Characterization of a Fission Yeast SUMO-1 Homologue, Pmt3p, Required for Multiple Nuclear Events, Including the Control of Telomere Length and Chromosome Segregation

Fission yeast strains, media, and methods.

Fission yeast strains were grown and used as described by Moreno et al. (62). All fission yeast strains used in this study are listed in Table 1. Standard genetic methods and staining with 4′,6-diamidino-2-phenylindole (DAPI) were as described elsewhere (62). Cell number was determined with a Sysmex CDA-500 cell counter. Transformation of S. pombe was achieved by the lithium acetate (72) or the electroporation (35) method. Vegetative growth under nonselective conditions was in YPD (2% peptone, 1% yeast extract, 2% glucose) or YE (0.5% yeast extract, 3% glucose) medium, supplemented as required. The synthetic minimal medium used was EMM2 (60), supplemented as required. The thiamine-regulatable promoter nmt1 was repressed by adding 5 μg of thiamine per ml to EMM2 medium (55).

Cloning of a genomic fragment and the full-length cDNA of the pmt3⁺ gene.

A genomic fragment of the pmt3⁺ gene was cloned by screening of an ordered S. pombe cosmid library (61) with the pmt3⁺ cDNA obtained by two-hybrid screening. The full-length cDNA of the pmt3⁺ gene was isolated by the screening of a λZapII S. pombe cDNA library (from about 5 × 10⁴ plaques) (34), using plaque hybridization with the pmt3⁺ cDNA obtained by two-hybrid screening as a probe. Sequencing was carried out on both strands by using an ABI PRISM dye terminator cycle sequencing ready reaction kit (Perkin-Elmer). Multiple alignments were carried out by using the CLUSTAL W program (86).

Construction of the pmt3 deletion strain.

A 4.2-kb EcoRI genomic fragment containing the pmt3⁺ gene was subcloned from a pmt3⁺-containing cosmid after restriction mapping. The 1.8-kb ura4⁺ gene was inserted at the unique ApaLI site of a 4.2-kb pmt3⁺ genomic fragment which had been previously blunt ended and ligated to an XbaI linker. A 6.0-kb fragment carrying the pmt3::ura4 construct was transformed into diploid strain TP4-1D/TP4-5A (h⁺/h⁻ his2/+ leu1-32/leu1-32 ura4-D18/ura4-D18 ade6-M216/ade6-M210) (36). Stable transformants were isolated, and gene disruption was confirmed by Southern blot analysis. For analysis of the phenotype of haploids with a disruption of the pmt3⁺ gene, tetrad analyses were performed by standard methods (62).

Analysis of HU and UV responses.

To test the response to hydroxyurea (HU), wild-type (HM123 [h⁻ leu1-32]), pmt3Δ (JN630 [h⁻ leu1-32 ura4-D18 pmt3Δ::ura4]), cdc2-3w (JN632 [h⁻ leu1-32 cdc2-3w]), and cds1Δ (JN633 [h⁻ leu1-32 ura4-D18 cds1Δ::ura4]) cells were grown to a mid-log phase (0.5 × 10⁷ to 1 × 10⁷ cells/ml) in YE medium at 30°C. HU was added to the culture to a final concentration of 12 mM at time zero. Portions were taken and plated on YE plates, and cell viability was determined (66).

To test UV sensitivity, wild-type (HM123) cells, pmt3Δ cells (JN630), and pmt3Δ cells containing plasmid pREP1x-pmt3, which expresses pmt3⁺, were grown to mid-log phase (0.5 × 10⁷ to 1 × 10⁷ cells/ml) in EMM2 supplemented with leucine or EMM2 containing 5 μg of thiamine per ml (for pmt3Δ cells containing pmt3⁺ expression plasmid) at 30°C. Then, cells were plated in duplicate onto EMM2 plates supplemented with leucine or EMM2 plates containing thiamine (5 μg/ml) at 800 cells per plate and irradiated in a Stratalinker (Stratagene) at doses ranging from 0 to 150 J/m² (66). Cell viability was determined by counting colonies following incubation at 30°C for 4 to 6 days.

Determination of minichromosome loss rates.

Ch10 is a linear minichromosome that carries the sup3-5 tRNA gene, which suppresses the ade6-704 mutation (70). A colony color assay was performed to measure the stability of Ch10. S. pombe ade6-704 cells form red colonies. If the minichromosome Ch10 is stably maintained in the ade6-704 cells during colony formation, sup3-5 on Ch10 will suppress the ade6-704 mutation and the cells will form white Ade⁺ colonies. The rate of loss of Ch10 from wild-type (YM75 [h⁺ leu1-32 ura4-D18 ade6-704 Ch10]) and pmt3Δ (YM76 [h⁺ leu1-32 ura4-D18 pmt3Δ::ura4 ade6-704 Ch10]) cells under normal conditions was calculated by two different methods (67). Each experiment was carried out at least four times.

In the first method, YM75 and YM76 strains were grown in the absence of adenine to select for Ch10-harboring cells. Several colonies were dispersed in 1 ml of YE medium. Cell number was measured in a cell counter. The cells were then diluted into YE medium, sonicated briefly, and plated onto YE plates at a concentration of ∼10³ cells/plate. After 3 to 5 days growth at 30°C, the proportion of Ade⁻ (completely pink) colonies was calculated. The overall loss rates of Ch10 from wild-type and pmt3Δ cells were taken to be an average of the loss rates observed in four experiments.

In the second method, cultures of cells were grown under selective conditions to a density of 1 × 10⁶ to 3 × 10⁶ cells per plate to ensure maintenance of Ch10. The cells were then harvested, resuspended in an equal volume of YE medium, and incubated for three generations. Samples were removed at the beginning and the end of the incubation period, and cell number and the proportion of Ade⁻ segregants were determined as follows. The wild-type and pmt3Δ cells were plated on EMM2 plates supplemented with adenine (7 μg/ml), which renders Ade⁻ colonies pink. The rate of loss of Ch10 was calculated by using the following formula, previously described by Murakami et al. (67): rate of loss = 1−e^(1/n)lnR_n/R₀ where R₀ and R_n are the proportions of Ade⁺ cells at 0 and n generations after removal of selection, respectively.

Measurements of telomere length.

S. pombe chromosome DNA was prepared by a glass bead-phenol extraction protocol (29). Chromosome DNA was digested with ApaI, resolved on a 1% agarose gel, and transferred onto a nylon membrane (GeneScreen Plus; NEN). The 0.3-kb ApaI-EcoRI fragment of pAMP2 (52), which contains the S. pombe telomeric repeat sequences, was used as a probe for telomeric repeat sequences in the Amersham ECL system.

Expression of Pmt3p and HA₃ tagging Pmt3p in S. pombe.

The pmt3⁺ cDNA from the two-hybrid screen was digested with BamHI and XhoI and cloned into the BamHI-XhoI sites of pREP1 after inserting a XhoI linker into the SmaI site of the multicloning site downstream of the thiamine-regulatable nmt1 promoter (55). The resulting construct was termed pREP1x-pmt3. Tagging of Pmt3p with three copies of the hemagglutinin (HA) epitope was done as follows. The pmt3⁺ cDNA was amplified by PCR using oligonucleotide primers located at either end of the open reading frame, and the amplified DNA was digested with NotI and BamHI. Primers were 5′-TAGCGGCCGCATGTCTGAATCACCATCA-3′ (a sense primer to create a NotI site at the start codon of pmt3⁺ cDNA; the NotI site is underlined) and 5′-CCGGATCCAAGGCATAGATGGGTGCA-3′ (an antisense primer to create a BamHI site downstream of pmt3⁺ cDNA; the BamHI site is underlined). The NotI-BamHI fragment of pmt3⁺ cDNA was cloned into the NotI-BglII sites of the strongest nmt1 promoter (nmt) HA₃-tagging vector pSLF173 (23) after converting an ura4 marker to a LEU2 marker and into the weakest nmt1 promoter (nmt**) HA₃-tagging vector pSLF373 (23) after converting an ura4 marker to a LEU2 marker. The resulting constructs were termed pSLF173L-pmt3 and pSLF373L-pmt3, respectively.

Preparation of S. pombe cell lysates and Western blot analysis.

Whole-cell extracts from S. pombe cells for Western analysis were prepared by the boiling sodium dodecyl sulfate (SDS)-glass bead method as described below (50). Ten milliliters of logarithmic-phase S. pombe cell cultures (about 5 × 10⁶/ml) was harvested. Pellets were washed twice with water, resuspended in 100 μl of water, and heated at 90°C for 5 min. Then 120 μl of 2× Laemmli buffer (4% SDS, 20% glycerol, 0.6 M β-mercaptoethanol, 0.12 M Tris-HCl [pH 6.8]) containing 8 M urea was added to the samples, which were vigorously vortexed with an equal volume of acid-washed glass beads for 3 min and then heated again 90°C for 5 min. The whole-cell extracts were analyzed by SDS-polyacrylamide gel electrophoresis using 10% polyacrylamide gels and then transferred to Immobilon transfer membranes (Millipore) by using a wet-type transfer system. Anti-HA monoclonal antibody 12CA5 was purchased from BAbCo. Horseradish peroxidase-conjugated goat anti-mouse secondary antibody was purchased from Bio-Rad. Western blot detection was done with the ECL Plus system as described by the manufacturer (Amersham).

GFP tagging of Pmt3p.

The plasmid for green fluorescent protein (GFP) tagging of Pmt3p was constructed as follows. For the GFP fusion at the N terminus of Pmt3p, GFP(S65A) (63) was amplified by PCR with a 5′ primer and a 3′ primer. Primers were 5′-CCGGATCCGGCGGCCGCATGAGTAAAGGAGAAGAA-3′ [a sense primer to create a BamHI site before the start codon of GFP(S65A); the BamHI site is underlined] and 5′-GCGAATTCCTTTTGTATAGTTCATCCATGC-3′ [an antisense primer to remove the stop codon of GFP(S65A) and to create an EcoRI site; the EcoRI site is underlined]. The PCR product of the GFP(S65A) fragment whose stop codon was removed was digested with BamHI and EcoRI (fragment 1). For preparation of the pmt3⁺ promoter region, the pmt3⁺ genomic DNA was amplified by PCR with a 5′ primer and a 3′ primer. Primers were 5′-GATATCTTGAATAACTTC-3′ (a sense primer containing an EcoRV site, which is complementary to the 5′-terminal sequence shown in Fig. 1A; the EcoRV site is underlined) and 5′-GCGGATCCAGATACTATATAAAATC-3′ (an antisense primer to create a BamHI site, which is complementary to the region [positions −25 to −9 in Fig. 1A]) upstream of the initiator ATG of pmt3⁺; the BamHI site is underlined). The PCR product of the pmt3⁺ promoter region was digested with EcoRV and BamHI (fragment 2). For preparation of the pmt3⁺ coding and 3′ noncoding regions, the pmt3⁺ genomic DNA was amplified by PCR with a 5′ primer and a 3′ primer. Primers were 5′-GCGAATTCTGAATCACCATCAGC-3′ (a sense primer to create an EcoRI site and to remove the initiating ATG codon of pmt3⁺, which is complementary to the region [positions +4 to +20 in Fig. 1A] just downstream of the initiator ATG of pmt3⁺; the EcoRI site is underlined) and 5′-AAGCTTCAAGAAAATTTAGC-3′ (an antisense primer containing a HindIII site, which is complementary to the 3′-terminal sequence shown in Fig. 1A; the HindIII site is underlined). The PCR product of the pmt3⁺ coding and 3′ noncoding regions was digested with EcoRI and HindIII (fragment 3). Fragments 1, 2, and 3 were cloned together into the EcoRV-HindIII site of pBluescript II KS+ (77) to construct a plasmid for GFP tagging of Pmt3p (pBS-GFPpmt3). The resulting plasmid, pBS-GFPpmt3, contained the native promoter of pmt3⁺ and the full pmt3⁺ coding region, whose N terminus was ligated in frame to the GFP gene. To integrate the GFP-tagged pmt3⁺ gene onto the genome, the EcoRV-HindIII fragment containing the pmt3⁺ promoter, the GFP-tagged pmt3⁺ coding region, and the 3′ noncoding region of pBS-GFPpmt3 was isolated and cloned into the SmaI-HindIII site of an integration vector, pYC11 (13). The resulting plasmid, pYC11-GFPpmt3, was used for transformation of the wild-type strain (YM77 [h⁺ leu1-32 ura4-D18 ade6-704]), and stable Leu⁺ transformants (YM78 [h⁺ leu1-32 ura4-D18 ade6-704 pmt3⁺:pYC11- GFPpmt3]) were selected.

Fluorescence microscopy.

Immunofluorescent staining and in situ hybridization of cells were done according to a protocol provided by Y. Chikashige (11), which is a modification of the protocols of Chikashige et al. (12) and Dernburg et al. (17). For GFP-Pmt3-expressing cells, cells were fixed with 3% formaldehyde at room temperature for 60 min. TAT1 mouse monoclonal anti-α-tubulin antibody (90) and anti-Sad1 rabbit polyclonal antibody (27) were used to stain microtubule and spindle pole bodies (SPB), respectively. Secondary antibodies used were goat anti-mouse immunoglobulin G (IgG) Cy3-conjugated antibody, goat anti-rabbit Cy3-conjugated antibody (Jackson Laboratory), goat anti-rabbit Oregon green-conjugated antibody (Molecular Probes), and donkey anti-mouse Cy5-conjugated antibody (Amersham). DNA was stained with DAPI (1 μg/ml). Microscopic images were obtained by using a Delta Vision system (Applied Precision) with an Olympus oil immersion objective lens (Dplan Apo 100/NA1.3). Several z-axis sections of 0.1-μm intervals were combined by the quick projection program to avoid overlooking any signals within a cell.

Nucleotide sequence accession number.

The nucleotide sequence of pmt3⁺ has been deposited in the GenBank database under accession no. AB017187.

Identification of S. pombe Smt3p/SUMO-1 homologue.

Two-hybrid screening of an S. pombe cDNA library was done with an accessory factor of DNA polymerase δ, PCNA, as bait. Two novel cDNAs were isolated, one of which was found to encode a homologue of the S. cerevisiae Smt3p after DNA sequencing. The SMT3 gene was originally isolated as a multicopy suppressor of the S. cerevisiae mif2^ts mutation (59) and was suggested to be a novel type ubiquitin-related factor. Therefore, we named this gene pmt3⁺ (S. pombe homologue of SMT3). Although we confirmed the in vivo interaction between PCNA and Pmt3p by immunoprecipitation (data not shown), we have found no genetic relationship between PCNA and Pmt3p.

We searched and obtained the full-length pmt3⁺ cDNA and the corresponding 4.2-kb genomic DNA fragment as described in Materials and Methods. Genomic sequence analysis indicated that the pmt3⁺ gene contains three exons (total of 351 bp) and two introns with appropriate consensus splicing signal sequence (Fig. 1A). Since the sequences upstream of the putative initiator ATG include an in-frame stop codon, we concluded that this cDNA contains the full-length pmt3⁺ gene and that the original cDNA from two-hybrid screening also contained the full-length gene (Fig. 1A). The presence and positions of introns were confirmed by sequencing of the genomic pmt3⁺ gene and the pmt3⁺ cDNA. The pmt3⁺ open reading frame encodes a predicted protein of 117 amino acids with a molecular mass of 13 kDa. Amino acid sequence comparisons revealed that S. pombe Pmt3p is 39% identical to S. cerevisiae Smt3p and human SMT3C (SUMO-1) (47), 13% identical to S. cerevisiae Rub1p (44) and mouse Nedd8 (39), and 11% identical to ubiquitin (Fig. 1B and 1C). The conserved diglycine (GG) sequence corresponding to the C-terminal Gly75 and Gly76 of ubiquitin, which was shown to be crucial for conjugation between ubiquitin and ubiquitin pathway enzymes, was completely conserved among ubiquitin and ubiquitin-related proteins, including Pmt3p (Fig. 1B and C). However, ubiquitin Lys48, required for the generation of ubiquitin polymers, was replaced by Gln at the same position in Pmt3p (Gln83), which suggests that Pmt3p may not form polymers as proposed for SUMO-1 (Fig. 1C) (4).

Gene disruption of the pmt3⁺ gene and phenotypes of pmt3Δ cells.

Gene disruption of pmt3⁺ was carried out as described in Materials and Methods. Precise replacement of one of the two chromosomal copies of pmt3⁺ in an S. pombe diploid strain by the pmt3::ura4 fragment was confirmed by Southern blot analysis of genomic DNA (Fig. 2A and C). All dissected tetrad spores of the Ura⁺ diploid gave two Ura⁻ colonies and two tiny Ura⁺ colonies (Fig. 2B). Thus, the pmt3Δ cells were viable but grew more slowly than wild-type cells (the doubling time of pmt3Δ cells was about 5 h at 30°C in YE medium) (Fig. 2E). DAPI staining and microscopic analysis indicated that about 23% of pmt3Δ cells display aberrant cellular and nuclear morphologies (Fig. 2D); 6% were generally elongated (Fig. 2D-1), 4% showed a displaced nucleus (Fig. 2D-2 and 2D-3), and 2% showed enucleated daughter cells (Fig. 2D-4), indicating that cells had undergone cytokinesis without prior completion of mitosis. In addition, the chromatin of 4% of the cells was highly condensed (Fig. 2D-5 and 2D-6), indicating strong mitotic delay. Furthermore, 7% of the cells displayed the typical cut phenotype where cells remain connected by threads of chromatin (Fig. 2D-7 and 2D-8). These phenotypes are reminiscent of the constant-cut phenotypes of rad31⁺ and hus5⁺ mutants or disruptants (3, 21, 79) (see Discussion). Missegregation of the chromosomes was also confirmed by flow cytometric (fluorescence-activated cell sorting [FACS]) analysis. Wild-type cells showed the normal 2C DNA content peak. The profile of pmt3Δ cells showed a decrease of the number of cells with 2C DNA content and an increase in the number of cells with a DNA content greater or less than 2C DNA content (at times −2 and 0 h in Fig. 3C). This profile is consistent with the faulty chromosome segregation observed by DAPI staining.

In addition to the constant-cut phenotype described above, we found that pmt3Δ cells showed sensitivities to a number of stresses including high temperature, DNA damage induced by UV light or the DNA-damaging agent methyl methanesulfonate (MMS), inhibition of DNA synthesis by HU. The pmt3Δ cells incubated for 12 h at 36°C were 35% as viable as those incubated at 30°C (Fig. 2E). The frequency of elongated and cut phenotype cells after incubation for 12 h at 36°C was higher than at 30°C (data not shown). A plate concentration of 0.0025% MMS severely inhibited growth of the pmt3Δ cells but had little effect on the growth of wild-type cells (Fig. 2E). pmt3Δ cells were moderately sensitive to low doses of UV radiation, as their viability decreased more than 10-fold after a 150-J/m² dose of UV radiation (Fig. 3A). In addition to the sensitivities to high temperature, UV, and MMS, a plate concentration of 5 mM HU severely inhibited growth of the pmt3Δ cells (Fig. 2E). The sensitivity of pmt3Δ cells to HU was examined quantitatively and compared to those of wild-type, cdc2-3w, and cds1Δ cells. As shown in Fig. 3B, pmt3Δ cells showed moderate sensitivity to HU treatment like that of cdc2-3w cells. During an 8-h incubation in the presence of 12 mM HU, the viability of pmt3Δ cells dropped more than 10-fold, while the viability of wild-type cells remained largely unchanged. The viabilities of cdc2-3w and cds1Δ cells, which are replication checkpoint-defective strains, dropped more than 50- and 1,000-fold, respectively (Fig. 3B). Furthermore, FACS analysis showed that the majority of pmt3Δ cells arrest with a 1C DNA content like that of wild-type cells after exposure to HU (Fig. 3C). These results suggested that pmt3Δ cells retain S/M phase checkpoint control (see Discussion).

Requirement of Pmt3p for chromosome segregation.

Since the DAPI staining and FACS profile described above suggested faulty segregation of chromosomes in pmt3Δ cells, we tested the sensitivity of pmt3Δ cells to the microtubule-destabilizing agent thiabendazole (TBZ). TBZ is known to bind tubulin molecules and to inhibit their polymerization. Increased sensitivity to microtubule-destabilizing benzimidazole compounds is a characteristic of defective microtubule components or microtubule-interacting proteins. A plate concentration of 10 μg of TBZ per ml severely inhibited the growth of the pmt3Δ cells but had little effect on the growth of wild-type cells (Fig. 4A). This result also suggested that pmt3⁺ may be involved in assembly of the microtubule system.

To analyze the chromosome segregation defect in pmt3Δ cells in more detail, we measured the rate of loss of a nonessential minichromosome from wild-type and pmt3Δ cells. Wild-type and pmt3Δ cells containing the ade6-704 allele at the genomic ade6 locus and a single copy of the nonessential centromeric minichromosome Ch10 (70) were constructed (YM75 and YM76, respectively [Table 1 and Materials and Methods]). The stability of minichromosome in wild-type and pmt3Δ cells was determined by two different methods described in Materials and Methods. In the first method, the ratio of cells that had lost the minichromosome in pmt3Δ cells (30%) was 50-fold higher than that in wild-type cells (0.6%). In the second method, the rate of minichromosome loss for wild-type cells was 3.4 × 10⁻⁴ per generation under normal growth conditions (Fig. 4B). However, pmt3Δ cells showed more than a 100-fold increase in minichromosome loss rate, with a loss rate of 5.6 × 10⁻² per generation. These results suggest that pmt3Δ cells lack the functions required for accurate chromosome segregation, which may account for the longer doubling time, the aberrant cell morphology, and the low viability of pmt3Δ cells.

Striking increase in telomere length in pmt3Δ cells.

We found another interesting phenotype in pmt3Δ cells, that loss of Pmt3p function caused a striking increase in telomere length. S. pombe chromosomes I and II contain ApaI restriction sites which are located on the centromere-proximal side of the telomeric repeats (84). When the chromosome DNA of wild-type cells was digested with ApaI and analyzed for telomere length by Southern blot analysis with a telomeric repeat DNA fragment as the probe, a band smear of about 0.3 kb, which corresponds to the length of the telomeric repeat in wild-type cells, was detected (Fig. 5, lane 1). As shown in lane 2, disruption of the pmt3⁺ gene caused elongation of telomeres up to a length intermediate (0.8 to 1.2 kb) between that of the wild-type and taz1Δ cells (15), although the signal intensities were not quantitatively estimated in this experiment.

We investigated whether the increase in telomere length could be restored by reintroduction of the gene. pmt3Δ cells were transformed with pSFL373L-pmt3, which carries the HA₃-tagged full-length pmt3⁺ cDNA under the control of the inducible nmt1** promoter (the weakest nmt1 promoter). pmt3Δ transformants containing pSFL373L-pmt3 displayed near-normal phenotypes of pmt3⁺ cells when grown in the absence of thiamine (induced conditions) but not in the presence of thiamine (repressed conditions). The elongated telomere did not recover its normal length under repressed conditions (Fig. 5, lane 3). First, we cultured the cells under repressed conditions and then removed thiamine to induce the expression of pmt3⁺. Chromosomal DNA was prepared every 18 generations after the removal of thiamine. As seen in Fig. 5, lane 3 to 10, the telomeres became gradually shorter and achieved a length of about 0.5 kb after about 90 generations of growth under induced conditions. This result suggested that the elongated telomeres were shortened by a mechanism that maintains telomere length in the pmt3⁺ background. The rate of telomere shortening in pmt3Δ cells expressing pmt3⁺ was approximately 4 to 8 bp per generation, which roughly correlates with the length of the consensus repeat unit of the S. pombe telomere. When pmt3⁺ gene expression was again repressed by the addition of thiamine at 90 generations after induction (lanes 9 and 10), the lengths of the telomeres immediately increased and attained the level seen in pmt3Δ cells within 18 generations (lanes 3, 8, 9, and 10). The amounts of HA₃-tagged Pmt3 protein expressed by the plasmid were confirmed by Western blotting (lanes 3 to 10). These results strongly indicated that pmt3⁺ was directly or indirectly involved in the maintenance of correct telomere length. This is the first demonstration that one of the Smt3p/SUMO-1 family proteins is involved in telomere length maintenance.

Pmt3p-protein conjugation in S. pombe.

The results mentioned above suggest that Pmt3p is required in multiple cellular events. We speculated that Pmt3p exerts its function through its covalent attachment to target proteins like its homologues SUMO-1 and Smt3p. To confirm that S. pombe Pmt3p is conjugated to cellular proteins in vivo, Pmt3p was amino-terminally tagged with three copies of the HA epitope. The gene encoding this fusion protein was placed under the control of the inducible nmt1 promoter (pSLF173L-pmt3). Western blot analysis using an anti-HA antibody revealed the expression in pmt3Δ cells of HA₃-Pmt3p in the presence of thiamine. Migration of the HA₃-tagged Pmt3p was far slower than that predicted from the sequence. This is consistent with the previous reports that both mammalian SUMO-1 and S. cerevisiae Smt3p show anomalously slow migration on SDS-gels (6, 33, 48). In addition, several other anti-HA-reactive proteins with sizes ranging from about 40 to more than 175 kDa were observed as distinct bands (Fig. 6). Considering the pleiotropic phenotypes of pmt3Δ cells, these bands most likely represent HA₃-Pmt3p conjugated to proteins of different sizes whose identities remain to be determined, although some of the bands may be degradation products of the higher-molecular-mass species. We also detected a similar but less clear conjugation pattern in the strain with the integrated GFP-Pmt3 fusion protein described below (data not shown).

Localization of Pmt3p.

Various phenotypes of pmt3Δ cells described above predict that several nuclear proteins are regulated by conjugation of Pmt3p. To determine the intracellular localization of Pmt3p, GFP (63, 69) was fused to the N terminus of Pmt3p. The fusion gene under the control of the pmt3⁺ promoter was integrated into wild-type cells (see Materials and Methods). We confirmed that the pmt3Δ phenotypes were rescued by expression of the GFP-Pmt3 fusion protein, indicating that this fusion protein was functional (data not shown). The green fluorescence of the fusion protein was observed predominantly in the nucleus. Interestingly, intense signals appeared as one or more spots in the nuclei. To identify the nature of the spots, the cells expressing the fusion protein were fixed and stained with anti-Sad1 antibody, which recognizes SPB. The pattern of GFP fluorescence in the fixed cells was not significantly different from that of the live cells (data not shown). In interphase cells, the SPB signals always colocalized with the fluorescent spots. In Fig. 7A, one of the two spots of the green fluorescence signal can be seen to be colocalized with the anti-Sad1 signal. To examine the localization of GFP-Pmt3p during mitosis, cells expressing the fusion protein were fixed and stained with the anti-α-tubulin antibody. The GFP-Pmt3p spots were not detected at the spindle poles during prometaphase and metaphase. Instead, a relatively bright region or intense spots were observed between the spindle poles (Fig. 7B). After anaphase, GFP-Pmt3p spots reappeared at the spindle poles. The behavior of the bright signals at the SPB is reminiscent of centromeres which cluster and associate with SPB throughout the cell cycle except around metaphase in S. pombe (24). We compared centromere positioning in wild-type and pmt3Δ cells by using a combination of immunofluorescence and in situ hybridization with a probe specific for centromeric repeated sequences. The positioning of centromeres in pmt3Δ cells was indistinguishable from that in wild-type cells (Fig. 7C), suggesting that Pmt3p is not essential for centromere clustering at the SPB. Another feature of GFP-Pmt3p localization is that the GFP signal, which was dispersed throughout the nucleus in interphase, seemed to be excluded from chromosome DNA by condensed chromosomes during mitosis (Fig. 7B). These observations indicate that Pmt3p may be required for kinetochore and/or SPB functions during chromosome segregation.