Suppressors of Bir1p (Survivin) Identify Roles for the Chromosomal Passenger Protein Pic1p (INCENP) and the Replication Initiation Factor Psf2p in Chromosome Segregation

Abstract

Fission yeast Bir1p/Cut17p/Pbh1p, the homolog of human Survivin, is a conserved chromosomal passenger protein that is required for cell division and cytokinesis. To study how Bir1p promotes accurate segregation of chromosomes, we generated and analyzed a temperature-sensitive allele, bir1-46, and carried out genetic screens to find genes that interact with bir1⁺. We identified Psf2p, a component of the GINS complex required for DNA replication initiation, as a high-copy-number suppressor of the bir1-46 growth defect. Loss of Psf2p function by depletion or deletion or by use of a temperature-sensitive allele, psf2-209, resulted in chromosome missegregation that was associated with mislocalization of Bir1p. We also found that the human homolog of Psf2p, PSF2, was required for proper chromosome segregation. In addition, we observed that high-copy-number expression of Pic1p, the fission yeast homolog of INCENP (inner centromere protein), suppressed bir1-46. Pic1p exhibited a localization pattern typical of chromosomal passenger proteins. Deletion of pic1⁺ caused chromosome missegregation phenotypes similar to those of bir1-46. Our data suggest that Bir1p and Pic1p act as part of a conserved chromosomal passenger complex and that Psf2p/GINS indirectly affects the localization and function of this complex in chromosome segregation, perhaps through an S-phase role in centromere replication.

Chromosomal passenger proteins play key roles in coordinating cell division and cytokinesis. The passenger protein Survivin has been implicated in control of chromosome biorientation, attachments between the kinetochore and spindle microtubules, and spindle checkpoint surveillance, in addition to promotion of apoptosis (reviewed in reference 2). Survivin is highly expressed in many human cancers and may contribute to oncogenic transformation (3). However, because loss of Survivin function in human cells and in mice results in pleiotropic phenotypes such as chromosome misalignment and missegregation, centrosome abnormalities, defects in mitotic spindle formation and checkpoint signaling, and polyploidy (9, 32, 34, 56, 58), the primary function of Survivin in cell division remains unclear. To study Survivin function, we used fission yeast as a model system. The Survivin homolog in fission yeast, Bir1p (also called Cut17p or Pbh1p), is known to be essential for cell viability, equal chromosome segregation, and cytokinesis (38, 43, 44, 55). Here, we further characterize Schizosaccharomyces pombe Bir1p and identify functional interactions between Bir1p, the inner centromere protein homolog Pic1p (S. pombe INCENP), and Psf2p, a protein involved in DNA replication.

Survivin and Bir1p belong to the IAP (inhibitor of apoptosis) family of proteins (11) that contain BIR (baculovirus IAP repeat) domains at the N terminus and a long helix at the C terminus (10, 39, 57). Survivin/Bir1p acts as part of a conserved complex that includes the protein kinase Aurora B (S. pombe Ark1p) and the inner centromere protein INCENP (S. pombe Pic1p) (7, 33). The chromosomal passenger proteins localize to the kinetochore during metaphase but translocate to the spindle midzone as chromosomes segregate at anaphase. Survivin/Bir1p affects the localization and protein kinase activity of Aurora B/Ark1p (7, 23, 38, 41, 49, 53). Aurora B is thought to monitor spindle microtubule tension at the kinetochore and promote biorientation of sister kinetochores to opposite spindle poles and is also involved in spindle disassembly (6, 8, 13, 22, 32, 52).

To characterize the role of fission yeast Bir1p in chromosome segregation, we generated and analyzed a temperature-sensitive allele, bir1-46, and identified a novel protein, Psf2p, as a high-copy-number suppressor of the bir1-46 growth defect. We independently identified fission yeast Psf2p in a screen for mutants that block rereplication (16). Psf2p homologs in Saccharomyces cerevisiae and Xenopus laevis were recently described as components of the GINS protein complex, which consists of four subunits (Sld5p/Cdc105p, Psf1p/Cdc101p, Psf2p/Cdc102p, and Psf3p) and is essential for DNA replication (27, 30, 51). The GINS complex associates with replication origins and loads proteins that are required for replication initiation, such as Cdc45p (51). In addition, the GINS complex is required for replication elongation, possibly as a component of active replication forks (27, 30, 51). We also found that Pic1p (S. pombe INCENP), which we identified previously as a protein that interacts with the Aurora B protein kinase Ark1p (33), acts as a high-copy-number suppressor of the bir1-46 growth defect. In this work, we further characterize the relationship between the chromosomal passenger proteins Bir1p and Pic1p, and we demonstrate that the replication initiation factor Psf2p is required for Bir1p localization and function in chromosome segregation.

MATERIALS AND METHODS

Yeast strains and genetic methods.

All S. pombe strains (Table 1) in this study were derived from the wild-type strains 972 (h⁻) and 975 (h⁺). Standard genetic techniques and media have been described previously (1, 18, 37). Cells were synchronized in G₂ using lactose gradients (reference 4 and references therein). Small G₂ cells were collected by centrifugation through 10 to 40% lactose in YES medium, and then released to restrictive temperature (36°C) for 4 h. Cell cycle progression was monitored according to septation index (a marker for cells in S phase) and number of binucleate cells (cells in mitosis).

TABLE 1.

Yeast strains used in this study

Strain	Genotype	Source or reference
Sp1	h−ura4-D18 leu1-32 his3-D1 ade6-M216	Laboratory stock
Sp109	h−/h+ura4-D18/ura4-D18 leu1-32/leu1-32 his3-D1/his3-D1 ade6-M216/ade6-M210 bir1+/bir1Δ::his3+	This study
Sp157	h−ura4-D18 leu1-32 his3-D1 ade6-M210 bir1Δ::his3+ pREP3X-bir1+	This study
Sp287	h−ura4-D18 leu1-32 his3-D1 ade6-M216 bir1-46	This study
Sp367	h−leu1-32 his3-D1 ade6-M210 mts3-1	16
Sp476	h−/h+ura4-D18/ura4-D18 leu1-32/leu1-32 his3-D1/his3-D1ade6-M216/ade6-M210 psf2+/psf2Δ::his3+	This study
Sp483	h−/h+ura4-D18/ura4-D18 leu1-32/leu1-32 his3-D1/his3-D1 ade6-M216/ade6-M210 pic1+/pic1Δ::his3	This study
Sp499	h−his3-D1 ade6-M216 ura4-D18 leu1-32 psf2Δ::his3+pDS672a-Psf2p-myc	This study
Sp500	h−his3-D1 ade6-M216 ura4-D18 leu1-32 psf2Δ::his3+pSGP572a-Psf2p-GFP	This study
Sp605	h−/h+ura4-D18/ura4-D18 leu1-32/leu1-32 his3-D1/his3-D1ade6-M216/ade6-M210 sld5+/sld5Δ::his3+	This study
Sp613	h−leu1-32 cut17-275	38
Sp631	h+psf2-209 ura4-D18 leu1-32 his3-D1 ade6	16

The wild-type bir1⁺ gene was amplified from a cDNA library by PCR (46) and subcloned into the plasmid pREP3X, which contains the nmt1 (no message in thiamine) promoter. To generate strain Sp109, which carries the bir1 deletion in the chromosome but remains viable due to ectopic expression of Bir1p, a wild-type diploid strain had one copy of bir1⁺ deleted by one-step disruption with his3⁺. The deletion was confirmed by two independent genomic PCR analyses (data not shown). Tetrad analysis of Sp109 resulted in 2:2 segregation of viable spores to inviable spores (data not shown), consistent with previous reports that S. pombe bir1⁺ is essential for cell viability (43, 55). Sp109 was transformed with pREP3X-bir1⁺ followed by random spore analysis and replica plating. Spores that were both His⁺ and Leu⁺, indicative of Δbir1 and nmt1-bir1⁺, were selected to generate the inducible Bir1p expression strain Sp157. The Sp157 strain is sensitive to thiamine, which represses Bir1p expression from the nmt1 promoter.

To generate a temperature-sensitive allele of bir1, the wild-type bir1⁺ gene was amplified from genomic DNA using PCR and then subcloned into the plasmid pUR19, which contains an autonomous replication sequence. The pUR19-bir1⁺ plasmids were subjected to hydroxylamine mutagenesis as described previously (reference 50 and references therein). Briefly, 10 μg of plasmid DNA was incubated in 500 μl of mutagenesis buffer (1 M hydroxylamine, 50 mM sodium pyrophosphate, pH 7.0, 2 mM EDTA, 100 mM NaCl) at 75°C for 6 to 10 min. Mutagenized plasmids were desalted by use of a QIAEX II gel extraction kit (QIAGEN) and transformed into Sp157. Replica plating was used to identify transformants that were His⁺, Ura⁺, Leu⁻, and temperature sensitive, indicating Δbir1 cells that were kept alive by a temperature-sensitive bir1 mutant expressed from a plasmid. The plasmids were isolated and retransformed into Sp157 to confirm that the phenotype resulted from mutation of bir1. One temperature-sensitive allele, bir1-46, was generated. To replace the endogenous bir1⁺ gene with bir1-46, a linear DNA fragment that contained bir1-46 was transformed into Sp157. Transformants were plated on solid rich medium (YES) to repress Bir1p expression from the nmt1 promoter and enrich for cells containing bir1-46 in place of Δbir1. Transformants were replica plated and His⁻ and Leu⁻ colonies were selected to generate strain Sp287.

To confirm that the wild-type bir1⁺ gene was replaced by the bir1-46 allele, the bir1-46 strain was crossed to a wild-type strain, and the resulting diploid was analyzed by tetrad dissection. Tetrads showed 2:2 segregation of wild-type to temperature-sensitive spores (data not shown). In addition, we performed Southern blot analysis using both wild-type and bir1-46 genomic DNA. Both genomic DNA samples showed an identical hybridization pattern (data not shown), indicating that the wild-type bir1⁺ allele is precisely replaced by the bir1-46 mutant allele on the chromosome.

To conduct the high-copy-number suppressor screen, Sp287 (bir-46) was transformed with either a cDNA or genomic DNA library. Transformants were incubated at 28°C for 20 h followed by further incubation at 34°C for 5 days. Colonies grown at 34°C were streaked out on YES plates and cultured at 28°C for 4 to 5 days to lose the plasmids containing library DNA, followed by replica plating and incubation at 34°C to confirm that growth at restrictive temperature was dependent on the library DNA.

The cDNA and genomic DNA of psf2⁺ were amplified from wild-type genomic DNA, as psf2⁺ has no introns. The psf2⁺ cDNA was subcloned into pDS672a to generate the Psf2p-myc fusion construct. To generate an S. pombe strain that contains the psf2 deletion in the chromosome but is kept alive by ectopically expressed Psf2p-myc, one copy of psf2⁺ was disrupted with his3⁺ in a wild-type diploid strain and then transformed with pDS672a-psf2⁺-myc. Random spore analysis and replica plating identified spores that were both His⁺ and Ura⁺, indicative of Δpsf2 and nmt1-psf2⁺-myc, to generate the Psf2p-myc expression strain (Sp499). To deplete Psf2p, Sp499 was cultured in the presence of 5 μg/ml thiamine.

The pic1⁺/Δpic1 strain (Sp483) was generated by deleting one copy of pic1⁺ in a diploid strain using one-step disruption with his3⁺. The disruption was confirmed by genomic PCR (data not shown).

A diploid strain with one copy of sld5⁺ deleted (Sp605) was also generated using one-step disruption with his3⁺ and confirmed by genomic PCR (data not shown).

Spore germination and immunofluorescence staining.

Spore germination was performed as described previously (37). Briefly, diploid cells were cultured in 10 ml YES at 32°C to late log phase (optical density at 595 nm [OD₅₉₅], 0.8 to 1.0). Cells were collected by centrifugation, and then the pellet was resuspended in 200 ml of ME mating medium followed by 3 days incubation at 25°C to induce sporulation. Spores were collected by centrifugation and then resuspended in 20 ml of 2.5% Glusulase solution followed by incubation at 28°C for 24 h. Glusulase-treated spores were washed five times with rinse buffer (0.17% yeast nitrogen base) followed by centrifugation through 40 ml of 25% glycerol. Purified spores were cultured in synthetic medium without histidine for 16 h and then collected for immunofluorescence staining.

Yeast immunofluorescence staining was conducted as previously described (19) except that Zymolyase 100T (Seikagaku) was used to digest cell walls (0.5 mg/ml for 10 to 13 min at room temperature). Cells were incubated with an anti-α-tubulin monoclonal antibody (clone B-5-1-2, 1:3,000 dilution; Sigma) overnight at room temperature and then washed three times with PEMBAL buffer {100 mM PIPES [piperazine-N,N′-bis(2-ethanesulfonic acid)], pH 6.9, 1 mM EDTA, 1 mM MgSO₄, 1% bovine serum albumin, 0.1% NaN₃, 100 mM lysine}. Cells were then incubated with Alexa Fluor 594-conjugated anti-mouse immunoglobulin G (IgG; 1:600 dilution; Molecular Probes) for 3 to 4 h at room temperature and then washed again three times with PEMBAL. DNA was stained with DAPI (4′,6′-diamidino-2-phenylindole). Cells were viewed using a fluorescence microscope (Olympus 1X70), and images were acquired and deconvolved using DeltaVision software (Applied Precision).

Antibody production.

Polyclonal antisera against S. pombe Pic1p, Psf2p, and Bir1p were raised in rabbits using His-tagged Pic1 (first 200 amino acids), His-tagged Psf2p (full length), and His-tagged Bir1p (first 330 amino acids), respectively, as antigens. Anti-Pic1p, anti-Psf2p, and anti-Bir1p antibodies were affinity purified as described previously (21), with minor modifications. In short, GST-Pic1p, GST-Psf2p, and GST-Bir1p fusion proteins were individually purified and immobilized on glutathione-agarose beads using dimethylpimelimidate (5 mg/ml). Polyclonal antisera against Pic1p, Psf2p, or Bir1p were diluted 1 to 6 in phosphate-buffered saline (PBS) and passed through Pic1p beads, Psf2p beads, or Bir1p beads, respectively, followed by washes using PBS and PBS with 500 mM NaCl. Bound antibodies were eluted with 10 volumes of 0.2 M ethanolamine (pH 11), followed by 0.1 M glycine (pH 2.9). Eluted antibodies were neutralized using 1 M Tris-HCl (pH 7) and then dialyzed against PBS overnight. For immunofluorescence staining, anti-Pic1p antibodies (1.0 μg/ml), anti-Psf2p antibodies (1.5 μg/ml), or anti-Bir1p antibodies (1.0 μg/ml) were used in overnight incubations and detected with anti-rabbit Alexa Fluor 488 antibodies (1:600 dilution; Molecular Probes).

Polyclonal antiserum against human PSF2 was raised in rabbits with full-length PSF2 as antigen. Anti-PSF2 antibodies were affinity purified using the modified protocol described above. To localize endogenous human PSF2, purified anti-PSF2 antibodies (0.1 to 0.5 μg/ml) were used as the primary antibody and detected with Alexa 488-conjugated anti-rabbit IgG antibodies (1:500 dilution; Molecular Probes) as the secondary antibody.

Cell culture, immunofluorescence staining, and small interfering RNA (siRNA).

The full-length human PSF2 cDNA was amplified from a HeLa cDNA library (20) by PCR and then subcloned into the plasmid pEGFP-N1 (Clontech) to generate the PSF2-EGFP fusion construct. HeLa cells cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS) were grown in 6-well plates with glass coverslips to ∼30% confluence. Cells were transfected with the PSF2-EGFP fusion construct using Effectene reagent (QIAGEN) according to the manufacturer's protocol (1:25 ratio).

Cells were fixed, permeabilized, and blocked as previously described (24). The coverslips were incubated with primary antibodies for 1 h at room temperature. Primary antibodies used were as follows: mouse anti-α-tubulin monoclonal antibody (1:2,000 dilution, clone B-5-1-2; Sigma), rat anti-α-tubulin monoclonal antibody (1:100 dilution, clone YOL 1/34; Oxford Biotechnology), or human polyclonal anticentromere antibodies (1:50 dilution, ANA-C; Sigma). After four washes with PBS, coverslips were incubated with secondary antibody for 1 h at room temperature. Secondary antibodies used were as follows: Texas Red-conjugated anti-mouse IgG antibody (1:250 dilution; Southern Biotechnology Associates), Texas Red-conjugated anti-rat IgG (1:300 dilution; Jackson ImmunoResearch), or Alexa 488-conjugated anti-human IgG or anti-rabbit IgG (1:500 dilution; Molecular Probes). After three washes with PBS, coverslips were incubated with PBS containing Hoechst dye (0.5 mg/ml) for 10 to 15 min at room temperature to stain the DNA. Cells were viewed using a fluorescence microscope (Olympus 1X70), and images were acquired and deconvolved using DeltaVision software (Applied Precision).

For videomicroscopy, a HeLa cell line expressing histone H2B-GFP (26) was cultured in polylysine-coated Delta T dishes (0.17-mm diameter; Bioptechs Inc.), followed by transfection using buffer or siRNA as described above. Transfected cells were synchronized with a double block of 2 mM thymidine. Twelve hours before capturing cell images, the culture medium was switched to CO₂ independent medium (Gibco/Invitrogen) with 10% FBS. Cells were then viewed using a fluorescence microscope (Olympus 1X70), and images were acquired every minute using the time-lapse program of the DeltaVision software (Applied Precision). The cell culture was maintained at 37°C using the Bioptechs heating system.

To deplete PSF2 in human cells, siRNA oligonucleotides were designed to match a region of the amino terminus of the PSF2 coding region (nucleotides 111 to 129; sense, CCCUGGUUUACCCGUGGAA-deoxyribosylthymine [dT]dT; antisense, UUCCACGGGUAAACCAGGG-dTdT; Dharmacon Research). A second set of siRNA oligonucleotides against nucleotides 334 to 352 were designed (sense, CGAAGCUCCUGUUAAAUGA-dTdT; antisense, UGAUUUAACUGGUGCUUCG-dTdT). The siRNA oligonucleotides were deprotected and annealed according to the manufacturer's protocol. siRNA transfection into HeLa cells was performed as previously described (14), with minor modifications. Briefly, HeLa cells cultured with antibiotic-free Dulbecco's modified Eagle's medium and 10% FBS were seeded in 12-well plates (15,000 cells in 1 ml medium per well). For each well, 6 μl of 20 μM annealed siRNA was mixed with 100 μl of OptiMem I (GibcoBRL/Invitrogen). In a separate tube, 6 μl of Oligofectamine (GibcoBRL/Invitrogen) was mixed with 24 μl of OptiMem I. After 7 min of incubation at room temperature, the two solutions were combined, mixed gently by inversion, and incubated for another 20 min at room temperature. OptiMem I was then added to bring up the final volume to 200 μl, and this mixture was added to cultured cells. Transfection efficiency was 80 to 90%, as determined by fluorescein isothiocyanate-conjugated 21-mer DNA oligonucleotides (data not shown).

Three days after transfection, total RNA was isolated using TRIZOL reagent according to the manufacturer's protocol (GibcoBRL/Invitrogen) and used for reverse transcription-PCR (RT-PCR). Reverse transcription was conducted in a 10-μl reaction mixture, using 0.5 μg of total RNA and Superscript II reverse transcriptase (GibcoBRL/Invitrogen) according to the manufacturer's protocol. After reverse transcription, 1.5 μl of RT reaction mixture was subjected to PCR (0.2 mM deoxynucleoside triphosphate and 0.1 μM of forward and reverse primers) using Taq DNA polymerase (QIAGEN). The PCR solution was incubated at 93°C for 3 min, followed by 30 cycles of amplification (92°C for 35 s, 52°C for 40 s, and 72°C for 1 min).

Protein lysates from HeLa cells treated with buffer only or PSF2 siRNA were prepared by lysing cells in modified CSK buffer (10 mM PIPES, pH 6.8, 100 mM NaCl, 300 mM sucrose, 1 mM MgCl₂, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 units/ml aprotinin, 50 μM NaF, 50 μM β-glycerophosphate, 0.5% Triton X-100) for 20 min at 4°C. Lysates were centrifuged at low speed for 5 min and then analyzed by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by immunoblotting with antibodies against human PSF2 and tubulin.

Fluorescence-activated cell sorting was carried out on HeLa cells treated with buffer only or PSF2 siRNA as follows. Cells were collected by trypsinization and then resuspended in 70% ethanol for 30 min. Cells were then washed in PBS and resuspended in PBS that contained 4 μg/ml propidium iodide. Cells were analyzed on a FACScan (Becton Dickinson) using Cell Quest software.

HeLa cells treated with buffer only or PSF2 siRNA were incubated with the microtubule inhibitor vinblastine (VBL; Sigma) for 14 h. Cells were then collected, immunostained with the anticentromere antibody ANA-C (Sigma) used at 1:50, and counterstained with Hoechst dye to detect the DNA. At least 100 cells per each of three experiments were scored for cells in metaphase.

HeLa cells treated with buffer only or PSF2 siRNA were incubated in the presence or absence of nocodazole and then stained with anti-CENP E antibodies (59); DNA was counterstained with Hoechst dye. The distance between sister kinetochores in at least 10 cells per condition per experiment was measured using DeltaVision software.

RESULTS

Bir1p is required for chromosome structure and segregation. To study Bir1p function in S. pombe, we generated a temperature-sensitive allele, bir1-46, and characterized its phenotypes. The bir1-46 cells did not form colonies at 34°C and displayed a slow-growth phenotype even at 28°C, a temperature that is fully permissive for wild-type cells (Fig. 1A). As with cells in which Bir1p was depleted (43, 55; data not shown) or that had a bir1⁺ degron mutant, in which the Bir1p protein was inactivated by degradation upon shift to restrictive temperature (45), bir1-46 cells cultured at restrictive temperature exhibited chromosome missegregation involving unequal nuclear division and cells with multiple DAPI-stained bodies characteristic of lagging chromosomes or chromatids (Fig. 1B). Some bir1-46 cells grown at restrictive temperature exhibited the cell untimely torn (cut) phenotype (data not shown). bir1-46 is a recessive allele that contains a single mutation that changes amino acid residue 976 from cysteine to tyrosine. This mutation maps to a predicted coiled-coil region near the C terminus and, interestingly, lies very close to the mutation in another temperature-sensitive allele of bir1, cut17-275 (A990T) (Fig. 1C) (38).

FIG. 1.

Characterization of S. pombe bir1-46. (A) The bir1-46 mutant is temperature sensitive. Wild-type (Sp1) and bir1-46 (Sp287) strains were grown on YES plates for 3 days at permissive temperature (28°C) or restrictive temperature (34°C). The bir1-46 strain exhibits slow growth at permissive temperature and displays little or no growth at the restrictive temperature. (B) bir1-46 cells exhibit chromosome missegregation at the restrictive temperature. Wild-type (Sp1) and bir1-46 (Sp287) cells were cultured at restrictive temperature (36°C) for 4 h, and then cells were stained with DAPI to detect DNA. Examples of cells with lagging chromosomes or unequal chromosome segregation are shown in the bir1-46 panels. Scale bar, 10 μm. (C) Protein sequence alignment of the C-terminal helical/coiled coil of human (Hs) Survivin (residues 93 to 142) with S. cerevisiae (Sc) Bir1p and S. pombe (Sp) Bir1p. Bold letters indicate conserved amino acid residues. (D) Wild-type (Sp1), bir1-46 (Sp287), and cut17-275 (Sp613) cells were cultured at 22°C until mid-log phase (OD₅₉₅, 0.4 to 0.6). Cells were reinoculated into YES medium to an OD₅₉₅ of 0.1 and cultured at 36°C. Every 2 h, aliquots of cells were taken and the cell density (OD₅₉₅) was measured. (E) For the 0 and 2-h time points, aliquots of wild-type, bir1-46, and cut17-275 cells were collected and stained for DAPI and tubulin to detect DNA and spindles. One hundred anaphase cells per strain were scored to determine the ratio of defective to normal chromosome segregation. (F) Examples of tubulin (red) and DAPI (blue) staining in cells from the 2-h time point from panel E.

To compare the mitotic defects of the two mutant alleles of bir1, we analyzed the phenotypes of bir-46 and cut17-275 at the restrictive temperature (36°C). The growth curves of bir1-46 and cut17-275 cells leveled off after 2 to 4 h at 36°C, consistent with a delay or arrest in cell division (Fig. 1D). Notably, after just 2 h at 36°C, all of the mitotic bir1-46 and cut17-275 cells exhibited chromosome missegregation (Fig. 1E and 1F). In contrast, wild-type cells continued to grow and retained normal cellular morphology even after 8 h at 36°C (Fig. 1D, 1E, and 1F). We conclude that both mutant alleles of bir1, bir1-46 and cut17-275, cause similar defects in chromosome segregation during mitosis.

To examine whether Bir1p loss of function affects spindle microtubules, we treated bir1-46 cells with the microtubule inhibitor thiabendazole (TBZ). Even at the permissive temperature, bir1-46 cells were unable to grow in the presence of TBZ (data not shown), similar to cut17-275 cells (38). Moreover, we found that bir1-46 cells at permissive temperature were sensitive to DNA damage induced by UV and the ribonucleotide reductase inhibitor hydroxyurea (HU) (data not shown), as reported for cut17-275 cells (38). These data suggest that Bir1p may affect chromosome structures required for DNA repair or for interactions with spindle microtubules.

Pic1p and Psf2p suppress bir1-46.

To identify genes functioning in concert with bir1⁺, we performed a high-copy-number suppressor screen to isolate genes that complemented the growth defect of bir1-46 at the restrictive temperature. We identified a genomic DNA fragment that restored the growth of bir1-46 cells at 34°C (data not shown). This DNA fragment contained three complete coding regions: arg2⁺, ura5⁺, and a novel gene that we named bsh3⁺ (Bir1 suppressor high copy number). Arg2p and Ura5p are involved in arginine and uracil synthesis, respectively. During the course of this study, we independently identified bsh3⁺ as psf2⁺ (SPBC725.13c), the homolog of budding yeast PSF2 (16). Deletion of S. pombe bsh3⁺/psf2⁺ from the complementing library DNA fragment abolished rescue of the bir1-46 growth defect at 34°C (data not shown), indicating that psf2⁺ is a high-copy-number suppressor of bir1-46. Moreover, high-copy-number expression of the psf2⁺ gene alone was able to rescue the growth defect of either of the bir1 mutants, bir1-46 or cut17-275, at restrictive temperature (Fig. 2A and 2B). High-copy-number psf2⁺ also suppressed the UV and TBZ sensitivities of bir1-46 (data not shown).

FIG. 2.

Identification of high-copy-number suppressors of bir1-46. (A) psf2⁺ and pic1⁺ suppress bir1-46 temperature sensitivity. bir1-46 cells (Sp287) were transformed with pUR19 (vector alone), pUR19-bir1⁺, pUR19-psf2, or pUR19-pic1⁺. Transformants were streaked on YES plates followed by incubation at either 28°C or 34°C for 3 days. The right side of the panel indicates the arrangement of strains on the YES plates. (B) psf2⁺ suppresses cut17-275. The cut17-275 strain was transformed with pUR19 (vector only), pUR19-bir1⁺, or pUR19-psf2⁺. Transformants were streaked on YES plates followed by incubation at either 22°C or 34°C for 3 days. The right side of the panel indicates the arrangement of strains grown on the YES plates.

In addition to the high-copy-number suppressor screen, we also performed reversion analysis to identify extragenic suppressors of the TBZ sensitivity of bir1-46. In the course of this screening, we found that overexpression of the S. pombe INCENP homolog pic1⁺ (SPBC336.15-SPBC685.01) complemented the growth defect of bir1-46 at 34°C (Fig. 2A). We had previously observed that overproduction of the Aurora B protein kinase Ark1p suppresses bir1-46 temperature sensitivity (33). These genetic interactions are consistent with the model in which Bir1p interacts with Pic1p and Ark1p in a conserved protein complex (33, 41, 44).

Pic1p is essential for chromosome segregation and cytokinesis.

We first identified S. pombe Pic1p in a two-hybrid screen for proteins that interact with Ark1p (33). To further characterize Pic1p, we carried out tetrad analysis to determine whether pic1⁺ is an essential gene. Tetrads from a diploid strain disrupted for one copy of pic1⁺ exhibited a 2:2 ratio of viable to inviable spores or only a single viable spore (Fig. 3A). In all cases, the viable spores were His⁻ (data not shown), indicating that they contained the wild-type pic1⁺ gene and that pic1⁺ is essential for cell viability. To examine the phenotypes of Δpic1 cells, we induced a diploid pic1⁺/Δpic1::his3⁺strain to sporulate and then selectively germinated the Δpic1::his3⁺ spores in medium lacking histidine. The Δpic1 spores generated cells with the cut phenotype as well as cells that were elongated, cells that contained multiple DAPI-stained bodies, and cells with multiple septa (Fig. 3B). These phenotypes indicate that pic1⁺ is important for equal chromosome segregation and cytokinesis.

FIG. 3.

Pic1p is an essential chromosomal passenger protein required for equal segregation. (A) pic1⁺ is essential. Diploid S. pombe cells with one copy of pic1 deleted (Sp483) were subjected to sporulation followed by tetrad dissection. Spores were separated and grown on YES plates at 28°C for 5 days; spores from a single tetrad are placed vertically in this figure. (B) Δpic1 cells display defects in chromosome segregation and septation. pic1⁺/Δpic1 diploid cells (Sp483) were sporulated in medium lacking histidine to selectively germinate spores with deletions of Δpic1. After 24 h, cells were stained with DAPI to examine nuclear morphology. Shown are four examples of aberrant nuclear morphology in Δpic1 cells: cut cells, unequal nuclear division, multiple DAPI-stained bodies, and multiple septation. (C) Pic1p exhibits localization typical of a chromosomal passenger protein. Affinity-purified anti-Pic1p antibodies (green) were used to stain wild-type (Sp1) and bir1-46 (Sp287) cells grown at the restrictive temperature. Cells were also stained for tubulin (red) and DAPI (white in the DAPI images and blue in the merged images) to visualize the spindles and DNA, respectively. Shown are representative examples of cells in interphase, metaphase, anaphase, and telophase.

Both Bir1p and Ark1p associate with the kinetochore during metaphase, then translocate to the elongating spindle during anaphase and remain at the spindle midzone at cytokinesis (38, 44). Importantly, the localization of Ark1p is dependent on Bir1p (41, 44). We analyzed the localization of Pic1p in wild-type and bir1-46 mutant cells to determine whether Pic1p displays a localization pattern similar to that of Bir1p and Ark1p and whether this localization requires Bir1p. We found that during metaphase, the endogenous Pic1p protein localized to a single nuclear focus coincident with the DAPI-stained chromosomes and then associated with the elongating spindle during anaphase (Fig. 3C). Notably, Pic1p did not localize properly in bir1-46 mutant cells at restrictive temperature, with most cells lacking discrete Pic1p staining (Fig. 3C). Taken together, these results imply that Pic1p is a chromosomal passenger protein that forms a functional complex with Bir1p and Ark1p and that formation or stabilization of this complex is impaired in the bir1-46 mutant.

Psf2p is an essential protein that affects chromosome segregation.

The novel suppressor of bir1-46 that we identified, S. pombe psf2⁺, encodes a 183-amino-acid protein without recognizable motifs or domains. There are Psf2p homologs in other species, including a human protein (DC5/CGI-122/HSPC037) that we have also named PSF2 (see below), S. cerevisiae (Psf2p), Xenopus laevis (Psf2p), Mus musculus (AK014776) and Caenorhabditis elegans (F31C3.5). The S. cerevisiae and Xenopus Psf2p proteins, as well as S. pombe Psf2p, are known to affect DNA replication (16, 27, 30, 51).

As an initial characterization of Psf2p, we tested whether S. pombe psf2⁺ is essential for cell viability. We deleted one copy of psf2⁺ in a diploid strain and carried out tetrad analysis. About half of the tetrads analyzed contained two viable spores and two inviable spores, while the remaining tetrads contained only one viable spore (Fig. 4A). psf2 mutants exhibit defects in meiotic chromosome segregation (16), suggesting that the single viable spores in the tetrads may result from a dosage effect. All of the viable spores were His⁻ (data not shown), indicating that they contained the wild-type psf2⁺ gene. We conclude that psf2⁺ is essential for cell viability.

FIG. 4.

Psf2p is an essential protein required for chromosome segregation. (A) psf2⁺ is an essential gene. Tetrad analysis of diploid S. pombe cells with one copy of psf2 deleted (Sp476). Separated spores were grown on YES plates at 28°C for 5 days; spores from each tetrad are arranged vertically in this figure. (B) Deletion phenotypes of Δpsf2. Spore germination of psf2⁺/Δpsf2 diploids (Sp476) was carried out to selectively allow growth of Δpsf2 spores. The Δpsf2 cells were stained with DAPI to detect chromosomal DNA; the small, DAPI-stained bodies are mitochondrial DNA. Shown are examples of aberrant nuclear morphology in Δpsf2 cells such as cut cells, unequal nuclear division, and cell elongation. (C) Depletion of Psf2p. Psf2p-myc cells (Sp499) were cultured in either nondepleting or depleting medium (containing thiamine) for 36 h. Total cell extracts were prepared, and protein lysates were analyzed by 13% SDS-PAGE followed by immunoblotting using affinity-purified anti-Psf2p antibodies (top panel). The transferred membrane was stripped and then reprobed with an antitubulin antibody as a loading control (bottom panel). Lane 1, whole-cell extract (30 μg) prepared under conditions where Psf2p-myc is expressed; lane 2, whole-cell extract (30 μg) prepared at 36 h after Psf2p depletion by addition of thiamine to the medium. (D) Depletion of Psf2p results in chromosome missegregation. Psf2p-myc cells (Sp499) were cultured in medium containing thiamine for 36 h to repress Psf2p expression. Cells were then collected and stained for tubulin (red) and DAPI (blue) to detect spindles and DNA. A black and white image of the DAPI staining is shown on the right. The arrow indicates a cut cell. (E) The temperature-sensitive psf2-209 mutant arrests cell growth. Wild-type (Sp1) and psf2-209 (Sp631) cells were cultured at 28°C until mid-log phase (OD₅₉₅, 0.4 to 0.6). Cells were reinoculated into YES medium to an OD₅₉₅ of 0.05 and cultured at 36°C. Every 4 h, aliquots of cells were taken and the cell density (OD₅₉₅) was measured. (F) Chromosome missegregation in the psf2-209 mutant at restrictive temperature. For the 0-, 8-, 12-, and 16-h time points in the time course shown in panel E, wild-type (Sp1) and psf2-209 (Sp631) cells were stained with DAPI to visualize the DNA. The psf2-209 cells at the 8-, 12-, and 16-h time points are elongated and exhibit chromosome missegregation. (G) Time course of chromosome missegregation in psf2-209 cells. Wild-type (Sp1) and psf2-209 (Sp631) cells were synchronized in G₂ using lactose gradients, and then cells were released to restrictive temperature (36°C). Aliquots of cells were collected every 20 min, and DNA was stained with DAPI to monitor progression through the cell cycle and nuclear morphology. Binucleates mark cells in mitosis, and septation indicates cells in S phase. Cells containing unequal chromosome segregation, lagging chromosomes, or cut cells were classified as missegregation events.

To investigate whether Psf2p affects chromosome segregation, we conducted a spore germination assay on diploid cells heterozygous for Δpsf2. psf2⁺/Δpsf2::his3⁺ diploid cells were induced to sporulate, and then the Δpsf2 spores were germinated in medium without histidine. We observed chromosome missegregation events in 65% of the mitotic Δpsf2 cells (Fig. 4B and data not shown), including cut cells, cells with unequal nuclear division, and cells with multiple DAPI-stained bodies characteristic of lagging chromosomes.

We also investigated the phenotypes associated with depletion of Psf2p. Although we were unable to epitope tag Psf2p at its endogenous locus, we constructed a strain in which a Psf2p-myc fusion protein was overexpressed under control of the nmt1 promoter. This overexpressed fusion protein complemented the inviability of the Δpsf2 strain (data not shown). Under repressing conditions (36 h in the presence of thiamine), little or no Psf2p-myc was detected by immunoblotting, indicating that the protein was significantly depleted (Fig. 4C). Depletion of Psf2p led to the formation of microcolonies that could not be further propagated (data not shown), with cells that were elongated or that displayed defects in chromosome segregation, including failure to segregate DNA on an elongated spindle, unequal nuclear division, and cells with multiple DAPI-stained bodies (Fig. 4D). In addition, a small fraction of Psf2p-depleted cells exhibited cut or multiple-septation phenotypes (data not shown).

Recently, we identified a temperature-sensitive allele of S. pombe psf2⁺, psf2-209, in a screen for mutants that blocked DNA rereplication (16). The psf2-209 mutant (R133K) itself displays a delay in S-phase entry or progression (16). At restrictive temperature (36°C), we found that the growth rate of psf2-209 cells was delayed relative to that of wild-type cells (Fig. 4E). Moreover, most psf2-209 cells became elongated with abnormal nuclear morphology, exhibiting characteristics such as unequal chromosome segregation and multiple DAPI-stained bodies by 8 h at restrictive temperature, although wild-type cells retained a normal nuclear morphology (Fig. 4F) (16). Thus, Psf2p function is important for equal chromosome segregation.

Psf2p could have a direct role in promoting accurate mitosis or could affect chromosome segregation indirectly through its S-phase role in DNA replication. To distinguish these possibilities, we synchronized wild-type and psf2-209 cells in G₂ phase using lactose gradients and then monitored cell cycle progression and cellular morphology in cells released to restrictive temperature. Wild-type cells maintained normal cellular morphology as they passed through mitosis into S phase (Fig. 4G). In contrast, chromosome missegregation was evident in a small subset of the psf2-209 cells released to restrictive temperature, and the frequency of missegregation events increased as cells progressed through the first mitosis (Fig. 4G). We observed a threefold increase in chromosome missegregation (5% to 15%) as psf2-209 cells passed from G₂ through the first mitosis at restrictive temperature (Fig. 4G). However, the majority of psf2-209 cells did not display these phenotypes until 8 h at restrictive temperature (Fig. 4F) (16). As an alternative means to test the role of Psf2p in chromosome segregation, we constructed a cdc25-22 mutant strain that contained the repressible nmt-Psf2p-myc fusion protein. We repressed Psf2p expression by the addition of thiamine and then shifted cells to restrictive temperature (36°C) to arrest the cells at G₂/M. We collected the G₂/M cells by lactose gradient centrifugation and then released them back to permissive temperature (25°C) to monitor cell morphology and chromosome missegregation. Cells in which Psf2p expression had been shut off by the addition of thiamine displayed a similar, slight increase in the number of cells with chromosome missegregation in the first mitosis (14% compared with 3% in cells where Psf2p expression was not shut off) (data not shown). The small fraction of cells with chromosome missegregation might, however, arise from S-phase cells where Psf2p was inactivated before DNA replication was completed. Thus, it remains unclear whether Psf2p affects mitotic chromosome segregation directly or indirectly through its S-phase role.

Psf2p is required for Bir1p localization.

To investigate the functional relationship between Psf2p and Bir1p, we tested whether Psf2p affects Bir1p localization. We raised anti-Bir1p antibodies and confirmed their specificity by immunoblot analysis. Although we were unable to reliably detect Bir1p from wild-type extracts, a protein of the size predicted for Bir1p was observed in extracts prepared from cells with mutations of the proteasome subunit Mts3p (17) and in extracts prepared from cells overproducing Bir1p (Fig. 5A). Bir1p may normally be present only at low levels within the cell and may undergo rapid, proteasome-dependent degradation in wild-type cells.

FIG. 5.

Psf2p is required to localize Bir1p. (A) Anti-Bir1p antibodies recognize Bir1p. Wild-type (Sp1) and mts3 (Sp367) cells and wild-type cells overexpressing Bir1p (Sp157) were grown at the permissive temperature until mid-log phase (OD₅₉₅, 0.4 to 0.6). Cells were then diluted into YES medium to an OD₅₉₅ of 0.2 and cultured at either the permissive or restrictive temperature for an additional 6 h. Total cell extracts were prepared, and proteins were analyzed by 8% SDS-PAGE followed by immunoblotting with affinity-purified anti-Bir1p antibodies (top panel). The transferred membrane was stripped and then reprobed with an antitubulin antibody as a loading control (bottom panel). Lanes: wild type, whole-cell extract (30 μg) of the wild type cultured under the permissive condition (25°C); bir1-HA mst3 25°C, whole-cell extract (30 μg) of mts3 mutant cultured under the permissive condition (25°C); bir1-HA mst3 36°C, whole-cell extract (30 μg) of mts3 mutant cultured under the nonpermissive condition (36°C); wild type + bir1-HA, whole-cell extract (30 μg) from a wild-type strain overexpressing Bir1p (25°C). (B) Bir1p is a chromosomal passenger protein. Wild-type cells (Sp1) were grown to early log phase (OD₅₉₅ = 0.4), and cells were immunostained for Bir1p (green) and tubulin (red). DNA was counterstained with DAPI (white in the DAPI images and blue in the merged images). Cells in interphase, metaphase, anaphase, and telophase are shown. Arrowheads indicate sites of Bir1p accumulation. (C) Specificity of Bir1p localization. Anti-Bir1p (green) and antitubulin (red) immunostaining was carried out on Δbir1 cells germinated from the heterozygous diploid bir1⁺/Δbir1 (Sp109). DNA was counterstained with DAPI (white in the DAPI images and blue in the merged images). (D) Mislocalization of Bir1p in Δpsf2 cells. Δpsf2 spores germinated from strain Sp476 (psf2⁺/Δpsf2::his3⁺) were stained for Bir1p (green) and tubulin (red), followed by DAPI staining to detect DNA (white in the DAPI images and blue in the merged images). Bir1p localization is not detected in some cells (top). Cells that do contain Bir1p exhibit mislocalization involving association with one of the two nuclei (arrowhead) and along part of the mitotic spindle. (E) Depletion of Psf2p leads to mislocalization of Bir1p. Psf2p-myc was depleted by culturing strain Sp499 in medium containing thiamine for 36 h, and then cells were collected and stained for Bir1p (green) and tubulin (red) and counterstained with DAPI (white in the DAPI images and blue in the merged images). Two of the mitotic cells display Bir1p staining that is more extensive than the area of the nucleus, marked by DAPI staining (arrowheads). The other two mitotic cells exhibit no distinct Bir1p staining. The interphase cells display nuclear Bir1p staining. (F) Mislocalization of Bir1p in psf2-209 cells. psf2-209 cells (Sp631) were collected 8 h after the shift to restrictive temperature and then immunostained for Bir1p (green) and tubulin (red). DNA was visualized with DAPI (white in the DAPI images and blue in the merged images). Bir1p staining is undetectable in some cells (top). In cells with Bir1p staining, Bir1p localizes to one end of the spindle or forms aggregates that are unassociated with the DNA (middle). Interphase cells display nuclear localization of Bir1p.

We used these antibodies to analyze Bir1p localization within whole fission yeast cells. Consistent with previous reports (38, 44), we found that Bir1p exhibited predominantly nuclear localization during interphase (Fig. 5B) and concentrated to a discrete spot inside the nucleus during metaphase (Fig. 5B). Typical of chromosomal passenger proteins, Bir1p associated with the elongating mitotic spindle during anaphase (Fig. 5B) and then localized to the central region of the mitotic spindle during late anaphase and/or telophase (Fig. 5B). Bir1p staining was not observed in Δbir1 cells derived from germinating spores (Fig. 5C), indicating that the staining pattern we observed in wild-type cells was specific for Bir1p.

Notably, in cells where Psf2p was deleted or depleted, Bir1p was mislocalized in the majority (76%) of anaphase cells. In the absence of Psf2p, Bir1p localization was often undetectable in mitotic cells (Fig. 5D and E). In those cells that did exhibit Bir1p staining, Bir1p often formed a diffuse aggregate larger than the area of the nucleus or spindle, or appeared to localize to one of the two nuclei or at one end of the spindle (Fig. 5D and E). However, Bir1p nuclear localization was retained in at least some interphase cells (Fig. 5E). Similarly, in psf2-209 mutant cells at restrictive temperature, nuclear localization of Bir1p was observed in some interphase cells, but Bir1p was mislocalized in most anaphase and telophase cells (Fig. 5F). Thus, the localization or stabilization of Bir1p during mitosis requires Psf2p function.

We identified Sld5 (Spbp4H10.21c), another component of the GINS complex, in a two-hybrid screen for Psf2-interacting proteins (data not shown). To characterize sld5⁺, we generated a diploid strain in which one copy of the gene was deleted and carried out tetrad analysis (see Fig. S1A in the supplemental material). Only two or fewer spores from each tetrad were viable (see Fig. S1A in the supplemental material), and these spores were His⁻ (data not shown), indicating that they contained the wild-type sld5⁺ gene and that sld5⁺ is essential for cell viability. We carried out spore germination assays to examine chromosome segregation and localization of the Bir1p protein in Δsld5 cells (see Fig. S1B and C in the supplemental material). We found that deletion of sld5⁺ resulted in chromosome missegregation similar to that observed with deletion of psf2⁺ (see Fig. S1B in the supplemental material). Furthermore, the Bir1p protein was mislocalized in Δsld5 germinated spores (see Fig. S1C in the supplemental material). The similar phenotypes caused by loss of function of Sld5p or Psf2p strongly suggest that these proteins affect Bir1p localization and chromosome segregation through their role in replication as part of the GINS complex.

Psf2p is a nuclear protein.

The requirement for Psf2p in equal chromosome segregation and Bir1p localization during mitosis raised the possibility that Psf2p itself acts as a chromosomal passenger protein. We examined Psf2p localization using an epitope-tagged version of Psf2p (Psf2p-green fluorescent protein [GFP]) that we overproduced and that complemented the inviability defect of Δpsf2, suggesting that it encodes a functional protein (data not shown). We found that Psf2p-GFP exhibited nuclear localization (Fig. 6A). Interestingly, the Psf2p-GFP fusion protein also localized to the region connecting the nuclei in cells in anaphase and telophase, suggesting that Psf2p may associate with the elongating mitotic spindle (Fig. 6A). We also generated antibodies against Psf2p, which detected nuclear staining of overproduced, but not endogenous, Psf2p (Fig. 6B). However, we were unable to integrate epitope-tagged versions of psf2⁺ (psf2⁺-GFP and psf2⁺-myc) at the endogenous psf2⁺ locus under control of the native promoter, precluding further determination of whether the endogenous Psf2p protein acts as a chromosomal passenger protein.

FIG. 6.

Localization of Psf2p/PSF2. (A) Psf2p localization in fission yeast cells. Wild-type cells overexpressing Psf2-GFP (Sp500) were stained with DAPI to detect DNA. Shown are examples of Psf2p-GFP fluorescence in different stages of the cell cycle (G₁/S, G₁, or DNA synthesis phase of the cell cycle; A/T, anaphase/telophase; I, interphase). Arrows mark the region of the cells that correspond to the elongating mitotic spindle. (B) Anti-Psf2p antibodies detect overproduced Psf2p in the cell nucleus. Cells overexpressing Psf2p (Sp500) were immunostained with affinity-purified anti-Psf2p antibodies. DNA was counterstained with DAPI. (C) PSF2-GFP localization in HeLa cells. Twenty-four hours after transfection with PSF2-EGFP, HeLa cells were stained for tubulin (red) to visualize spindle microtubules and Hoechst dye (blue) to detect the DNA. Shown are examples of PSF2-EGFP nuclear localization in interphase and cytokinesis, spindle association at metaphase, and midbody localization at cytokinesis.

The human homolog of Psf2p, PSF2 (DC5/CGI-122/HSPC037) is 29.5% identical to the S. pombe protein and has not been characterized. We generated a PSF2-EGFP chimera that we ectopically expressed in HeLa cells to examine the cellular localization of human PSF2. The PSF2-EGFP fusion protein exhibited predominantly nuclear localization in interphase cells and during cytokinesis (Fig. 6C). In addition, we observed weak PSF2-EGFP fluorescence associated with the mitotic spindle during metaphase and anaphase and at the midbody during cytokinesis (Fig. 6C). However, we have been unable to detect endogenously expressed human PSF2 in HeLa cells (data not shown). In both fission yeast and mammalian cells, the endogenous Psf2p/PSF2 protein may be present only at low levels or at specific times during the cell cycle, or the epitopes recognized by the antibodies may be hidden when Psf2p/PSF2 is assembled into the GINS complex. Our studies suggest that overproduced Psf2p/PSF2 is a nuclear protein and may possibly act as a chromosome passenger protein.

Depletion of human PSF2 causes defects in chromosome congression and cell growth arrest.

To investigate the role of human PSF2 in chromosome segregation, we used siRNA (14) to deplete PSF2 from HeLa cells. We used RT-PCR to verify that PSF2 expression was inhibited. While PSF2 was readily amplified from total RNA prepared from cells mock transfected with buffer alone, little or no PSF2 was amplified from RNA prepared from cells treated with PSF2 siRNA (Fig. 7A). The total RNA prepared from both samples displayed similar quality and concentration (Fig. 7A). In addition, SURVIVIN was amplified in similar amounts from total RNA treated with buffer alone or with PSF2-siRNA (Fig. 7A). Depletion of the PSF2 protein was confirmed by immunoblot analysis of protein lysates prepared from cells treated with PSF2 siRNA (Fig. 7A). Thus, the PSF2 protein was efficiently and specifically depleted by its cognate siRNA.

FIG. 7.

Characterization of human PSF2. (A) siRNA-mediated depletion of PSF2 from HeLa cells transfected with either PSF2 siRNA or buffer alone as a control. (Left) Three days after transfection, total RNA samples were isolated and subjected to 1% agarose gel electrophoresis, followed by ethidium bromide staining. Lane 1: 0.5 μg of total RNA prepared from cells treated with buffer only; lane 2, 0.5 μg of total RNA prepared from cells treated with PSF2 siRNA. (Middle) Total RNA samples (0.5 μg) were subjected to RT-PCR using PSF2 DNA oligonucleotides and SURVIVIN DNA oligonucleotides as a control. PCR products were analyzed by 0.7% agarose gel electrophoresis followed by ethidium bromide staining. Lanes 1 and 2, full-length SURVIVIN amplified from total RNA samples prepared from cells treated with buffer only or PSF2 siRNA, respectively; lanes 3 and 4, full-length PSF2 amplified from total RNA samples prepared from cells treated with buffer only and PSF2 siRNA, respectively. (Right) Depletion of the PSF2 protein was assessed by immunoblotting of protein lysates prepared from cells treated with buffer only or PSF2 siRNA, using antibodies against the PSF2 protein and tubulin at 3 days after transfection. (B) PSF2 siRNA causes an increase in number of cells with sub-G₁ DNA content. One hundred hours after transfection, cells were fixed with cold 70% ethanol, resuspended in PBS, and stained with propidium iodide (4 μg/ml). Flow cytometry was performed with a Becton Dickinson FACScan, and data were analyzed using the Cell Quest software. (C) Aberrant nuclear morphology in PSF2 siRNA cells. One hundred hours after transfection, cells were stained with Hoechst dye to detect DNA. (D) Unaligned chromosomes in cells treated with PSF2 siRNA. Seventy-two hours after transfection, cells were immunostained for tubulin (red) and ANA-C (green) to detect spindles and centromeres, respectively, and counterstained with Hoechst dye to detect DNA. Cells were viewed with an Olympus microscope, and images were acquired and deconvolved using the DeltaVision system.

At 100 h after transfection, HeLa cells treated with PSF2 siRNA had increased only five- to sixfold in number (two to three generations), while cells treated with buffer only increased 20- to 22-fold (>5 generations) (data not shown), suggesting that depletion of PSF2 results in cell growth arrest. To examine DNA content, we performed fluorescence-activated cell sorting. Cells treated with only buffer exhibited a normal cell cycle profile in which the majority of cells had a 2N DNA content, indicative of G₁ phase (Fig. 7B). The cells treated with PSF2 siRNA had a similar profile, except that the number of cells with sub-G₁ DNA content was increased at least eightfold (Fig. 7B), possibly due to aneuploidy resulting from chromosome missegregation.

Microscopic analysis revealed that 63% of HeLa cells treated with PSF2 siRNA displayed abnormal or fragmented nuclei (micronuclei, Fig. 7C), while more than 90% of control cells retained normal nuclear morphology (Fig. 7C). We also observed unaligned chromosomes in PSF2-siRNA-treated cells (Fig. 7D), which may reflect a defect in chromosome congression or segregation. We observed similar phenotypes of cell growth arrest and chromosome missegregation with a second, independent siRNA oligonucleotide directed against PSF2, indicating that these phenotypes are due to specific depletion of the PSF2 protein (data not shown).

To further examine the chromosome missegregation phenotypes resulting from siRNA depletion of PSF2, we monitored cell cycle progression using live cell microscopy. In control cells stably expressing GFP fused to histone H2B, the chromosomes efficiently condensed, congressed to the metaphase plate, and then segregated equally to opposite spindle poles within 45 to 60 min (see Fig. S2A in the supplemental material). In contrast, in cells treated with PSF2 siRNA, individual chromosomes or chromatids displayed a defect in congression to the metaphase plate, and cell division was delayed by at least 2 h (see Fig. S2B in the supplemental material). The phenotypes of cells treated with PSF2 siRNA, taken together with the phenotypes of fission yeast cells mutant for Psf2p, strongly support a conserved function for Psf2p/PSF2 in promoting equal chromosome segregation and cell viability.

The live cell analysis of cells in which PSF2 was depleted suggested the presence of unattached or mono-oriented chromosomes. To test whether human PSF2 is involved in kinetochore attachment to spindle microtubules, we incubated HeLa cells treated with buffer only or PSF2 siRNA with the antimicrotubule drug vinblastine. Low concentrations of VBL cause subtle defects in microtubule organization and dynamics, such as a decrease in the number of microtubules attached to the kinetochore, and a decrease in kinetochore spindle tension (25). As a result, VBL treatment delays progression from metaphase to anaphase (48, 61). In the presence of low concentrations of VBL, a much higher percentage of metaphase cells were observed in the population treated with PSF2 siRNA than in that treated with control cells (Table 2). Cells depleted of human PSF2 thus exhibit increased sensitivity to low doses of VBL, suggesting that PSF2 may promote kinetochore attachment to the spindle microtubules.

TABLE 2.

Vinblastine sensitivity of PSF2 siRNA-treated HeLa cells

siRNA	Cells in metaphase (%)a
	0 nm VBL	1 nm VBL	0.5 nM VBL	0.1 nM VBL
Mock	0.3	13.6	1.7	1.5
PSF2	3.2	36.2	23.6	10.3

^aData are averages from three experiments.

We also measured the distance between sister kinetochores, as marked by CENP-E staining, in HeLa cells treated with PSF2 siRNA (Table 3). We found that at prometaphase, the distance between paired kinetochores was shorter in cells treated with PSF2 siRNA than in cells treated with buffer only (Table 3). However, during metaphase, there was no significant difference in the distance between paired kinetochores aligned at the metaphase plate, although the distance between sister kinetochores remained shorter in the unaligned chromosomes of PSF2-depleted cells (Table 3). We additionally examined the distance between sister kinetochores not under tension from the mitotic spindle by treating cells with nocodazole to depolymerize microtubules. The distance between sister kinetochores in nocodazole-treated HeLa cells was also decreased when PSF2 was depleted. Together, our studies of PSF2 depletion in HeLa cells suggest that human PSF2, like S. pombe Psf2p, is required for normal mitotic chromosome segregation and may affect kinetochore tension or attachment to the mitotic spindle.

TABLE 3.

Sister kinetochore distance in HeLa cells depleted of PSF2^a

siRNA	Average distance between sister kinetochores (μm)
	Nocodazole treated	Prometaphase	Metaphase aligned	Metaphase unaligned
Mock	0.63 ± 0.015 (53)	0.80 ± 0.020 (50)	1.18 ± 0.032 (50)	NA
PSF2	0.53 ± 0.010* (50)	0.62 ± 0.019* (50)	1.23 ± 0.027 (32)	0.79 ± 0.042 (27)

^a*, P < 0.0001. NA, not available. Numbers in parentheses are numbers of cells.

DISCUSSION

Bir1p as a model for Survivin function.

Vertebrate Survivin affects multiple aspects of cell division, including chromosome biorientation, attachment to spindle microtubules, cytokinesis, and apoptosis (reviewed in reference 2). Although fission yeast Bir1p is much larger than Survivin, both proteins have similar overall topologies, possessing one (Survivin) or two (Bir1p) BIR domains at the N terminus and a coiled-coil region at the C terminus. Interestingly, although these proteins display only weak sequence homology overall, and Survivin fails to complement the inviability defect of Δbir1 (H.-K. Huang and T. Hunter, unpublished data), Bir1p and Survivin also contain another, previously unrecognized conserved motif in a helical region at their very C termini (Fig. 1C) that appears to be functionally important.

The mutations in both bir1 mutants, cut17-275 and bir1-46, map to conserved residues in this C-terminal motif (Fig. 1C). While the Cut17-275 mutant protein localizes to the cell nucleus, it is unable to associate with the mitotic spindle or recruit the Aurora B protein kinase Ark1p (38). Similarly, the last 36 amino acids of human Survivin, which include most of the conserved residues in the C-terminal motif, are required for its localization (49). Interestingly, the C terminus of S. cerevisiae Bir1p is sufficient to rescue the minichromosome loss phenotype of Δbir1 null mutant cells and interacts with the kinetochore protein Ndc10 (60). Taken together, these data strongly suggest that the C-terminal region of Bir1p/Survivin has a conserved role in the localization and/or activity of kinetochore-associated proteins.

Bir1p/Survivin and chromosomal passenger complexes.

Ark1p localization and protein kinase activity depends on Bir1p (41, 44). We find that localization of Pic1p, another chromosomal passenger protein, also requires Bir1p. While we have been unable to demonstrate a physical interaction between Pic1p and Bir1p (H.-K. Huang, J. D. Leverson, and T. Hunter, unpublished data), Pic1p interacts with Ark1p in two-hybrid assays (33). Furthermore, both Pic1p (this work) and Ark1p (33) suppress bir1-46 temperature sensitivity. Loss of function of Pic1p also causes chromosome segregation defects similar to those observed in bir1 and ark1 mutants (33, 38, 41, 44, 45). These data suggest that Bir1p, Ark1p, and Pic1p function as part of a conserved chromosomal passenger protein complex, as previously described in other systems (7, 15, 23, 28, 33, 41, 47).

In vertebrates, the Aurora B kinase associates with INCENP even in the absence of Survivin (15), suggesting that INCENP functions as a regulatory subunit that directs Aurora B localization and/or activation. Indeed, the recently determined crystal structure of Aurora B bound to a fragment of INCENP revealed a partially activated state of the kinase, which was fully activated upon phosphorylation of INCENP (47). About half of the Aurora B/INCENP complexes in cells interact with Survivin and a novel, conserved protein, Borealin (15). Aurora B also interacts with the third Aurora kinase family member, Aurora C (35). Interestingly, the RCC1-like protein TD-60, which acts as a guanine exchange factor for the small G protein Rac1, is a chromosomal passenger protein that does not appear to associate with Aurora B or Survivin (15, 36). Current evidence suggests that multiple complexes of chromosomal passenger proteins exist in the cell, and these may contribute to distinct aspects of chromosome alignment and segregation.

Links between Bir1p and S phase.

How does overexpression of Psf2p, which has an essential activity in S phase, complement the defective function of Bir1p, which serves as a part of a chromosomal passenger complex in mitosis? In principle, Psf2p might suppress deficiency of Bir1p directly, by enhancing its mitotic function, or indirectly, by acting in S phase to promote assembly of structures required for Bir1p localization and function. A growing number of replication-associated proteins that affect mitosis are known. Mammalian ORC6, a component of the origin recognition complex involved in DNA replication initiation, displays a localization pattern characteristic of chromosomal passenger proteins, suggesting that it has a direct role in mitosis (42). In budding yeast, another origin recognition complex subunit, Orc2p, promotes mitosis by maintaining sister chromatid cohesion at centromeres after S phase (K. Shimada and S. Gasser, personal communication). In fission yeast, the protein kinase Hsk1p, which activates replication origins to fire, likely promotes mitosis indirectly by establishing centromere cohesion during S phase (4).

Although they might have distinct roles in mitosis, Psf2p or Sld5p loss of function results in similar chromosome segregation defects, and we favor the idea that these proteins affect the function of Bir1p and the Bir1p-Ark1p-Pic1p complex indirectly, through their S-phase roles as subunits of the GINS complex. In particular, we suggest that the GINS complex promotes efficient replication through centromeres, which replicate early in fission yeast (29). Loss of GINS function is associated with reduced firing of replication origins and delayed S-phase progression (16, 27, 30, 51). The timely replication of centromeres may be required for proper assembly of centromeric chromatin and multiprotein complexes such as cohesin and the kinetochore. Bir1p localization is known to depend on sister chromatid cohesion (38), which is established during S phase (54) and may be mediated by heterochromatin (5, 40). In this regard, we found that human Survivin interacts with the human cohesin subunit SA2 (Scc3) in a two-hybrid assay (H.-K. Huang and T. Hunter, unpublished data). In S. pombe, psf2-209 mutant cells display normal localization of the heterochromatin protein Swi6 and the kinetochore protein Ndc80 even at restrictive temperature (H.-K. Huang, J. M. Bailis, and T. Hunter, unpublished data), suggesting that gross assembly of centromeric chromatin assembly is normal. However, it remains possible that Psf2p/GINS affects the assembly of cohesin or other kinetochore components that in turn promote the localization or stabilization of the Bir1p passenger complex.

It is also possible that bir1 mutants sustain S-phase defects that are suppressed by Psf2p overproduction. S phase is slightly delayed in the cut17-275 mutant relative to wild type (38), and there may be region-specific defects in DNA replication (such as through centromeres) although bulk DNA synthesis is completed. The cut17-275 mutant cells (38), like bir1-46 cells and psf2-209 cells (J. M. Bailis and T. Hunter, unpublished data), are sensitive to HU, which inhibits replication fork progression. However, bir1-46 cells are competent for the replication checkpoint (J. M. Bailis and T. Hunter, unpublished data). Moreover, psf2 mutants require the Rad3 checkpoint protein kinase for viability (16). The inviability of bir1 and psf2 mutant cells in HU may result from defects in chromatin condensation or other aspects of chromosome structure that inhibit DNA repair (38; J. M. Bailis and T. Hunter, unpublished data).

Models for Bir1p function and its regulation by Psf2p.

One role of the Bir1p/Survivin complex may be to couple spindle microtubule attachment to sensing of tension (reviewed in reference 31). S. cerevisiae Bir1p binds the centromere-associated CBF3 complex (60), and recent evidence suggests that Bir1p and Sli5p (INCENP) may act to connect the CBF3 complex to spindle microtubules (S. Sandall and A. Desai, personal communication). Consistent with this possibility, S. pombe bir1 mutants are sensitive to the microtubule drug TBZ (38). Overproduction of Psf2p might suppress defects in Bir1p function by enhancing chromosome structures required for kinetochore-microtubule attachment or tension. We found that HeLa cells depleted of PSF2 were sensitive to the microtubule drug VBL and exhibited a shorter distance between sister kinetochores in prometaphase cells and nocodazole-treated cells. However, loss of Psf2p function in S. pombe did not confer sensitivity to spindle microtubule drugs, and we were unable to identify biochemical or genetic interactions between Psf2p and α- or β-tubulin or the kinetochore subunits Dis1p and Ndc80p (H.-K. Huang and T. Hunter, unpublished data).

We suggest that GINS-mediated replication of centromeres is critical for proper centromeric chromatin and kinetochore assembly and that subtle alterations in these structures may manifest in different ways in S. pombe than in vertebrate cells. We observed defects in chromosome segregation with loss of Psf2p/PSF2 function in both fission yeast and human cells. However, S. pombe psf2 mutants, unlike HeLa cells depleted of PSF2, were not sensitive to disruption of spindle microtubules. Moreover, while Bir1p was mislocalized in S. pombe cells deficient for the GINS components Psf2p and Sld5p, we did not observe equivalent mislocalization of human Survivin in HeLa cells depleted of PSF2 (H.-K. Huang and T. Hunter, unpublished data). In S. pombe cells, overproduction of Psf2p in bir1 cells may promote timely replication through centromeres or enhance centromeric cohesion, thereby stabilizing residual, partially functional Bir1p complexes at the kinetochore. Conversely, centromere replication may be inefficient or delayed when Psf2p function is lost, resulting in defects in chromatin condensation, cohesion, and Bir1p association with kinetochores. S. pombe kinetochores only attach a few microtubules (12); thus, reduction in microtubule binding or defects in kinetochore attachment are not sufficient to prevent anaphase but result in delayed anaphase and chromosome missegregation. In contrast, a reduction in kinetochore attachment in HeLa cells depleted of PSF2 may result in a defect in congression and initiation of anaphase. Future work should resolve how Psf2p/PSF2 activity contributes to kinetochore assembly and function in mitosis.