Characterization of a Schizosaccharomyces pombeStrain Deleted for a Sequence Homologue of the Human Damaged DNA Binding 1 (DDB1) Gene*

Human damaged DNA-binding protein (DDB) is a heterodimer of p48/DDB2 and p127/DDB1 subunits. Mutations in DDB2 are responsible for Xeroderma Pigmentosum group E, but no mutants of mammalian DDB1 have been described. To study DDB1, theSchizosaccharomyces pombe DDB1 sequence homologue (ddb1+ ) was cloned, and a ddb1deletion strain was constructed. The gene is not essential; however, mutant cells showed a 37% impairment in colony-forming ability, an elongated phenotype, and abnormal nuclei. The ddb1Δstrain was sensitive to UV irradiation, X-rays, methylmethane sulfonate, and thiabendazole, and these sensitivities were compared with those of the well characterized rad13Δ,rhp51Δ, and cds1Δ mutant strains. Ddb1p showed nuclear and nucleolar localization, and the aberrant nuclear structures observed in the ddb1Δ strain suggest a role for Ddb1p in chromosome segregation.

Human damaged DNA-binding protein (DDB) is a heterodimer of p48/DDB2 and p127/DDB1 subunits. Mutations in DDB2 are responsible for Xeroderma Pigmentosum group E, but no mutants of mammalian DDB1 have been described. To study DDB1, the Schizosaccharomyces pombe DDB1 sequence homologue (ddb1 ؉ ) was cloned, and a ddb1 deletion strain was constructed. The gene is not essential; however, mutant cells showed a 37% impairment in colony-forming ability, an elongated phenotype, and abnormal nuclei. The ddb1⌬ strain was sensitive to UV irradiation, X-rays, methylmethane sulfonate, and thiabendazole, and these sensitivities were compared with those of the well characterized rad13⌬, rhp51⌬, and cds1⌬ mutant strains. Ddb1p showed nuclear and nucleolar localization, and the aberrant nuclear structures observed in the ddb1⌬ strain suggest a role for Ddb1p in chromosome segregation.
XP-E cells are only mildly defective in nucleotide excision repair (NER) of DNA damage, and microinjection of purified DDB corrects these deficiencies (10). Moreover, DDB stimulates, but is not necessary for, NER in vitro (11), but it appears to be necessary for normal global genomic NER (12). It associates with the histone acetyltransferase, cAMP-response element-binding protein-binding protein/p300, which is believed to be important in altering chromatin structure. Datta et al. (13) proposed that DDB has a stimulatory role in normal global genomic NER of UV-induced DNA lesions by recruiting cAMPresponse element-binding protein-binding protein/p300 to the site of damage to render chromatin more accessible to the DNA repair machinery.
DDB interacts with other cellular proteins, suggesting that it may be multifunctional. DDB2 binds to the cell cycle regulatory transcription factor, E2F1, and in the presence of DDB1 it acts as a negative regulator of G 1 /S cell cycle progression following UV-induced DNA damage (14). Both DDB1 and DDB2 are ubiquitinated by Cul-4A, a member of the cullin family of proteins (15)(16)(17). Cul-4A is believed to be a ubiquitin-protein isopeptide ligase (type E3) and is involved in G 1 /S phase progression in Caenorhabditis elegans (18). The expression of DDB2 is cell cycle-regulated, peaking at the G 1 /S boundary (17). DDB2, but not DDB1, mRNA and protein levels are induced 2-to 3-fold by UV irradiation (19). DDB1 binds to the apolipoprotein B (apoB) gene regulatory factor 2 (BRF-2), and the DDB1-BRF-2 heterodimer has been suggested to be required for optimum-specific apoB gene expression (20). DDB1 also binds to viral transcriptional transactivators, including the hepatitis B virus X protein (HBVx) (21) and the V proteins from paramyxovirus, simian parainfluenza virus 5, mumps virus, human parainfluenza virus 2, and measles virus (22). DDB2 also has been shown recently to bind HBVx protein (23). Because nuclear levels of DDB increase in late G 1 , DDB has been proposed to participate in nuclear functions of HBVx during the late G 1 phase of the cell cycle. Despite this information, the precise physiological role(s) of DDB still remain elusive.
We have identified three domains that are highly conserved in DDB1 sequence homologues of Homo sapiens, Mus musculus, Drosophila melanogaster, C. elegans, Dictyostelium discoideum, Arabidopsis thaliana, S. pombe (24), and Oryza sativa. 2 Two other reports assigned the same S. pombe predicted polypeptide to the DDB1 family of proteins (25,26). The overall identity between the human and S. pombe (previously published as NCBI accession number SPAC17H9.10c) sequences is 26%; the identities in domains 1, 2, and 3 are 45, 40, and 21%, respectively. To our knowledge, no sequence homologue of DDB1 has been found in Saccharomyces cerevisiae.
S. pombe has been used extensively to study DNA repair and cell cycle control (27,28). S. pombe appears to be more closely related to animal cells with respect to cell cycle regulation and chromosome structure and segregation than is S. cerevisiae (29) from which it diverged 330 -420 million years ago (30). In this study we report the cloning of the S. pombe sequence homologue of the human DDB1 gene, ddb1 ϩ , and the characterization of the ddb1 deletion mutant. ddb1 ϩ is not an essential gene, and the ddb1⌬ cells show an elongated phenotype and nuclear abnormalities. The sensitivity of the ddb1 deletion strain to a variety of DNA-damaging agents, to hydroxyurea, and to thiabendazole was compared with those of the null strains rad13 (the gene encoding for the homologue of the human XPG protein (31)), rhp51 (the gene encoding for the homologue of the human RAD51 protein (32)), and cds1 (a gene implicated in the intra-S checkpoint response, the homologue of the human CHEK2 (33,34)).

EXPERIMENTAL PROCEDURES
Strains, Media, and Growth Procedures-Unless otherwise indicated, cells were grown at 30°C in rich (YES) medium supplemented as needed with 250 mg/liter leucine, uracil, adenine, and histidine and 5 mg/liter of thiamine. Strains are described in Table I (36). To determine colony-forming ability, cultures in mid-log phase at 30°C in YES medium were counted with a hemocytometer and then properly diluted and plated onto YES plates that were incubated for 4 days at 30°C in the dark.
Cloning of S. pombe ddb1 ϩ -An open reading frame (ORF) encoding a predicted polypeptide that shares 26% sequence identity and 43% homology with that of the human DDB1 protein had been identified by BLASTp search of the NCBI databases (24). The ORF, designed here as ddb1 ϩ , is localized on chromosome I, Cosmid C17H9, and has been published previously as NCBI accession number SPAC17H9.01. The coding sequence of the putative gene was cloned by PCR using as template S. pombe genomic DNA. The PCR product was inserted into the Escherichia coli pCR T7/CT-TOPO vector (Invitrogen) in-frame with a C-terminal His 6 tag. The ddb1 ϩ -His 6 -encoding DNA fragment was removed with NdeI and PmeI and subcloned into the pREP2 plasmid (35) under the control of the strong thiamine-repressible nmt1 promoter to form the pFZ1 plasmid.
Overexpression of Ddb1p-His-Cells of the haploid strain 146 (Table  I) were grown to mid-log phase in MB medium supplemented with 250 mg/liter leucine, uracil, adenine, and histidine at 30°C and transformed with pFZ1 using a LiAc protocol (36) to form strain 146pFZ1. Ura ϩ transformants were selected in YE agar containing adenine, leucine, and histidine. Overexpression of Ddb1p-His was confirmed by immunoblot analysis of a total cell lysate from 0.2 A 595 of a logarithmically growing cell culture. For this procedure, a lysate was prepared by vortexing the cells with acid-washed glass beads in SDS gel loading buffer (50 mM Tris, pH 6.8, 100 mM dithiothreitol, 2% SDS, 0.1% bromphenol blue, 10% glycerol) and then briefly sonicated on ice prior to being electrophoresed through a 4 -20% gradient SDS-PAGE gel (Bio-Rad). Protein was then transferred to a nitrocellulose membrane that was probed with a mouse monoclonal antibody raised against His 5 (catalog number 346690; Qiagen).
Construction of the ddb1 Deletion Strain FZ150 -Strain 146pFZ was used to produce the deletion of ddb1 ϩ by one step replacement (37). The kan r module was amplified by PCR using the KanMX2 plasmid as template. The primers were designed to produce a deletion of the complete ORF of ddb1 ϩ . The PCR product was purified using the Wizard DNA clean up system (Promega) and transformed into 146pFZ1 using a LiAc protocol (36). Transformant colonies were screened for homologous recombination by colony PCR using the primer Ϫ183 ddb1F (5Ј-AACCATCAACATCGTTGTTCG-3Ј), which is complementary to a region 5Ј of the ddb1 ϩ ORF, and 373R-KanMX2 (5Ј-CGGATGTGATGT-GAGAACTG-3Ј), which is complementary to the kan r module. The replacement of the ddb1 ϩ gene with the kan r module was confirmed by PCR using the primers Ϫ183ddb1F and 3887R-ddb1 (5Ј-ACAAAAAGT-GTGTAGGCTTGG-3). The latter sequence is located 668 nucleotides downstream from the ddb1 ϩ stop codon. To select for colonies of the ddb1⌬ strain that have lost pFZ1, cells were plated on YES agar plates containing 500 g/ml 5-fluoroorotic acid to select for ura4 Ϫ cells. Surviving colonies were then plated on YES agar plates containing G418 to assure the presence of the kan r gene. The resulting strain, which has the kan r gene substituted for the ddb1 ϩ gene and lacks the plasmid, is designated FZ150.
Construction of Ddb1p-Myc Fusion-Ddb1p was fused to a Myc tag at the C terminus by homologous recombination at the chromosomal ddb1 ϩ locus as described (37). The plasmid pFA6a-13Myc-kanMX6 (a kind gift of Dr. J. Q. Wu, Yale University) was used as template for the PCR reaction. Expression of the fusion protein was confirmed by immunoblot analysis using the monoclonal antibody 9E10 (BAbCo) directed against the Myc tag. The strain that expresses the Ddb1p-Myc fusion protein was designated FZ29.
UV and X Irradiation-Exponentially growing cells in YES medium at 30°C were diluted in water, plated onto YES agar plates, exposed as indicated to a 254-nm light source or a tungsten x-ray source and then incubated at 30°C in the dark. Colonies were counted after 4 -5 days.
For UV irradiation in liquid medium, exponentially growing cells in YES medium were washed twice with water and resuspended in water at 5 ϫ 10 6 cells/ml. Thirty-ml aliquots of the suspension were irradiated with 254 nm of light in a 15-cm Petri dish as described by Yasuhira et al. (38) and then the cells were collected by centrifugation, suspended in YES medium, incubated at 30°C for the times indicated, and fixed in 70% ethanol for 4,6-diamono-2-phenylindole dihydrochloride (DAPI) staining or in 100% methanol for indirect immunofluorescence.
MMS, H 2 O 2 , HU, and Thiabendazole (TBZ) Sensitivity Spot Tests-Exponentially growing cells in YES medium were resuspended in water at a density of 10 7 cells/ml, and 10-l aliquots of 1:10 serial dilutions were spotted onto YES medium plates containing increasing concentrations of MMS, H 2 O 2 , HU, or TBZ as indicated. Sensitivity to the compounds was observed after 3-4 days at 30°C. For exposure to MMS in liquid, cultures in YES medium were brought to 0.033 or 0.15% MMS at 30°C. Aliquots were removed at the indicated times, fixed in 70% ethanol for DAPI staining, or diluted 1:2000 and plated onto solid YES medium. Colonies were counted after 3-5 days at 30°C in the dark.
Indirect Immunofluorescent Microscopy-For ␣-tubulin immuocytologic staining, cells in mid-log phase were fixed by the double-aldehyde method (39). Briefly, cells in YES medium were brought to 3.7% formaldehyde, 0.2% gluteraldehyde and incubated with shaking for 90 min at 30°C, washed with phosphate-buffered saline, 1 mM EGTA and then digested with Novozym and zymolase. Cell membranes were permeabilized with 1% Triton X-100 and then the cells were treated with fresh 1 mg/ml sodium borohydride in phosphate-buffered saline containing 1 mM EGTA and 1.2 M sorbitol for 5 min. The cells were then incubated in PELBS (phosphate-buffered saline containing 1 mM EGTA, 0.1 M Llysine-HCl, 1% bovine serum albumin, 0.1% sodium-azide) for 30 min at room temperature before the addition of an anti-␣-tubulin antibody (monoclonal immunoglobulin YOL1/34; Accurate Chemical and Scientific Co.) diluted 1:10. Cells were washed in PELBS and then the secondary antibody, a Cy3-conjugated goat anti-rat antibody (Jackson Immunoresearch) diluted 1:50, was added.
For Ddb1p-His and Ddb1p-Myc cellular localization studies, mid-log phase cells were fixed in 100% methanol. The cells were treated as above with the omission of the sodium borohydride. The primary antibody was a rabbit polyclonal antibody directed against the His tag (His probe (G-18)) (Santa Cruz Biotechnology, Inc.) diluted 1:100 or a mouse monoclonal antibody directed against the Myc-tag (9E10; BabCo) diluted 1:200. As appropriate, the secondary antibodies were a donkey anti-rabbit, fluorescein isothiocyanate-labeled, or a donkey anti-mouse, Cy3-labeled antibody (Jackson Immunoresearch) diluted 1:50.
This study For Nop1p cellular localization, mid-log phase cells were fixed in 100% methanol. The primary antibody was mouse monoclonal D77 antibody (a generous gift of Dr. J.-P. Aris, University of Florida, Gainesville) diluted 1:1000. The secondary antibodies were a Cy3-or a fluorescein isothiocyanate-labeled donkey anti-mouse antibody (Jackson Immunoresearch) diluted 1:50.
For double staining of Ddb1p-His and Nop1p, mid-log phase cells fixed in 100% methanol were treated as described above and incubated overnight with the rabbit polyclonal antibody, His probe (G-18) diluted 1:100. The cells were than washed, incubated for 6 h in the Cy3-labeled donkey anti-rabbit secondary antibody diluted 1:50, fixed again in 100% methanol, washed, blocked in PELBS for 30 min, and incubated overnight with the mouse monoclonal D77 antibody. The cells were finally washed and incubated for 6 h with the fluorescein isothiocyanatelabeled donkey anti-mouse secondary antibody diluted 1:50. DNA was visualized with DAPI as described by Moreno et al. (36).
All slides were analyzed for immunofluorescence stain with a Zeiss Axiophot epifluorescence microscope equipped with an Optronics DEI 750 low light, cooled CCD color video in the Biological Imaging Facility of the College of Natural Resources at the University of California, Berkeley. Images were captured on a Macintosh computer and analyzed with Adobe Photoshop 6.0.
Confocal microscopy was performed on a Zeiss 510 confocal laser scanning microscope and captured using Zeiss 510 software on a Silicon Graphics O 2 R12k UNIX work station running Windows NT 4. Images were transferred to a Macintosh computer and analyzed using Adobe Photoshop 6.0. Cell lengths were measured using the public domain NIH Image program available at rsb.info.nih.gov/nih-image/.

Cloning of ddb1 ϩ and Construction of the ddb1⌬ Strain-
The sequence of the apparent S. pombe homologue of human DDB1 contains an open reading frame of 3219 nucleotides, does not contain introns, and encodes a predicted protein of 1072 amino acids with a molecular mass of 120 kDa (24 -26). The human DDB1 cDNA contains 3423 nucleotides, encoding a predicted protein of 1140 amino acids with a molecular mass of 127 kDa (1). The ddb1 ϩ coding sequence was amplified by PCR using S. pombe genomic DNA as template and cloned into the E. coli pCR T7/CT-TOPO vector to be in-frame with a C-terminal His 6 tag.
To attempt to construct a strain lacking ddb1 ϩ from a diploid strain, four independent heterozygote ddb1 ϩ /ddb1⌬ clones were obtained by gene replacement in diploid cells (37). However, each of the recombinant clones failed to sporulate. Therefore, production from a haploid strain was pursued. Because it was not known whether a deletion mutant of ddb1 ϩ would be viable, the C-terminal His-tagged ddb1 ϩ ORF was subcloned under the control of the heterologous, thiamine-regulated nmt1 promoter into the pREP2 shuttle vector that contains a ura4 ϩ marker (35) to form pFZ1. The pFZ1 plasmid was then transformed into the haploid 146 strain to form 146pFZ1. Expression of the fusion protein in the 146pFZ1 strain was confirmed by immunoblot analysis using antibody directed against the His tag. 146pFZ did not show any abnormal phenotype. This strain was then used to obtain the FZ150pFZ strain that carries a deletion of the entire ORF of the chromosomal ddb1 ϩ gene. The ddb1 ϩ gene was replaced by a kan r module by homologous recombination and confirmed by PCR as described under "Experimental Procedures." Two independent clones that had undergone homologous recombination at the ddb1 ϩ locus were grown on plates containing 5-fluoroorotic acid to induce the loss of pFZ1. Cells from the two clones showed indistinguishable phenotypes, so all further studies utilized one of them, designated FZ150.
The colony-forming ability of the ddb1⌬ strain, FZ150, was 63% that of the parental 146 strain. Microscopic examination of strain FZ150 cells stained with DAPI revealed a large number of abnormally long cells (Fig. 1A). Aberrant nuclear morphologies, including unequal distribution of DNA to daughter cells such as seen in cut (cell untimely torn) mutants (40), nuclear DNA fragmentation, and elongated nuclei were observed in 22% (66/300) of the deletion mutant cells (Fig. 1A). Thus Ddb1p appears to be required for proper chromosome segregation under normal growth condition in a sizeable fraction of these cells.
To show that these abnormal characteristics were a direct consequence of the null allele, pFZ1, which contains the ddb1 ϩ -His 6 fusion, or the empty vector, pREP2, were transformed back into the FZ150 mutant cells. FZ150pFZ1 cells showed normal nuclear morphology, whereas FZ150 pREP2 cells showed a large number of cells with abnormal nuclei (data not shown). The pFZ1 also specifically restored a normal length distribution (see Fig. 1B and Table II).
ddb1⌬ Cells Are Sensitive to UV Irradiation-Because human DDB has been shown to have a role in the processing of DNA damage following UV exposure (10), the recovery following UV irradiation of strain FZ150 was scored by colony-forming ability. For comparison, colony-forming abilities were scored for a rad13⌬ mutant that is null for the homologue of the human XPG protein (31), a rhp51⌬ mutant that is null for the homologue of the human RAD51 (32), and a cds1⌬ mutant that is null for the homologue of human CHEK2 and is implicated in the DNA replication checkpoint (33, 34) ( Fig. 2A).
The ddb1⌬ strain, FZ150, was more sensitive than the parental strain, 146, to killing by UV irradiation, especially in the low dose range in which the shoulder of the survival curve seems to have been replaced by an abnormally high sensitivity. Thus, at doses up to 50 J/m 2 , the mutant cells showed greater sensitivity than the parental, but at higher doses the surviving cells appear to decay at the same rate as the parental. This result could indicate the apparent presence of two subpopulations of cells, one with high sensitivity to UV irradiation compared with the parental strain and the other with normal sensitivity. A comparison of the two curves indicates that roughly 40% of the mutant cells were hypersensitive to UV irradiation.
The LD 50 following UV irradiation of the ddb1⌬ cells was estimated from Fig. 2A to be 30 -50 J/m 2 , versus 135 J/m 2 for the parental strain. LD 50 s of 17 and 21 J/m 2 were similarly estimated for the rad13⌬ mutant and the rhp51⌬ mutant, respectively. The relatively modest UV sensitivity of the ddb1⌬ strain compared with that of the rad13⌬ strain is reminiscent of the slight UV sensitivity of human XP-E cells, which are defective in DDB activity relative to strains defective in NER (41). Instead, the sensitivity of the ddb1⌬ cells resembled that of the cds1⌬ cells, which had an LD 50 of ϳ60 J/m 2 . cds1 ϩ is required to delay the passage through S phase following exposure to UV irradiation (42).
One explanation for the morphological abnormalities seen in unirradiated ddb1⌬ mutant cells and for the presence of a subpopulation of cells that are hypersensitive to UV irradiation would be abnormal cell cycle arrest following DNA damage. To investigate a possible role for Ddb1p in the G 2 /M checkpoint, which is the major checkpoint in S. pombe (27), the lengths of the parental and the ddb1 null mutant cells were measured 1, 2, and 3 h following UV irradiation. If the G 2 /M checkpoint were invoked, it would be expected that a fraction of the population would become elongated in response to the irradiation. Twenty-five and 150 J/m 2 were used for the ddb1⌬ and parental strains, respectively, to achieve roughly the same survival. The mutant strain showed an increased percentage of elongated cells, from 39% of unirradiated cells to 49% 3 h post UV irradiation. On the other hand, the fraction of elongated parental cells increased from 2 to 31%. The mutant cells clearly showed an elongation response, but one cannot conclude whether the increase in the frequency of elongated cells in response to the radiation was significantly different from that of the parental cells because of (i) the large number of such cells in the mutant population prior to irradiation and (ii) the necessity to use different doses of irradiation for the two cell strains to keep the survival in the same range for the two. In any event, qualitatively it appears that strain FZ150 is able to invoke the G 2 /M checkpoint.
Flow cytometric analysis of unirradiated and UV-irradiated parental and FZ150 cultures was performed using the nucleic acid stain Sytox green (data not shown). Compared with the parental cells, the unirradiated ddb1⌬ cells showed a broader peak indicating heterogeneity in DNA content. However, the abnormally large heterogeneity in cell size of the ddb1⌬ strain rendered the flow cytometric data difficult to interpret, especially following UV irradiation.
Microscopic examination of cells after DAPI staining was also performed following UV irradiation. In the ddb1⌬ cells, the percentage of cells with aberrant nuclear morphologies increased from 22% in unirradiated cells to 33% 3 h after irradiation. In the parental strain, however, 7 and 4% of the cells showed aberrant nuclear morphology 2 and 3 h post UV irra-diation, respectively. In the ddb1⌬ mutant, as in the parental strain, the major class of nuclear abnormality consisted of unequal chromosome segregation, and the second most abundant class consisted of fragmented chromosomes (data not shown). Taken together these results suggest a minor role, if any, for Ddb1p for invoking the G 2 /M DNA damage checkpoint.
Sensitivity of the ddb1⌬ Cells to X-rays, Methylmethane Sulfonate (MMS), HU, and H 2 O 2 -Like the case with UV irradiation, the sensitivity to X-rays was abnormal at the lower doses (Fig. 2B). Extrapolation of the exponential part of the curve indicated that roughly 40% of the mutant cells was hypersensitive to X-rays, the same value estimated for UV irradiation.
The ddb1⌬ mutant was also sensitive to MMS (Fig. 3). This sensitivity might be attributed to the fact that alkylated bases are repaired by DNA glycosylases, giving rise to apurinic/apyrimidinic sites, and mammalian DDB binds with high affinity to abasic sites (4). The ddb1⌬ strain is more sensitive to chronic MMS exposure than is the rad13⌬ mutant and less sensitive than is the rhp51⌬ mutant in a spot plate assay (Fig. 3A). Both rad13 ϩ and rhp51 ϩ play a significant role in the repair of DNA damaged with MMS (43). As expected (42), the cds1⌬ strain showed only a slight sensitivity to MMS.
To compare the various strains for sensitivity to acute exposure to MMS, the cells were treated in liquid culture with 0.15% MMS and then plated on MMS-free plates. The same relative sensitivities among the strains were found, except that the ddb1⌬ and the rad13⌬ strains appeared to be equally sensitive in this case (Fig. 3B). The ddb1⌬ cells were also examined by fluorescence microscopy after exposure to liquid medium containing 0.033% MMS. Aliquots were removed FIG. 1. Cells carrying the ddb1⌬ allele show nuclear abnormalities and an elongated phenotype. A, DAPI staining. Cells of strains 146 or FZ150, as indicated, were fixed in 70% ethanol, DAPI-stained, and examined by fluorescence microscopy as described under "Experimental Procedures." The arrows indicate DNA abnormalities that include nuclear fragmentation, elongated nuclei, and cut phenotype. B, distribution of cell lengths. Cells of strains as indicated that had been growing exponentially in rich medium were placed on a thin layer of 1% agarose with YES medium solidified on a slide, covered with a coverslip, and examined with a light microscope. Pictures of the living cells were taken using a CCD camera, and cell lengths were measured using the NIH Image program. Mean lengths and the fraction of cells over 15 m are given in Table II. hourly, fixed with ethanol, stained with DAPI, and examined for nuclear abnormalities. No increase in nuclear aberrations in either the parental 146 or the mutant FZ150 strains was detected for up to six h of exposure (data not shown). The increase in the frequency of nuclear abnormalities in ddb1⌬ cells after UV irradiation but not after exposure to MMS suggests a role for Ddb1p in responding specifically to UV damage as is found for mammalian DDB. However, the basis of the sensitivity of ddb1⌬ cells to MMS is unclear.
The ddb1⌬ mutant was not abnormally sensitive to H 2 O 2 on plates (Fig. 4A). H 2 O 2 causes DNA damages similar to those induced by X-rays, with the exception of two-ended doublestrand breaks (12), and mammalian DDB does not interact with DNA double-strand breaks. Similarly to the ddb1⌬ cells, the rad13⌬, rhp51⌬, and cds1⌬ strains also do not appear to be abnormally sensitive to H 2 O 2 . Resistance to H 2 O 2 of a rad13⌬ strain was also observed by Yasuhira et al. (38).
The sensitivity of ddb1⌬ cells to HU was also investigated by a spot plate assay (Fig. 4B). HU inhibits ribonucleotide reductase and induces cell cycle arrest at the beginning of S phase. The ddb1⌬ strain was more sensitive than the parental or the rad13⌬ strains to HU but considerably less sensitive than the rhp51⌬ or cds1⌬ strains. Therefore, the ddb1⌬ cells are only slightly defective either in the intra-S checkpoint or in recombinational DNA repair. However, the basis of their sensitivity to HU is not clear.
Ddb1p Is Localized in the Nucleus and the Nucleolus-When a human DDB1-GFP fusion protein was transiently expressed at high levels from a plasmid in human diploid fibroblasts, it was localized predominantly in the cytoplasm, though some was seen in the nucleus (44). To study the cellular localization of S. pombe Ddb1p, strain FZ29, which carries a Ddb1 ϩ -Myc fusion under the control of the natural ddb1 ϩ promoter, was constructed as described under "Experimental Procedures." Expression of the recombinant protein was confirmed by an immunoblot assay with anti-Myc antibody. As observed by indirect immunofluorescence, Ddb1p-Myc localized both to the nucleus and the nucleolus, with darker staining in the nucleolus (Fig. 5, A-C). No signal was observed in the cytoplasm. Analysis of mitotic cells showed that Ddb1p-Myc remained associated with the nucleolar material during mitosis (Fig. 5, A-C, white arrows). As a comparison for the nucleolar staining, an antibody against Nop1p/fibrillarin (45) was used. Nop1p localizes to the nucleolus throughout the cell cycle. In Fig. 5D, one cell in interphase and one in mitosis (white arrow) are shown. Clearly the Ddb1p-Myc was present in the nucleolus. Unfortunately colocalization experiments using the anti-Myc and the anti-Nop1p antibodies could not be performed in this experiment, because both were mouse monoclonal antibodies.
As noted above, expression of a Ddb1p-His 6 fusion protein  Table I. FIG. 3. Cells carrying the ddb1⌬ allele are sensitive to MMS. A, chronic exposure. Parental and mutant cells were grown into log phase in YES medium at 30°C and then resuspended in water at a density of 10 7 cells/ml. Ten-l aliquots of 1:10 serial dilutions were spotted onto solid YES medium containing MMS as indicated and allowed to grow at 30°C for 4 days. B, acute exposure. Cells were grown into log phase in YES medium at 30°C then, at time 0, MMS was added to 0.15%. At the indicated times aliquots were diluted in water and plated onto solid agar medium. Colony-forming ability was determined after 4 days at 30°C in the dark. Strains are described in Table I. from the pFZ1 plasmid rescued the ddb1⌬ phenotype of strain FZ150. These cells (FZ150pFZ1) were used to investigate the intracellular localization of overexpressed Ddb1p-His protein using antibody raised against the His tag. In an exponentially growing culture in which Ddb1p-His was strongly overexpressed under the control of the nmt1 promoter in the absence of thiamine, the protein accumulated in the cytoplasm (Fig 6A). However, when thiamine was added to the growth medium to reduce the level of expression, Ddb1p-His appeared to be present somewhat in the cytoplasm and the nucleus but to be most concentrated in the nucleolus (Fig. 6B, yellow triangles). Colocalization analysis using the Nop1p/fibrillarin antibody confirmed that Ddb1p-His accumulates in the nucleolus in both interphase (Fig.  6B, yellow triangles) and mitotic (Fig. 6B, white arrows) cells.
Transiently overexpressed DDB1-GFP in human fibroblasts was predominantly cytoplasmic but translocated to the nucleus following UV irradiation (44). Therefore, Ddb1p-His cellular localization following UV irradiation was investigated in logarithmically growing FZ150pFZ1 cells in the presence or the absence of thiamine to regulate the level of Ddb1p expression. Cells were UV-irradiated with 50 or 150 J/m 2 and then allowed to recover in YES medium and were sampled hourly for 3 h. The cellular localization of the protein did not change after UV irradiation, however, remaining predominantly nucleolar when expressed at reduced levels in the presence of thiamine (data not shown). In summary, with a physiological level of expression, Ddb1p showed an almost exclusively nuclear and nucleolar localization.
Nucleolar Segregation in ddb1⌬ Cells-Because Ddb1p stained most intensely in the nucleolus, the nucleolar distribution within the ddb1⌬ cells was examined by indirect immunofluorescence using the Nop1p antibody (Fig. 7). Nop1p localizes at the site of rDNA transcription and remains associated with the nucleolar region during mitosis (46). Therefore it is a marker for rDNA genes throughout mitosis. S. pombe rDNA genes form two clusters of 100 -150 10.9-kb tandem repeats at both ends of chromosome III (47,48) so that the Nop1p antibody staining can also indicate patterns of chromosome III segregation. Strikingly, whereas staining patterns appeared to be normal in non-mitotic cells, in 53% (17/32) of the ddb1⌬ mitoses observed, Nop1p and therefore chromosome III, segregated unequally between the daughter cells (Fig. 7, right panels; arrows indicate mitoses). By contrast, Nop1p segregated equally in all of the mitoses observed in the parental cells (33/33) (Fig. 7, left panels; arrows indicate mitotic cells). Moreover, in 28% of the aberrant mitoses in the ddb1⌬ cells, the Nop1p signal is present only in one of the daughter cells. These results suggest that the chromatin abnormalities observed, at least in some of the mutant cells, were because of chromatid non-disjunction or premature disjunction.
Microtubule Staining in ddb1⌬ Cells and Sensitivity to TBZ-To investigate whether the failure of normal chromosome segregation in strain FZ150 is a consequence of a role of Ddb1p in microtubule formation, indirect immunofluorescence with an antibody directed against ␣-tubulin was carried out with exponentially growing cells. Fig. 8A shows a mitotic cell that displays lagging chromosomes. Indeed, 30% (6/20) of the mitosis scored showed such lagging chromosomes on the spindle.
Lagging chromosomes, which are considered to be single chromatids (49), represent a common phenotype in S. pombe mutants that are defective in chromosomal segregation, including rik1, clr4, clr6, and swi6 mutants (50 -53). It is noteworthy that Rik1p has been reported to share secondary structure features with the DDB1 family of proteins (26). Rik1p, Clr4p, Clr6p, and Swi6p are required for proper heterochromatin assembling at centromers, telomeres, mating loci, and rDNA and show high sensitivity to the microtubule destabilizing drug, TBZ (50 -53). The sensitivity to TBZ is thought to be because of the direct or indirect interaction of the products of these genes with microtubules at the kinetochore (52). Indeed, the ddb1⌬ FZ150 strain was also sensitive to TBZ (Fig. 8B), suggesting FIG. 4. Sensitivity of the ddb1⌬ cells to H 2 O 2 and HU. Cells were grown to log phase in rich medium at 30°C and then resuspended in water at a density of 10 7 cells/ml. Ten-l aliquots of 1:10 serial dilutions were spotted on solid YES medium containing H 2 O 2 (A) or HU (B) as indicated and allowed to grow at 30°C for 4 days in the dark. Strains are described in Table I. that Ddb1p also might interact with the centromeric heterochromatin. However, rhp51⌬ and the cds1⌬ mutants (but not a rad13 null mutant) showed similar sensitivity to TBZ so that other explanations of the TBZ sensitivity are possible. DISCUSSION DDB from mammalian cells is a heterodimer of DDB1 and DDB2 (2)(3)(4). As noted in the Introduction, DDB1 is an evolutionarily conserved protein; sequence homologues have been identified in mammals, worms, insects and plants. DDB2, however, is less conserved, and DDB2 sequence homologues have been identified only in mammals and plants (24). 3 Members of the DDB1 family do not contain nuclear localization signals, whereas all of the DDB2 homologues contain multiple nuclear localization signals (1). 4 Nevertheless, when the Ddb1p-Myc fusion protein was expressed under the control of the natural chromosomal promoter, an almost exclusively nuclear/nucleolar localization was observed. In human cells, DDB2 is proposed to bind to DDB1 in the cytoplasm and translocate it to the nucleus (44,54). This hypothesis is supported by the finding that two naturally occurring XP-E mutations of DDB2, 82TO and 2RO, are deficient in stimulating the nuclear accumulation of DDB1 (54). Therefore, one might expect there to be a functional homologue of human DDB2 in S. pombe that is responsible for the import of Ddb1p into the nucleus. Alternatively, it is also possible that a S. pombe DDB2 functional homologue does not exist. In mammalian cells, DDB2 is induced in response to UV irradiation (19), and DDB1 is subsequently transported to the nucleus. Moreover, this induction is dependent upon p53 (55). By contrast, S. pombe Ddb1p does not relocate following UV irradiation, because it is constitutively present in the nucleus, and S. pombe does not have a p53 homologue.
The FZ150 ddb1⌬ strain was slightly sensitive to UV irradiation compared with the NER-defective rad13⌬ strain, suggesting that Ddb1p does not play a principal role in NER of 3 Unpublished data. 4 Unpublished data. UV-induced DNA damage. The FZ150 strain also showed a slight sensitivity to X-rays. In both cases, these cells showed the presence of one subpopulation of about 40% with a high sensitivity to the radiation and a second subpopulation with normal sensitivity. Thirty-nine percent of the FZ150 cell population also had an elongated phenotype, suggesting that these elongated cells could be the subpopulation that is sensitive to irradiation. However, because it appeared that roughly the same proportion of cells do not form colonies, this correlation may be coincidental as it may be that the elongated cells do not replicate.
The basis of the sensitivity of the ddb1D cells to MMS remains unclear, because no increase in aberrant nuclei was detected following exposure to MMS in liquid culture. Possibly Ddb1p is involved in base excision repair, but then a sensitivity to H 2 O 2 would also have been expected. It is interesting that the rhp51D strain is also sensitive to MMS but not to H 2 O 2 , suggesting a possible role of Ddb1p in recombinational repair rather than in excision repair.
Ddb1p appears also to be required for proper chromosomal segregation under normal growth conditions. Observations of chromosome III with indirect immunofluorescence of the rDNA-associated protein, Nop1p, showed that the chromatin abnormalities observed in at least some of the ddb1⌬ cells appeared to be because of chromatid non-disjunction or premature disjunction. Neuwald and Poleksic (26) used hidden Markov models of structural repeats to predict the presence of ␤-propeller domains in the DDB1 family of proteins and in other structurally related proteins, including the fission yeast Rik1p-silencing protein. rik1 ϩ belongs to a class of S. pombesilencing genes that includes crl4 ϩ , clr6 ϩ , swi6 ϩ , and hst4 ϩ . These genes have been implicated in silencing of the matingtype loci, the telomeres, the centromeres, and the rDNA repeats and also to be essential elements in the assembly of a heterochromatin-like structure (50 -53, 56). Mutations in rik1, as well as in clr4, clr6, swi6, and hst4, also result in a defective chromatin structure leading to abnormal chromosome segregation. Most similar to the ddb1⌬ phenotype is the hst4⌬ phenotype. These cells show the same elongation and abnormal nuclear morphology as the ddb1⌬ cells (56). Moreover, like Ddb1p, Hst4p shows a preferential nucleolar localization, and the hst4⌬ cells are sensitive to UV irradiation (56).
Genomic instability in the rik1, clr4, clr6, swi6, and hst4 mutants is thought to be caused by a defect in centromeric silencing, resulting in defective centromeric function (50 -53).
Because clr4, rik1, swi6, and hst4 deletion mutants are highly sensitive to the microtubule-destabilizing drug TBZ, a direct or indirect interaction of the products of these genes with microtubules at the kinetochore was proposed (52). The ddb1⌬ strain is also sensitive to TBZ, indicating that the ddb1⌬ strain might also be defective in an interaction of the centromere with microtubules. Indeed the chromosomal segregation abnormalities observed in the ddb1⌬ cells, the presence of lagging chromosomes, and the Ddb1p nucleolar localization, together with the structural homology proposed between Ddb1p and Rik1p, suggest that Ddb1p might help to maintain normal heterochromatin structure at centromeric regions and possibly also at rDNA repeats. Should this be the case, the absence of Ddb1p could result in mispositioning of other unidentified proteins and consequent improper centromere assembly and genomic instability. In this light, the sensitivity of the ddb1⌬ strain to a variety of DNA-damaging agents could be interpreted as the result of inefficient DNA repair because of a general chromatin assembly defect rather then a specific defect in a DNA repair pathway.
The organization of rDNA genes and of the centromeric regions in S. pombe has been shown to be more closely related to that in animal cells than that in S. cerevisiae (47,48,(57)(58)(59)(60)(61)(62). A function for the S. pombe Ddb1p in maintaining proper chromatin structure in the rDNA genes and the centromeric regions might then explain why a DDB1 homologue is not present in S. cerevisiae but is found in S. pombe. Aravind et al. (63) compared 4,344 protein sequences from S. pombe with all available eukaryotic sequences and identified those genes that are conserved in S. pombe and non-fungal eukaryotes but are missing or highly diverged in S. cerevisiae. Their results suggested coelimination of functionally interacting sets of proteins. In particular, a pattern of coelimination was seen among proteins involved in chromatin remodeling, including Swi6p, Clr4p, Rik1p, Ddb1p, and the cullin 4A ortholog. As previously noted, both human DDB1 and DDB2 are ubiquitinated by Cullin 4A.
A stimulus for this study was ultimately to obtain clues to the function of mammalian DDB1, because naturally occurring mutations of DDB1 are not known. Although it is not clearly established that the mammalian and S. pombe sequence homologues are functional homologues, as well, the very potent defects that result as a consequence of the S. pombe ddb1 deletion would suggest that DDB1 mutations in mammalian cells are likely to be lethal. Moreover, the chromosome segregation defect observed in the ddb1⌬ mutant might suggest FIG. 8. Microtubule staining in ddb1⌬ cells and sensitivity to TBZ. A, confocal microscopic image of a mitotic ddb1⌬ cell that shows an abnormal mitotic phenotype and lagging chromosomes. The sample was stained with anti-␣-tubulin antibody and DAPI as described under "Experimental Procedures." B, sensitivity to TBZ. 10-Fold serial dilutions of log phase cells spanning the range from 10 5 to 10 cells were spotted onto YES agar plates containing TBZ as indicated. Plates were photographed after 4 days at 30°C in the dark. Strains are described in Table I. that subtle alterations in the DDB1 gene such as those caused by single nucleotides polymorphisms could lead to genomic instability and therefore to increased susceptibility to tumorigenesis.