Structure and Assembly of the Nup84p Complex

Abstract

The Nup84p complex consists of five nucleoporins (Nup84p, Nup85p, Nup120p, Nup145p-C, and Seh1p) and Sec13p, a bona fide subunit of the COPII coat complex. We show that a pool of green fluorescent protein–tagged Sec13p localizes to the nuclear pores in vivo, and identify sec13 mutant alleles that are synthetically lethal with nup85Δ and affect the localization of a green fluorescent protein–Nup49p reporter protein. In the electron microscope, sec13 mutants exhibit structural defects in nuclear pore complex (NPC) and nuclear envelope organization. For the assembly of the complex, Nup85p, Nup120p, and Nup145p-C are essential. A highly purified Nup84p complex was isolated from yeast under native conditions and its molecular mass was determined to be 375 kD by quantitative scanning transmission electron microscopy and analytical ultracentrifugation, consistent with a monomeric complex. Furthermore, the Nup84p complex exhibits a Y-shaped, triskelion-like morphology 25 nm in diameter in the transmission electron microscope. Thus, the Nup84p complex constitutes a paradigm of an NPC structural module with distinct composition, structure, and a role in nuclear mRNA export and NPC bio- genesis.

Introduction

In eukaryotic cells, transport between the nucleus and cytoplasm takes place through the nuclear pore complexes (NPCs) (Ohno et al. 1998). EM combined with image reconstruction has revealed many structural details of the NPC framework (for review see Stoffler et al. 1999). Accordingly, one of the most prominent NPC substructures is the central spoked ring, which exhibits eightfold symmetry and is embedded within the plane of the double nuclear membrane. The spoke complex is embraced by two ring assemblies that face the nuclear and cytoplasmic sites of the nuclear envelope, respectively. The cytoplasmic ring carries eight globular particles, from which eight short fibrils protrude into the cytoplasm. The nuclear ring is capped by a highly regular structure called the nuclear basket. Although the known vertebrate nucleoporins do not exhibit high sequence homology with their yeast counterparts, the overall NPC structure (spoke complex and outer and inner rings) is conserved between yeast and vertebrates (Yang et al. 1998). Furthermore, peripheral structural elements such as cytoplasmic fibrils, nuclear baskets, and NPC-attached intranuclear filaments were recently reported to be also present in yeast NPCs as revealed by EM (Fahrenkrog et al. 1998; Strambio-de-Castillia et al. 1999).

The molecular dissection of the yeast NPC is close to completion, i.e., most of the yeast nucleoporins are known to date. This fast progress was possible because multiple sources of information, e.g., yeast genetic screens with nucleoporin mutants, affinity-purification of tagged nucleoporins, and identification of their associated partner proteins, a NPC isolation procedure, and completion of the yeast genome sequencing project, could be successfully combined (for reviews see Rout and Wente 1994; Doye and Hurt 1997; Fabre and Hurt 1997). When nucleoporin mutants are analyzed for an altered NPC structure and function, two distinct phenotypical defects become apparent: (a) inhibition of nucleocytoplasmic transport reactions such as nuclear mRNA export or nuclear protein import; and (b) structural abnormalities of the NPCs and the nuclear membrane. Interestingly, distinct nucleoporin mutants exhibit both defects, whereas others have only transport or structural defects, respectively. To the first category belong several members of the Nup84p complex, a large assembly that consists of Nup84p, Nup85p, Nup120p, Nup145p-C, Seh1p, and Sec13p. Nucleoporin members of this complex exhibit nuclear mRNA export defects, but not inhibition of nuclear protein import (Aitchison et al. 1995; Goldstein et al. 1996; Siniossoglou et al. 1996; Teixeira et al. 1997). Therefore, it was suggested that the Nup84p complex is linked to the mRNA export machinery, consistent with the recent finding that Nup85p interacts with the mRNA exporter complex Mex67p–Mtr2p (Segref et al. 1997; Santos-Rosa et al. 1998). However, the Nup84p complex is also linked to the biogenesis pathway of NPCs, since mutation of members of the Nup84p complex causes severe defects in the organization and distribution of NPCs within the nuclear envelope. The most apparent phenotype is NPC clustering, connected with nuclear envelope abnormalities such as long nuclear membrane extensions, which penetrate deeply into the cytoplasm (Siniossoglou et al. 1996). A different structural defect associated with nup116 and gle2 mutants is a nuclear membrane seal over the NPCs (Wente and Blobel 1993; Murphy et al. 1996; Bailer et al. 1998). Thus, nucleoporins have specific roles in the biogenesis, distribution, and structural integrity of NPCs and nucleocytoplasmic transport.

In contrast to transport routes through the pores, little is known about how NPCs assemble and become incorporated into the nuclear membrane. Progress has been made in the Xenopus oocyte system, which can be used to study and manipulate NPC assembly (Goldberg et al. 1997), but mechanistic aspects still remain unknown. Interesting in this context is the finding that the Nup84p complex contains one unexpected protein, Sec13p, which could point to a vesicular biogenesis step during NPC formation. Sec13p is a subunit of the COPII complex, which is involved in vesicle transport from the ER to the Golgi apparatus (Kaiser and Ferro-Novick 1998).

Here, we have further investigated the role of the bona fide COPII coat subunit, Sec13p, as the most unexpected member of the Nup84p complex. We show that a pool of Sec13p colocalizes with clustered NPCs in nup133 ⁻ cells. Furthermore, sec13 temperature-sensitive (ts) alleles were obtained that are genetically linked to NUP85 and exhibit an abnormal NPC organization. Gel filtration chromatography revealed that Sec13p is a specific member of the large Nup84p complex. Finally, we demonstrate by negative staining and rotary shadowing that the highly purified Nup84p complex exhibits a distinct Y-shaped structure with an average diameter of 25 nm.

Materials and Methods

Yeast Strains, Microbiological Techniques, Plasmids, and DNA Manipulations

The yeast strains used in this study are listed in Table . For construction of double disruption strains, haploid strains of opposite mating type and carrying the indicated gene disruptions and gene mutations with an appropriate shuffle plasmid (URA3 containing ARS/CEN plasmid) were mated. The haploid strain RSY282 (kindly provided by Dr. Randy Schekman, University of California, Berkeley, Berkeley, CA), which harbors the sec23-1 mutation, was mated to strain RS453 of opposite mating type. Diploids growing on selective medium were sporulated, and after tetrad analysis haploid double mutants were selected. Standard DNA manipulations (restriction analysis, end filling, ligation, PCR amplification, and DNA sequencing) and microbiological techniques (growth and transformation of yeast and E. coli strains, plasmid recovery, mating, and tetrad analysis) were done as described earlier (Siniossoglou et al. 1996). The following plasmids were used: pUN100, ARS1/CEN4 plasmid with the LEU2 marker; pRS314 and pRS316, ARS4/CEN6 plasmids with the TRP1 and URA3 markers, respectively; pASZ11-ADE2-green fluorescent protein (GFP)-NUP49, a SacI/BamHI fragment containing the NUP49–GFP fusion was inserted into the SacI/BamHI site of pASZ11-ADE2; and pASZ11-ADE2-GFP-PUS1 (Siniossoglou et al. 1998).

Table 1.

Yeast Strains

	Yeast strains
RS453	MATa/α,ade2/ade2,his3/his3,leu2/leu2,trp1/trp1,ura3/ura3
NUP120-ProtA	Mata,ade2,his3,leu2,trp1,ura3,nup120::HIS3 (pRS315-LEU2-NUP120-ProtA)
NUP84-ProtA	Mata,ade2,his3,leu2,trp1,ura3,Δnup84::HIS3 (pUN100-LEU2-NUP84-ProtA)
nup84−	MATa,ade2,his3,leu2,trp1,ura3,nup84::HIS3
nup85Δ	MATa,ade2,his3,leu2,trp1,ura3,Δnup85::HIS3 (partial disruption/deletion)
seh1−	MATa,ade2,his3,leu2,trp1,ura3,seh1::HIS3
SEC13 shuffle	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3 (pRS316-URA3-SEC13-Myc)
ProtA-NUP85	MATa,ade2,his3,leu2,trp1,ura3,Δnup85::HIS3 (pUN100-LEU2-ProtA-NUP85)
NUP84-ProtA	MATa,ade2,his3,leu2,trp1,ura3,nup84::HIS3 (pUN100-LEU2-NUP84-ProtA)
ProtA-TEV-NUP85	MATa,ade2,his3,leu2,trp1,ura3,nup85::HIS3 (pUN100-LEU2-ProtA-TEV-NUP85)
seh1(Δ248-288)-ProtA	MATa,ade2,his3,leu2,trp1,ura3,seh1::HIS3 (pRS315-LEU2-seh1 (Δ248-288)-ProtA)
SEC13-GFP	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3 (pUN100-LEU2-SEC13-GFP)
SEC13-GFP/nup133−	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,nup133::HIS3 (pUN100-LEU2-SEC13-GFP)
sec13-3	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3 (pUN100-LEU2-sec13-3)
SEC13-ProtA	Mata,ade2,his3,leu2,trp1,ura3,sec13::HIS3 (pUN100-LEU2-SEC13-ProtA)
SEC13/seh1−	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,seh1::HIS3 (pRS316-URA3-SEC13-Myc)
SEC13/nup84−	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,nup84::HIS3 (pRS316-SEC13-Myc)
SEC13/nup85Δ	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,nup85Δ::HIS3 (pRS316-SEC13-Myc)
sec13-34/nup85Δ	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,nup85Δ::HIS3 (pUN100-LEU2-sec13-34)
sec13-14/nup85Δ	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,nup85Δ::HIS3 (pUN100-LEU2-sec13-34)
SEH1-ProtA	MATa,ade2,his3,leu2,trp1,ura3,seh1::HIS3 (pRS316-URA3-SEH1-ProtA)
sec13-3/NIC96-GFP	MATa,ade2,his3,leu2,trp1,ura3,sec13::HIS3,nic96::HIS3 (pUN100-LEU2-sec13-3,pRS314-TRP1-NIC96-GFP)
sec23-1	MATα,ade2,his3,leu2,trp1,ura3,sec23-1
NUP145-ProtA	MATa,ade2,his3,leu2,trp1,ura3,nup145::HIS3 (pUN100-LEU2-NUP145-C-ProtA)

Construction of Fusion Genes and NUP85 and SEH1 Truncation Mutants

To construct protein A (ProtA)-TEV-Nup85p, which contains the tobacco etch virus (TEV) protease cleavage site (ENLYFQG) between the ProtA tag and NUP85 ORF, an NdeI site was generated with a primer that hybridizes at the ATG start codon of NUP85 (5′-TTT TTC ATA TGA CAA TCG ATG AT-3′), and with the help of an internal primer hybridizing 1.32 kb after the initiator ATG, a PCR fragment was generated and digested with NdeI/SalI. The resulting fragment was subcloned into the NdeI/SalI sites of the pROEX-1 vector (GIBCO BRL Life Technologies). The NdeI site in this vector is six nucleotides downstream of the last codon coding for the TEV recognition site. An EcoRV site, nine nucleotides upstream of the first codon coding for the TEV site, was used to release an EcoRV/SalI fragment from the pROEX-1-TEV-NUP85 (NdeI/SalI) construct. This fragment, containing the TEV site and part of the NUP85 ORF, was inserted into a pBluescript vector containing the NOP1 promoter joined to the ProtA tag, cut with EcoRV/SalI. This pBS-NOP1::ProtA-TEV-NUP85 (NdeI/SalI) construct was finally cut with SacI/NcoI to release a NOP1::ProtA-TEV-NUP85 (SacI/NcoI) fragment and was then ligated into a pUN100-NUP85 plasmid, previously cut with SacI/NcoI. The TEV-NUP85 junction in the final pUN100-ProtA-TEV-NUP85 was verified by DNA sequencing. For the construction of the SEC13-GFP and SEH1-GFP fusions, the pUN100-SEC13-ProtA and pRS316-SEH1-ProtA plasmids (Siniossoglou et al. 1996) were digested with BamHI and the sequence coding for the ProtA tag was replaced by a BamHI GFP cassette encoding the S65T/V163A variant of the GFP.

For the construction of the ProtA-Nup85pΔN truncation mutant, a 1.7-kb DraI fragment encoding residues 82–744 plus 70 bp from the 3′ untranslated region was subcloned into pBluescript-pNOP1-ProtA, cut with HindIII, and blunt-ended. The resulting plasmid coded for an NH₂-terminal tagged ProtA-Nup85p-C in which the L in position 82 (TTA) was kept (CTA) (LNCLR-QAFM). The sequence of the junction between the ProtA and the COOH-terminal portion of Nup85p is LQEFDIKL. Next, a 2.8-kb BamHI/XhoI fragment containing the NOP1::ProtA-NUP85-C fragment was subcloned into pRS314-TRP1, cut with BamHI/HindIII. To construct a ProtA-Nup85ΔCp mutant, a 2.25-kb SacI/partial-BglII fragment containing the pNOP1-ProtA cassette plus the first 453 amino acid residues of Nup85p (MTI-NEMLED) was subcloned from pUN100-NOP1::ProtA-NUP85 (Siniossoglou et al. 1996) into a pRS315-LEU2 vector and cut with SacI/BamHI. Next, a SmaI terminator sequence, containing several stop codons in the three possible reading frames, was cloned into the BglII-adjacent SmaI site of the poly-linker.

Isolation of sec13 ts Alleles

The SEC13-ProtA fusion gene was amplified by PCR under conditions that impair the fidelity of the Taq DNA polymerase. The buffer conditions were 6.5 mM MgCl₂, 0.5 mM MnCl₂, 1 mM dGTP, dCTP, dTTP, and 0.2 mM dATP. Around 20 ng of template DNA and 5 U of Taq polymerase were used. The PCR product was gel-purified and digested with SacI/HindIII to release the SEC13-ProtA fragment and to insert it in a pUN100-LEU2, cut with SacI/HindIII. The ts library was first transformed into E. coli and 5,000 transformants were obtained from which the plasmid DNA was isolated. 78% of the plasmids had SEC13-ProtA inserts, with a mutagenesis rate of 2% as judged by DNA sequencing of six randomly selected clones. The sec13::HIS3 disrupted strain complemented by pRS316-SEC13-Myc was transformed with the pUN100-LEU2-SEC13-ProtA library. About 2,200 yeast transformants selected on SDC-ura-leu plates were replica-plated on 5-fluoro-orotic acid plates. The 800 colonies that were growing on 5-fluoro-orotic acid were plated on yeast extract–peptone–d-glucose and tested for thermosensitive growth at 37°C. Finally, 17 sec13 mutants with a thermosensitive growth defect at 37°C were obtained.

Affinity-Purification of ProtA Fusion Proteins, TEV Cleavage, and Gel Filtration

Affinity-purification of ProtA fusion proteins by IgG-Sepharose chromatography was done according to Siniossoglou et al. 1998. 15 g of spheroplasts derived from the nup85::HIS3 strain and complemented by plasmid-borne ProtA-TEV-NUP85 was used for the purification. To isolate the Nup84p complex under native conditions, the ProtA-TEV-Nup85p fusion protein was isolated by IgG-Sepharose chromatography as described (Siniossoglou et al. 1998). However, instead of eluting at acidic pH, a second washing of the IgG-Sepharose beads was performed using 5 ml TEV cleavage buffer (0.01% Triton X-100, 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA). The Sepharose beads were then incubated in 200–500 μl of cleavage buffer containing 0.7 mM DTT and 10 U of recombinant TEV protease (Life Technologies) per gram of lysed cells for 70 min on a rotating wheel at 16°C. The supernatant containing the TEV protease–released nucleoporin complex was collected by centrifugation, and the beads were washed once with cleavage buffer before a final elution with 1 ml of 0.5 M acetic acid, pH 3.4. For gel filtration chromatography, the TEV-derived supernatant was concentrated by ultrafiltration (NanoSpin Plus; Gelman) and applied onto a Superdex 200 gel filtration column connected to an FPLC system (Amersham Pharmacia Biotech). Samples were chromatographed in 150 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM EDTA. The flow rate was 0.3 ml/min and 0.6-ml fractions were collected. 25 μl of each fraction was analyzed by SDS-PAGE followed by silverstaining or Western blotting using antibodies specific for the different components of the Nup84p complex.

EM

Glycerol spraying/low-angle rotary metal shadowing of the highly purified Nup84p complex was performed as described (Häner et al. 1997). For negative staining, a 10-μl drop of sample was placed on a carbon-collodion–coated grid that had been freshly glow-discharged in air. After washing with 100 μl of buffer, the sample was stained by washing with 100 μl of 1% (wt/vol) aqueous uranyl acetate and blotted dry. Micrographs were taken at 80 kV in a Philips 400T electron microscope at a magnification of 36,000. Thin section EM of yeast sec13 ts mutants was done as described in Segref et al. 1997.

Molecular Mass Determination of the Isolated Nup84p Complex

For mass measurement by scanning transmission EM (STEM), 5-μl aliquots of the sample were adsorbed for 1 min to glow discharged thin carbon films that spanned a thick fenestrated carbon layer covering 200-mesh/inch gold-plated copper grids. The grids were then blotted and washed (with subsequent blotting) on five drops of 100 mM ammonium acetate, which had been prepared using quartz bidistilled water. They were then immediately frozen in liquid nitrogen and freeze-dried at −80°C overnight in the STEM. A Vacuum Generators STEM HB-5 interfaced to a modular computer system (D-8035 Gauting; Tietz Video and Image Processing Systems GmbH) was employed. Details of the instrument's calibration for mass measurement may be found in Engel 1978 and Müller et al. 1992. Annular dark-field images were recorded digitally from the unstained freeze-dried samples at 80 kV accelerating voltage. The nominal magnification was 200,000 and recording doses were in the range of 300 electrons/nm². The 512 × 512 pixel images were evaluated using the program package IMPSYS as outlined in Engel and Reichelt 1988 and Müller et al. 1992. The mass values were corrected for beam-induced mass loss (Müller et al. 1992) according to the behavior of other proteins in their mass range (our unpublished data). The resulting absolute values were then displayed in histograms and fitted by Gauss curves.

Sedimentation velocity and low-speed sedimentation equilibrium runs of the purified Nup84p complex were performed on an Optima XLA analytical ultracentrifuge (Beckman Instruments) equipped with an absorption optical system (Terzi et al. 1994). The sample was analyzed at 8°C (sedimentation velocity) or 4°C (sedimentation equilibrium) in 50 mM Tris-HCl, pH 8.0, containing 300 mM NaCl, 0.5 mM DTT, and 1.5 mM DTT, with the protein concentration adjusted to OD₂₈₀ ranging between 1.0–0.1. Sedimentation velocity runs were performed at 52,000 rpm in a 12-mm epon double-sector cell, which was filled with 0.12 ml sample in one sector and the same volume of buffer in the other. Sedimentation coefficients were corrected to water by a standard procedure (Eason 1986). Sedimentation equilibrium runs were performed at 6,400, 7,200, and 8,000 rpm and at protein concentrations adjusted to OD₂₈₀ of 1.0, 0.2, and 0.15. An average molecular mass was determined using linear regression analysis so as to obtain the best linear fit of logarithmus A versus r² plots. A partial specific volume of 0.73 ml/g was used for the calculations.

Miscellaneous

SDS-PAGE and Western blot analysis were performed according to Siniossoglou et al. 1996. In situ hybridization of poly(A)⁺ RNA, indirect immunofluorescence using anti-Nsp1p antibodies, and fluorescence microscopy of GFP-labeled yeast cells were performed according to Santos-Rosa et al. 1998.

Results

A Pool of GFP-tagged Sec13p Localizes In Vivo to the Nuclear Pores

Sec13p is a component of the COPII coat complex in yeast and is involved in vesicular transport (Salama et al. 1993; Barlowe et al. 1994; Kaiser and Ferro-Novick 1998). In addition, Sec13p was found to coenrich with the Nup84p nucleoporin complex (Siniossoglou et al. 1996). To find out whether Sec13p can be detected at the nuclear pores in living cells, a GFP-tagged SEC13 under the control of its authentic promoter was expressed as a functional fusion protein in the otherwise lethal sec13::HIS3 null mutant. Sec13p-GFP exhibits both a punctate nuclear/ER membrane and cytoplasmic staining (Fig. 1 a). The Sec13-GFP signal in the cytoplasm tends to be concentrated in spots underlying the plasma membrane, which is suggestive of the peripheral ER. In contrast, GFP-tagged Seh1p predominantly stained the nuclear envelope with little staining in the cytoplasm, thus having characteristics of a classical NPC protein (Fig. 1 a). Since the nuclear envelope staining of Sec13p-GFP was punctate, it is possible that this represents nuclear pore labeling. We therefore used the nup133::HIS3 knockout strain in which the NPCs lose their even distribution and cluster in one or few areas within the nuclear envelope. Indeed, a pool of Sec13-GFP, like Nup49-GFP, clustered at the nuclear envelope in the nup133 ⁻ mutant (Fig. 1 a). Indirect immunofluorescence using antibodies against the nucleoporin Nsp1p confirmed that the clustered NPCs colocalize with the Sec13p-GFP signal (Fig. 1 b). On the other hand, another nuclear membrane/ER protein, Spo7p-GFP, did not cluster with NPCs in the nup133 ⁻ mutant (Fig. 1 b; see also Siniossoglou et al. 1998). This shows that a pool of Sec13p is specifically associated with NPCs under steady state conditions, which is in agreement with our biochemical data for the Nup84p complex (see below).

Figure 1.

A pool of Sec13p-GFP localizes to the nuclear pores. (a) The sec13 ⁻ /nup133 ⁻ double mutant complemented by either SEC13-GFP and NUP133 (sec13::HIS3) or by SEC13-GFP alone (nup133::HIS3) was analyzed by fluorescence microscopy (upper panel). In addition, Spo7p-GFP (Siniossoglou et al. 1998) and GFP-Nup49p were tested in the nup133 ⁻ cells (upper panel). For comparison, the seh1::HIS3 disruption mutant complemented by plasmid-borne SEH1-GFP was analyzed in the fluorescence microscope (lower panel). (b) The sec13 ⁻ /nup133 ⁻ double mutant expressing Sec13p-GFP was analyzed by indirect immunofluorescence microscopy using a monoclonal anti-Nsp1p antibody. It was also stained for DNA.

Novel Thermosensitive sec13 Mutants that Affect Distribution of the NPC Reporter GFP-Nup49p and Nuclear Morphology

Whereas the Nup members of the Nup84p complex (Nup84p, Nup85p, Nup120p, and Nup145p-C) have a function in both NPC distribution and nuclear mRNA export (Siniossoglou et al. 1996), the role of Sec13p and Seh1p within this complex remains unknown. Therefore, we screened for novel sec13 mutants that are impaired in the organization and the distribution of NPCs within the nuclear membrane, similar to the defects seen with some of the other Nup84p complex mutants. The SEC13 gene was mutagenized in vitro by PCR-based random mutagenesis (see Materials and Methods) and inserted into an ARS/CEN plasmid. Several thermosensitive sec13 alleles were isolated from this mutagenized bank. In total, 17 ts alleles of SEC13 (which do not allow growth at 37°C) were characterized further (for growth properties of some of these alleles, see Fig. 2 a). These sec13 ts mutants were transformed with a plasmid encoding GFP-Nup49p, which serves as a nuclear pore reporter protein, to visualize impairment of NPC assembly (Bucci and Wente 1998) or NPC distribution within the nuclear membrane (Belgareh and Doye 1997). Strikingly, most of these novel sec13 mutants show an abnormal intracellular distribution of GFP-Nup49p (as well as of Nic96p-GFP; data not shown) when shifted to the restrictive temperature. This was particularly evident for those mutants that grow slowly already at intermediate temperatures (e.g., 32°C) (Fig. 2, a and b, e.g., sec13-3). Other mutant alleles, which are less impaired in growth at higher temperatures (e.g., sec13-14 and sec13-34), exhibit less strong defects in GFP-Nup49p distribution (Fig. 2, a and b, e.g., sec13-34). When GFP-Nup49p distribution was abnormal in sec13 mutant cells, this nucleoporin reporter was often found in clusters at the nuclear membrane or aggregated in the cytoplasm. Interestingly, the nuclear morphology, as revealed by the intranuclear reporter Pus1p-GFP (Siniossoglou et al. 1998), appears normal in these sec13 mutants (Fig. 2 b). Another COPII mutant, sec23-1, also displayed an irregular distribution of GFP-Nup49p (Fig. 2, a and b; sec23-1). However, we noticed that in the case of sec23-1 mutant, some cells exhibit slightly elongated nuclei, although not comparable to the highly altered nuclear envelope morphology that has been recently reported for spo7 ⁻ and nem1 ⁻ cells (Siniossoglou et al. 1998). In conclusion, the distribution of the nuclear pore reporter GFP-Nup49p (i.e., a ring-like nuclear envelope staining) can be altered in sec13 thermosensitive mutants.

Figure 2.

Novel sec13 thermosensitive alleles that show defects in GFP-Nup49p location and nuclear membrane morphology. (a) Growth properties of novel sec13 thermosensitive mutants. Precultures derived from the sec13::HIS3 disrupted strain transformed with ARS/CEN plasmids containing the indicated SEC13 wild-type and sec13 ts alleles were diluted in growth medium and equivalent amounts of cells (diluted in 10⁻ steps) were spotted onto yeast extract–peptone–d-glucose plates. All SEC13 and sec13 alleles are tagged with one IgG-binding domain from ProtA in their COOH termini. For control, the sec23-1 ts mutant was also included in this analysis. It was grown for 3 d at the indicated temperatures. (b) Location of GFP-Nup49p and GFP-Pus1p in SEC13 ⁺ and sec13 and sec23-1 ts mutants as determined by fluorescence microscopy. Cells were grown at 23°C and shifted for 5 h to either 32 or 35°C. (c) Thin-section EM of SEC13, sec13-3, sec13-14, and sec13-34 cells, shifted for 5 h to 35°C. Filled arrowheads point to normal nuclear pores, arrows to nuclear envelope herniations, and the open arrow to a long nuclear membrane extension with a herniation morphology at the tip. Bars, 0.5 μm.

To analyze the nuclear envelope morphology of the sec13 mutants at the ultrastructural level, we performed thin section EM (Fig. 2 c). Structural abnormalities can be seen in these sec13 ts mutants, which is consistent with the aberrant distribution of the GFP-Nup49p reporter (see Fig. 2 b). Remarkably, the sec13-14 and sec13-34 cells develop extended clusters of herniated NPCs, similar to the herniations seen in nup116Δ and gle2Δ mutants (Wente and Blobel 1993; Bailer et al. 1998) (Fig. 2c), whereas spindle and nucleolar morphology was not altered (data not shown). However, the cells still contain normal nuclear pores in nonherniated areas of the nuclear envelope. In the sec13-3 mutant, a particular strong ER membrane accumulation is also seen. It is therefore possible that (a) the nuclear envelope clustering of the GFP-Nup49p reporter in sec13 ts mutants (e.g., sec13-14 and sec13-34; see Fig. 2 b) corresponds to the clustered nuclear envelope herniations at the ultrastructural level, and (b) the cytoplasmic aggregates of GFP-Nup49p seen in some of the mutant sec13 cells correspond to pore herniations at the tips of the nuclear envelope–derived ER outgrowths. Taken together the GFP-Nup49p localization and the EM data show that the nuclear envelope and NPC organization is impaired in the sec13-3, sec13-14, and sec13-34 mutants.

The sec13-14 and sec13-34 Alleles Exhibit a Conditional Synthetic Lethal Interaction with the nup85Δ Disruption Mutation

Since some of the sec13 ts alleles described above affect the distribution of the Nup49p nucleoporin reporter, we tested for a functional overlap with members of the Nup84p complex. A possible genetic interaction was analyzed between the novel sec13 ts alleles and the partial nup85Δ disruption, which expresses a COOH-terminal portion of the protein (Siniossoglou et al. 1996), the nup84::HIS3, or the seh1::HIS3 disruption (for strain constructions see Materials and Methods). It has been reported previously that the sec13-1 ts allele, which is impaired in ER-to-Golgi transport (Pryer et al. 1993), is not synthetically lethal with the nup85Δ and seh1::HIS3 alleles (Siniossoglou et al. 1996), for example. However, we noticed that two mutant alleles, sec13-14 (data not shown) and sec13-34 (Fig. 3 a), are synthetically lethal with nup85Δ at 32°C, a temperature at which the individual mutants sec13-34, sec13-14, and nup85Δ can grow very well (Fig. 3 a; see also Fig. 2 a). The observed synergistic growth inhibition between sec13-34 with nup85Δ is specific, since the nup84::HIS3/sec13-34 and the seh1::HIS3/sec13-34 double mutants are viable at 32°C (data not shown). In contrast, we do not find a synthetically lethal interaction between the seh1 null mutant and the various sec13 mutants, although Seh1p shares significant sequence similarity with Sec13p (see also Discussion). These data show that Sec13p not only physically associates with the Nup84p complex, but that it is also functionally linked to a particular member of this subcomplex, which is Nup85p.

Figure 3.

Genetic interaction between the sec13-34 and nup85Δ mutant alleles. (a) Growth curves of the sec13-34/nup85Δ double mutant and of the corresponding sec13-34 and nup85Δ single mutants at 23 and 32°C. The nup85Δ allele is derived from a partial gene disruption and expresses a Nup85p protein which is NH₂-terminally truncated (Siniossoglou et al. 1996). (b) Affinity-purification of Sec13p-ProtA from NUP85/SEC13, nup85Δ/SEC13, and nup85Δ/sec13-14 strains grown at 23°C or shifted for 12 h to 32°C. The purified wild-type or mutant Sec13p-ProtA fusion proteins were analyzed by Western blotting using anti-Nup84p antibodies. Indicated are the positions of Sec13p-ProtA and Nup84p.

Nup145p-C, Nup120p, and Nup85p-C, but not Nup84p and Seh1p, Are Essential for Assembly of the Nup84p Complex In Vivo

To identify the essential components for Nup84p complex assembly in vivo, the Nup84p complex was purified from different mutant strains. Either Seh1p-ProtA or Nup84p-ProtA served as baits for the purification, since these two fusion proteins copurify particularly well with all components of the complex, even from cells with wild-type SEH1 or NUP84. When Seh1p-ProtA was purified from the nup84 deletion mutant, all other components of the complex, except Nup84p, were present in stoichiometric amounts (Table ). When the complex was isolated from the seh1 deletion strain by purification of Nup84p-ProtA, all components of the Nup84p complex including Sec13p, but lacking Seh1p, were present (Table ). However, when Nup84p-ProtA was isolated from a nup120 deletion strain, no other complex member coenriched (Table ). Seh1p-ProtA isolated from either the nup120 disruption strain or the nup145 deletion strain expressing only the nonessential NH₂-domain (Teixeira et al. 1997) contained substoichiometric amounts of Nup85p, but no other members of the complex. When the mutant Sec13p encoded by sec13-14 and tagged with ProtA was affinity-purified from the partial disruption nup85Δ mutant (Siniossoglou et al. 1996), it largely lacked Nup84p and the other complex members (Fig. 3 b; data not shown). This all shows that Nup145p-C, Nup120p, and Nup85p-C, but not Nup84p and Seh1p are essential for assembly of the Nup84p complex.

Table 2.

Nup84p Complex Assembly in Nucleoporin Mutants

Strain background	Nup84p-ProtA associated proteins	Seh1p-ProtA associated proteins
nup84::HIS3	Nup85p, Nup120p, Nup145Cp, Seh1p, Sec13p	Nup85p, Nup120p, Nup145Cp, Sec13p
nup120::HIS3	−	Nup85p
nup145ΔC	ND	Nup85p
seh1::HIS3	Nup85p, Nup120p, Nup145Cp, Sec13p	Nup84p, Nup85p, Nup120p, Nup145Cp, Sec13p
sec13-34	ND	Nup84p, Nup85p, Nup120p, Nup145Cp

Assembly of the Nup84p complex in nucleoporin mutants. Nup84p-ProtA or Seh1p-ProtA were affinity-purified from nup84::HIS3, nup120::HIS3, seh1::HIS3, nup145ΔC, and sec13-34 strains. The components, which copurified, are listed in the table. ND, not determined.

Since Seh1p interacts directly with Nup85p (see above; Siniossoglou et al. 1996), we wanted to analyze which part of Nup85p is responsible for this interaction. Therefore, Nup85p truncation constructs lacking either the NH₂-terminal (Nup85pΔN) or COOH-terminal domain (Nup85pΔC) were made and tagged with ProtA to facilitate biochemical purification. To test their functionality, both truncated proteins were transformed into a nup85::HIS3 deletion mutant lacking the entire NUP85 ORF. The Nup85pΔN mutant grows well at 30°C, but is thermosensitive at 37°C. In contrast, the Nup85pΔC strain grows extremely slowly at 30°C, and cannot grow above 32°C (Fig. 4 a). Western blot analysis of extracts derived from the strains expressing the Nup85p truncation mutants showed that the two proteins are expressed at similar levels at the restrictive temperature (data not shown). Affinity-purification of ProtA-Nup85ΔNp revealed that all components of the Nup84p complex except Seh1p copurified (Fig. 4 b, Coomassie and Western blot). Conversely, affinity-purified ProtA-Nup85pΔC contained Seh1p, although in reduced amounts, and completely lacked the other four members of the complex (Fig. 4 b). Shorter NH₂-terminal fragments of Nup85p could not be stably expressed in yeast, therefore it was not possible to define the minimal Seh1p-binding sequence within Nup85p. When the distribution of the GFP-Nup49p NPC reporter was analyzed in these mutants, NPC clustering was seen in both Nup85pΔC and Nup85pΔN strains, although clustering was less striking in the case of Nup85pΔN (Fig. 4 c). Furthermore, nuclear mRNA export is severely impaired at the restrictive temperature in both mutants (Fig. 4 d).

Figure 4.

Distinct domains within Nup85p are required for interaction with the various members of the Nup84p complex. (a) Growth of nup85ΔC and nup85ΔN strains at 23, 30, and 35°C. A nup85::HIS3 null deletion mutant was transformed with the two nup85 truncation constructs and the corresponding strains were spotted onto yeast extract–peptone–d-glucose plates at the indicated temperatures. (b) Affinity-purification ProtA–tagged Nup85p full-length (FL) and of the Nup85pΔN (Δ1–182) and Nup85pΔC (Δ453–745) constructs. Shown is a Coomassie-stained gel of the purified complexes (the fusion proteins are indicated by asterisks, and the components of the Nup84p complex by lines), and a Western blot (right panel, only the relevant area from each blot is shown) using anti–ProtA, anti-Nup84p, anti-Nup145Cp, anti-Seh1p, and anti-Sec13p antibodies. (c) Deletion of the Nup85p COOH-terminal domain causes NPC clustering. The corresponding strains were transformed with a plasmid expressing GFP-Nup49p, shifted for 3 h to 37°C and the distribution of this NPC reporter was analyzed in the fluorescence microscope. (d) Both the COOH and NH₂ domains of Nup85p are required for efficient nuclear mRNA export. The nup85ΔC and nup85ΔN mutants as well as a wild-type control (NUP85 ⁺) were grown for 3 h at 37°C, fixed, and hybridized with a oligo(dT)-FITC probe to reveal poly(A)⁺ RNA distribution. Cells were also stained with Hoechst 33258 (DNA).

These data show that Nup85p contains two distinct domains, a COOH-terminal part, which is required for binding to Nup120p, Nup145p-C, Nup84p, and Sec13p, and an NH₂-terminal part, which binds to Seh1p. Both domains of Nup85p are required for nuclear mRNA export.

The Highly Purified Nup84p Complex Elutes as a Large Assembly during Gel Filtration

In the past and as described above, the affinity-purified Nup84p complex was eluted from the IgG-Sepharose beads through a denaturing low pH treatment (Siniossoglou et al. 1996). To release the purified Nup84p complex from the IgG beads under nondenaturing conditions, the TEV proteolytic cleavage site (a seven amino acid long cleavage site specific for the tobacco etch virus protease) was engineered between the ProtA tag and Nup85p (Fig. 5 a, left panel). This ProtA-TEV-Nup85p fusion construct was functional, since it can complement the nup85::HIS3 null mutant (data not shown). When ProtA-TEV-Nup85p was affinity-purified as described previously, and the IgG-Sepharose beads were incubated with recombinant TEV protease, cleaved Nup85p was released from the column together with the other members of the complex (Fig. 5, a and b, E1 and E2). However, the cleaved off ProtA moiety remained on the IgG-Sepharose beads. A smaller amount of ProtA-Nup85p that was not cleaved by the protease was eluted from the beads by a final pH 3.5 treatment (Fig. 5 a, HAc).

Figure 5.

Purification of a native Nup84p complex. (a) TEV protease induced release of affinity-purified ProtA-TEV-Nup85p from IgG-Sepharose beads. Schematic representation of the ProtA-TEV-Nup85p fusion protein consisting of two IgG binding domains (ProtA), a 10 residue long spacer sequence, the 7 residue long TEV cleavage sequence (underlined) with the cleavage site (indicated by an arrow), followed by the Nup85p sequence. After TEV elution from the beads, a first and second eluate (E1 and E2) were obtained. Finally, a low pH eluate (HAc) was collected. All eluates were analyzed by SDS-PAGE and Western blotting using anti-Nup85p and anti-ProtA antibodies. The positions of the ProtA-TEV-Nup85p and the cleaved Nup85p, lacking the ProtA tag are indicated. (b) Gel filtration chromatography of the affinity-purified and TEV-released Nup84p complex. The eluates E1 and E2, which were combined and concentrated to yield the load fraction (L) were analyzed by FPLC Superose 6 column chromatography. 0.6-ml fractions were collected and 25 μl of fractions 1–27, E1 and E2, and the load (L, 5% of the injected material) were analyzed by SDS-PAGE and silverstaining. The six subunits containing Nup84p complex peak in fraction 9, the Nup85p-Seh1p heterodimer in fraction 15. Fractions 9 and 15 were also analyzed by Western blotting using anti-Nup85p, anti-Nup84p, anti-Seh1p, and anti-Sec13p antibodies, respectively. The positions of the components of the Nup84p complex are indicated. Heavy and light IgG chains are evident in fractions 15 and 16, and the TEV protease in fractions 19 and 20. The Superose 6 column was calibrated with several molecular weight marker proteins of 670, 440, 252, and 158 kD (Bio-Rad).

To further purify the released Nup84p complex, which still contained contaminants such as heat shock proteins, IgGs, and TEV protease (see Fig. 5 b, E1 and E2), and to determine its size and structure, the TEV eluate was concentrated by ultrafiltration and applied to an FPLC Superose 6 gel filtration column (see Materials and Methods). SDS-PAGE of the collected fractions revealed that all six components of the Nup84p complex, including Sec13p, exactly cofractionate in only a few fractions, yielding an essentially pure Nup84p complex (Fig. 5 b, fractions 8–10). The column was calibrated with a protein standard of known molecular weight. Accordingly, the Nup84p complex elutes in a peak corresponding to ∼700 kD (Fig. 5 b; see below also).

Interestingly, Nup85p and Seh1p form a second peak (Fig. 5 b, fraction 15) that does not contain the other members of the complex (Fig. 5 b; note that these fractions contain IgG heavy and light chains). Accordingly, besides being organized in a large complex, Nup85p and Seh1p can form a heterodimeric complex, which is in agreement with our data described above. Thus, affinity-purification of the Nup84p complex (via tagged Nup85p), its elution from the affinity matrix by TEV cleavage, and final gel filtration chromatography yielded a highly pure preparation of the Nup84p complex.

The Highly Purified Nup84p Complex Exhibits a Y-Shaped Structure and a Molecular Mass of 375 kD

The purified Nup84p complex, which elutes as a large complex as determined by gel filtration chromatography (see also Fig. 5 b), may exhibit a distinct structure that could correspond to one of the above mentioned structural modules of the NPC. To analyze the structure of purified Nup84p complex by EM, both negative staining and glycerol spraying/low-angle rotary metal shadowing were performed. As a sample we used the Nup84p complex preparation derived from the gel filtration column (corresponding to Fig. 5 b, fractions 8–10), which is essentially pure as judged from the silverstained SDS-polyacrylamide gel (Fig. 5 b). This sample of the Nup84p complex yielded a prevalent morphology in the electron microscope. Independent of the applied specimen preparation procedure (i.e., negative staining or low-angle rotary metal shadowing), the Nup84p complex exhibits a predominantly Y-shaped, triskelion-like structure with an overall diameter of ∼25 nm (Fig. 6, a and b). In these projections, the Nup84p complex exhibits an approximate mirror symmetry, suggesting that the complex harbors a corresponding twofold axis of symmetry. Although the complexes appear to adsorb with a preferred orientation on the grid, there is a variation in molecular morphology, which indicates a degree of flexibility in the links between the arms and the stalk. Thus, the Nup84p complex may be a rather flexible structure. However, since the preparation is to a certain extent heterogenous in the electron microscope (only ∼60% of the negatively stained structures >10–15 nm show a Y-shaped form; see Fig. 6 a), it is possible that this is due to partial subunit disassembly during preparation.

Figure 6.

The highly purified Nup84p complex displays a Y-shaped, triskelion-like structure with a molecular mass of ∼375 kD. Electron microscopic appearance of the highly purified Nup84p complex after negative staining (a) and glycerol/rotary metal shadowing (b). In a, a large overview picture is shown, on which Y-shaped structures are highlighted by circles. Selected Y-shaped molecules from this picture were further processed by cut and paste and magnification using Adobe Photoshop, and are shown at the edge of the overview picture. In b, overview pictures (I) and a gallery of selected Nup84p particles (II) are shown. For comparison, shadowed IgG particles are also depicted in b. (c) Mass determination of the highly purified Nup84p complex by STEM. Histogram displaying the mass data determined for 619 particles. Correction has been made for beam-induced mass loss. The two Gaussian curves fall at 368 ± 122 kD (n = 438) and 698 ± 123 kD (n = 158), respectively. The number of particles giving rise to the peaks (n) was estimated by measuring up to and away from the point of peak overlap. The inset reveals a field from an annular dark-field image recorded of an unstained freeze-dried Nup84p sample.

According to the gel filtration data, the molecular mass of the Nup84p complex was estimated to be ∼700 kD (see Fig. 5 b). This value is almost twice that of the calculated molecular mass of 442 kD, assuming that all six subunits are present in one copy per complex. However, due to its nonglobular Y-shaped structure, the complex may exhibit an abnormal gel filtration behavior. Therefore, we determined the molecular mass of the purified Nup84p complex by quantitative STEM and analytical ultracentrifugation. Whereas for STEM predominantly particles with a similar average diameter (i.e., 25 nm) as in negative stain were depicted, in unstained freeze-dried samples, the typical Y-shaped, triskelion-like morphology readily observed with negatively stained Nup84 complex was more difficult to delineate in the STEM (Fig. 6 c, inset). Nevertheless, all except the very smallest particles were selected for measurement, and their absolute mass values were calculated, pooled, and displayed in a histogram (Fig. 6 c). This histogram was fitted by two Gauss curves peaking at 368 ± 122 kD (n = 438 particles) and 698 ± 123 kD (n = 158 particles), respectively (i.e., after correction to zero mass loss). To a first approximation, these two peak values indicate the sample to consist of a mixture of a monomeric (∼75%) and a dimeric (∼20%) Nup84p complex, with a small amount (∼5%) of larger (>1 MD) aggregates. However, the standard deviations are too large to simply stem from fluctuations in the background scattering, thus indicating some heterogeneity of the two Nup84p populations. It is conceivable that the bona fide Nup84p complex exhibits a rather limited stability with its more peripheral subunits, having a tendency to dissociate, so that each of the two Gauss curves represents several distinct Nup84p particles, i.e., containing different numbers of subunits.

The highly purified Nup84p complex was also subjected to analytical ultracentrifugation. By sedimentation equilibrium analysis, a predominant species with an average mass of 375 ± 13 kD was depicted at speeds ranging between 6,400 and 8,000 rpm and protein concentrations adjusted to OD₂₈₀ of 1.0, 0.2, and 0.15. The corresponding major S_20,w value was 5.6 by sedimentation velocity at 52,000 rpm, an S value being consistent with a flat Y-shaped 375 kD particle as deduced from EM of negatively stained samples (Fig. 6, a and b), mass determination by STEM (Fig. 6 c), and sedimentation equilibrium centrifugation analysis. According to these mass determination data, the Nup84p complex exists predominantly in a monomeric form when isolated from yeast.

Discussion

Now that most of the yeast nucleoporins have been cloned and characterized, the next goal is to put them into a higher order biochemical and structural context. Here, we report on the molecular and structural characterization of the yeast Nup84p complex. This NPC subassembly consists of six subunits, five of which are bona fide nucleoporins (Siniossoglou et al. 1996). When the essential nucleoporins of this complex are mutated, inhibition of nuclear mRNA export and defects in the distribution of NPCs and in the organization of the nuclear membrane occur. In the case of Nup85p, a physical and genetic link to the mRNA export machinery could be shown (Segref et al. 1997; Santos-Rosa et al. 1998), whereas Nup84p genetically interacts with two novel nuclear/ER membrane proteins, Spo7p and Nem1p, required for formation of a spherical nucleus (Siniossoglou et al. 1998). However, the role of the sixth member of the Nup84p complex, Sec13p, a bona fide subunit of the COPII coat complex, remained unclear. We now demonstrate that a pool of Sec13p functions as a nucleoporin. Several lines of evidence support a role for Sec13p at the nuclear pore. First, we found that a pool of Sec13p-GFP colocalizes with NPCs in a clustering nucleoporin mutant. Apparently, this reflects the same pool of Sec13p, which physically interacts with the Nup84p complex. Second, the distribution of the GFP-Nup49p, which serves as a nucleoporin reporter, is altered in several of the newly isolated ts sec13 mutants. Strikingly, thin sectioning EM of the sec13-3, sec13-14, and sec13-34 cells shifted to the restrictive temperature reveals abnormalities within the nuclear membrane and an accumulation of intracellular membranes, which could be nuclear envelope–attached ER. In addition, NPC herniations arise within the nuclear envelope of sec13 ts mutants, typically found in other nucleoporin mutants such as nup116Δ and gle2/nup40Δ. This all shows that Sec13p is required for normal nuclear envelope and NPC biogenesis. Interestingly, another COPII mutant, sec23-1, also exhibits defects in the distribution of the GFP-Nup49p nucleoporin reporter. Mutants blocking COPII function accumulate ER membranes at the nonpermissive temperature. Thus, it is possible that the abnormal NPC distribution in the sec23 and/or sec13 mutant is due to the accumulation of nuclear pores into these ER membranes, before their incorporation into the nuclear membrane. Alternatively, there might be a more direct link between COPII vesicle budding from the ER and the biogenesis of nuclear pores. Third, two ts alleles, sec13-14 and sec13-34, are synthetically lethal with the nup85Δ mutant at 32°C, showing that Nup85p and Sec13p perform overlapping or redundant functions at the NPC. Surprisingly, combined seh1 and sec13 mutant alleles do not exhibit such a genetic interaction.

Our biochemical analysis revealed that Nup120p, Nup145p-C, and the COOH-terminal domain of Nup85p are essential for the assembly of a core complex in vivo and thereby may serve as an assembly core or structural scaffold of the complex. Evidently, Nup84p, Seh1p, and Sec13p reside more peripherally within the Nup84p complex. Strikingly, the three members of the core complex exhibit a strong link to the mRNA export machinery (Fabre et al. 1994; Aitchison et al. 1995; Goldstein et al. 1996; Siniossoglou et al. 1996; Teixeira et al. 1997). Hence, it is conceivable that the Nup84p complex consists of a structural scaffold, which at the same time is involved in nuclear mRNA export (e.g., by interacting with mRNA export factors such as Mex67p and Mtr2p; see also Santos-Rosa et al. 1998). In contrast, the more peripheral, dispensable members of the complex (Nup84p, Seh1p, Sec13p) control the distribution of NPCs within the nuclear membrane as well as nuclear envelope organization. In the case of the nonessential Seh1p, we could show that it directly binds to the NH₂-terminal domain of Nup85p as one subunit of the core complex; this interaction is stable up to high salt concentrations (Siniossoglou et al. 1996), and a separate Nup85p–Seh1p heterodimeric complex is even detected in vivo (this study). The role of this heterodimeric Nup85p–Seh1p complex is presently unknown, but it could play a direct role in nucleocytoplasmic transport or represent a stable intermediate during NPC assembly.

The highly purified Nup84p complex exhibits a defined structure as revealed by EM of negatively stained as well as glycerol-sprayed/low-angle rotary metal-shadowed specimens. Although not appearing 100% homogeneous in the EM, >50% of the highly purified Nup84p complex exhibits a distinct Y-shaped or triskelion-like morphology with an overall particle diameter of ∼25 nm. We have estimated that there are between 8 and 16 copies of the Nup84p complex per NPC (Lutzman, M., and E. Hurt, unpublished data). Hence, the Nup84p complex makes a significant contribution to the overall mass of the NPC. Because of its apparent twofold symmetry, one possibility is that in situ, the Nup84p complex may assemble into an octagonal ring-like array, for instance, via interaction of the Y-shaped arms of adjacent complexes. In such a model, the third arm of each Nup84p complex unit remains free, and thus could interact with other components of the NPC, for example with a cytoplasmic fibril, and/or with transport factors. Alternatively, the Nup84p complex may oligomerize so that all three arms of a Y-shaped molecule are engaged in binding to other Y-shaped Nup84p complex molecules, thereby forming a coat- or cage-like structure. Finally, we cannot exclude that individual Y-shaped Nup84p particles exist within the structural framework of the NPC. Further structural and biochemical analysis of the Nup84p complex and its organization into higher-order complexes, including identification of its interacting components at the NPC, should help to answer these open questions.

Given the finding that the central NPC framework is conserved from yeast to vertebrates (Yang et al. 1998), it would be predicted that a homologue to the Nup84p complex might also exist in higher eukaryotes. Indeed, a recent study reported the identification of a mammalian NPC subcomplex containing the homologues of cleaved yeast Nup145p-C and Nup84p. Intriguingly, this complex also contains mammalian Sec13p and a Sec13p-related protein showing that the function of Sec13p at the nuclear pore has been evolutionarily conserved (Fontoura et al. 1999). It is therefore conceivable that the overall structure of the NPC is built from distinct subcomplexes or structural modules, which have been conserved during eukaryotic evolution.

In summary, the Nup84p complex appears to be a paradigm for the structural and functional analysis of an NPC subcomplex or a structural module. Amenable to yeast genetics and cell biological methods in vivo, as well as to biochemical purification and EM in vitro, it may be possible to reconstruct its 3-D structure, and to fit, i.e., to position and orient it into a 3-D mass density map of the entire NPC. Last but not least, it may be possible to use the purified Nup84p complex as a seed to reconstitute step-by-step an NPC in vitro from its purified nucleoporins and/or distinct subcomplexes.