Evidence That Vault Ribonucleoprotein Particles Localize to the Nuclear Pore Complex

Journal of Cell Science 106, 23-29 (1993) 23 Printed in Great Britain © The Company of Biologists Limited 1993

Evidence that vault ribonucleoprotein particles localize to the nuclear pore complex

Diane C. Chugani1,*, Leonard H. Rome2 and Nancy L. Kedersha3 1Department of Pharmacology, 2Department of Biological Chemistry and the Mental Retardation Research Center, UCLA School of Medicine, Los Angeles CA 90024, USA 3ImmunoGen, Inc., Cambridge, MA, USA *Author for correspondence at present address: Department of Pediatrics, Children’s Hospital of Michigan, Wayne State University, 3901 Beaubien Blvd, Detroit, MI 48201-9985, USA

SUMMARY

Vaults are cytoplasmic ribonucleoprotein organelles NPCs. The present study demonstrates that vaults that are highly conserved among diverse eukaryotic specifically associate with nuclei by both immunoblot- species. Their mass (12.9 MDa), diameter (26-35 nm) ting and immunofluorescence. Immunogold EM con- and shape (two halves, each with eightfold radial sym- firmed that vaults associate with the nuclear envelope metry) have recently been determined and are similar in tissue sections and with NPCs of isolated nuclei. to those ascribed to the central plug (or transporter) of the nuclear pore complex (NPC). The size and eightfold symmetry of the vault particle make it conducive to Key words: vaults, nuclear pore complex, ribonucleoprotein interacting physically in a complementary manner with particle, nuclear transport

INTRODUCTION Milligan, 1982), also termed the NPC transporter (Akey and Goldfarb, 1989), which consists of two equivalent halves Vaults are barrel-like cytoplasmic ribonucleoprotein parti- each with eightfold symmetry (Akey, 1990) having a cles (Kedersha and Rome, 1986) that are highly conserved measured diameter of 30-35 nm by cryo-electron and broadly distributed among eukaryotes ranging from microscopy (Akey, 1989; Reichelt et al., 1990). The aver- humans to Dictyostelium (Kedersha et al., 1990). These age mass of the central plug of Xenopus laevis NPCs is organelles have been well characterized at the molecular estimated to be 13 MDa using STEM (Reichelt et al., 1990). level, and contain four major proteins and a small RNA, In summary, the dimensions, mass and geometry of the cen- with the major vault protein (MVP; 95-104 kDa, depend- tral plug of the NPC are comparable to those of the vault ing on the species) comprising approximately 75% of the particle. In order to address the issue of whether vaults con- mass of the vault particle (Kedersha et al., 1990). The mass stitute NPC plugs/transporters, we undertook studies to of rat liver vaults as determined by scanning transmission determine if vaults are associated with the nucleus and, electron microscopy (STEM) is 12.9 MDa, and the dimen- more speciﬁcally, with the NPC. sions of the particle measure 34 nm ´ 60 nm by negative stain, 26 nm ´ 49 nm by cryo-electron microscopy of frozen-hydrated vaults, and 35 nm ´ 59 nm by STEM (Ked- ersha et al., 1991). When vaults are spread on polylysine- MATERIALS AND METHODS coated mica, freeze-etched and viewed by electron microscopy, a single vault barrel can be shown to open into Western blot analysis of isolated nuclei two ﬂower-like structures in which eight rectangular petals Nuclei were prepared from Dictyostelium discoidium and rat liver are joined by hooks to a central ring (Kedersha et al., 1991; as described by Davis and Blobel (1986). Samples of nuclei were model shown in Fig. 1). extracted in buffer containing either 1% Triton X-100 detergent Vaults display structural similarities to the central plug or NaCl (150 mM or 500 mM) for 30 min at room temperature. of the nuclear pore complex (NPC). NPCs are octagonal Nuclei were separated from extracted protein by a 10 min cen- organelles that span the inner and outer membranes of the trifugation in a microfuge at 12,000 g. Pellets, supernatants and untreated nuclei were adjusted to the same volume and equiva- nuclear envelope, providing a channel for nucleocytoplas- lent samples were boiled for 10 min in SDS sample buffer. West- mic transport (for reviews see: Franke et al., 1981; Ding- ern blots were performed as described by Towbin et al. (1979). wall and Laskey, 1986; Newport and Forbes, 1987; Gerace Vault antiserum raised against vaults from Dictyostelium (pre- and Burke, 1988; Feldherr and Akin, 1990). Inserted into pared as described by Kedersha and Rome, 1986) was used at a the center of the channel is the central plug (Unwin and dilution of 1:1000. 24 D. C. Chugani, L. H. Rome and N. L. Kedersha

Immunofluorescence of fibroblasts and rat liver rabbit antisera raised against rat liver vaults (1:2000; rabbit N2 nuclei [13]), preimmune serum (1:2000 dilution), or polyclonal rabbit Normal rat fibroblasts were grown to near confluency on glass antibody raised against the nuclear pore protein gp62 from rat coverslips and fixed for 10 min in methanol (- 20°C). Rat liver (1:500, generously provided by Douglass Forbes, UCSD), fol- nuclei, prepared as described above, were washed twice in PBS lowed by a 1 h incubation with goat anti-rabbit FITC (1:200 dilution, Boehringer-Mannheim). Nuclei and fixed cells were mounted (phosphate buffered saline; 10 mM NaH2PO4, 150 mM NaCl, pH 7.4). A milky suspension of nuclei in PBS was allowed to adhere in Gelvatol mounting medium (Fukio et al., 1987) and viewed on to gelatin-coated slides in a well made with a paraffin pen. After a Nikon Microphot FX microscope equipped with epifluorescence 10 min, the buffer was aspirated and the nuclei were fixed with optics and appropriate filters for the detection of fluorescein. In 4% paraformaldehyde in PBS for 10 min. addition, fixed cells were viewed on a Leitz confocal fluorescence Fixed cells and nuclei were blocked for 30 min with 5% normal microscope system. goat serum in PBS. Nuclei were then incubated with polyclonal Immunogold electron microscopy of isolated rat liver nuclei Nuclei were fixed by suspension in 1% glutaraldehyde for 10 min, pelleted in a clinical centrifuge (~800 g, 10 min) and suspended in TBS (Tris-buffered saline; 50 mM Tris-HCl, 154 mM NaCl) for 90 min, followed by pelleting and resuspension in 5% normal goat serum in TBS for 30 min. Pellets were suspended in affinity-purified vault antibody (1 mg/ml) or preimmune IgG (1 mg/ml) and incubated for 1 h with occasional mixing. Nuclei were washed by pelleting through 10 ml of 0.25 M sucrose in TBS. The nuclei were resuspended in 0.2% bovine serum albumin in TBS containing goat anti-rabbit IgG/gold conjugate (5 nm colloidal gold, Janssen). Nuclei were washed as above and suspended in 1% glutaraldehyde. The fixed nuclei were pelleted in a microfuge, osmicated and embedded in Spurr’s resin. Thin sections were viewed on a JEOL 1200EX electron microscope.

Immunogold electron microscopy in tissue sections Adult male Sprague Dawley rats were anesthetized with halothane and perfused through the left cardiac ventricle with 200 ml PBS, followed by 500 ml 4% paraformaldehyde, 0.1% glutaraldehyde in PBS through the left cardiac ventricle. Intes- tine and spleen (vault-rich tissues as determined by western blot- ting) were removed, cut into 1 mm blocks and incubated in the same fixative for 1 h. Tissue was washed in PBS and incubated for 30 min in 1% aqueous osmium tetroxide. Tissue was then dehydrated by stepwise transfer to 75% ethanol and embedded in LR White. Thin sections were mounted on nickel grids. Grids were floated for 10 min on 20 ml PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgCl2, pH 6.9), and 20 min on 5% normal goat serum in PHEM buffer. Affinity-purified vault antibody was prepared as previously described (Chugani et al., 1991). The preimmune IgG fraction from the same rabbit was isolated from whole serum using Protein A- Sepharose (Pharmacia). Affinity-purified vault antibody and preimmune IgG were used at 25 mg/ml in PHEM buffer containing 5% normal goat serum. Grids were floated on 20 ml of Fig. 1. Model of the vault organelle as the NPC plug. (A) The barrel-like structure of an intact vault. (B) Intact vault shown on the primary antibodies for 1 h and washed 4´ 5 min in 0.2% end as it would appear in an en face view of the NPC. (C) The bovine serum albumin (BSA) in PHEM buffer. Grids were then vault is depicted in an open conformation in which each half-vault floated on goat anti-rabbit IgG/gold conjugate (5 nm; Janssen) unfolds into a ‘flower’ consisting of eight petals, each joined by a diluted 1:2 in 0.2% BSA in PHEM buffer for 1 h, followed by curved filament to a common central ring. (D) Hypothetical model washing as above. In some experiments, grids were then flo a t e d of the vault as the central plug of the NPC viewed en face in open on wheat germ agglutinin/10 nm gold conjugate (Sigma) diluted conformation in the center of the pore. The four images were 1:5 in 0.2% BSA in PHEM buffer for 1 h, followed by wash- generated by clockwise rotation of the top half-vault flower with ing as above. Sections were then fixed for 2 min in 2% glu- respect to the bottom one, each flower shown in a different shade taraldehyde, washed 3 times in distilled water, and stained for of gray. The ‘spokes’ of the NPC are shown in black, and the pore 2 min with 10% uranyl acetate. Sections were viewed on a JEOL channel is shown in white. 1200EX electron microscope. Vaults localize to the NPC 25

RESULTS a group of NPC glycoproteins termed nucleoporins (Holt and Hart, 1986; Hanover et al., 1987), as well as numer- Immunoblot analysis of washed nuclei ous other cellular glycoproteins, and has been used exten- Isolated nuclei from Dictyostelium and rat liver were puri- sively both to label NPCs and to block nucleocytoplasmic fied from detergent-lysed cells by pelleting through 2.3 M transport (Akey and Goldfarb, 1989; Finlay et al., 1987; sucrose (Davis and Blobel, 1986). Vault protein was present Yoneda et al., 1987; Dabauvalle et al., 1988; Adam et al., in purified nuclei (Fig. 2, lanes G and H). Washing isolated 1990). WGA-rhodamine staining demonstrated punctate nuclei with high salt or the detergent Triton X-100 removed binding throughout the cytoplasm, but the loci were larger some but not all of the vault protein (Fig. 2, lanes A, C, E than those observed with vault staining (Fig. 3B). WGA, for Dictyostelium; I, K, M for rat liver nuclei). These data in addition, diffusely stained the entire cell surface. A indicate that vault particles reversibly associate with the prominent perinuclear rim of WGA staining, which has nuclear fraction. The fact that some vault protein remains been demonstrated in isolated nuclei and has been reported with nuclei even after harsh washing suggests that the inter- for nuclear pore antigens (Snow et al., 1987), was not appar- action of vaults with nuclei is specific. These results are ent with WGA staining of whole cells. Nuclear membrane consistant with the behavior of the NPC plugs, which are staining was apparently masked by an abundance of cyto- partially lost from NPCs during nuclei isolation procedures plasmic WGA staining in the perinuclear region. Perinu- and have been reported to be transiently associated with the clear staining with vault antibody could be unmasked by NPC rather than integral components (Reichelt et al., 1990). Vault and WGA immunofluorescence in cultured fibroblasts Immunofluorescent labeling of cultured fibroblasts with anti-vault antisera produced a fine punctate pattern of staining widely distributed throughout the cytoplasm (Fig. 3A). Cells did not display staining typical of nuclear membrane antigens, although some perinuclear staining was apparent in cells that displayed a relatively low abundance of vault immunofluorescent loci (arrowheads, Fig. 3A). These cells were double-stained with wheat germ agglutinin (WGA) conjugated to rhodamine, to determine if vaults were coincident with NPCs at the level of light microscopy. WGA binds to N-acetylglucosamine residues that are present on

Mr ´ 10- 3

Fig. 2. Western blot analysis of vault protein in isolated Dictyostelium nuclei. Lane A, 1% Triton X-100 pellet; lane B, 1% Triton X-100 supernatant; lane C, 150 mM NaCl pellet; lane D, 150 mM NaCl supernatant; lane E, 500 mM NaCl pellet; lane F, 500 mM NaCl supernatant; lane G, untreated nuclei, and vault protein in isolated rat liver nuclei lane H, untreated nuclei; lane I, 1% Triton X-100 pellet; lane J, 1% Triton X-100 supernatant; lane Fig. 3. Immunofluorescence localization of vaults and WGA in K, 150 mM NaCl pellet; lane L, 150 mM NaCl supernatant; lane normal rat fibroblasts. Cells were double-stained with: (A) anti- M, 500 mM NaCl pellet; lane N, 500 mM NaCl supernatant. vault antisera followed by goat anti-rabbit fluorescein, and Dictyostelium major vault protein is a doublet at Mr 95,000. Rat (B) wheat germ agglutinin conjugated to rhodamine, and viewed major vault protein is at Mr 104,000. on a Nikon Microphot FX microscope. Bar, 0.5 mm. 26 D. C. Chugani, L. H. Rome and N. L. Kedersha confocal microscopy. A non-confocal image (300 nm z- stained with anti-vault antibody consistently displayed axis) is shown (Fig. 4A) together with a confocal image clusters of 6-12 gold particles per vault (Fig. 6, top right). (20 nm z-axis) of the same cell (Fig. 4B). The confocal While gold labeling of the isolated nuclei was sparse, image shows markedly enhanced vault immunoreactivity at the rim of the nucleus. Vault immunolocalization on isolated nuclei Another experiment was performed to test whether vault staining of the nuclear membrane was masked by the abundance of vaults in the cytoplasm. Vault immunofluorescent staining was performed on isolated nuclei. I m m u n o fluorescent labeling of isolated rat liver nuclei with anti-vault antibody produced specific punctate staining on the surface of the nuclear envelope (Fig. 5A). The staining was less uniform and less intense than that obtained using antibodies specific for the NPC polypep- tide p62 (Fig. 5C), yet the staining was distinctly positive as compared to the nonimmune control (Fig. 5B). It would be expected that an antibody recognizing the plug would stain less than that of an integral pore component, due to loss during the isolation and staining procedures as dis- cussed above. The presence of vaults was examined at the EM level using anti-vault antibody and gold-conjugated secondary antibodies on isolated fixed nuclei prior to embedding and sectioning. Only clusters of gold particles were considered to represent a vault, since isolated vaults

Fig. 4. Immunofluorescence localization of normal rat fibroblasts Fig. 5. Immunofluorescence localization of vaults on isolated stained with anti-vault antisera viewed (A) in a nonconfocal mode nuclei. (A) Anti-vault antibody; (B) pre-immune serum; (C) anti- (300 nm z-axis) and (B) in a confocal mode (20 nm z-axis) on a p62 antibody. The focus shown is mainly on the surface of the Leitz confocal fluorescence microscope system. N, center of nuclei to best display punctate nature of staining. No label was nucleus. Bar, 0.25 mm. seen inside the nuclei with either antibody. Bar, 0.5 mm. Vaults localize to the NPC 27

analysis of 50 gold clusters revealed that over one third Vault immunolocalization in tissue sections could be unambiguously assigned to structures morpho- Vault distribution in situ was examined using immunogold logically identiﬁed as pores. A collage of immunogold staining on LR White-embedded sections of rat spleen and labeled NPCs is shown (Fig. 6). intestine, tissues high in vault expression (D. Chugani and

Fig. 6. Top panel: isolated vaults visualized with uranyl acetate, both unstained (left) and stained with anti-vault antibody followed by goat anti-rabbit IgG/gold conjugate (10 nm colloidal gold) (right). Note that vaults are largely composed of protein and, therefore, are not visualized in sections, due to their inability to Fig. 7. Immunogold electron microscopy in tissue sections. take up the stain. Bottom four panels: immunogold staining of (A) Intestine labeled for vaults (5 nm gold) and WGA (10 nm isolated rat liver nuclei. Clusters of gold particles (5 nm) can be gold); (B) spleen labeled for vaults; (C) intestine, a higher seen to decorate NPCs. Controls performed with nonimmune IgG, magniﬁcation of A; (D) intestine showing invagination in nuclear displayed few gold particles and were devoid of gold clusters at membrane labeled for vaults (5 nm gold) and WGA (10 nm gold). NPCs (not shown). Bar, 100 nm. Bars, 100 nm. 28 D. C. Chugani, L. H. Rome and N. L. Kedersha

N. Kedersha, unpublished data). Vault-specific gold clus- plug and the cytoplasmic ring on the cytoplasmic face of ters were seen throughout the cytoplasm of the cells, con- the NPC (Pante et al., 1992). In addition, WGA-gold sistent with the immunofluorescence studies, and at the labeled the distal ring of structures referred to as nuclear nuclear periphery (Fig. 7), often at circular profiles that baskets on the nuclear face of the NPC. The lack of WGA appeared to represent NPCs (Fig. 7D). As this technique binding, therefore, may be another trait that vaults and the results in poor nuclear membrane morphology, NPCs were plug have in common. double-labeled using 5 nm gold secondary antibody for Structural details of the central plug have been visual- vault-immunoreactivity and 10 nm gold conjugated to ized using eightfold averaging enhancements of cryo-elec- WGA. Coincident staining for vaults and WGA at the tron micrographs of Necturus oocyte NPCs (Akey, 1990). nuclear periphery (Fig. 7A, C, D) suggested that vaults were This method produces different density patterns of the associated with NPCs. plug/transporter, suggesting that the plug can display different conformations. In an attempt to reconcile these data with the hypothesis that vaults constitute the NPC plug, we DISCUSSION have devised a model that incorporates the rotation of one half of an open vault with respect to its other half (Fig. Because of similarities in structural characteristics between 1D). the vault particle (Kedersha et al., 1991) and the NPC plug The present study demonstrates that vaults are found in (Reichelt et al., 1990; Akey, 1990), we asked whether some association with the nuclear membrane and NPCs, but the portion of vaults interacts with nuclei. The present studies data do not conclusively establish that the vault particle were undertaken to determine whether vaults associate with actually forms the NPC central plug. These data are equally nuclei and NPCs. Previous subcellular fractionation studies compatable, for example, with the vaults interacting with had demonstrated that the vast majority of vault particles the plug through common structural features. A definitive were located in the cytoplasm (Kedersha and Rome, 1986). answer as to whether vaults are in fact the plugs might This was confirmed by immunofluorescence localization of require the isolation of this elusive NPC component. Since vaults in cultured fibroblasts (Fig. 3A). Staining of whole the bulk of the vaults are located in the cytoplasm, we cells did not, for the most part, demonstrate the character- suggest that vaults may act as shuttles between the nucleus istic nuclear rim staining that has been demonstrated for and the cytoplasm. Functional and genetic studies are in nuclear antigens, although some perinuclear staining was progress that may determine whether vaults are involved in apparent in cells that displayed a relatively low abundance nucleocytoplasmic transport. of cytoplasmic vault immunofluorescent loci (Fig. 3A, arrows). The lack of nuclear rim staining is likely due to We thank Drs Robert Ackermann and Elizabeth Neufeld for masking of the nuclear signal by the abundance of vaults their helpful comments on this manuscript, Sharon Belkin for the located in the cytoplasm as seen with WGA (Fig. 3B), since drawings in Fig. 1, Birgitta Sjostrand for technical assistance in a rim of nuclear membrane staining could be discerned by preparing samples for electron microscopy and Andrew Jacobson confocal microscopy (Fig. 4B). for help with confocal microscopy. This work was supported by DOE grant DE-FC03-87ER60615 and USPHS grant GM 38097. Isolated nuclei contained significant amounts of vault The UCLA electron microscopy facility was supported by protein, most of which was not removed by harsh washing USPHS-1-P41 GM27566. with high salt or detergent. This suggests that the association of vault protein with nuclei is of high affinity and specific. The specificity of vault interaction was also REFERENCES demonstrated by immunolocalization of isolated nuclei at the EM level. Over one-third of the vault immunoreactiv- Adam, S. A., Sterne Marr, R. E. and Gerace, L. (1990). Nuclear protein ity on isolated nuclei occurred unambiguously at NPC pro- import in permeabilized mammalian cells requires soluble cytoplasmic files (Fig. 6). In tissue sections, vault immunoreactivity and factors. J. Cell Biol. 111, 807-816. WGA-gold colocalized at the nuclear envelope. Akey, C. W. (1989). Interactions and structure of the nuclear pore complexes revealed by cryo-electron microscopy. J. Cell Biol. 109, 955- Do vaults constitute the central plug/transporter? The 970. central plug/transporter of the NPC has not been isolated Akey, C. W.(1990). Visualization of transport-related configurations of the and, therefore, its biochemical constituents are not known. nuclear pore transporter. Biophys. J. 58, 341-355. Colloidal gold conjugated to wheat germ agglutinin has Akey, C. W. and Goldfarb, D. S. (1989). Protein import through the nuclear pore complex is a multistep process. J. Cell Biol. 109, 971-982. been localized to the central region of the nuclear pore com- Chugani, D. C., Kedersha, N. L. and Rome, L. H. (1991). Vault plex in spreads of oocyte nuclei (Akey and Goldfarb, 1989). immunofluorescence in brain: new insights regarding the origin of Some investigators have interpreted these results to indi- microglia. J. Neurosci. 11, 256-268. cate that the central plug/transporter is composed of the Dabauvalle, M.-C., Schultz, B., Scheer, U. and Peters, R. (1988). nucleoporins, a group of NPC-associated glycoproteins that Inhibition of nuclear accumulation of karyophilic proteins in living cells by microinjection of the lectin wheat germ agglutinin. Exp. Cell Res.174, bind WGA (Holt and Hart, 1986; Hanover et al., 1987). 291-296. This interpretation of WGA localization would argue Davis, L. I. and Blobel, G. (1986). Identification and characterization of a against vaults forming the plug/transporter, since rat liver nuclear pore complex protein. Cell 45, 699-709. vaults do not react with WGA (Kedersha and Rome, 1986). Dingwall, C. and Laskey, R. A. (1986). Protein import into the cell More recent studies, however, suggest that WGA does not nucleus. Annu. Rev. Cell Biol.2, 367-390. Franke, W. W., Scheer, U., Krohne, G. and Jarasch, E.-D. (1981). The label the central plug in projection images, but it appears nuclear envelope and the architecture of the nuclear periphery. J. Cell to label a spoke component at a site between the central Biol. 91, 39s-50s. Vaults localize to the NPC 29

Feldherr, C. M. and Akin, D. (1990). EM visualization of a novel ribonucleoprotein particle: large structures contain a single nucleocytoplasmic transport processes. Electron Microsc. Rev. 3, 73-86. species of small RNA. J. Cell Biol. 103, 699-709. Finlay, D. R., Newmeyer, D. D., Price, T. M. and Forbes, D. J. (1987). Newport, J. W. and Forbes, D. J. (1987). The nucleus: structure, function, Inhibition of in vitro nuclear transport by a lectin that binds to nuclear and dynamics. Annu. Rev. Biochem. 56, 535-565. pores. J. Cell Biol. 104, 189-200. Pante, N., Jarnik, M., Heitlinger, E. and Aebi, U. (1992). Structure, Fukio, Y., Yumura, S. and Yumura, T. K. (1987). Agar-overlay assembly and interactions of the nuclear lamina and the nuclear pore immunoﬂuorescence: high-resolution studies of the cytoskeletal complex. EMSA 92 Proc. 50, 490-491. components and their changes during chemotaxis. Meth. Cell Biol. 28, Reichelt, R., Holzenburg, A., Buhle, E. L. Jr., Jarnik, M., Engel, A. and 347-356. Aebi, U. (1990). Correlation between structure and mass distribution of Gerace, L. and Burke, B. (1988). Functional organization of the nuclear the nuclear pore complex and of distinct pore complex components. J. envelope. Annu. Rev. Cell Biol.4, 335-374. Cell Biol. 110, 883-894. Hanover, J. A., Cohen, C. K., Willingham, M. C. and Park, M. K. Snow, C. M., Senior, A. and Gerace, L. (1987). Monoclonal antibodies (1987). O-linked N-acetylglucosamine is attached to proteins of the identify a group of nuclear pore complex glycoproteins. J. Cell Biol. 104, nuclear pore: evidence for cytoplasmic glycosylation. J. Biol. Chem. 262, 1143-1156. 9887-9894. Towbin, H., Staehdin, T. and Gordon, J. (1979). Electrophoretic transfer Holt, G. D. and Hart, G. W. (1986). The subcellular distribution of of proteins from polyacrylamide gels to nitrocellulose sheets: procedure terminal N-acetylglucosamine moieties: localization of a novel protein- and some applications. Proc. Nat. Acad. Sci. USA 76, 4350-4354. saccharide linkage, O-linked GlcNAc. J. Biol. Chem. 261, 8049-8057. Unwin, P. N. T. and Milligan, R. A. (1982). A large particle associated Kedersha, N. L., Heuser, J. E., Chugani, D. C. and Rome, L. H. (1991). with the perimeter of the nuclear pore complex. J. Cell Biol. 93, 63-75. Vaults. III. Vault ribonucleoprotein particles open into ﬂower-like Yoneda, Y., Imamoto-Sonobe, N., Yamaizumi, M. and Uchida, T. structures with octagonal symmetry. J. Cell Biol. 112, 225-235. (1987). Reversible inhibition of protein import into the nucleus by wheat Kedersha, N. L., Miquel, M.-C., Bittner, D. and Rome, L. H. (1990). germ agglutinin injected into culture cells. Exp. Cell Res. 173, 586-595. Vaults. II. Ribonucleoprotein structures are highly conserved among higher and lower eukaryotes. J. Cell Biol. 110, 895-901. (Received 19 January 1993 - Accepted, in revised form, Kedersha, N. L. and Rome, L. H. (1986). Isolation and characterization of 20 May 1993)