INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Equal distribution of genetic material during eukaryotic cell division requires reorganization of chromosome structure in mitosis, known as mitotic chromosome condensation. Condensation results in compaction of chromosomes, such that the average distance between points along the chromosome is reduced approximately fivefold in higher eukaryotes and twofold in budding yeast (reviewed by Koshland and Strunnikov, 1996; Hirano, 1999). Condensation is thought to serve several functions. These include the reduction of the length of chromosome arms such that they are shorter than half the length of the mitotic spindle and thus can be completely segregated into daughter cells during cytokinesis. Condensation may also help to resolve entangled chromatin fibers and increase mechanical resistance of the chromosomes to the forces of the mitotic spindle.

Several factors involved in this process have been identified in various organisms. They are evolutionarily related, as judged by their sequences, pointing to conservation of the basic mechanisms of mitotic chromosome condensation. These factors include a so-called condensin complex, topoisomerases, histone H3, and a number of additional proteins identified by yeast mutations.

The best biochemically characterized chromosome condensation factors are the 8S and 13S "condensin" complexes, identified in the Xenopus egg extract system (Hirano et al., 1997). The 8S complex is important, but not sufficient for mitotic chromosome condensation. It consists of two SMC-type (structural maintenance of chromosomes) proteins, XCAP-C and XCAP-E. Their budding yeast homologues, Smc2p and Smc4p, have also been implicated in chromosome condensation (Strunnikov et al., 1995). The 13S condensin complex is necessary and sufficient to perform Xenopus mitotic chromosome condensation in vitro. It consists of five subunits, which in addition to XCAP-C and XCAP-E include three unrelated proteins: XCAP-D2, XCAP-G, and XCAP-H (Hirano et al., 1997). The 13S condensin complex is capable of binding DNA and using ATP to induce a global change in DNA configuration (Kimura and Hirano, 1997; Kimura et al., 1999).

In mitosis, XCAP-H, and to a lesser extent XCAP-G and XCAP-D2, subunits are hyperphosphorylated, and the complex is targeted to the chromosomes (Hirano et al., 1997). Cdc2 protein kinase is at least partly responsible for this phosphorylation, which is accompanied by a shift in electrophoretic mobility of these proteins (Kimura et al., 1998). This phosphorylation is necessary to activate the DNA reconfiguring activity of the condensin complex. It was hypothesized that this activity provides the driving force for mitotic DNA condensation (Hirano et al., 1997; Kimura et al., 1999). From the biochemical studies in Xenopus, it appears that the function of condensins is limited to mitotic compaction of chromatin.

Mutations in fission yeast Schizosaccharomyces pombe genes homologous to Xenopus condensins cause defective chromosome condensation in mitosis (Sutani et al., 1999). In this organism, mitotic phosphorylation of Cut3/SMC4 subunit, which is homologous to Xenopus XCAP-C, is required for mitotic relocation of condensins from cytoplasm to the nucleus (Sutani et al., 1999).

A mutation in the homologue of condensin subunit XCAP-H has been described in Drosophila (Bhat et al., 1996). It results in a mitotic chromosome segregation defect, in which the centromeres separate but chromosome arms do not get resolved. In contrast to the situation in Xenopus egg extracts depleted of condensins, no detectable defect in chromosome condensation could be observed in the barren mutant. The Barren protein was reported to interact with topoisomerase II and to activate its decatenating activity. It was hypothesized that the defect in topoisomerase II activation is responsible for the failure of chromosome resolution in mitosis in barren mutant embryos (Bhat et al., 1996).

Although chromosome condensation cannot be directly observed in budding yeast, it can be detected using fluorescence in situ hybridization (FISH), using either cosmid-size probes or probes that hybridize to the ribosomal DNA array (Guacci et al., 1994). Ribosomal DNA encompasses a region of ~500 kb on chromosome XII, and its condensation state can be visually assessed after hybridization of a fluorescent probe. In interphase, the rDNA appears as a diffuse area, whereas in mitotic cells it has a defined string-like or bead-like shape (Guacci et al., 1994).

A mutation in SMC2, the budding yeast homologue of XCAP-E, leads to a defect in mitotic chromosome condensation and segregation (Strunnikov et al., 1995). Mutant cells accumulate in mitosis while retaining relatively high viability. Some cells eventually undergo an abnormal division and arrest as unbudded cells (i.e., in the G1 phase of the cell cycle). When grown at permissive temperatures, the cells do not show a significant increase in the rate of chromosome loss. This set of characteristics is different in some respects from the phenotype of the yeast top2 mutants, which affect topoisomerase II (DiNardo et al., 1984; Holm et al., 1989). These cells attempt to segregate their chromosomes, which results in lethality. Unlike smc2, the top2 mutant also has an increased chromosome loss rate.

Condensation defect was also detected in a double mutant trf4 top1 (Castano et al., 1996). TRF4 was identified in a screen for mutations that are inviable in combination with topoisomerase I null mutation. Trf4p physically interacts with Smc1p and Smc2p. Its biochemical activities or cellular functions are unknown.

All five known Xenopus condensin subunits have highly similar homologues in the budding yeast genome. In addition to SMC2 and SMC4, there is BRN1, the homologue of the XCAP-H and Drosophila Barren, which is the focus of this work. We have also identified the yeast homologue of XCAP-G, YCG1, as a dosage suppressor of brn1 mutation. The homologue of XCAP-D2, named LOC7, was identified in a screen for genes necessary for sister chromatid separation and segregation (N. Bhalla and A. Murray [University of California, San Francisco, CA] Saccharomyces Genome Database entry). Here we explore the properties of BRN1 as a step to dissect the molecular mechanisms of mitotic chromosome condensation.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Deletion of BRN1 was accomplished by replacing the complete ORF of the gene with the KanMX4 marker, which confers resistance to G418 (Wach et al., 1994). This was done by PCR amplification of the KanMX4 module from the pFA6a-kanMX4 plasmid (Wach et al., 1994), using the primers containing 18-19 bp identity to the regions flanking the KanMX4 gene at their 3' ends, and 45 bp identity to the sequences flanking the BRN1 ORF at the 5' ends. The PCR product was transformed into a diploid yeast strain (W303 derivative), and G418-resistant colonies were tested for correct replacement of BRN1 using PCR, encompassing both 5' and 3' junctions.

Temperature-sensitive mutations of BRN1 were created by PCR-based mutagenesis or by chemical mutagenesis of the cloned gene. In the PCR experiment, we have separately mutagenized the regions approximately corresponding to the N-terminal, middle, and C-terminal one-third of the protein. The BRN1 gene in a TRP1 CEN plasmid was cut ("gapped") with BsrGI+NcoI, SphI, or Eco47III+SalI, respectively. The gapped plasmids were cotransformed with the corresponding PCR products into a brn1-1 + p(BRN1 URA3) strain, followed by eviction of the BRN1 URA3 plasmid on 5FOA-containing plates. Temperature-sensitive strains were selected and verified by plasmid rescue in Escherichia coli and retransformation into yeast. We have recovered one mutant resulting from the mutagenesis of the middle part of the gene (brn1-20), and several mutants in the C-terminal part of the gene. Mutants of the latter group had multiple substitutions, two of which were common to all alleles; we chose the brn1-34 allele, which has only two substitutions, for further analysis. Chemical mutagenesis with hydroxylamine, which produced the brn1-60 mutation, was performed as described (Sikorski and Boeke, 1991).

Chromosome condensation was assayed by FISH of the ribosomal DNA region, as described (Guacci et al., 1994). The probe was generated by PCR amplification of a fragment of rDNA repeat unit and labeled with biotin using the BioNick nick-translation system (Life Technologies, Grand Island, NY). Blind scoring of at least 100 cells in each preparation was used to determine the percentage of condensed chromosomes.

Sister chromatid cohesion and segregation were analyzed in a strain containing an array of Lac operator sequence repeats integrated at the LEU2 locus, close to the centromere of chromosome IV, and expressing a LacI::GFP fusion protein (Straight et al., 1996, 1997). The strain was crossed to brn1-60 mutant, and ts, green fluorescent protein-positive (GFP⁺) segregants were selected. Cells were fixed with 4% formaldehyde for 15 min, placed onto polylysine-coated slides, stained with 0.1 µg/ml DAPI, and mounted in Vectashield (Vector Laboratories, Burlingame, CA) for microscopy.

Flow cytometry was performed as described earlier (Ouspenski et al., 1995). Cells were fixed in 80% ethanol, treated with RNase A (1 mg/ml, 2 h at 37°C), and stained with 10 µg/ml propidium iodide. The samples were briefly sonicated just before analysis to disperse clumps.

Pulse-field gel electrophoresis was performed using the CHEF-DR II system (Bio-Rad, Hercules, CA) according to manufacturer's recommendations. Samples were run at 200 V with 120 s pulse time for 36 h.

Anti-Brn1p antibody was raised in a rabbit against the synthetic peptide IDMPIKNRKNDTHYL, corresponding to amino acids 457-471 of the predicted sequence. Affinity purification, immunoblotting, and immunofluorescence were done according to conventional procedures (Harlow and Lane, 1988; Pringle et al., 1989).

Immunofluorescent staining of yeast cells was done as described (Kilmartin and Adams, 1984), except that cells were fixed with formaldehyde for 30 min.

Immunoprecipitation was performed from cells containing pAS443 (2 mm SMC2::MYC6, a gift from A. Strunnikov) and pIL114 (CEN GAL->3HA::BRN1, this study), induced with galactose overnight. Cells (~10⁹) were broken with glass beads in 2 ml IP buffer (20 mM HEPES, pH 7.9, 150 mM KCl, 2 mM MgCl₂, 0.1 mM DTT, 10% glycerol, supplemented with protease and phosphatase inhibitors [Harlow and Lane, 1988]), and insoluble matter was removed by centrifugation (20,000 × g for 20 min). Extracts were supplemented with Triton X-100 to 0.1% and BSA to 1 mg/ml. After preclearing with protein G Sepharose, the extract was split in four, and each portion was incubated overnight with protein G beads preloaded with monoclonal antibodies to Myc (9E10), hemagglutinin (HA) (12CA5), tubulin (negative control, YOL1/34), or an affinity-purified rabbit polyclonal anti-Brn1p antibody described above. Beads were washed six times with IP buffer, boiled in SDS-containing sample buffer, and analyzed by immunoblotting.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

BRN1 mutations

The yeast gene corresponding to the ORF YBL097W, for which we use the name BRN1, has been pointed out as the possible homologue of the Drosophila Barren gene, on the basis of sequence homology (Bhat et al., 1996). It also has high sequence similarity to Xenopus condensin subunit XCAP-H and human BRRN1 (Hirano et al., 1997; Cabello et al., 1997). We cloned the BRN1 gene from a W300-derived strain and found that its sequence differs from the corresponding Saccharomyces Genome Database entry by one amino acid: glycine-495 rather than alanine. The difference may be due to strain polymorphism.

To explore the function of BRN1 in yeast, we deleted the ORF of the gene in a diploid strain, replacing it with the KanMX4 kanamycine resistance module (Wach et al., 1994). Sporulation and spore dissection of this strain demonstrated that BRN1 is essential for viability.

We created three independent temperature-sensitive alleles of the BRN1 gene. brn1-20 (K489E) and brn1-34 (K592E + E638G) alleles were obtained by error-prone PCR mutagenesis of the middle one-third and C-terminal one-third of the BRN1 ORF, respectively. brn1-60 mutation was generated by chemical mutagenesis of the plasmid containing BRN1 gene. This allele has the same mutation as brn1-20 (K489E), plus an additional substitution P490S. This second substitution can be viewed as a partial reversion, because it improves the growth of cells at subrestrictive temperatures (although they exhibit a tight arrest at 37°C). The mutant alleles were substituted for BRN1 in the genome by "pop in, pop out" gene replacement. The resulting brn1-20 mutant cells grow slower than wild type at all temperatures; brn1-34 cells grow at a wild-type rate at temperatures up to 35°C and stop growth at 37°C, but frequently give rise to spontaneous "revertant" colonies; and the brn1-60 mutant exhibits normal growth up to 35°C and a tight growth arrest at 37°C. Unless indicated otherwise, all experiments described here were performed with brn1-34 and brn1-60 alleles, and only the results obtained with brn1-60 are shown, because no significant differences in the phenotypes between these two alleles were detected.

BRN1 Is Necessary for Chromosome Condensation and Segregation, but Not for Sister Chromatid Cohesion

Chromosome Condensation. Because the BRN1 homologue in Xenopus, XCAP-H, is necessary for mitotic chromosome condensation, we tested brn1 mutant cells for a condensation defect. As a marker of mitotic condensation, we have assessed the state of the ribosomal DNA region of chromosome XII, which encompasses ~500 kb of DNA sequence. When visualized by FISH, the rDNA array appears as a diffuse mass in interphase, whereas in mitotic cells it forms defined string-like structures (Guacci et al., 1994) (Figure 1). In this experiment, exponentially growing cells were shifted to the restrictive temperature, and at the same time, nocodazole was added to arrest cells in mitosis. After incubation for 3.5 h, the cells were processed for FISH and analyzed by fluorescence microscopy. Mutant cells proceed though the cell cycle and arrest at mitosis under these conditions, as evidenced by the accumulation of large-budded cells. In brn1 mutants, the ribosomal DNA region is uncondensed in most cells at the restrictive temperature (Figure 1). The observed rDNA morphology in brn1 cells is indistinguishable from that of smc2 mutant (Strunnikov et al., 1995) (Figure 1), indicating that BRN1 is also necessary for proper mitotic condensation.

View larger version (9818K): [in this window] [in a new window]   Figure 1.   Mitotic chromosome condensation defect in brn1 mutant cells. Examples of rDNA array morphologies, visualized by fluorescence in situ hybridization, are shown. BRN1 (w.t.) or brn1-60 cells were incubated at indicated temperatures for 3.5 h with or without nocodazole (noc). rDNA morphology in smc2-8 cells is shown for comparison. The bottom right panel (noc->1h 37C) shows brn1-60 cells blocked in nocodazole at permissive temperature for 2.5 h, followed by a shift to 37°C in the continued presence of nocodazole. Arrowheads point to examples of rDNA morphology scored as "intermediate." The graph below shows percentages of cells with the indicated rDNA morphologies after 3.5 h at the restrictive temperature in the presence of nocodazole. Blind scoring of at least 100 cells in each preparation was performed. Cells that could not be unequivocally assigned to one of the two classes were scored as intermediate (see arrowheads above).

Sister Chromatid Cohesion. Some aspects of mitotic chromosome condensation are linked to sister chromatid cohesion, as illustrated by mcd1/scc1 mutations, in which sister chromatids separate prematurely and fail to condense properly in mitosis (Guacci et al., 1997; Michaelis et al., 1997). To determine whether BRN1 is necessary for chromatid cohesion, we tagged the centromeric region of chromosome IV in brn1 mutant cells with an array of Lac operator repeats (Straight et al., 1996). Expression of a fusion of LacI repressor with the GFP allows visualization of the centromeric region. In wild-type cells, sister chromatids remain attached until the onset of anaphase, and GFP fluorescence appears as a single spot. We tested brn1 mutants for maintenance of sister chromatid cohesion, when the cells were prevented from progression into anaphase by nocodazole. Midlog phase cultures were split into two halves, and one half was shifted to the restrictive temperature. At the same time, cell cycle progression was blocked by addition of nocodazole, and cells were examined by fluorescence microscopy 3 h later. Almost all cells produced a single fluorescent spot at both permissive and restrictive temperatures: 90-95% in wild-type as well as mutant cells (200 cells of each genotype scored). This indicates that Brn1p is not required to establish or maintain sister chromatid cohesion after DNA replication (Figure 2).

View larger version (117K): [in this window] [in a new window]   Figure 2.   Sister chromatid cohesion is normal in brn1 mutant cells. The centromeric region of chromosome IV in these strains is tagged with an array of Lac operator repeats and visualized by expressing LacI::GFP fusion protein. GFP fluorescence (bright dots) is shown overlaid onto DIC images of cells. Wild-type (A) and brn1-60 mutant (B) cells were incubated at 37°C for 3 h in the presence of nocodazole. The two sister centromeres appear as a single dot in >90% of wild-type and mutant cells under these conditions (200 cells of each type scored).

Chromosome Segregation in Mitosis. Chromosome segregation in brn1 mutants was followed in cells with chromosome IV centromere tagged with GFP, as in the above experiment. Shifting exponentially growing mutant cells to the restrictive temperature for 3 h resulted in accumulation of large-budded cells with a single nucleus. In some of these, the nuclei were elongated and traversed the bud neck (Figure 3A). In contrast to wild-type cells, where the centromeres in elongated nuclei were always separated, centromeric GFP signal appeared as a single dot in most mutant cells (70-85% depending on the allele; 200 cells scored). After 5 h at restrictive temperature, some mutant cells (30-60%) separate their centromeres, whereas the chromatin mass remains stretched through the bud neck, a morphology never observed in wild-type cells (Figure 3A, compare the rightmost cells). This likely reflects the failure of chromosome arms to compact properly, so that their length remains greater than half the mitotic spindle length.

View larger version (20K): [in this window] [in a new window]   Figure 3.   (A) Chromatid separation and segregation in BRN1 (w.t.) and brn1-60 mutant cells. Centromere of chromosome IV, visualized with LacI::GFP, is shown in green; cell nuclei, stained with DAPI (pseudocolored), are bright red; the outline of the cells is visible as red background. Exponentially growing cultures were shifted to 37°C for 3 h and processed for microscopy. (B) Examples of mitotic spindle morphology in BRN1 (w.t.) and brn1-60 cells incubated at 37°C for 3 h. Antitubulin immunofluorescence is shown in green; DNA stained with DAPI is pseudocolored red.

Cell Cycle Progression. To determine how the absence of BRN1 function affects cell cycle progression, we followed the mutant cell morphology by microscopy and the DNA content by flow cytometry. Because of the relatively high viability of mutant cells (Figure 4A), we could not assign the essential function of BRN1 to a defined cell cycle stage. When asynchronously growing cells were shifted to the restrictive temperature for 3 h, they accumulated at the large-budded stage, indicating a delay or arrest in G2 or mitosis (Figure 4B). The large-budded cells contain a single nucleus that either has not migrated to the bud or has traversed the bud neck and is abnormally stretched (Figures 3 and 4B).

View larger version (36K): [in this window] [in a new window]   Figure 4.   The effect of brn1 mutation on cell cycle progression. (A) Relative viability of brn1-60 cells. Midlog phase cultures were split into two halves, and one half was shifted to the restrictive temperature (37°C). At indicated time points, portions of each culture were plated out at a density of 50-500 cells per plate, and the numbers of resulting colonies, divided by the corresponding number of colonies at time zero, were plotted on the graph. (B) Cell cycle stage distribution (% of total; 200 cells scored) of brn1-60 cells at the restrictive temperature, compared with wild-type (BRN1) cells. Asynchronous cultures were incubated at 37°C for 3 h. (C) DNA content of BRN1 and brn1-60 cells arrested at the G1 phase with -factor and released at the restrictive temperature. Samples of cells were fixed at the indicated time points, stained with propidium iodide, and analyzed by flow cytometry. (D) Pulse-field electrophoresis of chromosomal DNA of BRN1 (lanes 1, 2) and brn1-60 mutant cells (lanes 3, 4) grown at 23°C (lanes 1, 3) or incubated at 37°C for 3 h. Lane 5, BRN1 cells treated with 100 mM hydroxyurea for 3 h.

Brn1p Interacts with Smc2p In Vivo

Because the Xenopus homologue of Brn1p, XCAP-H, is a part of the 13S condensin complex (Hirano et al., 1997), we sought to determine whether it interacts with other prospective condensin subunits in yeast cells. Brn1p tagged with HA epitope was coexpressed with Smc2p tagged with Myc (Strunnikov et al., 1995), and proteins were immunoprecipitated with anti-Myc antibody. Immunoblotting with anti-HA antibody revealed specific coimmunoprecipitation of Brn1p with Smc2p only when Myc-tagged Smc2p was present (Figure 5). Thus BRN1 and SMC2 encode components of the same molecular complex in yeast.

View larger version (35K): [in this window] [in a new window]   Figure 5.   Physical association between Brn1p and Smc2p. Total protein extracts from cells expressing HA-Brn1p (from GAL1 promoter on a CEN plasmid) and Myc-Smc2p (from endogenous promoter on a 2-µm plasmid) were immunoprecipitated with anti-Myc antibody (9E10) and analyzed by immunoblotting with anti-HA and anti-Myc antibodies, as indicated. WCE, Whole cell extract (input), 20 µl; Myc-IP, immunoprecipitation from 400 µl of WCE.

Genetic Interactions of BRN1 with Other Chromosome Condensation Factors

To identify proteins that functionally interact with BRN1, we performed a screen for dosage suppressors of brn1-60 mutation. Two overlapping sets of clones have been recovered multiple times (more than 50 independent clones in each set), one set encoding BRN1 itself and the other containing an uncharacterized ORF YDR325W. This gene is homologous to XCAP-G, a component of Xenopus 13S condensin complex, and likely encodes the corresponding condensin subunit in yeast. We name this gene YCG1 (Yeast CAP G). Overexpression of YCG1 restores growth of brn1-60 cells at 37°C to a nearly wild-type rate (Table 1). The suppression is allele-specific, because YCG1 suppresses the brn1-20 mutation only to a limited extent and does not suppress the brn1-34 allele.

To explore the functional relations of BRN1 with other condensins, we tested the gene for dosage interactions with SMC2. Overexpression of SMC2 from a high-copy plasmid failed to suppress temperature sensitivity of brn1 mutants (Table 1). The reciprocal experiment produced a similar outcome: overproduction of Brn1p did not rescue the smc2 temperature-sensitive mutation. Overexpression of YCG1 in smc2 mutant cells resulted in only marginal suppression, possibly reflecting a closer interaction of YCG1 with BRN1, as compared with SMC2.

Brn1p Is a Cell Cycle-regulated Nuclear Protein

We raised and affinity-purified an antibody to a peptide derived from the predicted Brn1p sequence. When this antibody was used for immunofluorescence, we were unable to detect the endogenous protein, presumably because of its low abundance. Overexpression of Brn1p from GAL1 promoter resulted in an uneven pattern of nuclear staining in some cells, possibly reflecting the subnuclear distribution of the protein (Figure 6). Specificity of immunofluorescence staining was confirmed by preincubating the antibody with the antigenic peptide, which abolished the staining.

View larger version (41K): [in this window] [in a new window]   Figure 6.   Intracellular localization of the Brn1p protein. (A) Immunofluorescent staining of cells overexpressing Brn1p from the GAL1 promoter on a CEN plasmid, using an affinity-purified anti-Brn1p antibody. (B) Nuclear DNA of the same cells stained with DAPI. (C) Phase-contrast image of the same cells. (D) Levels of Brn1p under induced (Gal) and uninduced (Glu) conditions.

The genome-wide survey of cell cycle regulation of gene expression in yeast showed that the level of BRN1 transcript is increased at G2/M (Cho et al., 1998), suggesting that the protein level may be regulated as well. Using the anti-Brn1p antibody, we compared endogenous Brn1p levels in wild-type cells arrested in G1 with -factor and in mitosis with nocodazole. The level of protein in mitosis is significantly higher than in G1 (Figure 7A). The dynamics of Brn1p level in the cell cycle was followed in cells synchronized at the G1/S boundary with hydroxyurea (Figure 7B). Protein abundance drops 45-60 min after release from the block, as the cells complete mitosis, followed by accumulation as the cells approach mitosis in the next cell cycle.

View larger version (28K): [in this window] [in a new window]   Figure 7.   Cell cycle regulation of Brn1p. (A) Levels of endogenous Brn1p in wild-type cells blocked in mitosis with nocodazole (N), or in G1 with -factor (), detected by immunoblotting with anti-Brn1p antibody. Equal amounts of total protein were loaded in the two lanes. (B) Brn1p in wild-type cells released from hydroxyurea arrest. Cells were collected every 15 min, and total protein extracts obtained from identical culture volumes were loaded onto each lane. Most cells have completed mitosis (78% unbudded cells, n = 200) 1 h after release, whereas at the 2 h point most cells were in G2 or M (73% cells with bud size more than one-half of the mother cell, n = 200).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Yeast Brn1p protein is similar in sequence to Xenopus condensin subunit XCAP-H, human and mouse BRRN1, S. pombe Cnd2, and a predicted protein from Arabidopsis thaliana (pairwise BLAST E values in the range of e²³ to e³⁴). This sequence homology reflects functional conservation, because the human protein can substitute for BRN1 function in yeast (our unpublished observation). Homology between these proteins is significantly higher than between Brn1p and Drosophila Barren (Blast E value = 0.004). This may account for some of the specific features of the barren mutant phenotype in Drosophila, as compared with the yeast mutant phenotype and functional data in the Xenopus system.

The sequence of BRN1, as well as its homologues in other species, contains several potential PEST sequences, which are characteristic of unstable proteins and may serve as signals for regulated degradation (Rechsteiner and Rogers, 1996). This is in agreement with our data that Brn1p level is regulated in the cell cycle. Perhaps the protein must be degraded for chromosome decondensation after completion of mitosis.

The lack of electrophoretic mobility change of Brn1p in the cell cycle is surprising, because XCAP-H exhibits a substantial shift attributable to hyperphosphorylation in mitosis (Kimura et al., 1998). It should be noted that although Brn1p, as well as XCAP-H and other homologues, has numerous consensus phosphorylation sites for several protein kinases, few of these sites are located within regions that are highly conserved between species. This raises the possibility that the mechanisms of regulation of the protein vary in different species.

Although some brn1 mutant cells permanently arrest in mitosis, a significant proportion of cells arrest with a terminal morphology characteristic of S-phase. This raises the possibility that Brn1p may have a function in interphase, possibly in DNA replication. Alternatively, mitotic BRN1 function may be necessary to "reset" chromatin for the next round of replication.

The mitotic defects of brn1 cells are similar to those described for the smc2 mutant (Strunnikov et al., 1995). In addition to chromosome condensation defect, this includes the arrest of some cells in mitosis, stretched nuclei, and terminal arrest in interphase of the second cell cycle. This indicates that the two genes are involved in a common cellular function, as suggested by their homology to subunits of the same molecular complex in Xenopus. There are some differences, however, which include the initiation of bud development by some brn1 cells before arrest. It remains to be determined whether this difference reflects functional specialization of the two proteins or is specific to alleles of the respective mutations. Specialization of condensin subunits is further illustrated by the fact that overexpression of YCG1, but not SMC2, can suppress brn1 lethality.

Like XCAP-H in Xenopus, yeast Brn1p is required for chromosome condensation in mitosis. This is in contrast to the phenotype of barren mutation in Drosophila, in which no condensation defect could be detected (Bhat et al., 1996). An additional difference is the centromere separation defect in our mutants. In Drosophila mutant, centromere separation occurs normally in mitosis, whereas chromosome arms remain interlocked and form chromatin bridges. These Drosophila phenotypes are observed in a null mutation of barren, so the differences cannot be explained by allele-specific defects.

According to the "superhelical tension" model of mitotic chromosome condensation (Hirano, 1999), XCAP-H functions in mitosis as a component of condensin complex. The activity of the complex is to introduce positive supercoils into DNA, leading to compaction of the chromatin fiber. A defect in this function should result in mitotic chromosomes that are less compact, but it does not predict a chromatid separation defect like the one observed in brn1 mutants in yeast. This may indicate that chromatin compaction is mechanistically necessary for resolution of sister chromatids. If this is the case, further development of the model is required to account for interdependence of chromatid condensation and separation. Alternatively, Brn1p may have a role in chromatid separation, which is separate from its function in condensation. It will be of interest to use the yeast homologues of other condensin subunits to dissect the molecular details of chromatin rearrangements during cell division.