An Unexpected Link between the Secretory Path and the Organization of the Nucleus*

Yeast sec mutations define the machinery of vesicular traffic. Surprisingly, many of these mutations also inhibit ribosome biogenesis by reducing transcription of rRNA and genes encoding ribosomal proteins. We observe that these mutants reversibly inhibit protein import into the nucleus, with import cargo accumulating at the nucleoplasmic face of nuclear pore complexes, as when Ran-GTP cannot bind importins. They also rapidly and reversibly relocate multiple nucleolar and nucleoplasmic proteins to the cytoplasm. The import block and relocation are antagonized by overexpression of yeast Ran, Hog1p kinase, or Ssa/Hsp70 proteins or by inhibition of protein synthesis. These nucleocytoplasmic signaling events document an extraordinary plasticity of nuclear organization.

Immunofluorescence and Electron Microscopy-Cells were fixed and processed as described (23). Formaldehyde was added directly to growing cultures for 10 min. Cells were then sedimented and resuspended for 10 min at room temperature in 3.7% formaldehyde, 10% methanol, 0.1 M potassium phosphate, pH 6.5; washed three times in buffer; and washed once in buffered 1.2 M sorbitol. They were then carefully spheroplasted, adhered to polylysine-coated slides, dehydrated, and immunostained.
For electron microscopy, cells were fixed in 2% formaldehyde, 0.1% glutaraldehyde, spheroplasted, and embedded in LR White. Thin sections were incubated with rabbit anti-␤-galactosidase (Cappel, catalog no. 55976), which had been freed from cross-reactivity by preadsorption on fixed yeast, and then with anti-rabbit Ig colloidal gold.
Immunoblotting-Glass bead extracts of cells prepared with buffered SDS were reduced, fractionated by SDS-polyacrylamide gel electrophoresis, blotted to nitrocellulose membranes, and probed with the indicated antibodies, using ECL reagents for signal detection.
Import Assays-Yeast carrying pGAL-NLS-lacZ were grown in raffinose dropout medium at 23°C and shifted to the indicated temperatures upon addition of galactose (2% final concentration) 1-2 h before fixation and immunostaining.
Yeast carrying pMIG1-GFP-lacZ were grown in glucose dropout medium and then diluted into dropout medium containing 5% glycerol until growth resumed and nuclear GFP could not be detected. (The preliminary growth in glucose medium is needed since growth in glycerol is slow.) To study import, cultures were shifted for 30 min to 23 or 37°C and then supplemented with glucose (2% final concentration) Ϯ cycloheximide. After 5 min, the cells were quickly washed and examined.
Yeast carrying pHOG1-GFP were grown overnight in glucose synthetic medium. At this point, Hog1p-GFP is readily detected in the cytoplasm. The cells were then preincubated for 30 min at 23 or 37°C and challenged by supplementation with NaCl (0.4 M final concentration) for 10 min before examination.

RESULTS
Nuclear Proteins Relocate to the Cytoplasm in sec Mutants-To explore the causes of the transcriptional inhibition which is observed in sec mutants, we localized several nuclear proteins in sec mutants and observed that they relocate to the cytoplasm. Fig. 1A, for example, illustrates relocation of the nucleolar prolyl hydroxylase, Fpr3p/Npi46p, at 37°C upon inhibition of vesicle exocytosis in sec1-1, upon inhibition of transport through the Golgi in sec7-1, and upon inhibition of several steps of membrane traffic in sec18 -1. Comparable relocation is also seen in sec12-4 and sec23-1, which block export from the ER. Several other nucleolar proteins also relocate under these conditions (Cbf5p, Nop1p, Nop4p, Nsr1p, and Ssb1p), as does the nucleoplasmic protein, Npl3p/Mtr13p (20, 24 -26). A substantially weaker effect is seen in sec63-1, which interrupts translocation into the ER. To interrupt membrane traffic in the absence of mutant proteins and temperature increase, we have used brefeldin A (27) and also observe distinct although less extreme relocalization of a nucleolar protein within 30 min at 30°C (Fig. 1B).
The relocation in sec mutants of Fpr3p and the yeast fibrillarin homologue, Nop1p, is strikingly inhibited by cycloheximide (Fig. 1A) or by the transcriptional inhibitor, thiolutin (data not shown). At least Fpr3p is known not to shuttle (20).
Moreover, relocation of Fpr3p and Nop1p is reversible: after incubating sec1-1, sec7-1, or sec18 -1 for 2 h at 37°C, return to 23°C for 2 h Ϯ 100 g/ml cycloheximide leads to reconcentration of these antigens in the nucleolus, unless ATP synthesis is inhibited. This evidence of reversibility demonstrates the significant stability of at least Fpr3p and Nop1p, thereby ruling out an alternate explanation of their appearance in the cytoplasm: ongoing synthesis and lack of import.
Nuclear Import Is Inhibited in sec Mutants-Our earlier studies have documented a close relation between relocation of nucleolar proteins to the cytoplasm and inhibition of nuclear import (20). We have therefore inquired whether the three sec mutants inhibit import of NLS-␤-galactosidase, which has the nuclear localization signal (NLS) of SV40 large T antigen, and other proteins that require distinct import equipment: Mig1p-GFP-␤-galactosidase, H2B-GFP, and Hog1p-GFP (19,21,22). Import of each of these proteins can be readily documented in wild type at 23°C or 37°C, but is strongly inhibited in the sec mutants ( Fig. 2A). When NLS-␤-galactosidase is induced for 1 or 2 h at 37°C, it is detected at the periphery of the nucleus and in the cytoplasm, with the peripheral distribution being most visible in sec1-1 and sec7-1 and the cytoplasmic signal most FIG. 1. Relocation of nuclear proteins. A, sec1-1 (AFY138), sec7-1 (AFY62), and sec18 -1 (RSY271) were grown at 23°C, transferred to 37°C for 2 h, fixed, and stained to localize Fpr3p using a monospecific polyclonal antibody from J. Thorner. Relocation is extensive within 30 min, but we illustrate a 2-h time point for comparison to the additional panels. Note that inclusion of cycloheximide (cy, 100 g/ml) stops relocation and that induction of yeast Ran (Cnr1p/Gsp1p) or Ssa1p from a GAL promoter or expression of Hog1p from a high copy plasmid protects against relocation. Mock induction of a control pRS316 derivative including a GAL1 promoter and polyadenylation site but lacking an open reading frame does not provide protection (data not shown). For the Ran and Ssa1p experiments, cells were grown in raffinose medium at 23°C, induced with 2% galactose for 1 h at 23°C, and then shifted to 37°C for 2 h. In each case, comparable observations were made for Nop1p. B, treatment of S. pombe wild type strain DS2 with 50 g/ml brefeldin A (ϩ) for 30 min at 30°C causes partial relocalization of Gar1p from the nucleolus to the cytoplasm, by comparison to controls lacking brefeldin (Ϫ), as judged by indirect immunofluorescence with a polyclonal antibody from M. Caizergas-Ferrer (61). These experiments were done with S. pombe because Saccharomyces cerevisiae is resistant to brefeldin A. C, two mutations that inhibit the Ran GTPase cycle and nuclear import, mtr1-1 (TK17) and rna1-1 (EE1b-6), cause partial relocation of Fpr3p upon incubation at 37°C for 2 h. Similar observations were made for Nop1p. These effects are most visible in synthetic medium.

FIG. 2. Protein import in sec mutants.
A, light microscopy. Cells carrying pGAL-NLS-lacZ were shifted to galactose medium for 2 h at 23°C or 37°C and processed for detection of ␤-galactosidase. Note the nuclear signal at 23°C and the signal at the nuclear periphery (white lines) or cytoplasm at 37°C in sec1-1 and sec7-1. In sec18 -1, only the cytoplasmic signal can be seen. Such inhibition of import is reversible upon subsequent temperature reduction to 23°C. Among other import cargoes that we have tested, although NLS-␤-galactosidase is stable, several are not fully retained if they are delivered to the nucleus at 23°C and the cells are then shifted to 37°C. Thus, although it is quite clear that multiple import cargoes do not accumulate in the nucleus at 37°C in sec mutants, this may in part result from secondary release. B, electron microscopy immunolabeling. sec7-1 (pGAL-NLS-lacZ) cells shifted to galactose medium for 2 h at 37°C were fixed and processed for immunogold detection of ␤-galactosidase. Note especially the signal traversing nuclear pores (black arrows) and at the inner face of the NPC (open arrows). N, nucleus. C, to evaluate the ability of several proteins to protect nuclear import, sec18 -1 carrying pMIG1-GFP-lacZ was transformed with a control plasmid (pGP315), pGAL-CNR1, pGAL-SSA1, or pHOG1. To induce Cnr1p or Ssa1p, the corresponding transformants were transferred to raffinose medium containing 2% galactose for 1 h at 23°C. As indicated under "Experimental Procedures," the cells were then shifted to medium containing 5% glycerol for 30 min at 23°C or 37°C. They were subsequently challenged with 2% glucose to attempt to cause import and examined after 5 min. Samples incubated at 23°C all showed obvious nuclear fluorescence (data not shown). The 37°C samples are illustrated. Note the cytoplasmic fluorescence in the control transformant and the nuclear signal in the other samples, indicative of import. visible in sec18 -1 ( Fig. 2A). Fig. 2 (B and C) shows that a discrete amount of the import cargo actually enters the extreme periphery of the nucleoplasm. This is quite different from the docking of cargo at cytosolic NPC-associated sites in the absence of an energy source (28). Thus, the sec mutants may inhibit release of import cargo from the nucleoplasmic face of the NPC, as when the import factor, importin ␤, cannot interact with the GTP-bound nuclear pool of the GTPase, Ran, which governs nucleocytoplasmic transport (29,30). As for relocation of nucleolar proteins, protein synthesis is required for inhibition of import; for example, Mig1p-GFP-␤-galactosidase normally enters the nucleus upon shift from raffinose to glucose medium and this entry is inhibited in the sec mutants at 37°C unless cycloheximide is present (data not shown).
Overexpression of Ran or Hog1p Kinase Protects sec Mutants Against Nuclear Perturbation-Since Ran cycle mutants inhibit nuclear import and since the site of inhibition of import in sec mutants suggests that there is a defect at the level of the Ran cycle, we have inquired whether the Ran-GAP mutant, rna1-1, and the Ran-GEF mutant, mtr1-1, relocate nucleolar proteins. As shown in Fig. 1C, this is the case. Moreover, overexpression of yeast Ran strikingly protects import and reduces the relocation of Fpr3p and Nop1p in sec1-1, sec7-1, and sec18 -1 (Figs. 1A, 2C, and 3B).
Although these observations are consistent with the notion that import is critical for retention of nuclear proteins, inhibition of the importin ␣/␤ import path in corresponding mutants does not cause relocation ((srp1-31, rsl1-1) (Refs. 14 -15 and 29, 31; data not shown). Furthermore, Ran does remain concentrated in the nucleus as judged by immunofluorescence in the sec mutants (data not shown) and, although more subtle modifications may occur, the abundance and 1D gel mobility of yeast Ran, Rna1p and Mtr1p/Prp20p do not change over 2 h at 37°C (Fig. 3A).
Some aspects of cell physiology in the sec mutants (increased concentration of cytosolic proteins, limitation of the cell perimeter, altered relation between the cell wall and plasma membrane, etc.) may resemble the effects of exposure to media of altered tonicity. We have therefore asked whether expression from a high copy plasmid of the key MAP kinases implicated in resistance to hypertonic and hypotonic stress (Hog1p or Mpk1p (the terminal kinases of the PKC signaling path)), ensures protein import and nucleolar coherence in the sec mutants at 37°C (12,32). Remarkably, studies of the import of Mig1p-GFP-␤-galactosidase and localization of Fpr3p or Nop1p show that Hog1p does have these effects (Figs. 1A and 2C). Mpk1p does not provide protection (data not shown).
Activation of Hog1p by hypertonic stress is accompanied by its phosphorylation and transient entry into the nucleus, its exit being mediated by the importin ␤ family member, Crm1p (19,33). Since the protection by Hog1p may also reflect its entry into the nucleus, we have monitored the distribution of Hog1p-GFP in sec mutants. We observe that it remains concentrated in the cytoplasm, and that this is true even upon hypertonic shock (data not shown). Moreover, experiments in which Hog1p-GFP is expressed in crm1-1 show that Hog1p-GFP cannot be trapped in the nucleus, i.e. it does not cycle through the nucleus under normal growth conditions. Any relevant targets of Hog1p kinase activity therefore may be external to the NPC. Nevertheless, upon overexpression there may be some increased nuclear titer of Hog1p. Interestingly, HOG1 is not essential, and a ⌬hog1 strain is not obviously deficient in nuclear import or nucleolar organization.
Role of Heat Shock Proteins-There has been only limited investigation of the heat shock response (HSR) in sec mutants (34,35). We observe that when sec1-1, sec7-1 or sec18 -1 is incubated at 37°C, there is a strong and sustained HSR, as judged by transcription from a reporter plasmid, pHSE 2-lacZ. The HSR exceeds that seen in wild type cells, both in intensity and in duration, with sec18 -1 being the strongest. Moreover, immunoblotting and pulse labeling of newly synthesized proteins show clear induction of Hsp104 and Hsp70 proteins in these mutants at 37°C. Nevertheless, by comparison to several temperative-sensitive strains that do not affect membrane traffic or relocate nuclear proteins (act1-2, cdc33-1, pre1-1, prp5-1, sec62-1), the HSR in sec mutants is of intermediate intensity. In fact, sec62-1 induces a HSR that is substantially stronger than all the other strains. Moreover, incubation of wild type at 37°C or treatment with doses of ethanol, arsenite, and dihydrosphingosine known to produce a HSR does not cause relocation. Thus, the HSR does not parallel relocation. Interestingly, inhibition of import of Mig1p-GFP-␤-galactosidase and relocation of the nucleolar proteins do not require induction of heat shock proteins that depend on heat shock transcription factor (Hsf1p), judging from examination of a sec1-1 hsf1 double mutant. Moreover, the hsf1 mutant (and ssa1) does not itself inhibit import or cause relocation (data not shown).
Although a HSR does not necessarily perturb the nucleus and although Hsf1p function is not needed to cause such perturbation, the impact of sec mutations on the nucleus appears similar to the effects of incubation at elevated temperatures. Incubation at 42°C, for example, also inhibits protein import and causes nucleolar proteins to relocate to the cytoplasm (20), and incubation even at 37°C selectively inhibits transcription of ribosomal protein genes (26,37). Since Ssa/Hsp70 proteins are needed to protect yeast against relocation at elevated temperature (20), we have inquired whether overexpression of Ssa1p at 37°C can protect the sec mutants against nuclear changes. Strikingly, Ssa1p overexpression does protect against inhibition of import and relocation (Figs. 1A and 2C). Expression of yeast Ran or Hog1p does not alter levels of Ssa proteins, as detected by Western blotting (data not shown). Thus, the protection due to Ran or Hog1p appears not to be mediated by increasing the amount of Ssa proteins.   3. Proteins of the Ran cycle. A, sec1-1, sec7-1, and sec18 -1 were incubated 2 h at 23°C or 37°C. An equal amount of protein from each strain was processed to detect yeast Ran (Cnr1p/Gsp1p), Ran-GAP (Rna1p), and Ran-GEF (Mtr1p/Prp20p) by Western blotting with monospecific polyclonal antibodies from P. Belhumeur and A. Hopper. Note that the titer and mobility of these proteins is not changed at 37°C. B, the same mutants were transformed with pGAL-CNR1, grown in raffinose medium at 23°C, and then induced (I) (or not (U)) by addition of galactose for 1 h at 23°C. Note the increased quantity of the GTPase upon induction. Scanning of the image shows 12%, 25%, and 68% induction in sec1-1, sec7-1, and sec18 -1, respectively. 38). Moreover, although we find that sec63 mutants do not have a strong phenotype, yeast carrying point mutations in Sec63p mislocalize an unidentified nucleolar protein and a newly synthesized SV40 large T antigen-NLS-invertase fusion to the cytoplasm (39). Strikingly, it has also been reported that Schizosaccharomyces pombe can be rendered resistant to brefeldin A by overexpression of a homologue of a Ran-binding protein or by mutating the exportin, Crm1p (27).
What Is the Common Signal/Mediator?-We suggest that sec1, sec7, and sec18 each send equivalent signals, which impact on Ran cycle function and cause relocation of nuclear proteins. We do not observe equivalent perturbation of the nucleus in sec mutants that block polypeptide translocation into the ER; however, this may be because of the major HSR that such mutants elicit.
Why does interruption of membrane traffic perturb the nucleus? We can exclude several possibilities.
(b) To learn whether relocation results from overaccumulation of newly synthesized proteins or lipids upstream from the secretion blocks, we have inquired: 1) whether relocation is seen in a sec61-2 sec18 -1 double mutant, since sec61-2 reduces translocation of newly synthesized proteins into the ER at the restrictive temperature (41); and 2) whether relocation can be eliminated by inhibitors of fatty acid and sterol biosynthesis (cerulenin, zaragozic acid; Ref. 42). These experiments do not support the hypothesis that overaccumulation in the ER causes relocation. Relocation in sec18 -1 sec61-2 is comparable to sec18 -1 and the inhibitors of lipid synthesis (which themselves do not affect the distribution of nucleolar proteins in wild type cells), do not protect the sec mutants against the impact of the 37°C incubation (data not shown).
(c) An unfolded protein response is induced by tunicamycin and tunicamycin does inhibit ribosome biogenesis (6,8), but tunicamycin treatment does not cause relocation of nucleolar proteins over 2 h at 30°C (data not shown).
(d) Cycloheximide or transcriptional inhibitors do not cause relocation of nucleolar proteins. Moreover, these inhibitors protect the nucleus of sec mutants. It therefore cannot be true that mere interruption of delivery of one or more newly synthesized proteins to the cell surface is sufficient to perturb the nucleus.
(e) "Unbalanced growth" may occur when the sec mutants continue to increase their mass but are unable to enlarge their surface area. A comparable imbalance has been suggested to underlie the "inositol-less death" that occurs when inositol auxotrophs are deprived of inositol, unless cycloheximide is present (43). Nevertheless, inositol withdrawal from an ino1 strain does not relocate nucleolar proteins over 4 h at 30°C (data not shown).
The Relation between Ssa Proteins, Protein Kinases, and Ran-As a conservative hypothesis, since overexpression either Ssa1p or yeast Ran protects the nucleus, we propose that interruption of membrane traffic sequesters or inactivates Ssa proteins, which are known to be critical for import (45,46) (Fig.  4). Evidence consistent with there being a reduction of levels of free Ssa proteins in sec mutants is provided by the induction of transcription from the HSR reporter plasmid, considering that reduction of free Hsp70/Ssa protein levels is known to induce a HSR (44). 2 Despite these indications of Ssa protein involvement, mere loss of Ssa function in ssa1 and hsf1 strains does not inhibit import or relocate nuclear proteins at 37°C. We therefore propose that interruption of membrane traffic must also cause other events that are functionally linked to components of the Ran cycle, e.g. phosphorylation of critical substrates.
Judging from our observations of protection by Hog1p and the recent report that protein kinase C (but not Hog1p or cAMP-dependent protein kinase) is required for sec perturbation of the nucleus (8), Hog1p and PKC paths play opposing roles, possibly acting on the same targets. We suggest that subsequent loss of Ran cycle function then inhibits protein import and that inhibition of import leads to relocation of nuclear proteins and therefore inhibits transcription. Once relocation of nuclear proteins begins, if key factors that are needed for import are also perturbed, the process may accelerate.
The cycloheximide sensitivity of nuclear perturbation by sec mutants may reflect a need for ongoing synthesis of some key protein (s), e.g. a protein that activates PKC. Since Hsp70 proteins participate in conformational maturation of newly synthesized proteins (47)(48)(49), an alternate possibility is that Hsp70 proteins are most available for import when protein synthesis is inhibited. Thus, in the presence of cycloheximide, 2 We have also obtained evidence of a functional interaction between Ssa1p and the Ran cycle; overexpression of Ssa1p suppresses the protein relocalization seen in Ran-GEF mutants (mtr1/prp20), and the allele specificity of this suppression is the same as for Cnr1p (yeast Ran) suppression of mtr1/prp20 mutants (19). This is consistent with biochemical observations of association between these proteins in animal cells. Several previous studies have also implicated Hsp70 proteins in nuclear import.
FIG. 4. Model of sequential events. We postulate that inhibition of membrane traffic inhibits Ran function and/or import. Once protein import is inhibited, nuclear/nucleolar proteins relocate, causing inhibition of transcription. Hog1p may activate Ran function or some other aspect of import. The inhibition of membrane traffic may itself limit levels of free Ssa proteins, which are known to be needed for import. Overexpression of any of the three underlined proteins protects the nucleus of sec mutants, as does inhibition of protein synthesis. The recent observation that PKC is required for ribosome synthesis to be inhibited (8) is summarized in the lower portion of the figure. As indicated by the dotted lines, PKC may in fact act on Ran function or other aspects of import.
although additional Ssa proteins are not synthesized, ambient levels may remain sufficient to protect the nucleus.
There may also be a specific molecular link between vesicular transport and nucleocytoplasmic transport. For example, both Sec13p and one ER membrane protein interact with nucleoporins (50,51), a member of the importin ␤ superfamily is implicated in both protein secretion and nuclear transport (52,53), and mutations in splicing factors can inhibit vesicular transport (54). Since GTPases of the Ras superfamily are critical for both vesicular transport and nuclear transport (55,56), it is also of interest that proteins have been identified which interact with both Ran and Rabs (57). Moreover, at least one Rab protein and Hsp70 interact with stress signaling paths (58,59).
Generality of Relocation of Nuclear Proteins-There are other circumstances in which nucleocytoplasmic transport is perturbed or nuclear proteins relocate to the cytoplasm, perhaps because of inhibition of protein import. Thus, several viruses that replicate in the cytoplasm of animal cells relocate specific nuclear proteins to the cytoplasm and the M protein of vesicular stomatitis virus inhibits nucleocytoplasmic transport, which depends on Ran (60). Although many nuclear proteins appear not to shuttle under normal growth conditions, they may do so under extreme circumstances.