Abstract
The Schizosaccharomyces pombe Rad32 protein is required for repair of DNA double strand breaks, minichromosome stability and meiotic recombination. We show here that the Rad32 protein is phosphorylated in a cell cycle-dependent manner and during meiosis. The phosphorylation is not dependent on the checkpoint protein Rad3. Analysis of a partially purified protein preparation indicates that Rad32 is likely to act in a complex. Characterisation of the rad32-1 mutation and site-directed mutagenesis indicate that three aspartate residues in the conserved phosphoesterase motifs are important for both mitotic and meiotic functions, namely response to UV and ionising radiation and spore viability.
Introduction
Recombination processes contribute to genetic diversity, are required for efficient chromosome segregation during meiosis and provide a mechanism for repairing DNA double strand breaks (DSBs) during the mitotic cell cycle. Understanding the molecular events that are associated with recombination has involved the characterisation, in both prokaryotes and eukaryotes, of mutants defective in the individual stages.
In Saccharomyces cerevisiae, extensive genetic analysis has resulted in the characterisation of a number of genes required for repair of DSBs (the RAD52 group of genes), many of which are also required for meiotic recombination (e.g. 1), as well as several genes required solely for meiotic recombination (2,3). Double mutant analysis of these repair and recombination mutants with spo13 (a mutant defective in the reductional division during meiosis; 4), has resulted in the division of the genes into ‘early’ and ‘late’ subgroups according to the time of their involvement in recombination.
The early events of recombination during both meiosis and the repair of DSBs include the processing of DNA ends at DSBs. Proteins required for this include the Rad50, Mre11 and Xrs2 proteins (3,5–8). In Escherichia coli, the proteins required for initiation of recombination include the RecBCD complex (e.g. 9). The RecB and RecC proteins are subunits of a recombination nuclease (exonucleaseV) and mutations in their genes can be suppressed by mutations in two further genes, sbcC (10) and sbcD (11), which encode subunits of another nuclease which has been proposed to cleave secondary structures formed at replication forks (12). The sbcC and sbcD genes have been shown to have sequence homology to the S.cerevisiae MRE11 and RAD50 genes (13).
The fission yeast Schizosaccharomyces pombe has recently proved to be a good model system for the study of DNA repair and damage response processes in eukaryotes (e.g. 14,15). From characterisation of the ∼35 radiation-sensitive (rad) mutants identified to date and isolation of homologues to known recombination repair genes in other organisms, a distinct set of genes required for recombination repair has been identified in S.pombe (e.g 16–19). This group includes rad22 (19), rhp51 (16), rhp54 (17) and rad32 (18) of which rad22, rhp51 and rhp54 are homologues of the S.cerevisiae RAD52, RAD51 and RAD54 genes, respectively.
Recent studies have indicated that the S.pombe rad32 gene shares structural and functional homologies with S.cerevisiae MRE11 (18), a gene required for an early stage of recombination (3). The rad32 null allele has been shown to be defective in meiotic recombination, repair of DSBs and to have a high rate of loss of minichromosomes (18). A search for motifs in the Rad32 protein has indicated structural similarities with the phosphoesterase motifs in the S.cerevisiae Mre11 and E.coli SbcC proteins (13).
We report here on the analysis of the S.pombe Rad32 protein. In particular we demonstrate that the protein is phosphorylated in a cell cycle-dependent manner and that the phosphorylation pattern is altered during meiosis. During mitotic growth, phosphorylation of Rad32 does not require the kinase activity of the checkpoint protein Rad3. Gel filtration studies suggest that Rad32 may form part of a complex. In addition we demonstrate, by sequence analysis of the rad32-1 allele and using site-directed mutagenesis, that three aspartate residues in the conserved phosphoesterase motifs are required for Rad32 function in both repair of DNA damage and during meiosis. In contrast a rad32 mutation (P93S) equivalent to one of the ‘separation of phenotype’ mutations in mre11S (7) has little effect on the function of the Rad32 protein. To investigate the nature of metal ions which may bind the phosphoesterase motifs we have studied the effect on cell growth and survival of altering the levels of Fe3+, Zn2+ and Mn2+ in the media, and these studies suggest that the Rad32 protein may bind Fe3+.
Materials and Methods
Strains, plasmids and growth media
The strains used are detailed in Table 1. The rad32 cDNA was obtained by PCR using two oligonucleotide primers (1N ATTCATATGCCAAATGACCCC) and (2B GGATCCCAATCAATCATCTAAA) using a cDNA library as template (20), and cloned into the E.coli expression vector pET15b (Novagen) and into pRH1 and pRH81 (which are modified versions of the S.pombe expression vectors pREP1 and pREP81, respectively; 21) with an NcoI-BamHI fragment containing a 6× His-tag coding sequence cloned adjacent to the nmt1 promoter. The pREP41-rad3 clone and rad3-d strain were from N. Bentley and A. Carr (22). Routine handling of cultures and growth media were as described previously (23). Media containing increased levels of metal ions we made as for standard S.pombe minimal media containing 10-fold increase in FeCl3 (2 mg/l), ZnSO4 (4 mg/l) or MnSO4 (4 mg/l).
Strains used and created in this study
Antibodies
Anti-Rad32 antisera were generated in rabbits against a 6× His-tag fusion protein containing the full-length Rad32 protein. The antibodies were affinity purified using Rad32 protein bound to Affigel-10 (Bio-Rad).
32P-labelling of yeast proteins
[32P]orthophosphate (2 mCi) was added to a 10 ml culture of sp.011 cells transformed with pRH1-Rad32 which had been grown in synthetic EMM1 medium (24) under selective conditions in the absence of thiamine for 48 h at 30°C. Cells were grown to a density of 5 × 106/ml and radiolabelling was carried out for 4 h. Cells were washed, resuspended in 5 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl pH 7.9, 6 M urea in the presence of a protease and phosphatase inhibitor cocktail (50 mM NaF, 0.1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 1 µg/ml leupeptin, 1 µg/ml aprotinin, 10 µg/ml soya bean trypsin inhibitor and 50 µg/ml TLCK) and broken using a Braun dismembranator. The resulting extract was spun at 2000 g for 3 min, the supernatant was removed and the beads washed in the same buffer. The resulting pooled sample was loaded onto a His·Bind resin column (Novagen) and Rad32-His fusion protein eluted in the presence of 6 M urea according to the manufacturer's instructions. All solutions contained the protease and phosphatase inhibitor cocktail.
Rad32 purification
The rad32 cDNA was cloned into the vector pRH81 to produce a 6× His-tag fusion protein (pMT3) and used to transform sp.011. The transformed strain was grown in selective medium. All protein purification procedures were carried out at 4°C. Cells (from 25 l of culture at 1 × 107 cells/ml) were collected by centrifugation and washed with cold Tris-buffered saline. The pellet was resuspended in 125 ml ice-cold buffer A (50 mM Tris pH 8.0, 25 mM p-nitrophenyl phosphate in the presence of the protease and phosphatase inhibitor cocktail). An equal volume of glass beads (0.4 mm diameter) was added to the cell suspension and cells were broken in a bead beater (Biospec Products) for 5 × 30 s with 5 min cooling between each burst. The supernatant was collected and the glass beads and cell pellet re-extracted with another 150 ml buffer A containing 300 mM NaCl with 2 × 30 s bursts in the bead beater. The supernatants were pooled and subjected to centrifugation at 18 000 r.p.m. The resulting supernatant was retained (Fraction I).
Fraction I was loaded onto a 70 ml Q-Sepharose column which was then washed with buffer A containing 100 mM NaCl without protease inhibitors. Proteins were eluted batchwise with buffer B (50 mM Tris pH 6.8, 500 mM NaCl containing the cocktail of phosphatase and protease inhibitors described for buffer A). Fractions containing Rad32 protein were identified by western blotting and pooled. Protein was precipitated by addition of ammonium sulphate to 60% and redissolved in 150 ml buffer C (25 mM sodium phosphate pH 6.8 containing phosphatase and protease inhibitors) to produce Fraction II. Fraction II was loaded onto a 30 ml S-Sepharose column and the proteins eluted with 50 mM HEPES pH 8.0, 400 mM NaCl containing protease and phosphatase inhibitors (Fraction III).
Fraction III was diluted with an equal volume of buffer D (50 mM Tris pH 8.0, 50 mM NaF and protease inhibitors) and loaded onto an 18 ml blue Sepharose colum and eluted with buffer D containing 800 mM NaCl. Protein was precipitated from Rad32-containing fractions with ammonium sulphate (60%) and then dissolved in 4 ml buffer D containing 150 mM NaCl (Fraction IV). Fraction IV was loaded onto a Sephacryl S-300 (16/60) column and eluted with buffer D containing 150 mM NaCl.
Rad32 protein is modified by phosphorylation in S.pombe. (a) Western analysis using anti-Rad32 antisera of protein extracts from E.coli transformed with pET15b-rad32 (lane 1) and S.pombe sp.011 (rad+) transformed with pRH81-rad32 (lane 2). (b) In vivo labelling with [32P]orthophosphate of sp.011 cells expressing Rad32 from pRH1-rad32, analysed by autoradiography (lane 1) and western analysis with anti-Rad32 antisera (lane 2).
Integration of mutated rad32 sequences at the rad32 locus
Site-directed mutagenesis of rad32 cDNA was carried out by standard procedures (25). In order to ensure integration of the mutated sequences at the correct locus, it was necessary to subclone the rad32 promoter (on a 660 bp PstI-NdeI fragment) upstream of the cDNA to provide homologous sequences for the integration. In order to do this, an NdeI site within the 660 bp promoter-containing fragment was removed by site-directed mutagenesis. To ensure that this had no effect on rad32 expression, a wild type rad32 cDNA cloned behind the mutated promoter-containing fragment was included in the study as a control. Integration of the mutated sequences into the rad32 locus was carried out by co-transformation of the rad32 null allele (in which the majority of the rad32 coding sequence is replaced by the ura4 gene) by the 2.8 kb linear fragments containing the appropriate mutation along with pAL19 (26). Transformed cells were selected on minimal medium in the absence of leucine. After 3 days, transformants were replica plated onto minimal medium containing both leucine and uracil plus 0.1% 5-fluoroorotic acid (5-FOA). Colonies appearing after 2 days were checked for loss of the ura4 gene by replica plating onto minimal agar plus and minus uracil.
Results
Rad32 protein is modified in S.pombe
The rad32 cDNA was cloned into the E.coli and S.pombe 6× His-tag expression vectors, pET15b and pRH81 respectively, and introduced into the appropriate host strains. Western analysis of fractions purified on Ni2+-agarose is shown in Figure 1a. In E.coli (lane 1), a single protein species of Mr 80 kDa is detected (the predicted Mr is 73.6 kDa) indicating that the Rad32 protein migrates somewhat slower than expected on SDS-PAGE. In S.pombe, in contrast to the situation in E.coli, multiple species are detected with at least three bands ranging in size from 80 kDa, corresponding to that detected in E.coli, to 110 kDa (Fig. 1a, lane 2). During the purification, it was noted that if phosphatase inhibitors were omitted, the slower migrating species disappeared (data not shown), suggesting that the ladder of bands may correspond to different phosphorylation states of Rad32.
To determine whether the Rad32 protein is modified by phosphorylation, in vivo32P-labelling was carried out on wild type (sp.011) S.pombe cells transformed with the pRH1-Rad32 plasmid. After 24 h incubation in thiamine-free medium to induce expression of the fusion protein, cells were labelled for 4 h and 6× His-tagged protein purified by Ni2+-agarose chromatography. The resulting fraction was subjected to SDS-PAGE followed by either autoradiography or western analysis using anti-Rad32 antisera (Fig. 1b). Both procedures identified multiple species, corresponding in size to those observed in Figure 1a, lane 2, indicating that indeed the Rad32 protein is phosphorylated in vivo.
Rad32 is hypophosphorylated at G2/M
The phosphorylation state of Rad32 was investigated in cells arrested at different stages of the cell cycle. Western analysis using anti-Rad32 antisera was carried out on protein extracts from six cell cycle mutants expressing Rad32 from pRH81-rad32 which had been arrested at the non-permissive temperature (Fig. 2a). The level of phosphorylation was similar in mutant strains arrested in G1 (cdc10), S phase (cdc22) and late S-G2 (cdc6, cdc17, cdc21) but was substantially reduced in cdc25 cells blocked at G2/M.
To further investigate phosphorylation during the cell cycle, cdc25 cells expressing Rad32 from pRH81-rad32, were incubated at 35.5°C for 3 h and then shifted down to 25°C to produce a synchronously cycling culture. Samples were taken at 20 or 30 min intervals and used for western blotting and analysis of the septation index (the peak in the septation index corresponds to G1/S). Figure 2b shows that following release from the cdc25 block, where phosphorylation levels are low, phosphorylation increases as cells enter the next cell cycle reaching a maximum around S phase after which the level drops to a minimum at G2/M.
Phosphorylation levels during meiosis
We have previously shown that Rad32 is required for meiotic recombination (18). We therefore wished to investigate whether there was any alteration in the phosphorylation patterns of Rad32 during meiosis. This was undertaken using an h90rad+ strain transformed with pRH81-rad32 grown in low nitrogen-containing liquid medium. Figure 2c shows that phosphorylation levels increase up to 6.5 h and have decreased by 21 h.
Phosphorylation is not dependent on the checkpoint protein Rad3
The Rad3 protein is required for the DNA damage and replication checkpoints (27). It has homology to a family of PI-3 kinases (22), and is required for phosphorylation of the checkpoint proteins Chk1 and Cds1 (28,29). To determine whether phosphorylation of Rad32 is dependent on the Rad3 protein, pRH81-rad32 was introduced (i) into a rad3-d strain and (ii) into sp.011 (rad+) cells singly and by cotransformation with pREP42-rad3. Western analysis of protein extracts from these strains indicated that phosphorylation levels of Rad32 were the same in cells deleted for rad3 and in those overexpressing rad3 (Fig. 2d) demonstrating that Rad3 is not required for phosphorylation of Rad32. We have also shown in a similar experiment involving transformation of a hhp1/hhp2 double deletion mutant (30) with pRH81-rad32 that phosphorylation is not dependent on either of the casein kinases Hhp1 and Hhp2 (data not shown). Additionally, we see no alteration in the pattern of phosphorylation of Rad32 following exposure of cells to 250 or 350 Gy ionising radiation (data not shown).
Phosphorylation of Rad32 is cell cycle-dependent, is altered during meiosis and is not dependent on the checkpoint protein Rad3. (a) Western analysis of S.pombe protein extracts from cell cycle mutants (transformed with pRH81-rad32) arrested at the non-permissive temperature (35.5°C) for 3 h, using anti-Rad32 antisera. (b) Western analysis of protein extracts from cdc25 cells (transformed with pRH81-rad32) released from a 3 h block at the non-permissive temperature. The culture was sampled at 20 or 30 min intervals and then protein extracts were prepared and the septation index determined. (c) An h90 strain obtained from sp.012 (rad+) was grown in low nitrogen-containing liquid medium to induce meiosis and samples taken at intervals for western analysis. Sample times following inoculation into low nitrogen medium were 0 (lane 1), 2.5 (lane 2), 5 (lane 3), 6.5 (lane 4) and 21 h (lane 5). (d) Western analysis of extracts from sp.011 cells transformed with pRH81-rad32 (lane 1), rad3-d (sp.379) cells transformed with pRH81-rad32 (lane 2) and sp.011 (rad+) cells cotransformed with pRH81-rad32 and pREP42-rad3 (lane 3).
Rad32 fractionates with an apparent molecular mass of 200 kDa on a gel filtration column
To further characterise the Rad32 protein, it was partially purified as described in the Materials and Methods. Gel filtration analysis on Sephacryl S-300 followed by western analysis indicated that the Rad32 protein fractionated with an apparent Mr of 180–220 kDa (data not shown), suggesting that Rad32 may form part of a complex. The nature of this complex is unknown but may be due to interaction of Rad32 with itself or other proteins. The formation of a complex by Rad32 is consistent with the data on S.cerevisiae and human Mre11 proteins which have been shown, using the yeast two-hybrid system, to interact with themselves, with Rad50 and with DNA ligase I (6,31).
Relationship of Rad32 to other phosphoesterases. (a) Sequence comparison of phosphoesterase motifs. The Swissprot accession numbers of the proteins are Human PP2b (calcineurin) Q08209, Rabbit PP-1 P08129, E.coli SbcD P13457, S.cerevisiae Mre11 P32829, human Mre11 P49959, S.pombe Rad32 Q09683. (b) Dendrogram showing evolutionary relationship of Rad32 to other selected phosphoesterases. Tree calculated using the ClustalV set of programs on DNASTAR. The Swissprot accession numbers are as follows: p2a2 Sc P23595, p2a1 Sc P23594, pp11 Sc P20604, ppz1 Sc P26570, ppz2 Sc P33329, rdgc Dm P40421, apah Ec P05637, apah Ka P27510, apah Hi P44751, dbr1 Sc P24309, exo1 T4 P04521, sbcd Bs P23479, asm Hs P17405, cn16 Ec P08331, usha Ec P07024, pp lambda P03772, p2c3 At P49599.
The rad32-1 mutation results in an Asp to Asn change in the phosphoesterase motif
The rad32-1 allele is sensitive to UV radiation, to virtually the same extent as the rad32 null allele (Fig. 4a). The rad32 gene was amplified by PCR from the rad32-1 mutant strain (18) and cloned into the T vector in order to allow identification of the mutation in the rad32-1 strain. Sequence analysis of three independent isolates indicated a single nucleotide substitution (G to A) resulting in an Asp to Asn mutation (D135N). This Asp residue occurs in a region which has been identified as being conserved in a family of phosphoesterases which include not only closely related members, such as S.cerevisiae and human Mre11 and E.coli SbcD (13), but also a range of protein phosphatases which include λ phosphatase (32), PP-1 (33) and calcineurin B (PP2b) (34). In particular, the family is defined by a cluster of motifs (13), which in Rad32 are located in the N-terminal region of the protein (Fig. 3a).
The evolutionary relationship between these phosphoesterases is shown in Figure 3b. This analysis indicates that as expected from the mutant phenotypes, Rad32 belongs to the sub-family which includes the human and S.cerevisiae Mre11 proteins.
Further analysis of residues within the phosphoesterase motif
To further investigate the importance of the conserved motifs for Rad32 function, site-directed mutagenesis was carried out. Analysis of the sequence of the motifs suggested that the conserved aspartate residues in Rad32 (D25, D65 and D135) may be involved in metal ion binding. (Note that D135 is the residue which is mutated in the rad32-1 strain.) All three aspartate residues were therefore each mutated to asparagine to create mutations D25N, D65N and D135N, and the effects observed when the mutated sequences were present in multiple or single copies in cells. An additional mutation was also included in the analysis as a control: this was a C to T mutation previously isolated by PCR mutagenesis, resulting in T517I.
Radiation sensitivities of rad32 alleles. (a) Comparison of rad32-1 and rad32-d alleles. (b) Effect of mutations present in multi-copy on the response to UV. sp.011 = rad+, 32d+X = rad32-d cells transformed with corresponding mutated sequence cloned behind the high level expression promoter version of nmt1 in pRH1. (c and d) Effect of the mutations integrated into the genome in single copy behind the rad32 promoter, 32-wt = wt cDNA cloned adjacent to rad32 promoter. (c) Response to UV. (d) Response to ionising radiation.
The results of introducing into rad32-d cells, the four rad32 mutations in high copy number on the pRH1 vector are shown in Figure 4b. The rad32-D25N and rad32-D135N mutations are least able to reverse the rad32-d UV radiation sensitive phenotype while the rad32-D65N and rad32-T517I mutations restore the levels of radiation resistance to close to wild type levels.
To determine whether the abilities of D65N and T517I to rescue the UV-sensitive phenotype are due to the presence of multiple copies of the mutant proteins, all the mutations were introduced into cells as integrated single copies (Fig. 4c). All strains carrying any one of the three D to N mutations displayed significant sensitivity to UV radiation with strains carrying the D25N and D135N mutations again displaying radiation sensitivities close to that of the rad32 null allele. Cells containing a single copy of the T517I mutation displayed no difference in radiation sensitivity from wild type cells or cells in which the wild type rad32 cDNA replaced the null allele in the genome. As expected, the results obtained with D135N are very similar to those obtained with the rad32-1 allele. A similar pattern of radiation sensitivity was observed in response to ionising radiation (Fig. 4d).
When the four mutations were introduced individually into wild type (sp.011) cells on a multicopy plasmid (pRH1) there was no change in radiation sensitivity or other phenotype indicating that none of these mutations acts as a dominant negative mutation (data not shown).
rad32-P93S mutation equivalent to one of the mre11S mutations has little effect on Rad32 function
The S.cerevisiae Rad50 and Mre11 proteins have been shown to be required for both DSB repair during mitosis and DSB formation and processing from DSB during meiosis (6,8,35). Mutations in both the RAD50 and MRE11 genes have allowed separation of these functions in that the rad50S and mre11S mutations allow initiation but not processing and repair of meiotic DSBs (5,7). To determine whether we could separate the functions of Rad32, we created the equivalent in Rad32 of one of the mutations in one of the mre11S alleles (7). In rad32 this is equivalent to a P to S substitution (P93S) and was created by site-directed mutagenesis and introduced into S.pombe as described above. Figure 4c and d shows that the mutation has no effect on the sensitivity to either UV or ionising radiation.
Effect of MMS and HU
Sensitivity of the rad32 alleles to MMS and HU
Methylmethane sulfonate (MMS) is a DNA damaging agent that ultimately results in DNA DSBs. We compared the MMS sensitivities of the site-directed mutants with those of the rad32 and rad22 null alleles and the DNA damage and replication checkpoint null allele rad3-d (Table 2). The rad32 null allele, like the rad22 and rad3 null alleles, is very sensitive to 0.005% MMS. The rad32-D25N and rad32-D135N mutants are also sensitive, but to a somewhat lesser extent. In contrast, the rad32-D65N and rad32-T517I mutations have little effect on the sensitivity to MMS. The rad32-P93S allele has 15% cell survival on 0.005% MMS, intermediate between wild type and rad32-d responses.
Hydroxyurea (HU) is an inhibitor of DNA replication and in wild type cells causes activation of the DNA replication checkpoint (27) through depletion of the dNTP pool. This replication arrest appears to result in the production of DNA structures which need to be processed prior to continued replication (e.g. 36). Wild type cells are able to survive this arrest and show 100% growth in the presence of 6 mM HU (Table 2). In contrast rad3-d, a DNA structure-dependent checkpoint mutant (22) is very sensitive and is unable to form colonies under these conditions. The rad22-d mutant and all of the rad32 alleles except rad32-P93S form microcolonies on 6 mM HU, a phenotype shared with other DNA damage response mutants such as rhp9-64 (36).
Spore viability is reduced in strains carrying mutations in the phosphoesterase motifs
We have previously shown that rad32 is required for meiotic recombination and spore viability (18). We therefore analysed the effects of the rad32 mutations on these two events. Table 3 shows the effects of the mutations on spore viabilities. In crosses to rad32-d it can be seen that two of the D to N mutations, rad32-D25N and rad32-D135N, reduce the spore viability to levels observed in rad32-d × rad32-d crosses. When crossing the D25N and D135N mutant alleles to themselves, the spore viability is similar, although slightly lower than that observed for the rad32-d null cross (the slightly different figures are likely to be due to the fact that only a small number of viable spores are detected and slight differences are therefore more noticeable). We also analysed meiotic recombination frequencies by crossing strains carrying the ade6-M26 and ade6-469 alleles (37). The rad32-D35N, rad32-T517I and rad32-P93S alleles displayed wild type levels of recombination while rad32-D25N and rad32-D135N alleles mirrored results obtained with the rad32-d null allele (data not shown).
Effect of rad32 mutations on spore viabilities
The slow growth and radiation-sensitive phenotypes of the D to N mutation-containing alleles can be partially rescued by increasing the Fe3+ concentration
The rad32-D25N, rad32-D65N and rad32-D135N mutations are present in motifs, which by analogy with published structures of the phosphoesterases calcineurin and mammalian serine/threonine protein phosphatase-1 (PP-1) are likely to be involved in metal binding. In the case of calcineurin, it is thought that the metal ions are Zn2+ and Fe3+ (32), while in PP-1, Mg2+, Ni2+, Cu2+ and Zn2+ have been ruled out leaving Mn2+, Fe3+ or Co2+ as possibilities (33). Since the three rad32 D to N mutations may reduce the protein's activity by increasing the Km for the required metal ion(s), we tested the effect of increasing the concentrations of metal ions on mutant protein function in vivo.
Deletion of the rad32 gene results in a decrease in cell viability during the normal mitotic cell cycle even in the absence of UV or ionising radiation-induced DNA damage (Fig. 5a). Additionally, under normal growth conditions rad32-d cells are 50% longer than wild type cells suggesting that a checkpoint has been activated (Fig. 5b). Analysis of cell viability and cell length in exponentially growing cultures of the new rad32 alleles indicates that the three D to N mutations all reduce the cell viability in the absence of induced DNA damage, although only the rad32-D25N and rad32-D135N mutations result in increased cell length. Curiously, the rad32-D65N mutation does not display the altered cell length phenotype although the cell viability is as low as for the other D to N mutant strains (Fig. 5b and c).
Cell size and cell viability of rad32 alleles. (a) % cell viability. Five-hundred cells were plated onto minimal medium (norm) and after incubation at 30°C for 3–4 days the number of colonies were counted and expressed as % cell viability, relative to the number plated. (b–d) Effect of 10-fold increase in concentrations of Fe3+ (Fe), Zn2+ (Zn) and Mn2+ (Mn) on (b) cell size (×5 mm), (c) cell viability and (d) cell viability following exposure to 170 J/m2 UV radiation.
In comparison to wild type and rad32-d cells which are unaffected, all three D to N mutant strains produced microcolonies on medium containing 10-fold increased Mn2+ concentration, presumably due to the observed decrease in cell viability (Fig. 5c and d). The increased cell length phenotype in rad32-D25N and rad32-D135N on standard minimal medium can be rescued by growth of mutant strains in 10× standard concentrations of Fe3+ and Mn2+. In addition, the low cell viability of rad32-D65N can be rescued partially by increased Fe3+ concentrations. Altering Fe3+ concentrations has no effect on wild type or rad32-d cells. Interestingly, altering the level of Zn2+ has no effect on either cell viability or cell length. The rad32-P93S mutation has no affect on cell length and is not affected by increasing the metal ion concentrations. These data suggest that the metal ion(s) bound by the Rad32 phosphoesterase motifs is not Zn2+ and may be Fe3+ or possibly Mn2+. The reason for the effects on cell viability seen with increased Mn2+ is not clear but may indicate that Mn2+ ions are competing for binding at the metal binding site(s).
Discussion
The homology between the Mre11 and Rad32 proteins suggests that they are likely to have related functions; however, analysis of the phenotypes of mre11 and rad32 mutants indicates that there are differences in the cellular roles of the two proteins. The mre11 gene is required for repair of DSBs during mitosis and for induction of meiotic recombination and spore viability but is not required for mating type switching or the normal mitotic cell cycle (although mre11 mutants display high frequency spontaneous recombination during mitosis) (3,6). Like Mre11, Rad32 is required for repair of DSBs and meiotic recombination (18) but it is also likely to be required in the alternative (non-nucleotide excision repair) UV-damage repair pathway (38) an additional role which may explain, at least in part, the UV- and HU-sensitive phenotypes of the rad32-d strain.
Phosphorylation of Rad32 protein is maximal around the time of S phase and minimal at G2/M. These data, along with the observed sensitivity to HU, suggest that if the phosphorylated form is the active form of Rad32, then the protein may be involved in processing structures occurring during S phase, having a role similar to that proposed for SbcC and SbcD (12). Phosphorylation levels also increase during meiosis reaching a peak at around 6 h after shifting cells to low nitrogen-containing medium (around the time taken to enter meiosis I; 39), consistent with a requirement for Rad32 in meiotic recombination.
The requirement for the three aspartate residues in motifs I, II and III corresponds well with the data obtained in an analysis of the λ-PPase (32) which is also a member of this family of phosphoesterases. Crystal structures have recently been obtained for two further members of this family, namely PP-1 (33) and calcineurin (34). From the two structures it can be seen that the aspartate residues equivalent to D25 and D65 (D64 and D92 in PP-1 and D90 and D118 in calcineurin) are both present at the active site and are involved in co-ordinating a metal ion which in the case of calcineurin, is thought to be Fe3+. The equivalent residues to D135 are glutamates in both PP-1 and calcineurin (E126 and E152, respectively) and these are present close to a second metal ion binding site. The importance of the phosphoesterase signature in Mre11 has also been shown recently, where an H to Y substitution in motif 4 results in a separation of function (mre11S) phenotype (40).
In creating the rad32-P93S mutation we were hoping to create the equivalent of the mre11S allele which results in separation of functions, where the mitotic functions are mainly intact but there is a severe defect in meiosis (7). The rad32-P93S mutant displays no mitotic defect, i.e. it has wild type levels of sensitivity to both UV and ionising radiation, is resistant to HU and displays very slight sensitivity to MMS. However, unlike mre11S, it is wild type in its ability to undergo meiosis, in that spore formation and spore viability are unaffected. The mre11S allele contains two mutations, namely P84S and T188I (7). We have only analysed the equivalent of the P84S mutation (this proline is conserved in S.cerevisiae, S.pombe, Caenorhabditis elegans, human, mouse, but not present in E.coli SbcD) and it is possible that both mutations are required for the S phenotype, although the amino acid equivalent of T188 is far less conserved in eukaryotes that P84.
The S.cerevisiae Mre11 protein has been shown recently to have 3′ to 5′ exonuclease activity (41); however, despite extensive analysis we have been unable to demonstrate nuclease activity in purified Rad32 preparations. This may be due to the labile nature of the phosphorylated and possibly active form of the protein or the need for the protein to act in a complex which may not survive the protein purification procedures. Indeed, our data on Rad32 being present in a complex (Fig. 3) support those relating to human and S.cerevisiae Mre11 from Petrini and co-workers (31,42,43) and Johzuka and Ogawa (6) and which indicate interactions with human ligase I, S.cerevisiae and human Rad50 and nibrin. SbcC and Rad50, which share structural similarities, have been shown to enhance the nuclease activities of SbcD and S.cerevisiae Mre11, respectively (12,41). To date there is no published sequence for an S.pombe homologue to rad50, although given the conserved nature of many of the DNA repair and recombination genes and the fact that human and S.cerevisiae homologues of rad50 exist (42,44), it is likely that there is also one in S.pombe. Should an S.pombe rad50 homologue be identified it would be interesting, by analogy with the E.coli SbcC and SbcD proteins (13), to co-express it with rad32 in order to improve the likelihood of observing nuclease activity should it exist in Rad32.
Acknowledgements
S.W. was supported by a BBSRC studentship. F.Z.W. thanks the Royal Society and the Wellcome Trust for travel grants. The work was supported in part by CRC grants SP2212/0101 and SP2212/0102.
Comments