Mouse RAD50 has limited epitopic homology to p53 and is expressed in the adult myocardium.

Previous studies have identified a 180-kDa mouse cardiomyocyte phosphoprotein with limited epitopic homology to p53. In this study, the protein was purified and partially sequenced. Oligonucleotide probes based on the available amino acid sequence data were used to isolate cDNA clones. Sequence analyses revealed that the clones encoded a protein with regional homology to the yeast RAD50 gene product. Expression of the mouse cDNA rescued the methyl methanesulfonate-sensitive phenotype in rad50 mutant yeast, indicating that the cardiomyocyte phosphoprotein is the mammalian homologue of the yeast RAD50 gene product. Fluorescence in situ hybridization analyses localized the mouse RAD50 gene to the A5-B1 region of chromosome 11. Northern blot analyses demonstrated a complex pattern of RAD50 expression during mouse development which was further complicated by the presence of several alternatively spliced transcripts. High levels of RAD50 expression was evident in the adult myocardium, a somewhat surprising observation given the absence of DNA synthesis in adult cardiomyocytes.

Adult cardiac myocytes are terminally differentiated cells, which exhibit little if any capacity to re-enter the cell cycle (1). In previous studies aimed at identifying potential therapeutic targets to promote cardiomyocyte proliferation, transgenic mice that express the SV40 large T antigen (T-Ag) 1 oncoprotein in the ventricles and/or atria were generated (2,3). These studies were based on the paradigm that the transforming activity of DNA tumor virus oncoproteins (as exemplified by T-Ag) resides largely in their ability to bind to, and consequently modify the activity of, endogenous cell cycle regulatory proteins (reviewed in Refs. 4 and 5). T-Ag expression in the hearts of transgenic mice led to the development of myocardial tumors composed of differentiated, proliferating muscle cells. Analysis of cell lines derived from the transgenic tumors revealed that the p53 and p107 tumor suppressors were prominent T-Ag-binding proteins (6). Two additional novel proteins of 193 and 380 kDa were also observed to bind, either indirectly or directly, to T-Ag (6). 2 It has become increasingly apparent that genomic instability plays an important role in tumor progression. This notion originally stemmed from the association observed between defects in DNA excision repair and the propensity for skin cancer in xeroderma pigmentosum patients (7). More recent genetic studies have established that inheritance of a mutation in the mismatch repair gene hMSH2 predisposes to hereditary nonpolyposis colon cancer (8 -10), thus providing a direct genetic link between genomic instability and tumor progression (11). Several lines of evidence have suggested a potential role for the p53 tumor suppressor gene product in the maintenance of genomic stability. For example, gene amplification was readily induced in fibroblasts cultured from patients with Li-Fraumeni syndrome or from p53-deficient mice, but not in cells with functional p53 (12,13). p53 has also been shown to preferentially bind to double-stranded DNA with insertion/deletion mismatches (14), although the functional consequences of this binding activity were not determined. A direct role for p53 in nucleotide excision repair has recently been identified. p53 can bind to and modify the activity of XPD and XPB, proteins that are required for TFIIH-associated nucleotide excision repair (15). p53 may also exert an indirect role in regulating nucleotide excision repair, as it is required for transcriptional activation of Gadd45 in response to ionizing radiation (16). In vitro studies have established that Gadd45 stimulates excision repair (17).
In addition to the T-Ag-binding proteins described above, a 180-kDa phosphoprotein that cross-reacted with anti-p53 monoclonal antibody mAb421 was also identified in the transgenic cardiomyocytes (6). Interest in this protein was piqued by the observation that its steady state levels were dramatically influenced by the confluence of the cell cultures. Limited epitopic homology similar to that observed between the 180-kDa cardiomyocyte phosphoprotein and p53 can sometimes reflect functional homology. This type of relationship is best exemplified by T-Ag and p68. p68 was originally identified as a protein that cross-reacted with anti-T-Ag antibody RA3-2C2 (18). Subsequent cloning analyses revealed that p68 was an RNA helicase (19). T-Ag is also known to exhibit RNA helicase activity, suggesting that the RA3-2C2 antibody detects a functionally conserved epitope in both T-Ag and p68. Based on these collective observations, we sought to further characterize the 180-kDa phosphoprotein present in transgenic cardiomyocytes. In this report, the phosphoprotein was isolated by preparative immune precipitation, partially sequenced, and cloned. Analysis of the cDNA clones indicated that the phosphoprotein is the mammalian homologue of the yeast RAD50 gene. RAD50 is a member of the RAD52 epistasis group, and is required for the repair of double-stranded DNA breaks as well as for synaptonemal complex formation in sporulating cultures. Complementation analyses indicated that the mouse RAD50 protein can rescue the DNA repair deficiency in rad50 mutant yeast. Prominent RAD50 expression was detected in the adult testes, myocardium, and lung. The high levels of cardiac RAD50 expression were surprising, given the post-mitotic nature of adult cardiomyocytes.
Protein Sequence Analysis-Gels of the mAb421 immune complex were electroblotted onto Millipore Immobilon-PSQ PVDF membranes (1 h, 250 mA constant current) as described (21). The membrane was stained with 0.1% Coomassie Blue R-250 in 50% methanol for 0.5 min, and then briefly destained with 10% acetic acid in 50% methanol to visualize the bands. The membrane was thoroughly washed with water and allowed to dry. Regions of the membrane containing proteins of interest were excised, wetted with 1 l of methanol, and reduced and alkylated with isopropylacetamide as described (22). The samples were then digested in 25 l of 0.1 M ammonium bicarbonate, 10% acetonitrile with 0.2 g of Lys-C (Wako Bioproducts, Richmond, VA) or 0.2 g of modified trypsin (Promega, Madison, WI) at 37°C for 17 h. The digested material was concentrated to 20 l on a Speed-Vac and directly injected onto a C18 0.32 ϫ 100-mm capillary column (LC Packing Inc., Amsterdam). The HPLC apparatus consisted of an Applied Biosystems (Palo Alto, CA) model 140A microgradient pump system, and a model 783 UV detector equipped with a Z-shaped flow cell (LC Packings, Inc.). The pump was operated at 50 l/min and connected to a Valco tee, which was used as a splitter to reduce the flow to 4 l/min (23). Solvent A was 0.1% aqueous trifluoroacetic acid, and solvent B was acetonitrile containing 0.07% trifluoroacetic acid. Peptides were eluted using a linear gradient of 0 -80% solvent B over 120 min and detected at 195 nm. Aliquots (0.2 l) of the capillary HPLC-purified peptides were mixed on the sample probe tip with 0.2 l of a saturated solution of with ␣-cyano-4-hydroxycinnamic acid (saturated in 50% acetonitrile, 2% trifluoroacetic acid). Mass spectra data were obtained with a LaserTec Research laser desorption linear time-of-flight mass spectrometer equipped with a 337-nm VSL-337 ND nitrogen laser (Laser Science Inc., Newton, MA). Automated protein sequencing was performed on a model 470A Applied Biosystems sequencer equipped with an on-line phenylthiohydantoin analyzer using modified cycles as described. Peaks were integrated with Justice Innovation software (Palo Alto, CA) using Nelson Analytical 760 interfaces. Sequence interpretation was performed on a DEC 5900 (24).
Isolation and Analysis of p180 cDNAs-Oligonucleotide probes (25) based on the peptide sequence data were used to screen an AT-2 cardiomyocyte cDNA library generated in GT11 (26). Plaque hybridizations, phage DNA isolation, and subcloning were performed using standard methodologies (27). Sequence was determined for both strands of all cDNA clones examined using the dideoxy chain terminating approach (Sequenase, U. S. Biochemical Corp.).
Reverse Transcriptase-PCR Amplifications-For first strand cDNA synthesis, reverse transcription was performed with 70 ng of total RNA and 100 pmol of random primers in a total volume of 10 l of 1 ϫ PCR (50 mM KCl, 10 mM Tris-HCl, pH 7.5, 1.5 mM MgCl 2 ), 1 mM each dNTP, 10 units of RNasin (Promega), and 10 units of avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim) at 42°C for 60 min. The volume was then adjusted to 50 l with 1 ϫ PCR containing 0.3 g of the sense and antisense oligonucleotide primers, and Taq polymerase (1 unit, Ampli-Taq, Perkin-Elmer). The mixture was overlaid with oil and amplified at 94°C (1 min) to 66°C (2 min) to 72°C (3 min) over 35 cycles. The amplification products were analyzed on agarose gels and visualized by UV epifluorescence following ethidium bromide staining.
Induction of DNA Damage in Mouse Fibroblasts-NIH-3T3 cells (ATCC) were plated at a density of 25,000 cells/cm 2 in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 g/ml penicillin, and 100 g/ml streptomycin. DNA damage was induced when the cultures reached near-confluence. To ascertain the sensitivity to MMS, drug was added at final concentrations ranging from 1 through 100 g/ml and the cells were cultured for an additional 48 h. Cells were then trypsinized and counted in quadruplicate using a hemocytometer, and the average cell count Ϯ standard error was determined. For DNA cross-linking experiments, 8-methoxypsoralen (MOPS) was added at final concentrations ranging from 0.002 through 20 g/ml 2 h prior to exposure to UV light. Irradiation was performed at 22°C using a bank of four T8 15-watt bulbs filtered to eliminate UVB radiation. The flux rate was measured at the media plane using a UVX-36 digital radiometer (UV Inc., San Gabriel CA) and was typically 1.2 milliwatts/cm 2 . UVA exposure was for 21 min for a total of 1.5 J/cm 2 . After irradiation, cultures were incubated for 48 h. Cells were then trypsinized and counted in quadruplicate using a hemocytometer. In both experiments, cells were harvested from replica dishes and processed for Northern blot analyses as described above.
Fluorescence in Situ Hybridization (FISH) Analyses-Lymphocytes isolated from adult mouse spleen were cultured at 37°C in RPMI 1640 medium containing 15% fetal calf serum, 3 g/ml concanavalin A, 10 g/ml lipopolysaccharide, and 5 M mercaptoethanol for 44 h. Bromode-oxyuridine was added to a final concentration of 0.18 mg/ml to induce synchronization, and the cells were cultured for an additional 14 h. The bromodeoxyuridine was removed, and the cells were cultured for an additional 4 h in ␣-minimal essential medium supplemented with thymidine (2.5 g/ml). To prepare chromosomes for FISH analysis, the cells were lysed by hypotonic treatment, fixed, and air-dried onto glass slides as described (32). A full-length mouse RAD50 cDNA was biotinylated using a BioNick labeling kit according to the manufacturer's protocol (Life Technologies, Inc.; see also Ref. 33). FISH analysis was then performed essentially as described (33,34). Briefly, the slides were baked at 55°C for 1 h, treated with RNase A, and the chromosomes denatured in 70% formamide, 2 ϫ SSC for 2 min at 70°C followed by dehydration with ethanol. The biotinylated RAD50 cDNA probe was denatured at 75°C for 5 min in 50% formamide, 10% dextran sulfate, and prehybridized to mouse C o t I DNA (Life Technologies, Inc.) for 15 min at 37°C. The chromosome preparations were then hybridized with the probe for 20 h at 37°C, washed, and visualized with an avidin-biotin (ABC) kit (Vector Laboratories, Burlingame, CA). The resulting FISH signal was superimposed upon the DAPI-banded chromosome pattern to ascertain the map position as described previously (34).

RESULTS
To further characterize the 180-kDa phosphoprotein present in transgenic cardiomyocytes, AT-1 tumor homogenates (35,36) were reacted with either mAb421 or with an anti-T-Ag polyclonal antibody. The immune complexes were harvested with protein A-Sepharose, displayed on a 7.5% SDS-polyacrylamide gel, transferred to a PVDF membrane, and visualized by Coomassie staining. As expected, a prominent band of 180 kDa was observed in the mAb421 immune complex but not in the anti-T-Ag immune complex (Fig. 1). Several additional T-Agbinding proteins identified previously in radiolabeled AT-2 cardiomyocytes (namely p193, p107, and p53; see Ref. 6) were also present in the AT-1 immune complexes. The region of the PVDF membrane containing the 180-kDa protein was excised and digested in situ with Lys-C or trypsin. The resulting fragments were then purified via capillary HPLC and their amino acid sequence determined. Although sequence data were obtained for 12 individual proteolytic fragments (which spanned more than 140 amino acid residues), no significant homologies with published protein sequences were detected in data base surveys. An oligonucleotide probe encoding the proteolytic fragment with the least degenerate amino acid sequence was therefore synthesized according to the rules set forth by Lathe (25).
Screens of an AT-2 cardiomyocyte cDNA library yielded a clone with a 764-bp insert encoding an open reading frame of 254 amino acids ( Fig. 2A, shaded box). Sequences corresponding to the proteolytic fragment encoded by the oligonucleotide probe, as well as those for three other proteolytic fragments from the 180-kDa protein, were present within the deduced amino acid sequence of the open reading frame ( Fig. 2A, vertical arrows). Subsequent screens of the AT-2 cardiomyocyte library ultimately yielded a set of overlapping clones, which spanned a total of 5088 bp (Fig. 2B). Sequence analysis of the cDNA clones identified an open reading frame of 3936 nucleotide residues (Fig. 3A). No other open reading frames in excess of 250 nucleotide residues were detected. A putative methionine initiation codon was identified at nucleotide residues 243-245. The sequences immediately flanking this methionine agreed well with the Kozak consensus sequence for eukaryotic initiation codons (ggtGCAAACATGT versus gccGCCACCATGG, respectively; see Ref. 37). The presence of an in-frame stop codon within the 5Ј flanking sequences indicated that the protein could not initiate prior to the methionine at nucleotide residues 243-245. The termination codon of the open reading frame was located at nucleotide residues 4179 -4181. An 895 nucleotide residue 3Ј-untranslated region and short poly(A) tail were also present. The open reading frame encoded a protein comprising 1312 amino acids and with a deduced molecular mass of 153 kDa (Fig. 3A). Amino acid sequences corresponding to all 12 proteolytic fragments obtained from the cardiomyocyte phosphoprotein were identified within the deduced amino acid sequence of the open reading frame (underlined amino acid res- idues, Fig. 3A). The discrepancy between the deduced molecular mass (153 kDa) versus the apparent molecular mass (approximately 180 kDa) most likely reflects carbohydrate moieties, as seven potential glycosylation sites were present in the primary amino acid sequence (Fig. 4). Data base surveys identified three regions of homology between the 153-kDa phosphoprotein and the yeast RAD50 gene product (38). These regions, located at amino acid residues 1-81, 129 -192, and 1117-1271 of the cardiomyocyte protein, exhibited 59%, 57%, and 65% sequence identity, respectively, with yeast RAD50 (Fig. 4).
Both proteins were 1312 amino acids in length, and the regions of homology were located at analogous positions. Based on this striking degree of homology, we have tentatively identified the 153-kDa cardiomyocyte phosphoprotein as the mouse RAD50 homologue.
The position of the mouse RAD50 gene was determined using FISH analyses. A full-length RAD50 cDNA probe was biotinylated and hybridized to mouse lymphocyte chromosome preparations. Under the conditions employed, a hybridization efficiency of 85% was obtained (i.e. 85 of 100 mitotic figures FIG. 3-continued examined exhibited paired FISH signal). Specific chromosomes were identified by virtue of their DAPI banding pattern. Comparison of the FISH signal with the DAPI banding pattern indicated that the mouse RAD50 gene is located on chromosome 11 (Fig. 5, A and B). Analysis of an additional 10 mitotic figures further localized the mouse RAD50 gene to the A5-B1 region of chromosome 11 (Fig. 5C).
Northern blot tissue surveys were performed to ascertain the developmental pattern of mouse RAD50 expression. A ϳ5.3-kb transcript was observed in most tissues from embryonic day 18 mice. Particularly high levels were present in the heart, liver, and thymus (Fig. 6A). A prominent 5.3-kb transcript was also detected in AT-2 cardiomyocyte RNA. By neonatal day 1.5, the pattern of RAD50 expression had dramatically changed. Expression in the brain, gut, liver, and skin was markedly reduced, whereas expression in the spleen was induced (Fig. 6B). Although steady state levels of the 5.3-kb transcript in heart and lung were similar to those observed in fetal mice, a second transcript of ϳ3.1 kb was also detected in these tissues. Most adult tissues expressed little or none of the 5.3-kb transcript ( Fig. 6C), although relatively high levels were detected in heart, lung, and aorta and very high levels were present in testes (Fig. 6C). Once again the 3.1-kb transcript was observed in the heart and lung, while an additional transcript of ϳ1.6 kb was observed in the testes. In each case, the RNA samples were hybridized with an 18 S rRNA oligonucleotide probe to confirm their integrity (Fig. 6).
Additional Northern blots were performed to define the relationship between the 5.3-, 3.1-, and 1.6-kb RAD50 transcripts. Probes were generated from the 5Ј end (nucleotide residues 119-1684), the middle region (nucleotide residues 1685-3516), and the 3Ј end (nucleotide residues 3517-4069) of the RAD50 cDNA. Probes from the middle region and 3Ј end hybridized to both the 5.3-kb and 3.1-kb transcripts in AT-1 cardiomyocyte RNA, while the 5Ј probe hybridized only to the 5.3-kb transcript (Fig. 7A). This observation indicated that the 3.1-kb splice variant was derived from the 3Ј half of the RAD50 gene. Given that cDNA clones 7 and 1.01 were ϳ3 kb in length and carried sequences from the 3Ј half of RAD50 (Fig. 2C), it is likely that they were derived from the 3.1-kb transcript. Se- quence analysis of these cDNA clones revealed a 127-bp deletion, which resulted in a change in reading frame (Fig. 3B). Reverse transcriptase-polymerase chain reaction (RT-PCR) amplification was used to determine if this 127-bp deletion was present in adult heart RNA, a tissue that expresses high levels of the 3.1-kb transcript. Oligonucleotide primers bracketing the deletion (sense primer, nucleotide residues 79 -108; antisense primer, nucleotide residues 489 -518; see Fig. 3B) produced two amplification products, one corresponding to the non-deleted sequences (566 bp) and one corresponding to the deleted sequences (439 bp, see Fig. 7B). Diagnostic sequence analysis confirmed the identity of the amplification products. Collec-tively these data suggest that the 3.1-kb transcript can encode a protein co-linear with the C-terminal portion of RAD50, encompassing amino acid residues 887-1312 of the full-length protein and with a predicted mass of 50 kDa.
Northern blot analyses of adult testes RNA revealed that the 5Ј probe hybridized to both the 5.3-kb and 1.6-kb transcripts, while probes from the middle region and 3Ј end hybridized only to the 5.3-kb transcript (Fig. 7A). This observation indicated that the 1.6-kb splice variant was derived from the 5Ј half of the RAD50 gene. Given that cDNA clone 15.2 was ϳ1.8 kb in length and carried sequences from the 5Ј half of RAD50, it is likely that it was derived from the 1.6-kb transcript (Fig. 2D). Sequence analysis indicated that clone 15.2 was identical to the full-length RAD50 transcript through nucleotide residue 1691. Beyond this point the nucleotide sequences were divergent, and the deduced amino acid sequence of the protein encoded by clone 15.2 terminated after only 13 additional amino acid residues (Fig. 3C). This clone also carried a 154-nucleotide residue 3Ј-untranslated region, as well as a poly(A) tail of 10 residues. RT-PCR analyses were performed to determine if the novel 3Ј sequences present in cDNA clone 15.2 were also present in RNA prepared from adult testes, a tissue that expresses high levels of the 1.6-kb transcript. RT-PCR amplification with primers encompassing the novel sequences (sense primer, residues 1365-1394; antisense primer, residues 1722-1751; see Fig. 3C) generated the expected 384-bp product (Fig. 7B). Diagnostic sequence analyses confirmed the identity of the amplification product. Collectively, these data suggested that the 1.6-kb transcript can encode a protein derived from the Nterminal half of RAD50 and with a predicted mass of 57.8 kDa. Finally, data base surveys revealed that the 3Ј-untranslated region of clone 15.2 encoded elements of mouse B1 repeats. As anticipated, Northern blot analyses with these sequences iden- A, northern blot of AT-1 cardiomyocyte (AT-1) and adult testes (TEST) RNA hybridized with a probe from the 5Ј, middle (MID), and 3Ј region of the RAD50 cDNA, respectively. The positions of the 28 S and 18 S rRNAs are indicated on the left, while the positions of the various RAD50 transcripts are indicated on the right. B, RT-PCR amplification of putative RAD50 splice variants. Lane 2 shows the amplification products from adult heart RNA using primers corresponding to nucleotide residues 79 -108 (sense primer) and 489 -518 (antisense primer). Products of 566 and 439 bp were obtained. Lane 3 shows the amplification products from adult testes RNA using primers corresponding to nucleotide residues 1365-1394 (sense primer) and nucleotide residues 1722-1751 (antisense primer) of clone 15.2. The expected 384-bp amplification product was obtained. Lane 1 shows molecular size standards; sizes are as indicated.
tified numerous transcripts (data not shown).
To determine if RAD50 expression in mammalian cells is modulated in response to DNA damage, NIH-3T3 cells were grown in the presence of increasing concentrations of MMS and cell viability was determined 48 h later (Fig. 8A). Approximately 25% of the cells were viable in cultures maintained at an MMS concentration of 30 g/ml. Northern blot analyses of RNA harvested from control and treated (30 g/ml MMS) cells demonstrated an increase in the steady state levels of RAD50 transcripts (Fig. 8C). In other studies, NIH-3T3 cells were exposed to UV light in the presence of increasing concentrations of MOPS. A concentration of 0.2 g/ml MOPS reduced the viability of NIH-3T3 cells to ϳ15% of that observed for untreated cells following UV irradiation (Fig. 8B). Northern analysis revealed an increase in steady state levels of RAD50 transcripts in cells treated with UV light plus MOPS, but not in cells treated independently with UV or MOPS (Fig. 8C). In each instance, rehybridization of the blot with an 18 S rRNA probe confirmed that similar amounts of RNA were examined. Thus, transcription of the mouse RAD50 gene appears to be increased in response to DNA damage.
The high degree of sequence homology strongly suggested that the 153-kDa mouse cardiomyocyte phosphoprotein is a structural homologue of the yeast RAD50 gene product. To ascertain the extent of functional homology, sequences encod-ing the mouse protein were subcloned into a yeast expression vector under the transcriptional regulation of the glyceraldehyde 3-phosphate promoter. The resulting construct, pGP153, was transfected into a rad50 mutant yeast strain (EI444). Stable transformants (designated EI444/pGP153) were selected by complementation of an auxotrophic leu2 mutation. As a negative control, transfectants carrying vector only were also produced (designated EI444/pGP). The strains were tested for their ability to grow in the presence of the DNA-damaging agent MMS. Wild type RAD50 yeast (strain EI417) were essentially unaffected by MMS at a concentration of 0.006%, exhibiting an average plating efficiency of ϳ70% as compared to cells plated in the absence of the drug (Fig. 9). In contrast, rad50 mutant yeast (strain EI444/pGP) exhibited on average a 1,500fold reduction in plating efficiency (0.049%) at this concentration of MMS. Expression of the 153-kDa mouse protein in rad50 mutant yeast (strain EI444/pGP153) significantly attenuated the effect of MMS; the plating efficiency (6.5%) was on average only 10-fold less than that observed for wild-type yeast and 130-fold greater than that for the rad50 strain (Fig. 9). Thus, the 153-kDa mouse protein can rescue (at least partially) the yeast rad50 mutation, indicating that the two proteins are functional homologues. DISCUSSION Previous studies aimed at characterizing the SV40 large T antigen-binding proteins in transformed cardiomyocytes identified a novel phosphoprotein recognized by anti-p53 antibody mAb421 (6). In this study, we show that this protein is the mammalian homologue of the yeast RAD50 gene product. Yeast rad50 mutants were initially isolated by virtue of their sensitivity to ␥-irradiation (39). Subsequent analyses have demonstrated that most rad50 mutants are unable to repair x-ray-induced DNA damage (specifically, double-stranded DNA breaks) and also are unable to undergo meiotic recombination (reviewed in Ref. 40). Deficiencies in meiotic recombination resulted from failure of synaptonemal complex formation, followed by catastrophic progression through meiosis I. As a consequence of this latter deficiency, sporulation in rad50 yeast is completely defective. Given that recombinogenic doublestranded breaks do not appear in the absence of RAD50 function (41), the slightly elevated levels of mitotic recombination (42) and successful mating type switching (43) observed in rad50 mutants is perplexing. Furthermore, rad50 strains exhibit essentially normal levels of homologous recombination but severely reduced levels of illegitimate recombination (44). Yeast RAD50 is a 153-kDa protein composed of an N-terminal nucleotide binding domain and two regions of extensive heptad repeats (38). Studies with purified RAD50 protein documented an ATP-dependent DNA binding activity in vitro, as well as transcriptional regulation of the RAD50 gene during meiosis in vivo (45). Despite this extensive body of data pertaining to RAD50 structure and the phenotypic analysis of rad50 mutant yeast strains, the precise function of the protein remains controversial.
The extent of structural homology observed between the mammalian and yeast RAD50 proteins is striking. Both proteins are 1312 amino acids in length, and regions exhibiting Ͼ50% amino acid identity and spanning a total of more than 300 residues are readily identified. The first region of homology extends from amino acid residues 1 through 81, and encompasses the nucleotide binding domain. Nucleotide binding domains consist of a hydrophobic ␤-strand and an ␣-helix separated by a loop containing the consensus sequence of A/G-X-X-X-X-G-K-S/T (46). The sequence of the mouse RAD50 loop (G-P-N-G-A-G-K-T) agrees well with this consensus sequence. Furthermore, the ␤-strand and ␣-helix motifs are largely conserved between yeast and mouse RAD50. The importance of the ATP binding domain to RAD50 function is underscored by the observation that amino acid substitutions within the loop motif completely abrogate activity in yeast RAD50 (40). Extensive sequence homology is also evident at amino acid residues 129 -192 and 1117-1271. Interestingly this latter region encompasses a rather large hydrophobic stretch, which is highly conserved between the mouse and yeast proteins (Ref. 38; Fig.  4). Mouse RAD50 also exhibits a limited degree of homology to several myosin heavy chain proteins, centromeric protein E, yeast USO1, and desmoplakin I and II. All of these proteins contain regions of heptad repeats, a structural motif that gives rise to a coiled coil conformation (47). Heptad repeats are also a prominent structural motif in yeast RAD50 (38).
Although expression of a mouse RAD50 transgene renders rad50 mutant yeast more than 100-fold less sensitive to MMS, the transfected cells are nonetheless ϳ10-fold more sensitive to DNA damage than wild type strains of yeast (Fig. 9). Several factors may contribute to the incomplete rescue of the rad50 phenotype. For example, the repair of double-strand DNA breaks in yeast is thought to require the activity of a multiprotein complex (48). This view is supported in part by the presence of three leucine zippers in mouse RAD50 (Fig. 4); leucine zippers are structural motifs that facilitate proteinprotein interactions (49). Indeed, recent studies have suggested that RAD50 activity is dependent upon binding to Mre1, a protein that, when mutated, gives rise to a similar spectrum of phenotypes observed in rad50 yeast (50). Subtle differences between the RAD50-binding proteins in yeast versus mouse would likely impact on the ability of the mammalian protein to rescue rad50 mutant yeast. Alternatively, the partial complementation might simply reflect the relatively low steady state levels of mouse RAD50 transgene transcripts in the transfected yeast.
The pattern of RAD50 transcription differs markedly from that observed for other mammalian DNA repair genes, as is exemplified by the ubiquitous expression of XRCC4 and RAD52 in adult mice (51,52). Indeed, RAD50 exhibits a complex pattern of transcriptional regulation during mouse development. Full-length RAD50 transcripts are detected in most embryonic tissues examined, consistent with its role in DNA repair in yeast. Similarly, the high levels of RAD50 observed in AT-1 and AT-2 cardiomyocytes are consistent with the high mitotic index observed in these cells. The reduction in RAD50 transcript levels observed in most adult tissues is thus not surprising, given that organogenesis is typically associated with cell cycle withdrawal and terminal differentiation. The pronounced ex-pression detected in adult testes is consistent with RAD50's role in meiotic reduction and sporulation in yeast. Based on this result, one would predict that high levels of expression would also be observed during oogenesis. In contrast, the persistent expression of RAD50 transcripts in the adult myocardium and lung a priori is contrary to the known functions of the protein. This is particularly true in the heart, where the inability of adult cardiomyocytes to proliferate is well established (1). While it may be argued that RAD50 activity may suppress the accumulation of mitochondrial mutations in the heart (an organ with particularly high energy demands), this is unlikely as the presence or absence of RAD50 activity does not impact on the ability to generate rho Ϫ mutants in yeast (53). It will be of interest to determine if myocardial RAD50 expression is cardioprotective, particularly in light of the susceptibility of this tissue to free radical injury resulting from ischemia/reperfusion.
Northern blot and RT-PCR analyses suggested that alternative splicing of the RAD50 gene can generate additional transcripts of 1.6 and 3.2 kb. The 1.6-kb transcript (represented by cDNA clone 15.2) can encode a 498-amino acid protein with a deduced molecular mass of 58 kDa. This protein would be co-linear with the N terminus of RAD50 and would contain the nucleotide binding domain. The 1.6-kb transcript is expressed exclusively in the testes, suggesting that it may function predominately in meiosis. The 3.1-kb transcript (represented by cDNA clones 7 and 1.01) can encode a protein of 426 amino acids, which is co-linear with the C-terminal portion of RAD50. It should, however, be noted that this protein would initiate at the seventh methionine residue in the transcript, which is located downstream from the 127-bp deletion (Fig. 3B). Although the initiation of translation at this methionine residue would violate the first-AUG rule, the nucleotide sequence (cgc-CAGCAAATGG) agrees well with the initiation codon consensus sequence (see above). Expression of this transcript is limited to the heart and lung (the only adult tissues other than the testes and aorta that exhibit appreciable expression of the full-length RAD50 transcript). The functional significance of this splice variant is presently unknown.
FISH analyses localized the mouse RAD50 gene to the A5-B1 region of chromosome 11. Other loci within this region include Flt-4 (FMS-like tyrosine kinase 4), Il-5 and Il-3 (interleukins 5 and 3), Tcf7 (T-cell specific transcription factor 7), Anx6 (annexin VI), Csfgm and Csfmu (colony stimulating factors), IRF-1 (interleukin regulatory factor 1), and Sparc (secreted acidic cysteine rich glycoprotein). Thus, although several interesting genes reside close to RAD50, none display a phenotype consistent with that presaged by rad50 mutant yeast. Furthermore, examination of syntenic maps failed to establish any genetic linkage between mouse RAD50 and mutants in human and rodent cell lines which exhibit defects in DNA repair. Consequently, assessment of the rad50 null phenotype in a mammalian system will likely require gene targeting experiments. Finally, the genetic position for mouse RAD50 reported here is substantiated in part by sequence analyses from P1 clones from the interleukin gene cluster on human chromosome 5. Data base surveys revealed a short open reading frame with homology to mouse RAD50. This region of human chromosome 5 is syntenic with A5-B1 region of mouse chromosome 11.
It is intriguing to speculate on the potential significance of the limited antigenic homology observed between RAD50 and p53. Our interest in RAD50 was kindled by virtue of its crossreactivity with mAb421 (6). This antibody recognizes a C-terminal epitope in p53 located at amino acid residues 371-380 (54). Other studies have shown that the C-terminal region of p53 (spanning amino acid residues 310 -390) is required for oligomerization (55) and for binding to single-stranded DNA and RNA (56,57). The C-terminal domain is also able to stably bind to double-stranded DNA fragments carrying insertion/ deletion mismatches (15). In this regard, it is of interest to note that biochemical analysis of yeast RAD50 revealed an ATP-dependent double-stranded and single-stranded DNA binding activity (45). Furthermore, sedimentation coefficient analyses suggested that purified yeast RAD50 can form dimers at 200 mM NaCl, and higher order oligomers at lower ionic strength. Given that both RAD50 and p53 possess the ability to oligomerize and bind to DNA, that the mAb421 epitope in p53 is localized within the oligomerization and DNA binding domains, and that RAD50 is recognized by mAb421, it is possible that the common structural epitope defined by this antibody in both p53 and RAD50 may also impart functional homology. Additional analyses are required to further explore this issue.
In summary, transformed cardiomyocytes express high levels of the RAD50 gene product. The observation that RAD50 is expressed in most fetal tissues, and that high levels of expression are detected in adult testes, suggests that the function of the mammalian protein may be similar to its yeast counterpart. This notion is supported by the observation that expression of the mouse RAD50 cDNA can rescue the rad50 phenotype in yeast, and by the up-regulation of RAD50 transcripts in NIH-3T3 cells treated with compounds that induce DNA damage.