Three-Dimensional Architecture of the Isolated Yeast Nuclear Pore Complex: Functional and Evolutionary Implications

Transcription

1 Molecular Cell, Vol. 1, , January, 1998, Copyright 1998 by Cell Press Three-Dimensional Architecture of the Isolated Yeast Nuclear Pore Complex: Functional and Evolutionary Implications Qing Yang,* Michael P. Rout, and Christopher W. Akey* * Department of Biophysics Boston University School of Medicine Boston, Massachusetts The Laboratory of Cellular and Structural Biology The Rockefeller University New York, New York Endocrinology and Reproduction Research Branch National Institutes of Health Bethesda, Maryland Summary We have calculated a three-dimensional map of the yeast nuclear pore complex (ynpc) from frozen- hydrated specimens, thereby providing a direct com- parison with the vertebrate NPC. Overall, the smaller ynpc is comprised of an octagonal inner spoke ring that is anchored within the nuclear envelope by a novel membrane-interacting ring. In addition, a cylindrical transporter is located centrally within the spokes and exhibits a variable radial expansion in projection that may reflect gating. The inner spoke ring, a transmem- brane spoke domain, and the transporter are conserved between yeast and vertebrates; hence, they are required to form a functional NPC. However, significant alterations in NPC architecture have arisen during evolution that may be correlated with differences in nuclear transport regulation or mitotic behavior. Introduction The nuclear pore complex (NPC) spans the nuclear envelope (NE) within a specialized pore membrane domain, formed by a fusion of the inner and outer nuclear mem- branes. The NPC functions as a channel for the nuclear import of proteins and snrnps, while providing an export path for trnas, mrnps, ribosomal subunits, and snrnps (reviewed in Gorlich and Mattaj, 1996; Nigg, 1997). Nuclear import can be divided into three major steps: targeting/docking, translocation, and substrate release (Newmeyer and Forbes, 1988; Richardson et al., 1988; Akey and Goldfarb, 1989; Pante and Aebi, 1996). Recent work has resulted in a paradigm for nuclear import and export, in which receptors mediate the binding of transport substrates containing a nuclear localiza- tion (NLS) or export sequence (NES) to repeat-con- taining nucleoporins (GLFG/FXFG) within the NPC. It is postulated that substrate translocation may utilize cycles of GTP hydrolysis mediated by Ran (Koepp and Silver, 1996; Nigg, 1997). In addition, the NPC is thought to provide passive diffusion channels that give the NE its characteristic permeability to small molecules (Paine et al., 1975; Akey and Radermacher, 1993; Akey, 1995; To whom correspondence should be addressed. Perez-Terzic et al., 1996), while peripheral channels may facilitate the redistribution of nuclear membrane proteins (Powell and Burke, 1990; Hinshaw et al., 1992, Akey and Radermacher, 1993; Wiese and Wilson, 1993). The three-dimensional structure of the vertebrate NPC has been studied in amphibian oocytes using electron microscopy and single-particle methods on frozenhydrated (Akey and Radermacher, 1993) and negatively stained specimens (Unwin and Milligan, 1982; Hinshaw et al., 1992). These studies have provided a low resolution model of this translocation machine (see Figure 1). The NPC is comprised of an octagonal spoke ring complex that is embedded in the NE and that encircles a central channel complex (termed the transporter; Akey, 1989, 1990). The spoke ring complex anchors more peripherally associated components, including the cytoplasmic filaments (Richardson et al., 1988; Jarnik and Aebi, 1991; Ris, 1991) and the nuclear basket (Jarnik and Aebi, 1991; Ris, 1991; Goldberg and Allen, 1992), which play a role in the docking of import (Richardson et al., 1988; Pante and Aebi, 1996; Rutherford et al., 1997) and export substrates (Kiseleva et al., 1996), re- spectively. Nucleocytoplasmic transport utilizes a central chan- nel within the NPC for import and export (Feldherr et al., 1984; Dworetzky and Feldherr, 1988). However, the mechanism of translocation across the NPC is currently unknown. Recently, atomic force microscopy and scan- ning electron microscopy on Xenopus (Goldberg and Allen, 1996; Perez-Terzic et al., 1996) and Chironomus NPCs (Goldberg and Allen, 1996; Kiseleva et al., 1998) have provided further observations of the NPC trans- porter (Akey and Radermacher, 1993). The NPC transporter is present in manually isolated NEs in a number of radially expanded configurations in projection, suggesting that the channel is gated (Akey and Goldfarb, 1989; Akey, 1990). The physical imprint of the trans- porter is revealed in thin sections of Balbiani ring mrnps and nucleoplasmin-gold particles caught traversing the NPC (Stevens and Swift, 1966; Feldherr et al., 1984; Mehlin et al., 1992), as these substrates are constrained within a Å diameter channel that extends across the NPC for about Å. The completion of the Saccharomyces genome se- quencing project (Clayton et al., 1997 and references therein) has resulted in an accelerated identification of yeast nucleoporins (nups) (Rout and Wente, 1994; Doye and Hurt, 1997). Moreover, the recent isolation of enriched fractions of yeast NPCs and NEs (Rout and Blobel, 1993; Strambio-de-Castillia et al., 1995) has opened the door to structural studies, as isolated ynpcs have dimensions similar to those in cells. However, yeast NPCs are both smaller and less massive than vnpcs ( 66 MDa versus 125 MDa; Reichelt et al., 1990; Rout and Blobel, 1993). Hence, a detailed three-dimensional (3-D) structure of the ynpc will allow an understanding of this fundamental size difference and provide a frame- work in which the position of yeast nucleoporins can be mapped. In this work, we present the 3-D structure of the ynpc

2 Molecular Cell 224 the method of Rout and Blobel (1993) and utilized computational methods to select the fraction with the highest degree of structural preservation. A representative subset of 8-fold symmetric ynpcs within this fraction were identified by rotational power spectra analysis, and the appropriate two- and three-dimensional maps were calculated. Figure 1. A Consensus Model of the Vertebrate NPC The overall structure is comprised of a wheel-like array of eight spokes connected to the inner spoke ring (ISR), which encircles the central channel complex (transporter [T] is pink). The spoke complex is framed top and bottom by the cytoplasmic (CR) and nuclear (NR) thin rings that serve as attachment sites for cytoplasmic particles (CP) and filaments and the nuclear basket. The spokes are embedded in the NE (light blue), are interconnected within the lumen to form a lumenal ring (LR), and are attached to the lamina (L). The outer and inner nuclear membranes are labeled (ONM) and (INM), respectively. A set of inner ring filaments interconnect the top and bottom of the central transporter to the thin rings and for clarity, are shown only on the cytoplasmic side (bold lines). determined by electron cryomicroscopy and image processing. The resulting maps have been compared with their vertebrate counterparts, to provide insights into the function and evolution of the NPC and its subassem- blies. Overall, the ynpc is smaller than the vertebrate NPC (vnpc) and maintains a conserved inner spoke ring and central transporter. In addition, yeast and vertebrate NPCs both interact with the NE at comparable radii using similar spoke domains. However, the ynpc does not have a lumenal spoke ring but rather has evolved a novel membrane-interacting ring that is comprised of 8 linear arms and 8 membrane-spanning spoke domains. In pro- jection maps, the central ynpc transporter is visualized in 3 possible transport-related configurations and demonstrates 2 related in transit forms, similar to that observed in vertebrate transporters (Akey and Goldfarb, 1989; Akey, 1990). The implications of these observations for the mechanism of nucleocytoplasmic transport are discussed. Results The resolution that is attainable with supramolecular assemblies in the electron cryomicroscope is often limited by the biochemical stability of the specimen. Therefore, we have characterized ynpc fractions isolated by Mass and Symmetry of Yeast NPCs We have used dark-field imaging in the scanning transmission electron microscope (STEM; Wall and Hainfeld, 1986) to estimate the mass average of the enriched and highly enriched fractions of ynpcs, isolated by the method of Rout and Blobel (1993). The data for highly enriched ynpcs is shown in Figure 2A and gives a Gaussian-fitted peak of 54.5 MDa (s 10.1). Similar mass values were obtained for enriched ynpcs and cross-linked ynpcs from the highly enriched fraction (Yang, 1997). The STEM mass data and earlier estimates obtained by light scattering and sedimentation methods (Rout and Blobel, 1993) overlap significantly and place the mass of the ynpc between 55 and 66 MDa. Previous measurements rely on bulk properties of the sample and are influenced by larger particles (Rout and Blobel, 1993), while the STEM measures individual particles to form a mass spectrum. Hence, the differences between the midpoint values may reflect the nonideal behavior of partially purified ynpcs. However, this mass estimate is consistent with the smaller size of the ynpc observed in thin sections of cells and nuclei (Rout and Blobel, 1993; Strambio-de-Castillia et al., 1995) and with our 3-D analysis. Isolated yeast NPCs are present in a range of masses, and images reveal varying degrees of elliptical distor- tion. These observations suggested that many of the ynpcs have sustained some physical damage during isolation. Therefore, we have carried out a detailed analysis of the preservation of detergent-solubilized ynpc fractions using 2-D averages (Frank et al., 1996; Akey, 1989) and rotational power spectra analysis (Crowther and Amos, 1971; Kocsis et al., 1995). Although the highly enriched fraction is biochemically cleaner, projection maps suggested that the enriched ynpcs have less circumferential distortions within the spokes (Yang, 1997). Hence, we chose the enriched fraction for two- and three-dimensional cryodata collection. After visual selection, rotational power spectra were calculated from the untilted ynpcs. This provided a one-dimensional diffraction pattern that was used to evaluate particle preservation, allowing us to choose the best ynpcs for further analysis. A histogram of rotational 8-fold power within the dataset is presented in Figure 2B. Particles with a low signal-to-noise ratio, strong elliptical distor- tions, or physical damage were excluded, based on the lack of a significant 8-fold harmonic in their power spectra (dotted curve in Figure 2C). Roughly half of the parti- cles demonstrated 8- and 16-fold harmonics in their spectra (Figures 2C 2E), reflecting the 8-fold symmetry of the ynpc (Rout and Blobel, 1993). Structure of the Isolated Yeast NPC and Interactions with the Nuclear Envelope We calculated projection maps for both detergent-solu- bilized (dform) and nuclear envelope associated (mform)

3 Architecture of the Yeast Nuclear Pore Complex 225 Figure 2. Mass and Rotational Power Spectra Analyses of Yeast NPCs (A) Mass histogram of 506 ynpcs obtained from dark-field STEM (peak mass and standard deviation are 54.5 Mda and 10.1 Mda). (B) Histogram of relative 8-fold power for particles within the ynpc dataset, plotted as a percentile. The distribution is skewed with a median of 28%. (C E) Averaged rotational power spectra from the bottom 1728, the best 1535 ynpcs (curves labeled [1] and [2] in [B] and [C]), from the top 10 (labeled [3] in [B] and [D]), and from the global transporter dataset ([GBT] in [E]). ynpcs in frozen-buffer and negative stain, respectively (Figures 3A and 3B). These maps were then compared with similar published maps of the vnpc (Figures 3C and 3D; Akey, 1995) to understand the structural basis of previously observed size differences and to identify the membrane-interacting spoke domain within the ynpc. In Figure 3A, the dform map of the ynpc has an outer diameter of 960 Å and is comprised of 8 spokes. Each spoke has 2 radial domains (circles labeled 1 and 2), with the proximal domains forming an inner spoke ring. The inner spoke ring encircles a central blurred disk that represents a global average of the transporter. An additional linear arm of density extends radially from the inner spoke ring and is connected circumferentially to both of the neighboring spoke 2 domains, to form a Figure 3. Two-Dimensional Projection Maps of Yeast NPCs and a Comparison with Vertebrate NPCs (A) Map of ynpcs without nuclear membranes in frozen buffer. Labels: radial spoke domains (labeled [1] and [2] with circles), inner spoke ring (ISR), linear arm (LA), central transporter (T), putative transport substrates (S). The position of the approximate spoke 2-fold axis is indicated. (B) Map of NE-associated ynpcs in negative stain. Two radial spoke domains are labeled with circles, and the position of the NE is indicated (*). (C) Map of vnpcs without nuclear membranes in frozen buffer. The radial arm is indicated (RA), and three radial spoke domains are labeled (1) (3) (with circles). The position of the NE is shown by contours from the appropriate difference map (Akey, 1995). (D) Map of vnpcs associated with the NE in frozen buffer (Akey, 1995). The three radial spoke domains (closed circles) and the NE (*) are indicated. Protein and membranes are open; scale bar 300 Å.

4 Molecular Cell 226 Table 1. Domain Positions and Dimensions in Yeast and Vertebrate NPCs Yeast NPC Vertebrate NPC Structural domains Di (Å) Dc Do Height Di (Å) Dc Do Height Transporter Inner spoke domain a Second spoke domain Membrane ring Nuclear membrane pore Entire NPC b 1000 c Vertebrate NPC dimensions are from Akey (1989), Akey and Radermacher (1993), and Akey (1995). Di/Dc/Do are inner, center and outer diameters. a Measured from Akey and Radermacher (1993). b Including the radial arms. c Including the cytoplasmic particles. ring. The spokes display approximate 2-fold symmetry The similar diameter of the yeast and vertebrate transporters when mirrored about the arrow labeled 2 (which bisects and the shared 8-fold symmetry of the spokes the spokes); however, the linear arm is displaced cir- would argue that the yeast transporter has 8-fold symmetry, cumferentially in a counter-clockwise direction from a and this property was used in the following analycumferentially local 2-fold axis (located between the spokes as sis. Classification was used to eliminate ynpcs with shown in Figure 3C), to interact preferentially with an poorly defined transporters and to sort the remaining adjacent spoke. This deviation from 822 symmetry may pore complexes into transporter groups (Akey, 1990; have arisen during purification, as similar circumferential Frank, 1990). This analysis resulted in projection maps distortions are present in vnpcs (Akey and Rader- of three transporter classes shown in Figures 4A 4C, macher, 1993). and a global average (see Figure 4D and averaged power A comparison of 2-D maps from yeast and vertebrate spectrum in Figure 2E). NPCs suggests a reason for the observed size differences. There are two prominent classes in which the transjection Vertebrate spokes have 3 radial domains in pro- porter ring encircles central material, that may represent with the positions of the innermost 2 domains transport substrates trapped within the channel during being equivalent to those in ynpcs (Figures 3A and 3C, spheroplast lysis (Figures 4B and 4C). The diameter of Table 1). A second projection map of the ynpc reveals the transporter in Figure 4C is 13% larger than in Figure that the spokes have 2 radial domains when the nuclear 4B, and this radial expansion was correlated with an membrane is present (Figure 3B, see circles). Hence, it increase in central substrate density (after scaling within seems unlikely that the best ynpcs have lost major the spokes). Indeed, the intensity of the central substrate spoke domains during solubilization and purification. disk was computationally truncated to provide an equiv- We suggest that the ynpc spoke may be intrinsically alent display of the spokes in Figures 4B and 4C. Transporters smaller than its vertebrate counterpart. with a similar in transit morphology were idensmaller Although yeast and vnpcs differ in lateral diameter tified previously in vnpcs (Figure 4E; Akey and Goldfarb, and mass, some features of their interaction with the 1989; Akey, 1990). In Figure 4A, the central transporter NE are conserved. In Figure 3B, the nuclear membrane ring surrounds an apparently empty central channel. is present as a strong band of circumferential density However, a comparison of the channel density with that interacts with the second spoke domain in ynpcs. background levels suggests that these transporters may This interaction occurs at a similar radius within vnpcs contain some residual material, as this class may have (Figure 3D and contoured difference map in Figure 3C; originated from the two in transit classes by the loss of Akey, 1995). However, there are notable differences between trapped substrates during isolation. Finally, the central yeast and vertebrates in the details of this interac- placement of the transporter within the inner spoke ring tion. For example, the ynpc has an additional linear arm creates a set of 8 low density features that may form that is not observed in the mform ynpc map, because diffusion pores in both species (open circles in Figures it is likely buried within the membrane. 4A and 4E). These data suggest that transporters are general features of all NPCs and reinforce the concept Visualization of Multiple Forms that the transporter represents a gated channel involved of the Central Transporter in nucleocytoplasmic transport (Akey and Goldfarb, The diameter of the globally averaged central transporter 1989; Akey, 1990; Akey and Radermacher, 1993). in ynpcs is similar to that observed in vnpcs ( Å; Akey, 1989, 1990). The vertebrate trans- Three-Dimensional Structure of the ynpc porter appears to have 8-fold symmetry, as 8 internal Three-dimensional maps were calculated of the ynpc ring filaments link the transporter to the cytoplasmic to provide insights into the architecture of this assembly. and nuclear thin rings (see Figure 1; Goldberg and Allen, However, we found that 3-D maps required ), and these data are consistent with previous label- ynpcs to provide an acceptable signal-to-noise ratio ing studies using MAb414 (Akey and Goldfarb, 1989). with these data. Hence, at this stage we computed a

5 Architecture of the Yeast Nuclear Pore Complex 227 Figure 4. Classification of ynpcs Reveals the Central Transporter (A) Transporter class 1 contains a central ring-like transporter encircling a central channel. One of eight diffusion channels is labeled with an open circle. (B) Transporter class 2 contains a transporter ring that encircles a central bright density that, by analogy with vnpcs, may be composed of substrates caught in transit (see Akey, 1990). (C) Transporter class 3 contains a central ring with a greater radial expansion than observed in class 2. (D) Global average of ynpc transporters. The spoke domains ([1] and [2]); the transporter ring (T), and the putative substrate disk (S) are labeled. (E) A radially expanded vertebrate NPC transporter (Akey, 1990). Note the similar in transit morphology in both vertebrate and yeast transporters. Scale bar 300 Å. 3-D map of the global transporter class (GBT class). centered on the spoke (Figure 5B). In this view, the spoke In addition, we used classification to obtain the most is composed of an inner spoke domain that spans the homogeneous spoke class (SC), irrespective of the qual- entire ynpc and a second domain located at the spoke ity of the associated transporter, and computed the midplane (dashed lines). These domains are joined at SC map. the top and bottom of the inner spoke ring by a bridging Central cross-sections 100 Å thick and oriented parallel density that forms an angle of 50 relative to the central to the original position of the NE are shown in Figures plane (closed circle in Figure 5B). Weak density present 5A and 5C, for the GBT and SC 3-D maps. In general, at the top and bottom of the spokes (open arrows, Figure the spokes are similar in both maps, but as expected, 5B) may represent disordered filaments and nuclear the central transporter is rather disordered in the SC baskets, as filaments were observed radiating from the map. In this view, the spokes form an inner spoke ring, ynpcs in dark-field STEM images (not shown) and peripheral comprised of spoke domain 1 and the proximal end of filament proteins such as Nup159p are retained the linear arms (inner closed circle, Figure 5C). In addi- in these preparations (Kraemer et al., 1995). The transporter tion, the second spoke domains are interconnected with in the GBT 3-D map (Figure 5B) contains a central the distal end of the linear arms (outer closed circle, bright density that may reflect substrates trapped within Figure 5C) to form a continuous ring. This feature is the cylindrical walls of the channel assembly (TW). Overlocated at a radius that would place a majority of the all, the isolated ynpc is a disk with dimensions of 960 ring within the nuclear membrane, while the most distal Å Å (Table 1). A strongly thresholded surface parts of the ring may protrude into the lumen. Hence, view of the spokes without the central transporter reveals we have named this novel feature the membrane ring. the connectivity of the spoke domains (Figure 5D). Additional evidence for this membrane ring was ob- The second radial spoke domain extends from the inner tained after detergent or heparin extraction of NEs or spoke ring with its attendant bridging density, and the ynpcs, respectively (Strambio-de-Castillia et al., 1995). linear arm is located between adjacent spokes. Extracted NEs show a filamentous meshwork in which The relationship between the central transporter, the ring structures are embedded, while strongly heparin- inner spoke ring, and the membrane ring is shown in ized ynpcs give rings with a diameter commensurate surface views of the GBT 3-D map (Figures 6A 6D). In with the size of the membrane ring. Figures 6A and 6B, the spoke transporter assembly is Prominent spoke features were revealed by calculating viewed obliquely from the bottom and top of the 3-D an angular projection that covered an arc of 22.5, volume, respectively. Face-on and side-on views of the

6 Molecular Cell 228 circle in Figure 6C), as observed in vertebrate NPCs (Akey and Radermacher, 1993). Hence, these internal channels may play a role in the passive diffusion of small molecules between the cytosol and nucleus. Discussion Rationale for Comparing Yeast and Vertebrate NPCs Yeast and vertebrate cells represent different aspects of the basic eukaryotic cell design. Generally, yeast cells are smaller, undergo rapid cell division with a closed mitosis, and have less complex cytoskeletons due to the external support of a cell wall. Vertebrate cells possess elaborations not found in yeast: they have a lamina and larger nuclei, and their NEs disassemble at mitosis. In addition, they differentiate into many cell types and undergo apoptosis. Given the evolutionary distance between yeast and vertebrates, it is not unexpected that the structure of their NPCs would differ. However, both yeast and vnpcs must contain the minimal components required to build a functional NPC. As a prelude to this study, it was necessary to prepare enriched fractions of ynpcs. However, isolated ynpcs display a range of physical deformations and mass. We therefore used rotational power spectra to select the best ynpcs for two- and three-dimensional maps, ac- counting for 2% 5% of the ynpcs. We suggest that the maps in this study are representative of the core ynpc at the current resolution of 80 Å in 2-D and Å in 3-D, excluding the putative peripheral filaments and baskets that are disordered. First, 16 yeast nups coenrich quantitatively with isolated NEs and ynpcs (Rout and Blobel, 1993; M. P. R., unpublished data). Second, all known nups ( 30) have been identified in the ynpc preparation using mass spectrometry (M. P. R. et al., unpublished data). Third, filaments were observed radiating from ynpcs in dark-field STEM images. As filament components such as Nup159p are retained in these preparations (Kraemer et al., 1995), the data indicate that peripherally associated nups are present in the isolated ynpcs. Thus, the best ynpcs are likely to contain a full (or nearly full) complement of nups. All existing evidence suggests that ynpcs are funda- mentally smaller than vnpcs. For example, ynpcs have a simple disk-like appearance in thin sections of whole cells, nuclei, and NEs but do not contain the distinctive features at the surface of the NE that correspond to the thin coaxial rings in vertebrate NPCs (Franke and Scheer, 1974; Rout and Blobel, 1993; Strambio-de-Cas- tillia et al., 1995). The overall dimensions of our structure are in excellant agreement with measurements from thin sections of yeast nuclei (Rout and Blobel, 1993), and the morphology is similar. Interestingly, the thickness of the NE lumen is markedly different in these two eu- karyotes: in yeast, the NE is Å thick (31 mea- surements from thin sections of 6 yeast nuclei in cells), while in vertebrates, the NE is Å wide (e.g., see Goldberg and Allen, 1996; Pante and Aebi, 1996). In vertebrate NPCs, the inner and outer nuclear mem- branes are in close apposition to the thin nuclear and cytoplasmic rings and their connecting spoke domains Figure 5. Selected Slices and Surfaces from Two Yeast NPC 3-D Maps (A) An average of 5 xy slices taken from the center of the GBT 3-D map. The spoke domains, transporter, and central substrate are labeled as before. The position of this slice is indicated in (B) (dashed lines). (B) Angular projection of the ynpc calculated from the GBT 3-D map. The inner spoke ring (1), the outer spoke domain (2), and the bridging region (closed circle) are marked. White arrowheads indicate peripheral density attributed to disordered filaments on both ynpc surfaces. The transporter wall (TW) and the putative central substrate (S) are labeled. (C) A central xy slice from the SC 3-D map (n 577). The spokes and membrane ring are labeled as before. Two domains of the linear arm are marked with circles, and the membrane interacting ring (MR) is formed by interactions between the linear arm (LA) and adjacent spoke domains (2). (D) A highly thresholded surface view of the GBT spoke ring is shown; spoke domain (2), the inner spoke ring (ISR), and the linear arm are marked. Scale bar 300 Å. 3-D map are shown in Figures 6C and 6D. The inner spoke ring encircles the central transporter and may function as a central organizing assembly within the ynpc. The membrane ring is formed by circumferential connections between the second spoke domains and the adjacent linear arm (Figure 6C), and when viewed from the side, this feature is not flat, but rather forms a sinusoidal ring (closed circles in Figure 6D). In the 3-D map, the central transporter is a cylinder with dimensions of Å. In addition, weak densities interconnect the top and bottom surfaces of the inner spoke ring and transporter; these features may be below the current resolution or partially disordered (not shown). Density identified as transport substrate in 2-D maps is visible at the center of the transporter ([S] in Figures 6A and 6C). Hence, the GBT 3-D map may represent transporters trapped in a number of differing configurations during spheroplast lysis. Interestingly, the transporter differs in the degree of radial expansion present at either end of the cylindrical channel. These differences may reflect an intrinsic size asymmetry of import and export substrates that is maintained in the averaged map. Finally, internal channels are located between the transporter and the inner spoke ring (open

7 Architecture of the Yeast Nuclear Pore Complex 229 Figure 6. Surface Views of a 3-D Map of the Yeast NPC with a Globally Averaged Transporter (Aand B) Obliqueviews of the ynpc revealing the bottom and top surfaces of the ynpc. The membrane ring (MR) interconnects adjacent spokes, making contact with the linear arm (*). The cylindrical transporter resides within the inner spoke ring (ISR) and contains putative central substrates (S). (C) An en face view of the ynpc reveals details of spoke domain architecture ([1] and [2]) and suggests that diffusion channels are located between the transporter and ISR (open circle). The membrane ring (MR), transporter (T), and putative central substrate (S) are labeled. (D) Side view of the dform ynpc indicates that the membrane ring interconnects adjacent spokes and the intervening linear arm in a sinusoidal manner (see closed circles). Scale bar 300 Å. in diameter (960 Å versus 1450 Å) and height ( 350 Å versus 800 Å) than the vnpc, and this difference re- flects the mass of the respective NPCs ( 60 MDa versus 125 MDa). However, vnpc domains located at high radius (the coaxial thin rings, cytoplasmic particles, and lumenal ring) form a more open structure, and this re- sults in an 5-fold volume increase relative to the ynpc. When viewed from the side, the vnpc is comprised of four major ring systems surrounding the central transporter, including the inner spoke ring, the cytoplasmic and nuclear thin rings, and the lumenal ring (Figures 1, 7C, and 7D). The isolated ynpc is simpler and is comprised of an inner spoke ring that encircles the central transporter and an outer membrane interacting ring. We did not detect structural equivalents of the thin cytoplasmic and nuclear rings, cytoplasmic particles, or the lumenal spoke ring in the ynpc (Figures 7A and 7B); however, the dimensions of our 3-D structure are in good agreement with measurements of ynpcs from thin sections of cells. Why is the vnpc larger than the yeast pore complex? Yeast and vertebrate NPCs have undergone divergent evolution from a common ancestral pore complex; hence, significant changes in morphology may have resulted from alterations in surface loop size within nup homo- logs. Thus, vertebrate nups are often larger than their yeast counterparts (e.g., vnup107p versus ynup84p). However, each NPC may also have a complement of nucleoporins that comprise domains not present in the other. A more useful comparison of these distantly related NPCs can be made from diagrammatic cross-sections of the structures (Figures 8A and 8B; Akey, 1995). Importantly, the equivalence of the inner spoke ring and position of the NE was used to align vertebrate and ynpcs (Franke and Scheer, 1974; Jarnik and Aebi, 1991; Akey and Radermacher, 1993; Goldberg and Allen, 1996). However, the thin coaxial rings are not present in the ynpc 3-D map (see next section). We suggest that major differences in the thickness of the NE lumen are correlated with the vertical size of the spokes and their mode(s) of interaction with the nuclear membrane. Hence, this observation signals a fundamental change in the architectural scale of the NPC in the two distantly related species. Additional evidence for the smaller ynpc is as follows. There is no lamin homolog in yeast; thus, domains of the nuclear thin ring and spoke surface that interact with the lamina in vertebrates (Jarnik and Aebi, 1991; Goldberg and Allen, 1996) are not required in ynpcs. Abundant transmembrane nups in vnpcs (e.g., gp210 and Pom121) are not present in yeast; hence, the corresponding lumenal spoke domains are absent in ynpcs. In addition, initial solubilization studies using both nuclei and NEs revealed ynpcs with a size and morphology similar to that found in the enriched preparation (Rout and Blobel, 1993). Finally, projection maps of both dform and mform ynpcs show a similar two-domain spoke morphology; yet these preparations are made very differently, and the nuclear membrane is expected to stabilize the spokes. When taken together, all of the morphological and biochemical data indicate that ynpcs are fundamentally smaller than vnpcs; hence, our 3-D map of the isolated ynpc is likely to represent the basic architecture of the spoke transporter assembly. Functional Implications of Differences in Yeast and Vertebrate NPC Structure Surface maps of yeast and vnpcs are presented to scale in Figures 7A/7B and 7C/7D. The ynpc is smaller

8 Molecular Cell 230 Figure 7. A Comparison between 3-D Structures of Yeast and Vertebrate NPCs (A) The yeast NPC as viewed from the putative cytoplasmic surface. The transporter is marked (T) and putative substrates are marked (S). (B) Side view of the yeast NPC showing the marked difference in height relative to the vnpc and the lack of thin rings on either surface. (C) The dform vnpc as viewed from the cytoplasmic surface. The transporter (T) is partly obscured by a ring of collapsed cytoplasmic filaments (CF) that emanate from the cytoplasmic particles ([CP], Akey and Radermacher, 1993). The radial arms are labeled (RA). (D) Side view of the vnpc reveals the cytoplasmic and nuclear thin rings (CR/NR) that are integral to the the spoke complex. A lumenal ring (LR) is formed by the lumenal spoke domain and adjacent radial arms (see closed circles). Scale bar 300 Å. as shown in Figure 8A and inset. In this alignment, the NE, although they share conserved transmembrane inner and central spoke domains of vnpcs have equivalent spoke domains. Notably, ynpcs have a novel linear arm domains in ynpcs, with the central or second spoke located between adjacent spokes that, together with domains making a similar transmembrane insertion. In spoke domain 2, forms a membrane ring (Figure 8B, addition, the radial and vertical dimensions of these left). The membrane ring may function to anchor the two domains are similar (Table 1). As cytoplasmic and inner spoke ring of ynpcs into the NE and provide cir- nuclear thin rings are not present in ynpcs, the spoke cumferential stabilization. Recently, Pom152p has been domains that support the thin rings are missing. These immunolocalized near the outer surface of the ynpc include the inner and outer vertical supports that form (Wozniak et al., 1994); hence, we suggest that Pom152p a part of the respective surfaces of the vertebrate spoke. may comprise part of the membrane ring (Strambio- As suggested, the thin coaxial rings and lumenal ring de-castillia et al., 1995). Moreover, many nup mutants may stabilize the larger vertebrate spokes and function display an altered NE morphology (Doye and Hurt, 1997) in transport or its regulation (Hinshaw et al., 1992; Akey that may reflect the close proximity of the spoke domains and Radermacher, 1993; Akey, 1995). to the nuclear membrane in the smaller ynpc. Cytoplasmic filaments and the nuclear basket provide A feature unique to the vnpc is a ring present within transport substrate docking sites (Richardson et al., the NE lumen, comprised of the lumenal spoke domains 1988; Kiseleva et al., 1996; Pante and Aebi, 1996) and and adjacent radial arms (Figure 8B, right). Gp210 is a are attached to the cytoplasmic and nuclear thin rings major membrane protein of the vnpc and is thought to in vnpcs (Jarnik and Aebi, 1991; Ris, 1991; Goldberg be a component of the lumenal ring (Greber et al., 1992; and Allen, 1992). Biochemical and morphological data Wozniak and Blobel, 1992; Akey and Radermacher, suggest that ynpcs also have these peripheral assem- 1993). Our data are consistent with the lumenal position blies (Rout and Blobel, 1993; Kraemer et al., 1995; of gp210, as ynpcs do not have a lumenal ring and M. P. R., unpublished data). Presumably, the primary there is no gp210 homolog in the yeast. During open attachment sites for these assemblies have been conserved, mitosis, vnpcs disassemble into soluble subcomplexes but may be augmented by the thin rings in verte- in response to phosphorylation (Macaulay et al., 1995; brates. For example, Nup88p has been implicated in the Favreau et al., 1996). As gp210 is specifically phosphory- attachment of the cytoplasmic filament protein CAN/ lated in mitosis (Favreau et al., 1996), this nup may play Nup214p to the vnpc (Bastos et al., 1997; Fornerod a role in NPC disassembly. Hence, the lack of gp210 et al., 1997). However, there is no yeast protein with and a lumenal ring may reflect the absence of NPC significant sequence similarity to Nup88p, although disassembly during closed mitosis in yeast. Nup159p is a yeast homolog of CAN/Nup214p; hence, Recently, it was postulated that spoke conformational Nup88p may represent a vertebrate-specific protein of plasticity may mediate the down-regulation of both acthe cytoplasmic rings or particles. tive transport and the diffusion of small molecules, in Yeast and vnpcs differ in their interactions with the response to calcium depletion from the endoplasmic

9 Architecture of the Yeast Nuclear Pore Complex 231 Figure 8. A Comparison between Models of Vertebrate and Yeast NPCs Suggests that Evolutionary Divergence Has Resultedin Dramatic Changes in the Architectural Scale of This Organelle (A) An idealized vertical section of the yeast (left) and vertebrate (right) NPCs, cut through the middle of the spokes. The ynpc is comprised of two radial spoke domains ([1] and [2]) with a bridging density between them; domain 1 encircles thecylindrical transporter. Putative cytoplasmic filaments and a nuclear basket are shown on the ynpc. Components of the vnpc are: cytoplasmic filaments (CF) and particles (CP), and the inner ring filaments (IRF) that connect the transporter with a centralchannel (CC) to the cytoplasmic (CR) and nuclear (NR) thin rings. Inset: Aligned spoke and transporter cross-sections from yeast and vertebrates. The yeast spoke is missing the vertebrate domains highlighted in pink, which include the vertical inner and outer domains (Vi/Vo) that support the cytoplasmic and nuclear thin rings (CR/NR) and frame the lumenal spoke domain (LS). The inner spoke ring and central spoke domains are conserved in both species. The outer and inner nuclear membranes (ONM/INM) make similar contacts with the C or second spoke domains; however, the overall thickness of the NE lumen mimics the vertical dimension of the NPC. The yeast transporter appears to be missing a central cylinder (pink) that gives the vertebrate transporter its hourglass shape. The yeast transporter may retain two possible gating assemblies (shown in gray). (B) An idealized cross-section of the yeast (left) and vertebrate (right) NPCs cut within thecentral plane, parallel to the NE. The NPCs show conserved and divergent interactions with the nuclear envelope (NE). The lumenal ring components (highlighted in pink) of the vnpc are missing in the ynpc, while the latter has evolved a novel membrane ring (dark blue). Yeast NPC domains labeled as in (A) with the addition of the linear arm (LA). Vertebrate NPC: inner (IS), central (CS), and lumenal spoke (LS) domains, radial arms (RA), transporter (T), and central channel (CC). reticulum (Akey, 1995; Greber and Gerace, 1995; framework of the pore complex. The inner spoke ring is Stehno-Bittel et al., 1995). Perhaps Ca 2 depletion from attached to similar second/central spoke domains that the NE lumen may play a role in the earliest stages penetrate the nuclear membrane. In addition, the proximal of mitosis and programmed cell death, by triggering a region of the linear arm in the ynpc (see Figures conformational change within the spokes that modu- 5A and 5C) either forms part of the inner spoke ring or lates the transport activity of the central transporter. As is attached to an ISR component. The inner spoke ring yeast do not undergo open mitosis or apoptosis, they provides a conserved framework in which the central may not require the larger spoke design. transporter is suspended (Figures 1 and 8). Moreover, these conserved spoke domains may play a role in seeding The Functions of Conserved Domains in NPCs NPC assembly. Many nup homologs are shared between yeast and How are substrates translocated through the NPC and vnpcs (e.g., yeast and human Nic96p; ynup170p/ peripheral assemblies? Structural data suggest that nu- ynup157p and vnup155p; ynup84p and vnup107p; clear transport may span two different environments. Doye and Hurt, 1997), and the transport factors are conserved First, substrate complexes that are bound to the cytovnpcs (Koepp and Silver, 1996). Therefore, yeast and plasmic filaments or nuclear basket must move to the are expected to share a common structure that transporter (Richardson et al., 1988; Kiseleva et al., 1996; mediates nuclear transport. A comparison between Pante and Aebi, 1996; Rutherford et al., 1997). In a second yeast and vnpcs reveals a conserved inner spoke ring step, substrates traverse a central channel within (Figure 8), which may function as the central organizing the NPC (Feldherr et al., 1984; Akey and Goldfarb, 1989;

10 Molecular Cell 232 Akey, 1990). In our studies, both yeast and vertebrate (crude, enriched, and highly enriched) were isolated with the prototransporters demonstrate a similar ring-like morphology col of Rout and Blobel (1993). To minimize damage, the final sample used for 3-D analysis was frozen only twice. Sample preparation: and an intrinsic ability to expand radially, when viewed l of enriched ynpcs were dialyzed against ml of in projection. This conserved feature may reflect an in- 10 mm Bis-Tris (ph 6.5)/0.1 mm MgCl 2 /20% DMSO/0.2% TX-100/ herent gating property of the NPC, as macromolecules 1:1000 solution P (2 mg of pepstatin, 90 mg of PMSF in absolute without NLSs or NESs must be excluded from the central ethanol) at 4 C for 8 12 hr; highly enriched ynpcs used 0.01% channel. Tween 20. To prepare frozen specimen, 10 l of sample was incu- The yeast transporter is a shortened cylinder, while bated for 40 min on carbon-coated and air glow discharged copper grids. Grids were manually double-blotted and plunged into liquid the vertebrate transporter is more elongated with a triethane in a humidity-controlled chamber (Dubochet et al., 1988). partite hourglass shape (Akey and Radermacher, 1993). Cryogrids were loaded into a Gatan cryoholder (model 626-DH) and Recent STEM images from Chironomus NPCs (Goldberg images recorded at 176 C at 13,000 on a Philips CM12 with low and Allen, 1996; Kiseleva et al., 1998) are consistent dose kit and specimen relocation hardware. The first zero of the with a tripartite transporter and suggest that there are contrast transfer function was at 1/70 Å, and 50/0 tilt image pairs two gate assemblies localized at the cytoplasmic and were collected for random conical tilt 3-D reconstruction (Rader- macher, 1988). nuclear faces of the transporter, which may restrict ac- Yeast NPCs associated with NEs were prepared by a modification cess to the central channel (Akey, 1990; Akey and Rader- of the method in Strambio-de-Castillia et al. (1995; M. P. R., unpubmacher, 1993). As the translocation mechanism is prob- lished data). Bound ribosomes were stripped from the NEs by 0.5 ably similar in yeast and vnpcs, we suggest that the M KCl and M puromycin treatment in 1.75 M sucrose/ yeast transporter is formed from two opposite-facing 10 mm Bis-Tris (ph 6.5)/0.1 mm MgCl 2 /20% DMSO for 10 min at gate assemblies, without the central extension that 20 C, followed by dialysis for 4 hr at 4 C in buffer without KCl/ puromycin. Stripped NEs were spun onto air glow discharged grids gives the vertebrate transporter its hourglass shape (Figand negatively stained with 2% uranyl acetate. Low dose images ure 8A, inset). The reasons for the central extension in were recorded at 13,000 with an appropriate defocus. the vertebrate transporter are unclear, but may reflect adaptations to allow transport regulation, a more efficient translocation of large substrates, or compensation Image Processing for the wider NE lumen. Finally, it has been suggested Tilt and zero micrograph pairs (n 42) were densitometered on an Eikonix with an image step size of 20 Å and ynpcs chosen with the that vaults and transporters may be related, given their WEB particles option (Frank et al., 1996), based on the circular shared 8-fold symmetry (Rome et al., 1991). However, morphology of the zero tilt particles. Initially, 10% 15% of the vaults have not been reported in yeast, and a search possible ynpcs were chosen (n 3264), and rotational power spectra of the complete genome does not detect vault protein were calculated to allow a nonbiased selection of particles homologs, indicating that transporters are distinct from (Crowther and Amos, 1971; Kocsis et al., 1995). The ratio of the vaults. 8-fold harmonic for each particle relative to the best particle was used to create a percentile distribution, which indicated that 1565 ynpcs had detectable 8-fold signal. Image processing was carried The Evolution of NPCs out on VMS/Alpha workstations using SPIDER (Frank et al., 1996). Our data suggest that NPCs have undergone a fundawhich Alignment of untilted ynpcs utilized a multireference approach, in mental change in their architectural scale between yeast the 10 best particles were aligned separately against the top 866 particles to generate new references; subsequently, a final best and vertebrates. Furthermore, a comparison of the two reference was calculated and used to realign the dataset. Resolution NPCs enforces new boundaries on our understanding was calculated using the differential phase residual for projections of what is required to build a functional NPC. What sort from half-datasets. Classification of the ynpcs into transporter and of NPC was directly ancestral to both yeast and vnpcs? spoke classes used dynamic clouds and hierarchical ascendant Two evolutionary scenarios could account for these difmerged classification (Frank, 1990; Akey, 1995), and final classes were ferences. In the first, the ynpc may have been streamwere using the dendrogram and visual inspection. The 3-D maps calculated using the R-weighted back projection method (Radlined from a larger common ancestral NPC, to function ermacher, 1988) and refined using projection onto convex sets (Akey in this small and rapidly dividing eukaryote. In a second and Radermacher, 1993), to correct for the missing cone. The 3-D model, early eukaryotes possessed a smaller and more maps in the text are comprised of the global transporter class (GBT, primitive NPC, which was retained by yeast while animal n 844) that contains all ynpcs with well-defined transporters, and cells expanded and elaborated their NPCs. Future studfiltered the best spoke class (SC, n 577). Final 3-D maps were Fourier- ies of NPCs in eukaryotes that diverged from the comto to 120 Å resolution, and surface models were thresholded mon ancestral cell before animals and fungi may distin- give a reasonable connectivity. The calculated volume of the GBT 3-D map used in the surface views corresponds to 150% of the guish between these possibilities. However, yeast NPCs expected mass of 60 MDa, similar to thresholds used for other low are likely to have retained or recapitulated features from resolution maps (Akey and Radermacher, 1993). an earlier and smaller NPC that is ancestral to all modern NPCs. With a completed yeast genome sequence in hand, the smaller ynpc will now serve as a focus for Mass Analysis of ynpcs future labeling and transport studies, which may provide The dform ynpcs were subjected to mass analysis by scanning transmission electron microscopy using a wet film sample preparapositional and functional information on individual nups tion method (Wall and Hainfeld, 1986), with TMV as a mass standard. and their roles in nucleocytoplasmic transport. In some cases, fixation with 0.1% formaldehyde/0.2% glutaraldehyde in sample buffer was done for 25 min. Images were recorded Experimental Procedures on the STEM1 at Brookhaven National Laboratory at 40 kv with a 2.5 Å diameter beam. The intensity of ynpcs was integrated after Specimen Preparation and Electron Microscopy background correction and multiplied by the TMV scale factor to NPCs were isolated from Saccharomyces uvarum (NCYC74) in detergent-extracted and membrane-associated forms. The fractions obtain the correct mass. Programs were provided by Dr. G. Sosinsky.

11 Architecture of the Yeast Nuclear Pore Complex 233 Acknowledgments new regular, fibrous lattice attached to the baskets of the nucleoplasmic face of the nuclear pores. J. Cell Biol. 119, This work was supported by grants from the NIH (C. W. A.). Goldberg, M.W., and Allen, T.D. (1996). The nuclear pore complex M. P. R. thanks G. Blobel and the HHMI for support. Q. Y. received and lamina: three dimensional structures and interactions deterpartial support from a Boston University GSRF. We thank I. V. Akey mined by field emission in-lens scanning electron microscopy. J. and R. Noltefor help with figures, J.Wall andm. Simon for assistance Mol. Biol. 257, with STEM data collection, E. Bullitt for critical comments, J. Kilmar- Gorlich, D., and Mattaj, I.W. (1996). Nucleocytoplasmic transport. tin for images of serial sectioned yeast cells, M. Radermacher for Science 271, helpful discussions, and L. Rome and P. Stewart for sharing the results of a vault database search. Greber, U.F., and Gerace, L. (1995). Depletion of calcium from the lumen of the endoplasmic reticulum reversibly inhibits passive diffusion and signal-mediated transport into the nucleus. J. Cell Biol. Received September 24, 1997; revised November 13, , Greber, U.F., Senior, A., and Gerace, L. (1992). A major glycoprotein of the nuclear pore complex is a membrane-spanning polypeptide References with a large lumenal domain and a small cytoplasmic tail. EMBO J. 9, Akey, C.W. (1989). Interactions and structure of the nuclear pore Hinshaw, J.E., Carragher, B.O., and Milligan, R.A. (1992). Architeccomplex revealed by cryo-electron microscopy. J. Cell Biol. 109, ture and design of the nuclear pore complex. Cell 69, Jarnik, M., and Aebi, U.J. (1991). Toward a more complete 3-D Akey, C.W. (1990). Visualization of transport-related configurations structure of the nuclear pore complex. Struct. Biol. 107, of the nuclear pore transporter. Biophys. J. 58, Kiseleva, E., Goldberg, M., Daneholt, B., and Allen, T.D. (1996). RNP Akey, C.W. (1995). Structural plasticity of the nuclear pore complex. export is mediated by structural reorganization of the nuclear pore J. Mol. Biol. 248, basket. J. Mol. Biol., 260, Akey, C.W., and Goldfarb, D.S. (1989). Protein import through the Kiseleva, E., Goldberg, M., Allen, T.D., and Akey, C.W. (1998). Strucnuclear pore complex is a multistep process. J. Cell Biol. 109, ture of the nuclear pore complex in Chironomus and the role of components in nuclear export of the Balbiani ring mesenger RNP. Akey, C.W., and Radermacher, M. (1993). Architecture of the Xeno- J. Cell Sci. in press. pus nuclear pore complex revealed by 3D cryo-electron microscopy. Kocsis, E., Cerritelli, M.E., Trus, B.L., Cheng, N., and Steven, A.C. J. Cell Biol. 122, (1995). Improved methods for determination of rotational symmet- Bastos, R., Ribas de Pouplana, L., Enarson, M., Bodoor, K., and ries in macromolecules. Ultramicroscopy 60, Burke, B. (1997). Nup84, a novel nucleoporin that is associated with Koepp, D.M., and Silver, P.A. (1996). A GTPase controlling nuclear CAN/Nup214 on the cytoplasmic face of the nuclear pore complex. trafficking: running the right way or walking RANdomly? Cell 87, J. Cell Biol. 137, Clayton, R.A, White, O., Ketchum, K.A., and Venter, J.C. (1997). The Kraemer, D.M., Strambio-de-Castillia, C., Blobel, G., and Rout, M.P. first genome from the third domain of life. Nature 387, (1995). The essential yeast nucleoporin NUP159 is located on the Crowther, R.A., and Amos, L.A. (1971). Harmonicanalysis of electron cytoplasmic side of the nuclear pore complex and serves in karyomicroscope images with rotational symmetry. J. Mol. Biol. 60, pherin-mediated binding of transport substrate. J. Biol. Chem. 270, Doye, V., and Hurt, E.C. (1997). From nucleoporins to nuclear pore Macaulay, C., Meier, E., and Forbes, D.J. (1995). Differential mitotic complexes. Curr. Opin. Cell Biol. 9, phosphorylation of proteins of the nuclear pore complex. J. Biol. Dubochet, J., Adrian, M., Chang, J., Homo, J.C., Lepault, J., McDow- Chem. 270, all, A.W., and Schultz, P. (1988). Cryo-electronmicroscopy of vitrified Mehlin, H., Daneholt, B., and Skoglund, U. (1992). Translocation specimens. Q. Rev. Biophys. 21, of a specific premessenger ribonucleoprotein particle through the Dworetzky, S.I., and Feldherr, C.M. (1988). Translocation of RNA- nuclear pore studied with electron microscope tomography. Cell coated gold particles through the nuclear pores of oocytes. J. Cell 69, Biol. 106, Newmeyer, D.D., and Forbes, D.J. (1988). Nuclear import can be Favreau, C., Worman, H.J., Wozniak, R.W., Frappier, T., and Cour- separated into distinct steps in vitro: nuclear pore binding and transvalin, J.C. (1996). Cell cycle dependent phosphorylation of nucleo- location. Cell 52, porins and the nuclear pore membrane protein gp210. Biochemistry Nigg, E.A. (1997). Nucleocytoplasmic transport: signals, mecha- 35, nisms and regulation. Nature 386, Feldherr, C.M., Kallenback, E., and Schulz, N. (1984). Movement of Paine, P.L., Moore, L.C., andhorowitz, S.B. (1975). Nuclear envelope a karyophilic protein through the nuclear pores of oocytes. J. Cell permeability. Nature 254, Biol. 99, Pante, N., and Aebi, U. (1996). Sequential binding of import ligands Fornerod, M., van Deursen, J., van Baal, S., Reynolds, A., Davis, to distinct nucleopore regions during their nuclear import. Science D., Murti, K.G., Fransen, J., and Grosveld, G. (1997). The human 273, homologue of yeast CRM1 is in a dynamic subcomplex with CAN/ Perez-Terzic, C., Pyle, J., Jaconi, M., Stehno-Bittel, L., and Clapham, Nup214 and a novel nuclear pore component Nup88. EMBO J. 16, D.E. (1996). Conformational states of the nuclear pore complex in duced by depletion of nuclear Ca 2 stores. Science 273, Frank, J. (1990). Classification of macromolecular assemblies stud- Powell, L., and Burke, B. (1990). Internuclear exchange of an inner ied as single particles. Q. Rev. Biophys. 23, nuclear membrane protein (p55) in heterokaryons: in vivo evidence Frank, J., Radermacher, M., Penczek, P., Ahu, J., Li, Y., Ladjadj, M., for the interaction of p55 with the nuclear lamina. J. Cell. Biol. 111, and Leith, A. (1996). SPIDER and WEB: processing and visualization of images in 3D electron microscopy and related fields. J. Struct. Radermacher, M. (1988). Three-dimensional reconstruction of single Biol. 116, particles from random and non-random tilt series. J. Electron Mi- Franke, W.W., and Scheer, U. (1974). Structures and functions of crosc. Tech. 9, the nuclear envelope. In The Cell Nucleus, ed. H. Busch. (New York: Reichelt, R., Holzenburg, A., Buhle, E.L, Jr., Jarnik, M., Engel, A., Academic Press Inc.), pp and Aebi, U. (1990). Correlation between structure and mass distri- Goldberg, M.W., and Allen, T.D. (1992). High resolution scanning bution of the nuclear pore complex and of distinct pore complex electron microscopy of the nuclear envelope: demonstration of a components. J. Cell Biol. 110,