|Home | About | Journals | Submit | Contact Us | Français|
Cdc48p from Saccharomyces cerevisiae and its highly conserved mammalian homologue VCP (valosin-containing protein) are ATPases with essential functions in cell division and homotypic fusion of endoplasmic reticulum vesicles. Both are mainly attached to the endoplasmic reticulum, but relocalize in a cell cycle-dependent manner: Cdc48p enters the nucleus during late G1; VCP aggregates at the centrosome during mitosis. The nuclear import signal sequence of Cdc48p was localized near the amino terminus and its function demonstrated by mutagenesis. The nuclear import is regulated by a cell cycle-dependent phosphorylation of a tyrosine residue near the carboxy terminus. Two-hybrid studies indicate that the phosphorylation results in a conformational change of the protein, exposing the nuclear import signal sequence previously masked by a stretch of acidic residues.
The Saccharomyces cerevisiae cell division cycle gene CDC48 has been characterized by a cold-sensitive mutant (cdc48-1, Moir et al., 1982 ) that arrests at 16°C as large budded cells with long aberrant microtubules spreading from an unseparated spindle pole body (Fröhlich et al., 1991 ). Cdc48p and its mammalian homologue p97/VCP (valosin-containing protein) play a central role in homotypic fusion. Isolated vesicles from yeast endoplasmic reticulum fuse after the addition of ATP and purified Cdc48p (Latterich et al., 1995 ). The fusion of rat Golgi-derived vesicles forming larger vesicles or cisternae requires, in addition to ATP, the addition of either VCP alone, or of N-ethylmaleimide-sensitive factor (NSF), soluble N-ethylmaleimide-sensitive factor attachment proteins (SNAPs), and protein p115 (Acharya et al., 1995 ; Rabouille et al., 1995 ). NSF and Cdc48p/VCP belong to the superfamily of AAA proteins and share a duplicated region of 230 amino acid residues (AAA box).
After subcellular fractionation, Cdc48p is found exclusively in the nuclear and microsomal fractions, while the 177,000 × g supernatant is essentially free of Cdc48p. Cdc48p is found mostly in the soluble fraction only if the cells have been broken by shaking with glass beads. VCP from porcine liver is mainly found in the microsomal fraction, soluble VCP (177,000 × g supernatant) is detected only if the tissue has been frozen before cell fractionation (Fröhlich et al., 1991 ). The protein appears to be predominantly membrane attached in these cells. In contrast, Xenopus oocytes contain soluble p97, which is found in the 100,000 × g supernatant of both the nucleus and of enucleated cells (Peters et al., 1990 ).
Egerton et al. (1992) found a tyrosine phosphorylation of VCP in cultivated murine T cells after stimulation of the T cell antigen receptor. The tyrosine at position 805 (of 806 residues) contributed to about 90% of the total phosphorylation (Egerton and Samelson, 1994 ). The corresponding position (834 of 835 residues) of Cdc48p is also a tyrosine residue.
We demonstrate cell cycle-dependent localization of both Cdc48p and human VCP. We identify a nuclear targeting sequence of Cdc48p and demonstrate that Cdc48p is tyrosine phosphorylated in vivo in a cell cycle-dependent manner. We show that nuclear import of Cdc48p is regulated by the phosphorylation and propose a molecular mechanism for the regulation.
Enzymes and chemicals used for immunofluorescence, Western blotting, molecular biology, and cell fractionation were obtained from Boehringer (Mannheim, Germany) or Sigma Chemical (Munich, Germany). Fluorescein-labeled sheep anti-rabbit IgG (F(ab′)2 fragment) and anti-mouse IgG, rhodamine-labeled sheep anti-rabbit IgG (F(ab′)2 fragment) and alkaline phosphatase-labeled sheep anti-rabbit IgG were from Boehringer; Cy3-labeled goat anti-mouse IgG was from Dianova (Hamburg, Germany); alkaline phosphatase-labeled goat anti-mouse IgG was from Sigma; mouse anti-phosphotyrosine antibodies 4G10 was from Upstate Biochemicals (New York, NY); mab 3–365-10 from Professor Anderer (Tübingen, Germany; Steinhilber et al., 1990 ), and rabbit anti-valosin antiserum (gift from M.J. Brownstein, Bethesda, MD; Koller and Brownstein, 1987 ) were used.
Phosphotyrosine phosphatase was obtained from Boehringer. Oligonucleotides were from MWG Biotech (Ebersberg, Germany). The Site-Directed Mutagenesis Kit was obtained from Stratagene (Heidelberg, Germany). For the two-hybrid assay the MATCHMAKER Two-Hybrid System from CLONTECH (Palo Alto, CA) was used. Bacto-Tryptone, Bacto-Peptone, and yeast extract were obtained from Difco Laboratories (Detroit, MI).
Temperature-sensitive Saccharomyces cerevisiae strains used were LH369 (cdc1-1), LH395 (cdc19-1), and LH127 (cdc20-1) (all described by Hartwell et al., 1973 ). KFY161 (MAT a lys2 his4) was used as a wild-type strain. Strain KFY247 MAT a/MAT α cdc48::URA3/ CDC48 his4-619/his4-619 leu2-3,112/leu2-3,112 ura3-52/ura3-52 was used for cloning of mutagenized cdc48 alleles in vector YEp52. Allele cdc48::URA3 was constructed by replacing a 1,160 base pair (bp) ClaI fragment containing the promoter region and 40% of the open reading frame of CDC48 by URA3. The transformed strains were sporulated on 2% potassium acetate plates, and the spores were segregated on YEP medium containing 4% glucose or 4% galactose.
For the two-hybrid assay, S. cerevisiae strain SFY526 (MATa ura3-52 his3-200 ade2-101 lys2-801 trp1-901 leu2-3, 112 canr gal4-542 gal80-538 URA3::GAL1-lacZ, CLONTECH; Bartel et al., 1993 ) was used. Unless stated otherwise, yeast cells were grown with shaking in liquid YEPD medium at 28°C.
Mutagenesis was performed using Escherichia coli XLmutS and E. coli XL1-Blue from Stratagene (La Jolla, CA).
Yeast expression vector YEp52 (Broach et al., 1983 ) was used for the GAL1 promoter regulated-expression of cdc48 alleles. Plasmid pcD1 containing the porcine VCP gene was a gift from M.J. Brownstein (Bethesda, MD).
The human amnion-derived cell line WISH (ATTC CCL 25) was grown as an adherent monolayer in DMEM medium containing 10% (vol/vol) fetal calf serum (Sigma), 2 mM glutamine, and 2000 U/ml each of penicillin and streptomycin.
For immunofluorescence microscopy, WISH cells were inoculated at a density of 105/cm2 on sterilized siliconized coverslips and incubated for 24 h at 37°C in an atmosphere of air containing 10% CO2. Rat hepatocytes were isolated and cultured as described previously (Gebhardt and Jung, 1982 ).
An exponentially growing culture of cdc20 mutants was transferred to a 37°C water bath and incubated for 3 h. The cells were then transferred to a 28°C water bath (t = 0) and incubated for various periods. An aliquot was taken every 15 min and processed for detection of phosphotyrosine, for immunofluorescence investigation, or for cell fractionation as described below.
WISH cells were inoculated at a density of 2.5 × 104/cm2 on sterilized siliconized coverslips and incubated for 30 h at 37°C in an atmosphere of air containing 10% CO2. After addition of 10 mM thymidine in growth medium to a final concentration of 1 mM, cells were further incubated for 12 h. The cell cycle block was released by washing three times with growth medium containing 1 μM deoxycytidine and incubation in the same medium (Madeo, Fröhlich, and Probst, unpublished method). To determine the proportion of mitotic cells indicating the degree of synchrony, cells were fixed with ice- cold ethanol:acetic acid:water (5:3:2) and the chromatin was stained with 0.025% crystal violet in 1% acetic acid for 5 min. While exponentially grown nonsynchronized cultures contained 2–3% mitotic cells, the synchronized cultures contained 25–30% mitotic cells 9 h after release of the thymidine block (Figure (Figure5A).5A). For Western blotting, cells were harvested by incubation with 0.05% trypsin (1–10 min) and boiled with 1% SDS for 15 min.
Aliquots (500 ml) of synchronized cdc20 cultures were collected at different times after reshifting the temperature to 28°C (see above). For subcellular fractionation, cells were digested with Zymolyase and the spheroplasts were harvested, washed twice, and homogenized by 40 strokes in a Dounce homogenizer as described by Gasser (1983) . The homogenate was centrifuged at 9,700 × g for 10 min to remove cell debris, nuclei, and mitochondria. The supernatant was centrifuged at 49,000 × g for 30 min to sediment the microsomal fraction, and the resulting supernatant was centrifuged at 177,000 × g for 90 min. The final supernatant was referred to as the cytoplasmic fraction. The procedure of Ide and Saunders (1981) was used for the isolation of yeast nuclei. Cells were digested with Zymolyase, and the spheroplasts were lysed and fractionated on a preformed Percoll gradient. Nuclei form a prominent band in the middle of the gradient. Contamination was calculated as the specific concentration (correlated to protein) of a marker molecule in relation to its concentration in the expected fraction. Alcohol dehydrogenase served as the marker for cytosol, and DNA (determined by microfluorometry, Cesarone et al., 1979 ) served as the marker for nuclei. Cross-contamination of the microsomal fraction with both markers, and of the nuclear fraction with alcohol dehydrogenase, was below 2%. Cdc48p was determined by Western blotting as described previously (Fröhlich et al., 1991 ). Anti-Cdc48p antiserum was diluted 1:1,250, and alkaline phosphatase- labeled anti-rabbit IgG antiserum was diluted 1:10,000. The Laemmli system (1970) with a 6% acrylamide separation gel was used for SDS-PAGE.
Cell extracts were obtained after vortexing the harvested cells with an equal volume of glass beads for 1.5 min. An equal volume of Tris-EDTA buffer was added, and the extract was centrifuged for 15 min in an Eppendorf centrifuge. The supernatant was used for Western blotting. For hydrolysis of phosphotyrosine, 50 μl of a cell extract from an exponentially growing wild- type strain and 1 μl (8 mU) of phosphoyrosine phosphatase were shaken for 3 h at 37°C.
For immunofluorescence, yeast cells were treated as described by Kilmartin and Adams (1984) with modifications. Cells were fixed for 20 min with 4.4% formaldehyde. After application to polylysine-coated slides, the samples were air dried instead of being treated with methanol/acetone.
Coverslips with adherent WISH cells were washed with ice-cold phosphate-buffered saline (PBS), treated with 4% formaldehyde (vol/vol in PBS) for 30 min at room temperature (22°C), washed three times with ice-cold PBS, and permeabilized at −20°C by sequential treatment with 100%, 50%, and 25% methanol (vol/vol in ddH2O) for 5 min each.
Cdc48p from Saccharomyces cerevisiae and VCP from porcine liver were purified to homogeneity (checked by SDS-PAGE with silver staining and by Western blotting) as described (Fröhlich et al., 1995 ). The proteins were used to immunize rabbits. Specificity of the antisera was checked by Western blots of whole cell protein extracts in which only single bands were recognized. The bands were indistinguishable in molecular weight from the purified antigens used to raise the antibodies. Furthermore, the bands were located at the same positions as those in Western blots stained with monoclonal anti-valosin antibodies from Koller and Brownstein (1987) . For immunofluorescence experiments, the antisera were affinity purified with purified Cdc48p or VCP, respectively, according to the protocol of Pringle et al. (1989) . Preincubation of the antisera with purified Cdc48p or VCP eliminates the intracellular signals, proving their specificity. The corresponding preimmune sera show no staining of the cells.
For immunofluorescence, rabbit anti-Cdc48p antiserum was diluted 1:32, rabbit anti-VCP antiserum was diluted 1:35, mouse anti-phosphotyrosine antibodies 4G10 were diluted 1:100, fluorescein-labeled sheep anti-rabbit IgG (F(ab′)2 fragment) was diluted 1:70, rhodamine-labeled sheep anti-rabbit IgG (F(ab′)2 fragment) was diluted 1:50, fluorescein-labeled sheep anti-mouse IgG (F(ab′)2 fragment) was diluted 1:100, and Cy3-labeled goat anti-mouseIgG was diluted 1:600. As a control, the anti-Cdc48p or anti-VCP antiserum was incubated with a tenfold molar excess of purified Cdc48p or VCP for 1 h at room temperature, centrifuged for 15 min at 20,000 × g, diluted, and immediately used for immunofluorescence. DNA was stained by a 5- min treatment with 1 μg/ml diaminophenylindole in yeast cells and with 1 μg/ml bisbenzimide in mammalian cells.
Western blotting was performed as described previously (Fröhlich et al., 1991 ). For the detection of phosphotyrosine, 5% bovine serum albumin was used for blocking. Anti-Cdc48p antiserum was diluted 1:1,250; anti-phosphotyrosine antiserum was diluted 1:1,000; alkaline phosphatase-labeled anti-rabbit IgG and anti-mouse IgG antibodies were diluted 1:10,000.
The 3′ end of CDC48 up to the internal SalI site at base 40 was replaced by an oligonucleotide pair (5′-AGCTTCCATGGGTGAAGAACACAAGCCATTGTTGGACGCTTCTGGTG-3′/3′-TCGACACCAGAAGCGTCCAACAATGGCTTGTGTTCTTCACCCATGGA5′), introducing a NcoI site at the START codon and a HindIII site directly before the open reading frame without altering the coded protein. A HindIII site was introduced 5′ of the terminator region of CDC48 by site-directed mutagenesis (oligonucleotide 5′-GAAAAAAGGGAAGCTTTAGGACCTCG-3′). The resulting HindIII fragment containing the complete CDC48 gene was cloned into HindIII- digested YEp52 (plasmid YEp52/CDC48). CDC48 was tagged by inserting a double- stranded oligonucleotide (5′-AGCTTATGTACCCATACGATGTTCCAGATTACGCTAGCTTGGGTGGTCC-3′/3′CATGGGACCACCCAAGCTAGCGTAATCTGGAACATCGTATGGGTACATA-5′) between the HindIII site before and the NcoI-site at the START codon. The resulting protein contains an insertion of 14 amino acid residues (Flu epitope) after the amino-terminal methionine, increasing its molecular mass by 1,513 Da. The tag has no apparent effect on protein function.
The codon of tyrosine834 of CDC48 was mutagenized to a phenylalanine codon (plasmid YEp52/cdc48Y834F) or a glutamic acid codon (plasmid YEp52/cdc48Y834E), respectively, and the nuclear localization sequence was destroyed (plasmid YEp52/cdc48nonuc) in plasmid YEp52/CDC48. The acidic domain was deleted in plasmid YEp52/cdc48Y834F (plasmid YEp52/cdc48Y834F/noacid). For site-directed mutagenesis, the Chameleon Site Directed Mutagenesis Kit from Stratagene was used. All mutations were confirmed by DNA sequencing (Table (Table1).1).
A NcoI fragment of CDC48 coding for the 56 amino-terminal amino acid residues of Cdc48p, a HinfI fragment coding for the 291 amino-terminal residues of Cdc48p, and HindII/DraI fragments coding for the 153 carboxy-terminal residues of Cdc48p, Cdc48pY834F, and Cdc48pY834E were isolated, their recessed 3′-ends were filled in with Klenow enzyme, and the fragments were ligated into vectors pGAD424 or pGBT9 (CLONTECH) digested with SmaI. An EcoRI/PvuII fragment of CDC48 coding for residues 173–578 (central part) was ligated into pGBT9 digested with EcoRI and SmaI. For the detection of interaction, combinations of a pGAD424 and a pGBT9-based plasmid were cotransformed into yeast strain SFY526 (Bartel et al., 1993 ), and transformants were selected on minimal medium (0.67% yeast nitrogen base, 2% glucose) supplemented with amino acids and bases lacking leucine and tryptophan. After 4 d of growth, yeast colonies were transferred onto Whatman No. 1 filters, disrupted by freezing in liquid nitrogen and thawing, and β-galactosidase activity was visualized by soaking the filter with a solution of 0.334 mg X-Gal and 2.7 μl β-mercaptoethanol per ml Z buffer (100 mM sodium phosphate, pH 7.0, 10 mM KCl, 1 mM MgSO4) and incubation at 30°C.
In exponentially grown S. cerevisiae wild-type cells (strain KFY161), immunofluorescence microscopy with anti-Cdc48p antiserum shows a stained nucleus in approximately 30% of the cells. The other cells have an unstained nucleus and a stain in the rest of the cell that is strongest in the vicinity of the nucleus (Figure (Figure1A).1A).
The dependency of the localization of Cdc48p from the stages of the cell division cycle was determined in synchronized cells. We used cdc20 mutant strain LH127 for cell synchronization because we found it to start proliferation fast and reproducibly after a reshift to the permissive temperature and to appear morphologically intact even after several hours at restrictive temperature. After 3 h incubation at 37°C, almost 100% of the cells are arrested with large buds (G2/M arrest) (Figure (Figure1F).1F). When the temperature is reshifted to 28°C (t = 0 min), cells begin to grow synchronously after a delay of 30 min. Up to t = 60 min, Cdc48p appears within the cytosol (Figure (Figure1B).1B). At t = 75 min, shortly before cell budding (G1/S-transition), most of the Cdc48p is concentrated in the nucleus of 90% of the cells (Figure (Figure1C).1C). After cell budding, Cdc48p redistributes and is again dispersed within the cytosol.
This cell cycle-dependent localization was demonstrated also by subcellular fractionation. Synchronized cultures of strain LH127 were harvested at different times (0 min, 15 min, 30 min, 45 min, 60 min, 75 min) after reshift to the permissive temperature (28°C), subcellular fractions were isolated and their Cdc48p content was determined by Western blotting (Figure (Figure1G).1G). The nuclear fraction shows an increase of Cdc48p from cells harvested at 0 min to cells harvested at 30 min followed by a decrease at later times. The microsomal fraction shows a complementary distribution with a decrease of Cdc48p from 0 min to 30 min followed by an increase. It should be noted that the cells are not fixed during the fractionation procedure, which includes an incubation (digest of the cell wall) of 60 min at 30°C, allowing intracellular processes to continue. The samples therefore do not correspond to the immunofluorescence samples harvested at the same time.
The localization of Cdc48p was investigated in cell division cycle mutants with an arrest point close to the time of Cdc48p relocalization. cdc1-1 and cdc19-1 mutants arrest as unbudded cells (some cdc1-1 cells terminate with a tiny bud). In contrast to most cell division cycle mutants, both cdc1 and cdc19 mutant cells do not enlarge after the termination of cell cycle development (Hartwell et al., 1973 ). After 3 h at the restrictive temperature (37°C), 85% of the cdc1 cells show most Cdc48p to be outside the nucleus (Figure (Figure1D).1D). When the temperature is shifted back to 28°C, Cdc48p accumulates in the nucleus of 80% of the cells within 2 h (Figure (Figure1E),1E), but the cells do not continue to proliferate.
In 95% of the cdc19 mutant cells arrested at 37°C for 4.5 h, a strong stain of Cdc48p appears in the nucleus (not shown).
Cdc48p contains a sequence near the amino terminus corresponding almost perfectly to a bipartite nuclear localization consensus and with high similarity to the SV40 large T antigen nuclear localization signal (Table (Table2).2). To investigate whether this sequence is responsible for the nuclear localization of Cdc48p, three of the five basic amino acid residues of the second part of the motive were changed to nonbasic residues (allele cdc48nonuc). When expressed from a GAL1 promoter in a cdc48::URA3-disrupted segregant, the mutant gene only allows germination if it is overexpressed (galactose medium), while a GAL1-regulated wild-type allele allows germination even under repressing conditions (glucose medium). On glucose media, the mutant strain grows more slowly than the wild type. Immunofluorescence localization with anti-Cdc48p antiserum shows a strong cytoplasmic staining and no detectable nuclear staining in 90% of the cells (Figure (Figure2A).2A).
The cell cycle-dependent relocalization of Cdc48p requires a regulation by a cell cycle-related signal. The inducible tyrosine phosphorylation of VCP at the tyrosine residue near the carboxy terminus observed by Egerton et al. (1992) and Egerton and Samelson (1994) is a potential molecular switch for the relocalization. To investigate the effect of the corresponding residue in Cdc48p, we mutated the tyrosine residue to a phenylalanine residue (cdc48Y834F) to simulate the nonphosphorylated form, and to a glutamic acid residue (cdc48Y834E) introducing a negative charge to mimic a phosphorylated tyrosine residue. When expressed from a GAL1 promoter in a cdc48::URA3 disruptant, both mutant genes restore spore germination and growth, but while the cdc48Y834E strain grows as fast as a wild-type strain, both on galactose- and on glucose-containing media, the cdc48Y834F strain has a doubled generation period (176 min vs. 90 min) on glucose media. Immunofluorescence localization shows Cdc48pY834F to exclude the nucleus of 90% of the cells (Figure (Figure2B),2B), while Cdc48pY834E is strongly concentrated in the nuclei of all cells (Figure (Figure2C).2C). After subcellular fractionation most of Cdc48pY834E is found in the nuclear fraction, and most of Cdc48pY834F is found in the microsome fraction (not shown).
The carboxy-terminal part of Cdc48p contains a stretch of acidic residues (residues 828–832) close to the site of tyrosine phosphorylation (E828EDDDLYS*). To determine its role in nuclear localization, it was deleted in the cdc48Y834F construct, resulting in the double-mutated allele cdc48Y834F/noacid. In contrast to the cdc48Y834F single mutant, the double mutant grows like wild type even on glucose medium. The Cdc48Y834F/noacid protein is visible in similar staining intensity in the cytosol and in the nucleus of 95% of the cells (Figure (Figure22D).
Wild-type cells and the Cdc48p tyrosine834 mutants were investigated for tyrosine phosphorylation by Western blotting with specific antibodies (Figure (Figure3A).3A). Only two of the four antibodies tested gave a signal with stimulated Jurcat cells used as a positive control. The two positively tested antibodies were used for the following experiments and always gave similar results. Exponentially grown wild-type cells show a strong phosphotyrosine band with the same electrophoretic mobility as Cdc48p. The band is by far the strongest in the molecular mass range of 50–200 kDa. The band completely vanishes if the cell extract is preincubated with phosphotyrosine phosphatase, excluding an unspecific reaction of the antisera. The signal is barely detectable in stationary cells. No phosphotyrosine signal is detectable in extracts of the cdc48Y834F mutant strain, while the cdc48Y834E mutant shows a weak signal with the anti-phosphotyrosine antibody, approximately 1 order of magnitude less intense than the wild type. When Cdc48p is tagged with the 1.5-kDa Flu epitope, the band is shifted to higher molecular mass (Figure (Figure3B),3B), proving the identity of the phosphotyrosinylated protein.
In subcellular fractions of wild-type cells, the nuclear fraction shows the strongest signal with anti-phosphotyrosine antibodies, but a weak signal is also detectable in the microsomal fraction. In subcellular fractions of the cdc48Y834E mutant strain, only the nuclear fraction gives a signal with anti-phosphotyrosine antibodies (not shown).
In synchronized cdc20 cells, a tyrosine-phosphorylated Cdc48p is only detected in cells harvested 60–75 min after release from the cell cycle arrest, corresponding to the phase of nuclear localization of Cdc48p (Figure (Figure33C).
To investigate whether the effect of the tyrosine phosphorylation on nuclear targeting works via a direct interaction of the carboxy-terminal region containing the phosphorylation site with the nuclear localization sequence close to the amino terminus, the two-hybrid system was used. Fragments of the CDC48 gene were fused to the DNA-binding and the activating domain of the GAL4 transcription activator and coexpressed in yeast strain SFY526, and the induction of a GAL1-regulated lacZ gene was monitored by β-galactosidase filter assay (Table (Table3).3). No interaction of the carboxy-terminal fragment of 153 amino acid residues with either the 56- or 291-residue amino-terminal fragment was observed. As a side effect, we found that the amino-terminal 56 amino acid residues fused to the DNA-binding domain alone activate the expression of the indicator gene; this effect is also present, although weaker, when the amino-terminal 291 aa residues of Cdc48p are fused to the DNA-binding domain.
However, both the amino-terminal and the carboxy-terminal fragment show interaction with the central part of Cdc48p (residues 173–578). The interaction of the carboxy-terminal part depends on amino acid 834: if the tyrosine residue is replaced by a phenylalanine, the interaction is significantly stronger, while no interaction is detectable if the residue has been changed to a glutamic acid.
Intracellular distribution of VCP, the mammalian Cdc48p homologue, was studied in the WISH cell line (human amnion derived, HeLa markers present) by immunofluorescence microscopy. WISH cells grow in adherent monolayers and have not lost contact inhibition; therefore cell division can be observed in a fairly physiological system.
During the interphase, cells are flattened, forming attenuated pseudopodia-like extensions. After bisbenzimide staining, the nucleus is clearly visible as a lentiform disk. A ring of small granules surrounding the nucleus is distinguishable from the cellular background in immunofluorescence with anti-VCP antiserum (Figure (Figure4A).4A).
As soon as the cells enter mitosis (detectable by a marked rounding of the cells and condensation of DNA into chromosomes), the distribution of the VCP signal changes rapidly. The intensity of the cytoplasmic staining increases. A bright spot, larger than the cytoplasmic granules, appears on each side of the metaphase plate (Figure (Figure4,4, B and C). Simultaneous staining against tubulin indicated that these spots correspond to the location of the centrosomes (not shown). VCP remains at the centrosomes during anaphase (Figure (Figure4D)4D) and cytokinesis. The signal begins to fade during telophase, leaving only the circumnuclear granular distribution characteristic of interphase cells. Similar patterns are observable in cultivated rat hepatocytes (not shown).
Affinity-purified and untreated sera gave identical staining patterns. Neither of the VCP patterns described above were observed in control experiments when preimmune sera were used instead of anti-VCP sera (not shown). The complete absence of a centrosomal staining in interphase cells should exclude potential fixation artifacts.
An immunofluorescence investigation of cultivated WISH cells with anti-phosphotyrosine antibodies showed a very similar pattern to that of anti-VCP antisera (Figure (Figure4).4). In mitotic cells the centrosomal region showed the most intense signal with both antibodies in a comparable intensity. The cytosolic granular staining was relatively weaker with the anti-phosphotyrosine antibody than with the anti-VCP antibody. Some cell–cell contacts stained with anti-phosphotyrosine antibody but not with anti-VCP antiserum (Figure (Figure44C).
WISH cells synchronized by thymidine treatment were investigated for phosphotyrosinylated proteins in a Western blot (Figure (Figure5B).5B). A band with the mobility of VCP is visible in extracts harvested 9–10 h after release of the thymidine block, corresponding to the period of the highest mitotic activity (25–30% cells in mitosis, Figure Figure5A).5A).
Both the CDC48 protein in yeast and its homologue VCP in mammalian cells are relocalized during the course of the cell cycle. In both types of cells, the proteins are located at vesicular structures within the cytoplasm, most probably the ER, during most of the cell cycle. However, in human WISH cells and rat hepatocytes, VCP is concentrated at the centrosomes during the entire mitosis, while in yeast cells Cdc48p accumulates within the nucleus for a short period at START. The temporary concentration of VCP at the centrosomes, in addition to the defect in spindle pole body duplication of arrested cdc48 mutants, suggests a role of VCP/Cdc48p in the proliferation of the spindle-organizing center. The divergent behavior of the highly similar proteins may be due to the different localization of their respective targets. While the centrosomes are freely accessible in the cytosol of mammalian cells, the yeast spindle pole is integrated in the nuclear envelope, which remains intact during the whole cell cycle. We assume that Cdc48p can only reach its target site at the spindle pole body from within the nucleus. The accumulation of the protein within the nucleus may eliminate the need for a specific targeting of Cdc48p to the spindle pole body. A similar behavior has been described for Dsk2p, which is involved in spindle pole duplication and is found inside the nucleus, but not attached to the spindle pole body (Biggins et al., 1996 ). The defined moments of relocalization during the cell cycle are in accordance with the role for Cdc48p and VCP proposed by us: the centrosome is duplicated during mitosis, while the spindle pole body is duplicated in late G1.
Our results suggest two cellular functions of both Cdc48p and VCP: an essential role in the cell division cycle, probably at the spindle pole body respectively the centrosome, and a function in the homotypic fusion of the ER. The two functions do not necessarily imply diverse molecular activities. It seems at least as likely that the difference lies primarily in the targets, while the molecular processes may be similar.
The amino-terminal region of Cdc48p contains a functional nuclear targeting consensus sequence as we demonstrated by site-directed mutagenesis. Elimination of the sequence not only prevents nuclear import, but makes a high accumulation of the mutated protein necessary for cell growth. Cell cycle progression apparently requires the presence of Cdc48p in the nucleus. On overexpression of the mutated gene, a small amount of the protein sufficient for its cell cycle function probably enters the nucleus, perhaps due to a marginal nuclear targeting activity of either the few basic residues of the mutated targeting signal or of another stretch of basic amino acid residues (residues 69–74). VCP lacks a nuclear targeting sequence in the corresponding region (see Table Table2),2), presumably because the centrosomes have become directly accessible during evolution, making the nuclear import unnecessary.
The relocation of both VCP and Cdc48p at a specific point in the cell cycle requires a regulator. A good candidate is the phosphorylation of a tyrosine residue at the penultimate carboxy-terminal position of VCP, which is triggered by stimulation of the T cell antigen receptor in cultivated murine T cells (Egerton et al., 1992 ; Egerton and Samelson, 1994 ).
We demonstrate that Cdc48p is tyrosine phosphorylated depending on the cell cycle phase. Phosphorylation occurs during the same period as nuclear localization of Cdc48p (late G1). A mutation of the tyrosine residue of Cdc48p near the carboxy terminus to phenylalanine, mimicking a nonphosphorylated tyrosine residue, abolishes tyrosine phosphorylation. The mutation has similar effects as the exchange of the nuclear targeting sequence: loss of nuclear import and growth defects when the gene is only expressed at low levels. In contrast, a mutation of the tyrosine residue to glutamic acid, which is the best imitation of a phosphotyrosine possible with the proteinogenic amino acids, results in a concentration of the protein in the nucleus. The abnormal morphology of the nucleus in the cdc48Y834E strain may be an effect of a lack of Cdc48p at the endoplasmic reticulum, resulting either in too little homotypic fusion activity, which would otherwise “clean up the odds and ends,” or in a defective protein coating of the endoplasmic reticulum membrane lacking the (otherwise rather abundant) Cdc48p. This morphological abnormality seems to have little effect on cell proliferation. The generation time is the same as that of wild-type cells.
In contrast to mammals, S. cerevisiae lacks dedicated tyrosine kinases but contains several “dual specificity” protein kinases which in vitro phosphorylate serine, threonine, and tyrosine residues (Hunter and Plowman, 1997 ). In vivo, tyrosine phosphorylation of S. cerevisiae has been shown in few examples, in addition to the autophosphorylation of some protein kinases (e.g., Spk1p [Stern et al., 1991 ] or Mck1p [Lim et al., 1993 ]) only for Cdc28p (Booher et al., 1993 ), mitogen-activated protein kinases (Errede et al., 1993 ), and immunophilin (Wilson et al., 1995 ). Thus, Cdc48p is the second example of a physiologically tyrosine phosphorylated S. cerevisiae protein other than a protein kinase, and, as for Cdc28p and the mitogen-activated protein kinases, tyrosine phosphorylation is phylogenetically conserved between yeast and vertebrates.
The regulation of nuclear import by a protein phosphorylation is a well established phenomenon (see Vandromme et al., 1996 , for review). However, in most cases described, the phosphorylation site (at either serine or threonine residues) is in the vicinity of the nuclear targeting sequence, and the effect of the phosphorylation is a prevention of nuclear import (Moll et al., 1991 ; Sidorova et al., 1995 ; Tagawa et al., 1995 ). The probable explanation of this effect is a neutralization or partial shielding of the positive charges of the nuclear targeting consensus by the negative charges of the phosphoric acid groups. In the case of Cdc48p, the phosphorylated site is a tyrosine residue, it is at the other end of the protein chain, and the effect of the phosphorylation is an induction of nuclear transport. Assuming a principally similar mechanism for the inactivation of the nuclear targeting, the stretch of acidic residues near the phosphorylation site might be able to mask the charge of the nuclear targeting sequence, provided that both elements are adjacent in the folded protein. A deletion of the acidic residues does indeed restore nuclear transport even in Cdc48pY834F. Nuclear transport of the double mutated protein is incomplete, indicating that the deletion has additional effects besides unmasking the nuclear targeting sequence, e.g., resulting in some misfolded protein accumulating in the cytosol. Two-hybrid experiments do not show a direct affinity of the parts of Cdc48p containing the nuclear targeting signal and the carboxy-terminal region, but both regions interact with the central part of the protein. The interaction of the carboxy terminus is dependent on residue 834: in the wild-type form, it shows little interaction with the central part of the protein; in case of the nonphosphorylatable phenylalanine, the interaction is stronger, while no interaction is found if residue 834 is a glutamic acid, mimicking a phosphotyrosine. We suggest that in the nonphosphorylated form of Cdc48p, attachment of the region of the nuclear targeting signal and of the carboxy terminus to the core of the protein brings them close to one another, resulting in a masking of the targeting sequence by the stretch of acidic residues. Tyrosine phosphorylation may result in a conformational change releasing the carboxy terminus and exposing the targeting sequence (Figure (Figure6).6). While direct data about the three-dimensional structure of Cdc48p is not available, the idea of a conformational change induced by the phosphorylation is supported by the observation that the phenylalanine mutant form shows an additional band in the protein degradation pattern compared with the wild-type and glutamic acid forms of Cdc48p (not shown), probably due to exposition of an otherwise protected target for proteolysis.
Our observation that some tyrosine phosphorylation can be detected in Cdc48pY834E, but not in Cdc48pY834F, is another indication of a structural difference between the phosphorylated and nonphosphorylated form of Cdc48p. Egerton and Samelson (1994) found some tyrosine phosphorylation of VCP at residues other than tyrosine 804. At least for Cdc48p, these additional phosphorylation events apparently only take place if residue 834 carries a negative charge, indicating an exposition of previously hidden residues due to an altered conformation.
It could be argued that the attachment of Cdc48p to cytoplasmic membranes might prevent its nuclear import, and that phosphorylation of tyrosine834 or its exchange to a glutamic acid residue might result in a release from the membranes. However, subcellular fractionation of a Cdc48pY834E strain shows that membrane attachment of the protein is not affected. While 80% of Cdc48pY834E is found in the nuclear fraction, only traces are found in the 170,000 × g supernatant while 20% of Cdc48pY834E is attached to the microsomes.
If tyrosine phosphorylation of Cdc48p is assumed to be the signal for nuclear targeting, the question arises what is the function of the tyrosine phosphorylation in mammalian VCP. Probably, the signal performs the corresponding role, directing the protein to the centrosomes. The colocalization of VCP with the antiphosphotyrosine signal in mitotic WISH cells could be interpreted as an indication, although it does not necessarily mean that both antibodies recognize the same protein. However, while components of both the centrosome (Wickramasinghe and Albertini, 1992 ) and the spindle pole body (Donaldson and Kilmartin, 1996 ; Friedman et al., 1996 ) are phosphorylated in a cell cycle-dependent manner, no tyrosine-phosphorylated protein has yet been described in these organelles.
VCP is the major tyrosine-phosphorylated protein in proliferating cells (Egerton et al., 1992 ). A Western blot of synchronized WISH cells indicates that the tyrosine-phosphorylated form of VCP is almost restricted to mitosis, which supports the idea that phosphorylation of VCP and its centrosomal localization might be linked.
Different anti-phosphotyrosine antibodies seem to recognize only a fraction of tyrosine-phosphorylated proteins, probably due to effects of neighboring residues, which makes them more protein specific. This additional specificity or the high abundance of Cdc48p may also be the reason why we have not detected any other tyrosine-phosphorylated protein in our Western blotting of yeast.
We thank Martin Latterich for stimulating discussions. Martin Sauerbeck and Peter Bohley generously provided us with cultivated WISH cells, and Rolf Gebhardt provided cultivated rat hepatocytes. We thank Michael J. Brownstein for sending us the porcine VCP gene and monoclonal antibodies against valosin. We are grateful to Stephan Sigrist for his donation of phosphotyrosine antibodies and for his help in printing the figures, to Wolfgang Hilt for the opportunity to use his micromanipulator, and to John Gatfield and Harold Taylor for critical reading of the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie.