Elevated Homologous Recombination Activity in Fanconi Anemia Fibroblasts*

It is widely believed that Fanconi anemia cells possess a reduced ability to repair inter-strand DNA cross-links. While the mechanism through which inter-strand DNA cross-links are removed from mammalian chromosomes is unknown, these lesions are repaired via homologous recombination in lower eukaryotes and bacteria. Based on the hypothesis that a similar mechanism of DNA repair functions in mammalian somatic cells, we measured homologous recombination activity in diploid fibroblasts from healthy donors, and Fanconi anemia patients. Somewhat surprisingly, homologous recombination levels in nuclear protein extracts prepared from Fanconi anemia cells were nearly 100-fold higher than in extracts prepared from control cells. We observed a similar increase in the activity of a 100-kDa homologous DNA pairing protein in extracts from Fanconi anemia cells. Transfection studies confirmed that plasmid homologous recombination levels in intact Fanconi anemia cells were substantially elevated, compared with control cells. These results suggest that inappropriately elevated levels of homologous recombination activity may contribute to the genomic instability and cancer predisposition that characterize Fanconi anemia.

Individuals suffering from the autosomal recessive disorder Fanconi anemia (FA) 1 have a greatly elevated risk for leukemia and other malignancies (1). The disease is genetically heterogeneous, with at least five distinct complementation groups having been identified (2). The major clinical features of this illness include progressive pancytopenia, and a number of congenital abnormalities (3). In affected individuals, the incidence of acute myelogenous leukemia is nearly 15,000-fold greater than the general population, and these individuals are also at a higher risk of developing solid tumors (1). While the recent cloning of the Fanconi anemia complementation group C and A genes (4 -6) have not provided insight into the cause of this disease, analysis of cultured cells isolated from affected individuals has provided intriguing clues to the possible molecular defects responsible for this disease. FA cells show enhanced levels of spontaneous and induced chromosome breakage (7) and are particularly susceptible to the clastogenic and cytotoxic effects of DNA damaging agents such as diepoxybutane and mitomycin-C (7). These in vitro findings, in conjunction with the clinical predisposition to cancer suggest that the primary genetic defects in FA may involve DNA repair or recombination genes. Since these cells are not particularly sensitive to ultraviolet or ionizing radiation, or to monofunctional alkylating agents such as ethyl methane sulfonate or methyl methane sulfonate, it appears likely that the affected DNA repair pathway is specific for the removal of DNA cross-links. In light of evidence that the nucleotide excision repair pathway functions to remove intra-strand cross-links (8,9), it would appear that the defect in FA cells involves specifically the removal of interstrand DNA cross-links. The recent observations that interstrand DNA cross-link binding activity is absent from protein extracts prepared from FA complementation group A cells (10), and that these cells possess a diminished ability to remove DNA cross-links in vitro (11,12) further support this hypothesis.
Numerous studies of lower eukaryotes and bacteria have provided strong evidence that lesions of this type are removed via homologous recombination (HR) in those organisms (13)(14)(15)(16)(17). A number of recent observations suggest that mammalian somatic cells also utilize HR to repair DNA damage. First, mammalian homologs of the yeast Rad51 protein (18), which is itself a functional homolog of the bacterial strand transferase recA (19,20), have been identified (21), and shown to be expressed in somatic cells. Second, it has been determined that the Rad51 protein localizes to discrete foci within murine somatic cells that have been exposed to DNA damaging agents. These foci were shown to be sites of unscheduled DNA synthesis, suggesting that they were sites of DNA repair activity (22). Third, mammalian homologs of another yeast recombinational repair gene, RAD52, have been identified (23)(24)(25). Overexpression of either the yeast (26) or human (27) RAD52 genes in mammalian somatic cells has been shown to enhance cellular homologous recombination, and confer resistance to DNA damaging agents, including the DNA cross-linking agent diepoxybutane.
Based on these observations, we hypothesized that the apparent DNA repair deficiency in Fanconi anemia cells could result from defective recombinational repair. In apparent contradiction to this model, a previous study indicated that HR levels in SV40-immortalized FA cells were indistinguishable from those seen in wild-type cells (28). However, we (29) and others (30) have demonstrated that HR levels are greatly elevated in both spontaneously and virally immortalized cells. These latter observations suggested that it would be more appropriate to examine homologous recombination levels in diploid cells from Fanconi anemia patients.
In this report we describe our results indicating that nuclear protein extracts prepared from Fanconi anemia cells obtained from two unrelated donors catalyze elevated levels of inter-plasmid homologous DNA recombination activity. We also demonstrate that these extracts possess substantially elevated levels of a 100-kDa homologous DNA pairing protein, that has been previously associated with immortalized cells (29,30). Finally, we show that plasmid recombination substrates introduced into Fanconi anemia cells undergo greatly elevated levels of intra-molecular homologous recombination. We discuss the possible relationship between this elevated HR activity, and the genomic instability and hypersensitivity to DNA crosslinking agents known to be associated with Fanconi anemia.

EXPERIMENTAL PROCEDURES
Cells and Culture Conditions--Diploid fibroblasts were obtained from the American Type Culture Collection (CeRel, HG261) or provided by Dr. Robert O'Dea, University of Minnesota. All diploid cells were used at passage numbers between 6 and 12. The human fibrosarcoma cell line HT1080 (31) was kindly provided by Dr. Raju Kucherlapati, Albert Einstein College of Medicine, Bronx, NY. Cells were maintained in Dulbecco's modified Eagle's medium, supplemented with 9% fetal bovine serum, in a humidified, 5% CO 2 environment.
Nuclear Protein Extracts-Extracts were prepared from growing cells exactly as described (32). Protein concentration was determined by the method of Bradford (33).
Cell-free Homologous Recombination Assay-HR catalyzed by cellfree extracts was performed essentially as described (32)  Briefly, 500 ng of each of these plasmids were incubated with nuclear protein extracts prepared from human fibroblasts, purified, and used to electroporate the recombination-defective Escherichia coli strain DH10B using a Life Technologies Cell-Porator. The ratio of kanamycinresistance (resulting from the generation of an intact neomycin phosphotransferase gene, see Fig. 1) to ampicillin resistance (both parental molecules confer ampicillin resistance) permits one to calculate the frequency with which HR has occurred within the extract. Control experiments were performed in parallel in which the two plasmids are separately incubated with the extracts and subsequently purified, combined, and used to transform bacteria. The frequency of HR seen in these controls was subtracted from that seen in coincubated samples to determine the frequency with which HR occurred within the extract. In the experiments described in this report this "background recombination" frequency was always 10 -20-fold lower than that seen in extracts prepared from cells possessing elevated levels of HR.
Homologous DNA Pairing Assay-The pairing-on-membrane (POM) assay described by Akhmedov et al. (36) was used to quantitate the amount of homologous DNA pairing activity present in nuclear protein extracts from normal, immortalized, or FA cells. Briefly, the assay measures the ability of protein immobilized on a nitrocellulose membrane to catalyze the pairing of linear double-stranded DNA (which is in solution) with homologous single-stranded circular DNA that is also immobilized on the membrane. This pairing is detectable because the double-stranded DNA is labeled by incorporation of radioactive nucleotides using the Klenow fragment of E. coli DNA polymerase. This double-stranded DNA contains a region of approximately 1.5 kb that is not homologous to the immobilized single-stranded DNA. Depending upon where this DNA is linearized, it is possible to create molecules that lack complementarity on either or both ends (see Fig. 4). When double-stranded DNA lacking any complementarity to the immobilized single-stranded DNA is used in these experiments, we do not detect any pairing activity.
Intracellular Homologous Recombination Assay-Homologous recombination activity within human fibroblasts was measured with plasmid pSV2neoDR:DL(D) (see below) as described previously (29). Briefly, we utilized calcium phosphate co-precipitation (37) to cointroduce a replication-defective plasmid pRSVEdl884 (38) encoding the SV40 large T antigen along with the plasmid recombination substrate. Following a 48-h incubation, low molecular weight DNA was recovered (39) and digested with the restriction endonuclease DpnI, to remove unreplicated plasmid DNA. The recombination substrate pSV2neoDR:DL(D) will replicate during this incubation period due to the presence of the SV40 large T antigen. As a consequence, methyl adenine moieties (introduced into the plasmid during its amplification in the bacterial host that produced it) will be replaced by non-methylated adenine residues, rendering these plasmids resistant to the restriction endonuclease DpnI. The pRSVEdl884 plasmid, which lacks an origin of repli-cation, along with plasmid molecules that have not entered the cell nucleus during transfection, or have not replicated, will be degraded by the restriction endonuclease. This restriction-digested material is used to transform recombination-deficient E. coli, and the bacteria plated on ampicillin or ampicillin plus kanamycin containing plates. As in the cell-free recombination described above, the number of ampicillin-resistant colonies indicates the total number of plasmids recovered, while the number of kanamycin-resistant colonies indicates the number of recombinant plasmids present. Control experiments indicate that the frequency of HR seen under these conditions is significantly greater than that which occurs when pSV2neoDR:DL(D) is directly introduced into E. coli. In all cases, the data represent the pooled results obtained from at least 6 separate transfections.
Plasmid DNA-Single-and double-stranded M13mp18 were obtained from New England Biolabs. The construction of plasmids pSV2neoDL (34), pSV2neoDR (34), and pRSVEdl884 (37) 2C). Plasmid pSV2neoDR:DL(D) was constructed by introducing a Klenow-treated HindIII,SmaI fragment of pSV2neoDL (containing the neomycin phosphotransferase gene) into the Klenow-treated EcoRI site of pSV2neoDR. Note that the transcriptional orientation of both the neomycin phosphotransferase (NEO) genes within this plasmid is counterclockwise. Single-stranded circular molecules were derived from ss-neoDR which was made by introducing the NEO gene from pSV2neoDR into M13mp18.

RESULTS
Rationale-We wanted to test the hypothesis that FA cells possess altered levels of homologous DNA recombination activity. A number of well established assays may be used to measure HR activity. In all cases, the approach involves the detection of recombinant, or joint, molecules that have been formed from two distinct, yet homologous, DNA sequences. In general, these assays involve measuring HR in either nuclear protein extracts, or in whole cells that have been transfected with recombination substrate. Ideally, to provide compelling evidence that HR activity in FA cells is altered relative to normal cells, it would be desirable to obtain similar results using both cell-free as well as intact cell systems. Toward this end, we initially measured HR activity in cell-free extracts from FA cells, using two different recombination assay systems. The FA cells used in this study were non-transformed, diploid fibroblasts derived from two unrelated individuals.
Elevated Levels of HR Activity in Nuclear Protein Extracts from FA Fibroblasts-Nuclear protein extracts (32) were prepared from diploid fibroblasts from two unrelated FA patients, and from normal donors. We initially measured the ability of these different extracts to catalyze HR between two doublestranded circular plasmids, pSV2neoDL and pSV2neoDR (see Fig. 1). Using the recombination substrate pair of pSV2neoDL and pSV2neoDR, HR levels in the FA cells were elevated 10 -20-fold, relative to the normal cells, in which HR activity was undetectable (Table I). Interestingly, the HR activity in the FA cells was nearly as great as that seen in the transformed cell line HT1080.
To extend this finding, we repeated the HR assays using the substrate pair of double-stranded pSV2neoDL, linearized with the restriction enzyme SalI, and single-stranded ssneoDR (see Fig. 1). This substrate pair yields lower levels of background HR activity (see "Experimental Procedures"), providing a better opportunity to determine the extent to which HR in FA cell-derived extracts is elevated, relative to extracts from normal cells. As Table I indicates, we observed that HR levels in the FA-derived extracts were between 50 and 100-fold greater than those seen in extracts from normal cells, and nearly indistinguishable from those seen in the immortalized HT1080 cell extracts. HR activity levels seen in this latter cell line are similar to those seen in other immortal cell lines (29,30).
Molecular Analysis of the Products of the Cell-free HR Reaction-The cell-free HR assay relies upon complementation of the strand-transferase deficiency of recA Ϫ E. coli (32). The results presented indicate that joint molecules can be formed between two circular double-stranded molecules (substrate pair A, Fig. 1), or between a single-stranded circular molecule and a double-stranded linear molecule (substrate pair B, Fig.  1). However, it is conceivable that a combination of enzymatic activities (for example, the concerted actions of an endonuclease and an annealing activity) within the extract could generate a joint molecule. Indeed, a number of reports indicate that such a process, referred to as single-strand annealing, can be detected in cultured mammalian cell lines and in Xenopus oocytes (40 -43). In principle, therefore, it is possible that single-strand annealing, as opposed to "traditional" HR is respon-  sible for the formation of kanamycin-resistant colonies represented in Table I. Fortunately, it is possible to distinguish between these two recombination pathways (44 -46). As Fig.  2A illustrates, the single-strand annealing pathway predicts that the kanamycin-resistant product of HR between doublestranded pSV2neoDL and pSV2neoDR will be monomeric pSV2neo. In contrast, since authentic HR is conservative, this pathway (Fig. 2B) will yield two NEO genes, one of which will be full-length. Note that the nature of this second allele will depend upon a host of factors, including, for example, whether gene conversion or reciprocal recombination has occurred. The major distinguishing feature of these two pathways is that single-strand annealing yields a single NEO gene, whereas authentic homologous recombination produces two genes. Since co-transformation of E. coli with two plasmid molecules does not occur to an appreciable extent in our system (data not shown) the demonstration that kanamycin-resistant bacteria harbor two alleles of the NEO gene would strongly support the hypothesis that authentic HR is responsible for the generation of kanamycin-resistant colonies.
Plasmid DNA was isolated from a number of kanamycinresistant bacterial colonies derived from cell-free HR reactions (using the substrate pair of pSV2neoDL and pSV2neoDR) catalyzed by FA cell-derived nuclear protein extracts. This material was treated with the restriction enzyme PstI, and separated on a 1% agarose gel. As described under "Experimental Procedures," digestion with this endonuclease permits one to identify the presence of either or both of the NEO genes. An ethidium bromide-stained gel of two such digests is presented in Fig. 2C. Lanes 1-3 contain pSV2neoDL, pSV2neoDR, and pSV2neo wild-type, respectively. Lanes 4 and 5 contain DNA from kanamycin-resistant colonies obtained from a cell-free HR experiment. As the figure indicates, in both cases, bands corresponding to full-length as well as both deletion alleles are present (see legend to Fig. 2 and "Experimental Procedures"). Electrophoresis of uncut plasmid DNA from these two colonies indicates that they both contained a single, dimeric plasmid (not shown). These two observations confirm that both of these colonies arose following a single reciprocal (conservative) HR event, such as is depicted in Fig. 2B. An expanded analysis of this type determined that 83% (25/30) of the kanamycin-resistant colonies harbored two NEO genes, while 17% (5/30) contained a single, wild-type gene. (In 5 out of 30 cases, we recovered one wild-type NEO gene and one NEO gene harboring both the pSV2neoDL-derived and pSV2neoDR-derived deletions. In these latter cases, we concluded that a single reciprocal recombination event had occurred. It is noteworthy that we detect both reciprocal and non-reciprocal, or apparent gene conversion, events in this system. In addition, intra-molecular recombination appears to occur with some frequency within the E. coli host, converting dimeric molecules to mixed monomers.) As these results indicate, in the great majority of cases (83%), conservative HR, rather than non-conservative single-strand annealing is responsible for the formation of joint molecules within the nuclear protein extracts.
A 100-kDa Homologous DNA Pairing Protein Is Highly Active in Extracts Prepared from FA Cells-The analysis of recombinant plasmid DNA described above is consistent with our hypothesis that authentic HR has occurred within the nuclear protein extracts. To obtain additional, independent support that HR activity present in the FA (and immortalized) fibroblasts, and absent in the normal fibroblasts was responsible for the results depicted in Table I, we turned to the POM assay (36). This in situ assay demonstrates the presence of homologous DNA pairing activity, and has been used to detect a 100-kDa recombination protein in nuclear extracts prepared from immortalized mammalian somatic cells (36). We have confirmed this observation, and shown that this protein is inactive, or absent, in extracts prepared from a number of normal somatic cells, including rat liver, murine embryo fibroblasts, and normal human diploid dermal fibroblasts (29,47). As shown in Fig. 3, this 100-kDa homologous DNA pairing protein is active in nuclear protein extracts from FA fibroblasts, but not in extracts from normal fibroblasts. As a control, we performed identical experiments in which there was no homology between the single-and double-stranded DNA substrates. Under these conditions, we did not detect any DNA pairing activity in any of several experiments, confirming a number of previously published reports (29,36,47,48, and data not shown).
DNA Pairing Activity in FA-derived Extracts Requires Terminal Homology-We next asked whether homologous pairing activity could be detected between two otherwise complementary molecules if regions of non-homology were present on the ends of the linearized double-stranded substrate. Two parallel POM assays were conducted. In one, the radiolabeled DNA substrate was linearized, generating ends homologous to the immobilized DNA substrate (Fig. 4, column A), while in the second, the double-stranded substrate contained ends that were not homologous to the immobilized substrate (Fig. 4,  column B). The double-stranded substrates were end-labeled and used to carry out POM assays ("Experimental Procedures"). As column A of Fig. 4 indicates, when the ends of the double-stranded molecule were complementary to the immobilized single-stranded substrate, significant homologous pairing was catalyzed by the 100-kDa protein, while, the presence of heterologous free ends (column B, Fig. 4) POM activity was significantly reduced. This result indicates that a free homologous DNA end is required to form a joint molecule. Removal of only 52 base pairs of non-homology from one end of the doublestranded substrate would generate a homologous free end available for pairing with the immobilized single-stranded circular DNA. Our results therefore indicate that joint molecule formation is not accompanied by extensive nuclease activity, although we cannot rule out the possibility that limited nuclease activity may be present (see below).
Homologous DNA Pairing Activity Is Directional-Linearized double-stranded DNA substrates containing non-homology on either the 5Ј or 3Ј end were radioactively labeled and a POM assay performed (the "end" in question refers to the strand of the DNA duplex that is complementary to the singlestranded circular molecule immobilized on the membrane). As   FIG. 3. Homologous pairing activity detected in nuclear protein extracts prepared from FA fibroblasts. A, POM assays were performed using nuclear protein extracts prepared from fibroblasts isolated from a normal donor (lane 1), from two unrelated FA donors (lanes 2 and 3), or from the transformed fibrosarcoma line, HT1080 (lane 4). Fig. 4 indicates, when the complementary 3Ј end of the doublestranded DNA is homologous to the immobilized substrate (column C), extensive POM activity is detected. In contrast, when this region of the DNA duplex is non-homologous (column D) very little POM activity is associated with the 100-kDa protein. Since homologous pairing activity must initiate at least predominantly at free ends (Fig. 4, columns A and B), these latter observations indicate that homologous pairing between the two molecules initiates in a 5Ј to 3Ј direction with respect to the single-stranded molecule.
HR Levels Are Elevated in Intact FA Cells-The results presented in Table I and Figs. 2 and 3 strongly suggest that HR activity is elevated in FA fibroblasts, relative to normal cells. However, it is conceivable that the HR and joint molecule formation (POM) activities do not accurately reflect events occurring within intact cells. To address this issue, we performed a third series of experiments in which a plasmid recombination substrate was introduced via calcium phosphatemediated co-precipitation into normal, and transformed fibroblasts, as well as the diploid fibroblast strain CeRel (American Type Culture Collection, referred to in Tables I and II as FA1), derived from an FA patient. The substrate in these experiments contained a direct repeat of two defective hetero-alleles of the NEO gene, which, when intact, confer resistance in bacteria to the antibiotic kanamycin (substrate D, Fig. 1). As the results in Table II indicate, the frequency of extrachromosomal intra-plasmid HR in FA1 (ATCC HG261) cells is ϳ20fold greater in FA diploid fibroblasts, than in either of two normal control fibroblasts. Chi-square analysis reveals that these results are significant (p Ͻ 0.005).

DISCUSSION
The results presented here demonstrate that homologous recombination activity is substantially elevated in diploid fibroblasts obtained from Fanconi anemia patients. Using a number of independent assays, we estimated that the HR activity present in FA cells, or in nuclear protein extracts prepared from these cells, was 20 -100-fold higher than that seen in normal cells or cell extracts. The levels of HR in the FA cells and cell-derived extracts were essentially indistinguishable from those seen in immortalized or transformed cells.
It is currently unknown whether elevated HR activity in diploid fibroblasts is a universal feature of Fanconi anemia. Fanconi anemia is genetically heterogeneous, with at least 5 distinct complementation groups having been identified. Since complementation analysis cannot be performed on fibroblasts, and since no lymphoid cell lines were developed from these patients, their complementation group status cannot be determined. It will therefore be of interest to perform additional experiments with diploid fibroblasts from patients of all the known complementation groups. This effort may be difficult, as there may well be more than 5 complementation groups comprising FA. Finally, in contrast to other recombination assays, the POM assay can be performed on a small number of cells. It may therefore be feasible to determine whether elevated joint molecule formation activity is associated with other, non-fibroblast, cell types obtained from FA patients.
The use of multiple recombination assays provides confidence in our conclusions. For example, the presence of intraplasmid recombination within intact FA cells does not by itself indicate that elevated levels of recombination enzymes are present in these cells. First, it is known that damaged DNA is a preferential substrate for HR (reviewed in Ref. 49). Second, it has also been reported that FA cells contain elevated levels of a clastogenic substance (50). The elevated levels of cellular recombination in FA cells could therefore result from enhanced levels of recombinogenic lesions within plasmid DNA introduced into those cells. However, the presence of greatly elevated levels of HR activity in nuclear protein extracts argues against this idea. First, it is difficult to believe that sufficient damage would occur within the plasmid DNA incubated with FA nuclear extracts to significantly stimulate recombination. Even complete linearization of one of the DNA substrates only modestly (about 10-fold, not shown) stimulated recombination, while the FA extracts are as much as 50 -100-fold more active in recombination than are normal cell extracts (Table I). Second, it is unlikely that any DNA damaging activity present in FA extracts would remain associated with the 100-kDa protein, as it must, to influence the POM activity.
Similarly, the demonstration that plasmid molecules introduced into FA cells recombine at elevated frequencies confirms that the HR activity observed in nuclear protein extracts from these cells accurately reflect events occurring within the intact cells. We previously conducted a similar analysis of a large number of normal and immortal cells, and reached an identical conclusion (29).
At this time, we do not know why HR levels are elevated in the FA cells we have analyzed. However, a number of features known to be associated with these cells suggest several possibilities. For example, elevated levels of endogenous DNA damage present in FA cells may induce expression of one or a number of recombination enzymes. Alternatively, it is possible that elevated levels of clastogenic molecules could directly induce expression of these enzymes. Finally, it is possible that the well known G2 cell-cycle arrest of FA fibroblasts may play a role in inducing, either directly, or indirectly, in inducing HR activity. It may be possible to provide support for these hypotheses through appropriate manipulations of normal diploid fibroblasts, followed by careful measurements of cellular HR activity levels. FIG. 4. Homology requirements for pairing activity. M13neoDR DNA was linearized with various restriction enzymes to generate (A) two homologous ends, (B) no homologous ends, (C) homology at the 3Ј end of the "plus" strand (the immobilized single-stranded M13mp18 is complementary to the "minus" strand), or (D) homology at the 5Ј end of the plus strand. A series of POM assays were performed, and the amount of radioactivity present on the nitrocellulose membrane in the region corresponding to 100 kDa was determined using a Phosphor-Imager (Molecular Dynamics, Sunnyvale, CA). Do elevated levels of HR contribute to the cellular phenotype observed in FA? In general, the lack of elevated sister chromatid exchange, the apparent non-homology dependent nature of the genomic rearrangements associated with FA cells, as well as their enhanced sensitivity to DNA cross-linking agents (recent results indicate that mammalian cells with elevated HR levels are resistant to DNA damaging agents, Refs. 26 and 27) would suggest that the answer to this question is negative. However, it is possible that grossly elevated levels of homologous strand pairing activity could lead to the inappropriate initiation of aberrant recombination events in FA cells. Initiation of recombination by the 100-kDa protein could lead to the formation of DNA double-strand breaks (or other mutagenic DNA lesions), thus explaining the high levels of spontaneous chromosome aberrations seen in FA cells. The overexpressed homologous DNA pairing protein could interfere with recombinational repair, perhaps through competitive binding to essential recombination proteins or even to DNA recombination substrates thereby explaining the sensitivity of these cells to DNA cross-linking agents.
The elevated levels of HR activity seen in FA cells may also play a role in oncogenesis. As mentioned above, patients with this disease suffer greatly elevated levels of both leukemia and solid tumors. In addition, FA cells in vitro are dramatically more sensitive to SV40-mediated transformation (51)(52)(53)(54). Recently, it has been shown that introduction of a wild-type copy of the appropriate FA gene into primary fibroblasts substantially reduces this sensitivity to SV40 transformation (55). Taken together, these results suggest that FA cells are predisposed to cancer. It is interesting to consider the possibility that elevated levels of HR may play an important role in generating genomic rearrangements, thereby hastening the somatic genetic alterations that result in cancer. In this context, it is noteworthy that the levels of HR activity seen in diploid FA fibroblasts are essentially indistinguishable from those seen in immortalized and transformed cells.