Characterization of the Human Rad51 Genomic Locus and Examination of Tumors with 15q14–15 Loss of Heterozygosity (LOH)¹

Genomic instability is a common denominator in the vast majority of human cancers (1). The high number of genetic alterations found in tumor cells has led to the suggestion that an increased mutation rate (mutator phenotype), likely in combination with selection and clonal expansion, provides the requisite for tumorigenesis (2, 3). While up to 30% of colorectal tumors display instability in the length of repeated sequences (MSI;⁴ Ref. 4), more than 70% of colorectal tumors exhibit gene amplifications and gross chromosomal changes that include chromosome translocations and alteration in chromosome numbers (aneuploidy; Ref. 5). Most MSI tumors harbor alteration(s) in the DNA mismatch repair genes (hMSH2, hMLH1, hPMS2, and hMSH6) that inactivate a pathway for repair of single-nucleotide and small insertion/deletion loop-type mismatches. In addition to MSI, inactivation of the mismatch repair machinery leads to an increase in spontaneous mutation rates, as would be predicted for a mutator phenotype (for recent reviews, see Refs. 6 and 7). A genetic basis for the 70% of colorectal tumors that exhibit gene amplifications and gross chromosomal changes has been suggested in experiments that demonstrate the persistence of the chromosomal aneuploidy phenotype (chromosomal instability) in colorectal cell lines (8). These studies support an aneuploidy mutator phenotype that is induced by as yet unidentified mechanisms.

Defects in recombination/repair pathways that process DNA damage, such as double-strand breaks, have been proposed to contribute to the observed genomic instability phenotype in some human cancers (1). Xeroderma pigmentosum, ataxia telangiectasia, Bloom’s syndrome, Fanconi’s anemia, and Nijmegen breakage syndrome are hereditary disorders in which chromosome breaks are known to contribute to an overall susceptibility to cancer (9, 10). Moreover, both the Bloom’s syndrome and Nijmegen breakage syndrome gene products are related to bacterial or yeast homologues known to participate in the repair of broken chromosomes (11, 12).

The hRAD51 protein has been suggested to play a role as a tumor suppressor gene for several reasons: (a) hRAD51 is a member of the RecA/RAD51 family of proteins, which have been implicated in DNA recombination/repair processes (13); (b) the hRAD51 protein appears to associate either directly or indirectly with p53 (14), BRCA1 (15), and BRCA2 (16), three other known tumor suppressors; (c) RAD51 has been shown to function in Saccharomyces cerevisiae mutation avoidance (17), and the frequency and spectrum of a rfa-1 mutant (the large subunit of the consensus cellular single-stranded binding protein) are enhanced by a rad51 null mutation (18); and (d) the distribution of hRAD51 appears to be altered in response to DNA damage as well as along paired meiotic chromosomes in spermatocytes derived from ATM^−/− mice (15, 19). Moreover, c-abl, an oncogene and downstream effector in the ATM DNA damage response pathway (20, 21), has been shown to phosphorylate hRAD51 and alter its ability to bind DNA in vitro (22). These results have led to the proposal that hRAD51 dysfunction in somatic cells might also contribute to genomic instability and cancer.

While both Escherichia coli RecA and the S. cerevisiae RAD51 proteins are not required for viability, mice containing a homozygous disruption of the Rad51 gene display early embryonic lethality (23), and attempts to generate null rad51 cell lines have failed (24). These findings would appear to limit a possible role for hRad51 as a tumor suppressor gene. Interestingly, mammalian cells contain at least eight additional RAD51 homologues that may provide some redundancy in function or functional interaction. For example, it has been shown that XRCC3, one of these RAD51 homologues, interacts with hRAD51, and XRCC3 mutant cells display radiation sensitivity and increased chromosome aberrations (25, 26).

Here, we have characterized the hRad51 genomic locus by defining the intron/exon boundaries and genetically mapping it relative to markers used previously in LOH studies of human tumors. We found that hRad51 was located in the center of a region at chromosome 15q14–15, which is frequently deleted in breast tumors and metastatic brain tumors (27). We examined a cohort of these tumors that displays LOH at 15q14–15 for hRad51 mutations and promoter methylation. While we found no alterations of the hRAD51 locus in the DNA derived from any of these human tumors, this study provides the tools necessary to further understand the regulation of hRAD51 expression and to study its potential role in tumorigenesis.

Oligonucleotide primers specific for hRad51 (R11, 5′-AGTGCTGCAGCCTAATGAGAGT-3′; and R12, 5′-TGTGACTACTGACCTGTCTCCT-3′) were used to screen a human BAC library (Version IV; Research Genetics, Huntsville, AL) by PCR. DNA from positive clones was isolated using the Qiagen (Chatsworth, CA) DNA purification kit. The DNA was sequenced on an Applied Biosystems automated sequencer, and intron-exon boundaries were determined by comparison with the cloned cDNA sequence as well as consensus splice site junctions. Intron sizes were confirmed by PCR followed by agarose gel electrophoresis. An additional 1.5 kb of 5′-untranslated sequence were sequenced and analyzed for potential transcription factor binding sites as well as the percentage of G+C bp and CpG sites.

The location of hRad51 was determined using primers R11 and R12 to screen the Genebridge-4 radiation hybrid panel (Version RH02.02; Research Genetics; Huntsville, AL). PCR products were visualized by agarose gel electrophoresis, and data were submitted to the Whitehead Institute/MIT Center for Genome Research for interpretation. Using the same method, we determined the genomic position of marker GAAA1C11 (also known as D15S1232) using forward primer R33 (5′-CCAGAGAGATCTTTCCCCAT-3′) and reverse primer R34 (5′-TTGCTCCACTGTTTTCTCAG-3′), and we determined the genomic position of marker D15S641 using forward primer R31 (5′-AACAAAGGGAGACCCTCATC-3′) and reverse primer R32 (5′-GACACCCCAGTAGCAATGAG-3′). Information on the original characterization of these markers was obtained from the Cooperative Human Linkage Center.⁵

Native tumor specimens and corresponding normal blood were obtained from patients treated at the University Hospital-Bonn between 1990 and 1998 during surgery or at autopsy. Tissues were collected without bias for age, sex, or ethnic background. All tumors were classified according to the WHO guidelines. The tumor specimens were examined microscopically before phenol DNA extraction to exclude contamination by nontumorous tissue. DNA was isolated as described previously (27). Twenty brain metastases of tumors from various tissue origins and their normal counterparts were analyzed for mutations in the coding region of the hRad51 gene. All tumors examined displayed LOH at 15q14–15 (Ref. 27; Table 2). A second set of 41 breast carcinomas (21 of which displayed LOH at 15q14–15) was examined for hRad51 mutations (Table 2).

Details of the primer sets used to amplify the hRad51 exons are shown in Table 1. Approximately 20 ng of genomic DNA were used as template in 20-μl reactions that contained 1.0 unit of AmpliTaq Gold (Perkin-Elmer); 200 μm each of dATP, dGTP, and dTTP; 40 μm dCTP; and 1.0 μCi of [α-³²P]dCTP. Cycles were as follows: 95°C for 12 min, followed by 30 cycles of 94°C for 30 s; 55.5°C for 30 s; 72°C for 45 s; followed by 5 min at 72°C. The reactions were heated at 94°C for 5 min and placed immediately on ice until loading. Samples were run at room temperature on 0.5× MDE gels (FMC BioProducts). Exons to be sequenced were amplified in a similar fashion, except that 200 μm dCTP was used. The PCR products were then isolated using agarose gel electrophoresis and the Qiagen Gel Extraction kit, and approximately 5 ng of PCR products were sequenced with an Applied Biosystems Automated Sequencer.

Methylation of the predicted CpG island of the hRad51 promoter was performed using the bisulfite method as originally described by Clark et al. (28). In brief, approximately 50 ng of DNA were digested with ApaI and then incubated with 2.5 m sodium metabisulfite (Amresco) and 100 mm hydroquinone (Sigma) at pH 5.0 in a total volume of 240 ml at 55°C for 16 h. Subsequently, the DNA was desalted, treated with 0.3 m NaOH at 37°C for 15 min, and precipitated. Primers were designed such that they flanked the region of highest CpG density as well as several Sp1 sites within the putative hRad51 promoter upstream of exon 1 (see Fig. 3, ; 5′-TGAGGGATTGGGGTAGGAGTA-3′, 5′-CCAACCTTCTACACACAAC-CCAA-3′, 5′-GTAAAAAGGGAAGAGGGTAGTTTG-3′, 5′-ACAACCC-AAATAAATTACAATTCCCAACT-3′). PCR products (274 bp) were synthesized by a nested PCR approach and sequenced directly. DNA treated with M.SssI, which methylates CpG sites, were used as a positive control. There did not appear to be any differences in the amplification of methylated versus unmethylated substrate DNA. We estimate that 20% of methylated CpGs in the substrate DNA at most sites could be detected.

The hRAD51 gene was mapped previously by fluorescence in situ hybridization to chromosome 15q15.1 (29). This genomic locus is adjacent to a region (15q14) that has been reported to be frequently lost in brain metastases of various tumors (27). To test whether hRad51 is involved in the development of these tumors, we first established the genomic location of the hRad51 gene and many of the markers used in the tumor study relative to markers contained in a radiation hybrid panel (27). Radiation hybrid mapping revealed that hRad51 is located between the most commonly deleted markers in these tumors, 26.7 cR telomeric from marker GAAA1C11 and 20 cR centromeric from marker D15S641 at 15q14–15 (Fig. 1). We also determined the location of nine polymorphic markers used to define the region of LOH on chromosome 15q14–15 in 21 metastatic brain tumors and 21 primary breast carcinomas. These data also indicate that the hRad51 genomic locus is slightly more centromeric than reported previously but is clearly within the common region of LOH in both the brain and breast tumors.

We determined the genomic structure of the hRad51 gene to screen tumors for possible hRad51 mutations. A BAC clone (21B18) containing the hRad51 gene was identified by PCR screening of a human BAC DNA library (Research Genetics) using primers specific for hRad51. DNA from clone 21B18 was isolated and sequenced. Intron sizes were determined by long-range PCR (Boehringer Mannheim). PCR products could not be obtained for introns 5 and 6, suggesting intron sizes of >10 kb. The hRad51 gene consists of 10 exons and spans at least 30 kb (Fig. 2,A). All exon-intron boundaries follow the GT-AC rule. The translation start codon is located in exon 2 (Fig. 2,B, underlined), and the average size of the coding exons is 112 bp. Further sequencing of the region 5′ of the first exon revealed that noncoding exon 1 contained a CpG island that was approximately 990 bp in size (Fig. 3). This putative promoter region contains several Sp1 recognition sites but lacks a TATA box. Further characterization of that region is in progress and will be published elsewhere. Sequences were submitted to GenBank (GenBank accession numbers AF165088, AF165089, AF165090, AF165091, AF165092, AF165093, and AF165094).

All coding exons from 21 metastatic brain tumors and 42 breast carcinomas (including 21 that clearly displayed LOH at chromosome 15q14–15) were analyzed by the single-strand conformational polymorphism method using the primer pairs listed in Table 1. In addition, complete coding exons of 10 brain metastases were sequenced. No alterations in the coding sequence of hRAD51 were detected (data not shown). These results suggest that nucleotide mutations of hRAD51 that might lead to altered function of the hRAD51 protein were unlikely to be present in the cohorts of tumors that displayed LOH surrounding the hRAD51 genetic locus.

We detected a large CpG island imbedded within the putative hRad51 promoter region. CpG methylation may reduce or abolish expression of the hRad51 gene, as has been shown for several tumor suppressor genes and repair genes including hMLH1 (30) and the O⁶-methylguanine methyltransferase gene (31). To rule out the possibility of both normal allele contamination and biallelic tumors, we examined promoter methylation in a region including the highest CpG density and several Sp1 binding sites in samples that clearly displayed LOH covering the hRad51 genomic locus (Table 2; see dot samples). We found no evidence of promoter methylation in any of these samples in the region examined (data not shown), suggesting that hRad51 expression is not impaired in tumors that display LOH at chromosome 15q14–15. These experiments must be tempered by the results of the bisulfite methylation detection method (28), in which there appeared to be a somewhat higher background between nucleotide −3693 and −3717).

In this study we have: (a) characterized the structure of the hRad51 genomic locus, which provided the tools necessary to specifically amplify hRad51 coding regions from genomic DNA; and (b) tested the possibility that hRad51 could function as a tumor suppressor gene in metastatic brain tumors and breast carcinomas that show high frequencies of LOH at chromosome 15q14–15.

Sequencing of the hRad51 genomic locus revealed 10 exons, and a large (approximately 1 kb) CpG island covers the putative promoter region, which also includes the first (noncoding) exon. This CpG island appears TATA-less and is similar to typical housekeeping promoters (32). The presence of several putative Sp1 promoter binding sites is consistent with the observed cell cycle-dependent expression of hRad51 (33). Previous reports have shown that both protein and mRNA levels rise at the G₁-S-phase boundary and remain elevated through the G₂-M phase (34). This G₁-S-phase pattern of expression is similar to that of yeast Rad51 (35), possibly reflecting an S-phase-specific role other than recombination for RAD51 (36). Perhaps the most interesting difference between mammalian RAD51 and its yeast and bacterial family members is that its expression is not induced by DNA-damaging agents (37). Both the E. coli RecA and yeast RAD51 proteins are induced over 1000-fold during the SOS and RAD9 damage responses, respectively (38). While the mammalian p53 damage checkpoint/response has been compared to these pathways, we did not find p53 consensus transcription-activation sites in the hRad51 promoter. Furthermore, recent reports suggest a p53/p21-dependent decrease in hRad51 mRNA in response to ionizing irradiation (39). This distinction may classify hRad51 more closely to genes such as PCNA, whose mRNA levels increase at the G₁-S-phase boundary but are down-regulated by p53 (40).

While hRAD51 mRNA levels are reduced in response to irradiation, proteins levels appear to persist (37), and hRAD51 becomes part of distinct nuclear foci (41). Other distinct nuclear foci of hRAD51 have been identified during S phase (36) and in B cells induced to undergo class switching (42). Thus, hRad51 appears to be regulated in at least two ways: (a) transcriptionally, by genes that confer a proliferative potential, as well as by checkpoint signaling pathways that regulate DNA damage responses; and (b) at the protein level, where interactions with other molecules leads to distinct cellular localization in hRAD51 nuclear foci. It is possible that hRAD51 regulation may occur at the transcriptional level in a cell type-, cell cycle-, or damage response-coordinated manner. Additional studies will be required to determine the context of the hRad51 promoter and understand its regulation at the molecular level.

We have determined the chromosomal location of hRad51 relative to markers used in LOH analysis of human tumors (27). hRad51 was found to be slightly more centromeric than reported previously (15q14 bordering 15q15) and central to a commonly deleted region in brain metastases from several tumor types and breast carcinomas. However, we did not uncover any mutations in the hRad51 coding region in a cohort of 42 tumors that displayed LOH in the chromosome 15q14–15 region. In addition, we found no indication that expression of hRad51 might be down-regulated by hypermethylation of the CpG island in the putative promoter region. It is therefore likely that hRad51 is not the candidate tumor suppressor gene in these tumors. These results support the notion that hRAD51 may be an essential gene, a redundant gene, or is dispensable and/or independent of the BRCA1/BRCA2 tumor suppressor pathway(s).

We thank Hans-Jürg Alder and the Kimmel Nucleic Acids Facility for oligonucleotide synthesis and sequencing.