Nup145p is an essential yeast nucleoporin involved in nuclear export of polyadenylated RNAs. We demonstrate here that Nup145p is cleaved in vivo to yield two functionally distinct domains: a carboxy‐terminal domain (C‐Nup145p) which is located at the nuclear pore complex (NPC) and assembles into the Nup84p complex, and a GLFG‐containing amino‐terminal domain (N‐Nup145p) which is not part of this complex. Whereas the essential C‐Nup145p accomplishes the functions required for efficient mRNA export and normal NPC distribution, N‐Nup145p, which is homologous to the GLFG‐containing nucleoporins Nup100p and Nup116p, is not necessary for cell growth. However, the N‐Nup145p becomes essential in a nup188 mutant background. Strikingly, generation of a free N‐domain is a prerequisite for complementation of this peculiar synthetic lethal mutant. These data suggest that N‐ and C‐domains of Nup145p perform independent functions, and that the in vivo cleavage observed is of functional importance.
Nuclear pore complexes (NPCs) are evolutionarily conserved structures that allow passage of soluble material through the nuclear envelope. While the aqueous channels of the NPCs allow passive diffusion of small molecules (<40 kDa), most proteins and RNAs are translocated through the pore by a process involving steps of GTP hydrolysis (for review see Gorlich and Mattaj, 1996). Active nucleocytoplasmic transport through NPCs is bi‐directional, signal‐mediated and involves different soluble shuttling factors (for review see Gerace, 1995; Koepp and Silver, 1996).
Yeast NPCs are of estimated molecular weight 66 MDa and could be composed of about 100 distinct proteins (called Nups or nucleoporins; Rout and Blobel, 1993). The role of NPC proteins in transport reactions is not yet elucidated even if there are numerous yeast nucleoporin mutants that accumulate mRNA inside the nucleus or are impaired in nuclear protein import under restrictive conditions. Nup1p, Nsp1p, Nic96p, Nup49p and Nup57p are yeast NPC proteins implicated in protein uptake into the nucleus (Nehrbass et al., 1990; Grandi et al., 1993, 1995b; Bogerd et al., 1994). For some of them, a physical interaction with the cytoplasmic receptors that recognize nuclear localization sequences (NLSs) has been demonstrated, suggesting that these nucleoporins may serve as docking sites for the import complexes (Belanger et al., 1994; Dingwall et al., 1995; Iovine et al., 1995; Rexach and Blobel, 1995). Nuclear pore proteins may also act as a receptor for factors involved in export reactions, since a cellular protein (Rab/hRip), which contains several FG repeat sequences typically found in nucleoporins, binds to a nuclear export sequence (NES) of the HIV Rev protein which, itself, exports unspliced and partially spliced viral mRNA from the nucleus to the cytoplasm (Stutz and Rosbash, 1994; Bogerd et al., 1995; Fritz et al., 1995).
Efficient mRNA export from the nucleus to the cytoplasm similarly requires a large number of yeast nucleoporins. Indeed, mutations in Nup1p, Nup82p/Nle4p, Nup116p, Rat7/Nup159p, Rat3/Nup133p, Nup49p and some components of the Nup84p sub‐complex (Nup84p, Nup85p/Rat9p, Nup120p/Rat2p) lead to the accumulation of mRNA in the nucleus (Wente and Blobel, 1993; Bogerd et al., 1994; Doye et al., 1994; Aitchison et al., 1995a; Gorsch et al., 1995; Grandi et al., 1995a; Hurwitz and Blobel, 1995; Kraemer et al., 1995; Li et al., 1995; Goldstein et al., 1996; Siniossoglou et al., 1996). In some cases, like in Nup116p deletion mutants, it has been suggested that a defect in export of polyadenylated RNAs might be indirectly caused by structurally sealed NPCs (Wente and Blobel, 1993). However, many other of these mutants do not exhibit structural NPC abnormalities, suggesting that these nucleoporins might directly participate in RNA export reactions. That an NES‐containing protein essentially involved in mRNA export, Gle1p, is found in association with NPCs suggests that at least some export factors, like import factors, have transient interactions with nucleoporins (Murphy and Wente, 1996). It is likely that some of these NPC components would interact with mediators of mRNA export, like the hnRNP proteins, which are tightly bound to mRNA, accompany mRNA during passage from the site of synthesis to the cytoplasm, and shuttle between the nucleus and the cytoplasm (Dreyfuss et al., 1993).
Nup145p is one of the NPC proteins whose depletion from yeast cells causes a nuclear retention of poly(A)+ RNAs, but does not directly affect nuclear protein import (Fabre et al., 1994). Upon biochemical purification, Nup145p is cleaved into two distinct moieties: a carboxy‐terminal domain (C‐Nup145p) which does not share any sequence homology to known proteins, and an amino‐terminal domain (N‐Nup145p) that has homology to two other yeast nucleoporins, Nup100p and Nup116p (Fabre et al., 1994; Wente and Blobel, 1994). This homology includes the presence of GLFG repeat sequences and a nucleoporin RNA‐binding motif (NRM) which binds to homopolymeric RNA in vitro (Fabre et al., 1994). Here we report that cleavage of Nup145p occurs in vivo and describe that the two different domains perform separate functions. The C‐Nup145p assembles into the Nup84p complex and is required for normal NPC distribution and mRNA export. The N‐Nup145p is not essential for cell growth, but becomes crucial if other nucleoporins are mutated.
Nup145p is cleaved in vivo
Although the NUP145 gene potentially codes for a 145 kDa nucleoporin, it is detected as two proteolytically cleaved products of 65 and 80 kDa (Fabre et al., 1994; Wente and Blobel, 1994). To determine whether the event that cleaves Nup145p occurs in vivo and thus may have a functional relevance, maturation of Nup145p was followed in living cells by pulse–chase analysis. For this purpose, ProtA–Nup145p, an amino‐terminally tagged form of the protein that can functionally replace the essential Nup145p (Fabre et al., 1994), was used. Exponentially growing yeast cells were pulse‐labelled for 5 min with [35S]methionine/cysteine before chase with excess of unlabelled amino acids. ProtA–Nup145p was then immunoprecipitated from cell aliquots taken after various time points of chase (see Materials and methods). After pulse‐labelling of yeast cells, a precursor polypeptide of expected molecular weight (∼136 kDa) was detected (Figure 1). During the chase, this precursor band was still visible after 5 min, but started to vanish after 10 min. Concomitant to the disappearance of the precursor, a new band of 57 kDa which corresponds to the tagged N‐Nup145p is detected. Similar results were obtained when wild‐type Nup145p was immunoprecipitated with an antibody specific for the amino‐terminal domain of the protein (data not shown). When a mutant form of Nup145p (ProtA–Nup145NC1p, see below), which is not cleaved, was used in a similar pulse–chase experiment, solely the full‐length ProtA–Nup145NC1p was detected (not shown). These experiments indicate that Nup145p is initially synthesized as a precursor, before it is cleaved in vivo into an amino‐ and carboxy‐terminal domain. These data also explain why under steady‐state conditions Nup145p is mainly found in its cleaved two forms, N‐Nup145p and C‐Nup145p (Fabre et al., 1994; Wente and Blobel, 1994).
Cleavage site of Nup145p is conserved during evolution
To determine the cleavage site within Nup145p and the residues crucial for in vivo processing, intragenic mutants impaired in Nup145p maturation were sought. The generation of a 65 kDa amino‐terminal and 80 kDa carboxy‐terminal fragment and sequence data derived from an amino‐terminal peptide of C‐Nup145p (Wente and Blobel, 1994) suggested that cleavage should occur between F605 and S606 (Figure 2). We therefore performed a random PCR mutagenesis of the DNA surrounding the S606 encoding region and obtained a collection of non‐cleavable (NC) NUP145 mutant alleles (see Materials and methods). DNA sequencing of the mutant alleles revealed that non‐cleavable forms of Nup145p carry mutations in F606→S (NC1) and H604→P (called NC2), respectively. An additional W608→R or L mutation is found in both mutants. The W608 mutation alone (NC3), however, did not inhibit the cleavage reaction (Figure 2A).
Databases were searched for proteins sharing homology with Nup145p. Nuclear pore proteins Nup116p and Nup100p from yeast, and Nup98p from human and rat, were found similar to N‐Nup145p (Fabre et al., 1994; Wente and Blobel, 1994; Radu et al., 1995). In addition, a fragment of an EST sequence of Caenorhabditis elegans was found to match the NRM of Nup145p (Fabre et al., 1994). Release of the entire genomic sequence corresponding to this cDNA in AceDB (ZK328.5B) reveals that homology with Nup145p is extended (Figure 2B). In particular, amino acids that were defined as important for Nup145p maturation (see above) are conserved between hNup98, rNup98, ZK328.5B and Nup145p. This cDNA might encode a putative C.elegans nuclear pore protein since the most amino‐terminal domain contains several GLFG motifs (not shown).
The non‐cleavable mutant proteins Nup145NC1p and Nup145NC2p were functional, since they could fully complement the growth arrest of a Δnup145 null strain (not shown; see Materials and methods). These data confirm our earlier conclusion that cleavage of Nup145p into an N‐ and C‐terminal domain is not required for its essential in vivo function, at least not in a genetic background in which other nucleoporins are intact (see also later).
Post‐translational maturation has a role in intracellular targeting of Nup145p
To determine whether the cleavage of Nup145p has a role in the localization of the two generated domains, subcellular localization of N‐Nup145p when part of the wild‐type precursor was compared with that of N‐Nup145p when it is expressed separately from C‐Nup145p and when it is not releasable from the latter (Figure 2C). Nup145p, N‐Nup145p and Nup145NC1p were thus tagged at their amino‐terminal ends with the green fluorescent protein (GFP, see Materials and methods). GFP–Nup145p, like GFP–Nup145NC1p, extends from M1 to Y1318, GFP–Nup145Np extends from M1 to F605, and GFP–Nup145Cp (or ProtA–Nup145Cp used in this trans‐complementation experiment) extends from S606 to Y1318. Subcellular localization was examined in either the Δnup145 null strain, or in nup133− cells, in which the nuclear pores cluster (Doye et al., 1994). Confocal microscopic analysis of living cells is shown in Figure 2C. As expected, GFP–Nup145p shows a punctuate nuclear envelope staining (see also Fabre et al., 1994). This staining co‐localizes with clustered NPCs in the nup133− strain (not shown). A similar localization is observed with GFP–Nup145NC1p, indicating that cleavage is not required for proper localization of the amino‐terminal end of the protein at the NPC. However, when GFP–Nup145Np is localized in the presence of C‐Nup145p provided in trans, the N‐terminal domain shows both an NPC and a nucleoplasmic staining.
To determine whether N‐ and C‐moieties of Nup145p depend on each other to be localized at the NPC, GFP–Nup145Np or GFP–Nup145Cp were independently expressed. When tested for functionality, GFP–Nup145Cp alone could fully complement the Δnup145 null mutant. In contrast, expression of GFP–Nup145Np could only partly complement the lethal phenotype of the Δnup145 null mutant. This is seen by a reduced growth rate of the mutant at 30°C (t1/2 = 3.5 h as compared with t1/2 = 1.2 h for wild‐type cells) and growth arrest of cells at 37°C. This result shows that C‐Nup145p performs the most crucial function for cell viability, whereas N‐Nup145p is dispensable for cell growth. Confocal analysis showed that GFP–Nup154Np is localized both at the NPCs (clustered because of the lack of C‐Nup145p; see below) and in the nucleoplasm, while GFP–Nup145Cp is located only at the nuclear pores.
Taken together, these data indicate that both N‐ and C‐domains of Nup145p contain signals for NPC targeting, but association of N‐Nup145p with nuclear pores is efficient only when it is derived from in vivo cleavage of the full‐length Nup145p.
The carboxy‐terminal domain of Nup145p assembles into the Nup84p nucleoporin complex
In order to unravel the function of N‐Nup145p and C‐Nup145p, interacting components for both domains were searched. For this purpose, the amino‐terminal ends of each domain were tagged with protein A which allows affinity purification under non‐denaturing conditions by IgG–Sepharose chromatography (Figure 3A). ProtA–Nup145Np extends from S247 to F605, and ProtA–Nup145Cp from S606 to Y1318 and complement Δnup145 null mutant in a manner identical to the GFP fusion proteins.
When interacting proteins were analysed by SDS–polyacrylamide gel electrophoresis followed by Coomassie blue staining (Figure 3B) and Western blotting (Figure 3C), it was found that ProtA–Nup145Cp (estimated molecular weight of 96 kDa, Figure 3B, lane 2; and corresponding Western blot, Figure 3C, lane 2) is in physical association with several prominent bands. The pattern of these major co‐purifying bands strikingly resembles the pattern of a previously described complex which includes Nup120p, Nup85p, band IV, Nup84p, Seh1p and Sec13p (Siniossoglou et al., 1996; Figure 3B, lane 1). This suggested that C‐Nup145p could be part of this nucleoporin complex. Band IV was isolated from an SDS–polyacrylamide gel in an essentially pure form; this was possible when the complex was affinity purified via ProtA–Nup84p from a nup84− strain in which Nup85p was not full‐length, but amino‐terminally truncated (Siniossoglou et al., 1996). Mass spectroscopic analysis unequivocally showed that band IV is the carboxy‐terminal domain of Nup145p. In addition, band IV is recognized by antibodies raised against C‐Nup145p (data not shown). By Western blotting it was further verified that Nup85p, Seh1p (Figure 3C), Nup84p and Sec13p (data not shown) are major co‐purifying bands of the ProtA–Nup145Cp eluate. The 120 kDa band is likely to be Nup120p, because it co‐migrates with the Nup120p band in the Nup84p complex (Figure 3B, compare lanes 1 and 2). Peptide sequence analysis of a 70 kDa band (Figure 3B, lane 2) showed that it corresponds to Hsp70p. However, Hsp70p is often detected in protein purification, arguing that it may be an unspecific contaminant (Craig et al., 1993). These data demonstrate that C‐Nup145p is part of Nup84p sub‐complex.
When full‐length ProtA–Nup145p was affinity purified, a 57 kDa band corresponding to ProtA–Nup145Np bound efficiently to the IgG–Sepharose column, but the bands corresponding to the other members of the Nup84p complex (Nup120p, Nup85p, C‐Nup145p, Nup84p, Seh1p and Sec13p) were also associated (Figure 3B and C, lane 4). However, these bands were clearly sub‐stoichiometric as compared with results obtained with ProtA–Nup145Cp (Figure 3B, lane 2). These data indicate that N‐Nup145p may not be part of the complex after cleavage. To test for this, ProtA–Nup145NC1p was isolated by IgG–Sepharose chromatography. It was observed that Nup120p, Nup85p, Nup84p, Seh1p and Sec13p could be recovered in approximately stoichiometric ratios (Figure 3B, lane 3). Taking these results together, we suggest that Nup145p precursor is targeted to the NPC Nup84p sub‐complex, and after cleavage, the carboxy‐terminal domain remains in interaction with it, while the amino‐terminal part is (largely) released from the complex (see also Discussion).
To determine further the possible interacting components of free N‐Nup145p, affinity purification was performed using a lysate derived from strain FYEF95 (ΔC) which expresses solely ProtA–Nup145Np. Although ProtA–Nup145Np could be efficiently affinity purified, none of the components of the Nup84p complex was associated in significant amounts (Figure 3B, lane 5). This result was also confirmed by Western analysis using antibodies against Nup85p and Seh1p (Figure 3C, lane 5). An identical pattern was obtained when a ProtA–Nup145Np fusion protein containing the GLFG repeats was used (not shown). The stronger band which migrates in the 70 kDa region, also detected in ProtA–Nup145p eluate, was analysed and shown to correspond to Hsp70p. These data indicate that under these conditions N‐Nup145p does not have stable interacting partners.
Yeast cells lacking the carboxy‐terminal domain of Nup145p are impaired in mRNA export and NPC distribution
The data described above indicated that C‐Nup145p assembles into the Nup84p complex. As members of this complex (e.g. Nup84p, Nup85p and Nup120p) participate in mRNA export and normal NPC distribution (Siniossoglou et al., 1996), we suspected that C‐Nup145p performs a similar function. Nuclear export of polyadenylated RNA was examined in the FYEF95 (ΔC) strain (Figure 4A). At 30°C, a temperature at which this strain grows, albeit very slowly (see also above), ∼30% of cells accumulate mRNA inside the nucleus. Following transfer to 37°C, progressively more and more cells show nuclear poly(A)+ accumulation, with 100% of cells positive after 7 h (Figure 4A). In contrast, no apparent defect in nuclear protein import could be detected as judged by the nuclear accumulation of a Matα‐2–lacZ reporter protein (not shown). Interestingly, >80% of the FYEF95 (ΔC) cells were able to recover growth if shifted back from 37°C to 30°C, indicating that the mutant is thermoreversible.
In this mutant, a nucleolar disintegration was also observed. The nucleolus was fragmented into several foci in ∼10% of the cells at 30°C and in 30% of the cells when shifted for 7 h to 37°C. Furthermore, we show that lack of C‐Nup145p induces nuclear pore clustering. This was demonstrated by the use of a nucleoporin tagged with GFP (GFP–Nup49p; Belgareh and Doye, 1997). Whereas in wild‐type cells GFP–Nup49p gives a typical ring‐like staining of the nuclear envelope, the same NPC reporter construct clusters in 100% of FYEF95 (ΔC) cells, even when grown at 30°C (Figure 4A). This is in agreement with the observation in Figure 2C that GFP–Nup145Np, when expressed alone, is partly localized in clustered NPCs. NPC clustering was further analysed by electron microscopy (Figure 4B). In a wild‐type strain, a normal nuclear and NPC morphology was observed. In the strain lacking C‐Nup145p, nuclear envelope morphology appears unaffected, but NPCs heavily cluster at one site of the nuclear membrane and no double nuclear membrane could be seen between the clustered NPCs (Figure 4B). On the other hand, in strain FYEF94 (ΔN) which expresses solely C‐Nup145p, NPCs show no tendency for clustering (Figure 4C). This is consistent with the observation that GFP–Nup145Cp is targeted to apparently normal NPCs (see Figure 2C).
We conclude that, like other members of the Nup84p NPC sub‐complex, the carboxy‐terminal domain of Nup145p, but not its amino‐terminal part, is required for a normal distribution of NPCs within the nuclear membrane, as well as for normal nuclear export of polyadenylated RNAs.
Isolation of a synthetically lethal mutants in which the amino‐terminal domain of Nup145p is required for cell growth
Since N‐Nup145p alone does not stably interact with other NPC components and is not required for cell growth, a colony sectoring assay was designed to identify mutated genes that show synthetic lethality when N‐Nup145p is missing (see Materials and methods). A synthetically lethal mutant (called nsl7) could be obtained from such a screen. nsl7 was complemented by the cleavable ProtA–Nup145p and ProtA–Nup145Np, but not by ProtA–Nup145Cp (Figure 5). This shows that the presence of the amino‐terminal domain of Nup145p in nsl7 is essential for cell growth. We next tested whether cleavage of Nup145p is a prerequisite to complement the synthetic lethal phenotype of this mutant. Strikingly, ProtA–Nup145NC1p, which is no longer cleaved in vivo (see also Figure 2), was strongly impaired in restoring growth of the nsl7 strain on 5‐FOA‐containing plates (Figure 5A). We therefore conclude that in a certain mutant background a separate Nup145p amino‐terminal domain is essential for cell growth. If the release of this domain from the precursor protein is inhibited from full‐length Nup145p, it cannot perform its function.
The mutation which caused synthetic lethality in nsl7 was complemented with a yeast genomic library insert containing NUP188 which encodes one of the major NPC constituents in yeast (Nehrbass et al., 1996; Zabel et al., 1996; see Materials and methods). This suggested that a recessive mutation of NUP188 in nsl7 is responsible for synthetic lethality when N‐Nup145p is missing. This idea was confirmed in two ways. First, transformation of nsl7 cells with a wild‐type NUP188 could complement the nsl7 mutant, while a mutated form of NUP188 (nup188ΔSpeI) could not (Figure 5A). Second, by combining the disruption allele of NUP145 with a knockout allele of NUP188, it was found that the requirement for a free N‐Nup145p was identical to the one observed in nsl7 (Figure 5B). To determine whether the mutation in nsl7 corresponds to a nup188 null or truncation allele, we recovered the genomic mutated form of NUP188 and tagged it. Full‐length Nup188p was seen by Western blot analysis (not shown), arguing that synthetic lethality may be due to a more subtle mutation within Nup188p.
In order to test if the genetic interaction between N‐Nup145p and Nup188p reflects a physical association, IgG affinity purification was performed in a strain which co‐expresses a functional Myc–Nup188p and ProtA–Nup145p (see Materials and methods). Under conditions which allowed complete solubilization of Myc–Nup188p and detection of the Nup84p NPC sub‐complex, we could not find Myc–Nup188p in the eluate fraction of ProtA–Nup145p (data not shown). However, this result does not exclude that Nup188p physically interacts with N‐Nup145p in vivo, but this association is not stable for our biochemical purification.
We also analysed whether phenotypic abnormalities are associated with the mutation of NUP188 in the nsl7 strain. Only morphological defects in NPC organization revealed by electron microscopy could be detected (Figure 4D). These abnormalities include NPCs which are only found on the nucleoplasmic face of the nuclear envelope and nuclear envelope herniations and are similar to those observed in strains mutated in NUP188 (Nehrbass et al., 1996). To determine further the cause of the synthetic lethality between the lack of N‐Nup145p and mutation in NUP188, we analysed whether in the nsl7 mutant, depletion of N‐Nup145p might have an effect on nuclear transport processes. To test for this, NUP145 was expressed under the inducible GAL10 promoter (see Fabre et al., 1994) in the constitutive presence of C‐Nup145p. We found that depletion of Nup145p in such a context had no effect on mRNA accumulation, nor on NLS‐dependent nuclear protein import (data not shown).
Nup188p interacts physically or genetically with several other nucleoporins including Pom152p (Nehrbass et al., 1996), Nic96p (Zabel et al., 1996) and Nup157p (Aitchison et al., 1995b). We, therefore, tested whether N‐Nup145p is required for cell growth when these nucleoporins are mutated. No synthetic lethality was observed in yeast strains in which N‐Nup145p was absent and POM152 or NUP157 were disrupted. In the case of NIC96, which encodes an essential nucleoporin (Grandi et al., 1993), two mutant alleles were tested. While a thermosensitive mutant which maps in the central domain of Nic96p (nic96‐1; Grandi et al., 1995b) requires N‐Nup145p in order to grow, a mutant allele which lacks the non‐essential carboxy‐terminal domain of Nic96p (nic96Δ532–839; Grandi et al., 1995b) does not (Figure 5C). These results demonstrate that N‐Nup145p becomes essential if other components of the nuclear pores, i.e. Nup188p and Nic96p, are mutated (see also Discussion).
One important finding of this paper is that a proteolytic cleavage of nucleoporin Nup145p creates, in vivo, two domains that function separately. Nup145p processing is the first example of a proteolytic maturation of a nuclear pore protein. While this maturation is dispensable for the activity of C‐Nup145p, it is essential for the function of N‐Nup145p.
Our data have shown that the cleaved C‐domain of Nup145p is incorporated into a large nucleoporin complex consisting of Nup120p, Nup85p, Nup84p, Seh1p and Sec13p. This gives a clue as to how C‐Nup145p gets targeted to the nuclear membrane and stably associates with nuclear pores. The Nup84p complex performs an essential role in nuclear pore biogenesis, and deletion of any Nup‐member from the complex (e.g. Nup120p, Nup85p and Nup84p) has drastic consequences for nuclear envelope and NPC organization (Siniossoglou et al., 1996). Lack of C‐Nup145p also leads to an extensive NPC clustering, reminiscent of the phenotype observed in nup84− cells, but different from the grape‐like NPC structures and nuclear envelope distortions observed in nup85− and nup120− cells (Aitchison et al., 1995a; Goldstein et al., 1996; Siniossoglou et al., 1996). Thus, individual members of the Nup84p complex including C‐Nup145p perform similar, but not identical functions in NPC and nuclear membrane biogenesis. Altogether these results confirm and extend recent results obtained through analysis of C‐terminal truncations of Nup145p (Dockendorff et al., 1997; Emtage et al., 1997). On the other hand, lack of N‐Nup145p has no effect on NPC distribution; the grape‐like NPC structures previously observed in a mutant strain in which the amino‐terminal domain of Nup145p was deleted may have been caused by a concomitant reduction in the expression of the carboxy‐terminal domain of Nup145p (Wente and Blobel, 1994). Beside their overlapping roles in NPC biogenesis, components of the Nup84p complex are required for efficient mRNA export. In this respect, mutants lacking C‐Nup145p are not distinguishable from other Nup mutants of the Nup84p complex.
N‐Nup145p, when provided as a free entity, is targeted to the NPCs and becomes nucleoplasmic. A similar dual intracellular location was found for the GLFG nucleoporin Nup98p from Xenopus, which is homologous to N‐Nup145p, Nup116p and Nup100p (Powers et al., 1995). Interestingly, Nup98p was recently shown to be involved in nuclear export of mRNA, rRNA and snRNA (Powers et al., 1997). It is, therefore, attractive to speculate that, upon travelling between the intranuclear compartment and the nuclear pores, Nup98p is involved in RNA export mechanisms. We do not have direct evidence for participation of N‐Nup145p in RNA transport reactions; however, the NRM domain (nucleoporin RNA‐binding motif) found in N‐Nup145p was shown to be genetically linked to a corresponding domain in Nup116p and Nup100p (Fabre et al., 1994). This domain binds homopolymeric RNA in vitro and deletion of NUP100 in combination with deletion of both NRMs of Nup116p and Nup145p leads to defective RNA export reactions (Fabre et al., 1995). First analysis of synthetic lethality between the lack of N‐Nup145p and the NUP188 mutant allele shows that defects in mRNA export and NLS‐dependent nuclear protein import are not the primary cause of cell death. It will be interesting to determine whether under these conditions other nuclear export or import pathways are affected.
Post‐translational proteolytic maturation of proteins is a widely used mechanism to control and regulate many different cellular processes, including protein targeting and translocation, assembly of protein complexes, enzymatic activation/inactivation, signal transduction and viral biogenesis (for reviews, see Palmenberg, 1990; Rehemtulla and Kaufman, 1992; Resnick and Zasloff, 1992; Little, 1993). Post‐translational modifications of NPC proteins, like the attachment of N‐acetylglucosamine residues to the protein backbone of FXFG‐repeat sequence‐containing nucleoporins, have been reported (Davis and Blobel, 1987; Holt et al., 1987; Starr and Hanover, 1990). Although the role of this modification is not clear, it may occur concomitantly with distinct assembly steps of nucleoporins into NPC structures (Davis and Blobel, 1987; Starr and Hanover, 1990). In the case of Nup145p, the half‐time of precursor maturation is ∼10 min, suggesting that maturation does not occur immediately after completion of Nup145p precursor synthesis. This delay in processing, together with our biochemical and localization results, suggests that assembly of Nup145p into the NPC precedes the cleavage reaction. After processing, Nup145p N‐ and C‐domains perform separate functions and even physically dissociate. We note, however, that simultaneous deletion of the two domains leads to an intragenic synthetic lethality, since the individual deletions can complement the lethal phenotype of a Δnup145 null mutant, at least at 30°C. Thus, the two domains may be involved in similar pathways.
The sequence around the site of proteolytic cleavage consists of the tetrapeptide 603‐N‐H‐F‐S‐606 which does not resemble the consensus cleavage site of any known protease. The cleavage site does, however, show some similarities to consensus sites for protein splicing in inteins (reviewed by Shub and Goodrich‐Blair, 1992; Cooper and Stevens, 1995). In such cases, excision of an internal segment (called intein) of a polypeptide and religation of the flanking regions (called exteins) is catalysed by the intein itself. In intein–extein junctions, serines or cysteines are found and histidine and asparagine amino acids are conserved in the carboxy‐terminal end of the intein. These residues are essential for the protein splicing activity. Thus, Nup145p may catalyse self‐cleavage by a mechanism similar to protein splicing, except that no religation of peptides is needed. Alternatively, a specific protease may exist in yeast to cleave Nup145p. Accordingly, mutation of the conserved residues would cause inhibition of self‐cleavage or recognition by a specific protease. There may be an exposed hinge region between the amino‐ and carboxy‐terminal domain of Nup145p which is easily attacked by (a) cellular protease(s). Upon mutation of this hinge region, a change of conformation would render the site less accessible for proteolytic attack. Interestingly, secondary structure predictions suggest the existence of a β‐sheet structure around the cleavage site which is disrupted in mutants of processing of Nup145p. However, this putative conformational change does not affect the functions of C‐Nup145p and N‐Nup145p. Indeed, non‐cleavable mutants of Nup145p can complement the NUP145 deletion and restore the temperature‐sensitive growth phenotype of a strain deleted for the NRM domains of NUP100 and NUP116, while the carboxy‐terminal domain alone cannot (Fabre et al., 1994 and unpublished results).
What could be the reason for the proteolytic cleavage of a nuclear pore protein in living cells? One clue as to why this may occur is the observation that Nup145p precursor cleavage becomes essential in a mutant background in which another nuclear pore protein, Nup188p, is mutated. Lethality could be explained either because: (i) cleavage is necessary; (ii) N‐Nup145p is not provided as a free entity; or (iii) C‐Nup145p remains as a part of the precursor. Yet, the cleavage is not a prerequisite for the function of C‐Nup145p because the non‐cleavable form of Nup145p is functional in the nup188‐mutated background (on condition that free N‐Nup145p is provided). In addition, the fact that N‐Nup145p, when expressed separately, can complement the synthetic lethal phenotype of the nsl7 mutant argues that cleavage per se is not required for functionality of the amino‐terminal domain and that this domain has an important role after cleavage. This raises the question of why these two Nup145p domains are covalently linked to each other in the initial phase of Nup145p biogenesis. One possibility, suggested by our results, is to ensure correct and/or efficient targeting of N‐Nup145p to the NPCs. Another explanation is that, by this mechanism, stoichiometric expression of both domains is guaranteed. Finally, it could be that the Nup145p precursor has a temporary function on its own during NPC assembly, before the two halves are cleaved and perform separate functions. It will be now of great interest to find out how nucleoporin cleavage is mechanistically linked to NPC biogenesis and/or function.
Materials and methods
Yeast strains, media and microbiological techniques
Standard yeast complex and minimal media were used (Sherman et al., 1986). Yeast transformation was performed with the one‐step procedure (Chen et al., 1992) or by electroporation (modified from Meilhoc et al., 1990). Yeast strains used in this study are listed in Table I (plasmids are indicated between brackets).
FYEF88 was constructed according to Fairhead et al. (1996). In order to avoid putative chromosomal rearrangements between NUP145‐containing plasmids and the genomic copy, the ORF of NUP145 was completely deleted from its ATG start codon to the TAA stop codon. This was performed by transforming yeast FYBL2‐5D (see Table I and Fairhead et al., 1996) with the complementary regions of 5′ and 3′ intergenes of NUP145 cloned in recombinant plasmids containing overlapping fragments of the selection marker URA3 (named pUR and pRA). Since NUP145 was known to be essential, FYBL2‐5D was previously transformed with a centromeric plasmid carrying NUP145 linked to the TRP1 marker. Transformants were then selected on glucose‐containing plates lacking uracil and tryptophan.
In order to recover the URA3 marker, chromosomal deletants containing two I‐SceI sites flanking URA3 (provided by pUR and pRA) were transformed with a plasmid expressing the I‐SceI gene under the control of a galactose‐inducible promoter and the LEU. d selection marker. Clones growing in galactose medium were then checked for the absence of URA3 and verified for correct genomic organization by Southern blotting. This strain was further transformed with a centromeric plasmid containing NUP145–URA3 and loss of the TRP1‐carrying plasmid was performed by successive replica plating. The strain FYEF88 or Δnup145, was used for all plasmid transformations described below.
Pulse–chase labelling and immunoprecipitation
Cells were grown at 30°C in minimal medium (2% glucose, 0.67% nitrogen base) supplemented with the required nutrients. After collecting the cells by centrifugation, the pellet was resuspended in the same medium to yield a final cell density of 4.5×108 cells/ml. After a 5 min pre‐incubation, cells were labelled for 5 min with 250 μCi/ml, 1000 Ci/mmol l‐[35S]methionine, l‐[35S]cysteine (ICN) and chased with a 300‐fold excess of non‐radiolabelled amino acids. Aliquots of 0.3 ml (1.55×108 cells) were withdrawn at the indicated time points of the chase and mixed with 1 ml of 1% SDS, 25 mM methionine, 25 mM cysteine and 1 mM Pefabloc (Boehringer). Whole‐cell extracts were prepared by the glass bead lysis method. Samples were then centrifuged at 4°C and supernatants were precleared by three incubations each of 1 h at 4°C with 100 μl protein A–Sepharose (1 g/50 ml) in 9 ml of TNET buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 0.1% Triton X‐100, 0.02% SDS). Finally, 2 μg of antibodies against protein A (Sigma) were added to the pre‐absorbed lysate and it was incubated for 1 h at 4°C, before 150 μl of protein A–Sepharose CL‐4B (20 mg/ml) were used to precipitate the antigen–antibody complexes (overnight incubation at 4°C on a turning wheel). ProtA–Sepharose beads were finally washed three times with TNET buffer. Bound antigen was eluted with 50 μl of 2× Laemmli's sample buffer and 20 μl were subjected to SDS–8% polyacrylamide gel electrophoresis. Gels were fixed, dried and autoradiographed.
Mutagenesis of the cleavage site within Nup145p
A PCR‐based protocol was used (Morrison and Desrosiers, 1993) for random mutagenesis of DNA encoding amino acids 600–608 of Nup145p. Briefly, two degenerate oligonucleotide primers (DG1: 5′‐TNCAAAGNCAANCATNNTAGCANTNGGGG‐3′; DG2: 5′‐AAATGCNAAAANGATNGACTNTGAAAGTCC‐3′) were synthesized to generate random mutations in the putative cleavage site. These two oligonucleotides, used in combination with two ‘outside’ primers, generated two PCR‐derived fragments which overlapped within 24 nucleotides of the randomly mutagenized region. These two PCR products were mixed, annealed and used in a third PCR reaction with the two ‘outside’ oligonucleotides, allowing mutagenized DNA to be subcloned into the XbaI–AatII sites of ProtA–NUP145 (Fabre et al., 1994), cloned into pUN100 (Elledge and Davis, 1988). This library of NUP145 mutagenized within the putative cleavage site was used to transform the shuffle strain FYEF88 and transformants expressing a non‐cleavable Nup145p were identified by Western blot analysis using an antibody against protein A, coupled to horseradish peroxidase (Dakopatts, Denmark). DNA fragments derived from PCR amplification were sequenced and mutations responsible for lack of cleavage were identified. Position W608 was found to be mutated in all our mutants and in most clones of the library. However, this mutation either alone or in combination with a mutation in F600 affects the cleavage reaction only slightly.
Construction of ProtA–NUP145 and GFP–NUP145 gene fusions and detection of the fusion proteins in yeast
Plasmid expressing the fusion ProtA–NUP145N was obtained by substituting the ProtA–NUP145 cassette described above by in‐frame fusion of PCR products encoding the N‐terminal region of NUP145 and NUP145 terminator. Cloned PCR fragments (636 bp) were verified by sequencing. Translation of NUP145N ends with the sequence VNHL(605) which corresponds to the predicted carboxy‐terminal end of N‐Nup145p, except for replacement of F605→L (numbers correspond to the positions in the wild‐type sequence of Nup145p). Similarly, plasmid expressing the fusion ProtA–NUP145C was obtained by substituting the ProtA–NUP145 cassette by in‐frame fusion of PCR products corresponding to the NOP1 promoter–protein A cassette (Fabre et al., 1994) and the region encoding for the carboxy‐terminal domain of Nup145p. Cloned PCR fragments (2047 bp) were verified by sequencing. Translation of the junction ProtA–NUP145C is LINI607WGL (bold letters indicate residues added between protein A and C‐Nup145p sequence). Expression of each construct under the authentic promoter was obtained by replacing the NOP1 promoter by a PCR product corresponding to the NUP145 promoter. In‐frame GFP fusions were obtained by replacement of protein A tag from previous constructs by GFP sequence which was isolated as a restriction fragment from pVD241GFP (containing the S65T/V163A GFP variant, a generous gift from V.Doye).
Each ProtA or GFP fusion protein was verified to be expressed to similar levels by Western blot analysis. Whole‐cell extracts of haploid strains expressing the different ProtA or GFP fusion constructs were prepared from cells grown in YPD medium or in selective media if required and according to Fabre et al. (1994). An equivalent of 0.05 OD600 of cells was analysed on SDS–8% polyacrylamide gels, followed by immunoblotting using IgG coupled to horseradish peroxidase (1:5000; Dakopatts, Denmark) to detect ProtA fusion proteins or using anti‐GFP polyclonal antibodies (1:1000; Clontech) followed by incubation with anti‐rabbit‐HRP (1:5000; Sigma).
Affinity purification of ProtA fusion proteins
Affinity purification of ProtA fusion proteins by IgG–Sepharose chromatography from whole‐cell extracts under non‐denaturing conditions was done according to Grandi et al. (1993) and Siniossoglou et al. (1996). Strains expressing the fusion proteins were spheroplasted and lysed in 1% Triton X‐100, 150 mM KCl, 20 mM Tris–HCl, pH 8, 5 mM MgCl2, supplemented by a cocktail of protease inhibitors. After centrifugation (27 000 g, 10 min), the supernatant was recovered and loaded on an IgG–Sepharose column (Pharmacia). After several washes, elution of the attached proteins from the ProtA‐tagged fusions was done by acidic pH.
For detection of a physical interaction between ProtA–Nup145p and Myc–Nup188p, lysis and ProtA purification were performed as above, but an anti‐c‐Myc monoclonal antibody (9E10, Boehringer‐Mannheim) at 10 μg/ml and a HRP‐sheep anti‐mouse antibody (secondary antibody, dilution 1:5000) were used for Western blot analysis.
To immunolocalize the nucleolar marker Nop1p, exponentially growing cells were fixed in 3.7% formaldehyde for 30 min. Cells were converted to spheroplasts using 0.5 mg/ml zymolyase 100.000T as described by Wimmer et al. (1992), immobilized on poly‐lysine‐coated coverslips and then incubated for 15 min in PBS/0.2 M glycine and 30 min in PBS/0.2% BSA. Nop1p was detected with monoclonal antibody A66 in a dilution of 1:100. Secondary antibodies, donkey anti‐rabbit IgG coupled to Cy3 or donkey anti‐mouse coupled to FITC (Jackson Immunoresearch Laboratories, Inc.) were diluted to 1:100. GFP–Nup49p was observed directly on fixed cells. Hoechst 33258 was used to stain DNA. Cells were examined in a Leitz DMRB fluorescence microscope.
To localize GFP fusion proteins in the cell, growing cells were transferred in Sykes Moore Chambers (Bellco, Vineland, NJ) filled with 1.2 ml selective medium and directly observed with a Leica inverted microscope equipped with a ×100/1.4 objective lens. Scanning was performed with a True Confocal Scanner LEICA TCS 4D and GFP fluorescence signal was detected with the fluorescein channel. Acquisitions of four focal planes, 0.8 to 1.2 μm apart, were made.
Mass spectrometry analysis
Bands of interest were excised from the gel and subjected to proteolytic digestion with an excess of trypsin as described (Wilm et al., 1996). Then 0.3 μl of digestion supernatant were deposited directly into an acidified water droplet placed on top of a fast evaporation matrix surface. Peptide mass maps were recorded on a Bruker Reflex Matrix Assisted Laser Desorption Ionization Time of Flight mass spectrometer equipped with a 337 nm N2 laser (Bruker‐Franzen, Bremen, Germany).
20 ml of cells growing in exponential phase were washed once with Buffer A (40 mM Kpi, pH 6.5, 0.5 mM MgCl2) and fixed overnight at 4°C with 2 ml of buffer A containing 2% fresh paraformaldehyde (Merck), 2% glutaraldehyde (Sigma). After treatment with a reducing buffer (0.2 M Tris–HCl, pH 9.4, 20 mM EDTA, 0.1 M β‐mercaptoethanol) cells were spheroplasted for 3 h in phosphate–citrate buffer containing 0.2 mg/ml of zymolyase 20T and 100 μl of glusulase. After washing three times with 0.1 M Na‐acetate, pH 6.1, cells were post‐fixed in 2% osmium tetroxide and 1% uranyl acetate and further dehydrated in acetone and propylenoxide. Embedding was performed in propylenoxide/epon. Forty nm sections collected on nickel grids were contrasted by staining with uranyl acetate and Reynold's lead. Specimens were visualized with a Zeiss EM109 electron microscope operating at 80 kV.
Analysis of poly(A)+ RNA export
Export of polyadenylated RNA from the nucleus was examined according to Fabre et al. (1994) using the FYEF95 (ΔC) mutant strain shifted to 37°C for 0.5, 1, 3, 5, 7 and 10 h, or grown at 30°C. The in situ hybridization steps were conducted exactly as described by Amberg et al. (1992). The oligo(dT) probe was directly coupled to Cy3. Photographs were taken with the 100× objective on a Leitz DMRB fluorescence microscope.
Isolation of synthetic lethal mutants
The synthetic lethal screen was performed on the basis of the ade2/ade3‐dependent red/white sectoring colony assay (Wimmer et al., 1992) using a strain deleted for the NUP145 gene (FYEF121, see Table I) and complemented by both an ARS–CEN–URA3–ADE3–NUP145 plasmid and an ARS–CEN–TRP1–ProtA–NUP145C. This strain performs a distinct red/white colony sectoring on SDC‐trp plates at 30°C. UV‐induced mutagenesis was performed as described by Wimmer et al. (1992) allowing a survival rate of 10%. From 60 000 screened colonies, one non‐sectoring mutant fulfilled the requirement to be synthetically lethal with the lack of N‐Nup145p but not with the lack of C‐Nup145p. Accordingly, this sl mutant (nsl7) was complemented by ARS–CEN–LEU2–ProtA–NUP145N. nsl7 was further transformed with a yeast genomic library inserted into pUN100 (ARS–CEN–LEU2‐based plasmid), as described by Wimmer et al. (1992). Three transformants displaying a red/white sectoring phenotype were obtained. Restriction analysis from plasmid DNA recovered from these transformants showed that two plasmids contained a genomic insert corresponding to NUP145, while one contained an unrelated insert of 12 kb. Sequencing of the ends of this latter insert revealed that it contained NUP188. Complementing activity was finally restricted solely to NUP188 (Zabel et al., 1996) by transforming nsl7 with pUN100 plasmids carrying either the NUP188 wild‐type gene or a nup188ΔSpeI mutant gene, which contains an internal out‐of‐frame deletion. Only the full‐length NUP188 was able to complement synthetic lethality in strain nsl7.
Pairwise combinations between disruption of NUP145 and NUP188, NUP157 or NIC96 were obtained by mating FYEF121 (Table I) with respectively NP188‐4‐1‐2, NP157‐2.1 (Aitchison et al., 1995b), nic96‐1 or nic96(Δ532–839) (Grandi et al., 1995). Tetratypes were selected after sporulation and double disruption was verified by Southern blot analysis. Haploids with the correct pattern were further transformed by ProtA–NUP145N and ProtA–NUP145NC1 (carried on ARS–CEN–LEU2 plasmids; see Table I). Synthetic lethality was tested by checking either growth on 5‐FOA or, in the case of FYEF178 (Table I), the capacity to segregate ProtA–NUP145C‐carrying plasmid. Since NIC96 disruption is lethal to cells, their viability is dependent on the LEU2 plasmid which carries the different alleles of NIC96. Thus, only the requirement for N‐Nup145p could be tested.
In the case of POM152, pairwise disruption could not be obtained through mating. Therefore, POM152 was directly disrupted in FYEF121 according to Fairhead et al. (1996).
We are very grateful to G.‐F.Richard, V.Doye and D.Tollervey for numerous discussions and critical reading of the manuscript, C.Fairhead for advice in FYEF88 construction and A.Perrin for help in database analyses. We thank V.Doye for the gift of GFP–Nup49p and for Nup188p‐expressing plasmids and R.Wozniak for NUP188‐, NUP157‐ and POM152‐disrupted strains. M.T.T. is a recipient from Junta Nacional de Investigação Científica e Tecnológica (PRAXIS XXI/ BD/5226/95). E.C.H. is a recipient of a grant from the Deutsche Forschungsgemeinschaft (SFB352). This work was supported by CNRS, FRM, Université Pierre et Marie Curie and Institut Pasteur.
- Copyright © 1997 European Molecular Biology Organization