A Human Homologue of the Schizosaccharomyces pombe rad1 + Checkpoint Gene Encodes an Exonuclease*

In the fission yeast Schizosaccharomyces pombe the rad1 + gene is required for both the DNA damage-dependent and the DNA replication-dependent cell cycle checkpoints. We have identified a human homologue of the S. pombe rad1 + gene, designated Hrad1, as well as a mouse homologue: Mrad1. Two Hrad1alternative splice variants with different open reading frames have been identified; one codes for a long form, Hrad1A, and the other encodes a short form because of N-terminal truncation, Hrad1B. Hrad1A has 60% identity to the S. pombe rad1 +sequence at the DNA level and 49% identity and 72% similarity at the amino acid level. Northern blot analysis indicates elevated levels of expression in testis and cancer cell lines. Chromosomal localization by fluorescence in situ hybridization indicates thatHrad1 is located on chromosome 5p13.2–13.3. This region is subject to loss of heterozygosity in several human cancers. Hrad1 also shares homology with the Saccharomyces cerevisiaeRAD17 and Ustilago maydis REC1 proteins. REC1 has previously been characterized as a 3′ → 5′ exonuclease with a C-terminal domain essential for cell cycle checkpoint function. We have expressed and purified polyhistidine-tagged fusions of Hrad1A and Hrad1B and show that HisHrad1A has 3′ → 5′ exonuclease activity, whereas HisHrad1B lacks such activity. The biological functions of the two proteins remain to be determined.

In the fission yeast Schizosaccharomyces pombe the rad1 ؉ gene is required for both the DNA damage-dependent and the DNA replication-dependent cell cycle checkpoints. We have identified a human homologue of the S. pombe rad1 ؉ gene, designated Hrad1, as well as a mouse homologue: Mrad1. Two Hrad1 alternative splice variants with different open reading frames have been identified; one codes for a long form, Hrad1A, and the other encodes a short form because of N-terminal truncation, Hrad1B. Hrad1A has 60% identity to the S. pombe rad1 ؉ sequence at the DNA level and 49% identity and 72% similarity at the amino acid level. Northern blot analysis indicates elevated levels of expression in testis and cancer cell lines. Chromosomal localization by fluorescence in situ hybridization indicates that Hrad1 is located on chromosome 5p13.2-13.3. This region is subject to loss of heterozygosity in several human cancers. Hrad1 also shares homology with the Saccharomyces cerevisiae RAD17 and Ustilago maydis REC1 proteins. REC1 has previously been characterized as a 3 3 5 exonuclease with a C-terminal domain essential for cell cycle checkpoint function. We have expressed and purified polyhistidine-tagged fusions of Hrad1A and Hrad1B and show that HisHrad1A has 3 3 5 exonuclease activity, whereas HisHrad1B lacks such activity. The biological functions of the two proteins remain to be determined.
For any dividing cell it is essential to maintain the integrity of the genome ensuring that DNA replication occurs once per cell cycle and that identical chromosomal copies are distributed equally to two daughter cells at mitosis. When cells are subjected to conditions that interfere with DNA replication or spindle assembly or that cause damage to DNA, cell-cycle progression halts, permitting cell cycle phase completion or DNA repair. These control mechanisms are referred to as "checkpoints" (reviewed in Refs. [1][2][3]. The loss of checkpoint control in mammalian cells, most notably through the inactivation of p53, results in genomic instability leading to the amplification, rearrangement, or loss of chromosomes, events that commonly occur in cancer cells (2,4).
Much of our current knowledge about checkpoint control has been obtained from studies using budding (Saccharomyces cer-evisiae) and fission (Schizosaccharomyces pombe) yeast (reviewed in Ref. 5). In the fission yeast at least two distinct checkpoint pathways have been identified: the DNA replicationdependent checkpoint pathway that ensures that S phase is completed before M phase is initiated and the DNA damagedependent checkpoint pathway that acts to halt the cell cycle when genomic integrity is compromised (5). The products of six genes, rad1 ϩ , rad3 ϩ , rad9 ϩ , rad17 ϩ , rad26 ϩ , and hus1 ϩ , have been identified in S. pombe as essential components of both checkpoint pathways (5). In S. cerevisiae the two checkpoint pathways are genetically separable; four gene products, RAD9, RAD17, RAD24, and MEC3, have been identified as essential for the DNA damage-dependent checkpoint pathway, and three gene products, POL2, RFC5, and DPB11, are essential for the DNA replication-dependent checkpoint pathway (reviewed in Refs. [3][4][5]. Several of the S. pombe checkpoint genes have structural homologues in the budding yeast and further conservation across eukaryotes has recently been demonstrated with the cloning of two human homologues of S. pombe rad3 ϩ , ATM (ataxia telangiectasia mutated) (6) and ATR (ataxia telangiectasia and rad3 ϩ related) (7,8), and a human homologue of S. pombe rad9 ϩ , Hrad9 (9).
The identification and characterization of human homologues of yeast checkpoint genes provides clear evidence that at least some checkpoint pathways are conserved between mammals and yeast. Currently little is known about the biochemistry of checkpoint control. The genetic data in yeast suggest that a complex of proteins mediates the monitoring of replication-specific structures and damaged DNA (10). Furthermore, recent biochemical studies in S. pombe and humans suggest that the cell cycle arrest in response to DNA damage is brought about by the activation of a signal transduction pathway involving the putative protein kinases ATM/ATR and Hchk1, resulting in inhibitory phosphorylation of Cdc25 and subsequent stabilization of the inhibitory Tyr 15 phosphorylation of Cdc2 (7,(11)(12)(13)(14).
Although a great deal of progress has been made in identifying the kinase components of the signal transduction pathway mediating cell cycle arrest (6 -8, 12), there has been little progress in identifying the sensing mechanisms that activate the checkpoint pathways. The S. pombe rad1 ϩ gene acts early in the checkpoint pathway (15) and presumably plays a role in the transmission of information regarding the state of the genome to the checkpoint control mechanism. The rad1-1 mutant is extremely sensitive to all DNA damaging agents (15,16) and fails to invoke a G 2 arrest in response to ionizing radiation yet is DNA repair-competent, indicating a loss in the ability to recognize the presence of damaged DNA (15). The rad1-1 mutant is also sensitive to hydroxyurea, indicating an uncoupling between the completion of S phase and entry into mitosis (15), and is synthetically lethal when combined with mitosis-promoting mutations such as wee1-50 (15). The S. pombe rad1 ϩ gene has been cloned (17), and the predicted amino acid sequence shows significant sequence similarity to the Ustilago maydis protein Rec1 (18) and S. cerevisiae RAD17 (20). Mutations in these genes lead to cell cycle checkpoint defects and other phenotypes associated with alterations in DNA repair and recombination (19,20).
Purified Rec1 protein has been shown to have a 3Ј 3 5Ј exonuclease activity (21), suggesting that Rec1/Rad1/RAD17 may be involved in modifying DNA damage (22,23). Studies of Rec1 truncation mutants have demonstrated that the nuclease activity is located in the N-terminal half of the protein. However the Rec1-dependent checkpoint function requires the Cterminal region of the protein and does not require the exonuclease activity (24,25). In addition, Rec1 mutants that lack exonuclease activity have a 100-fold higher rate of spontaneous mutation, indicating that the exonuclease function may be required for mismatch repair (24). Thus it would appear that the Rec1 protein is bifunctional, having an N-terminal exonuclease domain and a C-terminal checkpoint domain.
In this report we describe the cloning and characterization of a novel human cDNA, designated Hrad1, which is highly similar to the S. pombe rad1 ϩ checkpoint gene. We also report the cloning of a mouse homologue designated Mrad1. We show that Hrad1 is subject to alternative splicing giving rise to two potential open reading frames (ORFs) 1 coding for a full-length and a truncated form of Hrad1, designated Hrad1A and Hrad1B, respectively. We have expressed in Escherichia. coli and affinity purified N-terminal polyhistidine-tagged fusions of Hrad1A and Hrad1B and show that Hrad1A but not Hrad1B has 3Ј 3 5Ј exonuclease activity. Hrad1B corresponds to the C-terminal domain of Rec1 that has been implicated solely in checkpoint function. Thus we can speculate that Hrad1A plays a role in the recognition and processing of DNA ends and also in checkpoint function, whereas the role of Hrad1B may be restricted to checkpoint control.

EXPERIMENTAL PROCEDURES
Cloning and Sequencing of Human and Mouse rad1-A search for sequences similar to S. pombe rad1 ϩ was carried out using the TBLASTN program (26) against the GenBank TM data base. Deduced amino acid sequences were aligned using the CLUSTALW program, and similarity was determined with a blosum62 amino acid substitution matrix. An expressed sequence tag (EST) clone with significant sequence similarity to S. pombe rad1 ϩ was identified and obtained from the Integrated Molecular Analysis of Genomes and their Expression (I.M.A.G.E.) Consortium.
DNA sequencing was carried out on double-stranded plasmid DNA with dye terminator chemistry as prescribed by the manufacturer (Perkin-Elmer/Applied Biosystems), and the products were resolved on an ABI Prism™ 377 Automated Sequencer. The complete insert sequence of the EST clone was determined.
To extend this sequence and to examine the possibility of alternative splicing, 5Ј and 3Ј RACE-PCR was carried out according to the manufacturer's instructions (CLONTECH).
For the 5Ј RACE, two gene-specific primers were designed to the 5Ј end of the putative Hrad1 ORF: GSP1 (5Ј-ATTTGTAGGACTTCACT-CGTCATAT-3Ј) and GSP2 (5Ј-TGTGTATTGATTTTGCAGACT GTCA-3Ј). These were used in a nested PCR reaction with Marathon-Ready human placental cDNA (CLONTECH) as the template. The first PCR reaction made use of the GSP1 PCR primer for the 3Ј end, combined with the AP1 primer from CLONTECH that is complementary to an adaptor ligated to the 5Ј end of all Marathon-Ready cDNAs. The second PCR reaction started from 1-5 l of the first and used the GSP2 in combination with a nested AP2 primer (CLONTECH). The reaction conditions for the first RACE-PCR were 30 cycles of 30 s at 95°C, 30 s at 65°C and 1 min at 72°C. The subsequent nested PCR used two-step cycles, each consisting of a 30 s denaturation step at 95°C and a combined annealing and elongation step for 2 min, which was carried out at 72°C for the first five cycles, at 70°C for the five subsequent cycles and at 68°C for the remaining 25 cycles.
For the 3ЈRACE, two gene-specific primers were designed to the 3Ј end of the putative Hrad1 ORF: GSP3 (5Ј-CAAGGTTATGGTTAC-CCTTTGATG-3Ј) and GSP4 (5Ј-TCAATACACAGGAACCTGAGGAG-3Ј). These were used in a nested PCR reaction with Marathon-Ready human fetal cDNA (CLONTECH) as the template. The first PCR reaction made use of the GSP3 PCR primer for the 5Ј end, combined with the AP1 primer from CLONTECH that is complementary to an adaptor ligated to the 3Ј end of all Marathon-Ready cDNAs. The second PCR reaction started from 1-5 l of the first and used the GSP4 primer in combination with a nested AP2 primer (CLONTECH). After a 1-min denaturation at 94°C, the reaction conditions for the first RACE-PCR were 30 cycles of 30 s at 95°C and 4 min at 68°C, concluding with a 7-min extension at 72°C. The subsequent nested PCR also used twostep cycles, each consisting of a 30 s denaturation step at 94°C and a combined annealing and elongation step for 4 min, which was carried out at 72°C for the first five cycles, at 70°C for the five subsequent cycles and at 68°C for the remaining 25 cycles. A final 7-min extension step at 72°C concluded the PCR.
The PCR reaction products were resolved by agarose gel electrophoresis, and specific reaction products were excised, purified using the QIAQuick gel extraction kit (Qiagen), and ligated into the pCR2.1-TOPO vector (Invitrogen). The insert sequence of 10 independent clones was determined and compared with the putative Hrad1 cDNA sequence. The sequences obtained separated into two classes generating two ORFs that we designated Hrad1A and Hrad1B.
To determine the DNA sequence of mouse rad1 (Mrad1), mouse EST clones were identified by screening the GenBank TM with Hrad1, and three independent EST clones were obtained from the I.M.A.G.E. consortium. The insert sequence for each clone was determined and aligned as described for Hrad1.
Northern Blot Analysis-Two multiple human tissue Northern blots (CLONTECH) and a human cancer cell line Northern blot (CLON-TECH) were hybridized with a full-length Hrad1B cDNA probe, labeled with [␣-32 P]dCTP by random hexamer priming with the Prime-a-Gene Kit (Promega). The blots were hybridized in ExpressHyb solution according to the manufacturer's instructions (CLONTECH) and washed at high stringency (0.1 ϫ SSC, 0.1% SDS, 50°C, 2 ϫ 20 min) and exposed to Kodak X-Omat autoradiography film with intensifying screens at Ϫ70°C. The blots were then rehybridized with a human ␤-actin probe (CLONTECH) labeled with [␣-32 P]dCTP as described above to verify the intactness of the RNA samples and confirm the presence of comparable amounts of RNA across lanes.
After stripping (0.1 ϫ SSC, 0.1% SDS, 90°C, 2 ϫ 20 min), one of the blots was rehybridized with an oligonucleotide specific to the alternatively spliced region of Hrad1A, OAP103 (5Ј-CTGGAATATTTCAG-GAGTTTAAA-3Ј), to discriminate between the transcripts. The oligonucleotide was end-labeled in a standard reaction using T4 polynucleotide kinase and [␥-32 P]ATP and hybridized to the blot at 37°C, and the blot was washed as described for the previous hybridizations.
In addition, a multiple mouse tissue Northern blot (CLONTECH) and a mouse developmental blot (CLONTECH) were hybridized with a mouse rad1 cDNA probe (AvaI/PstI fragment) labeled by random hexamer priming as described above. The blots were washed and exposed to film and then rehybridized with the control ␤-actin probe; all of these steps were carried out as described for the previous hybridizations.
Bacterial Expression and Purification of HisHrad1A and HisHrad1B-The following primers were used to amplify the complete coding regions of Hrad1A and Hrad1B: OAP105 (5Ј-GCACAGATCTC-CCCTTCTGACCCAACAGATC-3Ј) and OAP110 (5Ј-GCACAGATCTT-GTTACCAAGGTTATGGTTAC-3Ј) for the 5Ј end of Hrad1A and Hrad1B, respectively, each in combination with the 3Ј PCR primer OAP033 (5Ј-GCCGAATTCTCAAGACTCAGATTCAGGAA-3Ј). The PCR product was digested with BglII and EcoRI and cloned into the corresponding sites of pRSETB (Invitrogen). This creates a fusion protein with the following amino acid sequence N-terminal to the second codon of Hrad1A or Hrad1B: MRGSHHHHHHGMASMTGGQQMGRDLYDD-DDKDPSS. The insert sequence of both constructs was verified.
The plasmids were transformed into E. coli BL21(DE3) cells, and a single colony of each was inoculated into 10 ml of LB medium (containing 10 g/liter Bacto-Tryptone (Difco), 5 g/liter Bacto Yeast extract (Difco), and 10 g/liter NaCl (Merck), brought to pH 7 with 10 N NaOH and supplemented with 100 g/ml ampicillin) and incubated at 250 rpm at 30°C overnight. 4 ml of the overnight culture was added to 400 ml of medium in a 2-liter Erlenmeyer flask and incubated until an A 600 of 0.5 was reached. High level expression was induced by the addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.5 mM and a further incubation for 20 h before the cells were harvested by centrifugation. Lysates from the harvested cells were separated into the soluble and insoluble fractions using standard procedures (30).
In the case of Hrad1A, approximately 20% of the expressed protein as judged by Coomassie staining of SDS-polyacrylamide gels was found to be present in the soluble fraction. Affinity purification was carried out using Ni-NTA-agarose essentially as described by the manufacturer (QIAGEN). The soluble material was incubated with the Ni-NTA-agarose in batch for 1 h at 4°C on a revolving wheel. The column was then poured and washed with 5 column volumes of sonication buffer (50 mM Tris, pH 8.0, 50 mM NaCl, 1 mM EDTA, 0.5 mM PMSF) followed by Buffer RB, pH 8.0 (10 mM Tris, 10% glycerol, 1 mM 2-mercaptoethanol, 0.5 mM PMSF), Buffer RB, pH 6.3 and Buffer RB, pH 5.5. The bound protein was eluted in Buffer RB, pH 8.0, containing 250 mM imidazole. Nine 1-ml fractions were collected and analyzed by SDS-polyacrylamide gel electrophoresis. Those fractions containing HisHrad1A were pooled and dialyzed overnight (Spectra/Por 6,000 molecular weight cut-off membrane) at 4°C against 1 liter of Buffer RB. The dialyzed material was centrifuged for 1 h at 4°C and 100,000 ϫ g. No pellet was visible by naked eye inspection. For Hrad1B, the expressed protein was found to be almost exclusively present in the insoluble fraction. Inclusion bodies were prepared (30) and dissolved in Buffer A (6 M guanidine HCl, 0.1 M NaH 2 PO 4 , 0.01 M Tris, pH 8.0, 1 mM 2-mercaptoethanol, 0.5 mM PMSF) at 5 ml/g wet weight. The soluble material was incubated with Ni-NTAagarose for 1 h. The column was poured and washed with 5 column volumes of Buffer A, followed by Buffer B (8 M urea, 0.1 M NaH 2 PO 4 , 0.01 M Tris, pH 8.0, 1 mM 2-mercaptoethanol, 0.5 mM PMSF) and Buffer C (8 M urea, 0.1 M NaH 2 PO 4 , 0.01 M Tris, pH 6.3, 1 mM 2-mercaptoethanol, 0.5 mM PMSF). The bound protein was then eluted in 10 ml of Buffer B containing 250 mM imidazole. The protein was refolded by rapid 10-fold dilution in Buffer RB and dialyzed against 1 liter of Buffer RB, pH 8.0, with several changes. The dialyzed material was centrifuged for 1 h at 4°C and 100,000 ϫ g. No pellet was observed. The final protein concentration was determined by Bradford assay (Bio-Rad), and an aliquot was separated on a 12% SDS-polyacrylamide gel to examine the purity and integrity.
Exonuclease Assays-The DNA used as exonuclease substrate was plasmid pBSII (KS ϩ ) (Stratagene), digested to completion with Sau3A. The DNA was either labeled at the 3Ј termini by incorporation of [␣-32 P]dCTP (ϳ3000 Ci/mmol) using the Klenow fragment of E. coli DNA polymerase or at the 5Ј termini with [␥-32 P]ATP (ϳ3000 Ci/mmol) using T4 polynucleotide kinase, prior to filling in 3Ј recessed termini with unlabeled dNTPs.
The exonuclease assay was carried out in a 20-l volume containing 50 mM Tris acetate, pH 9, 10 mM magnesium acetate, 1 mM dithiothreitol, 0.1 mM EDTA and substrate DNA. Exonuclease assays were carried out with 2 nmol (as total DNA nucleotide) of end-labeled DNA containing 1.6 pmol of 32 P-nucleotide as the terminal label at a specific activity of 2 ϫ 10 3 cpm/pmol. Reactions were started by the addition of enzyme, incubated at 37°C for 30 min unless otherwise stated and terminated by the addition of 30 l of an ice-cold solution of salmon sperm DNA (0.5 mg/ml) in 25 mM EDTA and 50 l of 10% trichloroacetic acid. The mixture was held on ice for 10 min and centrifuged at 4°C and 10,000 ϫ g for 10 min. A 75-l aliquot of the supernatant was removed and mixed with 400 l of scintillant (UltimaGold™, Packard) for determination of radioactivity by liquid scintillation counting (Tri-Carb 2100TR, Packard).
Fluorescence in Situ Hybridization Studies (FISH)-Chromosomal localization was carried out by SeeDNA Inc. (Toronto, Ontario, Canada). Lymphocytes isolated from human blood were cultured in ␣-minimal essential medium supplemented with 10% fetal calf serum and phytohaemagglutinin at 37°C for 68 -72 h. The lymphocyte cultures were treated with bromodeoxyuridine (0.18 mg/ml) to synchronize the cell population. The synchronized cells were washed three times with serum-free medium to release the block and recultured at 37°C for 6 h in ␣-minimal essential medium with thymidine (2.5 g/ml). Cells were harvested, and slides were prepared by using standard procedures including hypotonic treatment, fixation, and air drying.
The Hrad1B complete ORF was gel purified and biotinylated with dATP using the Life Technologies, Inc. BioNick labeling kit (15°C, 1 h) (31). The procedure for FISH detection was performed as described previously (31,32). Briefly, slides were baked at 55°C for 1 h. After RNase treatment, the slides were denatured in 70% formamide in 2 ϫ SSC for 2 min at 70°C followed by dehydration with ethanol. Probes were denatured at 75°C for 5 min in a hybridization mix consisting of 50% formamide and 10% dextran sulfate. Probes were loaded on the denatured chromosomal slides. After overnight hybridization, slides were washed, and the signal was detected. FISH signals and the 4Ј,6diamidino-2-phenylindole (DAPI)-banding pattern were recorded separately by taking photographs, and the assignment of the FISH mapping data with chromosomal bands was achieved by superimposing FISH signals with DAPI-banded chromosomes (33).

Identification of Human and Mouse
Homologues of S. pombe rad1 ϩ -A human EST cDNA clone (accession number AA029300) was identified in the EMBL DNA data base using the TBLASTN homology searching program (26) with the S. pombe rad1 ϩ -derived amino acid sequence as the query. This clone was obtained from the I.M.A.G.E. Consortium (EST 470124), and DNA sequence analysis of the 1.2-kb insert (Figs. 1 and 2A) revealed an ORF that was highly similar to S. pombe Rad1. The predicted amino acid sequence of this ORF, however, was significantly shorter than the S. pombe Rad1 protein. To ensure that we had identified the complete sequence of the putative Hrad1 cDNA, we carried out 3Ј and 5Ј RACE-PCR. The 3Ј RACE clones that we obtained were all identical to the original EST cDNA sequence. The 5Ј RACE clones that we obtained separated into two classes. The first was identical to our original cDNA clone, and the second contained a 119nucleotide insertion presumably originating from an alternatively spliced mRNA. This additional sequence changed the reading frame immediately upstream of the initiation codon in our original cDNA sequence and resulted in a second and longer ORF (Figs. 1 and 2A). We have named the long and short ORFs Hrad1A and Hrad1B, respectively, and their sequences have been deposited into GenBank TM under accession codes AJ004974 and AJ004975, respectively). Hrad1A has 60% identity to the S. pombe rad1 ϩ sequence at the DNA level and 49% identity and 72% similarity at the amino acid level (Fig. 2B). The Hrad1A cDNA encodes a protein with a predicted molecular mass of approximately 31.5 kDa, and the Hrad1B protein has a predicted molecular mass of approximately 19.5 kDa.
A subsequent search of the data bases resulted in the identification of several mouse ESTs homologous to the Hrad1 cDNA. Three clones were obtained from the I.M.A.G.E. Consortium (accession codes AA387891, AA387463, and AA386981) and sequenced, resulting in the identification of a single ORF that encodes a protein with a predicted molecular mass of approximately 31.5 kDa (data not shown) (the nucleotide sequence was deposited into GenBank TM under accession code AJ004976). Mouse and human Rad1A are 91% identical at the amino acid level. None of the mouse clones sequenced corresponded to Hrad1B.
Northern Blot Analysis of Human and Mouse rad1-The transcript profiles of human and mouse rad1 were examined by probing several multiple tissue Northern blots (CLONTECH), a cancer cell line Northern blot (CLONTECH), and a mouse developmental Northern blot (CLONTECH). Three transcripts of approximately 5, 3, and 1.3 kb were identified for Hrad1, and all were elevated in the cancer cell lines (Fig. 3A). All three transcripts were present in differentiated tissues with the highest levels in testis, skeletal muscle, placenta, and heart. There was a specific increase in the shortest transcript in testis. The alternative splice we identified represents a difference of 119 nucleotides, which is too small to account by itself for the differences between the three transcripts.
To discriminate between the transcripts we probed the blot containing the testis sample with an oligonucleotide (OAP103) derived from the alternatively spliced region. This hybridized specifically to the 3-kb transcript indicating that the 1.3-kb transcript, which is highly elevated in testis does not contain the alternatively spliced region and therefore does not correspond to Hrad1A (Fig. 3B). Further searches of the EST data bases have revealed a possible alternative 3Ј-untranslated region (accession number AA464502), which may explain at least one of the other transcripts (data not shown).
A single transcript of approximately 2.2 kb was identified for mouse rad1 and was expressed at comparable levels in all tissues and at all stages of development tested (Fig. 3C). We have not established whether mouse rad1 is subject to alternative splicing as demonstrated for Hrad1. Our Northern blot experiment does not provide the resolution to display two transcripts differing by a 119-nucleotide alternative splice.
Complementation of S. pombe rad1-Complementation has often been used to demonstrate biological activity for mammalian homologues of yeast proteins (34,35). We examined whether Hrad1 could complement the UV irradiation (DNA damage-dependent checkpoint) or hydroxyurea (DNA replication-dependent checkpoint) sensitivity phenotypes of a S. pombe rad1::ura4 ϩ strain. Hrad1A and Hrad1B were cloned into the S. pombe expression vector pREP3x (28) and transformed into wild-type and rad1::ura4 ϩ cells. Transformants were exposed to varying doses of UV or transiently exposed to 10 mM hydroxyurea as described previously (16). We observed no complementation of the UV or hydroxyurea sensitivity phenotypes (data not shown).
Hrad1 Transcription Is Not DNA Damage-inducible-Many genes involved in the response to DNA damage are induced by exposure to DNA damaging agents. To examine whether this was the case for Hrad1 we exposed human HaCaT cells to 30 J/m 2 of UV radiation, a dose sufficient to cause DNA damage and invoke a cell cycle arrest (36). We prepared RNA at various time points up to 8 h after exposure and then determined the effect upon Hrad1 transcript level using an RNase protection assay. The level of the Hrad1 transcript remained constant throughout the time course (data not shown). Consequently, we conclude that Hrad1 transcription is not increased in response to UV-induced DNA damage.
Affinity Purified HisHrad1A Exhibits Exonuclease Activity-The sequence of Hrad1 is highly similar to that of REC1 from U. maydis, and purified hexahistidine-tagged Rec1 has been shown to have exonuclease activity (21). To test whether Hrad1 shared this biochemical activity, both Hrad1A and Hrad1B were expressed in E. coli as N-terminal hexahistidine-tagged fusion proteins using the pRSETB inducible expression vector (Invitrogen). The proteins designated HisHrad1A and HisHrad1B were purified by immobilized metal affinity chromatography (Qiagen). Bacterially expressed HisHrad1A was found predominantly in the insoluble fraction; however, approximately 20% of the protein as judged by Coomassie staining of SDS-polyacrylamide gels was present in the soluble fraction, allowing purification under nondenaturing conditions. HisHrad1B was found exclusively in the insoluble fraction and was isolated from inclusion bodies under denaturing conditions and refolded by rapid dilution. HisHrad1A and HisHrad1B have predicted molecular masses of approximately 35.5 and 23.5 kDa, respectively. The electrophoretic mobility of the purified proteins on SDS-polyacrylamide gels was in agreement with these predictions (Fig. 4).
First we assayed purified HisHrad1A and HisHrad1B for the ability to release labeled 5Ј or 3Ј terminal nucleotides from linear double-stranded DNA. HisHrad1A exhibited clear nucleolytic activity (Fig. 5, A and B), releasing radionucleotide from both the 5Ј and the 3Ј termini with a 6-fold preference for the 3Ј end. HisHrad1B did not show any nucleolytic activity. This may be a consequence of the different purification regimes or may reflect a true difference in biological activity.
The conditions for optimal activity were examined and found to be similar to those observed for REC1 (21). Maximum activity was obtained in a low ionic strength buffer at pH 7.4 -9.0 and containing Mg 2ϩ (Table I). Activity was unaffected by ATP or dATP but was significantly reduced by 100 mM NaCl. The replacement of Mg 2ϩ with Zn 2ϩ or the addition of 10 mM EDTA resulted in complete loss of activity (Table I).
The substrate specificities were also examined in greater detail. HisHrad1A has an approximately 6-fold preference for 3Ј termini under these assay conditions and is approximately 2-fold more active when single-stranded DNA is provided as substrate (Fig. 5B). The activity associated with HisHrad1A differs from known Mg 2ϩ -dependent exonucleases (Table II), strongly suggesting that the activity we observed is inherent in the Rad1A gene product.
Chromosomal Localization of Hrad1-The chromosomal position of Hrad1as determined to establish whether a loss of heterozygosity associated with Hrad1 might be linked with any known disease. The 1.2-kb cDNA corresponding to Hrad1B (EST 470124) was used as a probe for FISH analysis. Under the conditions used, the hybridization efficiency was approximately 69% for the probe (among 100 checked mitotic figures, 69 showed signals on one pair of chromosomes). The DAPI banding pattern was used to establish that Hrad1 localizes to the short arm of chromosome 5. No additional locus was picked up by FISH detection under the conditions used. The detailed   position was further determined based upon the analysis of 10 photographs leading to the conclusion that Hrad1 is located on human chromosome 5p13.3-13.2 (Fig. 6). Loss of heterozygosity of this region of chromosome 5 has been linked to a variety of human neoplasias including lung cancer (37,38).

DISCUSSION
In S. pombe, cell cycle checkpoint arrest in response to DNA damage or inhibition of replication is dependent on multiple proteins. Six gene products have been identified that act early in the process of checkpoint control, and to date human homologues of two of these proteins have been identified, Hrad9 (9) and ATR (7,8). Based upon sequence homology and biochemical activity, we have identified a third component, Hrad1, a human homologue of he S. pombe rad1 ϩ cell cycle checkpoint gene. Hrad1 is highly similar to S. pombe rad1 ϩ , S. cerevisiae RAD17, and the REC1 gene of U. maydis. Mutations in these genes lead to similar cell cycle checkpoint defects and other phenotypes associated with alterations in DNA repair and recombination (16,19,20).
We have shown that Hrad1 is subject to alternative splicing giving rise to two ORFs, a long form, designated Hrad1A, and an N-terminal truncation, designated Hrad1B. The Hrad1 gene has at least three transcripts of 5, 3, and 1.3 kb present in all tissues that we examined and that are increased in all the cancer cell lines examined, suggesting either that transcription of Hrad1 is proliferation-dependent or that it is increased in response to the genomic instability of cancer cell lines. In testis we observed an increase specifically in the 1.3-kb transcript. We have demonstrated that this transcript does not correspond to Hrad1A, suggesting that it corresponds to Hrad1B or an as yet unidentified third form of Hrad1. Several yeast cell cycle checkpoint genes play important roles in meiosis (39) and recently ATM and ATR (the human homologues of S. pombe rad3) were shown to be highly expressed in testis where they interact with meiotic chromosomes. This suggests a direct role for these proteins in recognizing and responding to DNA strand interruptions that occur during meiotic recombination (40). Hrad1 could form part of this recognition complex in association with ATM or ATR.
When expressed in an S. pombe rad1::ura4 ϩ strain, both Hrad1A and Hrad1B failed to complement the phenotypes associated with loss of checkpoint function. However, this should not be taken as evidence against functional homology because failure to complement has been shown for other checkpoint genes such as ATR and Hchk1 (7,12). 2 The Hrad1 amino acid sequence is significantly similar to the U. maydis REC1 protein, indicating that Hrad1 may have a 3Ј 3 5Ј exonuclease activity. We have confirmed the predicted 3Ј 3 5Ј exonuclease activity for Hrad1A by demonstrating that Hrad1A expressed as a hexahistidine fusion protein and puri-fied from E. coli is an exonuclease with a 6-fold preference for 3Ј termini. The biochemical properties of HisHrad1A were very similar to REC1 (21). The truncated Hrad1B, expressed as a hexahistidine fusion protein, lacks detectable exonuclease activity. This may reflect the in vivo biological activity or may be a consequence of the purification regimes. To fully address this question, purification of endogenous Hrad1B from mammalian 2 A. Parker, unpublished results.  cells or from a heterologous expression system that yields native protein will be required.
An analysis of C-terminal truncations of the Rec1 protein has established that the C-terminal portion of the Rec1 protein is not essential for exonuclease activity but is crucial for G 2 -M checkpoint function (24,25). The checkpoint-compromised U. maydis rec1-1 mutant results from a C-terminal truncation lacking 122 amino acids (24), including a block of sequence containing four periodically spaced leucines that is conserved in Hrad1 and S. pombe Rad1 but not in S. cerevisiae RAD17 (24). This may represent a domain important in interactions with other checkpoint components. In addition, the rec1-5 mutant that lacks exonuclease activity has a 100-fold higher rate of spontaneous mutation, indicating that the exonuclease function may be required for mismatch repair (24). Thus, the Rec1 protein appears bifunctional, having an N-terminal exonuclease domain and a C-terminal checkpoint domain. This type of separation of enzymatic and checkpoint domains has also been described for S. cerevisiae DNA polymerase ⑀ (41). The bifunctional nature is likely to be conserved in Hrad1A based upon our biochemical studies and the protein sequence similarity between Hrad1A and the U. maydis Rec1 protein. However, in humans we have identified an additional component, Hrad1B, which corresponds to the C-terminal region of Rec1 implicated solely in checkpoint control. We cannot discount the possibility that the alternatively spliced mRNA coding for Hrad1B is not translated. However, if the Hrad1B mRNA leads to the production of a protein, then what role might it play in DNA metabolism or checkpoint control? In fission yeast, rad1 is required for both the DNA damage-and DNA replication-dependent checkpoints, whereas in S. cerevisiae RAD17 is only required for the DNA damage-dependent checkpoint. It has been demonstrated that RAD17 functions in conjunction with RAD24 and MEC1 to activate DNA degradation (23), leading to the suggestion that there is a requirement to process single-or double-stranded breaks such that single-stranded DNA is exposed to activate the checkpoint (22,23). If we consider Hrad1 as part of a complex that acts as a surveillance mechanism monitoring the genome and communicating the presence of DNA damage or unreplicated DNA to the cell cycle machinery, then it is conceivable that in higher eukaryotes two types of surveillance complexes exist separating the DNA damage-and DNA replicationdependent checkpoint pathways. Hrad1A could be specifically required for the DNA damage-dependent checkpoint, able to monitor the genome via the C-terminal checkpoint domain and process single-and double-stranded breaks via the N-terminal exonuclease domain. A link between the mismatch repair system and the DNA damage-dependent checkpoint has been demonstrated (42). Hrad1B may have a function in the DNA replicationdependent checkpoint that might require only the monitoring activity and not the nuclease function to identify replication forks. Hrad1B also contains the sequence shown to be essential for checkpoint function in a mutational analysis of S. pombe Rad1 (43). The increase in transcription that we observed in testis could also signify a role for Hrad1B in meiotic recombination in conjunction with ATR and ATM.
The loss of checkpoint function has been shown to lead to genomic instability even in the absence of exogenous DNA damage (44). In humans, the p53 gene and the ATM gene are not only required for the G 1 -S phase checkpoint (45,46) but also act as tumor suppressors (47)(48)(49)(50). It is likely that other checkpoint genes will act as tumor suppressors. The Hrad1 gene is located on chromosome 5p13.2-13.3. Loss of heterozygosity of this region of chromosome 5 has been associated with a tumor suppressor function most notably in human lung cancer (37,38). The CDK2/cyclin A-associated protein p45 (SKP2) has been mapped to 5p13 (51), but Hrad1 should also be con-sidered as a candidate tumor suppressor gene in this region.
The precise roles that Hrad1A and Hrad1B play in cell cycle control, DNA repair, and tumor suppression remain to be determined. This study represents a starting point to begin to unravel the potential multiple roles for the Hrad1 gene product.