axons and dendrites of older cells. The average rate of transport was about 1 µm/sec (range 0.2–2.8 µm/sec). We did not detect differences in the rate of transport between vesicles labeled with TfR and NgCAM, between transport in axons and dendrites, or between transport in the anterograde and retrograde directions. In stage 2 cells, the transport behavior of NgCAM- and TfR-labeled vesicles was essentially the same. Most neurites exhibited



7054

robust bidirectional transport of both types of vesicles. By developmental stage 3, when one of the neurites had grown distinctly longer than the others, the pattern of transport of TfR-containing vesicles had changed dramatically (Fig. 4), whereas there was no change in the transport of NgCAM. In stage 3 neurons, few TfR-labeled vesicles were present in the axon, and the amount of transport was significantly diminished compared to stage 2 neurites and the dendrites of the same cell. Over a 40-µm length of the axon (Fig. 4 b Upper), we could detect only two vesicles that moved in the anterograde direction. In contrast, we detected eight anterograde movements in a 20-µm length of one of the dendrites of this cell (Fig. 4 b Lower). To quantify the amount of transport in each neurite, we generated kymographs (Fig. 4 c), which track vesicle position (shown of the y axis) as a function of time (shown on the x axis; see Methods for details). The marked difference in the number of transport events between axon and dendrites is apparent (Fig. 4 c). Far fewer diagonal lines, corresponding to moving vesicles, are seen in the axon compared to all of the different dendrites. Analysis of the flux of TfR-GFP carrier vesicles in and

Fig. 2. The polarization of TfR to the dendritic plasma membrane parallels the exclusion of TfR-containing carrier vesicles from the axon. TfR-GFP was expressed by using a defective herpesvirus. Cell surface TfR was assessed by staining living cells with an anti-TfR antibody, whereas GFP fluorescence served as a measure of all expressed TfR, including that associated with intracellular vesicles, (a and b) On day 2, the polarization of TfR varied somewhat from cell to cell. In some cells (a), surface staining for TfR (Center) was absent from the axon (arrows), which was paralleled by the absence of axonal TfR-GFP fluorescence associated with intracellular vesicles (Right). Staining in dendritic processes (arrowheads) was readily observed with both labels. In other cells (b), surface staining and TfR-GFP fluorescence were present in the distal axon (arrows) at a level comparable to that in the dendrites (arrowheads). The GFP fluorescence illustrates all TfR present in cells, including carrier vesicles in dendritic and axonal processes. (Bar, 20 µm.) (c) On the basis of a cell-by-cell comparison, there was a close correlation between the degree of polarization of cell surface TfR and TfR-GFP fluorescence. The total fluorescence in all dendritic processes was expressed as a percentage of the total fluorescence in all neurites including the axon. Cells in culture for 1 day were infected with replication-defective herpesvirus encoding TfR-GFP. After 18 h, living cells were stained with antibody to detect protein expression on the cell surface.

Fig. 3. In stage 3 neurons, cell surface staining for NgCAM was restricted to the axon, whereas vesicles containing NgCAM-GFP were present in all processes (Left, phase contrast; Center, cell surface staining; Right, GFP; arrowheads denote dendrites). Cells in culture for 1 day were infected with replication-defective herpesvirus encoding NgCAM-GFP. After 18 h, living cells were stained with antibody to detect protein expression on the cell surface. (Bar, 20 µm.)



7055

out of the axon (i.e., the summed translocation of all moving vesicles) revealed a slight bias of transport in the retrograde direction. This finding is consistent with the idea that some of the TfR present in immature axons may be cleared by transport back to the cell body. In dendrites, the transport of TfR shows an anterograde bias. In the case of NgCAM, there was no difference in the number of anterograde transport events in the axon compared to the dendrites.

Fig. 4. Comparison of the transport of TfR-GFP carrier vesicles in axons and dendrites of stage 3 cells, (a) A stage 3 neuron expressing TfR-GFP (Right, phase contrast; Left, GFP fluorescence); note the higher level of TfR-GFP fluorescence in the dendrites compared to the faint fluorescence in the proximal axon. Vesicle transport in this cell was recorded over a period of 30 sec, capturing images every 600 msec. Movies 1–4 of these data are published as supplemental data on the PNAS web site, www.pnas.org. (Bar, 20 µm.) (b) Vesicle transport in the proximal axon (Upper) and a representative dendrite (Lower). The topmost panel shows an enlarged view of the axonal segment (boxed in a). The path of each vesicle that moved in the anterograde direction or the retrograde direction during the 30-sec recording is shown in the two succeeding panels. Lower shows an enlarged view of one dendrite (boxed in a), followed by the path of each vesicle that moved in the anterograde and retrograde directions. Many more vesicles travel into the dendrite than the axon. To enable the visualization of faint vesicles in the axon, contrast was enhanced relative to the dendrite. (c) To quantify transport, recordings from TfR-GFP-expressing cells were analyzed by using kymographs, which show anterogradely moving vesicles as diagonal lines with positive slope, whereas retrogradely moving vesicles are represented by lines with negative slopes. This analysis revealed that there is extensive anterograde vesicle traffic into each dendrite but few transport events in the axon.



7056

Fig. 5. Changes in the amount of transport of TfR and NgCAM during development in culture. In the case of carrier vesicles labeled with TfR-GFP, the number of anterograde transport events in stage 2 cells (square), which lack an axon, is roughly comparable to the number of events seen in dendrites throughout development (open circles). In contrast, the number of TfRGFP-containing vesicles entering the developing axon drops abruptly when the cells enter developmental stage 3 (filled circles). In the case of NgCAM, the number of anterograde transport events in dendrites remains constant during development, whereas the number of NgCAM-GFP-containing vesicles entering the axon increases gradually during development. Anterograde transport events were quantified by using kymograph analysis and normalized for the duration of the recording and the length of the neurite included in the image. The data for mature cells (>14 days in culture) were taken from ref. 1. We used kymograph analyses to assess changes in the amount of vesicle transport in axons and dendrites during development in culture, focusing on the delivery of vesicles carrying these marker proteins to the axonal and dendritic domains. These data are summarized in Fig. 5. Interestingly, the number of transport events in the undifferentiated neurites of stage 2 cells was comparable for both marker proteins. Moreover, the amount of transport found in stage 2 neurites was maintained in the immature dendrites of stage 3 cells. This observation was true for both TfR and NgCAM, even though NgCAM does not appear on the dendritic surface. Even in mature dendrites, whose arbors are many times longer than those of stage 3 cells, the amount of dendritic transport was unchanged. In contrast to the sustained levels of transport in the developing dendrites, we observed a marked decrease in the amount of TfR entering the axon of stage 3 cells. By 2 days, the frequency of anterograde transport into axons had already declined to half that seen in stage 2 neurites. By day 3, the selectivity of TfR transport, measured as the ratio of TfR vesicles entering the axon compared with the dendrites of the same cell, already approached mature levels (1). Although there was a profound change in the axonal transport of TfR that occurred at developmental stage 3, no change was seen in the axonal transport of NgCAM. The number of NgCAM anterograde transport events in the axons of stage 3 cells was not significantly different from that in the unspecified neurites of stage 2 cells. Instead, the amount of NgCAM transported into the axon increased gradually after day 2, eventually doubling by day 14. DISCUSSION To determine the time course of polarization of dendritic and axonal proteins, we used replication-defective herpesviruses and adenoviruses to express two dendritic proteins (TfR and LDLR) and an axonal protein (NgCAM) in hippocampal neurons at different times after plating. We found that TfR, LDLR, NgCAM, and its endogenous homolog, L1, were differentially distributed in neurons at stage 3 of development, as soon as the axon could be unambiguously identified. Polarization reached mature levels by day 5 in culture. These results are consistent with previous studies examining the polarization of the endogenous axonal membrane proteins synapsin I, synaptophysin, and L1 (6, 7), and the endogenous dendritic proteins TfR and telencephalin (10, 11). In contrast, studies examining the distribution of other dendritic membrane proteins, including GluR1, GluR2/3, GABAA receptors, and the LDLR-related protein, have concluded that these proteins are not detectably concentrated in dendrites until stage 4 (8, 9, 22). There are several possible explanations for this discrepancy. It could reflect real differences among the mechanisms that underlie the sorting of different classes of dendritic proteins. Alternatively, it could reflect the difficulty of accurately assessing the polarity of endogenous proteins early in development, when their expression levels are quite low and the extent of their polarization may be lower than in mature neurons. The situation is further complicated because some previous studies examined cells after fixation, thereby revealing intracellular as well as cell surface labeling. We believe that the method used here, based on expression of marker proteins at levels that makes their distribution easy to measure, offers a more accurate method to assess protein polarization in young neurons. One drawback of our approach is that overexpressing exogenous proteins could perturb the sorting machinery, resulting in an underestimate of protein polarization. It is difficult to imagine, however, that overexpression could lead to an overestimate of polarization. Using this approach, we obtained clear quantitative evidence that TfR and LDLR are preferentially concentrated in the dendritic membrane compared with the unpolarized marker pIgR665–668, and that this difference is evident by 1 day in culture. We also examined changes in the transport behavior of axonal and dendritic carrier vesicles. In stage 2 cells, the transport behavior of NgCAM- and TfR-labeled vesicles was essentially the same. This result is consistent with previous studies, which indicated that stage 2 neurites have not yet been specified as either axons or dendrites (3, 4). The transition from stage 2 to 3 is marked by the rapid and prolonged extension of a single



7057

neurite, which becomes specified as the axon. In discussing mechanisms that might regulate the delivery of new membrane needed for growth, Futerman and Banker (23) raised the possibility that the transport of carrier vesicles may be regulated in accordance with the rate of neurite elongation. According to this view, one might expect the transition from stage 2 to 3 to be accompanied by an increase in the number of axonal carrier vesicles entering the axon. Bradke and Dotti (4) have proposed that the transition from stage 2 to 3 is accompanied by a reorganization of intracellular transport, from multidirectional (into all neurites) to unidirectional (into the emerging axon). According to their model, this concerted change in transport affects a broad variety of organelles, including carrier vesicles conveying both axonal and dendritic proteins, as well as mitochondria and peroxisomes. Our analysis did not reveal the changes in transport predicted by either of these models. In the case of carrier vesicles containing the axonal protein NgCAM, we found no significant increase in the amount of transport into stage 3 axons compared with unspecified stage 2 neurites, nor was the amount of transport into the axon of stage 3 cells greater than into their dendrites. In the case of carrier vesicles conveying the dendritic protein TfR, we did not observe the increase in its axonal trafficking predicted by the Bradke and Dotti model. Instead, far fewer TfR carrier vesicles entered the nascent axon than entered the neurites of stage 2 cells; in stage 3 cells, TfR vesicles were preferentially transported to the dendrites, not to the axon. One important limitation in the current study is that the methods we used for expressing GFP constructs do not yield high levels of expression in very young neurons. Thus we were unable to assess transport before 2 days in culture. It is possible that there are changes in transport that occur concomitantly with axonal specification, but that these changes are not maintained throughout developmental stage 3. Alternative methods will be required to address this possibility. We have previously shown that in mature neurons, the polarization of NgCAM to the axonal plasma membrane does not depend on directed transport but instead involves events at the plasma membrane, most likely the preferential fusion of NgCAM carrier vesicles with the axonal membrane (ref. 1 and unpublished observations). Because cell surface NgCAM is polarized by stage 3, whereas NgCAM carrier vesicles are transported into both dendrites and axons at this stage, it is tempting to speculate that the same mechanism used in mature cells is responsible for the polarization of NgCAM early in development. What changes occur in neurons between developmental stages 2 and 3 that might initiate the polarization of cell surface proteins? In the case of dendritic proteins like TfR, it is highly likely that these changes involve the establishment of selective microtubule-based transport. We have shown that carrier vesicles containing TfR are preferentially transported into the dendrites at developmental stage 3, although the selectivity of this process is not as great as in mature cells. Moreover, our results show that at the early stages when TfRcontaining vesicles are not fully excluded from the axon, TfR is expressed on the axonal surface. This finding indicates that there is no additional quality control mechanism downstream of transport to prevent TfR-containing vesicles from fusing with the axonal membrane. It is thought that the selective microtubule-based transport that prevents the movement of dendritic carrier vesicles into the axon depends on regional biochemical differences within the neuron (1). These differences might take the form of biochemical differences among microtubules in different regions of the cell or of local differences in the regulation of components of the motor protein-carrier vesicle complex. Of the biochemical characteristics that distinguish axonal from dendritic microtubules in mature neurons, some have been shown to arise early in development. For example, although the microtubule-associated protein τ is uniformly distributed in stage 3 neurons, it is differentially phosphorylated in dendrites (24). Similarly, phosphorylated MAP1B is expressed in a proximodistal gradient in axons of cortical and sensory neurons (25, 26). These data suggest that a unique complement of kinase and phosphatase activities is present in developing axons. In addition to producing posttranslational differences in microtubule proteins, local differences in kinase or phosphatase activity could also regulate motor activity or the interaction of motor proteins with cargo vesicles, thereby inhibiting the delivery of dendritic carrier vesicles to the axon (27–29). Similarly, local posttranslational modifications could selectively regulate the vesicle fusion machinery, potentially inhibiting the fusion of NgCAM carrier vesicle in the dendritic domain (30). We thank Hannelore Asmussen, Jon Muyskens, and Barbara Smoody for the preparation of neuronal cultures, and Silvia LaRue, Julie Harp, and Sarah Godsey for excellent technical assistance. This work was supported by National Institutes of Health Grant NS17112. 1. Burack, M.A., Silverman, M.A. & Banker, G. (2000) Neuron 26, 465–472. 2. Dotti, C.G., Sullivan, C.A. & Banker, G.A. (1988) J. Neurosci. 8, 1454–1468. 3. Craig, A.M. & Banker, G. (1994) Annu. Rev. Neurosci. 17, 267–310. 4. Bradke, F. & Dotti, C.G. (2000) Curr. Opin. Neurobiol. 10, 574–581. 5. Esch, T., Lemmon, V. & Banker, G. (2000) J. Neurocytol. 29, 215–223. 6. van den Pol, A.N. & Kim, W.T. (1993) J. Comp. Neurol. 332, 237–257. 7. Fletcher, T.L., Cameron, P., De Camilli, P. & Banker, G. (1991) J. Neurosci. 11, 1617–1626. 8. Brown, M.D., Banker, G.A., Hussaini, I.M., Gonias, S.L. & VandenBerg, S.R. (1997) Brain Res. 747, 313–317. 9. Killisch, I., Dotti, C.G., Laurie, D.J., Luddens, H. & Seeburg, P.H. (1991) Neuron 7, 927–936. 10. Mundigl, O., Matteoli, M., Daniell, L., Thomas-Reetz, A., Metcalf, A., Jahn, R. & De Camilli, P. (1993) J. Cell Biol. 122, 1207–1221. 11. Benson, D.L., Yoshihara, Y. & Mori, K. (1998) J. Neurosci. Res. 52, 43–53. 12. Casanova, J.E., Apodaca, G. & Mostov, K.E. (1991) Cell 66, 65–75. 13. Herz, J. & Gerard, R.D. (1993) Proc. Natl. Acad. Sci. USA 90, 2812–2816. 14. Kowal, R.C., Herz, J., Goldstein, J.L., Esser, V. & Brown, M.S. (1989) Proc. Natl. Acad. Sci. USA 86, 5810–5814. 15. Lemmon, V. & McLoon, S.C. (1986) J. Neurosci. 6, 2987–2994. 16. Jing, S.Q., Spencer, T., Miller, K., Hopkins, C. & Trowbridge, I.S. (1990) J. Cell Biol. 110, 283–294. 17. Omary, M.B. & Trowbridge, I.S. (1981) J. Biol. Chem. 256, 12888–12892. 18. Goslin, K., Asmussen, H. & Banker, G. (1998) in Culturing Nerve Cells, eds. Banker, G. & Goslin, K. (MIT Press, Cambridge, MA), pp. 339–370. 19. Jareb, M. & Banker, G. (1998) Neuron 20, 855–867. 20. West, A.E., Neve, R.L. & Buckley, K.M. (1997) J. Neurosci. 17, 6038–6047. 21. Goslin, K., Schreyer, D.J., Skene, J.H. & Banker, G. (1990) J. Neurosci. 10, 588–602. 22. Craig, A.M., Blackstone, C.D., Huganir, R.L. & Banker, G. (1993) Neuron 10, 1055–1068. 23. Futerman, A.H. & Banker, G.A. (1996) Trends Neurosci. 19, 144–149. 24. Mandell, J.W. & Banker, G.A. (1996) J. Neurosci. 16, 5727–5740. 25. Mansfield, S.G., Diaz-Nido, J., Gordon-Weeks, P.R. & Avila, J. (1991) J. Neurocytol. 20, 1007–1022. 26. Bush, M.S., Goold, R.G., Moya, F. & Gordon-Weeks, P.R. (1996) Eur. J. Neurosci. 8, 235–248. 27. Thaler, C.D. & Haimo, L.T. (1996) Int. Rev. Cytol. 164, 269–327. 28. Sato-Harada, R., Okabe, S., Umeyama, T., Kanai, Y. & Hirokawa, N. (1996) Cell Struct. Funct. 21, 283–295. 29. Reese, E.L. & Haimo, L.T. (2000) J. Cell Biol. 151, 155–166. 30. Hsu, S.C., Hazuka, C.D., Foletti, D.L. & Scheller, R.H. (1999) Trends Cell Biol. 9, 150–153.


MOLECULAR ORGANIZATION OF THE POSTSYNAPTIC SPECIALIZATION

7058

Colloquium Molecular organization of the postsynaptic specialization

Morgan Sheng* Department of Neurobiology, and Howard Hughes Medical Institute, Massachusetts General Hospital and Harvard Medical School, 50 Blossom Street (Wel 423), Boston, MA 02114 A specific set of molecules including glutamate receptors is targeted to the postsynaptic specialization of excitatory synapses in the brain, gathering in a structure known as the postsynaptic density (PSD). Synaptic targeting of glutamate receptors depends on interactions between the C-terminal tails of receptor subunits and specific PDZ domain-containing scaffold proteins in the PSD. These scaffold proteins assemble a specialized protein complex around each class of glutamate receptor that functions in signal transduction, cytoskeletal anchoring, and trafficking of the receptors. Among the glutamate receptor subtypes, the N-methyl-Daspartate receptor is relatively stably integrated in the PSD, whereas the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor moves in and out of the postsynaptic membrane in highly dynamic fashion. The distinctive cell biological behaviors of Nmethyl-D-aspartate and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors can be explained by their differential interactions with cytoplasmic proteins. Excitatory synapses predominantly use glutamate as the neurotransmitter. When viewed by electron microscopy, excitatory synapses are characterized by an electron-dense thickening of the postsynaptic membrane, termed the postsynaptic density (PSD). Containing specific receptors for the neurotransmitter glutamate, as well as numerous receptor-associated proteins, the PSD can be regarded as a proteinaceous “organelle” specialized for postsynaptic signal transduction. A disk-like structure 30–40 nm thick and up to a few hundred nm wide, the PSD is relatively insoluble in nonionic detergents and can be purified to a considerable degree by differential centrifugation (1). Because it is a prime example of a subcellular molecular microdomain and contains the critical proteins involved in synaptic plasticity, the PSD has been intensively studied in recent years (reviewed in refs. 2 and 3). In neuronal excitatory synapses, glutamate receptors are cardinal components of the postsynaptic specialization and are highly concentrated in the PSD. Recent advances in understanding the molecular organization of the PSD has stemmed largely from studies of glutamate receptors and their interacting proteins. The major postsynaptic glutamate receptors include N-methyl-D-aspartate (NMDA) receptors, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, and the group I metabotropic glutamate receptors (mGluRs), which are linked to phospholipase C and phosphoinositide turnover. These glutamate receptors are specifically targeted to the postsynaptic membrane, indeed, even to specific subdomains within the postsynaptic specialization (4). THE NMDA RECEPTOR-PSD-95 COMPLEX NMDA receptors are a consistent feature of excitatory synapses of the forebrain, whereas AMPA receptor content is highly variable; indeed, a significant fraction of excitatory synapses lack AMPA receptors altogether (5–7). This differential regulation implies that NMDA and AMPA receptors use distinct mechanisms for synaptic targeting. NMDA receptors are heteromeric (probably tetrameric) complexes composed of NR1 and NR2 subunits (8). The four different NR2 subunits (NR2A to NR2D) possess long cytoplasmic tails, the C termini of which end in the conserved sequence -ESDV or -ESEV. This Cterminal peptide motif binds to the PDZ domains of PSD-95/SAP90, an abundant constituent of the PSD (9–15). PSD-95/SAP90 belongs to the MAGUK superfamily of proteins, which are characterized by the presence of PDZ domains, a Src homology 3 domain, and a guanylate kinase-like (GK) domain. PDZ domains are modular protein domains of 90 aa that are specialized for binding to C-terminal peptides in a sequence-specific fashion (16–18). The interaction between NR2 subunits of the NMDA receptor and PSD-95 is important for specific localization of NMDA receptors in the PSD (19, 20) and in the coupling of NMDA receptors to cytoplasmic signaling pathways (21, 22). For instance, by binding to neuronal nitric oxide synthase (nNOS), PSD-95 facilitates the activation of nNOS by NMDA receptor-mediated calcium influx (23, 24). In addition, PSD-95 is likely to aid in the anchoring of NMDA receptors to the postsynaptic cytoskeleton (25). The general concept of a postsynaptic scaffolding function for MAGUK proteins is supported by genetic studies of Discs large in Drosophila. Discs large (the fly homolog of PSD-95) is important for development of the neuromuscular junction in Drosophila larvae and required for synaptic localization of its binding partners: the Shaker potassium channel and the Fasciclin II adhesion molecule (26–28). It should be emphasized that in addition to NMDA receptors, PSD-95 probably organizes other membrane proteins (such as adhesion molecules, receptor tyrosine kinases, and ion channels) in the postsynaptic specialization of mammalian neurons. Thus PSD-95-associated proteins may serve anchoring and signaling functions that are not exclusively related to NMDA receptors. For instance, PSD-95 has been reported to bind kainate receptors, a less well-characterized class of ionotropic glutamate receptor that also exists at postsynaptic sites (29). The NMDA receptor/PSD-95 protein complex in the PSD is growing rapidly in size and complexity as newer technologies such as mass spectrometry are used to study its components (30). The total number of proteins in the PSD may be as high as a few hundred, especially if one includes proteins that are only weakly enriched in, or transiently associated with, the PSD. In general, our

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: PSD, postsynaptic density; NMDA, N-methyl-D-aspartate; AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; mGluR, metabotropic glutamate receptor; GK, guanylate kinase-like; IP3R, IP3 receptor. *E-mail: [email protected].



7059

understanding of the functional significance of proteins in the PSD lags behind the pace of their identification. Many of the proteins in the NMDA receptor/PSD-95 complex are specifically and highly enriched in the postsynaptic specialization. An example is SynGAP, a GTPase-activating protein for Ras, which has a C terminus that interacts with all three PDZ domains of PSD-95 (31, 32). The function of SynGAP remains unclear, but it may be involved in regulation of Ras activation in response to NMDA receptor stimulation. A protein termed SPAR, a GTPase protein for Rap, which binds to the GK domain of PSD-95, has been identified (D.Pak and M.S., unpublished observations). SPAR contains two domains that associate with actin and dramatically reorganize the actin cytoskeleton in heterologous cells. SPAR appears to regulate the size and shape of dendritic spines via its GAP activity, thus implicating Rap signaling in the control of postsynaptic structure. In addition to SPAR, the GK domain of PSD-95 family proteins binds to an abundant family of proteins in the PSD, termed GKAP (also named SAPAP or DAP) (33–36). The C terminus of GKAP in turn binds to the PDZ domain of Shank, a family of scaffold proteins containing multiple additional protein interaction domains including ankyrin repeats, Src homology 3 domain, and proline-rich motifs (37, 38). Via one of these proline-rich motifs, Shank interacts with Homer (37, 38), a cytoplasmic adaptor protein originally discovered by Worley and coworkers (39) as a binding partner of group I mGluRs. The NMDA receptor/PSD-95 complex therefore is potentially linked to mGluRs via Shank and Homer. The EVH1 domain of Homer binds to an internal sequence motif (consensus sequence PPXXF) in the proline-rich region of Shank and in the cytoplasmic tail of mGluR1/5 (40, 41). Homer proteins typically contain a coiled-coil domain that mediates self-association (41, 42). These “CC-Homers” multimerize to form multivalent complexes that can crosslink multiple proteins that contain the PPXXF motif (41). Several other proteins have been noted to contain the PPXXF Homer-binding consensus, including the IP3 receptor (IP3R), a downstream effector in the mGluR signaling pathway. Multimeric Homer has the potential therefore to link together mGluRs with IP3Rs, mGluRs with Shank, and IP3Rs with Shank. IP3Rs are concentrated in the smooth endoplasmic reticulum, an intracellular calcium store that extends into dendritic spines and often approaches the postsynaptic specialization (43). Thus the morphological basis exists in dendritic spines for a close interaction between postsynaptic mGluRs, the NMDA receptor complex, and intracellular calcium stores. It is believed that Homer brings IP3Rs into close proximity of the group 1 mGluRs, thereby allowing for more efficient coupling between surface mGluRs and intracellular calcium stores (40). Because Shank is a component of the NMDA receptor complex via binding to GKAP (37), the HomerShank interaction potentially links the group 1 mGluRs to the NMDA receptor and its associated proteins (38). Shank and Homer also may contribute to a functional coupling between NMDA receptors and intracellular calcium stores. Shank and Homer are highly and specifically enriched in the PSD and are located at the cytoplasmic face of the PSD (in contrast to PSD-95, which is located close to the postsynaptic membrane). This “deep” location within the PSD is well-suited for potential interactions of Shank and Homer with cytoplasmic proteins and the smooth endoplasmic reticulum. In addition, Shank and Homer could interact with the postsynaptic cytoskeleton that impinges on the cytoplasmic face of the PSD. Indeed, an interaction between Shank and the actin-binding protein cortactin has been discovered (37). Consistent with a role in cytoskeletal regulation, overexpression of Shank in cultured neurons induces enlargement of dendritic spines (C.Sala and M.S., unpublished work). The spine promoting effect depends on synaptic targeting of Shank and the ability of Shank to bind Homer. Thus Shank and Homer cooperate to induce enlargement of dendritic spines. In addition, Shank and Homer act synergistically to recruit IP3R to dendritic spines, presumably by direct binding of IP3R to Homer (C.Sala and M.S., unpublished work). Because they are indirectly associated with NMDA receptors and mGluRs, Shank and Homer may be able to couple morphological responses of dendritic spines to changes in synaptic activity. The NR1 subunit of the NMDA receptor also participates in a variety of interactions with specific cytoskeletal and signaling proteins (25). Together, the NMDA receptor subunits interact with a multitude of intracellular proteins, either directly or indirectly via scaffold proteins like PSD-95. The immediate envelope of protein interactions that anchors and integrates NMDA receptors in the PSD (the PSD-95 protein complex) can be regarded as a key modular subdomain of the postsynaptic specialization. REGULATED SYNAPTIC TARGETING OF AMPA RECEPTORS Although AMPA receptors also are concentrated at postsynaptic sites of excitatory synapses, the synaptic levels of AMPA receptors are much more heterogeneous than those of NMDA receptors. Some excitatory synapses contain NMDA receptors but not AMPA receptors, especially early in development (5–7). It is also apparent that a large fraction of AMPA receptors lies within intracellular compartments. The synaptic distribution of AMPA receptors can be altered by activity (44, 45), and recent studies suggest rapid activityregulated delivery of AMPA receptors to synapses (46–48). Thus the synaptic targeting of AMPA receptors appears to be regulated on a much shorter time scale than for NMDA receptors. The rapid movements of AMPA receptors into and out of the postsynaptic membrane has revealed a surprisingly dynamic regulation of the postsynaptic specialization. AMPA receptors are typically composed of heteromeric combinations of GluR1–4 subunits (8, 49), whose membrane topology is similar to that of NMDA receptor subunits. The C-terminal cytoplasmic tails of AMPA receptor subunits interact with a distinct set of cytoplasmic proteins than do NMDA receptors. These differential protein interactions presumably underlie the differential regulation of synaptic targeting of NMDA and AMPA receptor-channels. The GluR2 and GluR3 subunits of AMPA receptors share a C-terminal sequence (-SVKI) that interacts with the fifth PDZ domain of GRIP/ABP, a family of proteins containing six or seven PDZ domains (50– 52). GRIP is enriched in synapses in the brain, but to only a modest degree when compared with PSD-95. GRIP also differs from PSD-95 in being relatively abundant in intracellular compartments in dendrites and cell bodies of neurons, suggesting that GRIP may be involved in trafficking of AMPA receptors, rather than/in addition to synaptic anchoring (51, 53, 54). The fact that overexpression of the C-terminal tail of GluR2 in neurons inhibits synaptic clustering of AMPA receptors (50) is consistent with either an anchoring or trafficking role for GRIP. Blocking GluR2-GRIP interactions also prevents potentiation of synaptic responses, suggesting that binding to GRIP is involved in recruitment of functional AMPA receptors to the synapse (55). Similarly, mutation of the C terminus of GluR1 (which binds to the PDZ domain protein SAP97; ref. 56) also prevents its functional recruitment to synapses (47). Thus interactions between the C terminus of AMPA receptor subunits and PDZ domain scaffold proteins appear to be important for synaptic targeting and/or stabilization of AMPA receptors (57). In addition to GRIP/ABP, the C-terminal sequence of GluR2/3 mediates binding to PICK-1 (58), another PDZ-containing protein previously shown to bind protein kinase C (59). Phosphorylation of the C terminus of GluR2 prevents its binding to GRIP but not to PICK-1 (60–62), suggesting the



7060

possibility of a phosphorylation-dependent switch in AMPA receptor interaction with PDZ proteins. Phosphorylation-dependent changes in PDZ interactions could regulate the sorting of AMPA receptors during exocytosis and/or endocytosis (63). A surprising finding was that GluR2 binds to NSF, an ATPase involved in membrane fusion and vesicle trafficking (64–66). NSF binding is mediated by a membrane proximal segment of GluR2’s cytoplasmic tail, distinct from the C terminus that binds to GRIP or PICK-1. Surface expression of AMPA receptors is inhibited by peptides that block the GluR-NSF interaction, suggesting that NSF is involved in the insertion or stabilization of AMPA receptors in the postsynaptic membrane (67). The binding of NSF to AMPA receptor GluR2 subunits in particular seems to allude to the dynamic nature of the trafficking and regulation of AMPA receptors. It is possible that the NSFGluR2 interaction is relevant to synaptic plasticity by regulating the vesicle trafficking or protein unfolding of AMPA receptors (reviewed in ref. 68). Endocytosis of postsynaptic AMPA receptors is likely to be an important means of depressing excitatory transmission (69–71). The underlying dynamics and molecular mechanisms are being uncovered. Using immunofluorescence and surface biotinylation assays, a rapid rate of basal AMPA receptor endocytosis in cultured hippocampal neurons, which is further accelerated in response to synaptic activity, ligand binding, and insulin, has been measured (63). AMPA-induced AMPA receptor internalization is mediated in part by depolarization and calcium influx through voltage-dependent calcium channels and in part by a novel ligand-binding mechanism that is independent of receptor activation. The endocytosis of AMPA receptors depends on dynamin, but multiple signaling pathways converge on this final mechanism (63, 72). For instance, insulin- and AMPA-induced AMPA receptor internalization differentially depend on protein phosphatases; furthermore, they require distinct sequence determinants within the cytoplasmic tails of GluR1 and GluR2 subunits. Once internalized AMPA receptors can be sorted to different destinations. AMPA receptors internalized in response to AMPA stimulation enter a recycling endosome system, whereas those internalized in response to insulin diverge into a distinct (possibly degradative) compartment. Thus the molecular mechanisms and intracellular sorting of AMPA receptors are diverse and depend on the internalizing stimulus (63, 72). In contrast to AMPA receptors, NMDA receptors show negligible internalization over the time course of minutes to an hour (63). CONCLUDING REMARKS From the many recent studies reviewed above, a daunting picture is emerging of the molecular complexity of the postsynaptic specialization of glutamatergic synapses. Glutamate receptors (which are fundamental components of the postsynaptic membrane) use their cytoplasmic domains to interact with a variety of intracellular proteins. The receptor is thus anchored by, and integrated into, a sophisticated protein network that supports the receptor’s postsynaptic actions and that modulates the receptor’s activity. Within individual synapses, different subclasses of neurotransmitter receptors (e.g., AMPA, NMDA, and mGluRs) are segregated by differential protein interactions into distinct molecular environments that correspond to localized signaling microdomains. Examples include the PSD-95-based protein complex, which brings (among other things) calcium-regulated molecules into the sphere of influence of the NMDA receptor-calcium channel. These distinct microdomains are linked together in the overall PSD architecture by scaffold proteins such as Shank lying deep in the PSD. The NMDA receptor/PSD-95 complex can be regarded as a relatively stable core substructure of the PSD, whereas AMPA receptors and their associated proteins are much more dynamically regulated. The differential protein interactions of NMDA receptors and AMPA receptors ultimately will explain the contrasting cell biological behaviors of these two different glutamate receptors. Obviously, we have only reached a qualitative descriptive phase in the analysis of the molecular organization of the PSD. It will be critical to determine the stoichiometry and geometry of interactions involving glutamate receptors and other proteins, if we are to appreciate the true functional architecture of this postsynaptic organelle. It should be also clear that we have a rather static view of postsynaptic structure. An important future challenge is to uncover the developmental and activity-dependent regulation of the protein interactions that underlie the dynamics of the postsynaptic specialization. I am an Assistant Investigator of the Howard Hughes Medical Institute. This work is supported by the National Institutes of Health (Grant NS35050). 1. Carlin, R., Grab, D., Cohen, R. & Siekevitz, P. (1980) J. Cell Biol. 86, 831–845. 2. Ziff, E.B. (1997) Neuron 19, 1163–1174. 3. Kennedy, M.B. (1997) Trends Neurosci. 20, 264–268. 4. Takumi, Y., Matsubara, A., Rinvik, E. & Ottersen, O.P. (1999) Ann. N.Y.Acad. Sci. 868, 474–482. 5. Nusser, Z., Lujan, R., Laube, G., Roberts, J.D., Molnar, E. & Somogyi, P. (1998) Neuron 21, 545–559. 6. Petralia, R.S., Esteban, J.A., Wang, Y.-X., Partridge, J.G., Zhao, H.-M., Wenthold, R.J. & Malinow, R. (1999) Nat. Neurosci. 2, 31–36. 7. Takumi, Y., Ramirez-Leon, V., Laake, P., Rinvik, E. & Ottersen, O.P. (1999) Nat. Neurosci. 2, 618–624. 8. Dingledine, R., Borges, K., Bowie, D. & Traynelis, S.F. (1999) Pharmacol. Rev. 51, 7–61. 9. Cho, K.-O., Hunt, C.A. & Kennedy, M.B. (1992) Neuron 9, 929–942. 10. Kistner, U., Wenzel, B.M., Veh, R.W., Cases-Langhoff, C., Garner, A.M., Appeltauer, U., Voss, B., Gundelfinger, E.D. & Garner, C.C. (1993) J. Biol. Chem. 268, 4580–4583. 11. Kornau, H.-C, Schenker, L.T., Kennedy, M.B. & Seeburg, P.H. (1995) Science 269, 1737–1740. 12. Niethammer, M., Kim, E. & Sheng, M. (1996) J. Neurosci. 16, 2157–2163. 13. Sheng, M. (1996) Neuron 17, 575–578. 14. Kornau, H.C., Seeburg, P.H. & Kennedy, M.B. (1997) Curr. Opin. Neurobiol. 7, 368–373. 15. O’Brien, R., Lau, L. & Huganir, R. (1998) Curr. Opin. Neurobiol. 8, 364–369. 16. Ponting, C.P. (1997) Protein Sci. 6, 464–468. 17. Doyle, D.A., Lee, A., Lewis, J., Kim, E., Sheng, M. & MacKinnon, R. (1996) Cell 85, 1067–1076. 18. Songyang, Z., Fanning, A.S., Fu, C., Xu, J., Marfatia, S.M., Chishti, A.H., Crompton, A., Chan, A.C., Anderson, J.M. & Cantley, L.C. (1997) Science 275, 73–77. 19. Steigerwald, F., Schulz, T.W., Schenker, L.T., Kennedy, M.B., Seeburg, P.H. & Kohr, G. (2000) J. Neurosci. 20, 4573–4581. 20. Mori, H., Manabe, T., Watanabe, M., Sath, Y., Suzuki, N., Toki, S., Nakamura, K., Yagi, T., Kushiya, E., Takahashi, T., et al. (1998) Neuron 21, 571–580. 21. Migaud, M., Charlesworth, P., Dempster, M., Webster, L.C., Watabe, A.M., Makhinson, M., He, Y., Ramsay, M.F., Morris, R.G., Morrison, J.H., et al. (1998) Nature (London) 396, 433–439. 22. Sprengel, R., Suchanek, B., Amico, C., Brusa, R., Burnasheve, N., Rozov, A., Hvalby, O., Jensen, V., Paulsen, O., Andersen, P., et al. (1998) Cell 92, 279–289. 23. Brenman, J.E., Chao, D.S., Gee, S.H., McGee, A.W., Craven, S.E., Santillano, D.R., Wu, Z., Huang, F., Xia, H., Peters, M.F., et al. (1996) Cell 84, 757–767. 24. Sattler, R., Xiong, Z., Lu, W.Y., Hafner, M., MacDonald, J.F. & Tymianski, M. (1999) Science 284, 1845–1848. 25. Sheng, M. & Pak, D.T.S. (2000) Annu. Rev. Physiol. 62, 755–778. 26. Tejedor, F.J., Bokhari, A., Rogero, O., Gorczyca, M., Zhang, J., Kim, E., Sheng, M. & Budnik, V. (1997) J. Neurosci. 17, 152–159. 27. Thomas, U., Kim, E., Kuhlendahl, S., Ho Koh, Y., Gundelfinger, E.D., Sheng, M., Garner, C.C. & Budnik, V. (1997) Neuron 19, 787–799. 28. Zito, K., Fetter, R.D., Goodman, C.S. & Isacoff, E.Y. (1997) Neuron 19, 1007–1016. 29. Garcia, E.P., Mehta, S., Blair, L.A., Wells, D.G., Shang, J., Fukushima, T., Fallon, J.R., Garner, C.C. & Marshall, J. (1998) Neuron 21, 727–739. 30. Husi, H., Ward, M., Choudhary, J., Blackstock, W. & Grant, S. (2000) Nat. Neurosci. 3, 661–669.



7061

31. Kim, J.H., Liao, D., Lau, L.F. & Huganir, R.L. (1998) Neuron 20, 683–691. 32. Chen, H.J., Rojas-Soto, M., Oguni, A. & Kennedy, M.B. (1998) Neuron 20, 895–904. 33. Kim, E., Naisbitt, S., Hsueh, Y.-P., Rao, A., Rothschild, A., Craig, A.M. & Sheng, M. (1997) J. Cell. Biol. 136, 669–678. 34. Naisbitt, S., Kim, E., Weinberg, R.J., Rao, A., Yang, F.-C., Craig, A.M. & Sheng, M. (1997) J. Neurosci. 17, 5687–5696. 35. Takeuchi, M., Hata, Y., Hirao, K., Toyoda, A., Irie, M. & Takai, Y. (1997) J. Biol. Chem. 272, 11943–11951. 36. Satoh, K., Yanai, H., Senda, T., Kohu, K., Nakamura, T., Okumura, N., Matsumine, A., Kobayashi, S., Toyoshima, K. & Akiyama, T. (1997) Genes Cells 2, 415–424. 37. Naisbitt, S., Kim, E., Tu, J.C., Xiao, B., Sala, C., Valtschanoff, J., Weinberg, R.J., Worley, P.F. & Sheng, M. (1999) Neuron 23, 569–582. 38. Tu, J.C., Xiao, B., Naisbitt, S., Yuan, J.P., Petralia, R.S., Brakeman, P., Doan, A., Aakalu, V.K., Lanahan, A.A., Sheng, M. & Worley, P.F. (1999) Neuron 23, 583–592. 39. Brakeman, P.R., Lanahan, A.A., O’Brien, R., Roche, K., Barnes, C.A., Huganir, R.L. & Worley, P.F. (1997) Nature (London) 386, 284–288. 40. Tu, J.C., Xiao, B., Yuan, J.P., Lanahan, A.A., Leoffert, K., Li, M., Linden, D.J. & Worley, P.F. (1998) Neuron 21, 717–726. 41. Xiao, B., Tu, J.C., Petralia, R.S., Yuan, J.P., Doan, A., Breder, C.D., Ruggiero, A., Lanahan, A.A., Wenthold, R.J. & Worley, P.F. (1998) Neuron 21, 707–716. 42. Kato, A., Ozawa, F., Saitoh, Y., Fukazawa, Y., Sugiyama, H. & Inokuchi, K. (1998) J. Biol. Chem. 273, 23969–23975. 43. Spacek, J. & Harris, K. (1997) J. Neurosci. 17, 190–203. 44. Liao, D., Zhang, X., O’Brien, R., Ehlers, M.D. & Huganir, R.L. (1999) Nat. Neurosci. 2, 37–43. 45. Carroll, R.C., Lissin, D.V., von Zastrow, M., Nicoll, R.A. & Malenka, R.C. (1999) Nat. Neurosci. 2, 454–460. 46. Shi, S.H., Hayashi, Y., Petralia, R. S., Zaman, S.H., Wenthold, R.J., Svoboda, K. & Malinow, R. (1999) Science 284, 1811–1816. 47. Hayashi, Y., Shi, S.H., Esteban, J.A., Piccini, A., Poncer, J.C. & Malinow, R. (2000) Science 287, 2262–2267. 48. Malinow, R., Mainen, Z.F. & Hayashi, Y. (2000) Curr. Opin. Neurobiol. 10, 352–357. 49. Hollmann, M. & Heinemann, S. (1994) Annu. Rev. Neurosci. 17, 31–108. 50. Dong, H., O’Brien, R.J., Fung, E.T., Lanahan, A.A., Worley, P.F. & Huganir, R.L. (1997) Nature (London) 386, 279–284. 51. Wyszynski, M., Kim, E., Yang, F.-C. & Sheng, M. (1998) Neuropharmacology 37, 1335–1344. 52. Srivastava, S., Osten, P., Vilim, F., Khatri, L., Inman, G., States, B., Daly, C., DeSouza, S., Abagyan, R., Valtschanoff, J., et al. (1998) Neuron 21, 581–591. 53. Wyszynski, M., Valtschanoff, J.G., Naisbitt, S., Dunah, A.W., Kim, E., Standaert, D.G., Weinberg, R.J. & Sheng, M. (1999) J. Neurosci. 19, 6528– 6537. 54. Dong, H., Zhang, P., Song, I., Petralia, R.S., Liao, D. & Huganir, R.L. (1999) J. Neurosci. 19, 6930–6941. 55. Li, P., Kerchner, G.A., Sala, C., Wei, F., Huettner, J.E., Sheng, M. & Zhuo, M. (1999) Nat. Neurosci. 2, 972–977. 56. Leonard, A.S., Davare, M.A., Horne, M.C., Garner, C.C. & Hell, J.W. (1998) J. Biol. Chem. 273, 19518–19524. 57. Osten, P., Khatri, L., Perez, J.L., Kohr, G., Giese, G., Daly, C., Schulz, T.W., Wensky, A., Lee, L.M. & Ziff, E.B. (2000) Neuron 27, 313–325. 58. Xia, J., Zhang, X., Staudinger, J. & Huganir, R.L. (1999) Neuron 22, 179–187. 59. Staudinger, J., Zhou, J., Burgess, R., Elledge, S.J. & Olson, E.N. (1995) J. Cell. Biol. 128, 263–271. 60. Chung, H.J., Xia, J., Scannevin, R.H., Zhang, X. & Huganir, R.L. (2000) J. Neurosci. 20, 7258–7267. 61. Matsuda, S., Mikawa, S. & Hirai, H. (1999) J. Neurochem. 73, 1765–1768. 62. Matsuda, S., Launey, T., Mikawa, S. & Hirai, H. (2000) EMBO J. 19, 2765–2774. 63. Lin, J.W., Ju, W., Foster, K., Lee, S.H., Ahmadian, G., Wyszynski, M., Wang, Y.T. & Sheng, M. (2000) Nat. Neurosci. 3, 1282–1290. 64. Nishimune, A., Isaac, J.T., Molnar, E., Noel, J., Nash, S.R., Tagaya, M., Collingridge, G.L., Nakanishi, S. & Henley, J.M. (1998) Neuron 21, 87–97. 65. Osten, P., Srivastava, S., Inman, G.J., Vilim, F.S., Khatri, L., Lee, L.M., States, B.A., Einheber, S., Milner, T.A., Hanson, P.I. & Ziff, E.B. (1998) Neuron 21, 99–110. 66. Song, I., Kamboj, S., Xia, J., Dong, H., Liao, D. & Huganir, R.L. (1998) Neuron 21, 393–400. 67. Noel, J., Ralph, G.S., Pickard, L., Williams, J., Molnar, E., Uney, J.B., Collingridge, G.L. & Henley, J.M. (1999) Neuron 23, 365–376. 68. Lin, J.W. & Sheng, M. (1998) Neuron 21, 267–270. 69. Lüscher, C., Xia, H., Beattie, E.C., Carroll, R.C., von Zastrow, M., Malenka, R.C. & Nicoll, R.A. (1999) Neuron 24, 649–658. 70. Man, Y.H., Lin, J.W., Ju, W.H., Ahmadian, G., Liu, L., Becker, L.E., Sheng, M. & Wang, Y.T. (2000) Neuron 25, 649–662. 71. Carroll, R.C., Beattie, E.C., Xia, H., Luscher, C., Altschuler, Y., Nicoll, R.A., Malenka, R.C. & von Zastrow, M. (1999) Proc. Natl. Acad. Sci. USA 96, 14112–14117. 72. Beattie, E.C., Carroll, R.C., Yu, X., Morishita, W., Yasuda, H., von Zastrow, M. & Malenka, R.C. (2000) Nat. Neurosci. 3, 1291–1300.


A CELLULAR MECHANISM FOR TARGETING NEWLY SYNTHESIZED MRNAS TO SYNAPTIC SITES ON DENDRITES

7062

Colloquium A cellular mechanism for targeting newly synthesized mRNAs to synaptic sites on dendrites Oswald Steward*† and Paul F.Worley‡ Research Center, and Departments of Anatomy/Neurobiology and Neurobiology and Behavior, College of Medicine, University of California, Irvine, CA 92697; and ‡Department of Neuroscience, The Johns Hopkins University School of Medicine, Baltimore, MD, 21205 Long-lasting forms of activity-dependent synaptic plasticity involve molecular modifications that require gene expression. Here, we describe a cellular mechanism that mediates the targeting newly synthesized gene transcripts to individual synapses where they are locally translated. The features of this mechanism have been revealed through studies of the intracellular transport and synaptic targeting of the mRNA for a recently identified immediate early gene called activity-regulated cytoskeleton-associated protein Arc. Arc is strongly induced by patterns of synaptic activity that also induce long-term potentiation, and Arc mRNA is then rapidly delivered into dendrites after episodes of neuronal activation. The newly synthesized Arc mRNA localizes selectively at synapses that recently have been activated, and the encoded protein is assembled into the synaptic junctional complex. The dynamics of trafficking of Arc mRNA reveal key features of the mechanism through which synaptic activity can both induce gene expression and target particular mRNA transcripts to the active synapses. Information storage in the nervous system is thought to involve changes in synaptic potency that occur in response to particular patterns of activity. One candidate process is long-term potentiation (LTP), and the key features that make it an attractive candidate mechanism include: (i) LTP is long-lasting, enduring for hours and sometimes much longer, (ii) LTP is expressed selectively at synapses that have experienced particular patterns of activity (synapse specificity). (iii) LTP requires presynaptic activity in conjunction with a sufficient level of postsynaptic depolarization (the Hebb postulate), (iv) LTP can be induced by patterns of activity that central nervous system neurons actually exhibit, (v) The late stages of LTP, like the consolidation phase of memory, occur over a period of hours after the inducing event and require protein synthesis and perhaps the transcription of new gene products (for recent reviews, see refs. 1–4). Although it is the late, protein synthesis-dependent phase of LTP that is of particular interest as a candidate mechanism of information storage, relatively little is known about the actual cellular and molecular mechanisms that bring this enduring change about. There are three general possibilities: *Reeve-Irvine

(i) Plasticity could involve changes in the state of the existing molecules of the synapse (changes in phosphorylation state, or other posttranslational modifications). These sorts of changes are likely to account for the initial change in synaptic strength, but it is more difficult to explain how these sorts of changes could endure beyond a few hours. (ii) Plasticity could involve changes in the molecular composition of existing synapses. For example, there is increasing evidence that α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors are removed from the synapse during the induction of long-term depression and inserted during the induction of LTP. There also may be changes in other synaptic constituents. (iii) Plasticity could involve a structural change in synapses (increases or decreases in synapse size, formation of new synapses, or elimination of existing ones). If enduring synaptic modifications require the selective delivery of new molecular constituents to the synapses that are to be modified, this could be accomplished in three ways: (i) The key proteins critical for modification could be synthesized in the cell body and delivered selectively to the synapses that are to be modified through some selective transport process, (ii) The key proteins could be synthesized in the cell body and be widely distributed, with synapse specificity being conferred by a selective capture mechanism (see for example ref. 5). (iii) The key proteins could be synthesized on-site as the result of translation of mRNAs that are localized at synapses. These possibilities are not mutually exclusive, and different molecules could be targeted to synapses in different ways. The present review documents that neurons do possess a mechanism through which newly synthesized mRNA transcripts are targeted to active synapses where they mediate the local synthesis of proteins that become part of the synapse. This mechanism has been revealed through studies of the intracellular transport and synaptic targeting of the mRNA for a unique immediate early gene (IEG) called Arc (for activity-regulated cytoskeleton-associated protein). Arc, also known as Arg 3.1, is noteworthy because it is induced by neuronal activity like other IEGs, but the newly synthesized mRNA is rapidly delivered throughout dendrites. Moreover, intense synaptic activity causes the mRNA to localize selectively at synapses that had been activated. The induction of gene expression, delivery of the mRNA to dendrites, and synthesis of the protein occur during the first few hours after the inducing event—approximately the same time period in which protein synthesis-dependent synaptic modifications are occurring. Importantly, Arc protein is assembled into the matrix of the synaptic junctional complex (SJC), demonstrating that the mechanism can operate for protein constituents of the synapse. In what follows, we will summarize the key features of this mechanism and also propose a unifying hypothesis that may explain why certain synaptic proteins are locally synthesized. INDUCTION AND DENDRITIC TARGETING OF ARC MRNA AFTER INTENSE NEURONAL ACTIVITY Arc was initially discovered in screens for novel IEGs that are induced by neuronal activity in a protein synthesis-independent

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: LTP, long-term potentiation; ECS, electroconvulsive shock; SJC, synaptic junctional complex; NRC, N-methyl-Daspartate receptor complex; IEG, immediate early gene; psd, postsynaptic density; NMDA, N-methyl-D-aspartate. †To whom reprint requests should be addressed at: Reeve-Irvine Research Center, 1105 Gillespie Neuroscience Building, College of Medicine, University of California, Irvine, CA 92697 E-mail: [email protected].



7063

fashion (6, 7). Arc was unique because in contrast to the mRNAs of other IEGs, Arc mRNA rapidly migrates throughout the dendritic arbor of the neuron in which it is induced. This was discovered through in situ hybridization analyses of the distribution of Arc mRNA in the dentate gyrus after a single electroconvulsive shock (ECS). For example, by 2 h after a single ECS, newly synthesized Arc mRNA is distributed throughout the molecular layer of the dentate gyrus, which contains the dendrites of dentate granule cells (Fig. 1) whereas the mRNAs for other IEGs remain tightly localized to the region of the cell body.

Fig. 1. Newly synthesized Arc mRNA is selectively targeted to dendritic domains that have been synaptically activated. The photomicrographs illustrate the distribution of Arc mRNA as revealed by nonisotopic in situ hybridization in nonactivated dentate gyrus (A), 2 h after a single electroconvulsive seizure (B), and after delivering high-frequency trains to the medial perforant path over a 2-h period (C). Note the uniform distribution of Arc mRNA across the dendritic laminae after an ECS and the prominent band of labeling in the middle molecular layer after high-frequency stimulation of the perforant path. (D) Schematic illustration of the dendrites of a typical dentate granule cell and the pattern of termination of medial perforant path projections. HF, hippocampal fissure; GCL, granule cell layer. (A and B) [Reproduced with permission from ref. 28 (Copyright 2001, Elsevier Science)]. (D) [Reproduced with permission from ref. 9 (Copyright 1998, Elsevier Science)]. Because Arc is expressed as an IEG, the synthesis, intracellular trafficking, localization, and life history of Arc mRNA can be studied in a way that is not possible with mRNAs that are expressed constitutively. Evaluations of Arc mRNA distribution at different times after an ECS indicate that the mRNA reaches the most distal tips of the granule cell dendrites within 1 h after the inducing stimulus. The distance from the granule cell body layer to the distal tips of the dendrites is about 300 µm. Thus, Arc mRNA moves into dendrites at a rate of at least 300 µm per h (8). The evaluation of Arc expression at various times after ECS also revealed that Arc mRNA was present in dendrites only transiently. Peak levels of Arc mRNA were seen 1–2 h after a single ECS; thereafter, the levels of Arc mRNA declined, returning to near control levels after about 6 h (8). Interestingly, this is approximately the same time interval during which synaptic modifications are sensitive to inhibition of protein synthesis. NEWLY SYNTHESIZED ARC MRNA IS SELECTIVELY TARGETED TO SYNAPSES THAT HAVE RECENTLY BEEN ACTIVATED Subsequent studies of Arc revealed another remarkable feature—that newly synthesized Arc mRNA is selectively targeted to synapses that have been strongly activated (9). This was discovered initially in studies of Arc mRNA distribution after high-frequency stimulation of the entorhinal cortical projections to the dentate gyrus using a paradigm typically used to induce LTP. The projection from the entorhinal cortex to the dentate gyrus (the perforant path) terminates in a topographically organized fashion along the dendrites of dentate granule cells. Projections from the medial entorhinal cortex terminate selectively in the middle molecular layer of the dentate gyrus, whereas projections from the lateral entorhinal cortex terminate in the outer molecular layer. By positioning a stimulating electrode in different parts of the entorhinal cortex, it is possible to selectively activate a band of synapses that terminate on particular proximo-distal segments. High-frequency activation of the projections to middle dendritic domains (400-hz trains, eight pulses per train, delivered at a rate of 1/10 sec) strongly induces Arc expression. If high-frequency stimulation is continued as the newly synthesized mRNA migrates into dendrites, the mRNA localizes selectively in the middle molecular layer in exactly the location of the band of synapses that had been activated. This selective localization is evidenced by a prominent band of labeling for Arc mRNA in the middle molecular layer of the dentate gyrus (Fig. 1). When the medial perforant path is activated, the levels of labeling remain quite low in the outer molecular layer, indicating that newly synthesized Arc mRNA never migrates into the distal dendrites. This finding is in contrast to the situation after an ECS, where there are high levels of labeling through-out the molecular layer (compare Fig. 1 B and C), which suggests that as the mRNA enters the dendrites, it is somehow captured in the activated dendritic segments. An analysis of the distribution of Arc mRNA after various periods of stimulation (Fig. 2) further supports this idea. After 30 min of stimulation, Arc mRNA is still confined to the cell body layer. With continued stimulation, levels of labeling in the activated dendritic lamina increase progressively, whereas there is minimal, if any, increase in labeling in the nonactivated distal dendritic segments (the outer molecular layer). Thus, newly synthesized Arc mRNA appears to be captured by active synapses, preventing the further migration of the mRNA into more distal segments. ARC MRNA IS TARGETED TO DIFFERENT DENDRITIC DOMAINS DEPENDING ON THE POPULATION OF SYNAPSES THAT ARE ACTIVATED The intradendritic distribution of newly synthesized Arc mRNA is determined by the populations of synapses that are activated. For example, high-frequency stimulation of the lateral entorhinal cortex, which innervates distal dendritic segments, produces a band of labeling for Arc mRNA in the outer molecular layer. When the projections to proximal dendritic laminae are strongly activated, newly synthesized Arc mRNA localizes precisely in a band corresponding to the zone of activation. The synapses that terminate in this proximal dendritic lamina originate from large neurons in the hilus of the dentate gyrus. These project bilat



7064

erally, and so the projection system is called the commissural/ associational pathway. The synapses are excitatory, but stimulation of the pathway also evokes strong γ-aminobutyric acid (GABA)ergic inhibition via interneurons. Consequently, LTP can only be induced in this pathway when GABAergic inhibition is blocked (10). GABAergic inhibition can be blocked by positioning micropipettes containing bicuculline in the dentate gyrus during the period of high frequency stimulation. The diffusion of the bicuculline from the pipette blocks GABAergic inhibition locally in an area of about 1 mm diameter, thus enabling the induction of LTP (10). When the commissural pathway is activated under conditions of GABAergic blockade, Arc is strongly induced in the area surrounding the bicuculline-filled micropipette, and the newly synthesized mRNA localizes precisely in the inner molecular layer (9). Given this mechanism for targeting, it is likely, by extension, that Arc mRNA distribution also can be regulated on a finer scale, perhaps even on a synapse-by-synapse basis. The signal(s) that mediate this localization process remain to be defined.

Fig. 2. Analysis of the distribution of Arc mRNA after various periods of synaptic stimulation. The graph illustrates the distribution of Arc mRNA in the molecular layer of the dentate gyrus after various periods of stimulation of the medial perforant path. After 30 min of stimulation, Arc mRNA is still confined to the cell body layer. With continued stimulation, levels of labeling increase progressively in the activated dendritic lamina, whereas there is minimal if any increase in labeling in the nonactivated distal dendritic segments (the outer molecular layer). Thus, newly synthesized Arc mRNA appears to be captured by active synapses, preventing the further migration of the mRNA into more distal segments. [Reproduced with permission from ref. 9 (Copyright 1998, Elsevier Science)]. THE SELECTIVITY OF LOCALIZATION INVOLVES TARGETING OF MRNA TO ACTIVE DOMAINS AND MIGRATION/DEPLETION FROM INACTIVE REGIONS A noteworthy feature of the pattern of labeling produced by stimulation of the medial entorhinal cortex is that the levels of labeling in the activated dendritic lamina are higher than in the lamina containing the more proximal dendrites of granule cells. This is true even though the mRNA would have to move through the proximal dendrites en route to the activated lamina. A possible explanation for this pattern of labeling comes from recent findings about how organelles move in dendrites. For example, mitochondria and membrane vesicles exhibit both orthograde and retrograde movements, sometimes even reversing direction (see refs. 11 and 12). The same basic bidirectional movement is exhibited by fluorescently labeled RNA granules (13). Given these patterns of movement, it seems reasonable to expect that Arc mRNA also may move bidirectionally once the mRNA enters dendrites, shuttling back and forth unless and until the mRNA docks (is captured). In this situation, the docking of the mRNA in the activated lamina in response to synaptic activation would prevent retrograde movement of the mRNA back into proximal dendritic regions. Another possibility is that synaptic stimulation might actually cause Arc mRNA to migrate from inactive to active regions, depleting the mRNA from inactive segments. Evidence that this does occur has come from studies that use a different induction paradigm, designed to differentiate between the signals that induce Arc expression and those that mediate the localization. In this paradigm, Arc expression was induced by delivering an ECS. Then, the rat was anesthetized, and stimulation and recording electrodes were positioned so as to activate the medial perforant path on one side. The time between the ECS and the completion of the preparation for physiology was typically 30–45 min. Stimulus intensity was set so as to evoke an approximately half-maximal population spike (3- to 5-mV amplitude). Then, highfrequency trains (400-hz trains, eight pulses per train) were delivered to the perforant path beginning 1.5 h or 1.75 h after the ECS. Stimulation was delivered for 30 or 15 min, respectively, just before the animals were killed and perfused for in situ hybridization (in both cases, perfusion occurred 2 h after the ECS). The key to this experiment is that at 1.5 or 1.75 h post-ECS, Arc mRNA would have been present throughout the dendrites when the stimulation was initiated, in the pattern illustrated in Fig. 1 B. Remarkably, as little as 15 min of synaptic stimulation was sufficient to produce a prominent band of labeling for Arc mRNA in the middle molecular layer of the dentate gyrus (Fig. 3 B). Because it occurred so quickly, the development of the band is likely to represent redistribution of the Arc mRNA that is already in the dendrite rather than transport of Arc mRNA from the cell body. After 30 min of stimulation, the band became more distinct as levels of labeling decreased in the nonactivated laminae, especially in the outer molecular layer (Fig. 3 E and F). Thus, synaptic activation caused newly synthesized Arc mRNA to rapidly redistribute to the activated zone (as is evident with 15 min of stimulation), and depleted the mRNA from nonactivated regions of the dendrites (as seen with 30 min or more of stimulation). It is important to note that after prolonged periods of synaptic stimulation (2 h), the overall levels of labeling in the molecular layer are lower than on the side that received an ECS only (Fig. 2). This finding suggests that in addition to causing the newly synthesized mRNA to redistribute to active synaptic sites, synaptic activation may enhance mRNA degradation. This enhanced degradation could be linked to the targeting of the mRNA to the activated zone or could be caused by signals generated throughout the dendrite as a consequence of the intense depolarization. Local stabilization of mRNA is also seen in oocytes and developing embryos, where certain processes contribute to a generalized degradation of mRNA which is countered by a local stabilization in certain cytoplasmic domains (14).



7065

Fig. 3. Redistribution of Arc mRNA after localized synaptic activation. In these experiments, Arc expression was induced by delivering an ECS. Then, the rat was anesthetized, and stimulation and recording electrodes were positioned so as to activate the medial perforant path on one side. High frequency trains (400-hz trains, eight pulses per train) were delivered for 30 or 15 min, respectively just before the animals were killed and perfused for in situ hybridization (in both cases, perfusion occurred 2 h after the ECS). As little as 15 min of synaptic stimulation was sufficient to produce a prominent band of labeling for Arc mRNA in the middle molecular layer of the dentate gyrus (B). After 30 min of stimulation, the band became more distinct as levels of labeling decreased in the nonactivated laminae, especially in the outer molecular layer (E and F). [Reproduced with permission from ref. 28 (Copyright 2001, Elsevier Science)]. LOCALIZATION OF ARC MRNA IN ACTIVATED DENDRITIC LAMINAE IS ASSOCIATED WITH A LOCAL ACCUMULATION OF ARC PROTEIN Immunostaining of tissue sections from stimulated animals using an Arc-specific antibody revealed a band of newly synthesized protein in the same dendritic laminae in which Arc mRNA was concentrated (Fig. 4) (9). The fact that synaptic activation leads to the selective targeting of both recently synthesized mRNA and protein suggests that the targeting of the mRNA underlies a local synthesis of the protein. One additional important point revealed by immunocyto-chemistry is that newly synthesized Arc protein also is targeted to the nucleus. The significance of this dual targeting to active synapses and the nucleus is not yet known. ARC PROTEIN IS ASSEMBLED INTO THE POSTSYNAPTIC DENSITY (PSD)/N-METHYL-D-ASPARTATE (NMDA) RECEPTOR COMPLEX (NRC) An important clue to the function of Arc protein comes from recent evidence that the protein is concentrated in the psd and is one of a collection of proteins that are linked to the NMDA receptor. In the experiment illustrated in Fig. 5, subcellular fractions enriched in various types of cellular membranes were prepared by using established techniques (15). Band 1 contains myelin; band 2 contains nonspecialized plasma membrane; band 3 contains synaptic plasma membranes, and the pellet contains mitochondria. When band 3 is treated with detergent to remove membrane lipids and integral membrane proteins, the remaining insoluble residue represents a fraction that is highly enriched in insoluble proteins of the SJC. Western blot analyses of these subcellular fractions using antibodies against Arc and other synaptic molecules revealed that Arc protein is present at the highest relative levels in the SJC. This is the same distribution seen for other known components of the psd, like CAMKII. A similar line of evidence comes from studies of proteins that copurify with the NMDA receptor (termed NMDAR multiprotein complex or NRC, see ref. 16). In this study, the protein constituents of the NRC were identified by mass spectroscopy combined with large-scale immunoblotting. The immunoblotting experiments revealed that Arc (going by the name Arg 3.1 in that paper) was prominently represented in the NRC: Arc was not among the proteins that were detected



7066

by mass spectroscopy, however. It is interesting that a similar study, which used mass spectroscopy to identify protein constituents of the core psd, did not detect Arc (17). The starting material in that study was a psd fraction prepared by subcellular fractionation and detergent extraction, which is similar in composition to the SJC fraction. The study identified many of the same proteins that were present in the NRC, leading to the speculation that the NRC and the core psd may be different views of a subcellular structure specialized for postsynaptic signal transduction (18). The fact that Arc was not detected is probably because it is present in relatively low levels and is thus detectable by immunoblotting but not mass spectroscopy. It cannot be excluded, however, that the SJC fraction contains a slightly different complement of proteins than the psd fraction, and that Arc protein is extracted by the detergents in the preparation of the psd fraction.

Fig. 4. Arc protein accumulates in activated dendritic laminae in the same pattern as Arc mRNA. (A) Immunostaining of tissue sections from stimulated animals using an Arc-specific antibody revealed a band of newly synthesized protein in the same dendritic laminae in which Arc mRNA was concentrated. Note the sharp boundary (arrows) between the middle molecular layer (mml, the site of termination of the synapses that were activated) and the inner molecular layer. The fact that synaptic activation leads to the selective targeting of both recently synthesized mRNA and protein suggests that the targeting of the mRNA underlies a local synthesis of the protein. HF, hippocampal fissure; gcl, granule cell layer. (B) Newly synthesized Arc protein also is concentrated in the nucleus. Short arrows indicate examples of labeled nuclei. The experiments above were carried out in resting animals, and so it remains to be established whether the newly synthesized Arc protein that is induced by behavioral experience or synaptic activation is also targeted to the synaptic junctional region, and if so, over what time course. LOCAL PROTEIN SYNTHESIS: A MECHANISM MEDIATING COTRANSLATIONAL ASSEMBLY OF CERTAIN MOLECULES INTO THE SJC/NRC? Since the discovery of a selective localization of polyribosomes beneath postsynaptic sites, one key question has remained unanswered: why are certain proteins synthesized locally whereas most others are synthesized in the cell body? A hypothesis is suggested by the following: (i) The psd/NRC appears to be a highly organized multimolecular structure specialized for postsynaptic signal transduction (18). It seems very likely that proper signaling requires a precise stoichiometric relationship between the different molecules making up the complex.

Fig. 5. Evidence from subcellular fractionation experiments that Arc protein is concentrated at the synaptic junction. Shown is a slot blot of protein samples from subcellular fractions prepared according to the procedure of ref. 15 that have been stained with various antibodies. Band 1 contains myelin; band 2 contains nonsynaptic plasma membrane; band 3 contains synaptic plasma membranes (SPM). SJC is the fraction enriched in postsynaptic densities obtained by detergent extraction of band 3. Note that Arc protein is present at the highest relative levels in the synaptic plasma membrane and SJC fractions as are α and ß isoforms of CAMKII and fodrin, which are highly enriched in psd. (ii) The different protein components of the psd/NRC turn over at quite different rates. For example, Arc protein has a short halflife (a few hours). The other proteins have much longer half-lives (probably days), although the exact value is not known. The important implication of this fact is that the different molecular constituents of the psd/NRC would have to be replaced in existing psds by substitution. (iii) The molecular components of the psd/NRC are almost certainly linked together through precisely controlled intermolecular interactions. Creating these links probably requires that the proteins be in particular conformations. For certain other highly organized structures, proper protein-protein interactions may require cotranslational assembly. (iv) Finally, and most importantly, the mRNAs for several of the molecules that are part of the psd/NRC are present in dendrites. This is true of Arc, the a-subunit of CAMKII (19), and also shank. Together these facts suggest the hypothesis that certain proteins are locally synthesized because they must be assembled into the psd/ NRC complex by cotranslational assembly. It will be of considerable interest to take a closer look at whether any of the mRNAs encoding other constituents of the NRC complex are also present in dendrites (and conversely, whether the protein products of other dendritic mRNAs are part of the NRC). RIBOSOMES AT THE PSD? The hypothesis for cotranslational assembly predicts that ribosomes and other components of the translational machinery would have to be closely associated with the psd as they synthesize molecules that require cotranslational assembly. Moreover, given the targeting of Arc mRNA to active synapses, one might predict an increase in ribosomes associated with the psd after high-frequency stimulation. This issue is actually difficult to address with electron microscopic techniques because the electron density of mature postsynaptic densities is nearly the



7067

same as the electron density of ribosomes. If present as singlet ribosomes, it is very likely that they would be virtually invisible with conventional electron microscopy. In light of these ideas, it is of interest to assess the ultrastructural appearance of synapses that have experienced the intense activation used herein to induce Arc expression and targeting. Electron micrographs of synapses in the middle molecular layer of the dentate gyrus after 2 h of medial perforant path stimulation reveal striking modifications of spine shape (Fig. 6 compare A, control side, with B, stimulated). In particular, the synapses in the activated zone undergo a dramatic shape change and assume a chalice-like configuration that is highly reminiscent of the shapes exhibited by highly motile spines (20, 21). Similar shape changes have been described after brief periods of stimulation in a standard LTP paradigm (22). These shapes invite the speculation that high-frequency stimulation induces a period of intense spine motility. It is noteworthy, however, that one does not find obvious examples of polyribosomes near or embedded within the psd. Indeed, polyribosomes are difficult to find in the spines that exhibit the dramatic shape changes.

Fig. 6. Ultrastructural evidence for spine motility in synapses that have experienced intense synaptic activation. Illustrated are synapses in the middle molecular layer of the dentate gyrus on the control nonstimulated side (A) and after 2 h of high-frequency stimulation of the medial perforant path (B). Note that on the stimulated side, spines exhibit a chalise-like form that is remarkably similar to the form of highly motile spines. Animals received medial perforant path stimulation as described for 2 h and then were perfused with 2% paraformaldehyde/2% glutaraldehyde and prepared for electron microscopy. Photomicrographs then were taken in the middle molecular layer on the stimulated and control nonstimulated sides, den, dendrite; s, spine; t, terminal. These observations recall an earlier quantitative evaluation of synapse morphology after the induction of LTP that was, until now, rather curious—that fewer polyribosomes are detectable in and around synapses after inducing LTP (23). One interpretation of these observations is that strong synaptic activation triggers a translocation of ribosomes from the spine base or head to the psd, and that ribosomes that embedded in the electrondense psd become undetectable by conventional electron microscopy. It is important to note that our hypothesis of local synthesis of psd proteins revisits an hypothesis proposed in 1981 (24). As part of a study of synaptogenesis in the cerebellar cortex, Palacios-Pru et al. (24) provided electron microscopic images of what appeared to be ribosomes in close association with immature psds on developing spines of Purkinje cells in the cerebellum. Based on these images, it was suggested that during early development, the psd was synthesized by ribosomes that were actually in immediate contact with it (see also ref. 25). If it turns out that ribosomes are embedded within mature psds and mediate cotranslational assembly of components of the NRC, what was then a controversial hypothesis will have been vindicated. Clearly it is now important to explore this issue with modern immunocytochemical or other techniques. LESSONS FROM THE STUDY OF ARC. A CANDIDATE MECHANISM FOR PROTEIN SYNTHESIS-DEPENDENT SYNAPTIC MODIFICATION Studies of Arc reveal elements of a mechanism that is well-suited to mediate the sorts of molecular changes in synapses that are believed to underlie long-term synaptic plasticity. (i) The expression of the Arc gene is triggered by the patterns of synaptic activity that lead to enduring synaptic modification (LTP). (ii) Arc mRNA encodes a protein that is targeted to synapses, and also to the nucleus. (iii) The activity-dependent induction of Arc protein occurs during a time window that extends for a few hours after the inducing stimulus.



7068

(iv) Arc mRNA and protein are targeted to synapses that have experienced particular patterns of activity. (v) Arc is induced in response to the sorts of brief behavioral experience that can lead to long-lasting synaptic modifications. (vi) Finally, Arc mRNA is induced in neuron types that are thought to participate in enduring synaptic modification in response to behavioral experience (neurons in the hippocampus and cerebral cortex). There are a number of pieces of the puzzle that are still missing, however. First, it remains to be established whether Arc protein in fact plays a role in activity-induced synaptic modification. Additional clues about the role of the protein will likely come from studies of the protein itself and its interactions with other functional molecules of the postsynaptic density. Certainly, the fact that Arc may be linked to the NMDA receptor in some way is an important clue in this regard. But even if Arc does not play a direct role, the way that Arc is handled by neurons reveals the existence of previously unknown RNA trafficking mechanisms that could be used for sorting other mRNAs that do play a key role in bringing about activity-dependent modifications. The fascinating properties of Arc should not make us lose sight of the fact that other mRNAs are present in dendrites constitutively, including the mRNAs for molecules that have been strongly implicated in activity-dependent synaptic modification (the mRNA for the asubunit of CAMII kinase, for example). These mRNAs that are present constitutively provide an opportunity for local regulation of the synthesis of key signaling molecules via translational regulation. Hence, gene expression at individual synapses is likely to be regulated in a complex fashion. One level of regulation would be in the mRNAs available for translation (i.e., Arc). Another level might involve regulation of translation of the mix of mRNAs that are in place, including those present constitutively (a model of which might be the translational regulation of fragile-X, refs. 26 and 27). How this is coordinated and how all of these molecules actually fit in to the molecular consolidation process remains to be established. Thanks to Kelli Sharp and Jamie Zaffis for technical assistance. This work was supported by National Institutes of Health Grants NS12333 (O.S.) and MH 53603 (P.F.W.). 1. Bailey, C.H., Bartsch, D. & Kandel, E.R. (1996) Proc. Natl. Acad. Sci. 93, 13445–13452. 2. Mayford, M., Bach, M.E., Huang, Y.-Y., Wang, L., Hawkins, R.D. & Kandel, E.R. (1996) Science 274, 1678–1683. 3. Nguyen, P.V. & Kandel, E.R. (1996) J. Neurosci. 16, 3189–3198. 4. Jones, M.W., Errington, M.L., French, P.J., Fine, A., Bliss, T.V.P., Garel, S., Charnay, P., Bozon, B., Laroche, S. & Davis, S. (2001) Nat. Neurosci. 4, 289–296. 5. Frey, U. & Morris, R.G.M. (1997) Nature (London) 385, 533–536. 6. Link, W., Konietzko, G., Kauselmann, G., Krug, M., Schwanke, B., Frey, U. & Kuhl, K. (1995) Proc. Natl. Acad. Sci. 92, 5734–5738. 7. Lyford, G., Yamagata, K., Kaufmann, W., Barnes, C., Sanders, L., Copeland, N., Gilbert, D., Jenkins, N., Lanahan, A. & Worley, P. (1995) Neuron 14, 433–445. 8. Wallace, C.S., Lyford, G.L, Worley, P.F. & Steward, O. (1998) J. Neurosci. 18, 26–35. 9. Steward, O., Wallace, C.S., Lyford, G.L. & Worley, P.F. (1998) Neuron 21, 741–751. 10. Steward, O., Tomasulo, R. & Levy, W.B. (1990) Brain Res. 516, 292–300. 11. Ligon, L.A. & Steward, O. (2000) J. Comp. Neurol. 427, 340–350. 12. Silverman, M.A., Kaech, S., Jareb, M., Burack, M.A., Vogt, L., Sonderegger, D. & Banker, G. (2001) Proc. Natl. Acad. Sci. USA 98, 7051–7057. 13. Knowles, R.B., Sabry, J.H., Martone, M.E., Deerinck, T.J., Ellisman, M.H., Bassell, G.J. & Kosik, K.S. (1996) J. Neurosci. 16, 7812–7820. 14. Bashirullah, A., Cooperstock, R.L. & Lipshitz, H.D. (2001) Proc. Natl. Acad. Sci. USA 98, 7025–7028. 15. Cotman, C.W. & Taylor, D. (1972) J. Cell Biol. 55, 696–710. 16. Husi, H., Ward, M.A., Choudhary, J.S., Blackstock, W.P. & Grant, S.G.N. (2000) Nat. Neurosci. 3, 661–669. 17. Walikonis, R.S., Jensen, O.E., Mann, M., Provance, D.W.J., Mercer, J.A. & Kenedy, M.B. (2000) J. Neurosci. 20, 4069–4080. 18. Sheng, M. & Lee, S.H. (2000) Nat. Neurosci. 3, 633–635. 19. Burgin, K.E., Washam, M.N., Rickling, S., Westgate, S.A., Mobley, W.C. & Kelly, P.T. (1990) J. Neurosci. 10, 1788–1798. 20. Fischer, M., Kaech, S., Wagner, U., Brinkhaus, H. & Matus, A. (2000) Nat. Neurosci. 3, 887–894. 21. Kaech, S., Parmar, H., Roelandse, M., Bornmann, C. & Matus, A. (2001) Proc. Natl. Acad. Sci. USA 98, 7086–7092. 22. Desmond, N.L. & Levy, W.B. (1983) Brain Res. 265, 21–30. 23. Desmond, N.L. & Levy, W.B. (1990) Synapse 5, 139–143. 24. Palacios-Pru, E.L., Palacios, L. & Mendoza, R.V. (1981) J. Submicros. Cytol. 13, 145–167. 25. Palacios-Pru, E.L., Miranda-Contreras, L., Mendoza, R.V. & Zambrano, E. (1988) Neuroscience 24, 111–118. 26. Greenough, W.T., Klintsova, A.Y., Irwin, S.A., Galvez, R., Bates, K.E. & Weiler, I.J. (2001) Proc. Natl. Acad. Sci. USA 98, 7101–7106. 27. Weiler, I.J., Irwin, S.A., Klintsova, A.Y., Spencer, C.M., Brazelton, A.D., Miyashiro, K., Comery, T.A., Patel, B., Eberwine, J. & Greenough, W.T. (1997) Proc. Natl. Acad. Sci. 94, 5395–5400. 28. Steward, O. & Worley, P.F. (2001) Neuron, in press.


THINK GLOBALLY, TRANSLATE LOCALLY: WHAT MITOTIC SPINDLES AND NEURONAL SYNAPSES HAVE IN COMMON

7069

Colloquium Think globally, translate locally: What mitotic spindles and neuronal synapses have in common Joel D.Richter* Department of Molecular Genetics and Microbiology, University of Massachusetts Medical School, Biotech 4, Room 330, 377 Plantation Street, Worcester, MA 01605 Early metazoan development is programmed by maternal mRNAs inherited by the egg at the time of fertilization. These mRNAs are not translated en masse at any one time or at any one place, but instead their expression is regulated both temporally and spatially. Recent evidence has shown that one maternal mRNA, cyclin B1, is concentrated on mitotic spindles in the early Xenopus embryo, where its translation is controlled by CPEB (cytoplasmic polyadenylation element binding protein), a sequencespecific RNA binding protein. Disruption of the spindle-associated translation of this mRNA results in a morphologically abnormal mitotic apparatus and inhibited cell division. Mammalian neurons, particularly in the synapto-dendritic compartment, also contain localized mRNAs such as that encoding α-CaMKII. Here, synaptic activation drives local translation, an event that is involved in synaptic plasticity and possibly long-term memory storage. Synaptic translation of α-CaMKII mRNA also appears to be controlled by CPEB, which is enriched in the postsynaptic density. Therefore, CPEB-controlled local translation may influence such seemingly disparate processes as the cell cycle and synaptic plasticity. Many cells are remarkably polar. Neurons of the central nervous system, for example, have multiple extensions from the cell body, typically one axon and many dendrites. It stands to reason that this cellular polarity is dictated by the region-specific deposition of proteins and perhaps mRNAs. Vertebrate oocytes, whose radial symmetry would suggest a lack of morphological polarity, are actually characterized by considerable molecular polarity. Consider Xenopus oocytes, which sort many proteins and mRNAs to different locations, particularly along the animal-vegetal axis. This molecular asymmetry is inherited by the fertilized egg and is essential for the establishment of the body plan. Neurons and eggs both contain mRNAs whose translation is regulated both temporally and spatially. Although a number of factors mediate sequence-specific translation in these two cell types, one that has a central role is CPEB, the cytoplasmic polyadenylation element binding protein. In the synapto-dendritic compartment of mammalian hippocampal neurons, CPEB appears to stimulate the translation of α-CaMKII mRNA, which is essential for synaptic plasticity and long-term memory storage. In blastomeres of the developing Xenopus embryo, the control of cyclin B1 mRNA translation on mitotic spindles by CPEB is necessary for the integrity of the mitotic apparatus and for cell division. For both cell types, then, local translational control by CPEB mediates key biological functions. THE BACKGROUND One characteristic of early metazoan development is the mobilization of stored mRNAs into polysomes. In many cases, the stored mRNAs have relatively short poly (A) tails that are elongated at a time that is coincident with translational activation. During oocyte maturation, when oocytes re-enter the meiotic divisions after prolonged prophase I arrest, polyadenylation is stimulated by two cis-acting sequences in the 3 untranslated regions of responding mRNAs. The first is the hexanucleotide AAUAAA, which is also necessary for nuclear pre-mRNA polyadenylation, and the second is the cytoplasmic polyadenylation element (CPE), which has the general structure of UUUUUAU (1). The CPE is bound by the phospho-protein CPEB (2–4), and the hexanucleotide AAUAAA is bound by CPSF (cleavage and polyadenylation specificity factor), a group of factors that also promote nuclear pre-mRNA polyadenylation (5, 6). CPEB and CPSF, plus poly(A) polymerase (7, 8), comprise the core cytoplasmic polyadenylation complex. The identification of the core factors does not explain how cytoplasmic polyadenylation is initiated, nor does it explain the mechanism of translational dormancy or activation. An analysis of the early signaling events of Xenopus oocyte maturation revealed the stimulus for polyadenylation. Progesterone binding to an as-yet-unidentified surface-associated receptor leads to a transient but essential decrease in cAMP. This decrease is soon followed by the activation of Eg2, a member of the Aurora family of protein kinases (9). Active Eg2 phosphorylates CPEB on a single residue (4), which causes it (CPEB) to bind and recruit CPSF into an active cytoplasmic polyadenylation complex (10). By analogy with nuclear pre-mRNA polyadenylation, it is CPSF that recruits poly(A) polymerase to the end of the mRNA. Before oocyte maturation, mRNAs are actively repressed by the CPE, that is, by the same sequence that activates translation by promoting cytoplasmic polyadenylation. The mechanism by which the CPE could be bifunctional was indicated by experiments of de Moor and Richter (11), who demonstrated that efficient CPE-mediated repression requires a 5 cap (i.e., 7mG). This finding suggested that a factor that interacts with the CPE (i.e., CPEB) also could bind the cap (or cap binding proteins), which might limit access of the 5 end of the mRNA to initiation factors (for review of initiation factors, see ref. 12). Although there was no evidence that CPEB interacts with the cap, a CPEB-interacting protein was found to contain a peptide sequence that mediates its interaction with eIF4E, the cap binding factor (13). This factor, termed maskin, interacts with eIF4E in such a way as to preclude an association of eIF4E with eIF4G, thereby preventing the 40s ribosomal subunit from being correctly positioned on the 5 end of the mRNA. Because at least some eIF4E dissociates from maskin during oocyte maturation (and is coincident with polyadenylation), newly “liberated” eIF4E then is free to bind eIF4G and initiate translation. Consequently, maskin appears to belong to a class of proteins known as eIF4EBPs (14), which modulate cap-dependent trans

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: CPE, cytoplasmic polyadenylation element; CPEB, CPE binding protein; CPSF, cleavage and polyadenylation specificity factor. *E-mail: [email protected].



7070

lation by transiently interacting with eIF4E. Unlike the case with other known eIF4EBPs, however, maskin-mediated translational control is mRNA-specific because of its interaction with CPEB. LOCAL TRANSLATIONAL CONTROL AT THE MITOTIC APPARATUS At a late stage of oocyte maturation, after the activation of M-phase promoting factor (a heterodimer of cdc2 and cyclin B), 90% of the CPEB is destroyed (3). That which remains stable is highly localized to the cortex of the animal pole, which in the embryo will give rise to the ectoderm. After fertilization, CPEB remains concentrated in animal pole blastomeres. Within these cells CPEB, as well as maskin, is localized to the mitotic apparatus (15). At metaphase, these proteins are found along the length of the spindles, although there is a greater concentration of them toward the centrosomes. At prophase and prometaphase, the proteins are concentrated on centrosomes. Although the CPEB-activating kinase Eg2 also is found specifically on centrosomes, other proteins involved in polyadenylation-induced translation [poly(A) polymerase, CPSF, eIF4E], which although not particularly concentrated on the mitotic apparatus, are still coincident with it. These results, plus the observation that cyclin B1 mRNA is colocalized with CPEB on spindles, suggest that local polyadenylation-induced translation could take place on or near the mitotic apparatus (15). CPEB amino acid residues 168–211, which contain a PEST protein-protein interaction domain, mediate the interaction of this protein with microtubules in vitro and with centrosomes in vivo (15). When injected into embryos, a CPEB protein lacking these residues has little effect on the synthesis and oscillation of cyclin B1 protein during the cell cycle. However, this deletion mutant CPEB protein induces the “delocalization” of cyclin B1 mRNA and protein from mitotic spindles. The result of this delocalization is inhibited cell division and a malformation of the mitotic apparatus, which includes tripolar spindles, spindles detached from centrosomes, and multiple centrosomes. These data indicate that not only is regulated cyclin mRNA translation important for cell division in embryos, but that the critical translational event occurs in association with mitotic spindles. This finding implies that an important cell division-promoting activity of cyclin B1 protein must be directed to spindles. It is worth noting that cyclin protein is also present on the spindles of Drosophila embryos (16) and HeLa cells (17), where it also may have an essential function. LOCAL TRANSLATIONAL CONTROL AT SYNAPSES In the central nervous system, a single neuron may receive input signals from thousands of different cells. A dendrite that receives a signal from a given axon establishes a “tag” at the point of reception (i.e., the synapse), which distinguishes this stimulated synapse from the many others that are not stimulated (18). This tag establishes a history or memory of the stimulated synapse. Thus, synapses are considered to be “plastic” because their response to activation is influenced by their stimulation history. Two forms of synaptic plasticity, the longlasting phase of long-term potentiation and long-term depression, require new protein synthesis but not new mRNA synthesis (refs. 19–21; see also ref. 22). These observations, as well as others demonstrating that many of the components of the protein synthesis machinery, including mRNAs, are present in dendrites, suggest that local translational control by synaptic activation could underlie, at least partially, synaptic plasticity (23, 24). In mammals, CPEB was first thought to be relatively restricted to germ cells (25). However, subsequent studies showed it to also be present in the hippocampus, the portion of the brain that is responsible for long-term memory. Further analysis demonstrated that CPEB resides in the dendritic layer of the hippocampus, at synapses in cultured hippocampal neurons, and in the postsynaptic density of biochemically fractionated synapses (26). The presence of CPEB at synapses suggested a mechanism of translational control that could influence synaptic strength. It therefore became important to identify the synapto-dendritic mRNA(s) whose translation might be regulated by CPEB. The gene encoding α-CaMKII is necessary for long-term potentiation (27), α-CaMKII mRNA is present in dendrites (28), and αCaMKII protein levels increase upon synaptic stimulation (29, 30). These observations, plus the further revelation that the 3 untranslated region of α-CaMKII mRNA contains a CPE (26), suggested that this molecule could be a substrate for CPEB activity and undergo polyadenylation-induced translation. Because CPEB is present in the visual cortex as well as in the hippocampus, the effect of synaptic activity on CPEB-mediated translation could be tested by using dark-reared rats. In this paradigm, light exposure elicits massive synaptic activation in the visual cortex of rats raised in the dark. In such animals, light stimulation induced α-CaMKII mRNA polyadenylation and translational activation (26). Thus, CPEB may control local translation of this (and possibly other) mRNAs in the postsynaptic region and, by extension, synaptic plasticity. EXTANT QUESTIONS Although there are clear biological consequences of local CPEB-mediated translational control, many particulars remain obscure. For example, why must cyclin mRNA apparently be translated on spindles to effect cell division? If cyclin mRNA polyadenylation-induced translation is under cycle control, as suggested by the data of Groisman et al. (15), then what are the essential upstream signaling events? Is Eg2-mediated CPEB phosphorylation under cell cycle control, or is cytoplasmic polyadenylation, like nuclear polyadenylation, controlled at the level of poly(A) polymerase phosphorylation (31)? In the brain, many questions remain to be explored, such as whether CPEB is activated by Eg2-catalyzed phosphorylation, and most importantly, whether a CPEB knockout mouse would have impaired synaptic plasticity. Finally, the data of Groisman et al. (15) indicate that not only does CPEB regulate translation on spindles, but that it is also involved in localizing mRNA to the mitotic apparatus. Because several CPE-containing mRNAs are localized in dendrites (28), CPEB might influence this process in neurons as well. 1. Richter, J.D. (2000) in Translational Control, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 785–806. 2. Paris, J., Swenson, K., Piwnica-Worms, H. & Richter, J.D. (1991) Genes Dev. 6, 1697–1708. 3. Hake, L.E. & Richter, J.D. (1994) Cell 79, 617–627. 4. Mendez, R., Hake, L.E., Andresson, T., Littlepage, L.E., Ruderman, J.V. & Richter, J.D. (2000) Nature (London) 404, 302–307. 5. Bilger, A., Fox, C.A., Wahle, E. & Wickens, M. (1994) Genes Dev. 8, 1106–1116. 6. Dickson, K.S., Bilger, A., Ballantyne, S. & Wickens, M.P. (1999) Mol. Cell Biol. 19, 5707–5717. 7. Ballantyne, S., Bilger, A., Astrom, J., Virtanen, A. & Wickens, M. (1995) RNA 1, 64–78. 8. Gebauer, F. & Richter, J.D. (1995) Mol. Cell. Biol. 15, 3460–3468. 9. Andresson, T. & Ruderman, J.V. (1998) EMBO J. 17, 5627–5637. 10. Mendez, R., Murthy, K.G.K., Manley, J.L. & Richter, J.D. (2000) Mol. Cell 6, 1253–1259. 11. de Moor, C.H. & Richter, J.D. (1999) EMBO J. 18, 2294–2303. 12. Gingras, A.C., Raught, B. & Sonenberg, N. (1999) Anna. Rev. Biochem. 68, 913–963. 13. Stebbins-Boaz, B., Cao, Q., de Moor, C.H., Mendez, R. & Richter, J.D. (1999) Mol. Cell 4, 1017–1027. 14. Raught, B., Gingras, A.C. & Sonenberg, N. (2000) in Translational Control, eds. Sonenberg, N., Hershey, J.W.B. & Mathews, M.B. (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 245–294. 15. Groisman, I., Huang, Y.S., Mendez, R., Cao, Q., Theurkauf, W. & Richter, J.D. (2000) Cell 103, 435–447.



7071

16. Huang, J. & Raff, J.W. (1999) EMBO J. 18, 2184–2195. 17. Hagting, A., Karlsson, C., Clute, P., Jackman, M. & Pines, J. (1998) EMBO J. 17, 4127–4138. 18. Frey, U. & Morris, R.G. (1998) Trends Neurosci. 21, 181–188. 19. Kang, H. & Schuman, E.M. (1996) Science 273, 1402–1406. 20. Martin, K.C., Casadio, A., Zhu, H., Rose, J.C., Chen, M., Bailey, C.H. & Kandel, E.R. (1997) Cell 91, 927–938. 21. Huber, K.M., Kayser, M.S. & Bear, M.F. (2000) Science 288, 1254–1257. 22. Bear, M.F. & Malenka, R.C. (1994) Curr. Opin. Neurobiol. 4 389–399. 23. Bailey, C.H., Bartsch, D. & Kandel, E.R. (1996) Proc. Natl. Acad. Sci. USA 93, 13445–13452. 24. Schuman, E.M. (1997) Neuron 18, 339–342. 25. Gebauer, F. & Richter, J.D. (1996) Proc. Natl. Acad. Sci. USA 93, 14602–14607. 26. Wu, L., Wells, D., Tay, J., Mendis, D., Abbott, M.A., Barnitt, A., Quinlan, E., Heynen, A., Fallen, J.R. & Richter, J.D. (1998) Neuron 21, 1129–1139. 27. Silva, A.J., Stevens, C.F., Tonegawa, S. & Wang, Y. (1992) Science 257, 201–206. 28. Crino, P.B. & Eberwine, J. (1996) Neuron 17, 1173–1187. 29. Ouyang, Y., Rosenstein, A., Kreiman, G., Schuman, E.M. & Kennedy, M.B. (1999) J. Neurosci. 19, 7823–7833. 30. Scheetz, A.J., Nairn, A.C. & Constantine-Paton, M. (2000) Nat. Neurosci. 3, 211–216. 31. Colgan, D.F., Murthy, K.G., Prives, C. & Manley, J.L. (1996) Nature (London) 384, 282–285.


VASOPRESSIN MRNA LOCALIZATION IN NERVE CELLS: CHARACTERIZATION OF CIS-ACTING ELEMENTS AND TRANSACTING FACTORS

7072

Colloquium Vasopressin mRNA localization in nerve cells: Characterization of cis-acting elements and trans-acting factors Evita Mohr*†, Nilima Prakash‡, Kerstin Vieluf*, Carola Fuhrmann*, Friedrich Buck*, and Dietmar Richter* Hamburg, Institut für Zellbiochemie und klinische Neurobiologie, Martinistrasse 52, 20246 Hamburg, Germany; and ‡Department of Molecular Genetics, The Weizmann Institute of Science, 76100 Rehovot, Israel mRNA localization is a complex pathway. Besides mRNA sorting per se, this process includes aspects of regulated translation. It requires protein factors that interact with defined sequences (or sequence motifs) of the transcript, and the protein/RNA complexes are finally guided along the cytoskeleton to their ultimate destinations. The mRNA encoding the vasopressin (VP) precursor protein is localized to the nerve cell processes in vivo and in primary cultured nerve cells. Sorting of VP transcripts to dendrites is mediated by the last 395 nucleotides of the mRNA, the dendritic localizer sequence, and it depends on intact microtubules. In vitro interaction studies with cytosolic extracts demonstrated specific binding of a protein, enriched in nerve cell tissues, to the radiolabeled dendritic localizer sequence probe. Biochemical purification revealed that this protein is the multifunctional poly(A)-binding protein (PABP). It is well known for its ability to bind with high affinity to poly(A) tails of mRNAs, prerequisite for mRNA stabilization and stimulation of translational initiation, respectively. With lower affinities, PABP can also associate with non-poly(A) sequences. The physiological consequences of these PABP/RNA interactions are far from clear but may include functions such as translational silencing. Presumably, the translational state of mRNAs subject to dendritic sorting is influenced by external stimuli. PABP thus could be a component required to regulate local synthesis of the VP precursor and possibly of other proteins. Neurons of the central and peripheral nervous system have the capacity to deliver distinct mRNA species to locations outside their cell bodies. In the majority of cases, defined transcripts are sorted to the dendrites. The functional significance of this transport process appears obvious: dendrites are principally able to synthesize proteins on site because they are equipped with ribosomes and many of the components required for translation (1–4). Indeed, local translation in dendrites may play an important role in establishing at least certain forms of synaptic plasticity (5). By dendritic mRNA targeting, synthesis of components detrimental for nerve cell functions may be modulated in a spatial and possibly in a temporal manner. In mammalian nerve cells, unequivocal evidence for local translation in the axonal compartment, with the exception of the unmyelinated initial segment, has not been obtained (3, 6). Yet specific sorting of distinct mRNA species in vivo to axons of nerve cells, such as primary sensory neurons projecting to the olfactory bulb and hypothalamic magnocellular neurons, has been described (reviewed in ref. 3). Their functional role remains elusive. In rare cases, the same transcript species is targeted to axons and dendrites. An example to be discussed here is the mRNA encoding the vasopressin (VP) precursor protein. Work on RNA sorting in nerve cells has remained descriptive for many years. In the past, the “neuronal RNA localization community” had to recognize the pioneering role of other scientific disciplines that defined the molecular entities of the RNA localization machinery in non-neuronal cells, particularly in developing systems such as Drosophila oocytes and early embryos, but also in terminally differentiated cell types in species ranging from yeast to human. From these studies, it became clear that subcellular mRNA transport is highly complex and includes several components: (i) sequences within the RNA molecule, referred to as cis-acting elements, that may adopt secondary, tertiary, or even quaternary structures; (ii) a whole array of proteins, trans-acting factors, which mediate RNA sorting by binding either directly or indirectly to the mRNA to be transported; (iii) mechanisms of translational silencing and derepression; and (iv) components of the cytoskeleton as railway tracks and anchor sites of the ribonucleoproteins (reviewed in refs. 7–11). Conceivably, mRNA sorting in nerve cells should operate similarly. Indeed, cis-acting signals mediating dendritic transport of the mRNAs encoding the microtubule-associated protein 2 (MAP2; ref. 12), the VP precursor (13), and the a-subunit of Ca2+/calmodulin-dependent protein kinase II (14) and of the noncoding brain cytosolic 1 RNA (15) have been deciphered. In addition, trans-acting factors have been characterized that interact in vitro with the dendritic localizer elements of MAP2- (16, 17) and VP (18) mRNAs, even though their functional role in mRNA trafficking remains to be elucidated. Here, current knowledge of the molecular determinants of VP mRNA sorting in vivo and in primary cultured sympathetic nerve cells microinjected with eukaryotic expression vector constructs will be summarized. *Universität

MATERIALS AND METHODS Preparation of Protein Extracts. All steps were performed at 4°C or on ice. Cytosolic extracts were prepared by homogenizing (Dounce homogenizer) 30 g of rat brain tissue in 150 ml of homogenization buffer A containing 1 mM K-acetate, 1.5 mM Mg-acetate, 2 mM DTT, 10 mM Hepes, pH 7.8, and protease inhibitor (Complete, Boehringer Mannheim). The homogenate was centrifuged for 10 min at 16,500×g, and the supernatant fraction was saved. The sediment was resuspended in 100 ml of buffer A and centrifuged as above. The supernatant fractions were combined and centrifuged above a cushion of 30% sucrose in buffer A for 3.5 h at 90,000×g. The supernatant fraction

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: DLS, dendritic localizer sequence; MAP2, microtubule-associated protein 2; OT, oxytocin; PABP, poly(A)-binding protein; RRM, RNA recognition motif; SCG, superior cervical ganglion; SSTR, somatostatin receptor; VP, vasopressin; VP-RBP, VP mRNA-binding protein. Data deposition: The sequence reported in this paper has been deposited in the GenBank database (accession no. AF298278). †To whom reprint requests should be addressed. E-mail: [email protected].



7073

(S-90) was carefully removed. Solid ammonium sulfate was added to a final concentration of 45% saturation at 0°C. If necessary, the pH was adjusted to 7.8 with 1 M Tris-base. Precipitated proteins were sedimented for 30 min at 3,500×g. The supernatant fractions were discarded. Proteins were dissolved in 18 ml of 10 mM Hepes, pH 7.8/10 mM NaCl/2 mM DTT/1 mM EDTA/2% glycerol (vol/vol)/0.5 mM PMSF and desalted by using the same buffer and a 5-ml HiTrap desalting column (Amersham Pharmacia Biotech), according to the manufacturer’s instructions. Typically, the desalted eluent had a concentration of 7 mg/ml of protein, as determined with the Protein Assay Reagent (Bio-Rad) and BSA as a standard. Twenty-five milligrams of protein was subjected to heparin column chromatography (5 ml of HiTrap heparin column, Amersham Pharmacia Biotech). Proteins were eluted with 10.5 ml each (7 fractions, 1.5 ml each) of 10 mM Hepes, pH 7.8/2 mM DTT/1 mM EDTA/2% (vol/vol) glycerol/0.5 mM PMSF containing 0.1, 0.2, 0.3, 0.4, and 0.5 M NaCl, respectively. Protein fractions were snap-frozen in liquid nitrogen and stored at –80°C. Each fraction (1.5 µl) was tested for VP mRNA-binding activity by performing U V-crosslinking assays, as described (18). Fractions containing binding activity were further purified by affinity chromatography. Table 1. Biochemical purification and peptide sequence analyses reveal that VP-RBP is the rat PABP Protein purification Peptide sequence Percent identity to mouse PABP1 Heparin column: 0.2 M NaCl P-I: EFSPFGTITSAK 100%, amino acid, 313–324 Affinity purification: 0.1% SDS Heparin column: 0.5 M NaCl P-II: GYGFVHFETQEAAER 100%, amino acid, 139–153 Affinity purification: 1 M NaCl P-III: NFGEDMDDERL 100%, amino acid, 197–207 Heparin column: 0.5 M NaCl Affinity purification: 0.1% SDS Affinity Chromatography. Preparation of biotinylated VP mRNA. Full-size VP mRNA was prepared by using 5 µg of linearized template DNA and the RiboMAX T7-system (Promega) in a 100-µl assay. Biotin-16-UTP (Boehringer Mannheim) was included at a final concentration of 0.3 mM. In vitro transcripts were purified by using the RNeasy midi kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. Coupling of biotinylated RNA to streptavidin-coated paramagnetic particles. Streptavidin-coated paramagnetic particles [0.6 ml (1 mg/ml); Promega] was washed three times with 1 ml each of PBS (10 mM Na-phosphate, pH 7.4/150 mM NaCl). Three hundred picomol of biotinylated VP RNA in 1 ml of PBS was incubated with the particles for 30 min at room temperature on a rotating wheel. A coupling efficiency of 20–30% was routinely achieved. After removal of the RNA solution, the beads were washed twice with 1 ml of PBS and twice with 1 ml of binding buffer [10 mM Tris-HCl, pH 7.8/2 mM DTT/1.5 mM EDTA/10 mM KCl/4% (vol/vol) glycerol/6.7 µg/µl yeast tRNA] and Complete protease inhibitor. Affinity chromatography. Fractions obtained during heparin column chromatography containing binding activity were desalted as described by using binding buffer. For analytical affinity purification, 1.5–2.0 ml of desalted protein fractions (1 mg of protein) was incubated with biotinylated VP RNA coupled to 0.6 ml of streptavidin-coated paramagnetic particles for 20 min at room temperature after addition of heparin (final concentration 2.5 mg/ml) on a rotating wheel. Protein solution (unbound proteins) was removed and saved for later analysis. The particles were washed (2 min each) once with 1 ml of binding buffer containing heparin (2.5 mg/ml), twice with 1 ml each of binding buffer, and once with 200 µl of binding buffer. Bound protein was eluted with 100 µl of 0.1% SDS for 10 min at room temperature and stored at—20°C. In some cases, proteins were eluted first with the same volume of 1 M NaCl/10 mM Tris-HCl, pH 7.8/1.5 mM EDTA/2 mM DTT and Complete protease inhibitor for 30 min at room temperature, concentrated 10-fold by using Vivaspin columns (Sartorius), snap-frozen in liquid nitrogen, and stored at –80°C. For preparative purposes, affinity purification assays were scaled up 10fold. Peptide Sequencing. Proteins purified by affinity chromatography were separated by SDS/PAGE and stained for several hours with colloidal Coomassie blue (Roti-Blue, Roth, Karlsruhe, Germany) according to the protocol recommended by the manufacturer. The protein bands were cut out, washed in water (3×2 h), and treated with acetonitrile for 30 min. The shrunken gel pieces were rehydrated by addition of 1 µg of endoproteinase LysC (Roche Molecular Biochemicals) in 100 µl of digestion buffer (50 mM Tris-HCl, pH 8.5/1 mM EDTA) and incubated overnight at 37°C. The reaction was stopped by adding 1 µl of trifluoroacetic acid (TFA), and the supernatant fraction was collected. The gel pieces were sequentially incubated for 1 h with 100 µl each of reaction buffer, TFA/acetonitrile (50:50, vol/vol), and acetonitrile. All solutions were combined, and the proteolytic fragments were separated by narrowbore HPLC (130A, Applied Biosystems) on a C4 reverse-phase column (Vydac C4, 300 A pore size, 5 mm particle size, 2.1×250 mm). Peptides were eluted with a linear gradient (0–100% B in 50 min; solvent A: water/0.1% TFA, solvent B:70% acetonitrile/0.09% TFA) at a flow rate of 200 ml/min. Peptide-containing fractions detected at 210 nm were collected into siliconized tubes and frozen immediately. Peptide sequences (Table 1) were determined by standard Edman degradation on an automatic sequencer (476A, Applied Biosystems). Cloning of Rat Poly(A)-Binding Protein (PABP). A fragment of the rat PABP cDNA was amplified by the PCR after reverse transcription of rat brain RNA. Two fully degenerate primers were designed: forward primer, 5-TT[TC]GT[GATC]CA[TC]TT-[TC]GA [GA]AC[GATC]CA[GA]GA[GA]GC-3, deduced from the amino acid sequence FVHFETQEA of peptide P-II, and reverse primer, 5[GAT]AT[GATC]GT[GATC]-CC[GA]AA[GATC]GG[GATC]GA[GA]AA[TC]TC-3, deduced from the amino acid sequence EFSPFGTI of peptide P-I. Table 2. Comparison of axonal and dendritic VP mRNAs Axonal VP transcripts • have shorter poly(A) tails than transcripts located in the perikarya • are transported to axons after translation • are located in varicosities devoid of peptide hormones • are not associated with ribosomes and are therefore not translated. Dendritic VP transcripts • are identical in size to transcripts located in the perikarya • are transported to dendrites before translation • are located in parts of dendrites that contain ribosomes and small cisterns of rough endoplasmic reticulum • are most likely translated on-site.



7074

The amplification product was cloned into the SmaI site of pBluescript SK(+) (Stratagene). The sequence was determined by fluorescencebased dideoxy sequencing, by using the PRISM377 DNA Sequencer and PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems), according to the manufacturer’s protocol. A 453-bp insert with a very high degree of identity to the mouse PABP cDNA sequence [European Molecular Biology Laboratory (EMBL)/ GenBank accession no. X65553; nucleotides 758–1210] was isolated. A rat brain λ gt10 cDNA library was screened with the 32P-labeled rat PABP cDNA fragment by standard procedures (20). cDNA inserts from positive phage clones were excised with EcoRI and cloned into the EcoRI site of pBluescript SK (+). These constructs were analyzed by DNA sequencing. One clone consisting of 2,190 bp was obtained that contained the entire coding region for the rat PABP (EMBL/GenBank accession no. AJ298278). Analysis of cDNA was performed with the software package LASERGENE DNASTAR (Madison, WI). Nucleotide and amino acid sequences were compared with sequences in the EMBL and Swiss-Prot data libraries by using the program BLAST from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/Sitemap/index.html#BLAST).

Fig. 1. VP mRNA is sorted to the dendrites of primary cultured SCG neurons. (A) Schematic view of eukaryotic expression vectors microinjected into the cell nuclei of in vitro cultured SCG neurons. The expression of any inserted cDNA is driven by the cytomegalovirus (CMV) promoter. A short sequence of the bacterial ß-galactosidase (ß-gal) gene was included such that it forms part of the 5 untranslated region of the resulting transcripts. This sequence permits discrimination of mRNAs that are endogenously expressed in SCG neurons from those derived by transcription of the expression vector by performing in situ hybridizations with ß-gal anti-sense oligodeoxyribonucleotides. Addition of a poly(A) tail is mediated by the bovine GH (BGH) gene poly(A) addition sequence. (B-D) In situ hybridization analyses of cells microinjected with three different expression vector constructs. (B) Injection of a construct containing the rat VP cDNA inserted in sense orientation leads to transport of the mRNA into proximal and distal parts of the dendrites (arrows). Labeling of the axon has not been observed. (C) A cell is shown that expresses VP anti-sense transcripts. In this case, sorting out of the cell somata does not occur (arrowheads). (D) The vectorderived mRNA encoding α-tubulin is confined to the cell somata (arrowheads). Microinjected cells have been processed for in situ hybridization 18 h after injections. For experimental details, see ref. 13. Other Experimental Procedures. All other experimental procedures, such as preparation of primary cultured neurons, nuclear microinjections of eukaryotic expression vector constructs, preparation of riboprobes, in situ hybridizations, and UV-crosslinking assays, have been described in detail (13, 18). RESULTS AND DISCUSSION In Vivo, VP mRNA Is Sorted to Axons and Dendrites. The genes encoding the VP and structurally closely related oxytocin (OT) precursors are expressed in different populations of hypothalamic magnocellular neurons. The cells are peculiar because the peptide hormones are not only secreted from the nerve terminals in the posterior pituitary into the systemic circulation. Substantial amounts of VP and OT are also released from the dendrites into the brain. Thus, VP and OT have dual functions: first, they act as peptide hormones on diverse peripheral organs, and second, they have a role as neurotransmitters and/or neuromodulators in the central nervous system (21). Both VP and OT mRNAs are sorted to the axonal domain and to dendrites. Unlike other nerve cells, do magnocellular neurons possibly lack specific mRNA sorting mechanisms, for instance as a consequence of their secretory activity? Because the axonal and dendritic transcripts exhibit different characteristics (Table 2), this does not appear to be the case (3). (i) Axonal VP and OT transcripts have snorter poly(A) tracts than their counterparts in the cell bodies. (ii) Although the poly(A) tail of both mRNA species located in the cell somata significantly increases in length in response to osmotic challenge, this is not the case for the RNAs residing in the axon. (iii) In situ hybridizations combined with immunocytochemical analyses revealed that the peptide hormones, and their mRNAs are not colocalized within the axon. Hence, mRNA targeting to the axon is unlikely to result from sticking unspecifically to the neurosecretory granules, (iv) Transcripts are transported to the axon subsequent to translation, (v) There is no evidence for local translation of both mRNAs within the axonal compartment because they are not associated with ribosomes. The dendritic VP and OT mRNAs exhibit distinct characteristics: (i) A variant poly(A) tail length was not evident. The length of VP and OT mRNAs was of the same size in dendrites and cell somata. (ii) The mRNAs are sorted to dendrites before translation. (iii) Immunohistochemical studies at the ultrastructural level have confirmed synthesis of the VP and OT precursors in dendrites, in small cisterns of rough endoplasmic reticulum (ER). This observation is in line with the detection of VP mRNA by electron microscopic in situ hybridization in dendritic



7075

segments containing rough ER (for review, see ref. 3). A currently unresolved problem is the apparent lack of Golgi-like structures in dendrites of nerve cells. Even though Golgi marker proteins of the cis-, intermediate-, and trans-Golgi compartments are detectable immunocytochemically, these molecules are usually located in only one of the major dendrites, and they are frequently restricted to parts proximal to the cell body (22).

Fig. 2. Localization of VPRNA to the dendrites of microinjected SCG neurons is mediated by microtubules. Primary cultured SCG neurons have been injected with a eukaryotic expression vector containing the complete rat VP cDNA. The subcellular distribution of VP transcripts was detected by in situ hybridization by using a digoxigenin-labeled in vitro synthesized VP antisense RNA probe (for details, see ref. 13). Because the microinjection represents a stressful condition, the experiments were designed to allow for a period of recovery (4 h) before drugs were added. (A) Control 1 shows a nontreated cell fixed 4 h after injection. The VP RNA is largely confined to the cell body. Minor amounts are detectable in basal parts of the dendrites close to the cell body (arrow). (B) Control 2 shows a nontreated cell fixed 18.5 h after injection. The VP RNA has been transported to distal parts of the dendrites (arrows). (C) This neuron has been subjected to cokhicine treatment (0.2 µg/ml) for 18.5 h. The drug was added after a recovery time of 4 h after injection. VP RNA remains largely located in the cell body and in basal and very proximal parts of the dendrites (arrows). (D) Example of a neuron subjected to cytochalasin D treatment (0.15 µg/ml) for 18.5 h. The drug was added after a recovery time of 4 h after injection. Cytochalasin D does not inhibit transport of VP RNA to distal dendritic segments (arrows). CHARACTERIZATION OF DENDRITIC LOCALIZER SEQUENCES WITHIN VP MRNA. To define cis-acting sequences mediating VP mRNA sorting to nerve cell processes, eukaryotic expression vector constructs (Fig. 1 A) have been designed and introduced by nuclear microinjections into primary cultured neurons isolated from embryonic rat superior cervical ganglia (SCG) (13). When injections were done with a construct containing the cDNA in sense orientation, VP transcripts were detectable in the cell somata as well as in dendrites (Fig. 1 B). In contrast, VP antisense RNA remained confined to the cell bodies (Fig. 1 C), indicating that either a sequence motif or a secondary structure of the VP mRNA harbors information essential for mRNA transport. Vector-expressed α-tubulin transcripts (Fig. 1 D), like endogenous mRNA, are also entirely located in the perikarya (13). Further analyses revealed that the last 395 nucleotides, termed dendritic localizer sequence (DLS), containing part of the coding region and the complete 3-untranslated region (UTR), were able to confer dendritic targeting to a normally nonlocalized reporter transcript comparable to that achieved by the VP mRNA alone. Partial DLS segments spanning either its proximal or its distal half or the 3-UTR alone each confer only a moderate degree of dendritic localization to very proximal parts of the dendrites. Hence, the DLS contains several weak localizer elements, and these have to act in concert to mediate an efficient transport of the VP mRNA to the dendritic domain (3, 13).

Fig. 3. VP-RBP binds specifically to the DLS of the VP mRNA and is enriched in brain tissues. (A) Autoradiogram of UVcrosslinking analyses performed with 7.5 µg of rat brain cytosolic protein extract and 5 fmol of the radiolabeled DLS riboprobe [lacking a poly(A) tail]. Unlabeled competitor RNAs were added at a 100-fold molar excess. Lane 1, no competitor; lane 2, fullsize VP RNA (GenBank accession no. M25646); lane 3, dendritic targeting element of rat MAP2 mRNA (GenBank accession no. X51842, nucleotide residues 5383–5552); lane 4, full-size rat α-tubulin RNA (GenBank accession no. V01227). The positions of molecular size marker proteins is indicated on the Right. The arrow denotes the 85-kDa VP-RBP/RNA complex. All competitor RNAs represent the sense strands; none of the transcripts possess poly(A) tails, (B) Autoradiogram of UV-crosslinking analyses performed with 7.5 µg each of various rat tissue/cell line cytosolic protein extracts and 5 fmol of the radiolabeled DLS riboprobe [lacking a poly(A) tail]. Proteins were prepared from: lane 1, total brain; lane 2, hypothalamus; lane 3, heart; lane 4, lung; lane 5, spleen; lane 6, liver; lane 7, Rat I cells; lane 8, pheochromocytoma 12 cells. The positions of molecular size marker proteins are indicated on the Right. The arrow denotes the 85-kDa VP-RBP/RNA complex. In vivo but not in SCG neurons, VP mRNA is sorted to dendrites as well as to axons (13). Thus, SCG neurons apparently lack the machinery essential for sorting of distinct mRNA species to the axonal domain. Alternatively, low levels of mRNA might prevent their detection in axons. Dendritic Transport of VP mRNA Is Mediated by Microtubules. With rare exceptions, the cytoskeleton is indispensable when mRNAs are localized to distinct subcellular destinations. Microfilaments or microtubules are needed for mRNA sorting in non-neuronal cells in yeast, Xenopus oocytes, Drosophila oocytes and early embryos, and in mammalian cells (23, 24). Present knowledge about the type of cytoskeletal element as part of the mRNA transport machinery in nerve cells is still preliminary. Yet its



7076

identification is important because it provides valuable information concerning the motor proteins that link the ribonucleoprotein unit to microfilaments or microtubules. Earlier studies (25) suggest the involvement of microtubules in sorting newly synthesized poly(A)-RNA to dendrites. Transport of a defined newly synthesized mRNA species in the presence of cytoskeleton-disrupting drugs has not been investigated. We have analyzed dendritic targeting of VP mRNA in primary cultured SCG neurons subjected to depolymerization of either microtubules or microfilaments by colchicine and cytochalasin D treatment, respectively. As shown in Fig. 2, VP mRNA is directed toward the dendritic compartment along microtubules. Because microtubules in dendrites exhibit a mixed polarity (26), plus-end-and/or minusend-directed motor proteins such as kinesins and dynein are likely partners of the translocation complex. Microfilaments do not appear to be required for the transport process per se. However, the data do not rule out the possibility that the actin-based cytoskeleton may be necessary for mRNA anchoring within the neurites. Recently, evolutionarily conserved RNA-binding proteins have been identified that may play a role in transporting ß-actin transcripts in fibroblasts (27) and vegetal pole 1 (Vg1) mRNA in Xenopus oocytes (28, 29), respectively. Although ß-actin mRNA is transported along microfilaments, microtubules are prerequisite for Vg1 mRNA sorting to the vegetal hemisphere. On the basis of their high degree of similarity, it is conceivable that, by the interaction of such proteins (either directly or via additional adapter proteins) with different motor proteins, the ribonucleoprotein cargo could be transferred from microtubules to microfilaments and vice versa.

Fig. 4. Purification of VP-RBP by affinity chromatography. VP-RBP-containing fractions obtained after heparin column chromatography (eluted with 0.2 and 0.5 M NaCl, respectively) were individually subjected to affinity purification by using biotinylated full-size VP RNA [but lacking a poly(A) tail] immobilized on streptavidin-coated paramagnetic particles. Protein fractions obtained during affinity purification were separated by SDS gel electrophoresis and stained with Coomassie blue. (A) Affinity purification with proteins eluted with 0.2 M NaCl from the heparin column. (B) Affinity purification with proteins eluted with 0.5 M NaCl from the heparin column. Lane 1, 5 µg of protein eluted before affinity purification; lane 2, 5 µg of protein after incubation with VP RNA immobilized on streptavidin-coated paramagnetic particles (unbound proteins); lanes 3–5, 20 µl each of the wash fractions; lane 6, protein (20 µl) bound to VP RNA and eluted with 0.1% SDS. The asterisk denotes the major protein with an apparent molecular mass of 78 kDa. The protein marked by the arrowhead probably represents BSA (for details, see Fig. 6 legend). The positions of marker proteins (in kilodaltons) are indicated on the Left.

Fig. 5. The 78-kDa protein purified by affinity chromatography and eluted with 1 M NaCl retains binding activity to radiolabeled VP RNA. (A) Protein was first eluted with 1 M NaCl (left lane). Afterward, protein that remained bound to the VP RNA was eluted with 0.1% SDS (right lane). One-fiftieth each of the eluted protein was applied to the gel. The Coomassie-stained SDS gel is shown. The asterisk denotes the major protein with an apparent molecular mass of 78 kDa. The positions of marker proteins (in kilodaltons) are indicated on the Left, (B) Autoradiogram of UV-crosslinking analyses performed with fractions obtained during affinity purification and 5 fmol of the radiolabeled full-size VP RNA [lacking a poly(A) tail]. Lane 1, protein enriched by the heparin column chromatography before affinity purification; lane 2, protein after incubation with VP RNA immobilized on streptavidin-coated paramagnetic particles (unbound proteins); lanes 3–5, wash fractions; lane 6, protein bound to VP RNA, eluted with 1 M NaCl. The arrow denotes the 85-kDa protein/RNA complex. Characterization of Trans-acting Factors. Proteins (trans-acting factors) play an active role in sorting mRNA molecules to defined cytoplasmic locations. By using UV-crosslinking analyses, we have identified a protein enriched in brain tissue, termed VP-RNA-binding protein (VP-RBP), which interacts in a specific manner with the DLS of the VP mRNA (Fig. 3 A and B) but not with the 5-end of the VP mRNA, which lacks a role in dendritic mRNA localization (18). Moreover, the protein fails to bind to a variety of other transcripts, including a-tubulin mRNA and the dendritic targeting element of the MAP2 transcript (Fig. 3 A). These findings are complemented by recently published data that two trans-acting factors, MARTA1 and MARTA2, interact in vitro with the dendritic targeting element of MAP2 mRNA but not with the rat VP mRNA or with other transcripts known to be sorted to dendrites (16). Hence, the molecular determinants required for sorting of different mRNAs to dendrites appear to be surprisingly specific for a given transcript species. Presumably, this finding reflects the existence of different pathways that govern the correct temporal and spatial distribution of defined mRNAs that have been observed in nerve cells (1, 3). Purification of VP-RBP. Biochemical purification, including precipitation with 45% (NH4)2SO4 from cytosolic brain extracts and subsequent heparin column chromatography, was used to identify the molecular nature of VP-RBP. Binding activity, as revealed by formation of the 85-kDa protein/RNA complex, was detected in fractions eluted with 0.2 M and 0.5 M NaCl, respectively (data not shown). These fractions were separately



7077

subjected to affinity chromatography with biotinylated VP transcripts immobilized on streptavidin-coated paramagnetic particles. Proteins bound to VP RNA were eluted with 0.1% SDS and separated by denaturing PAGE. In both cases, that is when the 0.2 M and 0.5 M heparin column eluents were used as starting material, a major protein with an apparent molecular weight of about 78 kDa was eluted (Fig. 4 A and B, lane 6*). UV-crosslinking analyses performed with individual fractions obtained during affinity purifications revealed almost complete binding of the 78-kDa protein to the immobilized VP RNA (data not shown).

Fig. 6. The 78-kDa protein binds specifically to VP RNA. (A) Coomassiestained SDS gel of affinity chromatography assays performed with proteins obtained during heparin column chromatography (0.2 M NaCl-eluent) and different biotinylated RNAs immobilized on streptavidin-coated paramagnetic particles. Lane 1, proteins before affinity purification; lane 2, proteins eluted from paramagnetic particles coupled with biotinylated rat SSTR3 RNA (GenBank accession no. X63574, partial sequence corresponding to nucleotides 3010–3855); lane 3, proteins eluted from paramagnetic particles coupled with biotinylated rat VP RNA [lacking a poly(A) tail]; lane 4, proteins eluted from paramagnetic particles without RNA coupling; lane 5, coupling of biotinylated rat VP RNA but without addition of protein. The 78-kDa protein is denoted by the asterisk. The protein marked by the arrowhead probably represents BSA. The paramagnetic particles are stored in a buffer containing BSA. Residual amounts of this protein are obviously eluted from the particles with 0.1% SDS. (B) Autoradiogram of UV-crosslinking analyses performed with affinity-purified proteins and radiolabeled VP RNA [lacking a poly(A) tail]. Proteins obtained during heparin column chromatography (0.2 M NaCl-eluent) were affinity-purified with streptavidin-coated paramagnetic particles coupled with SSTR3 RNA (lanes 1 and 2) or full-size VP RNA (lanes 3 and 4). Lanes 1 and 3, protein before affinity purification; lanes 2 and 4, proteins after affinity purification (proteins not binding to the immobilized RNAs). The arrow denotes the 85-kDa VP-RBP/VP mRNA complex. The positions of molecular size marker proteins (in kilodaltons) are indicated on the Left. In repeated experiments, elution of the 78-kDa protein by buffers containing high concentrations of salt turned out to be variable and often was rather ineffective. Most of the protein came off the matrix by incubation with detergent (Fig. 5 A). However, as demonstrated by UV-crosslinking analysis, the salt-eluted protein retained RNA-binding activity forming a complex with radiolabeled VP mRNA identical in size to that seen with protein extracts before affinity purification (Fig. 5 B). Thus, the 78-kDa protein purified from rat cytosolic brain extracts most likely represents VP-RBP that, in UV-crosslinking analyses, gives rise to the protein/RNA complex with an apparent molecular weight of 85 kDa (ref. 18; see also Fig. 3).

Fig. 7. PABP present in cytosolic brain extracts and in partially purified chromatographic fractions exhibits different binding properties to VP mRNA. Autoradiogram of UV-crosslinking competition analyses performed with 5 fmol of radiolabeled full-size VP RNA [lacking a poly(A) tail] and (A) rat brain cytosolic proteins (S-130) or protein partially purified by precipitation with 45% (NH4)2SO4followed by heparin column chromatography and eluted with 0.2 M NaCl (B) or 0.5 M NaCl (C). Unlabeled competitor RNAs were added at 100-fold molar excess, and ribohomopolymers poly(A), poly(U), poly(G), and poly(C) were added at a concentration of 100 ng/assay. The following competitor RNAs were used: VP, complete rat vasopressin RNA; Tubulin, complete rat α-tubulin RNA; SSTRa-c, partial sequences derived from the rat SSTR3 mRNA (SSTR-a: nucleotide residues 451–1650; SSTR-b: nucleotide residues 1660–3010; SSTR-c: nucleotide residues 3010–3855). All competitor RNAs represent the sense strands; none of the transcripts possess poly(A) tails. Further control affinity purifications demonstrate specificity of VP-RBP interaction with VP mRNA (Fig. 6 A, lane 3*). (i) The protein fails to associate with an RNA fragment corresponding to part of the mRNA encoding the G-protein-coupled somatostatin receptor 3 (SSTR3, lane 2). As shown earlier, this RNA does not inhibit formation of the VP-RBP/VP RNA complex in competition UV-crosslinking assays (18). (ii) It shows no unspecific interaction with paramagnetic particles in the absence of RNA-coupling (lane 4). (iii) The protein is no constituent of particles coupled with VP mRNA and processed identically in the absence of protein (lane 5). Subsequent UV-crosslinking analyses (Fig. 6 B) give further confirmation: when affinity purification was done with coupling of SSTR3 RNA, binding of VP-RBP to VP mRNA is readily detectable in the protein pool before and after incubation with SSTR3 RNA-coated particles (lanes 1 and 2), whereas residual binding activity is largely depleted in the unbound protein pool when VP RNA was coupled to the matrix (lanes 3 and 4). VP-RBP Is the Rat PABP. Large-scale affinity purification by using liganded VP RNA and subsequent separation of eluted protein by SDS gel electrophoresis was used to isolate the 78-kDa protein in quantities sufficient for peptide sequencing. The



7078

following preparations were processed for protein identification: (i) Proteins eluted with 0.2 M NaCl from a heparin column, affinity purified and eluted with 0.1% SDS; (ii) proteins eluted with 0.5 M NaCl from a heparin column, affinity purified and eluted with 1 M NaCl; and (iii) proteins eluted with 0.5 M NaCl from a heparin column, affinity purified and eluted with 0.1% SDS. Proteins were digested separately with endoproteinase Lys-C. The resulting peptides were resolved by HPLC. The elution profiles were very similar, indicating that the proteins are either identical or at least highly related (data not shown). One peptide from each chromatographic separation (but exhibiting different retention times) was subjected to sequence analysis, revealing that the 78-kDa protein is the rat PABP. All peptides are 100% identical to parts of mouse PABP1 (Table 1). Cloning and cDNA sequencing showed that the rat PABP consists of 636-aa residues with a high degree of identity when compared with mouse (99.7%; ref. 19) and human (99.5%; ref. 30) PABP. Hence, VP-RBP will be referred to as PABP. We do not know why PABP elutes from the heparin column with 0.2 and 0.5 M NaCl, but several explanations are possible. First, covalent modifications, for example phosphorylated forms, of PABP with reduced affinity for the negatively charged ligand could exist. Second, PABP may be part of larger complexes displaying distinct binding characteristics to heparin. Currently, we cannot discriminate between these possibilities. However, UV-crosslinking competition analyses performed with VP mRNA indeed reveal different binding behaviors of PABP in cytosolic extracts and in partially purified protein pools (Fig. 7). PABP present in rat brain cytosolic extracts exhibits the highest degree of binding specificity. Complex formation is inhibited by a molar excess of unlabeled VP mRNA but not by a-tubulin transcripts and segments spanning various parts of the SSTR3 mRNA. With ribohomopolymer competitors, poly(A) competed as would be expected, whereas poly(U), poly(G), and poly(C) were inefficient (Fig. 7 A). When the same series of experiments was done with proteins eluted with 0.2 M NaCl from a heparin column, additional, albeit minor, competition was observed with one of the SSTR3 RNA segments (SSTR-b probe) and with poly(G) (Fig. 7 B). The lowest degree of specificity with respect to interaction with VP mRNA is observed for PABP present in the protein pool eluted with 0.5 M NaCl from the column (Fig. 1 C). These data clearly demonstrate that binding specificity of PABP to VP mRNA is determined by additional parameters, for instance a covalent modification or by other proteins that could alter PABP’s sequence selection by protein/protein interactions. Apparently, this “specificity factor,” whatever its nature, is brain specific. Evidence stems from UV-crosslinking analyses shown in Fig. 3 B. Even though PABP is known to be extremely abundant (31), most peripheral tissues and non-neuronal cell lines harbor much lower amounts of the protein in a form that is able to interact with VP mRNA compared with brain tissue. Possible Role of PABP in VP mRNA Metabolism. PABP harbors four highly conserved RNA recognition motifs (RRM; 80–100 aa in length) at the N-terminal part of the protein and a more divergent C-terminal auxiliary domain (32). It binds with high affinity to the poly (A) tail of mRNAs, thereby enhancing translation via interaction with initiation factors bound at the 5 end (33). Furthermore, it stabilizes mRNAs in a translation-dependent manner (34). Binding studies with individual RRMs or combinations thereof revealed several interesting features: single RRMs are unable to bind to poly (A). RRMs 1 and 2 have a high affinity for poly(A) identical to that of the full-size protein, whereas RRMs 3 and 4 have a much lower affinity for poly(A). In fact, binding of RRMs 3 and 4 to poly(U)- and poly(G)-sequences is much better than to poly(A) (35, 36). In yeast, PABP is essential for cell viability. Yet, whereas deletion of RRMs 1 and 2 alone still supported growth, removal of RRM 4 joined with C-terminal amino acid residues did not, suggesting critical functions of this sequence (35). Taken together, RRMs 1–4 are functionally diverse, and features other than high-affinity binding to poly(A) sequences are essential for cell viability. PABP is an extremely abundant protein. HeLa cells, for instance, have a roughly 3-fold excess of protein over binding sites on poly(A) mRNAs (31). Because PABP interacts with sequences other than poly(A) in vitro (31, 36), it probably has additional functions in mRNA metabolism. For instance, PABP is able to control translation of its own mRNA, and it does so by specific association with sequences of the 5-UTR (37, 38). Given the heterogeneous roles of PABP, it is conceivable that it may be involved in regulating the translational state of the VP (and possibly of other) mRNA. Accumulating evidence suggests that dendritically localized mRNAs are not translated until external stimuli trigger the activation of protein synthesis (5, 39). Translational silencing by PABP could, for instance, be accomplished by its binding to the DLS of the VP mRNA. A direct or indirect interaction of this (or these) molecule(s) with those that are bound to the poly(A) tract could inhibit translational stimulation, because it interferes with the interaction of poly(A) tail-bound PABP with translational initiation factors at the 5-end of the mRNA. A similar model involving PABP as part of a larger and preexisting protein complex has been proposed as a mechanism that regulates translation-dependent turnover of the c-fos mRNA (40). Several questions to be addressed in future experiments remain open, (i) Which RRMs of PABP participate in its interaction with DLS of the VP mRNA? (ii) What type of molecule (or modification) determines its binding specificity to the DLS? (iii) Does PABP also play a role in the metabolism of other dendritically localized mRNAs, and what exactly is that role? We thank Susanne Franke for expert technical assistance. This work is supported by the Deutsche Forschungsgemeinschaft and the Volkswagenstiftung (to D.R. and E.M.). Part of this work forms the Ph.D. thesis of Carola Fuhrmann. 1. Steward, O. (1997) Neuron 18, 9–12. 2. Kuhl, D. & Skehel, P. (1998) Curr. Opin. Neurobiol. 8, 600–606. 3. Mohr, E. (1999) Prog. Neurobiol. 57, 507–525. 4. Tiedge, H., Bloom, F.E. & Richter, D. (1999) Science 283, 186–187. 5. Schuman, E.M. (1999) Neuron 23, 645–648. 6. Alvarez, J., Giuditta, A. & Koenig, E. (2000) Prog. Neurobiol. 62, 1–62. 7. Bashirullah, A., Cooperstock, R.L. & Lipshitz H.D. (1998) Annu. Rev. Biochem. 67, 335–394. 8. Barbarese, E., Brumwell, C., Kwon, S., Cui, H. & Carson, J.H. (1999) J. Neurocytol. 28, 263–270. 9. Gonzalez, I., Buonomo, S.B.C., Nasmyth, K. & von Ahsen, U. (1999) Curr. Biol. 9, 337–340. 10. King, M.L., Zhou, Y. & Bubunenko, M. (1999) BioEssays 21, 546–557. 11. Lipshitz, H.D. & Smibert, C.A. (2000) Curr. Opin. Genet. Dev. 10, 476–488. 12. Blichenberg, A., Schwanke, B., Rehbein, M., Garner, C.C., Richter, D. & Kindler, S. (1999) J. Neurosci. 19, 8818–8829. 13. Prakash, N., Fehr, S., Mohr, E. & Richter, D. (1997) Eur. J. Neurosci. 9, 523–532. 14. Mori, Y., Imaizumi, K., Katayama, T., Yoneda, T. & Tohyama, M. (2000) Nat. Neurosci. 3, 1079–1084. 15. Muslimov, I.A., Santi, E., Hamel, P., Perini, S., Higgins, D. & Tiedge, H. (1997) J. Neurosci. 17, 4722–4733. 16. Rehbein, M., Kindler, S., Horke, S. & Richter, D. (2000) Mol. Brain Res. 79, 192–201. 17. Monshausen, M., Putz, U., Rehbein, M., Schweizer, M., DesGroseillers, L., Kuhl, D., Richter, D. & Kindler, S. (2001) J. Neurochem. 76, 155–165. 18. Mohr, E., Fuhrmann, C. & Richter, D. (2001) Eur. J. Neurosci., 13, 1107–1112. 19. Wang, M., Cutler, M., Karimpour, I. & Kleene, K.C. (1992) Nucleic Acids Res. 20, 3519.



7079

20. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) in Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press, Plainview, NY). 21. Morris, J.F., Pow, D.V., Sokol, H.W. & Ward, A. (1993) in Vasopressin, eds. Gross, P., Richter D. & Robertson, G.L. (John Libbey Eurotext, Paris), pp. 171–182. 22. Torre, E.R. & Steward, O. (1996) J. Neurosci. 16, 5967–5978. 23. Nasmyth, K. & Jansen, R.-P. (1997) Curr. Opin. Cell Biol. 9, 396–400. 24. Arn, E.A. & Macdonald, P.M. (1998) Cell 95, 151–154. 25. Bassell, G.J., Singer, R.H. & Kosik, K.S. (1994) Neuron 12, 571–582. 26. Baas, P.W., Black, M.M. & Banker, G.A. (1989) J. Cell Biol. 109, 3085–3094. 27. Ross, A.F., Oleynikov, J., Kislauskis, E.H., Taneja, K.L. & Singer, R.H. (1997) Mol. Cell. Biol. 17, 2158–2165. 28. Deshler, J.O., Highett, M.I., Abramson, T. & Schnapp, B.J. (1998) Curr. Biol. 8, 489–496. 29. Havin, L., Git, A., Elisha, Z., Oberman, F., Yaniv, K., Pressman-Schwartz, S., Standart, N. & Yisraeli, J.K. (1998) Genes Dev. 12, 1593–1598. 30. Grange, T., de Sa, C.M., Oddos, J. & Pictet, R. (1987) Nucleic Acids Res. 15, 4771–4787. 31. Görlach, M., Burd, C.G. & Dreyfuss, G. (1994) Exp. Cell Res. 211, 400–407. 32. Burd, C.G. & Dreyfuss, G. (1994) Science 265, 615–621. 33. Preiss, T., Muckenthaler, M. & Hentze, M.W. (1998) RNA 4, 1321–1331. 34. Coller, J.M., Gray, N.K. & Wickens, M.P. (1998) Genes Dev. 12, 3226–3235. 35. Burd, C.G., Matunis, E.L. & Dreyfuss, G. (1991) Mol. Cell Biol. 11, 3419–3424. 36. Kühn, U. & Pieler, T. (1996) J. Mol. Biol. 256, 20–30. 37. de Melo Neto, O.P., Standart, N. & de Sa, C.M. (1995) Nucleic Acids Res. 23, 2198–2205. 38. Bag, J. & Wu, J. (1996) Eur. J. Biochem. 237, 143–152. 39. Marin, P., Nastiuk, K.L., Daniel, N., Girault, J.-A., Czernik, A.J., Glowinski, J., Nairn, A.C. & Prémont, J. (1997) J. Neurosci. 17, 3445–3454. 40. Grosset, C., Cheu, C.A., Xu, N., Sonenberg, N., Jaquemin-Sablou, H. & Shyu, A. (2000) Cell 103, 29–40.


LOCAL TRANSLATION OF CLASSES OF MRNAS THAT ARE TARGETED TO NEURONAL DENDRITES

7080

Colloquium Local translation of classes of mRNAs that are targeted to neuronal dendrites James Eberwine*, Kevin Miyashiro, Janet Estee Kacharmina, and Christy Job Departments of Pharmacology and Psychiatry, University of Pennsylvania Medical Center, 36th and Hamilton Walk, Philadelphia, PA 19104–6084 The functioning of the neuronal dendrite results from a variety of biological processes including mRNA transport to and protein translation in the dendrite. The complexity of the mRNA population in dendrites suggests that specific biological processes are modulated through the regulation of dendritic biology. There are various classes of mRNAs in dendrites whose translation modulates the ability of the dendrite to receive and integrate presynaptic information. Among these mRNAs are those encoding selective transcription factors that function in the neuronal soma and ionotropic glutamate receptors that function on the neuronal membrane. Conclusive evidence that these mRNAs can be translated is reviewed, and identification of the endogenous sites of translation in living dendrites is presented. These data, as well as those described in the other articles resulting from this colloquium, highlight the complexity of dendritic molecular biology and the exquisitely selective and sensitive modulatory role played by the dendrite in facilitating intracellular and intercellular communication. Since the time of Cajal, it has been apparent that neurons have a striking morphological polarity. It was recognized with time that a neuron’s morphological appearance relates to the function of these cells. Neurons typically have a cell soma from which multiple dendrites and a single axon protrude. Presynaptic cells communicate with postsynaptic cells through direct connections between the presynaptic axon and the postsynaptic dendrite. Presynaptic information is processed within the dendrites and transferred to the cell soma where additional signal integration occurs. The axon, in turn, transfers the postsynaptic cells’ integrated response to the next postsyanptic neuron. Information processing in the dendrite is complex, involving both local dendritic and more global soma modulatory components. Dendrites are the primary locus of postsynaptic connectivity and can comprise >90% of the postsynaptic surface of some neurons. In general the presynaptic axon forms synapses on dendritic spines of the postsynaptic membrane. The establishment of such postsynaptic specializations reflects how neurons elaborate discrete plasma membrane domains that differ focally in their regulatory properties. Such focal domains are critical in determining specificity of information flow in the neuronal circuitry of the central nervous system. In response to prolonged periods of synaptic plasticity, including long-term potentiation and long-term depression, subsets of these synapses undergo a series of enduring changes in spine shape and density as well as electrophysiological characteristics. These changes result, in part, from local regulation of the functioning of the postsynaptic density (a protein-mRNA complex present in the dendritic spine). In this paper, we review some of the work from our laboratory detailing aspects of neuronal dendrite functioning. MRNA COMPLEXITY OF INDIVIDUAL NEURONAL DENDRITES It has been clear since the mid-1960s (1) that RNAs are localized in dendrites, with the first RNAs found being ribosomal RNA as visualized by electron microscopy. It took several more years for mRNAs to be conclusively shown to be localized to dendrites. For nearly a decade ribosomal RNA and a handful of mRNAs were the only ones known to exist in dendrites (2). In 1994, Miyashiro and colleagues (3) used mRNA amplification techniques to show that many more mRNAs existed in dendrites including the mRNAs for all of the ionotropic glutamate receptors and various mRNAs encoding proteins involved in modulating the translation of proteins. The repertoire of dendritically localized mRNAs was further expanded through the work of Crino and Eberwine (4) in which expression profiling and differential display showed the presence of more than 30 identified and many additional expressed sequence tag mRNAs in dendritic processes. These studies also established that there is molecular individuality in neuronal dendrites, because different processes can contain different mRNAs. These studies initially were controversial because some of the mRNAs, such as the ionotropic glutamate receptor subunit mRNAs, had been considered absent from dendrites based on the inability to detect the mRNAs by in situ hybridization. It is important to note that the earlier in situ hybridization papers suffered from a lack of technique sensitivity. With improved sensitivity many of these dendritically localized mRNAs, including members of the glutamate receptor family, have been shown to be present in dendrites by in situ hybridization methodologies. Indeed, using antisense RNA (aRNA) amplification of dendritic mRNA followed by differential display and microarray analysis, we estimate that 400 mRNAs can be localized to dendrites of rat hippocampal neurons in primary cell culture (Fig. 1) (3, 5, 6). In Fig. 1A, a cultured rat hippocampal neuron is shown. The processes marked HP 3–6 and HP 3–5 were independently harvested and the endogenous mRNA amplified with the aRNA procedure, and then differential display was performed on the dendritic aRNA. In Fig. 1 B, a comparison between the two processes of the differential display pattern using two different differential display primer sets is shown. The same 5 primer was used for each PCR while the 3 primer was varied (lanes A and C). Bands that comigrated between the different dendrites correspond to mRNAs that are localized to both dendrites. Fig. 1 C shows the display pattern generated when using nine different primer sets on the aRNA made from dendrite HP 3–5. The large number of bands highlights the complexity of the mRNA localized to dendrites. To put this number of 400 dendritically localized mRNAs into context, using the same methodologies we estimate that there are 10,000 different mRNAs expressed in the cell soma of these same

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: aRNA, antisense RNA; DHPG, (RS)-3, 5-dihydroxyphenylglycine; GFP, green fluorescent protein; CREB, cAMP responsive element binding protein. *To whom reprint requests should be addressed. E-mail: [email protected].



7081

hippocampal neurons. Consequently, 5% of the mRNAs complexity expressed in a neuron can be targeted to the dendritic domain.

Fig. 1. Differential display analysis of dendritic processes from a single hippocampal neuron in culture. (A) Isolated hippocampal cells free of overlapping processes from neighboring cells were identified in low-density cultures (5). Individual dendrites were harvested by transecting at the branch point and aspirated as described (3). (B) Comparisons of differential display products between dendrites with a single 5 primer, OPA-5 (Operon Technologies, Alameda, CA), in combination with anchor primers A (T11AC) or C (T11GC) show the presence of common (closed arrowheads) and unique (open arrowheads) PCR products. (C) When a single 5 primer (OPA-13) is used in combination with all nine anchor primers a large population of mRNAs are present within neuronal processes. The mechanism by which mRNAs are targeted to dendrites is thought to involve the binding of RNA-binding proteins to cis-acting elements in the transported mRNAs. There are more than 500 RNA binding proteins estimated to be encoded by the cellular genome. The cis-acting elements are likely composed of a primary nucleic acid sequence and a secondary RNA structure such that a specific RNAbinding protein recognition binding site is generated. The specificity of these cis-acting elements and the fact that only a limited number of RNAs are targeted to dendrites suggest that different RNAs may have different rates of transport into the dendrite. The regulated transport of mRNA into neuronal dendrites recently has become a topic of examination by a number of groups. One approach to determining the rates of mRNA transport into dendrites is to treat neuronal cells with various



7082

pharmacological agents that are known to modulate dendritic functioning followed by characterization of the mRNA populations that are present in the dendrite as a function of time after stimulation. Primary hippocampal neuronal cultures were stimulated for various time periods with a metabotrobic glutatmate receptor agonist DHPG [(RS)-3, 5-dihydroxyphenylglycine] that is also a known modulator of dendritic protein translation (7). At increasing exposure times the dendrites were harvested, the mRNAs were amplified, and the amplified products used as a probe to screen macroarrays and microarrays containing thousands of different cDNAs. Data presented in Fig. 2 show that this approach can be used to monitor the movement of mRNAs into the dendrite. The arrows in Fig. 2 A-D point to a pair of identical cDNA clones (one cDNA spotted in the two positions) that hybridize to the aRNA probe with greater intensity as a function of time after metabotrobic glutamate receptor stimulation. The aRNA probe used in this study was a pooled probe made from multiple dendrites from multiple neurons to ensure that differences in hybridization intensity are not caused by biological differences between individual neuronal dendrites.

Fig. 2. Time course for movement of mRNAs into the neuronal dendrite. Intact cultured rat hippocampal neurons were treated with DHPG. mRNA was harvested from dendrites at times of 0, 24, 32, and 45 min posttreatment, and aRNA was amplified. This probe was used to screen various types of arrays including the macroarrays shown. The hybridization intensity of selected spots corresponding to different immobilized cDNAs increased as a function of time after DHPG treatment. These data indicate that the movement of mRNA into the dendrite can be regulated by DHPG. FORMAL MOLECULAR PROOF OF PROTEIN TRANSLATION IN DENDRITES AND IDENTIFICATION OF IN VIVO TRANSLATION SITES The presence of ribosomes and mRNAs within dendrites suggests that the dendrites are translationally competent. Indeed various groups have shown by using immunohistochemistry that various protein components of the translational machinery are present within dendrites (8, 9). Additional data suggesting that translation can occur in dendrites was provided by monitoring radioactive amino acid incorporation into individual dendrites (10) (not an actual proof of translation) and through the use of synaptoneurosome preparations to show that radioactivity could be incorporated into a product whose synthesis was diminished by protein synthesis inhibitors (7, 11). Conclusive evidence that protein synthesis can occur in dendrites was provided by transfecting isolated dendrites with mRNA constructs that encode a protein fused to an epitope tag (c-myc) (4). This mRNA was lipid-encoated and applied to transected live dendrites by using the patch pipette as a delivery device. The reasoning behind this experiment was that the only way in which the myc epitope would be visualized is if the transfected mRNA was translated. Indeed, in the presence of brain-derived neurotrophic factor or neurotrophin-3 (to stimulate protein synthesis) the myc epitope was visible, thus providing direct evidence that proteins can be synthesized in dendrites. These data additionally provided support for stimulated protein synthesis in dendrites, which is compatible with the work in acute hippocampal slice preparations by Kang and Schuman (12). Protein synthetic machinery has been identified within dendrites and translation of exogenous mRNAs has clearly demonstrated that dendritic translation occurs independently of neuronal cell bodies. However, although dendritic ribosomes, mRNAs, and membraneous structures have been extensively characterized, the actual sites of protein synthesis in living dendrites have not been explored. To examine and characterize the endogenous dendritic translation sites, we used an assay that monitors protein synthesis in living dendrites whose cell bodies had been removed. These “isolated dendrites” were transfected with mRNA encoding green fluorescent protein (GFP) (Fig. 3 A-C) and green fluorescence that appeared upon translation of the GFP mRNA was recorded with a multiphoton laser-scanning microscope (Fig. 3 D and E). Over a time course of hours fluorescence was detected in some dendrites (Fig. 3 D and E). This “basal translation” was linear and nonsaturating over a time course longer than 3 h posttransfection (Fig. 3 E). These studies highlight the “site-specific” nature of translation within dendrites because the translation sites do not appear to move or, in other words, they are immobile in the dendrite. Visualization and additional characterization of these in vivo translation sites will permit a careful dissection of the kinetics and biology of stimulated protein synthesis in dendrites. DISCOVERY OF A SECOND MESSENGER SYSTEM IN DENDRITES The identification of multiple dendritically localized mRNAs and the evidence of local dendritic translation suggests that translation of mRNA in dendrites may produce proteins that serve various biological functions. Clearly, the mechanisms of signaling between presynaptic and postsynaptic neurons are quite complex. The production of action potentials was the first of these signaling mechanisms characterized. It has previously been shown that presynaptic activation of signal transduction cascades in the postsynaptic cell can converge on the postsynaptic nucleus to alter transcription factor activity, potentially resulting in activation or suppression of gene transcription. One example of how this can occur is shown in the intracellular convergence of signaling events that modify the activity of localized cAMP responsive element binding protein (CREB), resulting in an averaging of the cellular response to presynaptic stimulation. This averaging results in the loss of the spatially activated influences of presynaptic input on individual dendrites. Pertinent to this discussion is that upon analysis of amplified mRNA from isolated dendrites and growth cones of hippocampal neurons, select transcription factor mRNAs were shown to be present in this subcellular compartment (13). This discovery led to the hypothesis that synthesis of transcription factor proteins within dendrites would provide a novel and direct signaling pathway between



7083

the distal dendrite and the nucleus, resulting in modulation of gene expression important for neuronal functioning (13, 14). Among the transcription factor mRNAs shown to be present within developing dendrites was CREB (13). This initial mRNA discovery was expanded on to show that CREB is present in dendrites, that translation of CREB mRNA in isolated dendrites is feasible, and that CREB found in dendrites can interact with the cis-acting CRE DNA sequence using a novel in situ Southwestern assay (13). Further it was shown that CREB in dendrites is not transported to this site from the cell body because fluorescently tagged CREB microperfused into the soma did not move into the dendrites (13). In addition, CREB microperfused into dendrites was rapidly transported to the nucleus, its likely site of bioactivity. Lastly, using the isolated dendrite system it was shown that phosphorylation of Ser133 on CREB (which will greatly stimulate cAMPdependent gene expression) can occur in isolated dendrites independently of the nucleus (13). These data suggest the existence of a novel regulatory pathway in which transcription factors, synthesized and posttranslationally modified in dendrites, directly move to the neuronal nucleus to alter gene expression bypassing the integration of signal transduction pathways that converge on the nucleus (13, 14).

Fig. 3. Transfection of isolated dendrites with GFP mRNA results in fluorescence. (A and B) Phase images of 4-day-old primary hippocampal neurons grown on grided coverslips. Arrows indicate positions of cell bodies removed with a micropipette. (C) Two identical enlarged images of one dendrite from B. Bold black line indicates the position of the dendrite. (D) Four fluorescence images of dendrite in B before (0) and after (63, 133, and 203 min) transfection with GFP mRNA. Images were contrast-stretched in National Institutes of Health IMAGE for display purposes. (Scale bars = 50 µm.) (E) Mean fluorescence in isolated dendrites over a time course of hours. Black bar indicates period of transfection with GFP mRNA. n=3. DEMONSTRATION OF SYNTHESIS AND MEMBRANE INSERTION OF INTEGRAL MEMBRANE PROTEINS IN ISOLATED DENDRITES As mentioned previously in this article, our laboratory showed that glutamate receptor mRNAs are present in dendrites. This was an unexpected result because the proteins encoded by these mRNAs are integral membrane receptors and there is little electron microscopic evidence of classical rough endoplasmic reticulum or Golgi apparatus in the dendritic compartment. These are the subcellular structures necessary for synthesis and membrane insertion of integral membrane proteins into the cellular membrane. Immunohistochemical analysis at the light microscopy level has shown the presence of some protein components of the rough endoplasmic reticulum and Golgi in dendrites most notably at the base of the dendritic



7084

spine (8, 9, 15). The initial discovery of integral membrane protein encoding mRNAs in dendrites recently has been followed by showing that integral membrane proteins can indeed be synthesized in synaptic regions (Fig. 4) (16). This was accomplished by creating fusion constructs between glutamate receptor mRNA and a sequence encoding the c-myc epitope such that the translated glutamate receptor protein is extended with c-myc. Fig. 4 shows a GluR2-cMyc (c-myc engineered onto the C-terminal end of GluR2 protein) construct transfected into isolated dendrites. Transfected dendrites show low basal levels of GluR2-c-myc translation as evidenced by nearly undetectable GluR2-cmyc immunoreactivity (Fig. 4, A transmission image, B diaminobenzidine immunoreactivity). Upon DHPG stimulation GluR2-c-myc immunoreactivity is dramatically increased (Fig. 4 C and D).

Fig. 4. Stimulation of glutamate receptor mRNA translation in isolated dendrites. (A and C) The phase-contrast images of transected dendrites that correspond to dendrites that have been transfected with GluR2-c-myc shown in B and D, respectively. The dendrites in D have been treated with DHPG to stimulate protein synthesis, whereas those in B are DHPG naïve. Arrows point to individual dendrites that are visible in the paired panels. See ref. 16 for more information. The ability of the dendrite to translate an integral membrane protein, however, does not mean that the protein is inserted into the dendritic membrane. To examine this, we used the most current information concerning the topology of the glutamate receptor that shows that the N terminus of the protein is localized on external side of the cell membrane. Consequently, if a dendritically localized receptor is inserted into the membrane then the N terminus would be on the extracellular surface of the dendrite. Experimentally, the 5 end of GluR2 was engineered to contain the c-myc sequence so that translated c-myc-GluR2 would exhibit c-myc on the exterior of expressing cells. When this mRNA was transfected into isolated living dendrites and the dendrites were stimulated by various neurotransmitters, cmycglutamate receptor immunoreactivity was clearly visible within the dendrites (16). Further, when the dendritic membrane was not permeabilized the c-myc-glutamate receptor fusion construct was visualized on the cell surface while Map2, a highly abundant dendritic protein, was absent from the membrane but present in the cytoplasm (16). These data show that integral membrane proteins can be synthesized in dendrites and that these proteins can be inserted into the membrane, suggesting that a functional rough endoplasmic reticulum and Golgi exist in dendrites (16). These results suggest a potential mechanism for the modulation of the receptor repertoire under specific synapses through alterations in glutamate receptor subunit representation locally in response to synaptic stimulation. Not only is it likely that receptor subunit composition is altered, but the posttranslational modifications and association with accessory proteins also may differ for dendritically synthesized receptors as compared with cytoplasmically synthesized receptors. There are clear implications of these data in furthering our understanding of the Hebbian synapse. SUMMARY The localization of a subset of the cellular mRNAs (5%) to dendrites and the formal proof of local dendritic protein synthesis suggests that the dendrite serves a specialized role in regulating neuronal functioning. It is easy to envisage a dendrite as a passive entity capturing presynaptic information and passively transferring this information to the cell soma. To act in this manner the dendrite would not need to perform protein synthesis, it would just need to act as a cable, or wire, to propagate the information. The fact that stimulated protein synthesis occurs in dendrites, and data suggesting that this is important for various physiological properties, including long-term potentiation and long-term depression, suggests that incoming presynaptic information is either modified or modulated in postsynaptic dendrite. Alternatively, the postsynaptic dendrite may be modified in response to presynaptic input such that it responds differently to additional presynaptic input. As evidenced by data presented in this manuscript, and the work of many others, it is clear that both types of change occur. For example, a change in glutamate receptor repertoire at the synapse could alter the dendrite’s responsiveness to presynaptic glutamate challenge. As another example, if a presynaptic signal causes CREB to be synthesized and specifically phosphorylated in the dendrite with its subsequent movement to the nucleus, specific alterations in gene expression could be induced related to the type and intensity of presynaptic input (17). It is currently unclear which is the dominating regulatory feature of postsynaptic responsiveness to presynaptic input. The dendrite responds to presynaptic input through the generation of a combinatorial set of coordinated responses that are integrated into a coherent signal for propagation to the next postsynaptic neuron. Questions that arise from these considerations include: How many signal transduction pathways are modulated in the dendrite in response to a specific type of presynaptic input? Is the percentage of synapses that are modified by local protein synthesis as important as the proximal or distal position of the modified synapses on the dendrite? The complexities of the biological processes and information processing that occur within the dendrite suggest that the dendrite should be viewed as the primary site for filtering and modulating presynaptic input into the neuron. As highlighted in this manuscript and others from the National Academy of Sciences colloquium, data collected during the last decade have provided extensive information concerning the molecular composition of neuronal dendrites. With this foundation, the challenge for the coming decade will be to place this information into its functional context both in vitro and in vivo (14). Such information undoubtedly will be useful in many ways including (i) helping to develop drugs targeted to dendrite function and (ii) providing information that should be useful in the generation of better neural network algorithms. We thank Margie Moronski for preparing cultured rat hippocampal neurons for these studies. Jim Sanzo’s technical assistance with the multiphoton microscopy is greatly appreciated. This work was funded by National Institutes of Health Grants AG9900 and MH58561 to J.E.



7085

1. Bodian, D. (1965) Proc. Natl. Acad. Sci. USA 53, 418–425. 2. Garner, C., Tucker R. & Matus, A. (1988) Nature (London) 336, 374–377. 3. Miyashiro, K., Dichter, M. & Eberwine, J. (1994) Proc. Natl. Acad. Sci. USA 91, 10800–10804. 4. Crino, P. & Eberwine, J. (1996) Neuron 17, 1173–1187. 5. Buchhalter, J. & Dichter, M. (1991) Brain Res. Bull. 26, 333–338. 6. Weiler, I.J., Irwin, S., Klintsova, A., Spencer, C., Brazelton, A., Miyashiro, K., Comery, T., Patel, B., Eberwine, J. & Greenough, W. (1997) Proc. Natl. Acad. Sci. USA 94, 5395–5400. 7. Weiler, I.J. & Greenough, W. (1993) Proc. Natl. Acad. Sci. USA 90, 7168–7171. 8. Tiedge, H. & Brosius, J. (1996) J. Neurosci. 16, 7171–7181. 9. Gardiol, A., Racca, C. & Triller, A. (1999) J. Neurosci. 19, 168–179. 10. Davis, L., Banker, G. & Stewart, O. (1987) Nature (London) 330, 477–480. 11. Weiler, I.J., Wang, X. & Greenough, W. (1994) Prog. Brain Res. 100, 189–194. 12. Kang, H. & Schuman, E. (1996) Science 273, 1402–1406. 13. Crino, P., Khodakhah, K., Becker, K., Ginsberg, S., Hemby, S. & Eberwine, J. (1998) Proc. Natl. Acad. Sci. USA 95, 2313–2318. 14. Eberwine, J. (1990) in Dendrites, eds. Stuart, G., Spruston, N. & Hausser, M. (Oxford Univ. Press, Oxford), pp. 68–84. 15. Torre, E. & Steward, O. (1996) J. Neurosci. 16, 5967–5978. 16. Estee Kacharmina, J., Job, C., Crino, P. & Eberwine, J. (2000) Proc. Natl. Acad. Sci. USA 97, 11545–11550. 17. Eberwine, J., Job, C., Estee Kacharmina, J., Miyashiro, K. & Therianos, S. (2001) in Cell Polarity and Subcellular RNA Localization, ed. Richter, D. (Springer, New York), pp. 57–68.


CYTOSKELETAL MICRODIFFERENTIATION: A MECHANISM FOR ORGANIZING MORPHOLOGICAL PLASTICITY IN DENDRITES

7086

Colloquium Cytoskeletal microdifferentiation: A mechanism for organizing morphological plasticity in dendrites Stefanie Kaech*, Hema Parmar*, Martijn Roelandse, Caroline Bornmann, and Andrew Matus† Friedrich Miescher Institute, 4058 Basel, Switzerland Experimental evidence suggests that microfilaments and microtubules play contrasting roles in regulating the balance between motility and stability in neuronal structures. Actin-containing microfilaments are associated with structural plasticity, both during development when their dynamic activity drives the exploratory activity of growth cones and after circuit formation when the actin-rich dendritic spines of excitatory synapses retain a capacity for rapid changes in morphology. By contrast, microtubules predominate in axonal and dendritic processes, which appear to be morphologically relatively more stable. To compare the cytoplasmic distributions and dynamics of microfilaments and microtubules we made time-lapse recordings of actin or the microtubule-associated protein 2 tagged with green fluorescent protein in neurons growing in dispersed culture or in tissue slices from transgenic mice. The results complement existing evidence indicating that the high concentrations of actin present in dendritic spines is a specialization for morphological plasticity. By contrast, microtubule-associated protein 2 is limited to the shafts of dendrites where time-lapse recordings show little evidence for dynamic activity. A parallel exists between the partitioning of microfilaments and microtubules in motile and stable domains of growing processes during development and between dendrite shafts and spines at excitatory synapses in established neuronal circuits. These data thus suggest a mechanism, conserved through development and adulthood, in which the differential dynamics of actin and microtubules determine the plasticity of neuronal structures. Neuronal circuits need to maintain a delicate balance between stability and plasticity. On the one hand, the synaptic connections they make must be stable enough to support reliable signal transmission, while on the other, they must be sufficiently plastic to accommodate changes in connectivity that are necessary for the long-duration adaptation of behavior to sensory experience. How is neuronal structure organized and regulated to accommodate these diverse needs? Increasingly, experimental evidence implicates the neuronal cytoskeleton in regulating morphological plasticity in adult as well as developing tissue. More than any other cell type, neurons depend for their distinctive morphology on the cytoskeleton whose protein components are organized in a set of microdifferentiated compartments that mirror the polarized form of the cell and play a significant role in determining its development (1–3). Microfilaments and microtubules act together to guide and support the growth and differentiation of axons and dendrites. Whereas dynamic actin filaments drive the exploratory activity of growth cones as they respond to external guidance cues, microtubules stabilize the structure of the newly established process (4–10). Recent results suggest that a similar “division of labor” between the two cytoskeletal filament systems may persist in dendrites beyond the developmental period. In adult brain the highest concentrations of actin are associated with dendritic spines that form the postsynaptic component of most excitatory synapses (11–13). This dendritic spine actin retains a capacity for dynamic activity and can drive rapid changes in their shape (14–18). By contrast, the highest concentrations of the microtubule components, including tubulin and the microtubuleassociated proteins (MAPs), occur in the shafts of dendrites (19–23). This is consistent with ultrastructural studies showing microtubule bundles as the predominant cytoskeletal components of dendrite shafts whereas the cytoplasm of spines is characterized by a meshwork of fine filaments consistent with the predominance of actin-containing microfilaments (24–26). Despite these indications for separation between the two filament systems, the nature of the interface between the microtubule and microfilament domains has remained uncertain because of immunohistochemical data suggesting that MAP2, the major dendritic MAP, is present at postsynaptic sites and in dendritic spines (19, 27, 28). MAP2 can bind to actin in vitro (29, 30) so if it is present in spines this might suggest that it can act as a bridge between actin and microtubules at spine synapses. Studies using transfected fibroblastic cells have yielded diverse results regarding potential interactions between MAP2 and the actin cytoskeleton (31–35). Because cytoskeletal components are likely to be important in determining the locus of anatomical plasticity in dendrites (16, 18, 36–39) we have re-examined the distribution of actin and MAP2 in primary neurons by using fluorescent protein tags that allow the both the location and the dynamics of cytoskeletal proteins to be determined in living cell (40). The results show a striking compartmentalization of the cytoskeleton in dendrites with microtubule proteins limited to the dendritic shaft whereas actin is overwhelmingly concentrated in spines. This distribution is accompanied by a differentiation of dendrite structure into highly motile postsynaptic elements, the spines, and morphologically more stable elements, the dendrite shafts. METHODS Eukaryotic expression constructs containing actin and MAP2c and MAP2b tagged with green fluorescent protein (GFP) under control of chicken ß-actin sequences and techniques for preparing time-lapse recordings from transfected hippocampal neurons were as described (14, 33). The topaz spectral variant of GFP (41), here referred to as YFP (yellow fluorescent protein), was obtained from Packard Bioscience and was used to replace GFP in existing vectors by standard techniques. Transgenic mice expressing actin-GFP have been described (42). Transgenic mice expressing MAP2-GFP were generated by cloning a fragment containing the MAP2-GFP coding

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: MAP, microtubule-associated protein; GFP, green fluorescent protein; YFP, yellow fluorescent protein; NMDA, Nmethyl-D-aspartate. *S.K. and H.P. contributed equally to this work. †To whom reprint requests should be addressed at: Friedrich Miescher Institute, Maulbeer-strasse 66, 4058 Basel, Switzerland. E-mail: [email protected].



7087

sequence (33) into the pTSC vector containing a modified Thy-1 promoter (43). A 9.5-kb EcoRv/PvuI fragment was injected into oocytes of B6CF1 strain mice, and transgenic lines were established by standard techniques. Positive progeny were identified by PCR using GFPspecific primers and by Southern blot analysis. Organotypic slice cultures from transgenic mice were established as described by Gahwiler et al. (44). After at least 4 weeks in culture individual slices were mounted in purposebuilt observation chambers (Life Imaging Services, Olten, Switzerland) and perfused with artificial cerebrospinal fluid or Tyrode’s solution. No difference was apparent between buffers. Time-lapse recordings were made by using a Leica IRBE inverted microscope equipped with a Nipkow disk-microlens confocal system (Life Science Resources, Cambridge, U.K.). To display time-dependent changes in printed figures, a subtraction protocol was used to sum differences between images in time-lapse recordings by using METAMORPH software (Universal Imaging, West Chester, PA). The results were displayed by using a pseudocolor look-up table with dark blue indicating lack of change and red to yellow increasing amounts of motility (42). RESULTS Comparison of Actin and MAP2 Distributions Using Spectral Variants of GFP. To asses the distribution and dynamics of microfilaments and microtubules in dendrites, we prepared eukaryotic expression vectors containing actin labeled with GFP and MAP2c labeled with YFP. To compare their properties within the same dendrite, hippocampal neurons from 18-day rat embryos were simultaneously transfected with actin-GFP and MAP2c-YFP and maintained in dispersed cell culture for at least 3 weeks. By this time most excitatory synapses are made onto dendritic spines of mature appearance that are contacted by presynaptic terminals whereas earlier immature lateral filopodia are abundant (45–49). Fig. 1 A shows a living cell in such a culture visualized by phase-contrast microscopy (Left) and with filter sets selective for GFP (Center) or YFP (Right). Even at the low magnification shown in Fig. 1 A, punctate labeling along dendrites, indicative of actin-GFP accumulation in dendritic spines, was evident (Center). By contrast the same dendrites visualized by MAP2-YFP were smooth in appearance (Right), indicating the absence of MAP2 from dendritic spines. To study these distributions in more detail, actin-GFP and MAP2c-YFP images were taken at higher magnification and compared by assigning them contrasting colors (actin, green; MAP2c, red) and overlaying the images. Fig. 1 B shows the results of this procedure for a segment of dendrite from a doubly transfected cell in which the strong targeting of actin into spines and the contrasting restriction of MAP2 to the dendrite shaft is evident. The data shown in Fig. 1 B were taken from a time-lapse recording in which successive images were captured alternately by using the GFP or YFP filter sets. Such recordings show the same rapid dynamics of actin in dendritic spines described in previous studies (14, 42). By contrast, MAP2 showed no detectable dynamic activity over the 15 min of recording (see Movie 1, which is available as supplemental data on the PNAS web site, www.pnas.org). To represent this result in still images, six frames of actin-GFP and six frames of MAP2c-YFP, recorded alternately 30 s apart, were converted into profile outlines by using a computer routine. Each was assigned a different color and all six then were overlaid onto a single gray-scale fluorescence image from the same timelapse series. Changes in the shape of dendritic spines then are revealed by the separately colored outlines representing the successively recorded images (Fig. 1 C). By contrast, the same procedure applied to images of MAP2c shows no detectable change during the period of recording (Fig. 1 D).

Fig. 1. Actin and MAP2 differ in both distribution and dynamics in living hippocampal neurons. (A) Distribution of actin and MAP2 in a transfected hippocampal neuron in cell culture for 24 days, simultaneously expressing actin-GFP and MAP2c-YFP. The phase-contrast image (Left) shows the arrangement of the cell body and processes of the transfected cell interspersed with the network of axonal processes of untransfected cells. The original gray-scale images for actin-GFP (Center) and MAP2c-YFP (Right) images were prepared by using appropriate selective filter sets. (Bar=20 µm.) (B) Comparative distribution of actin and MAP2 in a dendrite segment produced by overlaying pseudocolored images for actin-GFP (green) and MAP2c-YFP (red). The high concentration of actin in dendritic spines (arrowheads) contrasts with the confinement of MAP2 to dendrite shafts. (Bar=2 µm.) (C and D) Time-dependent changes in the configuration of actin and MAP2 in dendrites. Six frames from a single time-lapse recording for actin-GFP (C) and MAP2-YFP (D) images, recorded alternately 30 s apart, were converted into profile outlines. Each outline was assigned a different color and overlaid onto a single gray-scale image from the same recorded sequence. Variations between the different color outlines indicate regions of morphological change that are evident in the actin images of dendritic spines (C) but are absent from the MAP2 images of the dendrite shaft (D). (Bar=2 µm.) Refer to supplemental Movie 1 for the original time-lapse sequence. This tight localization of MAP2 to dendritic microtubules was not only seen for the juvenile MAP2c splice variant but also for the high molecular weight MAP2b form that is expressed in the adult brain (50). Fig. 2 shows results for hippocampal neurons transfected with MAP2b-GFP. Like the embryonic MAP2c form, adult MAP2b is localized in dendrites but not within axons (arrow in Fig. 2 A and B). Both here and in higher magnification



7088

images MAP2b was bound to microtubules in dendrite shafts and did not enter dendritic spines (Fig. 2 C and D).

Fig. 2. MAP2 is absent from dendritic spines. (A and B) Like the embryonic low-molecular weight variant MAP2c, highmolecular weight MAP2b is confined to the somatodendritic domain of hippocampal neurons and is absent from the axon (arrow). Shown are both a phase (A) and a fluorescence (B) image of a live neuron transfected with GFP-tagged MAP2b and kept in culture for 14 days. (Bar=25 µm.) (C and D) This neuron transfected with MAP2b-GFP was maintained in culture for 4 weeks, by which time cells carry many dendritic spines. In the enlarged image (D) of the area outlined in C, the restriction of MAP2bGFP fluorescence to microtubule bundles in the dendritic shaft is obvious. No fluorescence is detected in spine protuberances from the dendrite. (Bars: C=15 µm; D=2 µm.) Time-Lapse Recording of MAP2-GFP in Tissue Slices from Transgenic Mice. Time-lapse recordings of GFP-labeled MAP2c in dendrites of dispersed cells consistently failed to show dynamic activity of the microtubule cytoskeleton over periods of up to 30 min. However, it remained possible that changes might occur over longer periods, particularly because transfection experiments using fibroblastic cells indicate that, although MAP2 slows microtubule dynamics, it does not inhibit them entirely (33). Indeed, substantial changes in configuration of the microtubule cytoskeleton are visible when time-lapse recordings are made from MAP2c-GFP transfected cells over periods of several hours (33). To address the question of whether comparable changes occur in dendrites we raised transgenic mice expressing MAP2c-GFP in central nervous system neurons (Fig. 3). Like the actin-GFP expressing animals we have previously described (42), MAP2c-GFP mice are indistinguishable from nontransgenic litter mates in morphology, fertility, and lifespan and show no obvious behavioral abnormalities nor deficits in the Morris water maze (H.P., P. Kelly, and A.M., unpublished data). This lack of overt effects of expressing exogenous MAP2 is consistent with results we previously obtained for transgenic mice expressing high levels of epitope-tagged MAP2c (51). In organotypic hippocampal slice cultures established from these animals MAP2c-GFP is readily detectable in dendrites with weaker expression occurring in cell bodies (Fig. 3 A). In more than 50 independently established cultures MAP2 was always limited to the shafts of dendrites (see, for example, Fig. 3 B). In several hundred cells examined within these cultures we have failed to find any evidence for the presence of MAP2c-GFP in dendritic spines. Confocal time-lapse recordings of MAP2c-GFP in hippocampal slices from transgenic mice showed a surprising lack of motility. Fig. 3 C shows data from a 4-week-old culture where the general distribution of MAP2c-GFP is shown by the single frame of original fluorescence data (Left). Fig. 3 C Right shows a “difference image” prepared by subtracting gray scale values for pixels in 30 successive frames and then summing the differences (see ref. 42). The values are displayed on a pseudocolor scale in which areas where there was little change are colored blue while those where large changes occurred appear red and white. As the overall blue color of Fig. 3 C Right shows, there was little change in the MAP2c-GFP image during the 10 min of recording. Similar time-lapse recordings of MAP2c-GFP in hippocampal slices were made for periods of up to 3 h (n=12). Fig. 3 D shows an example focused on a single dendrite recorded continuously for 3 h in which the blue coloration of the MAP2c-containing dendrite indicates the



7089

lack of change in the image (compare this to the actin-GFP pseudocolor image of dendritic spines shown below in Fig. 4 B). We considered the possibility that microtubules in dendrites might not show dynamic activity except under conditions of enhanced stimulation. Because both long-term potentiation of synaptic responses and stimulation induced increases in dendritic spine numbers are associated with activation of N-methyl-D-aspartate (NMDA) receptors (52–56) we made timelapse recordings of MAP2c-GFP in slices during exposure to NMDA or to the NMDA receptor antagonist MK-801. In neither case was any change in MAP2c-GFP images detectable in recordings of up to 2 h duration.

Fig. 3. Time-lapse recording of MAP2 in hippocampal tissue slices from transgenic mice stably expressing MAP2c-GFP. (A) Confocal GFP fluorescence image taken near the cell body layer of area CA1 in an organotypic slice culture established from an 11-day-old transgenic mouse and maintained in culture for 25 days. Nuclei in cell bodies are marked with *. (Bar=10µm.) (B) MAP2 localization in hippocampal neurons is limited to the shafts of dendrites. Single frame taken from a time-lapse recording of MAP2c-GFP fluorescence in the CA1 neuropil of a hippocampal slice maintained in culture for 4 weeks. (Bar=5 µm.) (C and D) Short-term time-lapse assay for MAP2 dynamics. (C Left) A single confocal gray-scale image of MAP2-GFP fluorescence in area CA1 of a 5-week-old hippocampal slice culture. (C Right) A pseudocolor “difference image” produced by summing gray-scale differences between images taken 30 sec apart over 10 min of time-lapse recording. Compare the overall lack of change in the MAP2 image (dark blue color) during the recording period to the high degree of change (green, yellow, and red) in actin images of similar configuration (Fig. 4.4 Right). (Bar=5 µm.) (D) Long-term timelapse assay for MAP2 dynamics. Original gray-scale fluorescence image (Upper) and pseudocolor difference image (Lower) of a dendrite segment followed over a 3-h time period. The dark blue color again indicates an overall lack of change in MAP2 distribution during this longer recording period. (Bar=5 µm.)



7090

Fig. 4. Time-lapse recording of actin dynamics in dendrite spines of hippocampal tissue slices from transgenic mice expressing actin-GFP. (A) (Left) An original fluorescence image in a single frame from a time-lapse recording in which frames were collected 30 sec apart. (Right) Changes in actin distribution over 10 min displayed by difference imaging using a pseudocolor scale (see text for details). Red and yellow patches indicate areas of high motility associated with dendritic spines. (Bar = 10 µm.) (B) Single gray-scale frame (Upper) and pseudocolor difference image at higher magnification. Shape changes are associated with dendritic spines (red and yellow patches) whereas the dendrite shaft shows little dynamic activity (Lower). (Bar=10 µm.) Actin-GFP Shows High Motility in Dendritic Spines of Transgenic Mice. Because MAP2-GFP in dendrites showed so little dynamic activity, we made comparable time-lapse recordings of actin-GFP in hippocampal slice cultures from transgenic mice. As previously reported (42), actin-GFP was concentrated in heads of dendritic spines (Fig. 4 A Left and B Upper). Time-lapse recordings made from these cultures confirmed that this spine actin is rapidly dynamic. This is shown by the pseudocolored difference images in Fig. 4 in which areas where there were large changes in the image during the 10 min of recording are colored red and yellow and areas where little change occurred appear blue. DISCUSSION Our data indicate that the cytoskeleton in neuronal dendrites is partitioned into distinct microtubule and microfilament domains associated with dendrite shafts and spines, respectively. This finding is in contrast to earlier immunocytochemical studies, which reported the presence of MAP2 in dendritic spines (19, 27, 28). A possible reason for this discrepancy is that the reaction product of immunoperoxidase staining used to detect MAP2 in the earlier studies spread from its origin at microtubules in the dendrite shaft into dendritic spines. Electron microscope studies generally confirm the results of our live cell imaging observations by showing that while microtubules are abundant in dendrites they are absent from spines that instead contain a meshwork of microfilaments consistent with the presence of high actin concentrations (11, 24–26). An exception is the presence of microtubules in large, branched spines in area CA3 of the hippocampus but these spines also contain ribosomes, multivesicular bodies, and mitochondria (57) emphasizing the special status conferred by their large size. Microtubules are not found in other CA3 spine types of more usual size, supporting the conclusion that microtubules generally do not extend into the spine cytoplasm. Based on the partitioning of microfilaments and microtubules between shaft and spine the dendrite cytoplasm can be considered, from the perspective of plasticity, as divided into separate microtubule (M) and actin (A) zones (Fig. 5). Interesting questions arise concerning events at the transition zone (T, Fig. 5). Because most neurons are postmitotic, their molecular components must be continuously replaced. For some proteins, including MAP2 (58), this is achieved by export of mRNA into dendrites, presumably followed by local synthesis of the protein product (59, 60). However, for most neuronal proteins, synthesis occurs in the cell body followed by transport into axons and dendrites. This process is best understood for membrane proteins that are transported on vesicles. These are conveyed along microtubules by motor molecules of the kinesin and dynein families, which can confer directional specificity toward axon or dendrite (61–64). The recent identification of a dendrite-specific kinesin, KIF17, bound to NMDA-2B glutamate receptor subunits as part of a vesicular complex (65) confirms the existence of a mechanism for transporting functional components of spine synapses along dendritic microtubules. The ultimate destination of the NMDA receptor subunits is the postsynaptic membrane, raising the question of how the vesicle that contains them travels from the microtubule transport system of the dendrite shaft to the postsynaptic membrane at the tip of a dendritic spine. Evidence for interactions between microtubule- and actin-based transport mechanisms near the cell surface (66, 67), together with the demonstration of a direct interaction between microtubule and actin transport motors (68), suggest that this transition may be accomplished by transfer of vesicles possessing motors for both systems from microtubules to microfilaments. A hypothetical scheme for this transfer is indicated diagrammati



7091

cally in Fig. 5 where vesicles move from microtubule to microfilament transport systems at the base of the spine. The management of this putative transition remains to be determined because a thin cortical layer of actin filaments is also present within dendrite shafts. The mechanisms responsible for delivering materials via the spine cytoplasm to sites in the postsynaptic junction have significant implications for synaptic plasticity in view of growing evidence for physical exchange of receptor molecules in the postsynaptic membrane of glutamatergic synapses (69–72).

Fig. 5. Hypothetical scheme for the partitioning of cytoskeletal microdomains between shaft and spine in dendrites. (Left) Part of dendrite in the region of a spine synapse. The axonal component (ax.), with its swollen presynaptic (pre.) bouton containing synaptic vesicles (sv.) is outlined in gray. It forms a synapse at the tip of a dendritic spine head. Inside the spine head the junctional region is marked by the postsynaptic density (psd.), a complex of scaffolding proteins that acts as the platform for assembling functional molecules such as neurotransmitter receptors and ion channels. The cytoskeleton of the dendritic spine is composed of actin filaments (barbed lines) that are inserted into the psd. The cytoskeleton of the underlying dendrite consists predominantly of microtubules (gray rods), which in dendrites are bidirectionally oriented so that some have the plus ends distally and others the minus end distally as indicated. This distribution of cytoskeletal filaments demarcates three cytoplasmic zones, an M zone in the dendrite shaft, where microtubules predominate, an A zone in the dendritic spine, where actin filaments predominate, and a T, or transition, zone. (Right) The expanded diagram shows the relationship of these zones to the delivery of materials to the synaptic domain as suggested by current evidence. Transport vesicles (blue filled circles) carry cargoes of functional molecules, such as NMDA receptors (pale blue symbols), bound for the postsynaptic membrane. These vesicles bear both microtubuledependent (M, kinesin and dynein) and actin-dependent (A, myosin) motor molecules. Transitory detachment of kinesin and dynein from microtubule tracks provides the opportunity for the myosin motors of transport vesicles to engage with the actin filaments of dendritic spines along which they travel to the synaptic domain. Single chevrons in the vicinity of the postsynaptic membrane represent the presence of labile actin filaments in this zone. The necessity of special mechanisms for transferring materials from shaft to spine raises the question of why such a partitioning of dendrite structure should exist at all. One possibility, suggested by the results of the present study, is that this separation is a specialization for regulating anatomical plasticity. As our time-lapse recordings show, the actin and microtubule domains are associated with distinct rates of plasticity. Whereas actin in dendritic spine defines a region of rapid morphological change occurring over seconds and minutes (14, 15, 17), time-lapse imaging of MAP2 suggests that microtubules in the dendrite shaft undergo little change in periods of up to 3 h. This does not exclude that dynamic changes in dendritic microtubules may occur over longer periods. Indeed time-lapse imaging of MAP2-containing microtubule bundles in transfected epithelial cells shows that gradual alterations in the configuration of the microtubule cytoskeleton can occur over periods of several hours (33). This finding suggests that MAP2-containing neuronal microtubules may have a capacity for morphological plasticity although at a rate intrinsically slower than that of actin filament arrays, which appear constantly motile in comparable recordings (14). That gradual changes in the extent and branching of dendrites can occur has been demonstrated by repetitive imaging of dendrites in superior cervical ganglia of adult rats where substantial changes in dendritic arbors have been documented over periods of weeks and months (73, 74). However, other studies support the idea that dendritic spines are the predominant site of activitydependent morphological plasticity in the brain in vivo (for example, refs. 17 and 75–78). Taken together these observations suggest that microdifferentiation of the dendritic cytoskeleton in mature neurons may be a cellular specialization for dividing the structural support of dendrites into two levels of stability. One of these, involving microtubules, appears to respond slowly, providing morphological stability to dendrite arbors while still allowing for long-term flexibility, whereas the other, involving motile actin filaments, allows for rapid, activity-dependent changes in synaptic structure. We thank Thierry Doll and Jean-Francois Spetz for assistance in preparing transgenic animals. 1. Matus, A., Huber, G. & Bernhardt, R. (1983) Cold Spring Harbor Symp. Quant. Biol. 48, 775–782. 2. Craig, A.M. & Banker, G. (1994) Anna. Rev. Neurosci. 17, 267–310. 3. Bradke, F. & Dotti, C.G. (1999) Science 283, 1931–1934. 4. Mitchison, T. & Kirschner, M. (1988) Neuron 1, 761–772. 5. Smith, S.J. (1988) Science 242, 708–715. 6. Gordon-Weeks, P.R. (1991) BioEssays 13, 235–239. 7. Avila, J., Dominguez, J. & Diaz-Nido, J. (1994) Int. J.Dev. Biol. 38, 13–25. 8. Tanaka, E. & Sabry, J. (1995) Cell 83, 171–176. 9. Heidemann, S.R. (1996) Int. Rev. Cytol. 165, 235–296.



7092

10. Letourneau, P.C. (1996) Perspect Dev. Neurobiol. 4, 111–123. 11. Matus, A., Ackermann, M., Pehling, G., Byers, H.R. & Fujiwara, K. (1982) Proc. Natl. Acad. Sci. USA 79, 7590–7594. 12. Fifkova, E. (1985) Brain Res. 356, 187–215. 13. Kaech, S., Fisher, M., Doll, T. & Matus, A. (1997) J. Neurosci. 17, 9565–9572. 14. Fischer, M., Kaech, S., Knutti, D. & Matus, A. (1998) Neuron 20, 847–854. 15. Dunaevsky, A., Tashiro, A., Majewska, A., Mason, C. & Yuste, R. (1999) Proc. Natl. Acad. Sci. USA 96, 13438–13443. 16. Halpain, S. (2000) Trends Neurosci. 23, 141–146. 17. Lendvai, B., Stern, E.A., Chen, B. & Svoboda, K. (2000) Nature (London) 404, 876–881. 18. Matus, A. (2000) Science 290, 754–758. 19. Matus, A., Bernhardt, R. & Hugh-Jones, T. (1981) Proc. Natl. Acad. Sci. USA 78, 3010–3014. 20. Bernhardt, R. & Matus, A. (1982) J. Cell Biol. 92, 589–593. 21. Burgoyne, R.D. & Cumming, R. (1984) Neuroscience 11, 156–167. 22. De Camilli, P., Miller, P.E., Navone, F., Theurkauf, W.E. & Vallee, R.B. (1984) Neuroscience 11, 817–846. 23. Huber, G. & Matus, A. (1984) J. Cell Biol. 98, 777–781. 24. Peters, A., Palay, S.L. & Webster, H.d.F. (1976) The Fine Structure of the Nervous System (Saunders, Philadelphia). 25. Landis, D.M. & Reese, T.S. (1983) J. Cell Biol. 97, 1169–1178. 26. Cohen, R.S., Chung, S.K & Pfaff, D.W. (1985) Cell. Mol. Neurobiol. 5, 271–284. 27. Caceres, A., Banker, G., Steward, O., Binder, L. & Payne, M. (1984) Brain Res. 315, 314–318. 28. Morales, M. & Fifkova, E. (1989) Cell Tissue Res. 256, 447–456. 29. Sattilaro, R.F. (1986) Biochemistry 25, 2003–2009. 30. Selden, S.C. & Pollard, T.D. (1986) Ann. N.Y. Acad. Sci. 466, 803–812. 31. Takemura, R., Okabe, S., Umeyama, T., Kanai, Y., Cowan, N.J. & Hirokawa, N. (1992) J. Cell Sci. 103, 953–964. 32. Weisshaar, B., Doll, T. & Matus, A. (1992) Development (Cambridge, U.K.) 116, 1151–1161. 33. Kaech, S., Ludin, B. & Matus, A. (1996) Neuron 17, 1189–1199. 34. Ozer, R.S. & Halpain, S. (2000) Mol. Biol Cell 11, 3573–3587. 35. Colella, R., Lu, C., Hodges, B., Wilkey, D.W. & Roisen, F.J. (2000) Brain Res. Dev. Brain Res 121, 1–9. 36. van Rossum, D. & Hanisch, U.K. (1999) Trends Neurosci. 22, 290–295. 37. Harris, K.M. (1999) Curr. Opin. Neurol. 9, 343–348. 38. Jontes, J.D. & Smith, S.J. (2000) Neuron 27, 11–14. 39. Parnass, Z., Tashiro, A. & Yuste, R. (2000) Hippocampus 10, 561–568. 40. Ludin, B. & Matus, A. (1998) Trends Cell Biol. 8, 72–77. 41. Ormö, M., Cubitt, A.B., Kallio, K., Gross, L.A., Tsien, R.Y. & Remington, S.J. (1996) Science 273, 1392–1395. 42. Fischer, M., Kaech, S., Wagner, U., Brinkhaus, H. & Matus, A. (2000) Nat. Neurosci. 3, 887–894. 43. Botteri, F.M., van der Putten, H., Wong, D.F., Sauvage, C.A. & Evans, R.M. (1987) Mol. Cell. Biol. 7, 3178–3184. 44. Gahwiler, B.H., Thompson, S.M., Audinat, E. & Robertson, R.T. (1991) in Culturing Nerve Cells, eds. Banker, G. & Goslin, K. (MIT Press, Cambridge, MA), pp. 379–411. 45. Bartlett, W.P. & Banker, G.A. (1984) J. Neurosci. 4, 1954–1965. 46. Papa, M., Bundman, M.C., Greenberger, V. & Segal, M. (1995) J. Neurosci. 15, 1–11. 47. Ziv, N.E. & Smith, S.J. (1996) Neuron 17, 91–102. 48. Boyer, C., Schikorski, T. & Stevens, C.F. (1998) J. Neurosci. 18, 5294–5300. 49. Fiala, J.C., Feinberg, M., Popov, V. & Harris, K.M. (1998) J. Neurosci. 18, 8900–8911. 50. Matus, A. (1988) Annu. Rev. Neurosci. 11, 29–44. 51. Marsden, K.M., Doll, T., Ferralli, J., Botteri, F. & Matus, A. (1996) J. Neurosci. 16, 3265–3273. 52. Bliss, T.V. & Collingridge, G.L. (1993) Nature (London) 361, 31–39. 53. Andersen, P., Moser, E., Moser, M.B. & Trommald, M. (1996) J. Physiol. (Paris) 90, 349. 54. Engert, F. & Bonhoeffer, T. (1999) Nature (London) 399, 66–70. 55. Maletic-Savatic, M., Malinow, R. & Svoboda, K. (1999) Science 283, 1923–1927. 56. Toni, N., Buchs, P.A., Nikonenko, I., Bron, C.R. & Muller, D. (1999) Nature (London) 402, 421–425. 57. Chicurel, M.E. & Harris, K.M. (1992) J. Comp. Neurol. 325, 169–182. 58. Garner, C.C., Tucker, R.P. & Matus, A. (1988) Nature (London) 336, 674–677. 59. Steward, O. & Banker, G.A. (1992) Trends Neurosci. 15, 180–186. 60. Schuman, E.M. (1999) Neuron 23, 645–648. 61. Brady, S.T. (1995) Trends Cell Biol. 5, 159–164. 62. Hirokawa, N. (1998) Science 279, 519–526. 63. Burack, M.A., Silverman, M.A. & Banker, G. (2000) Neuron 26, 465–472. 64. Terada, S. & Hirokawa, N. (2000) Curr. Opin. Neurobiol. 10, 566–573. 65. Setou, M., Nakagawa, T., Seog, D.H. & Hirokawa, N. (2000) Science 288, 1796–1802. 66. Gavin, R.H. (1997) Int. Rev. Cytol. 173, 207–242. 67. Goode, B.L., Drubin, D.G. & Barnes, G. (2000) Curr. Opin. Cell Biol. 12, 63–71. 68. Huang, J.D., Brady, S.T., Richards, B.W., Stenolen, D., Resau, J.H., Copeland, N.G. & Jenkins, N.A. (1999) Nature (London) 397, 267–270. 69. Nishimune, A., Isaac, J.T., Molnar, E., Noel, J., Nash, S.R., Tagaya, M., Collingridge, G.L., Nakanishi, S. & Henley, J.M. (1998) Neuron 21, 87–97. 70. Song, I., Kamboj, S., Xia, J., Dong, H., Liao, D. & Huganir, R.L. (1998) Neuron 21, 393–400. 71. Shi, S.H., Hayashi, Y., Petralia, R.S., Zaman, S.H., Wenthold, R.J., Svoboda, K. & Malinow, R. (1999) Science 284, 1811–1816. 72. Zhu, J.J., Esteban, J.A., Hayashi, Y. & Malinow, R. (2000) Nat. Neurosci. 3, 1098–1106. 73. Purves, D. & Hadley, R.D. (1985) Nature (London) 315, 404–406. 74. Purves, D., Hadley, R.D. & Voyvodic, J.T. (1986) J. Neurosci. 6, 1051–1060. 75. Valverde, F. (1967) Exp. Brain Res. 3, 337–352. 76. Coss, R.G., Brandon, J.G. & Globus, A. (1980) Brain Res. 192, 49–59. 77. Woolley, C.S. & McEwen, B.S. (1993) J. Comp. Neurol. 336, 293–306. 78. Moser, M.B., Trommald, M., Egeland, T. & Andersen, P. (1997) J. Comp. Neurol. 380, 373–381.


TRACKING THE ESTROGEN RECEPTOR IN NEURONS: IMPLICATIONS FOR ESTROGEN-INDUCED SYNAPSE FORMATION

7093

Colloquium Tracking the estrogen receptor in neurons: Implications for estrogen-induced synapse formation Bruce McEwen*†, Keith Akama*, Stephen Alves*, Wayne G. Brake*, Karen Bulloch*, Susan Lee*, Chenjian Li*, Genevieve Yuen*, and Teresa A.Milner‡ *Laboratory of Neuroendocrinology, The Rockefeller University, New York, NY 10021; and ‡Department of Neurology and Neuroscience, Weill Medical College of Cornell University, New York, NY, 10021 Estrogens (E) and progestins regulate synaptogenesis in the CA1 region of the dorsal hippocampus during the estrous cycle of the female rat, and the functional consequences include changes in neurotransmission and memory. Synapse formation has been demonstrated by using the Golgi technique, dye filling of cells, electron microscopy, and radioimmunocytochemistry. N-methyl-Daspartate (NMDA) receptor activation is required, and inhibitory interneurons play a pivotal role as they express nuclear estrogen receptor alpha (ERα) and show E-induced decreases of GABAergic activity. Although global decreases in inhibitory tone may be important, a more local role for E in CA1 neurons seems likely. The rat hippocampus expresses both ERα and ERß mRNA. At the light microscopic level, autoradiography shows cell nuclear [3H]estrogen and [125I]estrogen uptake according to a distribution that primarily reflects the localization of ERα-immunoreactive interneurons in the hippocampus. However, recent ultrastructural studies have revealed extranuclear ERα immunoreactivity (IR) within select dendritic spines on hippocampal principal cells, axon terminals, and glial processes, localizations that would not be detectable by using standard light microscopic methods. Based on recent studies showing that both types of ER are expressed in a form that activates second messenger systems, these findings support a testable model in which local, non-genomic regulation by estrogen participates along with genomic actions of estrogens in the regulation of synapse formation. The brain is widely responsive to gonadal hormones. Not only is the hypothalamus regulated by these hormones in relation to reproductive behavior and neuroendocrine physiology, but also structures like the hippocampus and midbrain serotonin system undergo sexual differentiation during perinatal development and are hormone responsive in maturity (1, 2). One of the processes regulated by ovarian hormones is the cyclic formation and breakdown of excitatory synapses on dendritic spines in the hippocampus (3). This finding was surprising because, until recently, the hippocampus was known as a brain region in which cell nuclear estrogen receptors (ER) are present in scattered inhibitory interneurons but not in principal neurons where spine formation occurs (4). Yet the effects of ovarian hormones on synaptic turnover were as impressive in the hippocampus as those in the ventromedial hypothalamus (5–7), a classic estrogen (E) target area of the brain for female sexual behavior (8). Moreover, effects of estrogens on hippocampal-dependent cognitive function are now recognized in rodents (9) and humans (10). Recent electron microscopic studies have revealed that ERs are expressed in hippocampus in non-nuclear locations within principal cells (11). This fact, along with the discovery that ER can couple to second messenger systems (12–14), has raised the possibility that ER may be involved in local signaling within neurons as well as regulating expression of genes via nuclear receptors in interneurons. Among the possible targets of local signaling is the translation of RNAs found in dendrites of hippocampal and other neurons. This review paper presents the state of current knowledge about the location of ER and progesterone receptors (PR) in hippocampus and the regulation of synapse formation by estradiol and removal by progesterone in CA1 pyramidal neurons. We start with a discussion of the functional significance of hippocampal synaptogenesis and then review what is known about the mechanism of synapse formation and the location of ER and their cellular mechanisms of action. The discovery of non-nuclear ER in dendritic spines, presynaptic nerve endings, and spineassociated glial cell processes has led us to propose a testable model for understanding the role of nuclear and non-nuclear ER in synapse formation. FUNCTIONAL SIGNIFICANCE OF ACTIONS OF ESTRADIOL IN THE HIPPOCAMPUS The functional significance of estrogen actions in the hippocampal CA1 region is evident from electrophysiological studies indicating that E treatment of ovariectomized rats produces a delayed facilitation of synaptic transmission in CA1 neurons that is N-methyl-D-aspartate (NMDA) mediated (15) and leads to an enhancement of voltage-gated Ca2+ currents (15, 16). This approach was significantly advanced by Woolley et al. (17), who used biocytin injection after recording from CA1 pyramidal neurons to visualize E induction of dendritic spines (17). Spine density correlated negatively with input resistance, and input/ output curves showed an increased slope under conditions where NMDA receptor-mediated currents predominated, whereas there was no increased slope where α-amino-3-hydroxy-5-methyl-4isoxazolepropionic acid (AMPA) receptor currents predominated (17). Other studies have shown that long-term potentiation sensitivity peaks on the afternoon of proestrus in intact female rats at exactly the time when excitatory synapse density has reached its peak (18). Proestrus is also the time of the estrous cycle when seizure thresholds in dorsal hippocampus are the lowest (19). Although activation of NMDA receptors in hippocampus is enhanced via AMPA receptors in some cases but not in others

This paper was presented at the National Academy of Sciences colloquium, “Molecular Kinesis in Cellular Function and Plasticity,” held December 7–9, 2000, at the Arnold and Mabel Beckman Center in Irvine, CA. Abbreviations: BDNF, brain-derived neurotrophic factor; CREB, cAMP response element-binding protein; CaMKII, calcium calmodulin kinase II; E, estrogen; P, progesterone; ER, estrogen receptor; PR, progesterone receptor; NMDA, N-methyl-D-aspartate; AMPA, αamino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; IR, immunoreactivity; GABA, γ-aminobutyric acid. †To whom reprint requests should be addressed. E-mail: [email protected].



7094

(20), the involvement of AMPA receptors in response to ovarian steroid manipulations is not known. It remains to be determined whether the E-induced synapses are so-called “silent” synapses with only NMDA receptors (21) or ones that contain AMPA receptors as well. In contrast to the efficacy of NMDA receptor inhibition for synapse formation (see below), blockade of AMPA receptors with the antagonist NBQX during E treatment failed to block synaptogenesis (22).

Fig. 1. Camera lucida drawings of apical dendrites of CA1 pyramidal neurons from ovariectomized rats either untreated (A) or treated (B) with estradiol and progesterone to induce spines. Scale bar = 10 µm. [Reproduced with permission from ref. 29 (Copyright 1990, Society for Neuroscience)]. Besides increasing NMDA currents, reducing seizure thresh-olds, and enhancing long-term potentiation in hippocampus, E treatment exerts effects on hippocampal-dependent learning and memory. Three types of effects have been reported. First, in the natural estrous cycle of the rat, a recent study has used a delayed matching-to-place task in female rats to show a close parallel between the temporal conditions by which E improves memory and the conditions for E to induce new excitatory synaptic connections in the hippocampus (9). Second, E treatment of ovariectomized rats has been reported to improve acquisition on a radial maze task as well as in a reinforced T-maze alternation task (23, 24). Third, sustained E treatment is reported to improve performance in a working memory task (25), as well as in the radial arm maze (24, 26). The effects of E replacement in rats are reminiscent of the effects of E treatment in women whose E levels have been suppressed by a gonadotrophin-releasing hormone agonist used to shrink the size of fibroids before surgery (10, 27). EXCITATORY SYNAPSE FORMATION IN THE HIPPOCAMPUS E treatment increases dendritic spine density on CA1 pyramidal neurons (Figs. 1 and 2). As observed by electron microscopy, E also induces new synapses on spines and not on dendritic shafts of CA1 neurons (28). There were no E effects on dendritic length or branching (3, 28, 29). Progesterone (P) treatment acutely enhances spine formation (Fig. 1). But, over a 12- to 24-h period, P caused the down-regulation of E-induced synapses (29, 30), as will be discussed further below. Estrogens do not act alone, and, in fact, ongoing excitatory neurotransmission is required for synapse induction, as shown by the finding that antagonists of NMDA receptors block E-induced synaptogenesis on dendritic spines in ovariectomized female rats (ref. 22 and Fig. 3). Because E treatment increases the density of NMDA receptors in the CA1 region of the hippocampus (17, 31), the activation of NMDA receptors by glutamate may lead the way in causing new excitatory synapses to develop.

Fig. 2. Number of dendritic spines per 10 µm obtained from the apical portion of the CA1 pyramidal cell dendritic tree. Values are the mean ± SEM for estrogen and estrogen plus 5-h progesterone treatment. E induces increased spine density, an effect that is enhanced by 5-h progesterone. **, Different from other groups, P< 0.01; *, different from E+P group, P< 0.05. [Reproduced with permission from ref. 29 (Copyright 1990, Society for Neuroscience)]. Spines are occupied by asymmetric, excitatory synapses and are sites of Ca2+ ion accumulation and contain NMDA receptors (32). NMDA receptors are expressed in large amounts in CA1 pyramidal neurons and can be imaged by conventional immunocytochemistry as well as by confocal imaging, in which individual dendrites and spines can be studied for colocalization with other markers (33–35). Confocal microscopic imaging showed that E treatment up-regulates immunoreactivity for the largest NMDA receptor subunit, NR1, on dendrites and cell bodies of CA1 pyramidal neurons, whereas NR1 mRNA levels did not change after E treatment that induces new synapses (35), suggesting the possibility that NR1 expression is regulated posttranscriptionally by E (Fig. 4). NUCLEAR ESTROGEN RECEPTORS IN THE HIPPOCAMPUS Adult CA1 pyramidal cells appear to lack detectable nuclear ER as shown by tritium autoradiography (36) and light microscopic immunocytochemistry (4, 37), whereas they show low levels of ERα and -ß mRNA by in situ hybridization (38, 39). Autoradiographic mapping studies of [3H]estradiol uptake in hippocampus showed a sparse distribution of interneurons in the CA1 region, as well as other regions of Ammon’s horn that contain nuclear E binding sites (36). This observation was confirmed by immunocytochemistry for ERα in the guinea pig hippocampus (37) and subsequently in the rat hippocampus (ref. 4 and Fig. 5). These findings and those from cell culture studies described below led to a hypothesis regarding the role of interneurons as trans-synaptic regulators of synapse formation. Two other mechanisms will then be considered: (i) that E acts via a novel non-genomic mechanism; or (ii) that there are low levels of genomic ERs that are undetectable by conventional immunocytochemistry. CELL CULTURE MODEL OF SYNAPSE FORMATION Recent studies revealed that E induces spines on dendrites of dissociated hippocampal neurons in culture by a process that is blocked by an NMDA receptor antagonist and not by an AMPA/ kainate receptor blocker (40). In a subsequent study, E treatment was found to increase the phosphorylation of cAMP response



7095

element-binding protein (CREB), and a specific antisense to CREB prevented both the formation of dendritic spines and the elevation in phosphoCREB immunoreactivity (IR; ref. 41).

Fig. 3. Number of dendritic spines per 10 µm obtained from the apical portion of the CA1 pyramidal cell dendritic tree. Values are the mean ± SEM for treatment of ovariectomized rats with either vehicle or E in the presence or absence of the competitive NMDA receptor blocker, CGP 43 487. NMDA blockade prevents E induction of spines. *, Different from E alone, P < 0.01. [Reproduced with permission from ref. 22 (Copyright 1994, Society for Neuroscience)]. In agreement with the in vivo data (4), ERα was located in the cultures on glutamic acid decarboxylase (GAD)-immunoreactive cells that constituted around 20% of total neurons; E treatment caused decreases in GAD content and the number of neurons expressing GAD. Mimicking this decrease with an inhibitor of γ-aminobutyric acid (GABA) synthesis, mercaptopropionic acid, caused an up-regulation of dendritic spine density, paralleling the effects of E (42). An additional factor in the formation of dendritic spines in the cell culture model is the neurotrophin, brain-derived neurotrophic factor (BDNF; ref. 43). Besides down-regulating GABA in inhibitory interneurons, E treatment also reduced BDNF by 60% within 24 h (43). This neurotrophin appears to be a negative regulator of dendritic spines; exogenous BDNF blocks E induction of dendritic spines whereas BDNF depletion mimicks E in inducing spine density (43). Interestingly, neurotrophins such as BDNF and neurotrophin-3 (NT-3) also increase the function of inhibitory and excitatory synapses in hippocampal cell cultures; moreover, BDNF causes an increase in axonal branching and length of GABAergic interneurons (44). NON-NUCLEAR ESTROGEN RECEPTORS Besides exerting delayed and prolonged effects via nuclear receptors, estrogens can have rapid effects on hippocampal and other neurons, sometimes involving coupling to second messenger systems, such as the phosphorylation of CREB (12, 13, 45). Our recent findings have compelled us to consider such a mechanism in relation to hippocampal synapse formation. What is becoming evident is that, besides the indirect, transsynaptic mechanism described above, local signaling by E also may be involved. A seminal study using transfection of ERα and ERß into Chinese hamster ovarian cells (46) revealed that both ERs are expressed in a form that couples to second messenger systems that are stimulated by E and blocked at least partially by non-steroidal estrogen antagonists. Previous studies had indicated that non-nuclear ERs can be seen at the light microscopic level in cultured cells (47) and also at the electron microscopic level in hypothalamus (48).

Fig. 4. Bar graphs depicting NMDA subunit R1 immunofluorescence intensity measurements in the somata (Left) and dendrites (Right) of the CA field of the hippocampus. For somatic intensity measurements (Left), there is a significant increase when comparing E and E+P with OVX control. **, P< 0.0001. In dendritic fields (Right), E and E+P treatments were increased compared with OVX control; *, P

Molecular Kinesis in Cellular Function and Plasticity (NAS Colloquium)

(NAS Colloquium) Molecular Kinesis in Cellular Function and Plasticity

(Sackler NAS Colloquium) Nanoscience

(NAS Colloquium) Plants and Population

(Sackler NAS Colloquium) Nanoscience

(NAS Colloquium) Computational Biomolecular Science

(NAS Colloquium) Geology, Mineralogy, and Human Welfare

(NAS Colloquium) Proteolytic Processing and Physiological Regulation

(NAS Colloquium) Science, Technology and the Economy

(NAS Colloquium) Science, Technology and the Economy

Molecular and Cellular Signaling

Molecular and Cellular Biophysics

Molecular and Cellular Signaling

Cellular and Molecular Neurobiology,

Molecular and Cellular Biophysics

Cellular and Molecular Neurobiology

Molecular and Cellular Signaling

(NAS Colloquium) The Neurobiology of Pain

Molecular Biology in Cellular Pathology

Molecular Biology in Cellular Pathology

(NAS Colloquium) Virulence and Defense in Host--Pathogen Interactions (National Academy of Sciences Colloquium Series)

Cellular and Molecular Methods in Neuroscience Research

Brucella: Molecular and Cellular Biology

Molecular and Cellular Iron Transport

Advances in Molecular and Cellular Endocrinology

Molecular and cellular biology research in psychiatry

Techniques in Molecular and cellular Biology

Cellular and Molecular Methods in Neuroscience Research

Molecular and Cellular MR Imaging

Cellulose. Molecular and Cellular Biology

Cellular and Molecular Methods in Neuroscience Research

Molecular Kinesis in Cellular Function and Plasticity (NAS Colloquium)

(NAS Colloquium) Molecular Kinesis in Cellular Function and Plasticity

(Sackler NAS Colloquium) Nanoscience

(NAS Colloquium) Plants and Population

(Sackler NAS Colloquium) Nanoscience

(NAS Colloquium) Computational Biomolecular Science

(NAS Colloquium) Geology, Mineralogy, and Human Welfare

(NAS Colloquium) Proteolytic Processing and Physiological Regulation

(NAS Colloquium) Science, Technology and the Economy

(NAS Colloquium) Science, Technology and the Economy

Molecular and Cellular Signaling

Molecular and Cellular Biophysics

Molecular and Cellular Signaling

Cellular and Molecular Neurobiology,

Molecular and Cellular Biophysics

Cellular and Molecular Neurobiology

Molecular and Cellular Signaling

(NAS Colloquium) The Neurobiology of Pain

Molecular Biology in Cellular Pathology

Molecular Biology in Cellular Pathology

(NAS Colloquium) Virulence and Defense in Host--Pathogen Interactions (National Academy of Sciences Colloquium Series)

Cellular and Molecular Methods in Neuroscience Research

Brucella: Molecular and Cellular Biology

Molecular and Cellular Iron Transport

Advances in Molecular and Cellular Endocrinology

Molecular and cellular biology research in psychiatry

Techniques in Molecular and cellular Biology

Cellular and Molecular Methods in Neuroscience Research

Molecular and Cellular MR Imaging

Cellulose. Molecular and Cellular Biology

Cellular and Molecular Methods in Neuroscience Research

Recommend Documents