MEMBRANE PROTEIN TRANSPORT A Multi-Volume Volumes
•
Treatise 1996
This Page Intentionally Left Blank
MEMBRANE PRO...
51 downloads
1215 Views
18MB 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
MEMBRANE PROTEIN TRANSPORT A Multi-Volume Volumes
•
Treatise 1996
This Page Intentionally Left Blank
MEMBRANE PROTEIN TRANSPORT A Multi-Volume Treatise Editor:
STEPHEN S. ROTHMAN University of California San Francisco, California
VOLUMES
•
1996
i^n) Greenwich, Connecticut
\M PRESS INC. London, England
Copyright © 7 996 by JAI PRESS INC. 55 Old Post Road, No. 2 Greenwich, Connecticut 06836 JAI PRESS LTD. 38 Tavistock Street Covent Garden London WC2E 7PB England All rights reserved. No part of this publication may be reproduced, stored on a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, filming, recording, or otherwise, without prior permission in writing from the publisher. ISBN: 1-55938-989-3 Manufactured in the United States of America
CONTENTS
LIST OF CONTRIBUTORS THE NUCLEAR PORE COMPLEX: TOWARD ITS MOLECULAR ARCHITECTURE, STRUCTURE, A N D FUNCTION Nelly Pante and Ueli Aebi
vii
1
STRUCTURE A N D FUNCTION OF MITOCHONDRIAL PRESEQUENCES Merritt Maduke and Da vid Roise
49
BACTERIAL TOXIN TRANSPORT: THE HEMOLYSIN SYSTEM Jonathan A. Sheps, Fang Zhang, and Victor Ling
81
PROTEIN SORTING TO THE YEAST VACUOLE Bruce F. Horazdovsky, Jeffrey H. Stack, and Scoff D. Emr
119
BACTERIAL EXTRACELLULAR SECRETION: TRANSPORT OF a-LYTIC PROTEASE ACROSS THE OUTER MEMBRANE OF ESCHERICHIA COLI Amy Fujishige Boggs and David A. Agard
165
MECHANISMS OF PEROXISOME BIOGENESIS: REGULATION OF PEROXISOMAL ENZYMES, A N D THEIR SUBSEQUENT SORTING TO PEROXISOMES Gillian M. Small
181
PEROXISOMAL TOPOGENIC SIGNALS A N D THE ETIOLOGY OF PEROXISOME-DEFICIENT DISEASE Yukio Fujiki
213
vi ATP BINDING CASSETTE PROTEINS IN YEAST Carol Berkower and Susan Michaelis MEMBRANE PROTEIN TRANSPORT IN EUKARYOTIC SECRETION CELLS Kaarin K. Concz and Stephen S. Rothman INDEX
CONTENTS 2 31
279 295
LIST OF CONTRIBUTORS
Ueli Aebi
M.E. Muller Institute for Microscopy, Biozentrum University of Basel Basel, Switzerland
David A. Agard
Howard Hughes Medical Institute and the Departments of Pharmaceutical Chemistry and Biochemistry University of California, San Francisco San Francisco, California
Carol Berkower
Department of Cell Biology and Anatomy The Johns Hopkins University School of Medicine Baltimore, Maryland
Amy Fujishige Boggs
Howard Hughes Medical Institute and the Departments of Pharmaceutical Chemistry and Biochemistry University of California, San Francisco San Francisco, California
Scott D. Emr
Division of Cellular and Molecular Medicine University of California, San Diego School of Medicine and Howard Hughes Medical Institute La Jolla, California
Yukio Fujiki
Department of Biology Faculty of Science Kyushu University Fukuoka, Japan
Kaarin K, Goncz
Cardiovascular Research Institute School of Medicine and Dentistry University of California, San Francisco San Francisco, California VII
vm
LIST OF CONTRIBUTORS
Bruce Horazdovsky
Department of Biochemistry University of Texas Southwestern Medical Center Dallas, Texas
Victor Ling
Division of Molecular and Structural Biology The Ontario Center Institute Departmentof Medical Biophysics University of Toronto Toronto, Ontario, Canada
Susan Michaelis
Department of Cell Biology and Anatomy The Johns Hopkins University School of Medicine Baltimore, Maryland
Merritt Maduke
Department of Chemistry and Biochemistry University of California, San Diego La Jolla, California
Nelly Pante
Department of Cell Biology and Anatomy The Johns Hopkins University School of Medicine Baltimore, Maryland
David Roise
Department of Chemistry and Biochemistry University of California, San Diego La Jolla, California
Stephen S. Rothman
Departments of Physiology and Stomatology Schools of Medicine and Dentistry University of California, San Francisco San Francisco, California
Gillian M. Small
Department of Cell Biology and Anatomy Mount Sinai School of Medicine New York, New York
Jonathan A. Sheps
Division of Molecular and Structural Biology The Ontario Center Institute Department of Medical Biophysics University of Toronto Toronto, Ontario, Canada
Jeffrey H. Stack
Division of Cellular and Molecular Medicine University of California, San Diego School of Medicine and Howard Hughes Medical Institute La Jolla, California
L ist of Contributors Fang Zhang
I Division of Molecular and Structural Biology The Ontario Cancer Institute Department of Medical Biophysics University of Toronto Toronto, Ontario, Canada
This Page Intentionally Left Blank
THE NUCLEAR PORE COMPLEX: TOWARD ITS MOLECULAR ARCHITECTURE, STRUCTURE, AND FUNCTION
Nelly Pante and Ueli Aebi
Abstract I. Introduction II. The Nuclear Envelope III. Structure of the Nuclear Pore Complex A. STEM Mass Analysis B. Structure of the Basic Framework of the NPC C. Cytoplasmic and Nuclear Ring D. Cytoplasmic Filaments and Nuclear Basket E. The Central Plug or Channel Complex F. A Consensus Model for the NPC IV. Toward a Molecular Architecture of the NPC A. Integral Membrane Proteins of the NPC B. Peripheral Membrane Proteins of the NPC C. Yeast NPC Proteins D. IsolationandCharacterizationof Distinct NPC Components V. Molecular Trafficking Through the NPC A. Passive Diffusion
Membrane Protein Ti-ansport Volume 3, pages 1-47. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 1
2 2 3 5 8 9 11 12 15 18 18 19 24 31 35 36 36
NELLY PANTE and UELI AEBI
B. Import of Nuclear Proteins C. Export/Import ofRNAs and RNP Particles V. Conclusions and Future Prospects Acknowledgments References
36 38 40 42 42
ABSTRACT The nuclear pore complex (NPC) is a -120 MDa macromolecular assembly embedded in the double-membraned nuclear envelope (NE) that mediates bidirectional molecular trafficking between the cytoplasm and the nucleus of interphase cells. Electron microscopy (EM) of negatively stained and frozen-hydrated specimens combined with three-dimensional (3-D) reconstruction has yielded the architecture of the basic framework of the NPC, and a number of specimen preparation methods and microscopy techniques have revealed the structure of distinct peripheral NPC components such as the cytoplasmic and nuclear ring, the cytoplasmic filaments, and the nuclear basket. Recently, significant progress has been made toward identifying, characterizing, and cloning and sequencing NPC proteins. However, to date only about two dozen NPC proteins, called nucleoporins, have been identified, a number accounting for less than 15% of the NPC mass. Some of these nucleoporins have been localized within the 3-D structure of the NPC; thus the molecular architecture of the NPC is starting to emerge. Nevertheless, the 3-D structure and functional significance of the different structural components and proteins of the NPC remain to be established. Relatively little is known about the molecular mechanisms underlying nucleocytoplasmic transport through the NPC. However, over the past few years there have been some advances in establishing the signals, receptors, and factors mediating nuclear import of proteins and snRNPparticles> and nuclear export ofRNAs and RNP particles. These functional studies have indicated that there may exist multiple signaling pathways for these different processes and substrates. Here we review recent advances toward the 3-D structure and molecular architecture of the NPC, and the molecular basis of nucleocytoplasmic transport of proteins, RNAs, and RNP particles through the NPC.
I. INTRODUCTION In eukaryotic cells the nucleus is separated from the cytoplasm by the nuclear envelope (NE), which enables each compartment to have its own distinctive composition and function. Beside separating the genetic machinery from protein synthesis, the NE allows nucleocytoplasmic trafficking; most of the RNA synthesized in the nucleus is transported to the cytoplasm where it is used for protein synthesis, while proteins required for nuclear function are synthesized in the cytoplasm and imported into the nucleus. This bidirectional exchange of material between the nucleus and the cytoplasm occurs through the nuclear pore complexes
Nuclear Pore Complex
3
(NPCs), large macromolecular assemblies that span the NE. The NPC allows passive diffusion of ions and small molecules, but molecules larger than 9 nm in diameter do not traverse the NPC freely. These are selectively imported into the nucleus by a signal-requiring and ATP-dependent mechanism. Macromolecular assemblies such as ribonucleoprotein (RNP) particles containing mRNA are also selectively exported from the nucleus by active transport. Despite significant progress in understanding the structure and function of the NPC, the molecular mechanism of nucleocytoplasmic transport through the NPC has remained elusive. Here we review recent advances made toward the elucidation of the architecture and biochemical composition of the NPC, and recent progress made toward the understanding of the mechanism of mediated transport through the NPC.
II. THE NUCLEAR ENVELOPE As illustrated schematically in Figure 1, the nuclear envelope (NE) is made up of a double membrane enclosing a lumen, the "perinuclear space." The outer nuclear membrane faces the cytoplasm and is continuous with the endoplasmic reticulum (ER). Thus the perinuclear space is contiguous with the lumen of the ER. On its cytoplasmic surface the outer nuclear membrane is often studded with ribosomes, as is the rough ER. The inner nuclear membrane faces the nucleoplasm and in its nucleoplasmic surface is lined by the nuclear lamina, a polymer of intermediate filament-like proteins, the nuclear lamins (Aebi et al., 1986; Gerace and Burke, 1988). The nuclear lamina provides a general framework for NE structure and is an anchoring site for interphase chromatin. The NPCs are interposed at irregular intervals between the inner and outer nuclear membranes, where the two membranes are fused to form the "pore membrane." The density of NPCs within the NE varies among different cell types, but it is roughly proportional to the metabolic activity of the cell (Maul, 1977): it ranges between 2-4 pores/|im^ for lymphocytes and 50-60 pores/|Lim^ for maXurQ Xenopus laevis oocytes. Because of the high density of NPCs and the ease of manually isolating its NE, amphibian oocytes have been extensively used in structural studies of the NPC (see below). However, due to the lack of a reproducible bulk isolation procedure, this preparation is not of practical use for biochemical studies: about 100 nuclei have to be isolated to obtain enough material to reveal most of the NPC proteins on a SDS-polyacrylamide gel. Therefore, most of the biochemical information about the NE has been obtained from rat liver preparations. These preparations yield a fraction enriched in NPCs attached to the nuclear lamina that has been successfully used for the identification and isolation of several NPC proteins (see Section IV) and for raising antibodies directed against a number of NPC proteins (Davis and Blobel, 1986; Snow et al., 1987). During cell division the NE disassembles and reassembles in a strictly coordinated manner (reviewed by Wiese and Wilson, 1993). Membrane fragmentation of the NE occurs after chromatin condensation at prometaphase when the NPCs
NELLY PANTE and UELI AEBI Outer nuclear membrane Perinuclear space Inner nuclear membrane
Nuclear pore complex Endoplj retlj
Cytoplasm Ions, small molecules
Passive
Proteins, snRNPs RNAs/RNPs
diffusion
ATP-dependent mediated transport
Figure 1. Schematic representation of the different components of the NE together with the molecular trafficking occurring across the nuclear envelope (NE) through the nuclear pore complexes (NPCs). The NE consists of an inner and an outer nuclear membrane enclosing the perinuclear space. The outer nuclear membrane is continuous with the endoplasmic reticulum (ER), so that the perinuclear space of the NE is contiguous with the lumen of the ER. The inner nuclear membrane is lined by the nuclear lamina (NL), a near-tetragonal meshwork made of intermediate filament-like proteins called nuclear lamins. The NPCs are interposed at irregular intervals between the inner and outer nuclear membrane. Bidirectional molecular trafficking between the nucleus and the cytoplasm occurs through the NPCs, which allow passive diffusion of ions and small molecules, and ATP-mediated transport of proteins, RNAs, and RNPs.
completely disappear and the nuclear lamina depolymerizes. Reassembly of the NE occurs during telophase when the membrane fragments aggregate at the chromosome surfaces and intact NPCs start to reappear, even on small pieces of NEs (Roos, 1973). Little is known about the mechanism of fragmentation and reformation of the NE, and the disassembly and reassembly of the NPCs during mitosis. Immunofluorescence microscopy with antibodies against NPC proteins has shown that some NPC proteins disperse throughout the cytoplasm during mitosis and become progressively concentrated around the periphery of the chromosomes in late anaphase and early telophase (Snow et al., 1987). The incorporation of NPC proteins at the reforming NE seems to be a stepwise process in which assembly of
Nuclear Pore Complex
5
protein constituents of the NPC proper precedes assembly of the peripheral components of the NPC (Byrd et al., 1994). Some aspects of nuclear lamina depolymerization and repolymerization during cell division have been well established (reviewed by Gerace and Burke, 1988). This process is regulated by specific phosphorylation-dephosphorylation of the nuclear lamins: during mitosis when the nuclear lamina depolymerizes, the nuclear lamins are phosphorylated, and during telophase when the nuclear lamina reassembles, they are dephosphorylated (Gerace and Blobel, 1980; Ottaviano and Gerace, 1985). It has been shown that phosphorylation of the lamins by the cell-cycle-specific cdc2 kinase induces lamina disassembly in vitro (Peter et al., 1990; Dessev et al., 1991). The mitotic phosphorylation sites have been mapped at either end of the lamin molecule (Ward and Kirschner, 1990; Peter et al., 1991), and phosphorylation of these sites results in the disassembly of in vitro formed lamin head-to-tail polymers (Peter et al., 1991).
III. STRUCTURE OF THE NUCLEAR PORE COMPLEX The structure of the NPC has been extensively investigated by different electron microscopy (EM) specimen preparation methods and imaging techniques (reviewed by Pante and Aebi, 1993, 1994), including scanning force microscopy (SFM) in a physiological buffered environment (Pante and Aebi, 1993; Goldie et al., 1994). As documented in Figure 2, a and b, when intact Xenopus oocyte NEs are spread on an EM grid and viewed in a transmission EM after negative staining, extensive arrays of NPCs are revealed. In these images the NPCs appear as round particles with a diameter of-125 nm (Unwin and Milligan, 1982; Reichelt et al., 1990; Jamik and Aebi, 1991). Depending on the face of the NE that is adsorbed to the EM support film, the NPCs reveal different morphologies: if the NE adsorbs to the EM film with its cytoplasmic face, so that its nuclear face is exposed, the NPCs appear irregularly stained and poorly preserved (see Fig. 2a); whereas when the cytoplasmic face is exposed, the NPCs appear well preserved and exhibit a distinct eightfold rotational symmetry (Fig. 2b). In cross sections of Epon-embedded intact Xenopus oocytes or Xenopus oocyte nuclei (Fig. 2c), the NPCs appear to span the NE (i.e., '^SO nm), and they reveal filamentous structures associated with both its cytoplasmic and nuclear periphery (see Fig. 2c, arrowheads). Thus both negatively stained and embedded/thin-sectioned NPCs clearly reveal that the NPC is a structure with eightfold rotational symmetry, but its cytoplasmic and nuclear periphery exhibit a high degree of asymmetry. EM images of negatively stained and embedded/thin-sectioned NEs (i.e., Fig. 2, a-c) reveal the morphology of the NPC in the presence of the nuclear double membrane, which is probably essential for stabilization of the structure of the NPC. However, information on the different structural components of the NPC comes from studies where the nuclear membranes have been solubilized. As documented in Figure 2d, when NEs are exposed to treatment with nondenaturing detergents.
Figure 2. Xenopus oocyte nuclear envelope (NE) after different preparations for EM and different chemical treatments, (a) Nuclear and (b) cytoplasmic face of negatively stained intact NEs. Depending on which face of the NE adsorbed to the EM grid, the NPCs revealed a different morphology, thus revealing the highly asymmetric architecture of the NPC with regard to its nuclear (a) and cytoplasmic (b) periphery, (c) Cross section of Epon-embedded NE revealing both cytoplasmic (large arrowheads) and nuclear (small arrowheads) filamentous structures associated with the NPC periphery. (d) Negatively stained NPCs after detergent treatment of a spread NE. Whereas intact NPCs (a-c) are embedded in the NE, the detergent-treated NPCs in (d) have detached from the nuclear envelope, revealing the basic framework of the NPC consisting of the plug-spoke complex. Scale bar, 100 nm (a-d).
Nuclear Pore Complex
7
the basic framework of the NPC is yielded more clearly. It consists of eight "spokes" embracing a "central pore," which is sometimes "plugged" with a "central channel complex" (also called "central plug" or "transporter"). Recently, the architecture of the basic framework of the NPC has been determined by 3-D reconstruction of both negatively stained (Hinshaw et al., 1992) and frozen hydrated (Akey and Radermacher, 1993) detergent-treated NPCs. These studies are discussed in more detail in Section IIIB. As documented in Figure 3, besides revealing the basic framework (i.e., the spoke complex), detergent treatment also releases two types of rings: (i) cytoplasmic rings, which are predominantly positively stained and are yielded by rolling intact nuclei on an EM grid (Jamik and Aebi, 1991), and (ii) nuclear rings, which are negatively stained and appear less massive than the cytoplasmic rings, both by comparison of their radial mass density profiles (Fig. 3, c and d) and from direct mass analysis (discussed in Section IIIA). Both the cytoplasmic and nuclear rings
Figure 3. Negatively stained intact NPCs and distinct NPC components. Electron micrographs, correlation averages (always including 20 eight-fold rotationally symmetrized NPCs; upper insets), and radial mass density profiles (lower insets) of intact NPCs (a), and distinct NPC components yielded after treatment of spread NEs with 0.1% Triton X-100 (b-d). (a) Membrane-bound, intact NPCs; (b) plug-spoke complexes; (c) cytoplasmic rings, which are also left behind after rolling intact nuclei on a carbon film-coated EM grid; and (d) nuclear rings. Scale bar, 100 nm (a-d). Adapted from Jarnik and Aebi (1991).
8
NELLY PANTE and UELI AEBI
contain associated filamentous structures that were first observed in embedded/thin-sectioned NEs but were believed to represent a specimen preparation artifact since they were not depicted by other EM preparation techniques (Franke and Scheer, 1970a,b; Franke, 1974). Recently, the use of more elaborate EM specimen preparations and imaging techniques has further documented the existence of these peripheral filamentous components of the NPC (Jamik and Aebi, 1991; Ris, 1991; Goldberg and Allen, 1992; Goldie et al, 1994). As will be discussed in more detail in Section HID, these studies have revealed that the cytoplasmic ring is decorated with eight short cytoplasmic filaments, whereas the nuclear ring is topped by a mechanically fragile "baskef formed by eight thin filaments joined distally by a terminal ring. In summary, the NPC is composed of (i) a basic framework made of eight spokes embracing a central pore, (ii) a central plug or channel complex, (iii) a cytoplasmic ring from which eight short, kinky filaments emanate, and (iv) a nuclear ring with a basket-like assembly attached. In this NPC architecture the basic framework is sandwiched between the cytoplasmic and nuclear ring. In the following sections, we first present mass measurements of the entire NPC and its different components, before we discuss various structural aspects of the different NPC components in greater detail. A. STEM Mass Analysis
Not only does the NPC have large dimensions (^125 nm in diameter and -70 nm high), but it also harbors an enormous mass. Quantitative scanning transmission EM (STEM) has revealed for the intact NPC a total mass of 124 MDa (i.e., 124 x 10^ Da; Reichelt et al., 1990). The membrane-bound NPC without the central channel complex has a mass of 112 MDa; thus the central plug is -^12 MDa. These authors also measured the mass of the different NPC components obtained after detergent treatment of NEs. Accordingly, as summarized in Table 1, the basic framework of the NPC as reconstructed in 3-D by Hinshaw et al. (1992) and Akey
Table 1. Masses of the Intact NPC and Its Major Structural Components Mass (MDaf Nuclear pore complex Basic framework Central channel complex Cytoplasmic ring Nuclear ring
124.0 ±11.0 51.7 ± 5.3 12.0 ± 1.1 32.0 ± 5.5 21.1 ± 3.7
Note: ^Determined by quantitative scanning transmission EM (STEM) and adapted from Reichelt et al. (1990).
Nuclear Pore Complex
9
and Radermacher (1993) (see Section IIIB) has a mass of 52 MDa without the central channel complex. Two types of rings were also measured in detergenttreated NEs: heavy (32 MDa) and light (21 MDa). Since the 32-MDa rings were also observed as "footprints" when nuclei were rolled back and forth on charged EM grids, they must represent the cytoplasmic rings with the remnants of one to several collapsed cytoplasmic filaments attached (see Section HID). The 21-MDa nuclear rings frequently revealed mass in the center (see Section HID), which due to its variation among rings was excluded from the mass measurements. Thus the sum of the mass of its principal components (i.e., the basic framework, the central plug, and the cytoplasmic and nuclear ring) yields a total mass of 117 MDa (i.e., 52+12 + 32 + 21 MDa) for the intact NPC. The small difference (-6%) between the measured mass of the intact membrane-bound NPC and the sum of its components may be due to loss of some of its peripheral components (i.e., the cytoplasmic filaments and nuclear baskets) during detergent treatment and/or specimen preparation. B. Structure of the Basic Framework of the NPC
The architecture of the basic framework of the NPC (i.e., the spoke complex) has recently been revealed in 3-D reconstructions of both negatively stained (Hinshaw et al., 1992) and frozen hydrated (Akey and Radermacher, 1993) NPCs obtained after detergent treatment. As documented in Figure 4a, the 3-D mass map of negatively stained, detergent-treated NPCs is built of eight multidomain spokes, with each spoke consisting of two identical halves. Hence the entire spoke complex yields 8-2-2 symmetry with one half-spoke representing the asymmetric unit. Each half-spoke is built from four distinct morphological domains termed the "annular," "column," "ring," and "lumenal" domains (see Fig. 4b). The lumenal domain extends into the lumen of the NE and is believed to contain as one of its constituents the glycosylated NPC protein gp210 (see Section IVA). Since the basic framework of the NPC as shown in Figure 4a has a mass of 52 MDa (Reichelt et al., 1990; see also Table 1), the mass of one half-spoke is ~3.3 MDa, that is, it is on the order of a ribosome. When the pore membrane is positioned in the negatively stained detergent-released NPC map, eight ~10-nm diameter "peripheral channels" are created between two adjacent spokes and the pore membrane border at a radius of -40 nm from the NPC center (see Fig. 4a). As speculated by Hinshaw et al. (1992), these peripheral channels may represent sites for passive diffusion of ions and small molecules, and they may also facilitate import of inner nuclear membrane proteins (SouUan and Worman, 1993; reviewed by Wiese and Wilson, 1993). The 3-D reconstruction of frozen hydrated detergent-treated NPCs (Akey and Radermacher, 1993) produced a similar 3-D map of the basic framework of the NPC. However, these authors included the central channel complex in their reconstruction. Thus in their 3-D map the spoke complex embraces an elaborate, barrel-like central channel complex. Also by including the central channel complex,
10
NELLY PANTE and UELIAEBI
this reconstruction yields eight internal channels between two adjacent spokes and the central channel complex. However, since this tomographic reconstruction is based on a relatively large missing tilt cone of information that affects the final 3-D map predominantly at low radii, the central mass of their 3-D model may have been peripheral nel
leral
channel
Figure 4. Surface renderings of the 52-MDa basic framework of negatively stained NPCs released from the NE upon detergent treatment, (a) Slightly tilted view of the basic framework of the NPC, which consists of eight multidomain spokes and exhibits strong 8-2-2 symmetry thus indicating that it is built of two identical halves relative to the central plane of the nuclear envelope. Note that because of its irreproducible appearance the central plug or channel complex has been omitted in this reconstruction. When the pore membrane is positioned in the map of negatively stained detergent-released NPC, eight ~10-nm diameter "peripheral channels" are created between two adjacent spokes and the pore membrane border at a radius of - 4 0 nm. (b) Three different side views of one multidomain spoke cut out from the basic framework of the NPC as shown in (a). Each half-spoke is built from four distinct morphological domains, annular (a), column (c), ring (r), and lumenal (I). Adapted from H i n s h a w e t a l . (1992).
Nuclear Pore Complex
11
overestimated despite application of a solvent flattening procedure to the reconstruction (Akey and Radermacher, 1993). In additipn, as we will discuss in Section HIE, other NPC components (i.e., the nuclear basket) may contribute to what in ice-embedded NPC images appears as the central plug. Moreover, since the abundance and morphology of the central channel complex depend on the isolation and preparation conditions employed (see Section IIIE), it is necessary to develop a procedure to control the reproducible appearance of the central plug among NPCs before computing a 3-D map of an irreproducible structure. C. Cytoplasmic and Nuclear Ring
The basic framework of the NPC (Fig. 4a) is sandwiched between a cytoplasmic and a nuclear ring. Both cytoplasmic (Fig. 2b) and nuclear (Fig. 2a) views of negatively stained intact NPCs revealed a ring at the periphery of the NPC. These rings readily detach from the NPC proper upon chemical or physical manipulation of the NE: (i) Xenopus oocyte NEs extracted with Triton X-100 revealed distinct rings lying next to spoke complexes (Unwin and Milligan, 1982; Jamik and Aebi, 1991; Hinshaw et al., 1992); (ii) thin rings are observed in osmotically shocked NEs (Akey, 1989); and (iii) rolling of isolated Xenopus oocyte nuclei on an EM grid (without spreading the NE on the grid) leaves cytoplasmic rings (so-called footprints) behind (Jamik and Aebi, 1991). Projection maps of correlation averaged NPCs and their corresponding radial density profiles clearly demonstrated that after such treatments the remaining NPCs had lost material at their periphery (compare Fig. 3a with Fig. 3b; see also Jamik and Aebi, 1991). Moreover, STEM mass measurements of the rings released after detergent treatment revealed the existence of two types: 32-MDa and 21-MDa rings (Reichelt et al., 1990; see Section IIIA and Table 1). Because the footprints left behind after rolling isolated nuclei on EM grids (see above) also revealed a mass of 32 MDa, it was concluded that the 32-MDa rings represented cytoplasmic rings, whereas the 21 -MDa rings represented nuclear rings. Thus although the cytoplasmic and nuclear rings have a similar appearance, they are different. The difference between these two rings has also been demonstrated by limited proteolysis (Goldberg and Allen, 1993). Accordingly, tryptic digestion of NPCs sequentially removes subunits of both rings, but the nuclear rings are more sensitive to proteolysis than are the cytoplasmic rings. The above described studies have clearly documented the existence of distinct cytoplasmic and nuclear rings. In contrast, it has been argued that these rings are an integral part of the basic framework of the NPC (Hinshaw et al., 1992; Akey and Radermacher, 1993). In the 3-D map of the basic framework the ring domain of the multidomain spokes defines two tenuous rings at the cytoplasmic and nuclear faces of the NPC (Fig. 4). Since according to these authors the basic framework of the NPC exhibits good 8-2-2 symmetry, these two tenuous rings should be identical or at least very similar. In contrast, the cytoplasmic and nuclear rings released upon detergent treatment of Xenopus oocyte NE preparations yield distinct masses (see
12
NELLY PANTE and UELI AEBI
above and Table 1; Reichelt et aL, 1990). Moreover, the basic framework of the NPC reconstructed from negatively stained NPCs after detergent treatment (Hinshaw et al., 1992) has a mass of 52 MDa without the central plug. Hence there must be additional components associated with this basic framework to account for the --110-MDa mass of the intact, unplugged NPC (Reichelt et al., 1990). Interestingly, the masses of the 32-MDa cytoplasmic ring and 21-MDa nuclear ring amount to just about what has to be added to the 52-MDa mass of the basic framework (i.e., 105 MDa) to arrive at a mass of approximately 110 MDa for the intact, unplugged NPC. Therefore, it is conceivable that the cytoplasmic and nuclear rings are not (or only partially) represented in the 3-D mass density maps of the basic framework of detergent-released NPCs (see Fig. 4a). D. Cytoplasmic Filaments and Nuclear Basket Recently, several independent investigations have confirmed the existence of filamentous structures associated with both the cytoplasmic ring and the nuclear ring of the NPC (Jamik and Aebi, 1991; Ris, 1991; Goldberg and Allen, 1992; Goldie et al., 1994). As illustrated in Figure 5, a and b, v^h^n Xenopus oocyte NEs are visualized in the EM after quick freezing/freeze drying/rotary metal shadowing, the cytoplasmic face of the NPC looks distinct from the nuclear face (see also Jamik and Aebi, 1991). Accordingly, the cytoplasmic face of the NPC is topped with a 100-110-nm outer diameter ring, from which eight kinky filaments protrude (see Fig. 5a, arrowheads), which have a tendency to collapse into themselves and thus often appear as short cylinders or "cigars." The presence of these cytoplasmic filaments is best documented in situations where they have bent to the side and adhered to filaments of adjacent NPCs, thus appearing as NPC connecting fibrils (see Fig. 5a, small arrows). As shown in Figure 5b, the nuclear face of the NPC accommodates a more tenuous, 90-100-nm outer diameter ring from which eight thin, 50-100-nm long filaments emanate and are joined distally by a 30-50-nm diameter terminal ring, thus forming a "basket" or "fishtrap." These cytoplasmic filaments and nuclear baskets make the NPC distinctly asymmetric relative to the plane oftheNE. High-resolution scanning EM of critical point-dried/metal-sputtered isolated NEs have revealed similar structures (Ris, 1991; Goldberg and Allen, 1992). In addition, in Triturus cristatus this technique has depicted the existence of an ordered fibrous nuclear lattice, termed the "NE lattice" or NEL, that is connected to the nuclear baskets via their terminal rings (Goldberg and Allen, 1992). The chemical composition and function of this NEL remain to be established. Remnants of such a lattice or filament system connecting adjacent baskets has also been observed in Xenopus oocyte NEs in the form of basket connecting filaments (Jamik and Aebi, 1991; Ris, 1991). More recently, the native cytoplasmic and nuclear NPC topography has been visualized by scanning force microscopy (SFM) of spread Xenopus oocyte NEs
Nuclear Pore Complex
13 ■^JTT
b # :
Figure 5. The NPC Is highly asymmetric with regard to its cytoplasmic and nuclear periphery. The cytoplasmic (a and c) and nuclear (b and d) face of NPCs revealed by transmission EM of quick-frozen/freeze-dried/rotary metal-shadowed intact Xenopus oocyte NEs (a and b) and by scanning force microscopy (SFM) of intact Xenopus oocyte NEs kept In physiological buffer (c and d). Relatively short cytoplasmic filaments (a; arrowheads), and nuclear baskets (b) are revealed by quick freezing/freeze drying/rotary metal shadowing. The resolution of the SFM images (c and d) is insufficient to resolve individual NPC-associated filaments; thus the cytoplasmic face of the NPC appears "donut-IIke" (c), whereas the nuclear face exhibits a "dome-like" appearance (d). The arrowheads In (a) point to cytoplasmic filaments protruding from the cytoplasmic ring of the NPC, whereas the small arrows in (a) mark filaments that have bent to the side and thereby adhered to filaments of adjacent NPCs, thus appearing as "NPC connecting fibrils." Scale bar, 100 nm (a-d).
kept in physiological buffer (Pante and Aebi, 1993; Goldie et al., 1994). In agreement with the results of dehydrated specimens (Jamik and Aebi 1991; Ris, 1991; Goldberg and Allen, 1992), corresponding SFM topographs revealed a high degree of asymmetry betw^een the nuclear and cytoplasmic periphery of the NPC. As documented in Figure 5, c and d, by SFM the cytoplasmic face of the NPC appears "donut-like," whereas the nuclear face exhibits a "dome-like" appearance. However, since the resolution in these SFM images is insufficient to resolve individual NPC-associated filaments, the in vivo conformation of these cytoplasmic
14
NELLY PANTE and UELI AEBI
and nuclear filaments has remained uncertain. For example, the cytoplasmic filaments have been described as "granules" (Stewart and Whytock, 1988), "short cylinders" (Jamik and Aebi, 1991), and "T-shaped" particles (Goldberg and Allen, 1992). At this stage it is difficult to determine which of these conformations, if any, represents the native one or to which extent these different conformations may merely represent preparation artifacts. The integrity of the nuclear baskets critically depends on divalent cations. In the presence of 0.5 mM MgCl2 or CaCl2, well-formed baskets are observed. In contrast, as documented in Figure 6a (see also Jamik and Aebi, 1991), if the divalent cations
Figure 6. The structural integrity of the nuclear baskets depends on divalent cations. (a) Nuclear face of a quick-frozen/freeze-dried/rotary metal-shadowed Xenopus oocyte NE depleted of divalent cations with 2 mM EDTA. (b) Nuclear face of a Xenopus oocyte NE after treatment with 2 mM EDTA prior to quick freezing/freeze drying/rotary metal shadowing. Depletion of divalent cations with 2 mM EDTA destabilizes the terminal rings and thereby causes disassembly of the nuclear baskets (a), whereas the addition of divalent cations after destabilization with EDTA causes reformation of the nuclear baskets (b). (c, d) Electron micrographs, correlation averages (always including 20 eight-fold rotationally symmetrized NPCs; upper insets), and radial mass density profiles (lower insets) of negatively stained intact NPCs after treatment with 2 mM EDTA (c) or in the presence of 0.5 mM MgCl2 (d). The mass density profile about the center of the NPC is significantly attenuated in the presence of 2 mM EDTA, a condition that causes disassembly of the nuclear baskets (see a). Scale bar, 100 nm (a-d). Adapted from Jamik and Aebi (1991).
Nuclear Pore Complex
15
are chelated by 2 mM EDTA or EGTA, the nuclear baskets become destabilized and are disrupted. If exogenous ATP is present during isolation and preparation of NEs, a similar effect is observed (our own unpublished results). Surprisingly, if after destabilization by EDTA or EGTA divalent cations are reintroduced, the nuclear baskets reform (see Fig. 6b). These results indicate that the nuclear baskets are dynamic structures in that they disassemble upon removal of divalent cations and reassemble upon their addition (Jamik and Aebi, 1991) and therefore could be directly involved in the active transport of proteins, RN As, or RNP particles through the NFC. As to their possible functional role, the cytoplasmic filaments have been implicated (i) as docking sites for import into the nucleus through the NPC and (ii) in delivering the docked material to the central channel complex for active translocation (Gerace, 1992). Some indirect evidence that is consistent with such a scenario is the observation that the cytoplasmic filaments sometimes bend around so that their distal end reaches down into the central pore (see Fig. 12, a and c, and Fig. 13, arrowheads). E. The Central Plug or Channel Complex
Often a massive, ~12-MDa particle, which has been termed the "central plug," "central channel complex," or "transporter," is depicted in the central pore of the NPC (see Fig. 2, b and d). The frequent lack of this central structure and its highly polymorphic appearance within a given NPC population has given rise to suggestions that at least in part it may represent particles (i.e., RNP particles) in transit rather than an integral component of the NPC (e.g., Jamik and Aebi, 1991; Gerace, 1992). Nevertheless, based on a computer analysis of several thousand NPCs from frozen-hydrated NEs, Akey (1990) has classified this central structure into four different groups that are related to different transport states of the NPC and termed "closed," "docked," "open/in transit," and "open." Based on this computer classification, Akey (1990) has proposed this central structure to represent the actual "transporter" and modeled it as a double-iris arrangement that can assume several distinct configurations as it actively transports molecules and particles through the NPC. Since, depending on the isolation and/or specimen preparation of the NEs, both the abundance and appearance of this central plug are highly variable (Unwin andMilligan, 1982;Reicheltetal., 1990; Jamik and Aebi, 1991), it is far from clear whether the different morphologies presented by Akey (1990) do indeed represent distinct transport-related states. Thus the extent to which this central structure is an integral component of the NPC, or whether it represents at least in part material in transit remains to be determined. Moreover, to establish the functional significance of the different morphologies of this central structure identified by Akey (1990), it is necessary to directly correlate these with corresponding transport assays. In view of the recently described peripheral components of the NPC (see Section HID and Fig. 5), it is also conceivable that a substantial fraction of what in proj ection
16
NELLY PANTE and UELI AEBI
appears as the central plug in fact represents remnants of the nuclear basket (including the terminal ring), which may have been squashed into the pore upon embedding the NPC in a thin layer of negative stain or a thin ice film, or material associated with it. In support of this notion, the mass density profile of negatively stained NPCs in the presence of EDTA (Fig. 6c), a condition that causes disassembly of the nuclear baskets (see above; Jamik and Aebi, 1991), is significantly attenuated when compared with that of NPCs isolated in the presence of millimolar amounts of divalent cations (Fig. 6d). Thus to more systematically investigate the nature and 3-D structure of the central plug or channel complex, it is first necessary to develop a method that yields a reproducible presence and appearance of this NPC component, and at the same time control the structural integrity of the nuclear baskets. Toward achieving this goal, we have been exploring a number of different buffers, incubation conditions, and chemical fixation protocols. As illustrated in Figure 7, we have found that 95% of the NPCs harbor a massive central plug when during isolation the NEs are stabilized with Cu-orthophenanthroline, an oxidizing agent causing S-S bridge formation.
Figure 7. Chemical manipulation of isolated NEs changes both the abundance and appearance of the central plug or channel complex, (a, b) Negatively stained Xenopus oocyte NEs that have been isolated in the presence of 0.1 mM Cu-orthophenanthroline, an oxidizing agent causing intra- and intermolecular S-S bridge formation, (c) Control, prepared as in (a) but in the absence of Cu-orthophenanthroline. Accordingly, --95% of the NPCs yield a massive central plug when they are stabilized with Cu-orthophenanthroline during isolation. Scale bar, 200 nm (a) and 100 nm (b-c).
Nuclear Pore Complex
17
Cytoplasmic fliumcnts
r
Outer membrane l.umenal domain
Inner membrane \
Nuclear ring
Nuclear basket
^
Terminal ring
Figure 8. Schematic representation of the consensus model of the membrane-bound NPC. Its major structural components include the basic framework (i.e., the spoke complex as shown in Figure 4a), the central plug or channel complex, the cytoplasmic and nuclear rings, and the cytoplasmic filaments and nuclear basket. The 52-MDa basic framework of the NPC has been adapted from the 3-D reconstruction of negatively stained detergent-released NPCs (Hinshaw et al., 1992). The cytoplasmic filaments and nuclear basket have been modeled based on EM data obtained by Ris (1991), Jarnik and Aebi (1991), and Goldberg and Allen (1992) (see also Figure 5, a and b). In this consensus model of the NPC, we have also pictured a cytoplasmic and nuclear ring in addition to the two tenuous rings defined by the ring domain of the spoke complex (see Figure 4a). The central plug or channel complex has been modeled as a transparent ellipsoidal particle to indicate the fact that its definite structure remains to be determined.
Recently, the central plug as it appears in frozen-hydrated NPCs after detergent treatment has been reconstructed in 3-D (Akey and Radermacher, 1993). Accordingly, it has a rather complex, hourglass-like structure '-62-nm long with distal and central diameters of-42 and ~32 nm, respectively. However, as we have discussed in Section IIIB, reconstruction errors due to sampling and limited tilt range accumulate predominantly at low radii (i.e., in the center of the 3-D mass map); thus the central mass in this 3-D reconstruction may be overestimated. On the other
18
NELLY PANTE and UELI AEBI
hand, this representation of the central plug is not entirely consistent with the double-iris-like model previously proposed by Akey (1990). Based on the similarity of its overall size and shape, octagonal symmetry, and mass, it has been proposed that the central plug may represent a "vault" ribonucleoprotein particle (Kedersha et al., 1991), with the vaults acting as transport vehicles, that is, carrying nuclear proteins as their cargo. Recent evidence for vault immunoreactivity at the NPC (Chugani et al., 1993) supports this idea. However, a more specific and comprehensive characterization of the central plug in terms of its molecular structure and, association with the basic framework of the NPC still has to be achieved before it can be identified with any known particle such as vaults. Similarly, any such candidate particles must be subjected to a more stringent analysis regarding their possible interaction or association with the NPC. F. A Consensus Model for the NPC The large amount of structural studies (reviewed above) have elucidated the structure of some of the components of the NPC. Thus a consensus model for the architecture of the NPC is starting to emerge. In the model presented in Figure 8 we have identified a number of distinct structural components. Accordingly, intact membrane-embedded NPC consists of a basic framework (i.e., the spoke complex shown in Fig. 4a) sandwiched between a cytoplasmic and a nuclear ring. These are two rings in addition to the two tenuous rings defined by the ring subunits of the spokes. The cytoplasmic ring is decorated with eight cytoplasmic filaments, and the nuclear ring is topped with a basket-like filamentous assembly. The central pore often harbors a central plug or channel complex whose definite structure and functional role remain to be established.
IV. TOWARD A MOLECULAR ARCHITECTURE OF THE NPC In contrast to the relatively large number of structural studies (see above), less is known about the chemical composition and molecular architecture of the NPC. Based on its molecular mass of about 120 MDa (Reichelt et al., 1990), it is believed that the NPC is composed of multiple copies (i.e., 8 or 16) of on the order of-100 different proteins. For several years gp210 and p62 have been the only two well-characterized NPC polypeptides. However, the production of specific antibodies, the use of molecular genetics, and the development of an isolation procedure for NPCs from yeast (Rout and Blobel, 1993) have recently enabled the identification, characterization, and cloning and sequencing of an increasing number of NPC proteins. Moreover, the combination of well-characterized antibodies with different EM specimen preparation methods has allowed localization of several of these proteins to distinct structural components of the NPC. Thus elucidation of the molecular architecture of the NPC has now definitely started.
Nuclear Pore Complex
19
As summarized in Table 2, three major groups of NPC proteins have so far been identified and characterized. These are (i) integral membrane proteins with only part of their mass residing in the NPC, which have been proposed to anchor the NPC to the nuclear membrane; (ii) peripheral membrane proteins that are not associated with or anchored in the nuclear membrane; members of this group have been called "nucleoporins" (Davis and Blobel, 1986) and denoted by NUPx, where X indicates the molecular mass in kilodaltons (Sukegawa and Blobel, 1993); and (iii) related yeast NPC proteins whose exact relationship to the vertebrate NPC proteins is not yet known. Following is a description of the currently identified members of these three groups.. A. Integral Membrane Proteins of the NPC
The first NPC protein identified in rat liver NEs was a 190-kDa glycoprotein, originally termed gpl90 (Gerace et al., 1982) but renamed gp210 on the basis of the molecular mass of 210 kDa deduced from its amino acid sequence (Wozniak et al., 1989). gp210 was classified as an integral membrane protein since it remained associated with the nuclear membrane even after extraction with alkaline pH or chaotropic agents (Gerace et al., 1982). gp210 contains asparagine-linked, high mannose-type oligosaccharides and therefore binds concanavalin A (ConA), a lectin specific for a-D-mannopyranose and related sugars (Gerace et al., 1982; Wozniak et al., 1989). The topology of gp210 has been determined by the use of site-specific antibodies and proteolytic digestions (Greber et al., 1990). Accordingly, gp210 consists of a large (95% of its total mass) NH2-terminal domain residing in the lumen of the NE, a single, 21-residue-long transmembrane segment that has been shown to be sufficient for sorting gp2l0 to the nuclear membrane (Wozniak and Blobel, 1992), and a short, 58-residue long COOH-terminal domain associated with the NPC (see Fig. 9). Unexpectedly, an antibody directed against the lumenal domain of gp210 inhibits both passive diffusion of small molecules and mediated nuclear import of proteins (Greber and Gerace, 1992). Based on its topology and abundance in the NPC, there have been speculations that the lumenal domain of gp210 forms part of the "knobs" (Jarnik and Aebi, 1991) or lumenal subunits (see Fig. 4) that have been shown to extend from the spokes radially into the lumen of the NE (Hinshaw et al., 1992; Akey and Radermacher, 1993). As a possible functional role, gp210 has been proposed to act as a membrane anchor for the NPC and/or to have a topogenic role in membrane folding during nuclear pore formation (Greber et al., 1990; Jarnik and Aebi, 1991; Gerace, 1992; Hinshaw et al., 1992). Recently, another transmembrane NPC protein has been identified, cloned and sequenced, and characterized (Hallberg et al., 1993). Based on its predicted amino acid sequence, it has a molecular mass of 121 kDa and thus has been termed POM121 (for pore membrane protein of 121 kDa). This protein was not extracted from rat liver NEs by 7.0 M urea, and therefore it was classified as an integral
Table 2. Classification of Cloned and Sequenced NPC Proteins
VP~ Integral membrane proteins of the NPC
Name
Molecular Characteristics of primary and Other properties and possible massa (kDa) predicted secondary structure functions
gp2 10
210
POM121
121
POM152 (yeast)
152
Location
References
Most of its mass (N- Gerace et al., 1982; 2 1-residue long transmembrane Bears N-linked (via Asp) domain between a 58-residue high mannose domain) resides in Wozniak et al., long C-domainb and a 1783oligosacharides. Reacts the lumen of the 1989; Greber et with Con A. Antibodies NE. al.. 1990. residue long N - d ~ m a i n . ~ against lumenal domain inhibits NPC function. Possibly anchors the NPC to the pore membrane. Most of its mass (C- Hallberg et al., 1993. 44-residue long transmembrane Binds WGA. Possibly domain) resides in anchors the NPC to the domain between a 28-residue the NPC proper. long N-domain and a 1127pore membrane. residue long C-domain. Repetitive XFXFG motifs at C-terminal third. 20-residue long transmembrane Reacts with ConA. Possibly Most of its mass (C- Wozniak et al., 1994. domain between a 175anchors the NPC to the domain) resides in pore membrane. the lumen of the residue long N-domain and a NE. 1142-residue long C-domain.
Peripheral membrane p62 proteins of the NPC
62d
0-linked glycoprote- NUP153 ins
153
NUP107 NUPI55
107 155
Cytoplasmic and nu- Starr et al., 1990; clear periphery of Carmo-Fonseca et the central plug or al., 1991; Cordes channel complex et al., 1991; Finlay et al., 1991; Guan et al., 1995. Four zinc finger motifs. Binds WGA. Exist as a homo- Terminal ring of the Sukegawa and Repetitive XFXFG motifs at nuclear basket. Blobel, 1993; oligomer of 2 MDa. Binds C-terminal 213. DNA in vitro. McMorrow et al., 1994; Cordes et al., 1993; Pante et al., 1994. Repetitive XFXFG, SVFG, Binds WGA. Exist as a Cytoplasmic Kraemer et al., 1994; filaments. FGG, and FGG motifs. complex with p75. Pante et al., 1994. Leucine zipper motif. Putative oncogene product associated with myeloid leukemogenesis. Leucine zipper motif. Does not bind WGA. Unknown Radu et al., 1994. Nonrepetitive motifs. Does not bind WGA. Unknown Radu et al., 1993.
Tprlp265 p180?
265
- 1600-residue long a-helical
NSPl
87
Non-0-linked proteins
Yeast NPC proteins
a-helical coiled-coil C-domain. Binds WGA. Exists as a Repetitive XFXFG motifs at complex (p62-p58-p54N-domain. p45). Required for NPC function.
coiled-coil region. Acidic C-domain. Repetitive XFXFG motifs.
Very prone to proteolysis with a major proteolytic product of 180kDa. Essential for cell growth.
-
Cytoplasmic filaments.
Byrd et al., 1994; Wilken et al., 1993
Unknown.
Hurt, 1988. (continued)
Table 2. (Continued)
QP~ XFXFG family
Name NUPl NUP2
Molecular Characteristics of primary and Olher properlies and possible massa (kDa) predicted secondary structure functions 114 95
NUP/NSP49 49 NbPMSP100 100
GLFG family
Loca tion
Essential for cell growth. Unknown. Not required for cell growth. Unknown. Forms a complex with NUPI. Essential for cell growth. Unknown.
Repetitive GLFG motifs at Ndomain. Repetitive GLFG motifs at N- Binds RNA in vitro. domain. RNA binding motifs.
Unknown.
NUP/NSP116 116
Repetitive GLFG motifs at N- Deletions in the nupll6 gene Unknown domain. RNA binding motifs. yields sealed NPCs. Binds RNA in vitro.
NUPINSPI45 145
Repetitive GLFG motifs at N- Deletionsldisruptions in the Unknown domain. RNA binding motifs. nupl45 gene yield clusters of sealed NPCs. Binds RNA in vitro. Nonrepetitive motifs. Forms a complex with NSPI Unknown and NUP49.
N1C96 Notes:
Repetitive XFXFG motifs. Repetitive XFXFG motifs.
96
aCalculated from the amino acid sequence. b ~ ~ ~ ~ - t e r mdomain. inal CNH,-terminal domain. d~ependingon species.
References Davis and Fink, 1990. Loeb et al., 1993; Belanger et al., 1994. Wente et al., 1992; Wimmer et al., 1992. Wente et al., 1992; Wimmer et al., 1992; Fabre et al., 1994. Wente et al., 1992; Wimmer et al., 1992; Wente and Blobel, 1993; Fabre et al., !994. Fabre et al., 199-4; Wente and Blobel, 1994. Grandi et al., 1993.
NPC
perinuclear space
9"^^°
T T ^ c
NPC perinuclear space ' \ ^
?\
perinuclear space * " N
CI 1
^
ox^
C P0M121
NPC
- ^ ^ POM152 (yeast) glycosylatjon sites XePXePXePXeP XFXFG motifs C-G
V—L-G-PF-.- Y motif
Figure 9. Schematic diagram of the domain architecture and membrane topology of the cloned and sequenced integral membrane proteins of the NPC deduced from their amino acid sequences. Three transmembrane NPC glycoproteins, gp210, P O M 1 2 1 , and the yeast POM152, have been thus far cloned and sequenced. They all reveal a distinct stretch of hydrophobic residues that is predicted to be a transmembrane segment transversing the pore membrane. gp210 contains a large NH2-terminal domain residing in the lumen of the NE, and a small COOH-terminal tail associated with the NPC. In contrast, P O M ! 21 consists of a short NH2-terminal tail residing in the lumen of the NE, and a long COOH-terminal domain (with a number of repetitive pentapeptide motifs XFXFG at its COOH-terminal end) associated with the NPC. Very much like gp210, most of the mass of yeast POM152 is predicted to reside in the lumen of the NE. In addition, the amino acid sequence of POM152 contains eight repetitive segments each 24 residues long, with the consensus sequence C-G V— L-G-PF—Y. The N-linked glycosylated residues of gp210 occur atthe lumenal domain close to the nuclear membrane (Greber et al., 1990). By analogy, the N-linked glycosylated sites of POM152 are speculated to be located in the lumenal domain (Wozniak et al., 1994). For more information about these proteins, see Table 2 and references therein. 23
24
NELLY PANTE and UELI AEBI
membrane protein. However, similar to some of the peripheral membrane proteins of the NPC (see below and Table 2), POM 121 binds wheat germ agglutinin (WG A), a lectin that recognizes O-linked A^-acetylglucosamine (GlcNac) residues. Moreover, the cDNA-deduced amino acid sequence of P0M121 has revealed the presence of a repetitive pentapeptide motif XFXFG (where X indicates any amino acid), which is also present in the members of the O-linked NPC glycoprotein family as well as some yeast NPC proteins (see below and Table 2). Although antibodies directed against POM 121 clearly labeled the NPC, the exact location of this protein within the NPC remains to be determined. Since the primary structure of POM 121 has revealed a 44-residue-long transmembrane domain sandwiched between a short NH2-terminal tail (28 residues long) and a long COOH-terminal domain containing the repetitive XFXFG motifs, it has been predicted that the small NHj-terminal tail resides in the lumenal domain (see Fig. 9). Thus, in contrast to gp210, most of the mass of P0M121 is predicted to reside within the NPC proper. Very much like gp210, POM 121 has been proposed to function as a membrane anchor of the NPC. The recent development of a procedure to isolate milligram quantities of NPCs from yeast (Rout and Blobel, 1993) has opened the possibility of more systematically identifying novel yeast NPC proteins. Using this approach, an integral membrane glycoprotein that reacts with ConA has been identified, and cloned and sequenced (Wozniak et al., 1994). The deduced amino acid sequence revealed a 152-kDa protein, termed POM 152, that does not share any similarity with the two vertebrate transmembrane NPC proteins except for a small region (19 residues long), similar to P0M121. Analysis of the amino acid sequence of POM 152 indicated that this protein contains a 20-residue long transmembrane domain between residues 175 and 196. Since the COOH-terminal domain (residues 196— 1337) of POM 152 contains three putative sites for A^-linked glycosylation, at least one of which is glycosylated, in analogy to gp210 (see above), this domain has been proposed to reside in the lumen of the NE (see Fig. 9). However, the exact topology of this protein remains to be established. B. Peripheral Membrane Proteins of the NPC O-Linked Glycoproteins
A group of at least eight peripheral membrane proteins of the NPC has been identified by immunological approaches in rat liver NEs (Davis and Blobel, 1987; Holt et al., 1987; Snow et al., 1987). These proteins, which have molecular masses ranging fi-om 35 to 250 kDa, contain up to 10-20 O-linked A^-acetylglucosamine (GlcNac) residues and therefore bind the lectin WGA. These O-linked glycoproteins appear to be involved in mediated nuclear import, which is inhibited by both monoclonal antibodies to these proteins (Dabauvalle et al., 1988a; Featherstone et al., 1988) and WGA(Finlay et al., 1987; Dabauvalle et al., 1988b). As summarized in Table 2, three of these O-linked glycoproteins have now been cloned and
Nuclear Pore Complex
25
sequenced: p62 (Starr et al, 1990;Carmo-Fonsecaetal., 1991;Cordesetal., 1991), NUP153 (Sukegawa and Blobel, 1993; McMorrow et al., 1994), and rat NUP214, the ~210-kDa glycoprotein originally identified by Snow et al. (1987) and recently demonstrated to be a homologue of human CAN (Kraemer et al., 1994), a putative oncogene product associated with myeloid leukemogenesis (Von Lindern et al., 1992). All three of these proteins contain multiple copies of a more or less degenerate pentapeptide (XFXFG) motif (see Fig. 10), which is considered to be a diagnostic feature for the O-linked NPC glycoprotein family. As illustrated in Figure 10, these repeats are clustered within each protein: within the NH2-terminal half of p62, and within the COOH-terminal domain of NUP153 and CAN. In addition, CAN contains multiple copies of the tripeptide motif FGQ that is also present in two yeast NPC proteins, NUPlOO and NUP116 (see Fig. 14; Wente et al., 1992; Wimmer et al., 1992), and of the degenerate tetrapeptide motif SVFG and the tripeptide motif FGG that have so far not been found in other NPC proteins. CAN also contains a leucine zipper motif that may represent a protein-protein dimerization domain (Von Lindern etal., 1992). As indicated in Figure 10, in addition to the NH2-terminal domain containing multiple copies of the XFXFG motif, the COOH-terminal half of p62 contains heptapeptide repeats characteristic of a-helical coiled-coil conformations. Recently, p62 has been expressed in Escherichia coli and the recombinant protein visualized in the EM after glycerol spraying/rotary metal shadowing (Buss et al., 1994). Accordingly, recombinant p62 appears as a 3 5-nm rod-shaped molecule with a slight protuberance at the NH2-terminal end, thus confirming the a-helical coiled-coil conformation of the COOH-terminal domain. Circular dichroism of recombinant p62 has indicated that the repetitive NH2-terminal domain may have a cross-P conformation (Buss et al., 1994). As we will discuss in Section IVD, within the NPC p62 exists as a complex with at least two other NPC proteins, p58 and p54 (Finlay et al., 1991; Guan et al, 1995). This complex is required for protein import into the nucleus (Finlay et al., 1991). NUP153 is unique among the 0-linked NPC glycoproteins identified and characterized to date in that its primary sequence harbors four zinc finger motifs, each containing two pairs of cysteine residues (Cys2-Cys2) (Sukegawa and Blobel, 1993; McMorrow et al., 1994). Since this type of zinc finger motif is found in DNA-binding proteins (reviewed by Coleman, 1992), a fragment of NUP153 containing these four motifs was expressed in E. coli and demonstrated to bind DNA in a zincdependent manner (Sukegawa and Blobel, 1993). This result has given rise to speculations about a possible role of NUP 153 in gating transcribable genes to the NPC, thus facilitating export of the transcribed RNA (Sukegawa and Blobel, 1993). 3'D Localization of Some of the O-Linked Glycoproteins Since WGA specifically binds to the O-linked NPC glycoproteins, several attempts have been made to localize these polypeptides by labeling NPCs with
a) 0-linked glycoproteins
N
b) Non-0-linkedproteins
~c
NUP107
H
NI I . . . 1 C Tprtp265 ,
CS/irwW
coiled-mila-helix
v
XFXFG motifs
q
XFXFG motifs alemating with
R SVFG, FGQand FGG motifs Leucine zipper motif
Figure 10. Schematic diagram of the domain architecture of the cloned and sequenced peripheral membrane proteins of the NPC deduced from their amino acid sequence. Depending on the content of Olinked N-acetylglucosamine (GlcNac) residues, two families of peripheral membrane proteins of the NPC have been distinguished: (a) 0-linked glycoproteins that contain several copies of a more or less degenerate pentapeptide motif XFXFG, and (b) non-@linked proteins that do not contain any repetitive sequence motifs. In addition, p62 ,contains a COOH-terminal a-helical coiled-coil domain, and NUP153 harbors four zinc finger motifs. In the case of CAN/NUP214/p250, the repetitive XFXFG motif alternates with repetitive SVFG, FGQ, and FGG motifs. The amino acid sequences of the three members of the non-Olinked protein family are unique. In the case of Tprlp265, it contains a very long (1600 residues) a-helical coiled-coil domain near its NH2-terminal end. For more information about these proteins, see Table 2 and references therein.
Nuclear Pore Complex
27
ferritin-tagged (Finlay et al., 1987; Hanover et al., 1987) or colloidal gold-tagged WGA(Akey and Goldfarb, 1989; Pante and Aebi, 1993; Pante et al., 1994). These labeling studies have revealed that WGA specifically binds to the NPC, and they have localized at least one of its binding sites to the terminal ring of the nuclear basket (Pante and Aebi, 1993; Pante et al., 1994). In projection, additional WGA binding sites have been identified at two distinct radial locations: (i) at low radii between 3.6 and 12.7 nm, and (ii) at higher radii between 18.2 and 36.2 nm (Pante e t a l , 1994). Using antibodies to individual NPC proteins, immunofluorescence microscopy and immuno-EM studies have localized these to the NPC. However, the relatively strong cross-reactivity of these antibodies has made the localization of individual proteins within the NPC difficult. For example, this protein was symmetrically located at both the nuclear and cytoplasmic faces of mouse liver NPCs by using a polyclonal mouse anti-p62 antibody (Cordes et al., 1991), whereas the same antibody labeled only the nucleoplasmic face of Xenopus oocyte NPCs (Cordes et al., 1991; Pante and Aebi, 1993). Thus the localization of p62 has remained ambiguous. To resolve this ambiguity, we have recently produced a monoclonal antibody, RL31, which reacts specifically with rat p62 (Guan et al., 1995). As illustrated in Figure 11, RL31 predominantly labels the central plug or channel complex of rat liver NPCs, both its nuclear and cytoplasmic periphery, although the labeling at its nuclear face is more frequent. By the use of a polyclonal antibody raised against a fusion protein expressed from a NUP153 cDNA construct, NUP153 has been unequivocally localized to the nuclear periphery of the NPC (Sukegawa and Blobel, 1993). By the same approach, CAN/NUP214 has been localized to the cytoplasmic periphery of the NPC (Kraemer et al., 1994). However, in these studies there was no clear identification of a particular NPC component(s) labeled by these antibodies. Recently, more specific localization of these two proteins has been reported. Using an antibody raised against an extract of nuclear matrix proteins that recognizes NUP153, Cordes et al. (1993) have localized this protein to intranuclear NPC-attached filaments, which among other structures may represent nuclear baskets that have been disrupted during sample preparation. More specifically, Pante et al. (1994) have identified this protein as a constituent of the nuclear basket with at least one of its epitopes residing in the terminal ring (see below). Accordingly, as documented in Figure 12, a monoclonal antibody, termed QE5, that by immunoblotting recognized three O-linked glycoproteins (p62, NUP 153, and p250), labeled both the cytoplasmic and the nuclear peripheries of Xenopus oocyte NPCs. On the cytoplasmic face of the NPC gold-conjugated QE5 labeled (i) the cytoplasmic filaments and (ii) sites down in the pore disposed toward the nuclear opening (Fig. 12, a and c). As can be seen in Figure 12, b and c, at the nuclear periphery, gold-conjugated QE5 labeled predominantly the nuclear baskets. To identify these distinct labeling sites with the different NPC proteins recognized by QE5, Pante et al. (1994) have used an anti-peptide antibody against human NUP 153 and a monospecific anti-p250 poly-
28
NELLY PANTE and UELI AEBI
100 nm Figure 11. Localization of the rat NPC protein p62 by immunoelectron microscopy. Isolated rat liver NEs were incubated with the monoclonal antibody RL31 (which recognizes rat p62 on Western blots), conjugated to 8-nm colloidal gold, and prepared for EM by embedding and thin sectioning. Shown are a view along a single cross-sectioned NE stretch together with a gallery of labeled NPCs cross sections. These cross sections revealed that RL31 predominantly labels the central plug or channel complex of rat liver NPCs, both its nuclear and cytoplasmic periphery, although the labeling at the nuclear side appears to be more frequent, c, cytoplasmic; n, nuclear side of the NE. Scale bar, 100 nm from Guan et al. (1995).
clonal antibody. As illustrated in Figure 12, c-e, labeling with these two antibodies revealed that NUP153 is a constituent of the terminal ring of the nuclear basket, whereas p250 is a constituent of the cytoplasmic filaments. Localization of NUP153 at the nuclear basket together with its four zinc finger motifs is consistent with the stabilizing effect of Zn^"^ on the nuclear baskets (see Fig. 6; Jamik and Aebi, 1991). The '-250-kDa glycoprotein recognized by QE5 also reacts with the RLl monoclonal antibody used by Snow et al. (1987) (B. Burke and R. Bastos, personal Figure 12. Localization of CAN/NUP214/p250 and NUP153 by immunoelectron microscopy, (a) Cytoplasmic and (b) nuclear faces of quick-frozen/freeze-dried/rotary metal-shadowed spread Xenopus oocyte NEs labeled with the monoclonal QE5 antibody conjugated to 8-nm colloidal gold. QE5, which recognizes p250, NLJP153, and p62 on Western blots, specifically labeled the cytoplasmic periphery of the NPC at (i) the cytoplasmic filaments (a, arrowheads) and (ii) sites down in the pore disposed toward the nuclear opening, (continued)
Nuclear Pore Complex
29
Figure 12: (continued) At the nuclear periphery of the NPC, the gold-conjugated QE5 labeled predominantly the nuclear baskets (b, arrowheads), (c) Gallery of selected examples of quick-frozen/freeze-dried/rotary metal-shadowed NPCs labeled with the monoclonal QE5 antibody, a polyclonal antibody against p250, and an anti-peptide antibody against human NUP153. The anti-p250 antibody exclusively labeled the cytoplasmic filaments, whereas the anti-NUP153 anti-peptide antibody exclusively labeled the terminal ring of the nuclear baskets, (d, e) cross sections and selected examples of Epon-embedded, Triton X-100-treated Xenopus oocyte nuclei labeled with antl-p250 antibody (d) and anti-NUP153 anti-peptide antibody (e). In agreement with the quick freezing/freeze drying/rotary metal shadowing results shown In c, these cross sections documented that the anti-p250 antibody exclusively labeled the cytoplasmic filaments, whereas the antl-NUP153 anti-peptide antibody exclusively labeled the terminal ring of the nuclear baskets, c, cytoplasmic; n, nuclear side of the NE. Scale bar, 100 nm (a-e). Adapted from Pante et al. (1994).
30
NELLY PANTE and UEUAEBI
100 nm Figure 13. Localization of Tpr/p265 by immunoeiectron microscopy. Isolated rat liver NEs were incubated with the monoclonal RL30 antibody (which specifically reacts with rat p265) conjugated to 8-nm colloidal gold and salt-washed with 0.5 M NaCI prior to embedding and thin sectioning. Shown are a view along a single cross-sectioned NE stretch together with a gallery of labeled NPCs cross sections. These cross sections revealed that RL30 exclusively labeled the cytoplasmic side of the NE. Gold particles were found to be associated predominantly with the cytoplasmic filaments of the NPC, which in some cases bent around and reached down into the central pore (see arrowhead), c, cytoplasmic; n, nuclear side of the NE. Scale bar, 100 nm. Adapted from Byrd et al. (1994).
communication). In addition, p250 is recognized by polyclonal antibodies raised against the cloned NH2- and COOH-terminal domain of human CAN (B. Burke and R. Bastos, personal communication). Therefore, p250 corresponds to the (9-linked glycoprotein CAN/NUP214. Non-O'Linked Proteins A group of at least 30 proteins that do not contain GlcNac have recently been identified in rat liver NEs (Radu et al., 1993). This group of proteins has been separated from the O-linked NPC glycoproteins by WGA-Sepharose chromatography, and they have been further isolated by SDS-4iydroxylapatite chromatography (Radu et al., 1993). However, so far only two of these polypeptides, NUP155 (Radu et al., 1993) and NUP107 (Radu et al., 1994), have been demonstrated to be bona fide NPC proteins. Both proteins have been cloned and sequenced, and their
Nuclear Pore Complex
31
deduced amino acid sequences do not reveal any of the repetitive sequence motifs (i.e., the XFXFG motif), which seem to be a diagnostic feature for the 0-linked NPC glycoproteins (see Table 2). As illustrated schematically in Figure 10, NUP107 contains a leucine zipper motif at its COOH-terminal end, which has been suggested to cause dimerization with a second leucine-zipper-containing polypeptide (Radu et al, 1994). Anti-peptide antibodies against both NUP 155 (Radu et al., 1993) and NUP 107 (Radu et al., 1994) were raised and used to label several types of cultured cells by immunofluorescence microscopy and immuno-EM. Both antibodies labeled the NPCs of these cells, thus confining that NUP 155 and NUP 107 are bona fide NPC proteins. However, localization of these two proteins to distinct NPC components remains to be determined. By using an autoimmune serum from a patient with overlap connective tissue disease, a NPC polypeptide of 180 kDa that does not bind WGA has recently been identified (Wilken et al., 1993). Affinity-purified antibodies from this serum were used to \3bQ\Xen0pus oocyte NEs (Wilken et al., 1993). Accordingly, this antibody bound to the cytoplasmic ring and associated fibers of the NPC, Using both monoclonal and autoimmune antibodies, Byrd et al. (1994) have identified a 265-kDa non-0-linked NPC protein in rat liver NEs. As documented in Figure 13, by immuno-EM this protein localizes to the cytoplasmic filaments of the NPC. It has further been shown that p265 is very prone to proteolysis with a major 175-kDa proteolytic product (Byrd et al., 1994), which could be identical to the 180-kDa polypeptide identified by Wilken et al. (1993). Supporting this notion, the autoimmune antibody used by Wilken et al. (1993) also reacted with a '-260-kDa protein. Most interestingly, amino acid sequences of peptides from p265 and its proteolytic fragment pi 75 revealed 80-90% identity with the primary sequence of human Tpr (translocated promoter region), a '-265-kDa protein whose NH2-terminal domain appears in oncogenic fusions with the met, trk, and ra/protooncogenes (Mitchell and Cooper, 1992). Therefore, p265 is the rat homologue of human Tpr (Byrd et al., 1994). Consistent with being a non-O-linked NPC protein, the amino acid sequence of Tpr lacks the repetitive XFXFG pentapeptide motif diagnostic for the O-linked NPC glycoproteins (see Table 2). In addition, Tpr contains a predicted a-helical coiled-coil region over 1600 residues long (see Fig. 11; Mitchell and Cooper, 1992). Based on this secondary structure, it is conceivable that Tpr, together with p250 (see Fig. 12; Pante et al., 1994), forms the backbone of the cytoplasmic filaments. However, because of its size Tpr may also extend into the cytoplasmic ring. C. Yeast NPC Proteins
The use of antibodies against vertebrate NPC proteins in concert with the design of genetic screens has allowed identification of at least nine yeast NPC proteins (see Table 2; reviewed by Fabre and Hurt, 1994). The number of these are expected to rapidly increase, now that a procedure to bulk isolate NPCs from yeast has been developed (Rout and Blobel, 1993). As mentioned above, one of these yeast NPC
32
NELLY PANTE and UELI AEBI
a) XFXFG family ls\Jiodlpo€i€ta€iaacifSj/^^
Nl
NSP1
)DOOOOOCX30000C3|C»Q|
Nl
C
loooooooooc^
NUP1
NUP2
b) GLFG family Ntocwaadl
^c
NUP/NSP49
Ic
NUP/NSP100
N[^S=i«»ia
NIC96
GLFG motifs
J^WHSIWSBi I
I
related domains
RNA-binding motif
Figure 14, Schematic diagram of the domain architecture of the cloned and sequenced yeast NPC proteins deduced from their amino acid sequence. Depending on the occurrence of highly repeated motifs in their amino acid sequence, three families of yeast NPC proteins have been distinguished: (a) the XFXFG family, which contains several copies of a more or less degenerate pentapeptide motif XFXFG clustered in the central part of each protein; (b) the GLFG family, which contains several copies of a degenerate tetrapeptide GLFG within their NH2-terminai domain; and (c) the third family includes the yeast nucleoporin interacting component NIC96, which does not contain any repetitive sequence motifs and forms a complex with NSP1 and NUP49 (Grandi et al., 1993). Three members of the second family, NUP100, NUP116, and NUP145, contain related domains, including an RNA-binding motif. For more information about these proteins, see Table 2 and references therein.
Nuclear Pore Complex
33
proteins, POM 152, is an integral membrane protein. The rest of them can be classified into three groups based on the occurrence of highly repeated sequence motifs (see Table 2 and Fig. 14). First, the XFXFG family contains several copies of a more or less degenerate pentapeptide motif (XFXFG) clustered in the central part of each protein (see Fig. 14a). This XFXFG motif is characteristic of the vertebrate O-linked NPC glycoproteins (see above and Fig. 10), but it is not clear whether any of these yeast NPC proteins are in fact glycosylated. Members of this group include NSPl (Hurt, 1988), NUPl (Davis and Fink, 1990), and NUP2(Loeb et al., 1993). A second group represents the GLFG family, which contains multiple copies of a degenerate tetrapeptide motif (GLFG) within their NH2-terminal domain (see Fig. 14b). Members of this group include NUP/NSP49, NUP/NSP100, NUP/NSPl 16 (Wente et al., 1992; Wimmer et al., 1992), as well as NUP/NSP 145 (Fabre et al., 1994; Wente and Blobel, 1994). Finally, the third group is defined by the yeast nucleoporin interacting component NIC96 (see Fig. 14c), which does not exhibit any of these repetitive sequence motifs (Grandi et al., 1993). Although most of the vertebrate and yeast NPC proteins contain short repetitive sequence motifs, in particular the XFXFG motif, no obvious relationship has yet been depicted between these vertebrate and yeast proteins. The only exception is NSP1, which is considered to be the yeast homologue of vertebrate p62: both proteins have the same domain structure (see Figs. 10 and 14), with 50% similarity in their COOHterminal domains (Carmo-Fonseca et al, 1991; Fabre and Hurt, 1994). Three members of the GLFG family, three proteins, NUP/NSP 100, NUP/NSP 116, and NUP/NSP 145, contain highly homologous sequence regions (see Fig. 14b) including an RNA-binding motif (Fabre et al., 1994). Fragments of NUP/NSPl 16 and NUP/NSP 145 including the RNA-binding motif were expressed as fusion proteins in E. coli, and these were demonstrated to bind RNA in vitro (Fabre et al., 1994). Based on these results, it has been suggested that NUP/NSP 116 and NUP/NSP 145 may play a role in RNA recognition and/or transport through the NPC (Fabre and Hurt, 1994; Fabre et al., 1994). Whereas antibodies directed against some of these yeast NPC proteins do label the NPC by immuno-EM, the exact localization of their epitopes has not yet been established. One of the difficulties in determining their localization is that the 3-D structure of the NPC in embedded/thin-sectioned yeast NEs or nuclei is poorly preserved. On the other hand, the yeast system has the advantage that the phenotype of a mutation in a particular NPC protein can be directly characterized. As a consequence, the genes for several yeast NPC proteins have been shown to be essential for cell growth (reviewed by Fabre and Hurt, 1994). Yeast mutants can also be examined in the EM to determine whether they perturb the structure of the NPC and/or the NE. For example, Wente and Blobel (1993,1994) have investigated the phenotype after deletion/disruption of the yeast NPC proteins NUP/NSPl 16 and NUP/NSP 145. Accordingly, a membrane seal was formed over the cytoplasmic face of the NUP/NSPl 16-deficient NPCs, which did not block nuclear export but caused the export substrate to accumulate within the cytoplasmic membrane
NELLY PANTE and UELI AEBI
34 Tpr/NUP180
CAN / p250
POM121
p62 Complex
NUP153
Figure 15. Schematic diagram summarizing the immunolocalization of characterized NPC protein epitopes within the 3-D architecture of the consensus model of the NPC. Tpr/p265 and CAN/NUP214/p250 exhibit epitopes at the cytoplasmic filaments (see Figures 12 and 13), whereas NUP153 exhibits an epitope at the terminal ring of the nuclear basket (see Figure 12). p62 epitopes are exposed at both the cytoplasmic and nuclear peripheries of the central plug or channel complex (see Figure 11). The transmembrane glycoprotein gp210 exhibits several epitopes in the lumen of the NE (Greber et al., 1990) where, based on its topology, most of its mass resides (see Figure 9; Greber et al., 1990). Epitopes for the transmembrane glycoproteins POM121 and POM152 are also shown, but they should not be taken literally because their exact localization remains to be determined. In contrast to gp210, most of the mass of POM121 is predicted to reside within the NPC proper. Very much like gp210, most of the mass of yeast POM152 is predicted to reside In the lumen of the NE.
herniations covering the NPCs (Wente and Blobel, 1993). Furthermore, deletion/disruption of the NH2-terminal end of NUP/NSP145 yielded yeast nuclei with clusters of numerous NPCs interconnected by a network of NE herniations (Wente and Blobel, 1994). Based on these results, it has been proposed that NUP/NSPl 16 and NUP/NSP145 are possibly involved in establishing specific NPC—NE interactions and/or mediating NPC biogenesis (Wente and Blobel, 1993; 1994). In summary, only 17 (8 vertebrate and 9 yeast) of the approximately 100 polypeptides of the NPC have thus far been identified, characterized, and cloned and sequenced (see Table 2). As illustrated in Figure 15, epitopes of 7 of these NPC
Nuclear Pore Complex
35
proteins have been localized within the 3-D NPC architecture. Since the native conformation of none of these proteins is currently known, it is difficult to map the entire protein within the NPC. However, identification of some of their epitopes with distinct structural components of the NPC has been the first step toward elucidation of the molecular architecture of the NPC (see Fig. 15). D. Isolation and Characterization of Distinct NPC Components
When assembled in the NPC, different NPC proteins may mutually interact, thus forming distinct subcomplexes within the NPC. In vertebrate species, it was first reported that some of the soluble NPC proteins contained in in vitro nuclear reconstitution extracts from Xenopus oocytes form a large supramolecular complex with a molecular mass of 254 kDa, which contains p68, the Xenopus homologue of rat p62, interacting with other NPC proteins (Dabauvalle et al., 1990). This complex has also been isolated and characterized at the molecular level from rat liver NEs (Finlay et al., 1991; Kita et al., 1993). Accordingly, it consists of p62 interacting with two other proteins of molecular mass 58 (p58) and 54 (p54) kDa. Finlay et al. (1991) estimated the mass of the p62 complex to be 550-600 kDa with a molar stoichiometry of 4:1:1 (p62:p58:p54). However, Kita et al. (1993) have reported a mass of 231 kDa and a molar stoichiometry of 1:1:2 (p62:p58:p54). Thus the molecular mass and molar stoichiometry of the p62 complex remain controversial. More recently, Guan et al. (1995) have developed a modified procedure to isolate the p62 complex from rat liver NEs. Accordingly, in addition to p58 and p54, p62 is associated with an additional NPC protein, p45. Mass analysis by quantitative STEM has revealed that the native p62 complex is a ~200-kDa particle containing one copy each of p62, p58, p54, and p45 (Guan et al., 1995). This "minimal" p62 complex is able to self-associate into higher order oligomers. Using the monoclonaaibody, Pante et al. (1994) have identified, in addition to the p62 complex, two distinct NPC complexes in extracts of BHK cells. Although this antibody recognizes p62, NUP153, and p250 on Western blots, it immunoprecipitates three additional polypeptides (p54, p58 and p75) that were found to be associated with p62 (p54 and p58) and with p250 (p75). In addition, NUP153 was released as a homo-oligomer of >1 MDa, most likely representing an octamer. Because several of its epitopes are located on the terminal ring of the nuclear baskets (see Fig. 12), it is conceivable that the NUP153 octamer defines the basic framework of the octameric terminal ring. Since the long a-helical COOH-terminal domain of yeast NSPl directs cytosolic proteins to the NPC, it was proposed that this protein might interact with other yeast NPC proteins (Hurt, 1990). Indeed, NSPl has now been shown to form a complex with three other proteins, NIC96, NSP49, and a novel yeast NPC protein of 54 kDa that has not yet been cloned and sequenced (Grandi et al., 1993). A second NPC complex has recently been identified in yeast (Belanger et al, 1994). Accordingly, it consists of the yeast proteins NUP1 and NUP2 interacting with Srpl, the product
36
NELLY PANTE and UELI AEBI
of a gene previously identified as a suppressor of mutants defective in RNA polymerase I (Yano et al., 1992).
V. MOLECULAR TRAFFICKING THROUGH THE NPC As illustrated schematically in Figure 1, two different types of nucleocytoplasmic transport across the NPC occur: (i) passive diffusion of ions and small molecules through an aqueous channel with a physical diameter of--9 nm (Paine et al., 1975): and (ii) mediated transport of proteins, RNAs, and RNP particles through a gated channel with a functional diameter of up to 26 nm (Feldherr et al., 1984). Although the mediated nucleocytoplasmic transport of different substrates uses the same machine, depending on the substrate it appears to occur via different signal pathways. Some of the signals, receptors, and factors mediating nuclear import of proteins or RNPs and nuclear export of RNAs have begun to be identified and isolated. In this section we focus on some recent advances made toward identification and characterization of different signals and factors required for molecular trafficking trough the NPC. For background information and further details, the reader is referred to some recent reviews covering this topic (Forbes, 1992; Gerace, 1992; Izaurralde and Mattaj, 1992; Mattaj et al., 1993; Newmeyer, 1993). A. Passive Diffusion
Early microinjection experiments of dextrans have demonstrated that the NPC has the properties of a molecular sieve (Paine et al., 1975). Accordingly, molecules larger than ^9 nm are excluded from the nucleus, while smaller molecules can passively diffuse across the NPC with a rate inversely proportional to their size. The recent 3-D reconstruction of the NPC (see Fig. 4; Hinshaw et al., 1992) from negatively stained preparations of detergent-treated Xe«o/7M5 oocyte NEs (see Figs. 2d and 3b) has indicated that there may exist eight ^ 10-nm diameter, slightly kinked channels at a radius of-40 nm that are located between two adjacent spokes and the pore membrane (see Section IIIB and Fig. 4a). These peripheral channels have been proposed to be sites for passive diffusion of ions and small molecules through the NPC (Hinshaw et al., 1992). In contrast, the 3-D reconstruction of ice-embedded NPCs (Akey and Radermacher, 1993) has revealed eight ~ 10-nm diameter channels at a radius of-32 nm located between two adjacent spokes and the central channel complex; these channels have also been speculated to represent sites for passive diffusion (Akey and Radermacher, 1993). Thus the location of these diffusional channels remains controversial. B. Import of Nuclear Proteins
Nuclear import of proteins has been the most extensively studied nuclear transport process (review by Forbes, 1992; Gerace, 1992; Newmeyer, 1993). In vitro and in vivo studies have demonstrated that nuclear protein import is highly selective, and requires ATP and cytosolic factors (Feldherr et al., 1984; Newmeyer et al..
Nuclear Pore Complex
37
1986; Richardson et al., 1988; Adam et al., 1990; Newmeyer and Forbes, 1990). Targeting of nuclear proteins to the NPC is specified by short amino acid sequences, called nuclear localization signals (NLSs), on the protein to be transported (reviewed by Dingwall and Laskey, 1991; Garcia-Bustos et al., 1991). Many studies have focused on the characterization of NLSs and identification of receptors that interact with NLSs to mediate import of nuclear proteins. As a consequence, two types of NLSs have been identified: (i) the simian virus 40 (SV40) large T-antigen type of NLS, which consists of a single contiguous stretch of basic amino acid residues (Chelsky et al., 1989); and (ii) the Xenopus nucleoplasmin type of bipartite NLS, which contains two interdependent basic domains separated by 10 intervening "spacer" residues (Robbing et al., 1991). A number of NLS-binding proteins have also been identified (reviewed by Gerace, 1992). However, in addition to a NLS-binding protein, cytosolic factors are required in order to stimulate mediated nuclear import of a protein in vitro (Adam et al, 1990; Newmeyer and Forbes, 1990). Recently, considerable effort has been made to identify, isolate, and characterize the cytosolic factors essential for nuclear import of proteins. Using a cell-free in vitro assay for nuclear protein import (i.e., Xenopus egg extracts, nuclei from any source, and a fluorescently labeled nuclear protein; Newmeyer et al., 1986), Newmeyer and Forbes (1990) have identified two cytosolic factors, one required for ATP-independent binding of proteins to the NPC, and the second one for translocation of the substrate through the NPC. The development of another in vitro transport assay consisting of digitonin-permeabilized cells (digitonin permeabilizes only the plasma membrane while leaving the NE intact) supplemented with exogenous cytosol and ATP (Adam et al., 1990) has demonstrated the requirement of multiple cytosolic factors for nuclear import of proteins (Adam et al., 1990; Moore and Blobel, 1992). Four cytosolic factors essential for import of nuclear proteins have recently been identified with this assay: (1) two NLS binding proteins of 54 and 56 kDa that have the properties of a functional import receptor (Adam et al., 1989; Adam and Gerace, 1991); (2) a component that interacts with O-linked glycoproteins of the NPC (Steme-Marr et al., 1992); (3) the ubiquitous cellular protein hsc70 (Imamoto et al., 1992; Shi and Thomas, 1992); and (4) the small GTPase Ran/TC4 (Melchior et al., 1993; Moore and Blobel, 1993). The latter has to be in an "active," GTP-bound state to mediate nuclear import, which is inhibited by GTP-y-S and other nonhydrolyzable GTP analogues (Melchior et al., 1993; Moore and Blobel, 1993). In addition to these four cytosolic factors, Adam and Adam (1994) have recently identified another cytosolic factor that binds to the NLS receptor, thereby mediating the binding of the protein-receptor complex to the NPC, but does not mediate the translocation step. The relationship of all these cytosolic factors, as well as their site(s) and mechanism of action, remains to be established. Although these factors are predominantly cytosolic proteins, some of them (i.e., the 54/56 NLS receptor, the hsc70 protein, and Ran/TC4) are also found in the nucleus, suggesting that they may function as shuttling carriers and thus may be recycled for several rounds of transport (Adam et al., 1989; Adam and Gerace, 1991; Gerace, 1992).
38
NELLY PANTE and UELI AEBI
Originally, nuclear import of proteins has been described as a two-step process: (1) binding of the nuclear protein to the NPC and (2) translocation of the NPCbound protein through the NPC (Newmeyer and Forbes, 1988; Richardson et al., 1988). The first step does not require ATP and is temperature independent, whereas the second step requires energy via ATP hydrolysis. However, as illustrated in Figure 16, the recent advances on identifying and characterizing NLSs, NLS receptors, and cytosolic factors suggest that the NPC-mediated transport pathway of proteins into the nucleus occurs by at least five distinct steps: (1) While in the cytoplasm, the protein to be imported is complexed to a cytosolic NLS-binding receptor via its specific NLS (Newmeyer and Forbes, 1988; Adam and Gerace, 1991), and this interaction may be stabilized by some cytosolic factor(s) (Adam and Adam, 1994). (2) Depending on the action of additional cytosolic factors, this protein-receptor complex then docks to an NPC by specific binding to some "peripheral" NPC component such as the cytoplasmic ring or the cytoplasmic filaments (Richardson et al., 1988; Steme-Marr et al., 1992). (3) From this peripheral docking site the protein-receptor complex is next delivered to the central channel complex, which harbors the actual transport machine. (4) Active translocation of the protein-receptor complex through the central channel complex occurs after channel gating to accommodate the particular size and shape of the proteinreceptor complex. (5) After release into the nucleus, the protein-^*eceptor complex dissociates, and the receptor may be recycled for further rounds of transport (Adam et al., 1989; Adam and Gerace, 1991). Some of these nuclear protein import steps are still hypothetical, and several issues remain elusive. For example, the site(s) and mechanism of ATP utilization, the site(s) and mechanism of action of the different cytosolic factors, the NPC ligands and the role of specific NPC components involved in the different transport steps, and last but not least, the nature and molecular mechanism of the gated channel. Thus although a number of functional aspects of the mediated import of proteins into the nucleus have now been established, there remain many questions and ambiguities to be resolved before the detailed molecular mechanism implicated in this process will be understood. C. Export/Import of RNAs and RNP Particles
In addition to nuclear proteins, the NPC also imports ribonucleoprotein (RNP) complexes into the nucleus. The import of U-rich small nuclear ribonucleoprotein (U snRNP) particles which contain a 5' trimethyl G cap and are complexed with proteins termed Sm has been studied in some detail (reviewed by Izaurralde and Mattaj, 1992; Mattaj et al., 1993). For example, nuclear import of U6 snRNP particles is inhibited by proteins bearing the SV40 T antigen NLS (Michaud and Goldfard, 1991); thus nuclear import of this particle seems to occur by a mechanism identical (or at least very similar) to that of nuclear import of proteins (see above). However, in the case of Ul, U2, U3, U4, and U5 snRNP particles, their nuclear import is not inhibited by proteins bearing an NLS, but it requires both the trimethyl
• : protein ;
^
%-
M: receptor
1. formation of proteinreceptor complex
Blj! ^
ilMii ^
2. docking of protein-receptor complex to NPC periphery (cytoplasmic filaments and/or cytoplasmic ring)
3. delivery of protein-receptor complex to central channel complex
liiii J Jio4 ^ * '"^^
^* ^*^^'^'^®' 9^^^^9 ^^^ translocation of protein-receptor complex
my ~^ n^l
i'
7
5. release and dissociation of protein-receptor complex
Figure 16. Schematic diagram of the nuclear import pathway of proteins through the NPC. In the first step, the protein to be transported associates with an NLS receptor this complex then docks to the cytoplasmic periphery (i.e., to the cytoplasmic filaments or the cytoplasmic ring) of an NPC, from where it is delivered to the central plug or channel complex for translocation. Each one of these steps appears to be mediated by the action of one or several cytosolic factors. Once in the nucleus, the protein-receptor complex dissociates, and the receptor may be recycled for another round of transport. Adapted from Gerace (1992). 39
40
NELLY PANTE and UELI AEBI
G cap and binding of Sm proteins (Fischer and Luhrmann, 1990; Hamm et al., 1990; Michaud and Goldfard, 1992; Fischer et al., 1993). Moreover, nuclear import of the U3 snRNP particle is not inhibited by an excess of free trimethyl G cap or by proteins bearing an NLS (Michaud and Goldfard, 1992). Thus there appear to exist at least three distinct signaling pathways for import of snRNP particles. Recently, Marshallsay and Luhrmann (1994), using the digitonin-permeabilized cell system developed by Adam et al. (1990), have demonstrated that nuclear import of RNP particles requires cytosolic factors, as does nuclear import of proteins (see Section VB). This approach now opens the possibility of identifying and characterizing both the specific signals and factors mediating the nuclear import of RNP particles. Compared with the nuclear import of proteins and RNP particles, the export of RNAs and RNP particles is still rather poorly understood. The NPC exports several classes of RNAs, including snRNAs, mRNAs, and tRNAs. Since they are packaged with proteins into RNP complexes, these different RNAs are probably exported in the form of RNP particles. Indeed, export of RNP particles through the NPC has been visualized by EM (Stevens and Swift, 1966; Mehlin et al., 1992). For example, export of Balbiani ring granules (premessenger RNP particles in the salivary glands of Chironomus) seems to be a polar process in that the 5' end of its RNA exits the nucleus first (Mehlin et al., 1992). Similar to nuclear protein import, export of RNAs from the nucleus is a signal-dependent, receptor-mediated process that requires energy in the form of ATP hydrolysis (Zasloff, 1983; Bataille et al., 1990; Dargemont and Kuhn, 1992). However, since the export of RNAs may be controlled at multiple levels within the nucleus, it has been difficult to determine the signals involved in this process and the nuclear site(s) where these signals exert their effect. Nevertheless, some of the signals and factors mediating nuclear export of RNAs and RNP particles are starting to emerge. In the case of mRNAs and snRNAs, it has been shown that the monomethylated RNA cap structures facilitate their nuclear export; thus the signal(s) may reside on the primary structure of the RNA (Hamm and Mattaj, 1990). Moreover, a nuclear cap-binding protein that might play a role similar to the NLS receptors in nuclear protein import (see Section VB) has been identified (Izaurralde et al., 1992). More recently, Jarmolowski et al. (1994) have demonstrated that the transport of various classes of RNAs is mediated by distinct rather than common essential factors. Thus export of RNAs and RNP particles may occur by different signaling pathways, as does the nuclear import of proteins and RNP particles (see above).
V. CONCLUSIONS AND FUTURE PROSPECTS Considering the size and complexity of the NPC, significant progress has been made over the past few years in determining its 3-D architecture and molecular composition, and in describing the mechanisms of mediated transport through the NPC. As a consequence, the 3-D structure of its basic framework has now been
Nuclear Pore Complex
41
determined to a resolution of just under 10 nm (see Fig. 4). Several distinct peripheral NPC components such as the cytoplasmic and nuclear ring, the cytoplasmic filaments and the nuclear basket have been identified (see Fig. 8); and about two dozen of its protein constituents have been characterized and are starting to be localized within the 3-D structure of the NPC (see Fig. 15). Some of the signals, receptors, and factors mediating nuclear import of proteins and RNP particles and nuclear export of RNAs and RNP particles have also been identified. Nevertheless, there remain a number of questions concerning the 3-D structure, chemical composition, and functional role(s) of the different structural components of the NPC. Recently there has been some progress in localizing the epitopes of several NPC proteins with distinct structural components of the NPC by immuno-EM (see Figs. 11—13 and 15). Thus the molecular architecture of the NPC is slowly but definitely emerging. However, even if there are several copies (i.e., 8 or 16 because of the 8-2-2 symmetry of the basic framework of the NPC) of the NPC proteins thus far identified, they represent only --15% of the NPC mass. Therefore, we have a long way to go before the complete architecture of the NPC will be unveiled at the molecular level. Toward this goal, the recent success of isolating NPCs in bulk from yeast (Rout and Blobel, 1993) has opened the possibility of more systematically identifying the protein constituents of yeast NPCs. Most importantly, this system offers the advantage of combining molecular genetics approaches with biochemical, structural, and functional analyses of the NPC. The next step toward a more complete molecular architecture of the NPC will be not only to identify NPC proteins with distinct NPC components, but to map individual proteins within the 3-D structure of the NPC and to determine their conformation and specific interactions with other proteins residing within distinct NPC components. Moreover, to eventually reconstitute functional NPCs in vitro, we also have to isolate and molecularly characterize distinct NPC components (e.g., the spoke complex, the cytoplasmic and nuclear rings, the cytoplasmic filaments, the nuclear basket, and the central plug or channel complex), determine their 3-D molecular architecture, and decipher their functional roles in terms of distinct steps implicated in mediated nucleocytoplasmic transport. Toward this goal, several NPC subcomplexes have recently been identified: i.e., the p62 complex, the p250-p75 complex, the NUP153 homo-oligomer, the yeast NSPl complex, and the yeast NUPl—NUP2-Srpl complex (see Section IVD). However, these complexes remain to be further characterized at the structural and molecular levels. The recent chemical characterization of some of the peripheral components of the NPC (i.e., the cytoplasmic filaments and the nuclear basket; see Figs. 11—13 and 15) opens the possibility of testing their functional properties and determining the functional roles of distinct NPC components. For example, the hypothesis that the cytoplasmic filaments might be initial docking sites for nuclear import of NLS-bearing proteins (see Fig. 16) can now be tested by the use of antibodies directed against the constituent proteins of these filaments (i.e., Tpr/p265 and CAN/NUP214/p250). Similarly, the role of the nuclear basket in the nuclear export
42
NELLY PANTE and UELI AEBI
of RNAs and RNP particles can be tested by using antibodies directed against NUP153, a constituent of the nuclear basket. In the effort to understand the mechanism of mediated nuclear import of proteins and RNP particles, the development of a digitonin-permeabilized cell system to study these processes in vitro (Adam et al., 1990) has made possible the identification of a number of signals, receptors, and factors mediating nuclear import of proteins and RNP particles. However, the NPC site(s) where they exert their mechanism of action remain to be established. Similarly, the signals and factors that mediate nuclear export of RNAs have begun to be elucidated, but the nuclear site(s) where they exert their effect is unclear. Nevertheless, these results have revealed the existence of multiple nuclear import-export pathways that might be specific for certain cell types or stages of differentiation. Finally, the recent success in imaging and at the same time manipulating NPCs in their native buffer environment by scanning force microscopy (see Fig. 5, c and d) provides us with the exciting possibility of directly correlating NPC structure with function.
ACKNOWLEDGMENTS The authors are indebted to C. Henn for designing and preparing Figures 1,4, 8, and 15. We thank Dr. R. Milligan (Scripps Research Institute, La Jolla) who provided the data of the 3-D reconstruction of negatively stained detergent-released NPCs that enabled us to produce Figures 4 , 8 , and 15. We are grateful to Dr. B. Burke (Harvard Medical School, Boston) and Dr. L. Gerace (Scripps Research Institute, La Jolla) for providing us with several anti-NPC antibodies. We thank Ms. U. Sauder for her help with embedding and thin sectioning; R. Wyss for help with Figures 9,10,14, and 16; and K. N. Goldie for providing the micrographs for Figure 5, c and d. We thank Ms. H. Frefel and Ms. M. Zoller for their expert photographic work. This work was supported by the M. E. Muller Foundation of Switzerland, by grants from the Swiss National Science Foundation and the Human Frontier Science Program (HFSP).
REFERENCES Adam, J. H. & Adam, S. A. (1994). Identification of cytosolic factors required for nuclear localization sequence-mediated binding to the nuclear envelope. J. Cell Biol. 125, 547-555. Adam, S. A. & Gerace, L. (1991). Cytosolic proteins that specifically bind nuclear localization signals are receptors for nuclear import. Cell 66, 837-847. Adam, S. A., Lobl, T. J., Mitchell, M. A., & Gerace, L. (1989). Identification of specifically binding proteins for a nuclear location sequence. Nature (London) 337, 276—279. Adam, S. A., Steme-Marr, R., & Gerace, L. (1990). Nuclear protein import in permeabilized mammalian cells requires soluble cytoplasmic factors. J. Cell Biol. 111, 807-816. Aebi, U., Cohn, J., Buhle, L., & Gerace, L. (1986). The nuclear lamina is a meshwork of intermediatetype filaments. Nature (London) 323, 560-564. Akey, C. W. (1989). Interactions and structure of the nuclear pore complex revealed by cryo-electron microscopy. J. Cell Biol. 109, 955-970. Akey, C. W. (1990). Visualization of transport-related configurations of the nuclear pore transporter. Biophys. J. 58, 341-355.
Nuclear Pore Complex
43
Akey, C. W. & Goldfarb, D. S. (1989). Protein import through the nuclear pore complex is a multistep process. J. Cell Biol. 109, 971-982. Akey, C. W. & Radermacher, M. (1993). Architecture of the Xenopus nuclear pore complex revealed by three-dimensional cryo-electron microscopy. J. Cell Biol. 122, 1—19. Bataille, N., Helser, T., & Fried, H. M. (1990). Cytoplasmic transport of ribosomal subunits microinjected into the Xenopus laevis oocyte nucleus: A generalized, facilitated process. J. Cell Biol. Ill, 1571-1582. Belanger, K. D., Kenna, M. A., Wei, S., & Davis, L. (1994). Genetic and physical interactions between Srplp and nuclear pore complex proteins NUPlp and NUP2p. J. Cell Biol. 126, 619-630. Buss, F., Kent, H., Stewart, M., Bailer, S. M., & Hanover, J. A. (1994). Role of different domains in the self-association of rat nucleoporin p62. J. Cell Sci. 107, 631-638. Byrd, D., Sweet, D. J., Pante, N., Konstantinov, K. N., Guan, T., Saphire, A. C. S., Mitchell, R J., Cooper, C. S., Aebi, U., & Gerace, L. (1994). Tpr, a large coiled coil protein whose amino terminus is involved in activation of oncogenic kineses, is localized to the cytoplasmic surface of the nuclear pore complex. J. Cell Biol. 127, 1515-1526. Carmo-Fonseca, M., Kern, H., & Hurt, E. C. (1991). Human nucleoporin p62 and the essential yeast nuclear pore protein NSPl show sequence homology and a similar domain organization. Eur. J. Cell Biol. 55, 17-30. Chelsky, D., Ralph, R., & Jonak, G. (1989). Sequence requirements for synthetic peptide-mediated translocation to the nucleus. Mol. Cell Biol. 9, 2487-2492. Chugani, D. C, Rome, L. H., & Kedersha, N. L. (1993). Localization of vault particles to the nuclear pore complex. J. Cell Sci. 106, 23-29. Coleman, J. E. (1992). Zinc proteins: Enzymes, storage proteins, transcription factors, and replication proteins. Annul Rev. Biochem. 61, 897-946. Cordes, V., Waizenegger, I., & Krohne, G. (1991). Nuclear pore complex glycoprotein p62 oiXenopus laevis and mouse: cDNA cloning and identification of its glycosylation region. Eur. J. Cell Biol. 55,31^7. Cordes, V., Reidenbach, S., Kohler, A., Stuurman, N., van Driel, R., & Franke, W. W. (1993). Intranuclear filaments containing a nuclear pore complex protein. J. Cell Biol. 123, 1333-1344. Dabauvalle, M.-C, Benevente, R., & Chaly, N. (1988a). Monoclonal antibodies to a M^ 68,000 pore complex protein interfere with nuclear protein uptake in Xenopus oocytes. Chromosoma 97, 193-197. Dabauvalle, M.-C, Schultz, B., Scheer, U., & Peters, R. (1988b). Inhibition of nuclear accumulation of karyophilic proteins by microinjection of the lectin WGA. Exp. Cell Res. 174, 291—296. Dabauvalle, M.- C, Loos, K., & Scheer, U. (1990). Identification of a soluble precursor complex essential for nuclear pore assemble in vitro. Chromosoma 100, 56-66. Dargemont, C. & Kiihn, L. C. (1992). Export of mRNA from microinjected nuclei oi Xenopus laevis oocytes. J. Cell Biol. 118, 1-9. Davis, L. I. & Blobel, G. (1986). Identification and characterization of a nuclear pore complex protein. Cell 45, 699-709. Davis, L. I. & Blobel, G. (1987). The nuclear pore complex contains a family of glycoproteins that includes p62: Glycosylation through a previously unidentified cellular pathway. Proc. Natl. Acad. Sci. USA 84, 7552-7556. Davis, L. I. & Fink, G. R. (1990). The NUPl gene encodes an essential component of the yeast nuclear pore complex. Cell 61,965-978. Dessev, G., lovcheva-Dessev, C, Bischoff, J. R., Beach, D., & Goldman, R. (1991). A complex containing p34^ ^ and cyclin B phosphorylates the nuclear lamin and disassembles nuclei of clam oocytes in vitro. J. Cell Biol. 112, 523-533. Dingwall, C. & Laskey, R. A. (1991). Nuclear targeting sequences—^A consensus? Trends Biochem. Sci. 16,478-481. Fabre, E. & Hurt, E. C. (1994). Nuclear transport. Curr. Opin. Cell Biol. 6, 335-342.
44
NELLY PANTE and UELI AEBI
Fabre, E., Boelens, W. C, Wimmer, C, Mattaj, I W., & Hurt, E. C. (1994). Nupl45p is required for nuclear export of mRNA and binds homopolymeric RNA in vitro via a novel conserved motif. Cell 78,275-289. Featherstone, C, Darby, M. K., & Gerace, L. (1988). A monoclonal antibody against the nuclear pore complex inhibits nucleocytoplasmic transport of protein and RNA in vivo. J. Cell Biol. 107, 1289-1297. Feldherr, C. M., Kallenbach, E., & Schultz, N. (1984). Movement of a karyophilic protein through the nuclear pores of oocytes. J. Cell Biol. 99, 2216-2222. Finlay, D. R., Newmeyer, D. D., Price, T. M., & Forbes, D. J. (1987). Inhibition of in vitro nuclear transport by a lectin that binds to nuclear pores. J. Cell Biol. 104, 189-200. Finlay, D. R., Meier, E., Bradley, P., Horecka, J., & Forbes, D. J. (1991). A complex of nuclear pore proteins required for pore function. J. Cell Biol. 114, 169-183. Fischer, U. & Luhrmann, R. (1990). An essential signaling role for the m G cap in the transport of Ul snRNPs to the nucleus. Science 249, 786-790. Fischer, U., Sumpter, V., Sekine, M., Satoh, T., & Luhrmann, R. (1993). Nucleocytoplasmic transport of U snRNPs: Definition of a nuclear localization signal in the Sm core domain that binds a transport receptor independently of the m G cap. EMBO J. 12, 573-583. Forbes, D. J. (1992). Structure and function of the nuclear pore complex. Annul Rev. Cell Biol. 8, 495-527. Franke, W. W. (1974). Structure, biochemistry and functions of the nuclear envelope. Int. Rev. Cytol. Suppl. 4, 71-236. Franke, W. W. & Scheer, U. (1970a). The ultrastructure of the nuclear envelope of amphibian oocytes: A reinvestigation. I. The mature oocyte. J. Ultrastruct. Res. 30, 288-316. Franke, W. W. & Scheer, U. (1970b). The ultrastructure of the nuclear envelope of amphibian oocytes: A reinvestigation. II. The immature oocyte and dynamic aspects. J. Ultrastruct. Res. 30, 317-327. Garcia-Bustos, J., Heitman, J., & Hall, M. N. (1991). Nuclear protein localization. Biochim. Biophys. Acta 1071, 83-101. Gerace, L. (1992). Molecular trafficking across the nuclear pore complex. Curr. Opin. Cell Biol. 4, 637-645. Gerace, L. & Blobel, G. (1980). The nuclear envelope lamina is reversibly depolymerized during mitosis. Cell 19,277-287. Gerace, L. & Burke, B. (1988). Functional organization of the nuclear envelope. Annu. Rev. Cell Biol. 4, 335-374. Gerace, L., Ottaviano, Y., & Kondor-Koch, C. (1982). Identification of a major polypeptide of the nuclear pore complex. J. Cell Biol. 95, 826-837. Goldberg, M. W. & Allen, T. D. (1992). High resolution scanning electron microscopy of the nuclear envelope: Demonstration of a new, regular, fibrous lattice attached to the baskets of the nucleoplasmic face of the nuclear pores. J. Cell Biol. 119, 1429-1440. Goldberg, M. W. & Allen, T. D. (1993). The nuclear pore complex: Three-dimensional surface structure revealed by field emission, in-lens scanning electron microscopy, with underlaying structure uncovered by proteolysis. J. Cell Sci. 106, 261-274. Goldie, K. N., Pante, N., Engel, A., & Aebi, U. (1994). Exploring native nuclear pore complex structure and conformation by scanning force microscopy in physiological buffers. J. Vac. Sci. Technol. B 12, 1482-1485. Grandi, P., Doye, V., & Hurt, E. C. (1993). Purification of NSPl reveals complex formation with "GLFG" nucleoporins and a novel nuclear pore protein NIC96. EMBO J. 12, 3061-3071. Greber, U. F. & Gerace, L. (1992). Nuclear protein import is inhibited by an antibody to a lumenal epitope of a nuclear pore complex glicoprotein. J. Cell Biol. 116, 15—30. Greber, U. F., Senior, A., & Gerace, L. (1990). A major glycoprotein of the nuclear pore complex is a membrane-spanning polypeptide with a large lumenal domain and a small cytoplasmic tail. EMBO J. 9, 1495-1502.
Nuclear Pore Complex
45
Guan, T., Muller, S., Klier, G., Pante, N., Blevitt, J. M., Haner, M., Paschal, B., Aebi, U., & Gerace, L. (1995). Structural analysis of the p62 complex, an assembly of 0-linked glycoproteins that localizes near the central gated channel of the nuclear pore complex. Mol. Biol. Cell 6,1591-1603. Hallberg, E., Wozniak, R. W., & Blobel, G. (1993). An integral membrane protein of the pore membrane domain of the nuclear envelope contains a nucleoporin-like region. J. Cell Biol. 122, 513-521. Hamm, J. & Mattaj, I. W. (1990). Monomethylated cap strucmres facilitate RNA export from the nucleus. Cell 63, 109-118. Hamm, J., Darzynkiewicz, E., Tahara, S. M., & Mattaj, I. W. (1990). The trimethyguanosine cap structure of Ul snRNA is a component of a bipartite nuclear targeting signal. Cell 62, 569-577. Hanover, J. A., Cohen, C. K., Willingham, M. C, & Park, M. K. (1987). O-linked N-acetylglucosamine is attached to proteins of the nuclear pore. J. Biol. Chem. 262, 9887-9894. Hinshaw, J. E., Carragher, B. O., & Milligan, R. A. (1992). Architecture and design of the nuclear pore complex. Cell 69, 1133-1141. Holt, G. D., Snow, C. M., Senior, A., Haltiwanger, R. S., Gerace, L., & Hart, G. W. (1987). Nuclear pore complex glycoproteins contain cytoplasmically disposed O-linked N-acetylglucosamine. J. Cell Biol. 104, 1157-1164. Hurt, E. C. (1988). A novel nucleoskeletal-like protein located at the nuclear periphery is required for the life cycle oiSaccharomyces cerevisiae. EMBO J. 7, 4323—4334. Hurt, E. C. (1990). Targeting of a cytosolic protein to the nuclear periphery. J. Cell Biol. Ill, 2829-2837. Imamoto, N., Matsuoka, Y., Kurihara, T, Kohno, K., Miyagi, M., Sakiyama, F., Okada, Y., Tsunasawa, S., & Yoneda, Y. (1992). Antibodies against 70-kD heat shock cognate protein inhibit mediated nuclear import of karyophilic proteins. J. Cell Biol. 119, 1047—1061. Izaurralde, E. & Mattaj, I. W. (1992). Transport of RNA between nucleus and cytoplasm. Semin. Cell Biol. 3, 279-288. Izaurralde, E., Stepinski, J., Darzynkiewicz, E., & Mattaj, I. W. (1992). A cap binding protein that may mediate nuclear export of RNA polymerase Il-transcribed RNAs. J. Cell Biol. 118, 1287—1295. Jarmolowski, A., Boelens, W. C, Izaurralde, E., & Mattaj, I. W. (1994). Nuclear export of different classes of RNA is mediated by specific factors. J. Cell. Bio. 124, 627-635. Jamik, M. & Aebi, U. (1991). Toward a more complete 3-D structure of the nuclear pore complex. J. Struct. Biol. 107,291-308. Kedersha, N. L., Heuser, J. E., Chugani, D. C, & Rome, L. H. (1991). Vaults. III. Vault Ribonucleoprotein particles open into flower-like structures with octagonal symmetry. J. Cell Biol. 112,225-235. Kita, K., Omata, S., & Horigome, T. (1993). Purification and characterization of a nuclear pore glycoprotein complex containing p62. J. Biochem. (Tokyo) 113, 377-382. Kraemer, D., Wozniak, R. W., Blobel, G., & Radu, A. (1994). The human CAN protein, a putative oncogene product associated with myeloid leukemogenesis, is a nuclear pore complex protein that faces the cytoplasm. Proc. Natl. Acad. Sci. USA 91, 1519-1523. Loeb, J. D. J., Davis, L., & Fink, G. F. (1993). NUP2, a novel yeast nucleoporin, has functional overlap with others proteins of the nuclear pore complex. Mol. Biol. Cell. 4, 209-222. Marshallsay, C. & Liihrmann, R. (1994). In vitro nuclear import of snRNPs: Cytosolic factors mediate m^G-cap dependence of Ul and U2 snRNP transport. EMBO J. 13, 222-231. Mattaj, I. W., Boelens, W., Izaurralde, E., Jarmolowski, A., & Kambach, C. (1993). Nucleocytoplasmic transport and snRNP assembly. Mol. Biol. Rep. 18, 79-83. Maul, G. G. (1977). The nuclear and cytoplasmic pore complex. Structure, dynamics, distribution and evolution. Int. Rev. Cytol. Suppl. 6, 75-186. McMorrow, I. M., Bastos, R., Horton, R., & Burke, B. (1994). Sequence analysis of a cDNA encoding a human nuclear pore complex protein, hnupl 53. Biochim. Biophys. Acta 1217, 219-223. Mehlin, H., Daneholt, B., & Skoglund, U. (1992). Translocation of a specific premessenger ribonucleoprotein particle through the nuclear pore studied with electron microscope tomography. Cell 69, 605-613.
46
NELLY PANTE and UELI AEBI
Melchior, F., Paschal, B., Evans, J., & Gerace, L. (1993). Inhibition of nuclear protein import by nonhydrolyzable analogues of GTP and identification of the small GTPase Ran/TC4 as an essential transport factor. J. Cell Biol. 123, 1649-1659. Michaud, N. & Goldfard, D. (1991). Multiple pathways in nuclear transport: the import of U2 snRNP occur by a novel kinetic pathway. J. Cell Biol. 112, 215-223. Michaud, N. & Goldfard, D. (1992). Microinjected U snRNAs are imported to oocyte nuclei via the nuclear pore complex by three distinguishable targeting pathways. J. Cell Biol. 116, 851-861. Mitchell, P. J. & Cooper, C. S. (1992). Nucleotide sequence analysis of human tpr cDNA clones. Oncogene 7, 383-388. Moore, M. S. & Blobel, G. (1992). The two steps of nuclear import, targeting to the nuclear envelope and translocation through the nuclear pore, require different cytosolic factors. Cell 68, 939-950. Moore, M. S. & Blobel, G. (1993). The GTP-binding protein Ran/TC4 is required for protein import into the nucleus. Nature (London) 365, 661-663. Newmeyer, D. D. (1993). The nuclear pore complex and nucleocytoplasmic transport. Curr. Opin. Cell Biol. 5, 395-407. Newmeyer, D. D. & Forbes, D. J. (1988). Nuclear import can be separated into distinct steps in vitro: Nuclear pore binding and translocation. Cell 52, 641-653. Newmeyer, D. D. & Forbes, D. J. (1990). An N-ethylmaleimide-sensitive cytosolic factor necessary for nuclear protein import: Requirement in signal-mediated binding to the nuclear pore. J. Cell Biol. 110,547-557. Newmeyer, D. D., Finlay, D. R., & Forbes, D. J. (1986). In vitro transport of a fluorescent nuclear protein and exclusion of non-nuclear proteins. J. Cell Biol. 103, 2091—2102. Ottaviano, Y. & Gerace, L. (1985). Phosphorylation of the nuclear lamins during interphase and mitosis. J. Biol. Chem. 260, 624^32. Paine, P. L., Moore, L. C, & Horowitz, S. B. (1975). Nuclear envelope permeability. Nature (London) 254 10^114. Pante, N. & Aebi, U. (1993). The nuclear pore complex. J. Cell Biol. 122, 977-984. Pante, N. & Aebi, U. (1994). Towards understanding the 3-D structure of the nuclear pore complex at the molecular level. Curr. Opin. Struct. Biol. 4, 187-196. Pante, N., Bastos, R., McMorrow, I., Burke, B., & Aebi, U. (1994). Interactions and three-dimensional localization of a group of nuclear complex proteins. J. Cell Biol. 126, 603-617. Peter, M., Nakagawa, J., Doree, M., Labbe, J. C, & Nigg, E. A. (1990). In vitro disassembly of the nuclear lamina and M phase-specific phosphorylation of lamins by cdc2 kinase. Cell 61,591-602. Peter, M., Heitliger, E., Haner, M., Aebi, U., & Nigg, E. A. (1991). Disassembly of in vitro formed lamin head-to-tail polymers by CDC2 kinase. EMBO J. 10, 1535-1544. Radu, A., Blobel, G., & Wozniak, R. W. (1993). Nupl55 is a novel nuclear pore complex protein that contains neither repetitive sequence motifs nor reacts with WGA. J. Cell Biol. 121, 1—9. Radu, A., Blobel, G., & Wozniak, R. W. (1994). Nupl07 is a novel nuclear pore complex protein that contains a leucine zipper. J. Biol. Chem. 269, 17600-17605. Reichelt, R., Holzenburg, A., Buhle, E. L., Jamik, M., Engel, A., & Aebi, U. (1990). Correlation between structure and mass distribution of the nuclear pore complex, and of distinct pore complex components. J. Cell Biol. 110, 883-894. Richardson, W. D., Mills, A. D., Dilworth, S. M., Laskey, R. A., & Dingwall, C. (1988). Nuclear protein migration involves two steps: Rapid binding at the nuclear envelope followed by slower translocation through the nuclear pores. Cell 52, 655-664. Ris, H. (1991). The 3-D structure of the nuclear pore complex as seen by high voltage electron microscopy and high resolution low voltage scanning electron microscopy. EMSA Bull. 21,54-56. Robbins, J., Dilworth, S. M., Laskey, R. A., & Dingwall, C. (1991). Two interdependent basic domains in nucleoplasmin nuclear targeting sequence: Identification of a class of bipartite nuclear targeting sequence. Cell 64, 615-623.
Nuclear Pore Complex
47
Roos, U.-P. (1973). Light and electron microscopy of rat kangaroo cells in mitosis 1: Formation and breakdown of the mitotic apparatus. Chromosoma 40, 43-82. Rout, M. R & Blobel, G. (1993). Isolation of the yeast nuclear pore complex. J. Cell Biol. 123, 771-783. Shi, Y. & Thomas, J. O. (1992). The transport of proteins into the nucleus requires the 70-kilodalton head shock protein or its cytoplasmic cognate. Mol. Cell Biol. 12, 2186-2192. Snow, C. M., Senior, A., & Gerace, L. (1987). Monoclonal antibodies identify a group of nuclear pore complex glycoproteins. J. Cell Biol. 104, 1143-1156. Soullan, B. & Worman, H. J. (1993). The amino-terminal domain of the lamin B receptor is a nuclear envelope targeting signal. J. Cell Biol. 120, 1093-1100. Starr, C. M., D'Onofrio, M., Park, M. K., & Hanover, J. A. (1990). Primary sequence and heterologous expression of nuclear pore glycoprotein p62. J. Cell Biol. 110, 1861—1871. Steme-Marr, R., Blevitt, J. M., & Gerace, L. (1992). 0-linked glycoproteins of the nuclear pore complex interact with a cytosolic factor required for nuclear protein import. J. Cell Biol. 116, 271—280. Stevens, B. J. & Swift, H. (1966). RNA transport from nucleus to cytoplasm in Chironomus salivary glands. J. Cell Biol. 31, 55-77. Stewart, M. & Whytock, S. (1988). The structure and interactions of components of nuclear envelopes from Xenopus oocyte germinal vesicles observed by heavy metal shadowing. J. Cell Sci. 90,409-423. Sukegawa, J. & Blobel, G. (1993). A nuclear pore complex protein that contains zinc finger motifs, binds DNA, and faces the nucleoplasm. Cell 72, 29-38. Unwin, P. N. T & Milligan, R. A. (1982). A large particle associated with the perimeter of the nuclear pore complex. J. Cell Biol. 93, 63—75. Von Lindem, M., Fomerod, M., van Baal, S., Jaegle, M., de Wit, T, Buijs, A., & Groveld, G. (1992). The translocation (6;9), associated with a specific subtype of acute myeloid leukemia, results in the fusion of two genes, dek and can, and the expression of a chimeric, leukemia-specific dek-can mRNA. Mol. Cell Biol. 12, 1687-1697. Ward, G. E. & Kirschner, M. W (1990). Identification of cell cycle-regulated phosphorylation sites on nuclear lamin C. Cell 61, 561-577. Wente, S. R. & Blobel, G. (1993). A temperature-sensitive NUP116 null mutant forms a nuclear envelope seal over the yeast nuclear pore complex thereby blocking nucleocytoplasmic traffic. J. Cell Biol. 123,275-284. Wente, S. R. & Blobel, G. (1994). NUP 145 encodes a novel yeast glycine-leucine-phenylalanine-glycine (GLFG) nucleoporin required for nuclear envelope structure. J. Biol. 125, 955—969. Wente, S. R., Rout, M. P., & Blobel, G. (1992). A new family of yeast nuclear pore complex proteins. J. Cell Biol. 119,705-723. Wiese, C. & Wilson, K. L. (1993). Nuclear membrane dynamics. Curr. Opin. Cell Biol. 5, 387-394. Wilken, N., Kossner, U., Senecal, J.-L., Scheer, U., & Dabauvalle, M.-C. (1993). Nupl80, a novel nuclear pore complex protein localizing to the cytoplasmic ring and associatedfibrils.J. Cell Biol. 123,1345-1354. Wimmer, C , Doye, V., Grandi, P., Nehrbass, U., & Hurt, E. C. (1992). A new subclass of nucleoporins that functionally interact with nuclear pore protein NSPl. EMBO J. 11, 5051—5061. Wozniak, R. W & Blobel, G. (1992). The single transmembrane segment of gp210 is sufficient for sorting to the pore membrane domain of the nuclear envelope. J. Cell Biol. 119, 1441—1449. Wozniak, R. W., Bartnik, E., & Blobel, G. (1989). Primary structure analysis of an integral membrane glycoprotein of the nuclear pore. J. Cell Biol. 108, 2083-2092. Wozniak, R. W, Blobel, G., & Rout, M. R (1994). POM 152 is an integral protein of the pore membrane domain of the yeast nuclear envelope. J. Cell Biol. 125, 31—42. Yano, R., Oakes, M., Yamaghishi, M., Dodd, J. A., & Nomura, M. (1992). Cloning and characterization of SRPl, a suppressor of temperature-sensitive RNA polymerase I mutations, in Saccharomyces cerevisiae. Mol. Cell. Biol. 12, 5640-5641. Zasloff, M. (1983). tRNA transport form the nucleus in a eukaryotic cell: Carrier-mediated translocation process. Proc. Natl. Acad. Sci. USA 80, 6436-6440.
This Page Intentionally Left Blank
STRUCTURE AND FUNCTION OF MITOCHONDRIAL PRESEQUENCES
Merritt Maduke and David Roise
Abstract I. Introduction A. Protein Import into the Mitochondrial Matrix B. Mitochondrial Presequences II. Structure of Presequences III. Surface Activity A. Insertionof Presequences into Monolayers B. Disruptionof Phospholipid Vesicles C. Effectof Length on the Properties of Presequences IV. Binding to Membranes A. Binding to Phospholipid Vesicles B. Binding to Mitochondrial Membranes C. Effects of Surface Potential on Binding to Membranes V. Import into Mitochondria VI. Import Into Phospholipid Vesicles VII. Model for Recognition of Mitochondrial Presequences A. Bindingof Precursors to Membranes/« Vivo B. TheMembranePotentialMay Determine Specificity of Import
Membrane Protein TVansport Volume 3, pages 49-79. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 49
50 50 50 52 55 56 56 56 57 58 58 59 62 65 68 70 70 71
50
MERRITT MADUKE and DAVID ROISE
C. Role ofMitochondrial Membrane Proteins VIII. Summary Acknowledgments References
72 74 74 75
ABSTRACT Mitochondrial presequences are responsible for the specific targeting of proteins to the matrix of the mitochondria. The presequences lack sequence homology but share common physical characteristics—positive charge and amphiphilicity—which make them ideally suited for interacting with the mitochondrial outer membranes. Structural studies have demonstrated that presequences adopt a-helical structure in the presence of negatively charged surfaces and that the a-helices are amphiphilic. Quantitative studies using synthetic presequences have shown that the presequences bind to lipid vesicles and to the lipids of mitochondrial membranes, and that the binding to these membranes is similar. Other studies have shown that the import of mitochondrial presequences into isolated yeast mitochondria and their import into protein-free phospholipid vesicles are also similar. These results are discussed, and a model for the recognition of mitochondrial presequences within cells is presented.
1. INTRODUCTION The goal of this chapter is to summarize the results of research performed in our laboratory on mitochondrial presequences and to describe how these results are relevant to the mechanism of transport of proteins into mitochondria. Our work has been directed at elucidating the mechanism of import of presequences into mitochondria. Understanding, in detail, how presequences interact with and are imported into mitochondria should serve as a first step toward understanding the mechanism of import of full-length precursor proteins into the organelle. Our results demonstrate that mitochondrial presequences, because of their physical characteristics, are ideally suited for their role as targeting signals for proteins destined for the mitochondria. A. Protein Import into the Mitochondrial Matrix
Most mitochondrial proteins are synthesized in the cytoplasm and must be imported into the organelle. Proteins that are destined for the interior of the mitochondria are usually synthesized as longer precursor forms with amino-terminal extensions. These extensions, called presequences, are responsible for the specific targeting of the protein to the mitochondria and are proteolysed once inside the matrix. When expressed on hybrid proteins through the use of gene fusions, mitochondrial presequences are capable of directing the import of even nonmito-
Mitochondrial Presequences
51
chondrial proteins into the mitochondria (Hurt et al., 1984; Horwich et al., 1985; Hurt et al., 1985b; Hurt and van Loon, 1986). A diagram representing the import of a precursor protein into the mitochondrial matrix space is shown in Figure 1. Listed in the figure are components that are involved in the protein import process. Cytosolic ATP and heat shock proteins are required for the import of many proteins (Deshaies et al., 1988; Murakami et al.. 1988; Gething and Sambrook, 1992; Stuart et al., 1994), presumably to maintain the precursors in a conformation that is competent for being imported (Eilers and Schatz, 1988). Other cytosolic factors may be involved in protein import (Firgaira et al., 1984; Ohta and Schatz, 1984; Ono and Tuboi, 1988, 1990a), although it has been observed that chemically pure precursors can be efficiently imported into purified mitochondria in the absence of added factors (Eilers and Schatz, 1986; Becker et al., 1992). The cytosolic factors may act to stabilize the precursor and prevent aggregation, and they thus may not be necessary for protein translocation under all conditions. Several outer and inner mitochondrial membrane proteins have been implicated in the import process and are listed in the figure. The proteins identified in yeast have been named Mas (for mitochondrial assembly) and ISP (for import site protein) (Schatz, 1993); the proteins from Neurospora crassa have been named either MOM (for mitochondrial outer membrane) or MIM (for mitochondrial inner
BINDING
\
CYTOSOL ATP Hsp70 other soluble factors?
OUTER MEMBRANE Mas20p/MOM19 Mas70p/MOM72 ISP42/MOM38 MOM30, MOMS. MOM7. MOM22 INNER MEMBRANE ISP45/MIM44 Mas6p/MIM23
MATRIX Hsp60 mHsp70 MPPa.p ATP
Figure 1. Protein import into the mitochondrial matrix. See text for explanation of steps and components.
52
MERRITT MADUKE and DAVID ROISE
membrane) (Kiebler et al., 1993); the numbers in each case refer to the apparent molecular mass (kDa) of each protein. These proteins have been identified by biochemical and/or genetic approaches. In contrast to the protein translocation process in the endoplasmic reticulum, which has recently been reconstituted from a limited number of purified components (Gorlich and Rapoport, 1993), none of the mitochondrial membrane proteins has yet been studied in a purified and reconstituted system. The exact functions of the components are, therefore, still debatable. The mitochondrial membrane potential (A^) also plays an important role in the translocation of precursors across the inner membrane. It is required for the initiation of transport, presumably because it induces electrophoresis of the positively charged presequences across the inner membrane (Pfanner and Neupert, 1985; Martin et al., 1991). Inside the matrix, ATP and heat shock proteins are required to facilitate protein folding and to limit aggregation (Stuart et al., 1994). The subunits of the mitochondrial processing peptidase (MPP-a, MPP-P) (Kalousek et al., 1993) are required for cleavage of the presequence. Cleavage of the presequence is not necessary for some proteins and in some cases may occur prior to the complete translocation of the precursor into the matrix space (Schleyer and Neupert, 1985). As shown in the model in Figure 1, an early step in the import of a precursor is likely to be the insertion of the presequence, as an a-helix, into the lipids of the mitochondrial outer membrane. This suggestion is consistent with many studies on mitochondrial presequences, which will be discussed. A precursor protein, bound to the membrane as shown, would diffuse laterally in the outer membrane until it reaches a site of translocation. For proteins destined for the matrix space, translocation occurs at contact sites, where the outer and inner mitochondrial membranes are closely associated (Schleyer and Neupert, 1985). It should be noted that the detailed structure of the membrane at the contact sites is unknown, and it is not clear whether these contact sites represent stable structures, or whether they form transiently as precursors are in the process of being transported (Schwaiger et al., 1987; Pon et al., 1989; Pfanner et al., 1990). Indeed, under some conditions the two membranes can become uncoupled as a precursor is being imported, and the protein may only reach the intermembrane space (Hwang et al., 1991; Wachter et al., 1994). It is not yet known whether transport across the two membranes at the contact site occurs through a translocation pore, directly through the lipid bilayer, or through some combination of protein and lipid; hence, a question mark is shown in the model. B. Mitochondrial Presequences Mitochondrial presequences are rich in positively charged residues and hydroxylated residues, and they generally lack negatively charged residues. They do not, however, contain any obvious consensus sequence. The lack of a consensus
Mitochondrial Presequences
53
sequence makes it puzzling how mitochondria specifically recognize the targeting information contained within the hundreds of matrix-targeted proteins. Some clues to this enigma come from studies on the physical properties of presequences. Roughly 10 years ago, it was noted that the presequence of yeast cytochrome oxidase subunit IV (CoxIV) could form an amphiphilic a-helix (Roise et al., 1986). This type of structure is common among polypeptides that interact with surfaces (Kaiser and Kezdy, 1983; Roise, 1993). Although the primary sequence of the CoxIV presequence lacks any clustering of hydrophobic or basic residues, a plot of the sequence on a helical wheel diagram shows that clustering of hydrophobic and basic residues occurs when the presequence forms an a-helix (Fig. 2). Experiments with synthetic peptides corresponding to segments of the CoxIV presequence confirmed that this sequence forms an amphiphilic a-helix (Roise et al., 1986). A peptide corresponding to the complete, 25-residue presequence was unstructured in aqueous solution but formed an a-helical structure in the presence of negatively charged detergent micelles. The peptide was able to insert into phospholipid monolayers and to disrupt phospholipid bilayers. These results verified the prediction that the peptide would form an amphiphilic a-helix. To establish that amphiphilicity is a general property of mitochondrial targeting sequences, 23 presequences were compared, and the hydrophobic moment of each was predicted (von Heijne, 1986). The hydrophobic moment is a numerical evaluation of the amphiphilicity of an a-helix (Eisenberg, 1984). It is calculated as a
HaN-MetLeuSerLeuAi^GlnSerMArgPhePheLysProAlaThrArgT^
ile "let [ys-*"8 _ 1 _ is .
Figure 2. Presequence of the yeast cytochrome oxidase subunit IV (CoxIV). The primary structure is shown with charges indicated and hydrophobic residues underlined. In an a-helical conformation, the side chains of the presequence of CoxIV are arranged so that the positively charged residues lie on one face of the helix and the hydrophobic residues lie on the opposite face.
54
MERRITT MADUKE and DAVID ROISE
vector sum of the hydrophobicities of the amino acids in a sequence. The hydrophobic moment is used to provide a more quantitative prediction of amphilicity than does simple inspection of the sequence on a helical wheel diagram. The analysis by von Heijne revealed that all the mitochondrial presequences examined have the potential to form amphiphilic a-helices (helices of high hydrophobic moment). This observation suggested that formation of a surface-active, positively charged amphiphilic helix may be a universal property of mitochondrial presequences. Experiments with precursor proteins have confirmed that amphiphilicity is a general feature of functional matrix-targeting presequences. With genetic techniques the presequence of any precursor protein can be manipulated as desired. The effect of changes in the presequence on the ability of a precursor to be imported into mitochondria either in vivo or in vitro can then be determined. Numerous site-directed and random mutagenesis studies have been performed on various precursor proteins, and the results generally indicate that positive charge and amphiphilicity are important for presequence function (reviewed in Roise, 1993). Mutations that decrease positive charge or hydrophobicity, or that introduce residues that disrupt the helix (glycine or proline), decrease import efficiency. In an extremely thorough study, a strong correlation between the hydrophobic moment of a predicted a-helix and the efficiency of import of the attached protein was observed (Bedwell et al., 1989). To test the hypothesis that mitochondrial presequences are recognized because of their general physical properties and not because of a particular amino acid sequence, artificial presequences were fused to a CoxIV protein lacking its own presequence (Allison and Schatz, 1986). These presequences were designed to avoid resemblance to any particular mitochondrial presequence and were composed solely of arginine, leucine, and serine, residues that occur frequently in mitochonddrial presequences. Several of the artificial presequences were capable of directing attached proteins to the mitochondrial matrix. The functional presequences were later found to be amphiphilic, whereas the artificial sequences that lacked hydrophobic residues and were nonfunctional were not amphiphilic (Roise et al., 1988). In a complementary approach, random DNA fragments were fused to a truncated CoxIV gene (Lemire et al., 1989). The precursor proteins generated by these gene fusions were analyzed for their ability to be imported into mitochondria in yeast cells. Approximately 25% of the random presequences were found to be capable of directing the protein into the mitochondrial matrix. Examination of the primary sequences revealed that the functional presequences were all positively charged and had the potential to form amphiphilic a-helices. The predicted hydrophobic moment of a given sequence was strongly correlated with the efficiency of the sequence as a targeting signal. Taken together, the results demonstrate that mitochondria recognize positively charged, amphiphilic a-helices at the amino-termini of precursor proteins.
Mitochondrial Presequences
55
The rest of the chapter will summarize recent experiments that confirm the role of mitochondrial presequences as amphiphilic a-helices and that suggest that direct interactions between presequences and the lipid bilayer are involved in the import of precursors into mitochondria. A model will be presented that describes how the physical properties of presequences are important for the association of precursors with the mitochondrial surface and how the transmembrane potential across the mitochondrial inner membrane may be responsible for the specific uptake of precursor proteins by mitochondria in vivo.
II. STRUCTURE OF PRESEQUENCES To determine how presequences function, it is important to understand the structural and physical properties of the presequences. By using synthetic peptides corresponding to mitochondrial presequences, it has been possible to measure these properties directly. Circular dichroism (CD) measurements have been used to determine the a-helical content of several synthetic presequences. All of the presequences are random coils in solution but adopt varying degrees of a-helical structure in the presence of detergent micelles, small vesicles, or trifluoroethanol (TFE, a helix-stabilizing solvent) (Epand et al., 1986; Roise et al., 1986; Aoyagi et al., 1987; Tamm and Bartoldus, 1990; Hoyt et al., 1991). The surfaces (detergent or lipid) generally must be negatively charged to induce the formation of helical structure. Nuclear magnetic resonance (NMR) studies have supplemented the CD studies by identifying the specific residues that compose the helices of several presequences. The CoxIV presequence was shown by CD to be 40-50% a-helical in the presence of sodium dodecylsulfate micelles (Roise et al., 1986). Nuclear Overhauser effects (NOEs) indicated that residues 3—11 of the CoxIV presequence form interresidue a-helical contacts in the presence of dodecylphosphocholine micelles (Endo et al., 1989). In a similar environment, the presequence of rat liver aldehyde dehydrogenase displayed two helical regions (Karslake et al., 1990). The NOEs, together with the rates of amide proton exchange, indicated that the COOH-terminal helix is more stable than the NH2-terminal helix. In TFE, the presequence of the P-subunit of yeast Fj-ATPase also adopted a-helical structure, between residues 4 and 10 and between residues 14 and 19 (Bruch and Hoyt, 1992). With this presequence, the NH2-terminal helix is the more stable of the two. Finally, two synthetic peptides corresponding to the targeting segments of two precursors that are not proteolytically processed in the mitochondrial matrix have been studied using NMR (Hammen et al., 1994). Residues 4-21 from the targeting sequence of rhodanese and residues 4—14 from thiolase were found to be helical in the presence of TFE. The results for the thiolase presequence in TFE and in micelles were similar. The general conclusion from these studies is that mitochondrial presequences tend to form a-helices in the presence of an appropriate surface. This conclusion is
56
MERRITT MADUKE and DAVID ROISE
consistent with the predictions from helical wheel projections of functional presequences.
III. SURFACE ACTIVITY A. Insertion of Presequences into Monolayers
The ability to insert into membranes is thought to be an important feature of mitochondrial targeting signals. This behavior has been confirmed by studies measuring the surface activity of synthetic presequence peptides. One measure of surface activity is the insertion of synthetic presequences into phospholipid monolayers. Monolayers are formed at the air—water interface by spreading phospholipids on the surface of an aqueous buffered solution. When a presequence is injected into the solution below the monolayer, an increase in surface pressure is observed if the presequence inserts into the monolayer. The initial surface pressure of the monolayer can be controlled by adjusting the surface area. If the increase in surface pressure upon addition of a presequence is measured at various initial surface pressures, the data can be linearly extrapolated to determine a limiting pressure at which the presequence will no longer be able to insert into the monolayer. For CoxIV presequences of 25 or 33 residues, the limiting pressure was 40-50 mN m"^ (Roise et al., 1986; Tamm, 1986). Although a monolayer is not a perfect model for a bilayer, some studies have suggested that phospholipid packing in a monolayer is most similar to that in a bilayer at a surface pressure of roughly 33 mN m~^ (Demel et al., 1975; Seelig, 1987). The ability of the wild-type CoxIV presequence to insert into monolayers at pressures higher than 40 mN m~' is consistent with the ability of this presequence to insert into membrane bilayers. In contrast, a mutant form of the CoxIV presequence that contained a two-residue deletion designed to disrupt the amphiphilic helix (Al 1,12-CoxIV) inserted into monolayers only up to 30 mN m~^ (Roise et al., 1988). This presequence also had less detergent-induced a-helical content than the wild-type presequence, and it directed the import of attached proteins less efficiently. The relatively low limiting pressure for insertion of the mutant presequence into monolayers suggests that the decreased function of the mutant presequence may be due to a decreased ability to insert into membranes. B. Disruption of Phospholipid Vesicles
Another assay used to measure surface activity is leakage of the dye 6-carboxyfluorescein from phospholipid vesicles. Carboxyfluorescein trapped within vesicles at a high concentration has a quenched fluorescence. Disruption of the vesicles by surface-active agents induces a leakage of the dye that can be measured as an increase in fluorescence as the dye dilutes into the solution. Peptides that correspond to functional presequences can disrupt vesicles in these assays (Roise et al.,
M itochondria I Presequences
57
1986, 1988; Skerjanc et al., 1987; Zardeneta and Horowitz, 1992), whereas the mutant version of the CoxIV presequence, Al 1,12-CoxIV (Section III A), is greatly decreased in its ability to disrupt fluorescein-containing vesicles (Roise et al., 1988). This result is consistent with the low limiting pressure that was observed for insertion of this presequence into monolayers. The disruptive effects of presequences on vesicles also appear to depend on the length of the synthetic sequence; short versions of functional presequences do not disrupt vesicles, whereas all synthetic presequences consisting of 22 or more residues and corresponding to functional presequences have been found to disrupt membranes (Roise et al., 1986, 1988; Hoyt et al., 1991; Zardeneta and Horowitz, 1992; Nicolay et al., 1994; see also Section IIIC). These results suggest that the ability to disrupt membranes is a common feature of functional presequences. C. Effect of Length on the Properties of Presequences
Comparison of synthetic presequences of different lengths supports the hypothesis that insertion into the membrane is required for function. The length of a synthetic presequence is an important determinant of the ability of the presequence both to insert into membranes and to inhibit protein import. It should be noted that import inhibition studies must be interpreted with caution — mitochondrial presequences are able to dissipate the membrane potential and can thus cause nonspecific inhibition of protein import. Nevertheless, protein import can be inhibited at concentrations of presequence that do not dissipate the membrane potential (Gillespie et al., 1985; Glaser and Cumsky, 1990b; Cyr and Douglas, 1991). This observation, along with the fact that presequences themselves can be efficiently imported into mitochondria (Glaser and Cumsky, 1990a; Pak and Weiner, 1990; Cyr and Douglas, 1991; Roise, 1992), suggests that the presequences are competing for a step in the normal import process and that inhibition studies can be a useful measure of presequence function. The ability of various CoxIV presequences to inhibit protein import and to disrupt vesicles is correlated with the length of the presequences. A peptide corresponding to the first 22 residues of the CoxIV presequence was found to inhibit reversibly the import of precursors into mitochondria (Glaser and Cumsky, 1990b). A shorter peptide, corresponding to the first 16 residues, was a much weaker inhibitor of import, even though it contained the 12 residues that are sufficient to import an attached protein (Hurt et al., 1985a,b). Furthermore, the disruptive effects on membranes of 25-residue and 33-residue versions of the CoxIV presequence were much greater than those of shorter 15-residue and 17-residue versions (Roise et al., 1986; Nicolay et al., 1994). Thus, even though the targeting information appears to be contained in the extreme NH2-terminal part of the presequence, the context of a longer sequence is required for efficient insertion into the membrane and for efficient inhibition of protein import.
58
MERRITT MADUKE and DAVID ROISE
Studies on the presequence of rat liver pre-omithine carbamyltransferase (pOCT) also show a correlation between the ability to form a surface-active helix and the ability to act as an efficient inhibitor of protein import. A synthetic peptide corresponding to residues 1—27 of pOCT increased in a-helical content in the presence of lipids and was able to disrupt large lipid aggregates (Epand et al., 1986). A shorter peptide (residues 16-27), which had five of the eight charges and was predicted to be amphiphilic, did not exhibit this behavior. The longer, surface-active peptide inhibited the import of precursor proteins into mitochondria; the shorter peptide did not (Gillespie et al., 1985). A peptide corresponding to 19 amino acids of the presequence of the Fj-ATPase P-subunit did not induce carboxyfluorescein leakage from vesicles, even at high ratios of peptide to lipid (Hoyt et al., 1991). Nor did this peptide efficiently compete with precursor protein for mitochondrial import. Peptides corresponding to the amino-terminal 32 or 51 residues of the precursor, however, were efficient inhibitors of protein import (Cyr and Douglas, 1991). Unfortunately, the effects of these longer peptides on lipid vesicles were not measured, but the longer peptides presumably would have induced carboxyfluorescein leakage. The 19-residue segment from the presequence of the FpATPase P-subunit was also found to insert into monolayers (Hoyt et al., 1991). A titration of the peptide indicated that it inserted efficiently at an initial surface pressure of 20 mN m~^ The ability of the peptide to insert at higher initial surface pressures was not tested, however. The relatively low maximal surface pressure increase observed suggests that the limiting pressure may be lower than 33 mN m~^ and that this peptide may not insert into bilayers. The low limiting pressure is consistent with the inability of the peptide to induce fluorescein leakage from vesicles and the inefficiency of the peptide as an inhibitor of protein import. In conclusion: several studies have shown that the length of a synthetic presequence correlates with the ability of the presequence both to insert into membranes and to inhibit mitochondrial protein import. Even though the information necessary for targeting proteins to mitochondria may be contained in a short segment, the context of a longer peptide or protein appears to be necessary for the presequence to function as an inhibitor of protein import. The correlation between the ability to inhibit protein import and the ability to insert into membranes, although not proof, is consistent with the hypothesis that presequence binding to lipids is the initial step in mitochondrial protein import.
IV. BINDING TO MEMBRANES A. Binding to Phospholipid Vesicles
A few studies have measured directly the binding of presequences to phospholipid vesicles. A fluorescently labeled CoxIV presequence was shown to bind with moderately high affinity (10^—10^ M~^) to sonicated unilamellar vesicles that
Mitochondrial Presequences
59
contained negatively charged lipids (Frey and Tamm, 1990). Measurements of the lateral diffusion of the presequence suggested that the presequence inserted either parallel to the surface of the membrane as a monomer or perpendicular to the surface as an oligomer. The former possibility is supported by measurements with phospholipid monolayers using a radioactively labeled CoxIV presequence (Tamm, 1986). An area of 560 ± 170 A^ per bound presequence was calculated by measuring changes in the surface area of the monolayer as a function of the amount of presequence bound. This value is consistent with an orientation of the a-helix parallel to the surface of the membrane. Molecular modeling suggests that, as a monomer, the CoxIV presequence would insert to a depth of 5-8 A into a bilayer (Tamm, 1991). Calorimetric techniques were used to measure the association of a synthetic human ornithine transcarbamylase (pOCT) presequence with sonicated unilamellar vesicles (Myers et al., 1987). The peptide associated strongly (--10^ M~^) with vesicles that contained negatively charged phospholipids. The emission maximum of the tryptophan fluorescence of the presequence was shifted 4 nm toward the blue upon binding of the presequence to the vesicles. This shift in the emission is an indication that the tryptophan became exposed to a more hydrophobic environment, although the shift is not as great as those observed for tryptophans contained in peptides that insert deeply into bilayers (Surewicz and Epand, 1984; McKnight et al., 1991). Since the tryptophan in the pOCT sequence is predicted to be on the hydrophilic face of the a-helix, the small blue shift in the fluorescence is consistent with insertion of the presequence parallel to the surface of the membrane, so that the hydrophilic face remains partially exposed to the solvent. A similar blue shift was observed for the rhodanese presequence, whose tryptophan is also predicted to reside on the hydrophilic face of the helix (Zardeneta and Horowitz, 1992). The binding of the presequence of rat pOCT to lipid vesicles was quantitated by using assays to measure vesicle disruption (Skerjanc et al., 1987). The affmity of this presequence for sonicated unilamellar vesicles was similar to that observed for the CoxIV presequence. Fluorescence energy transfer measurements showed that, although the presequence oligomerizes in solution, it probably associates with the lipid as a monomer. The binding of the presequence of pOCT was not dependent on the application of a transbilayer potential, with the negative charge inside. This result is in contrast to measurements of the binding of the CoxIV presequence to extruded large unilamellar vesicles, which showed that the binding was increased in the presence of a potassium difflision potential (de Kroon et al., 1991). B. Binding to Mitochondrial Membranes
The experiments with model membranes demonstrated clearly that presequences can insert directly into lipid monolayers and bilayers. The interactions between presequences and mitochondria have also been measured (Roise, 1992; Swanson and Roise, 1992). To monitor binding, the CoxIV presequence was labeled at its
60
MERRITT MADUKE and DAVID ROISE
sole cysteine with iodoacetamidofluorescein. The fluorescein label permitted detection of the presequence at nanomolar concentrations with minimal interference from mitochondrial chromophores. The fluorescent label does not appear to affect the physical properties of the peptide significantly. The ability of the fluorescent presequence to bind to isolated yeast mitochondria was confirmed by experiments in which the fluorescence was observed to cosediment with the mitochondria. In a more convenient and quantitative approach, binding of the fluorescein-labeled presequence to mitochondria was observed directly by a decrease in fluorescence that occurred upon binding. Atypical binding curve based upon this approach is shown in Figure 3A. The data fit well to a two-state model of binding in which the free presequence has a relatively high fluorescence and the bound form has a quenched fluorescence. Ofparticular interest was the observation that the binding was not saturable up to the highest concentration of presequence measured (0.1 juM); regardless of the total concentration of presequence, the same fraction of presequence was bound to a given amount of mitochondria (the same amount of quenching was observed). This result showed that the presequence was binding directly to the lipid bilayer of the mitochondrial outer membrane. If the presequence had bound to a protein on the surface of the mitochondria, the fraction of total presequence bound should have decreased at increasing concentrations of presequence as the binding protein became saturated. Since binding of the CoxIV presequence to mitochondria does not occur at discrete sites, the binding is best described as a partitioning equilibrium (Figure 3B). The partitioning is analogous to the distribution of a solute molecule in an organic extraction—for example, between water and hexane. In the case of presequence binding to mitochondria, the organic phase corresponds to the membrane surface. The partition coefficient can be obtained directly from an experimental value, A/5Q, which is the concentration of membrane required to bind half the presequence. This value can be used to calculate the amount of presequence bound to a given concentration of membrane. In the experiment shown in Figure 3 A, 50 pmol of presequence was bound to 0.024 mg of total mitochondrial protein at a concentration of mitochondria equal to the value of M5Q. The amount of even the most abundant mitochondrial outer membrane protein, porin, is only 2.9 pmol in this amount of mitochondrial protein (Freitag et al., 1982; Riezman et al., 1983); therefore, this calculation confirms that the presequence must have bound to the lipids of the outer membrane rather than to a protein. Similar observations were made for the binding of a radioactively labeled CoxIV presequence to rat liver mitochondria (Nicolay et al., 1994). The binding of the presequence to mitochondria is rapid, and it is reversible. Although the rate of binding was not measured directly, the quenching of presequence fluorescence occurs within the mixing time of an experiment (