Prp2o, the Saccharomyces cerevisiae homolog of the regulator of chromosome condensation, rcc1, interacts with double-stranded dna through a multi-component complex containing gtp-binding proteins

A ubiquitous and essential eukaryotic protein generally referred to as RCC1 (Regulator of Chromosome Condensation) is involved in the maintenance of interphase chromatin conformation and/or the position of the chromosomes in the nucleus and acts as a negative regulator of chromosome condensation (Enoch and Nurse, 1991; Nishimoto et al., 1978, 1981; Nishitani et al., 1991; Ohtsubo et al., 1987, 1989, 1991). The RCC1 family of proteins is characterized by a common motif of approximately 60 amino acids, which is repeated about seven times throughout the protein (Aebi et al., 1990; Ohtsubo et al., 1987). PRP20/SRM1, the RCC1 homolog in the budding yeast Saccharomyces cerevisiae has been shown to complement a ts mutant for RCC1 in mammalian cells (Ohtsubo et al., 1991). Furthermore, the human RCC1 gene restores the viability of the yeast ts mutant (Fleischmann et al., 1991). Other common features shared by the RCC1 family are: a functional location in the nucleus (Fleischmann et al., 1991; Ohtsubo et al., 1989); an association with a RAS-related GTP-binding protein, referred to in mammalian cells as Ran/TC4 (Belhumeur et al., 1993; Bischoff and Ponstingl, 1991b; Drivas et al., 1990); and an affinity for dsDNA resulting from the inter-action of the RCC1 with Ran/TC4-GTP (Bischoff et al., 1990; this report).

Morphological studies of the S. cerevisiae ts mutant allele, p rp 2 0 - 1, demonstrate that the budding yeast RCC1 homolog has a direct involvement in maintaining a functional organization of the nucleoplasm (Aebi et al., 1990). The Ran/TC4 homologs in S. cerevisiae encoded by the GSP 1 and GSP2 genes were also found to be required for the nuclear organizational activities of PRP2 0 (Belhumeur et al., 1993). Unlike more complex eukaryotes, S. cere - v i s i a e does not undertake the conventional structural changes observed during mitosis. Consequently, it becomes difficult to access the role Prp20 plays in the negative regulation of chromosome condensation. On the contrary, in the fission yeast Schizosaccharomyces pombe, where a more conventional and observable chromosome condensation occurs, the RCC1 and Ran/TC4 homologs (PIM1 and SPI1, respectively) were isolated as mutants on the basis of their ability to enter mitosis despite inhibition of DNA synthesis by hydroxyurea (Matsumoto and Beach, 1991). This S. pombe phenotype is similar to that found in tsBN2, a ts mutant of BHK cells, that contains a point mutation in the gene for the RCC1 protein (Kai et al., 1986; Uchida et al., 1990). From the S. pombe and mammalian mutant results it was proposed that RCC1 protein is somehow involved in monitoring and signalling the completion of DNA replication (Enoch and Nurse, 1990, 1991; Hartwell and Weinert, 1989). Recent evidence, however, indicates that RCC1 holds a more complex role than would be accountable if it were solely involved in a signalling mechanism (for review, see Dasso, 1993). Other mutant alleles of RCC1 subsequently isolated in both budding and fission yeast show effects on a wide variety of nuclear activities. In S. cerevisiae, ts mutants of the RCC1 homolog have been shown to affect RNA metabolism and nuclear morphology, PRP2 0 (Aebi et al., 1990; Forrester et al., 1992); pheromone response pathways, SRM1 (Clark and Sprague Jr, 1989); and mRNA transport, MTR 1 (Kadowaki and Tartakoff, 1992) and PRP2 0 (Aebi et al., 1990; Forrester et al., 1992). In S. pombe, ts mutations in the RCC1 homolog influence both chromosome condensation (PIM 1; Matsumoto and Beach, 1991), and chromosome decondensation (D CD 1; Sazer and Nurse, 1992). The pleiotropic effects on nuclear activities by the various yeast RCC1 mutants support a role for RCC1 in the maintenance of a proper nuclear organization. In such a model it is assumed that specific spatial arrangements of the nuclear components are required in the accurate completion of essential nuclear processes. Consequently, the pleiotropic nature of the RCC1 mutants could be explained if mutations in the protein disrupts the proper nucleoplasmic organization. To fully understand the multiple activities of RCC1 in the nucleus and during the cell cycle, it will be necessary to determine the biochemical interactions that the RCC1 protein has with the chromatin and the nuclear skeleton by more conventional means.

To investigate directly the multiple functions for Prp20, we first examined the dsDNA-binding ability of the protein. We have used an in vitro column-binding assay to examine the cellular requirements for Prp20 dsDNA binding. In extracts made from asynchronously grown wild-type strains, we found that Prp20 is readily retained on a dsDNA- cellulose column. We also found that the single base mutation in the prp20-1 mutant protein, which changes a glycine residue located near the carboxyl terminus to a glutamic acid, permanently eliminated the ability of the mutant protein to bind to dsDNA. Furthermore, yeast Prp20 protein produced in E. coli requires additional yeast proteins present in extracts of the prp20-1 strain to interact with dsDNA. Immunoprecipitation with an antibody directed against Prp20 revealed at least 6 other proteins that co-pre- cipitate with Prp20. Our conclusion that a multi-component complex is involved in the binding of Prp20 to dsDNA is further upheld by sizing-column-chromatography profiles of the eluents from the dsDNA column in which Prp20 (52 kDa) is eluted in a fraction of size range greater than 150 kDa. We also demonstrated that one of the proteins in the Prp20 complex is a GTP-binding protein of 23 kDa and have identified it as GSP1, the S. cerevisiae homolog of the human GTP-binding protein, Ran/TC4 (Belhumeur et al., 1993; Drivas et al., 1990). In Xenopus oocyte nuclear fusion assays it has been shown that RCC1 is required during the initiation phases of DNA replication (Dasso et al., 1992). We therefore examined the dsDNA-binding abilities of the Prp20 complex in arrested extracts from a set of the cell cycle division ts mutants that are inhibited at different points in the cell cycle. We found that only in cells arrested in the DNA replication phase of the cell cycle did the Prp20 complex lose its dsDNA binding character.

The following yeast strains were used in this study: EJ101 (MATα his1, prb1-1122, prc1-126), prp20/2C (MATa, prp20-1, ade2-101, his3-200, lys2-801, ura3-52, a temperature-sensitive mutant allele of PRP20) (Aebi et al., 1990), SS330 (MATa, his3-200, tyr1.ade2- 101, ura3-52, gal suc2, the parent strain of prp20/2C). PRP20/prp20/2C contains the wild-type full-length PRP20 gene in YEp352 vector transformed into prp20/2C. GSP1/prp20/2C and GSP2/prp20/2C contain GSP1 and GSP2 in the YEp352 vector transformed into prp20/2C, respectively (Belhumeur et al., 1993). Yeast cells were grown in YPD (1% yeast extract, 2% peptone, 2% glucose) or synthetic media as described by Sherman et al. (1986). All amino acids were added at the final concentration of 0.002%. Yeast cells were transformed by using the lithium acetate method (Ito et al., 1983). The cdc mutants used in this study are described by Hartwell (1973). PET-3a-PRP20 was provided by M. Aebi and was transformed into BL21(DE3) (Fleischmann et al., 1991; Studier and Moffat, 1986).

Total yeast extract was prepared by the modification of the method of Jong et al. (1985). Wild-type yeast strains were grown in YPD medium to mid-log phase. Cells were pelleted by centrifugation and then washed 2× in 1 × PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.4). The cells were then lysed by glass bead homogenization in the presence of 5 ml Buffer A (50 mM Tris-HCl, pH 8, 10% glycerol, 0.5 mM dithiothreitol (DTT), 1 mM Na₂-EDTA, and 1 mM phenylmethylsulfonyl fluoride (PMSF)) plus 1 mM Na₂-EGTA, 0.5 mM bacitracin, 1 mM benzamidine, and 1 M NaCl. The lysate was then centrifuged for 45 minutes at 8,000 rpm in the Sorvall GSA rotor to remove cell debris. Polyethyleneglycol 8000 in 2 M NaCl was added to the supernatant to a final concentration of 6% (w/v). It was kept on ice for 30 minutes after which time the nucleic acid precipitate was removed by centrifugation in the Sorvall GSA rotor for 30 minutes at 8,000 rpm. The resulting supernatant was dialysed extensively against Buffer A and equilibrated in Buffer A + 0.1 M NaCl.

To prepare extracts from the temperature-sensitive mutant allele of PRP20, prp20-1, and from the prp20-1 ts suppressor strains, GSP1 and GSP2, the cells were grown in selective media supplemented with the appropriate amino acids at room temperature and at 34°C, respectively. The prp20-1 extract was prepared in 0.3 M NaCl instead of 1 M NaCl.

The dsDNA-cellulose column was prepared with 2 ml of condensed bed volume of calf thymus dsDNA-cellulose (Sigma) packed into a 5 ml poly-prep chromatography column (Bio-Rad) and equilibrated with Buffer A + 0.1 M NaCl containing benzamidine, bacitracin and PMSF. Then 10 ml of the prepared cell extract were passed through the column at a rate of 0.2 ml/min- utes and washed with 20 ml of the same buffer. The proteins bound to the column were step-eluted with increasing concentrations of NaCl. The proteins in the collected fractions were precipitated with 7 volumes of acetone at -20°C.

The PRP20 gene in a bacterial expression vector, PET-3a-PRP20, which placed the gene under the T7 RNA polymerase inducible promoter was provided by M. Aebi, and transformed into the E. coli system, BL21(DE3). A bacterial culture bearing PET-3a- PRP20 was grown to mid-log phase, and the synthesis of the Prp20 protein was induced with 2 mM isopropyl-β-D-thiogalactopyra- noside (IPTG) for 3 hours. The Prp20 protein was present in inclusion bodies in E. coli. The extraction of the inclusion bodies containing Prp20 is based on the method by Nagai and Thogersen (1987). The inclusion body was then solubilized in 8 M urea in 1× PBS. The proteins were precipitated with 4 volumes of acetone on ice. The precipitates were separated by SDS-polyacry- lamide gel electrophoresis (PAGE) (Laemmli, 1970), and the Prp20 protein band was excised from the gel. The Prp20 was electro-eluted from the gel slice, and then used to immunize rabbits.

To prepare E. coli-expressed yeast Prp20 for in vitro dsDNA- binding assay, the inclusion bodies were resuspended in 8 M urea in Buffer A containing proteinase inhibitors and dialysed against Buffer A to remove urea. This protein preparation was not subjected to an acetone precipitation. The Prp20 protein preparation was then equilibrated to 0.1 M NaCl.

The protein concentrations of the cell extracts and purified samples were determined using the Bio-Rad protein assay kit with bovine serum albumin as a standard. The proteins were separated by 12% SDS-PAGE, transferred to the nitrocellulose in a semidry transfer cell (Bio-Rad) in transfer buffer (25 mM Tris-HCl, 192 mM glycine, pH 8.3, and 20% methanol), and incubated for 30 minutes in blocking solution (0.2% gelatin, 1% BSA, 1% goat serum). The blot was then incubated overnight with either Prp20 or Gsp1 antibodies at 1:1000 dilution in TPBS (0.05% Tween-20, 0.15 M NaCl, 0.01 M Na-phosphate, pH 6.8), washed and reacted with biotin-sp-conjugated affinity-pure goat anti-rabbit immunoglobulin G (Jackson ImmunoResearch, West Grove, PA) at 1:1000 dilution for 30 minutes at room temperature. The immunoperoxidase method was used to detect the antibody. Prestained protein molecular mass standards (BRL) were used for the SDS-PAGE. The preparation of the antibody to Gsp1 is described by Belhumeur et al. (1993).

The yeast cells (EJ101) were grown to saturation (A₆₀₀ = 5) in minimal media containing 1 mM NH₄SO₄, 5% glucose, and supplemented with the appropriate amino acids. The cells were washed and resuspended to an A₆₀₀ of 0.02 in minimal media with 0.1 mM NH₄SO₄. At mid-log phase, the cells were resuspended in a small volume of fresh media without NH₄SO₄ at the concentration of 2 A₆₀₀ of cells per ml. The cells were labelled for 3 hours in the presence of 250 μCi of H₂³⁵SO₄ (ICN, 43 Ci/mg S) at room temperature (Franzusoff et al., 1991). At the end of 3 hours, the labelled cells were washed twice in 1× PBS, and the lysate was prepared as described before for dsDNA-cellulose binding. The eluent from the dsDNA-cellulose column was dialysed to 0.1 M NaCl followed by immunoprecipitation with 5 μl of anti- Prp20 antibody.

GTP-binding proteins in yeast cell extracts were detected based on the methods derived from various sources (Lapetina and Reep, 1987; McGrath et al., 1984; Schmitt et al., 1986). As described earlier in the method of immunoblotting, the proteins were separated by 12% SDS-PAGE, and then transferred immediately to nitrocellulose paper without further treatment of the gel. In order to detect GTP-binding proteins, the blot was incubated for one hour in the blocking solution containing 1% gelatin, 0.1% Tween- 20, and 1× PBS at room temperature (Serafini et al., 1991). The blot was then rinsed briefly in the binding buffer (50 mM Tris- HCl, pH 7.5, 0.3% Tween-20, 5 mM MgCl₂, and 1 mM Na₂- EGTA). Binding of GTP was assayed with [α-³²P]GTP (3000 Ci/mmol, 1 μCi/ml of binding buffer) for 1 hour at room temperature (Rubins et al., 1990). After binding, the blot was rinsed several times with the binding buffer for 1-2 hours, it was then exposed to Kodak X-(OMAT)AR film with an intensifying screen at −80°C.

The yeast extract prepared as described above was passed through the dsDNA-cellulose column. After washing the column with 100 ml of binding buffer, the bound proteins were eluted with 0.8 M NaCl from the column and equilibrated to 0.1 M NaCl. This protein pool was concentrated by aquacide (Cal-Biochem) to a final volume of 0.2 to 0.25 ml, followed by filtration through 0.22 μm filter. The protein was then applied to a gel filtration column (1.5 cm × 30 cm) of Superose 12 HR (Pharmacia), which was preequilibrated with 0.1 M NaCl in buffer A, and then fractionated with the same buffer at a flow rate of 0.2 ml/minutes. The effluent fractions were monitored continuously at 280 nm and analyzed by SDS-PAGE and immunoblotting. The total protein was visualized by silver staining (Clark, 1991). Molecular mass standards for gel filtration were obtained from Bio-Rad and consist of the following proteins: thyroglobulin 670 kDa, bovine gamma globulin 158 kDa, chicken ovalbumin 44 kDa, equine myoglobin 17 kDa, vitamin B-12 1.35 kDa.

Early log phase cultures of cdc mutants 4, 8, and 13 were grown at 25°C and then shifted to 37°C until greater than 80% of the arrested population exhibited the terminal phenotype. To determine the placement of the nucleus, cells were fixed in 3.8% formaldehyde and incubated with Hoechst 33258 (0.5 μg/ml) for 30 minutes. The cells were then mounted in 90% glycerol and viewed under Nomarski DIC and UV illumination on a Zeiss Axioplan microscope. Indirect immunofluorescence staining was performed as described (Clark and Abelson, 1987) with the following modifications. The cells were fixed for 30 minutes and the rabbit anti-Prp20 antibody was used at a 1:300 dilution. Anti- Prp20 antibody was visualized using a donkey anti-rabbit IgG conjugated with Texas Red (Jackson ImmunoResearch). Silver staining of the nucleolus was done as described previously (Clark, 1991).

To verify that Prp20 has the ability to bind dsDNA in vitro, total yeast extract from an asynchronous culture of a wildtype yeast strain was prepared as described in Materials and Methods. The extract was passed through the dsDNA-cel- lulose column and the proteins that did not bind to the column were collected (flow-through). The column was then washed with 10 bed volumes (20 ml) of the binding buffer (wash fraction), and the proteins still retained by the dsDNA-cellulose were eluted with increasing NaCl concentrations from 0.2 M to 1.0 M. Proteins in each fraction were acetone-precipitated and separated by 12% SDS- PAGE. The immunoblot in Fig. 1 uses an antibody against Prp20 to reveal the column fractions containing Prp20. The majority of the Prp20 retained on the dsDNA column was eluted at 0.2 M NaCl and the remaining proteins were eluted from the column at 0.4 M NaCl. As a control, an identically prepared yeast extract was passed through a column containing only unmodified cellulose. Prp20 did not bind to this column (data not shown). The observed affinity of Prp20 to dsDNA is consistent with that of the mammalian RCC1 protein, which also eluted from the dsDNA column between 0.2 and 0.4 M NaCl. Lane Prp20 of Fig. 1 contains yeast Prp20 produced in E. coli, which was used as a positive marker for the SDS-PAGE. This heterologously expressed Prp20 showed an altered mobility on SDS-PAGE when compared with the Prp20 from yeast extract. Although the reason for this difference in mobility is not clear, it has been reported previously (Fleischmann et al., 1991). Nevertheless, this protein is specifically recognized by the Prp20 antibody and is functional with respect to its ability to bind to dsDNA in the presence of the inactivated mutant extract (see below).

The temperature-sensitive conditional lethal phenotype of the PRP20 mutant allele, prp20-1, is the result of a single point mutation at residue 457 in prp20, which changes a glycine to a glutamic acid (Fleischmann et al., 1991; Vijayraghavan et al., 1989). This glycine residue is conserved among RCC1 homologs, suggesting an essential role for this residue in the structure and/or function of the protein. This point mutation in prp20 is somehow responsible for the pleiotropic effects seen in this mutant strain, but does the lethality of the prp20-1 mutant correlate with a loss of in vitro dsDNA-binding ability in the mutant protein? To address this question, total cell extract was prepared from the prp20-1 strain, prp20/2C, grown at permissive temperature. The extract from this mutant strain was divided in half and then loaded onto the dsDNA column. Just prior to the loading, the extracts were incubated for 5 minutes, one at 23°C (permissive temperature) and the other at 34°C (non-permissive temperature). As shown in Fig. 2B (lane b) the heat treatment of the mutant extract completely abolished the dsDNA-binding ability of prp20, whereas the prp20 in the mutant extract prepared under the permissive condition retained its ability to bind to dsDNA (lane a).

This loss of dsDNA-binding ability was not due to the complete degradation of prp20 since the mature protein was observed in the flow-through (see Fig. 3, lane 1, *). A similar loss of dsDNA binding by prp20 was also obtained from an extract that was heat inactivated for 5 minutes at 37°C or from an extract prepared from the prp20/2c strain grown for 1 hour at the non-permissive temperature (data not shown). The parent strain of prp20/2C, SS330, did not lose its dsDNA-binding ability after the heat treatment (Fig. 2A). Furthermore, treatment of the mutant extract with either high salt (1 M NaCl) or freezing and thawing of the extract also resulted in the loss of dsDNA-binding ability of prp20 (data not shown). This inability of the inactivated mutant extracts to bind dsDNA was irreversible. The extreme sensitivity of the mutant extracts to high salt concentrations made it necessary to prepare the mutant extract at low salt (0.3 M NaCl).

To examine the possibility that Prp20 alone is responsible for its dsDNA-binding ability, yeast Prp20 was produced in E. coli. The dsDNA-binding ability of E. coli-produced yeast Prp20 was then tested by passing it through the dsDNA column under conditions identical to the ones used for the yeast extracts (Fig. 3). No Prp20 was detected in the 0.8 M NaCl eluent (lane 6). All E. coli-produced yeast Prp20 loaded on to the column was observed in the flowthrough as detected by immunoblotting (lane 4). In order to emphasize that E. coli-produced Prp20 alone does not bind to dsDNA, a 20-fold excess of protein was loaded in lanes 4-6 compared with lanes 1-3. The Prp20 antibody used for these immunoblots was raised against the E. coli- produced yeast Prp20 preparation. Thus, because of the overloading in lanes 4-6, the E. coli protein impurities that reacted with the antiserum became obvious. On yeast extracts alone the antibody only revealed the single Prp20 band.

The inability of bacterially produced yeast Prp20 to bind dsDNA could be due to one of two reasons: first, the E. coli extract may lack some yeast-specific post-translational modifications required for dsDNA binding of Prp20, or alternatively, the binding of Prp20 to dsDNA may require interaction with other yeast components. To determine which of these explanations is more probable, a reconstitution assay was developed in which the bacterially produced Prp20 protein was combined with the irreversibly inactivated prp20-1 mutant extracts in 1 M NaCl and then dialysed extensively against Buffer A + 0.1 M NaCl at 4°C. This reconstituted extract was then passed through the dsDNA-cellulose column and eluted at 0.8 M NaCl. Although neither bacterially produced wild-type Prp20 (Fig. 3, lane 6) nor the mutant prp20 in the in vitro-inactivated extract (Fig. 2B, lane b) bound dsDNA, the combination of the two inactive samples re-established the dsDNA-binding ability of bacterially produced Prp20 (Fig. 3, lane 3). The difference in the SDS-PAGE mobilities between E. coli- produced (arrowhead) and S. cerevisiae-produced (*) protein further confirms our results: analysis of dsDNA column eluent revealed only one band corresponding to E. coli-pro- duced Prp20 (lane 3). The mutant prp20, which does not bind to dsDNA, can be seen in the flow-through fraction (lane 1, *) as distinguished by its faster mobility on the SDS-PAGE. Furthermore, to exclude the possibility of the dsDNA binding by multimers formed between these two inactive proteins, we repeated the experiment using a truncated version of E. coli-produced Prp20, in which 13 carboxyl-terminal amino acid residues have been removed by site-directed mutagenesis. In the immunoblot from the dsDNA eluent of this experiment, only the lower Prp20 band, which corresponds to the truncated form of Prp20, was detected (data not shown).

To return to the question of post-translational modifications, the amino acid sequence analysis of Prp20 shows the presence of phosphorylation consensus sites for protein kinase CDC2/28, protein kinase C, and cAMP-dependent protein kinase. At this point there is no biochemical evidence that Prp20 is a phosphoprotein and, if it is, whether the modification by phosphorylation is important for its dsDNA-binding ability. However, it is unlikely that the extract, which had been extensively dialysed prior to the addition of the bacterially produced Prp20 and then redialysed before loading onto the column, would contain a sufficient quantity of small molecules, such as ATP or GTP. Also, since all the procedures were done at 4°C, including the column binding, any modification enzymes would have to be functional at that low temperature. Although we cannot completely exclude the possibility of involvement of post-translational modifications, we think that a more likely explanation for this reconstitution result is that other yeast proteins are required to establish the stable binding of the Prp20 complex to dsDNA. As described above, the preparation of extracts of the prp20/2C under certain conditions (i.e. high salt, heat treatment) abolishes the dsDNA- binding ability of prp20 (see Fig. 2). Any ancillary proteins required by Prp20 should remain functional in this prp20- inactivated extract. It is the interaction of the bacterially produced wt Prp20 with the ancillary proteins from this yeast extract that restores the dsDNA-binding ability of bac- terially produced yeast protein. Such an explanation is likely, for two proteins of >150 kDa and 94 kDa have been reported to co-precipitate with Prp20 antibody (Fleischmann et al., 1991). Proteins of similar sizes were also found to co-immunoprecipitate with RCC1 (Ohtsubo et al., 1987).

If Prp20 is binding to the dsDNA column through an interaction with other proteins, such a complex should have a higher molecular mass than the 52 kDa of Prp20. Fractions from the dsDNA column that contain Prp20 were thus chromatographed on a gel-filtration molecular-sizing column consisting of the Superose (Pharmacia) matrix with a size range of 50,000-300,000 Da. To ensure that the integrity of any dsDNA-binding Prp20 complex was maintained after salt elution from the dsDNA column, eluted fractions containing Prp20 were pooled, dialysed against Buffer A + 0.1 M NaCl, and reloaded onto the dsDNA column. All of the reloaded Prp20 was retained on the dsDNA column (data not shown). Therefore, in preparation for the Superose column chromatography, total wild-type yeast extract was partially purified by passing it through the dsDNA-cellu- lose column, eluting the bound proteins at 0.8 M NaCl. The eluent was dialysed back to the low-salt-binding condition (0.1 M NaCl) and concentrated by aquacide to 0.25 ml for loading onto the FPLC. Then, 0.4 ml fractions were collected from the FPLC, and assayed by immunoblot (Fig. 4A). The FPLC chromatogram (Fig. 4B) revealed Prp20 in fractions #10-#14 (Fig. 4A). Furthermore, another protein, Gsp1, was also found in fractions #10-#12 (Fig. 4A). Prp20 and Gsp1 were both found in fractions #10 and #11, corresponding to a Mr of between 150,000 and 200,000. The arrows in Fig. 4B represent the positions of the molecular mass standards. The coincidentally migrating Gsp1 protein is thus a candidate for a component of the Prp20 complex. Furthermore, the dsDNA-chromatographic profile of Gsp1 protein coincided with that of Prp20 (data not shown): Gsp1 also eluted from a dsDNA-cellulose column at NaCl concentrations between 0.2 M and 0.4 M. Similar to the mutant prp20 in the prp20/2C strain, Gsp1 in prp20/2C also failed to bind to dsDNA-cellulose when the extract was heat-inactivated. However, when this mutant extract is reconstituted with bacterially produced Prp20, Gsp1 and bacterially produced Prp20 both bound to the dsDNA column as detected in immunoblot using anti-Gsp1 and anti-Prp20 antibodies, respectively (data not shown). These results and others (see below) strongly suggest that Gsp1 is a part of the Prp20 complex, even though the exact stoichiometry of these components have not yet been established. Prp20 and Gsp1 in fraction #12 and Prp20 in fraction #13 and #14 likely represent the dissociation of the complex during chromatography. The isolation of the other members of this complex is being pursued, but here we will further examine the potential for a direct interaction between Gsp1 and Prp20.

Two yeast genes have been isolated by their ability to suppress the mutant phenotype of prp20-1 at non-permissive temperature, and are designated as GSP1 and GSP2 (Bel- humeur et al., 1993). The overexpression of GSP1 and GSP2 rescues the lethal phenotype of prp20-1 at 34°C. If the loss of dsDNA-binding ability of the mutant prp20 protein is important in the conditional loss of viability of the mutant strain, then the GSP1 and GSP2 suppression of prp20-1 lethality should be reflected in the re-establishment of dsDNA binding of the mutant strain prp20. Strains of prp20/2C containing 2μ plasmids carrying the genes for either GSP1, GSP2, or wild-type PRP20 were grown at 34°C. The extracts were prepared from these strains and tested for dsDNA-binding ability of mutant prp20. Fig. 5 shows the binding of prp20 to the dsDNA column in all three strains tested (lane 7,8,9). The mutant strain, prp20/2C itself is not viable at 34°C. The mature prp20 detected in the immunoblot indicated a direct correlation between the overexpression of the suppressors and a restoration of the dsDNA-binding ability of prp20 in the mutant strain. We also observed another Prp20 band of a higher molecular mass in the PRP20/prp20/2C strain. A similar observation was reported by Fleischmann et al. (1991) that the overproduction of Prp20 causes a shift in its SDS-PAGE mobility. We are unable to explain why this occurs, but it is reproducible and it does not affect the dsDNA-binding ability of the modified Prp20. The control experiment with prp20/2c containing only the plasmid vector showed that the vector itself did not contribute to the restored dsDNA-binding ability of mutant prp20. Furthermore, the overproduction of Gsp1 and Gsp2 has no effect on the quantity of Prp20 translation products as judged by immunoblot analysis (data not shown). This excludes the possibility that Gsp1/Gsp2 may control the protein synthesis in prp20/2c.

The amino acid sequence analysis of Gsp1 and Gsp2 revealed over 80% homology to the protein of a human gene, TC4 (Belhumeur et al., 1993). We shall only mention GSP1 from here on for brevity since GSP1 and GSP2 are two separate genes that code for the same protein. TC4 was originally isolated from teratocarcinoma cells during a search for genes with RAS-like GTP-binding characteristics (Drivas et al., 1990). TC4 and GSP1 both contain the consensus GTP-binding sequences, thus Gsp1 was tested for its specific binding to [α-³²P]GTP on nitrocellulose (see Materials and Methods). The autoradiogram revealed the presence of several GTP-binding proteins including Gsp1 and the binding of these proteins to [α-³²P]GTP was shown to be GTP-specific. Only GTP could eliminate this binding, while even 100-fold excesses of the other nucleotides, dGTP, CTP, UTP and ATP, could not interfere with [α- ³²P]GTP binding of these proteins (data not shown). Fig. 6A shows the [α-³²P]GTP blot of the FPLC fraction #11 from Fig. 4A. In this blot we have detected two other GTP- binding proteins of 32 kDa and 17 kDa (*), besides Gsp1. We cannot state categorically that these proteins are required components of the Prp20 complex. However, these two GTP-binding proteins (32 kDa and 17 kDa), coimmunoprecipitate with Prp20 and Gsp1 from the 0.4 M NaCl eluent of the dsDNA-cellulose column using the Prp20 antibody (Fig. 6B). In addition, anti-Prp20 or anti- Gsp1 antibody co-immunoprecipitated Gsp1 or Prp20, respectively, as shown by the subsequent immunoblotting of the immunoprecipitated proteins (data not shown). A doublet of 45-46 kDa is also seen in the ³⁵S-labelled proteins, which co-immunoprecipitates with either anti-Prp20 or anti-Gsp1, but does not bind [α-³²P]GTP.

RCC1 protein is required for the replication of sperm chromatin added to Xenopus nuclear extracts (Dasso et al., 1992). It was reported that the RCC1 protein does not play an enzymatic role in DNA synthesis but is somehow involved in the establishment of the pre-DNA replication complex (Dasso et al., 1992). We thus examined the possibility of an alteration in the dsDNA-binding ability of Prp20 complex at the various points of the cell cycle. A set of cell-division cycle ts mutants was chosen in respect to their interference with the normal progression of the cell cycle, that is, at the non-permissive temperature they exhibit a stage-specific arrest: at G1 (cdc28), in the initiation of DNA replication (cdc4 and cdc7), during DNA replication (cdc8), at the onset of mitosis (cdc13) and at nuclear separation (cdc15). Extracts were prepared from both the arrested (A) and the non-arrested (NA) cdc mutant cells. The dsDNA-binding ability of Prp20 in the cdc mutant extracts was analyzed as described above. Prp20 protein bound to dsDNA in all arrested extracts except those specific for DNA replication: cdc4, cdc7 and cdc8. The results for cdc4 and cdc8 are shown along with cdc13 as a positive control. The dsDNA-binding ability of Prp20 was observed in all of the non-arrested (NA) extracts (Fig. 7A).

The nature of the alteration(s) that causes the Prp20 and/or the Prp20 complex to lose its dsDNA-binding ability is not known at the present time. However, the loss of the protein’s affinity for dsDNA is not due to degradation of Prp20, since the mature protein was observed in the cell extracts (Fig. 7A). Whether our results reflect the normal S phase activity of the cell in vivo remains to be answered. To monitor the percentage of arrest in the cdc strains used in this set of experiments, the morphological characteristics of the terminal phenotypes of these mutants were followed. Fig. 7B shows the morphology of the entire cell (top) and the position of the nucleus (bottom) in the arrested cdc4, cdc8, and cdc13 cells. See the legend in Fig. 7 for the percentages of arrest.

In this study, we have determined that the in vitro dsDNA- binding ability of the Prp20 complex is dependent on the integrity of the complex. Also, we have shown that the loss of dsDNA-binding ability by the mutant prp20 is directly related to the lethal phenotype observed in prp20/2c. In the prp20/2c strain, the fragmentation of the nucleolus was observed at the non-permissive temperature, indicative of a disruption in nucleoplasmic organization (Aebi et al., 1990). This result might suggest that the dsDNA-binding function of the Prp20 complex is directly related to the maintenance of the order of the nucleoplasm. Again using the integrity of the nucleolus as a marker for a proper nucleoplasmic arrangement, the nucleolar structure in these cells was observed by silver staining (NO-Ag) (Fig. 8B, indicated with the arrows). Unlike prp20/2c, the arrested cdc8 cells show no changes in the nucleolar structure. Furthermore, immunofluorescence staining of the arrested cdc8 cells with the anti-Prp20 antibody locates Prp20 to the nucleus (Fig. 8A). The mutant prp20 of the prp20-1 allele is shown to leave the nucleus upon temperature shift (Amberg et al., 1993). These data indicate that the loss of dsDNA-binding ability of Prp20 during DNA replication does not have a direct effect on the maintenance of the nucleoplasmic organization.

Electron microscopic characterization of the PRP20 mutant cells have shown that the organization of the nucleoplasm is disrupted within 15 minutes of shifting the cells to the non-permissive temperature (Aebi et al., 1990). This rapid disorganization of nucleoplasmic structure is lethal to the cell. In an attempt to understand the role of this RCC1 homolog in nuclear organization, as well as, any possible involvement in cell cycle control, we set out to examine the dsDNA-binding aspects of Prp20 using an in vitro dsDNA-binding assay.

First, we determined that Prp20, like RCC1, has the ability to bind to dsDNA (Fig. 1). We have also demonstrated that the binding ability of Prp20 to dsDNA is directly related to cell viability in prp20/2C, a PRP20 mutant strain (Fig. 2). The importance of the dsDNA-binding ability of Prp20 is further strengthened by the effects the genes that suppress the prp20-1 mutant, GSP1 and GSP2, have on the nucleoplasm. We have shown that the overexpression of Gsp1 and Gsp2 not only restores the dsDNA-binding ability of the mutant prp20 (this report), but the microscopic analysis of these cells also demonstrates that increasing the cellular copy number of these proteins prevents the disorganization of the nucleoplasm at non-permissive temperature (Belhumeur et al., 1993). These genetic data strongly suggest that a protein complex is essential for Prp20 function. From cellular mapping studies of wild-type Prp20 and Gsp1, both proteins have a predominantly nuclear location (Belhumeur et al., 1993; Fleischmann et al., 1991). This nuclear location is directly and specifically affected in the prp20-1 ts mutant; at the non-permissive temperature prp20 and Gsp1 are both found in the cytoplasm (Amberg et al., 1993; R. Tam and M. W. Clark, unpublished results). These results implicate the association between Prp20 and Gsp1 being required for the nuclear localization of these proteins. The existence of the Prp20 complex is further corroborated in this report, first, by the reconstitution assays using E. coli-produced Prp20 and inactive prp20-1 extract, and second, by immunoprecipitations and sizing-column-chromatography that show this complex to be composed of 67 proteins, one of which is Gsp1. Recent studies of a truncated version of mammalian RCC1 have reached a similar conclusion: that the essential chromatin binding of RCC1 is propagated not by the protein alone, but through a protein complex containing RCC1 (Seino et al., 1992). Also, in mammalian cells, a small molecular mass GTP-binding protein of 25 kDa (Ran) has been found to co-purify with RCC1 (Bischoff and Ponstingl, 1991a,b). These observations indicate that the association of RCC1 with other proteins is a ubiquitous characteristic essential for its nuclear functions and that a GTP-binding protein is required in those critical activities.

In vitro nuclear envelope assembly assays using extracts from Xenopus eggs reveal that GTP hydrolysis is required in the assembly processes (Boman et al., 1992; Newport and Dunphy, 1992). Dasso et al. (1992), however, have shown recently that the presence of Xenopus RCC1 protein in the nuclear extracts is not required for nuclear envelope assembly. Their experiments, though, do demonstrate that RCC1 is essential for DNA replication. The effect of RCC1 on DNA synthesis is not a result of the protein’s direct enzymatic involvement in replication. Instead, RCC1 either maintains or promotes the formation of the DNA poly- merase/chromatin structural complex. Taken together, the vertebrate and yeast data implicate the GTP-binding ability of TC4/Gsp1 through its association with RCC1/Prp20 as a strong candidate for modulating the structural arrangements of the nucleoplasm. In addition, the influence that RCC1/Prp20 exerts upon the organization of nuclear components thus affects the initiation of DNA replication.

Even more intriguing, the Prp20 dsDNA-binding complex contains two other GTP-binding proteins of molecular masses 17 kDa and 32 kDa (Fig. 6). Multiple GTP- hydrolysis events could thus be required for the Prp20 complex activities. Although no direct evidence has been presented for an association of mammalian RCC1 with GTP-binding proteins other than Ran/TC4, a small number of GTP-binding proteins have been identified in the mammalian nucleus and nuclear envelope (Rubins et al., 1990; Seydel and Gerace, 1991). These nuclear GTP-binding proteins generally fall into two size ranges of 20 kDa to 30 kDa and 45 kDa to 70 kDa. Although their specific functions are not yet known, some of the mammalian nuclear GTP-binding proteins will be utilized in the nuclear envelope assembly process, while others will likely be involved in the RCC1-directed maintenance of the organization of the nucleoplasm. Furthermore, there is precedent for multiple GTP-binding proteins regulating complex cellular processes. In signal recognition particle (SRP)-mediated translocation of proteins across the RER membrane (Rapiejko and Gilmore, 1992) both the SRP (SRP54) and the α and β subunits of the SRP receptor (SRα and SRβ) are potential GTP-binding proteins. Site-directed mutagenesis experiments will be valuable in studying such a complex form of regulation. In consequence, the genes for the 17 kDa and 32 kDa GTP-binding proteins of the Prp20 complex will be sought using reverse genetics. A database search with the amino-terminal peptide sequence for the 17 kDa protein reveals that it is a unique protein (A. Lee and M. W. Clark, unpublished results). When the genes for the 17 kDa and 32 kDa are obtained they will be analyzed against mutations made in the Gsp1 and Gsp2 genes that have already been isolated (Belhumeur et al., 1993) to determine the involvement of GTP-hydrolysis in the regulation of the nuclear activities of Prp20.

The loss of the dsDNA-binding ability of Prp20 in the arrested extracts of the cdc4, cdc7, and cdc8 mutants implies that the Prp20 complex is altered when the cell is preparing for, and is in the process of DNA synthesis. This S-phase-specific alteration of the Prp20 complex appears to be reversible. In mutants that prevent mitosis (cdc13) or nuclear separation (cdc15), Prp20 is readily retained on the dsDNA-cellulose. At this time it is not clear how these dramatic changes in dsDNA-binding affinities in vitro are reflected in the binding of the Prp20 complex to the chromatin. However, it is reasonable to assume that the altered complex must also show a reduced affinity for the chromatin even if that affinity is not totally eliminated. The importance of this cell cycle-dependent behaviour for Prp20 is further manifested by the necessity for Xenopus RCC1 in the initiation of DNA replication (Dasso et al., 1992).

Analysis of the nucleolar morphology in the cdc8- arrested cells indicates that a segregation of the activities of the Prp20 complex may exist. The activity represented by the cell cycle-dependent dsDNA binding can be altered without affecting the essential role that Prp20 plays in the maintenance of nucleoplasmic arrangement. Utilizing extracts from the cdc mutants arrested in DNA replication, we can now investigate the molecular basis for the S-phase- specific alteration(s) of the Prp20 complex. In addition, the different stages of the cell cycle can be examined to determine if any alteration(s) of the Prp20 complex are involved in a DNA replication monitoring and signalling mechanism, as well as to dissect the multiple roles played by RCC1 in the eukaryotic nucleus.

We thank Drs J. Banks, H. Bussey and P. Lasko for useful discussions. We thank Dr M. Aebi for PET-3a-PRP20 and we are grateful to Dr Paul Lasko for the critical review of this manuscript.

This work was supported by a Postdoctoral Fellowship from the Medical Research Council of Canada (MRC) to P.B., a graduate fellowship from the Fonds pour la Formation de Chercheurs et l’Aide a la Recherche (FCAR) to R.T., and an Operating Grant to M.W.C. from the National Cancer Institute of Canada (NCIC).