Deficient regulation of DNA double-strand break repair in Fanconi anemia fibroblasts.

Fibroblasts from patients with Fanconi anemia (FA) display genomic instability, hypersensitivity to DNA cross-linking agents, and deficient DNA end joining. Fibroblasts from two FA patients of unidentified complementation group also had significantly increased cellular homologous recombination (HR) activity. Results described herein show that HR activity levels in patient-derived FA fibroblasts of groups A, C, and G were 10-fold greater than those seen in normal fibroblasts. In contrast, HR activity in group D2 fibroblasts was identical to that in normal cells. Western blot analysis revealed that the RAD51 protein was elevated 10-fold above normal levels in group A, C, and G fibroblasts, but was not altered in group D2 fibroblasts. HR activity levels in these former cells could be restored to near-normal levels by electroporation with anti-RAD51 antibody, whereas similar treatment of normal and complementation group D2 fibroblasts had no effect. These findings are consistent with a model in which FA proteins function to coordinate DNA double-strand break repair activity by regulating both recombinational and non-recombinational DNA repair. Interestingly, whereas positive regulation of DNA end joining requires the combined presence of all FA proteins thus far tested, suppression of HR, which is minimally dependent on the FANCA, FANCC, and FANCG proteins, does not require FANCD2.

linking agents such as mitomycin C and diepoxybutane (13)(14)(15). These cellular features, along with the propensity of FA patients to develop cancers, lead to the classification of FA as a DNA repair defect disorder (16). However, unlike other well defined DNA repair defect disorders such as xeroderma pigmentosum, ataxia telangiectasia, and Bloom's syndrome, cloning of the genes defective in FA cells has not identified the molecular defect(s) responsible for the disease (17)(18)(19)(20)(21)(22)(23)(24)(25)(26). However, analysis of the putative functions of these FA gene products has indicated that they interact in a common pathway (27,28), although the function of this pathway is not clear.
Investigation of DNA end joining activity both in intact patient-derived FA cells and in extracts prepared from these cells revealed a defect in DNA end joining activity (29 -32). As this type of double-strand break repair is thought to occur in cells via a non-recombinational mechanism, these data indicate that FA cells have a deficiency in non-recombinational DNA repair. However, there is a growing body of evidence that also links FA proteins to recombinational DNA repair. First, the FANCB and FANCD1 genes were recently shown to be identical to BRCA2 (26). Although its precise function is not known, cells lacking functional BRCA2 protein are nearly completely unable to repair restriction enzyme-generated chromosomal double-strand breaks via homologous recombination (HR) (33). Second, the FANCD2 protein co-localizes with BRCA1 and RAD51 proteins to discrete nuclear foci following exposure to ionizing radiation (27,34,35). BRCA1-deficient cells are also unable to properly utilize HR to rejoin chromosomal double-strand breaks (36), and it is well established that the recombination protein RAD51 plays an essential catalytic role in recombinational DNA repair (37,38). Since it is known that recombinational repair is an important mechanism in the repair of DNA crosslinks (39), these results support the view that FA cells have a deficiency in HR. However, one potential difficulty with this interpretation is that direct examination of HR activity in fibroblasts from two unrelated FA patients of unidentified complementation group status failed to detect a defect in this activity (40). Instead, it was shown that the level of intraplasmid HR activity in these cells was elevated 10 -20-fold above that seen in normal fibroblasts. Thus, the most consistent interpretation is that FA cells have a deficiency in the regulation of these pathways that repair DNA double-strand breaks.
To confirm the initial finding that FA cells have altered HR activity, as well as to gain insight into the nature of the regulation of DNA double-strand break repair pathways in FA cells, we first examined extrachromosomal HR activity in intact patient-derived FA cells of several known complementation groups. We found a 10-fold elevation in HR activity in FA cells belonging to complementation groups A, C, and G, which was reversed by genetic complementation via retroviral transduction. Furthermore, extracts from these cells also had an ϳ10fold increase in the level of RAD51 protein. Inhibition of RAD51 by co-electroporation of anti-RAD51 antibody into these cells restored HR activity to wild-type levels. Interestingly, FA fibroblasts of complementation group D2 did not display a similar elevation in either HR or RAD51 protein. Antibody-mediated inhibition of FANCD2 in intact normal cells showed that only DNA end joining activity was affected, whereas similar treatment of normal cells with anti-FANCC antibody altered both HR and DNA end joining activities. These data support the conclusion that the FA proteins deficient in the cells studied herein are involved in the regulation of both recombinational and non-recombinational DNA repair.

EXPERIMENTAL PROCEDURES
Cell Culture and Plasmid Constructs-Cells were maintained in a humidified 5% CO 2 -containing atmosphere at 37°C. All cell strains were obtained from the Oregon Health Sciences University Fanconi Anemia Cell Repository unless otherwise noted. Cell strains PD.715.F, PD.751.F, PD.792.F, PD.793.F, CCD-1059Sk, CCD-1056Sk, and CCD-1108Sk (American Type Culture Collection Cell Repository) are human diploid fibroblasts derived from normal subjects. The immortalized cell line HT1080 (American Type Culture Collection Cell Repository) was derived from a spontaneous human fibrosarcoma. Cell strains PD.720.F, PD.551.F, PD.145.F, and PD.352.F (referred to as A, C, D2, and G, respectively) are human diploid FA fibroblasts from patients of complementation groups A, C, D2, and G, respectively. The diploid cell strain PD.20.F, derived from an unrelated FA patient of complementation group D2, was used to prepare the nuclear extract used for Western blot analysis of RAD51; all other examinations of "D2" cells were performed with the PD.145.F cell strain. Cell strains 720-FAA, 551-FAC, and 352-FAG (referred to as A cor , C cor , and G cor , respectively) are retrovirally corrected FA cells that are infected with retroviruses encoding FANCA, FANCC, and FANCG cDNAs, respectively. Murine MPF60T and MPF62T cells are embryonic fibroblasts derived from mice homozygous for FancC⌬ exon 9 and from homozygous wild-type mice, respectively (41). Plasmids encoding the patient-derived L554P mutant FANCC allele (referred to as pL554P) and the wild-type FANCC allele were kindly provided by Dr. Maureen E. Hoatlin (Oregon Health Sciences University and Portland Veterans Affairs Medical Center, Portland, OR) (42).
Intracellular HR Assay in Intact Cells-Two similar but slightly different assays were utilized to measure HR in intact cells; both were described previously (40).
Intraplasmid HR-Briefly, intraplasmid extrachromosomal HR was assayed in human diploid fibroblasts by electroporating the plasmid substrate pSV2neoDR:DL(D) (40) along with plasmid pRSVEd1884 (43). The latter plasmid encodes the SV40 large T-antigen, expression of which drives replication of plasmids containing an SV40 origin of replication. It is noteworthy that pRSVEd1884 lacks such an origin of replication. The former plasmid contains an intact ␤-lactamase gene (AMP) that confers resistance to the antibiotic ampicillin, as well as a tandem repeat of two defective heteroalleles of the neomycin phosphotransferase gene (NEO). Each allele contains a different, non-overlapping deletion within the NEO gene (referred to as DR and DL), which renders it nonfunctional. HR occurring between these heteroalleles can regenerate an intact and functional NEO gene, which confers resistance to the antibiotic kanamycin. Following plasmid electroporation, mammalian cells were incubated for 48 h. Plasmid DNA was then recovered from the cells (44) and incubated with the restriction enzyme DpnI, which digests plasmids that have not replicated in mammalian cells. Finally, plasmids were electroporated into DH10B electrocompetent reporter bacteria. HR frequency in the mammalian cells was determined by comparing the number of resulting bacterial colonies obtained from growth on ampicillin-containing medium (referred to as "nonrecombinant colonies") with the number obtained from growth on ampicillin-and kanamycin-containing medium (referred to as "recombinant colonies"). In some experiments, antibody (0.4 g) was introduced into cells via co-electroporation with plasmid substrate.
Interplasmid HR-To determine the frequency of interplasmid HR in murine cells, plasmids pPYSV2neoDL and pPYSV2neoDR were coelectroporated into cells. Following a 48-h incubation period, the plasmid DNA was recovered (44) and used to transform DH10B electrocompetent bacteria as described above. Both plasmids harbor intact AMP genes and the mouse polyoma origin of replication, but each contains a different nonfunctional allele of the NEO gene (these alleles are in fact identical to those present in pSV2neoDR:DL(D)). Thus, as in the case above, the ratio of kanamycin-resistant to ampicillin-resistant bacterial colonies obtained provides a measure of the frequency of interplasmid HR. In some experiments, antibody (0.4 g) was co-electroporated into murine cells along with plasmid substrates for the indicated experiments.
Plasmid End Joining in Intact Cells-DNA end joining assays in intact cells were performed as described previously (32). Briefly, plasmid pSV2neo (which encodes the SV40 origin of replication) was linearized by treatment with the restriction enzyme EcoRI to produce cohesive DNA ends or SmaI to produce blunt ends, gel-purified, and quantitated. Five micrograms of plasmid pRSVEdl884 (which encodes the SV40 large T-antigen, but does not replicate itself) was co-electroporated into mammalian cells along with 1.25 g of linearized pSV2neo, and cells were incubated for 48 h. Plasmid DNA was subsequently recovered from the cells (44) and incubated with the restriction enzyme DpnI to digest plasmids that did not replicate in mammalian cells. This recovered plasmid DNA was then introduced into DH10B electrocompetent reporter bacteria and plated onto LB agar-containing Petri dishes. Percent end joining was determined by dividing the number of bacterial colonies obtained from mammalian cells electroporated with linearized pSV2neo by the number of bacterial colonies obtained from mammalian cells electroporated with circular pSV2neo in a parallel experiment.
Preparation of Protein Extracts-Nuclear protein extracts were prepared as previously described (31) and were used for Western blot analysis. Whole cell protein extracts were prepared by washing cells attached to culture dishes three times with ice-cold phosphate-buffered Tris, followed by incubation in 1 ml of ice-cold lysis buffer (125 mM Tris (pH 8.0), 375 mM sodium chloride, 0.25% SDS, 0.02% (w/v) sodium azide, 0.5% sodium deoxycholate, and 1 mM phenylmethylsulfonyl fluoride) for 15 min. Subsequently, the lysed cell solution was collected and centrifuged at 7000 ϫ g for 5 min at 4°C. Protein concentration of the supernatant was determined using the Bradford assay (45); these whole cell protein lysates were used for Western blot analysis.
Western Blot Analysis-Equal concentrations of protein extracts were electrophoretically resolved on SDS-polyacrylamide gels and transferred to nitrocellulose membranes. After a 1-h incubation in 5% bovine serum albumin in Tris-buffered saline, the membranes were probed with primary antibodies diluted in 5% bovine serum albumin in Tris-buffered saline for 1 h. Membranes were then washed three times with 0.1% bovine serum albumin in Tris-buffered saline and incubated with the appropriate second antibody for 1 h. This treatment was followed by three additional washes with 0.1% bovine serum albumin in Tris-buffered saline. Incubations with 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (Sigma) were then performed to detect antigens recognized by primary antibodies. Coomassie Blue-stained gels were prepared at the same time as all Western blots shown herein using the same concentrations of protein extracts.
Antibodies-Goat polyclonal anti-RAD51 antibody and rabbit polyclonal anti-AP endonuclease-1 antibody (REF-1) (Santa Cruz Biotechnology Inc.) were used 1:1000 and 1:2500 dilutions, respectively, for Western blot analysis. Human polyclonal anti-RAD51 antibody was kindly provided by Dr. Charles M. Radding (Yale University) and was used at a 1:2000 dilution for Western blot analysis. Rabbit polyclonal anti-RAD51 antibody and rabbit polyclonal anti-MRE11 antibody (Novus Biologicals Inc., Littleton, CO) were used at 1:2000 dilutions for Western blot analysis. Mouse monoclonal anti-MRE11 antibody (BD Biosciences) was used at a 1:500 dilution for Western blot analysis. Rabbit monoclonal anti-KU86 antibody (Serotec, Raleigh, NC) was used at a 1:500 dilution for Western blot analysis. The affinity-purified rabbit polyclonal antisera to human proteins FANCC and FANCD2 were kindly provided by Dr. Alan D. D'Andrea (Harvard Medical School) and were co-electroporated into intact cells. Rabbit polyclonal anti-RAD51B, anti-RAD51C, anti-RAD51D, anti-XRCC2, and anti-XRCC3 antibodies (Chemicon International, Inc., Temecula, CA) were used at 1:500 dilutions for Western blot analysis. Alkaline phosphataseconjugated goat anti-rabbit IgG and rabbit anti-goat IgG (Sigma) were used at 1:5000 dilutions for Western blot analysis.

FA Fibroblasts of Complementation Groups A, C, and G Have
Elevated Extrachromosomal HR Activity-It was previously shown that cells derived from two unrelated FA patients of unknown complementation groups had HR levels that were significantly greater than those seen in fibroblasts derived from normal subjects (40). To confirm and extend this observation, an intraplasmid HR assay (40) was used to determine the HR activity present in intact fibroblasts from FA patients from multiple known complementation groups. As shown in Fig. 1, FA fibroblasts from patients belonging to complementation groups A, C, and G all had significantly increased extrachromosomal HR frequency. The HR frequency observed was between 90 and 100 ϫ 10 Ϫ5 . In contrast, the diploid cell strain PD.792.F, derived from a normal subject, had an HR frequency of only 9 ϫ 10 Ϫ5 (Fig. 1). Similar examination of three other cell strains from normal subjects revealed HR frequencies of 9 ϫ 10 Ϫ5 , 10 ϫ 10 Ϫ5 , and 10 ϫ 10 Ϫ5 , respectively. These values were essentially identical to the HR frequencies of 9.5 ϫ 10 Ϫ5 and 9.9 ϫ 10 Ϫ5 previously reported in two different strains of diploid fibroblasts from unrelated normal donors (40). Thus, FA fibroblasts of complementation groups A, C, and G had a 9 -10-fold increase in HR frequency. Interestingly, examination of HR in fibroblasts of complementation group D2 revealed that they did not have elevated HR activity (Fig. 1). Instead, these cells displayed HR activity of ϳ9 ϫ 10 Ϫ5 , similar to that in all normal cells examined. The frequency with which the reporter bacteria catalyzed HR was determined to be Ͻ1 ϫ 10 Ϫ5 , indicating that the HR frequency results obtained reflect recombination events occurring in the mammalian somatic cells.
Because concerns over the influence of the SV40 large Tantigen in DNA repair processes have been raised (46), a second set of HR experiments were performed in which plasmid pRSVEdl884 was not co-electroporated into cells along with the recombination plasmid substrate. In these experiments, in which the SV40 large T-antigen was absent, the recombination plasmid substrate did not replicate. Nevertheless, we observed that HR frequency in diploid FA fibroblasts of complementation groups A, C, and G was elevated 8-, 9-, and 8-fold, respectively, above that observed in diploid fibroblasts from normal donors (data not shown).
Retrovirus-mediated gene transfer with the appropriate cDNAs has been shown to render FA cells resistant to bifunctional cross-linking agents (18,(22)(23)(24)(25)(26). Furthermore, this treatment has been shown to eliminate the sensitivity of FA cells to the cytotoxic effects of restriction enzyme-induced chromosomal DNA double-strand breaks (32). Therefore, we examined HR frequency in the retrovirally corrected counterparts of FA cells that displayed elevated HR activity in Fig. 1. As shown in Table I, retrovirally corrected FA fibroblasts had HR frequencies that were not statistically different from those seen in normal diploid fibroblast cells. These data therefore show that the elevation of HR activity in FA fibroblasts of complementation groups A, C, and G is reversible and is not likely to be the consequence of spontaneous immortalization or secondary mutations affecting HR.
RAD51 Protein Is Overexpressed in HA Fibroblasts Displaying Elevated HR Activity-RAD51 is the human homolog of the Escherichia coli recombination protein RecA (47), which functions in DNA double-strand break repair by searching for homologous sequences at sites of DNA double-strand breaks and facilitating strand pairing and exchange (37,38,48). Immortalized cells have been shown to have both elevated HR (49,50) and elevated RAD51 mRNA expression (51). These findings suggest that the elevated HR is due to the higher levels of expression of RAD51. This interpretation is supported by the finding that overexpression of RAD51 increases spontaneous plasmid HR activity in Chinese hamster ovary cells (52). We therefore investigated FA fibroblasts displaying elevated HR activity for a similar increase in RAD51 protein. Nuclear protein extracts were prepared from diploid fibroblasts from FA patients and from normal subjects. Protein concentration was determined, and equivalent amounts of total nuclear protein from these extracts were resolved by SDS gel electrophoresis and Western blot analysis performed using human polyclonal anti-RAD51 antibody (a generous gift from Dr. Charles M. Radding). Fig. 2A shows that RAD51 protein levels were elevated in extracts from group A, C, and G cells compared with an extract prepared from the normal cell strain PD.792.F. Fig.  2A also reveals that RAD51 levels were not elevated in an extract prepared from diploid FA fibroblasts of complementation group D2. It is noteworthy that normal and complementation group D2 fibroblasts both had relatively low levels of HR, whereas fibroblasts belonging to complementation groups A, C, and G had elevated levels of HR.
To confirm that the level of RAD51 present in the normal fibroblast cell extract is representative of that present in other normal diploid fibroblasts, we analyzed protein levels in nuclear extracts from seven other cell strains obtained from unrelated normal subjects. Fig. 2B shows that RAD51 protein levels in these extracts were similar, ranging from 0.9 to 1.3fold relative to that seen in the normal strain depicted in Fig.  2A. As a control, a nuclear extract from immortalized HT1080 cells was examined. Fig. 2B also shows that, as predicted, it had elevated RAD51 protein levels. To ensure that that the elevated levels of RAD51 expression detected in Fig. 2A were not due to unequal protein loading, we performed a control Western blot analysis using an antibody that specifically recognizes the AP endonuclease-1 protein. As shown in Fig. 2C, whereas there were slight well-to-well variations, each extract contained approximately the same amount of immunoreactive protein, consistent with our interpretation that the elevated levels of RAD51 detected in Fig. 2A are due to elevated protein expression in FA fibroblast cells.  To further confirm this conclusion, we examined the relative expression levels of RAD51 in whole cell lysates from control and FA fibroblasts. Equivalent amounts of protein from these extracts were resolved by electrophoresis, and RAD51 protein levels were determined. In these experiments, Western blot analysis was performed using a second anti-RAD51 antibody (Novus Biologicals Inc.). The results revealed that extracts from retrovirally corrected FA fibroblasts had RAD51 protein levels that were similar to those detected in extracts from normal and FA group D2 fibroblasts (Fig. 2D). Quantitation of RAD51 protein bands from multiple Western blots revealed that, whereas RAD51 levels were 8 -11-fold higher in extracts from FA fibroblasts from complementation groups A, C, and G, RAD51 protein levels in retrovirally corrected cells were not statistically different from those in normal or group D2 cell extracts. Additionally, Western blot analysis of these same extracts with a third anti-RAD51 antibody (Santa Cruz Biotechnology Inc.) revealed similar results, showing that RAD51 protein levels were elevated ϳ10-fold in whole cell protein lysates from FA fibroblasts of groups A, C, and G, whereas protein levels were relatively low in their retrovirally corrected counterparts and in normal and group D2 extracts (data not shown). Finally, to demonstrate that the different levels of RAD51 protein expression detected in Fig. 2D were not due to unequal protein loading, we performed Western blot analysis using anti-KU86 antibody. We (31) and others (30) have shown that KU86 protein levels are not elevated in FA cells compared with control cells. As shown in Fig. 2E, we observed that there was no significant difference in the relative expression levels of KU86 among these extracts, confirming that unequal protein loading is not responsible for the elevated levels of RAD51 detected in FA fibroblasts.
Subsequent to having observed that FA cells express elevated levels of RAD51, we became aware of a study by Digweed et al. (34), who reported that RAD51 expression levels are not elevated in FA fibroblasts from complementation groups A and G relative to those in normal or retrovirally corrected FA fibroblasts. The underlying source of this discrepancy was unclear especially since the Western blot analysis depicted in Fig. 2A was performed using the same antibody that Digweed et al. reported using (Novus Biologicals Inc.). However, due to the consistency with which we observed elevated RAD51 protein levels using three different anti-RAD51 antibodies, we were confident that our conclusion that FA cells of complementation groups A, C, and G possess elevated levels of RAD51 protein was correct. We nevertheless wished to uncover the source of this apparent discrepancy.
Upon careful examination, we noted that the conclusion of Digweed et al. (34) that RAD51 expression levels are not altered in FA cells was based upon comparing relative expression levels of RAD51 and MRE11. We therefore considered the possibility that MRE11 levels could conceivably differ between normal and FA fibroblasts. To test this hypothesis, we performed Western blot analysis on whole cell extracts from normal, FA, and retrovirally corrected FA fibroblasts using antibodies specific for RAD51 and MRE11 (both from Novus Biologicals Inc.). As shown in Fig. 3A, both MRE11 and RAD51 proteins were overexpressed in FA fibroblasts from complementation groups A, C, and G. Interestingly, MRE11 protein expression levels in retrovirally corrected FA cells were indistinguishable from those in normal cells. Similar results were obtained when Western blot analysis was performed on these same extracts with a second anti-MRE11 antibody (obtained from BD Biosciences) (data not shown). Again, Western blot  7)) and HT1080 cells (H). C, the same amounts of the same nuclear protein extracts from A were examined by Western blot analysis for AP endonuclease-1 protein expression. D, RAD51 Western blot analysis was performed on equal amounts of whole cell protein extracts prepared from normal diploid fibroblasts; patient-derived FA fibroblasts of complementation groups A, C, and G and their retrovirally corrected counterparts (A cor , C cor , and G cor , respectively); and patient-derived FA fibroblasts of complementation group D2. E, the same amounts of the same whole cell protein extracts from D were examined by Western blot analysis for KU86 protein expression.

FIG. 3. RAD51 and MRE11 protein levels are elevated in protein extracts from FA fibroblasts of complementation groups A, C, and G.
A, Western blot analysis of equal amounts of whole cell protein extracts from normal diploid fibroblasts (N); patient-derived FA fibroblasts of complementation groups A, C, and G and their retrovirally corrected counterparts (A cor , C cor , and G cor , respectively); and patient-derived FA fibroblasts of complementation group D2 was performed by probing the blot with both anti-RAD51 and anti-MRE11 antibodies. B, the same amounts of the same whole cell protein extracts from A were examined by Western blot analysis for AP endonuclease-1 expression.
analysis using anti-AP endonuclease-1 antibody ruled out the possibility that unequal protein loading was responsible for our finding (Fig. 3B). It thus appears that Digweed et al. (34) made an unfortunate choice in using MRE11 as a standard against which to determine relative RAD51 protein expression levels in FA and normal cells. Seen in light of the experiments depicted in Figs. 2 and 3, it is clear that our results and those of Digweed et al. are in complete agreement, i.e. the ratio of RAD51 to MRE11 protein expression does not vary between FA and normal fibroblasts. However, the critical point, which is clearly illustrated in Fig. 3, is that expression of both proteins is significantly elevated in FA fibroblasts belonging to complementation groups A, C, and G.
RAD51 and HR Are Elevated in FA Complementation Group C Models-The finding that overexpression of a patient-derived mutant FANCC gene (referred to as pL554P) rendered normal cells sensitive to both DNA cross-linking agents and induced chromosomal DNA double-strand breaks (32,42) prompted us to hypothesize that normal cells expressing this dominantnegative gene would have elevated RAD51 protein and HR activity. To test this hypothesis, Western blot analysis was performed on whole cell protein extracts from transgenic normal diploid fibroblasts overexpressing the pL554P FANCC gene and on extracts from unmodified normal cells. As shown in Fig. 4, RAD51 protein levels were 8 -9-fold higher in transgenic cell extracts compared with normal cell extracts. This increase in RAD51 protein was similar to that seen in FA fibroblast group C cell extracts (10 -11-fold) compared with their retrovirally corrected counterparts (Fig. 4A). HR frequency in transgenic normal diploid fibroblasts was also elevated compared with unmodified normal diploid cells (Fig. 4B). HR frequency in transgenic cells was increased 8-fold compared with unmodified normal cells. Control experiments showed that overexpression of the wild-type FANCC gene in normal diploid fibroblasts had no effect on RAD51 protein levels or HR frequency (data not shown).
A mouse model of FA with a mutation affecting the carboxyl terminus of the FANCC protein has been generated (41). Embryonic fibroblasts from these mice are hypersensitive to the cytotoxic effect of DNA cross-linking agents (41). We have found that these cells are also sensitive to DNA double-strand breaks induced by electroporation of bacterial restriction enzymes (32). Given these similarities to human FA cells, it seemed likely that murine FA cells would also have elevated RAD51 protein and HR activity. Western blot analysis of extracts from murine homozygous mutant FancC cells showed that the RAD51 protein was elevated compared with murine homozygous wild-type cell extracts (Fig. 5A). Quantitation re-vealed that the RAD51 protein was elevated 10 -11-fold. Fig.  5B shows that HR frequency in murine mutant FancC cells was also elevated ϳ9-fold compared with wild-type cells.
Protein Expression of RAD51 Paralogs Is Not Altered in FA Fibroblasts-Several RAD51 paralogs (RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3) have been identified and shown to function in cellular HR (53)(54)(55)(56)(57). These paralogs have also been suggested to play roles in DNA cross-link repair and maintenance of genomic stability (58 -60). Given the finding that FA cells of complementation groups A, C, and G have both elevated RAD51 protein and elevated HR activity and the inherent genomic instability and sensitivity FA cells display to DNA cross-linking agents, we hypothesized that FA fibroblasts may also have altered protein expression of one or more of these paralogs. Western blot analysis of whole cell protein extracts from FA and retrovirally corrected FA fibroblasts was performed with antibodies specific for the RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3 proteins. As shown in Fig. 6, no discernible difference in the levels of any of these proteins was seen in extracts from FANCC or retrovirally corrected FANCC fibroblasts. Similar results were obtained from extracts from other FA complementation groups (data not shown). These results show that, contrary to the expression of RAD51 and MRE11 proteins, there is no evident change in expression of FIG. 4. Normal diploid fibroblasts expressing a dominant-negative FANCC gene have elevated RAD51 protein and high HR activity. A, RAD51 protein (RAD51p) levels were examined by Western blot analysis of equal amounts of whole cell protein extracts prepared from normal diploid fibroblasts that were unmodified (N) or that overexpress the pL554P FANCC gene (NϩpL554P) and from FA fibroblasts from patients of complementation group C and their retrovirally corrected counterparts (C cor ). Protein levels were quantitated in three independent experiments and are expressed as the -fold increase versus unmodified normal diploid extracts. Error bars represent S.E. *, p Ͻ 0.05. B, HR frequency was examined in these same cells. Results are an average of at least three independent experiments. Error bars represent S.E. *, p Ͻ 0.05. these five proteins in FA fibroblast extracts. This is similar to data in previous reports showing that expression of four other proteins (KU86, KU70, XRCC4, and DNA ligase IV) in cell extracts from these cells is unaltered (30,31). This provides further evidence that the results presented in Figs. 2 and 3 showing elevated levels of RAD51 protein expression cannot be due to unequal protein loading.

Introduction of Antibodies Specific for RAD51 Results in Altered HR Activity in FA Fibroblasts of Complementation
Groups A, C, and G-It has previously been shown that overexpression of the RAD51 protein in mammalian cells results in a stimulation of chromosomal HR (52). This finding, along with the data presented above, suggests that the elevated HR activity seen in FA fibroblast of complementation groups A, C, and G is due to overexpressed RAD51 protein. To test this hypothesis, we examined the ability of an antibody specific for the RAD51 protein to attenuate HR activity in these FA cells. Antibodies have previously been shown to be effective at inhibiting protein function when introduced directly into cells by electroporation (61). Therefore, we examined the effect of cointroduced polyclonal anti-RAD51 primary antibody (Novus Biologicals Inc.) on intracellular HR activity in diploid FA fibroblasts. As shown in Table II, the presence of anti-RAD51 antibody reduced HR frequencies in FA fibroblasts of complementation groups A, C, and G to levels that were not statistically different from those seen in normal fibroblasts not treated with antibody. In contrast, the presence of anti-RAD51 antibody had no effect on HR frequencies in normal diploid or complementation group D2 fibroblasts (Table II). As a control, goat anti-rabbit IgG was co-introduced into normal and FA fibroblasts along with recombination plasmid substrates. As expected, no effect on HR was observed following this treatment (Table II). Similar to human FA cells, treatment of murine homozygous mutant FA cells with anti-RAD51 antibody reduced HR frequency to wild-type levels, whereas antibody treatment of murine wild-type cells had no effect (data not shown).
Effects of Introduction of Anti-FA Antibodies on Cellular Double-strand Break Repair-The data presented above indicate that the absence of functional FANCC protein results in elevated expression of the RAD51 protein, thereby leading to elevated intraplasmid HR activity. To confirm this result, we performed an HR experiment in which anti-FANCC antibody was co-electroporated into intact normal diploid fibroblast cells along with the recombination plasmid substrate. As shown in Fig. 7, normal fibroblasts treated with anti-FANCC antibody had significantly increased HR activity. A Ͼ8-fold increase in HR frequency was observed in these cells versus normal cells not treated with antibody. Additionally, anti-FANCC antibody introduction into both FA fibroblasts of complementation group D2 and retrovirally corrected FA fibroblasts of complementation group C resulted in a similar increase in HR frequency compared with cells not treated with antibody (Fig. 7). As expected, anti-FANCC antibody did not influence HR frequency in FA fibroblasts of complementation groups A, C, and G.
FA fibroblasts of complementation groups A, C, D2 and G have all been shown to have significantly decreased DNA end joining activity (32), whereas the results presented thus far indicate that only FA fibroblasts of complementation groups A, C, and G have elevated HR activity. To confirm this finding and to gain insight into the relationship between FA proteins and DNA double-strand break repair pathways, we examined both recombinational and non-recombinational DNA repair in the presence of either anti-FANCC or anti-FANCD2 antibody (generous gifts from Dr. Alan D. D'Andrea). We hypothesized that if FANCC and FANCD2 function in different capacities in the regulation of these DNA double-strand break repair pathways, specific inhibition of either of these proteins in normal cells would result in different DNA repair phenotypes. Intact normal diploid cells were therefore co-electroporated with anti-FANCC or anti-FANCD2 antibody along with plasmid substrates for either HR repair or DNA end joining. Table III FIG. 6. Protein levels of the RAD51 paralogs are not altered in FA fibroblast extracts. Equal amounts of whole cell protein extracts prepared from FANCC fibroblasts (C) and retrovirally corrected FANCC fibroblasts (C cor ) were examined by Western blot analysis using polyclonal anti-RAD51B, anti-RAD51C, anti-RAD51D, anti-XRCC2, and anti-XRCC3 antibodies.  shows that, as a control, normal fibroblasts treated with anti-IgG antibody had both HR frequencies and DNA end joining activities that were similar to those in cells not treated with antibody. Conversely, normal cells co-electroporated with anti-FANCC antibody displayed a concurrent increase in HR activity and decrease in DNA end joining activity (Table III). These alterations in both DNA repair pathways were of similar magnitude to those seen in patient-derived FA fibroblasts of complementation group C. Additionally, normal cells co-electroporated with anti-FANCD2 antibody had DNA end joining activities that were significantly decreased, whereas no change in HR activity was observed (Table III). DISCUSSION The results presented above indicate that human fibroblasts belonging to FA complementation groups A, C, and G have dramatically elevated levels of intraplasmid HR activity. These cells also display elevated levels of RAD51 protein, and results obtained from electroporation of anti-RAD51 antibody into these cells indicate that the elevated HR activity is RAD51-dependent. This elevated RAD51-dependent HR activity clearly results from the absence of the respective FA proteins since retrovirus-mediated gene correction restored HR activity to wild-type levels. Murine embryo fibroblasts from a FancC knockout mouse model also expressed elevated levels of RAD51 protein as well as elevated levels of interplasmid HR activity. In contrast, human FA fibroblasts belonging to complementation group D2 had neither elevated RAD51 protein expression nor elevated HR activity. We observed that anti-FANCC antibody introduced into normal fibroblasts led to elevated levels of HR activity, whereas similar treatment with anti-FANCD2 antibody was without effect on intracellular HR activity. Further analysis of these cells revealed that, in both cases, antibody treatment sensitized the cells to the cytotoxic effects of bifunctional cross-linking agents, proving that introduction of the antibodies inhibited the FANCC and FANCD2 proteins, respectively. Thus, we can conclude that a subset of FA proteins, minimally including the FANCA, FANCC, and FANCG proteins, but exclusive of the FANCD2 protein, function to suppress HR in fibroblasts.
In addition to the effects summarized above, electroporation with either the anti-FANCC or anti-FANCD2 antibody inhibited cellular DNA end joining activity, which was previously shown to be deficient in FA fibroblasts belonging to complementation groups A, C, D, and G (32). The finding that both end joining and HR are aberrant in FA fibroblasts raises the possibility that the two pathways are coordinately regulated by the FA proteins. It is clear that both end joining and HR pathways can, in principle, be utilized to repair similar types of DNA damage, including double-strand breaks and interstrand cross-links. Clearly, however, the ramifications for the cell of choosing one particular pathway over the other are significant. Although DNA end joining is frequently imprecise, it is less likely to generate chromosome translocation events compared with HR (62). On the other hand, HR is more likely to precisely repair a lesion, particularly if the lesion involves an interstrand cross-link (63,64). However, the existence of repetitive DNA provides a greater opportunity for potentially oncogenic translocation events to occur during HR. It is thus reasonable to propose that mammalian cells coordinately regulate these distinct DNA repair pathways, in effect "deciding" which pathway is utilized to repair a specific lesion. A number of observations are consistent with the idea that the FA proteins perform this role. First, although it has long been believed that FA cells are defective in DNA repair, the absence of recognizable sequence motifs within the FA proteins has cast doubt on the idea that they participate directly in DNA repair. Second, it has been difficult to understand why, if the FA proteins are essential players in DNA repair, they are not present in lower eukaryotes. Third, the FANCD2 protein interacts with a number of DNA repair and regulatory proteins and relocates within the nucleus of cells following induced DNA damage (27,35).
Based on the finding that FA group A, C, and G fibroblasts, but not D2 cells, have elevated HR, we propose that a signal transduced by the FANCA, FANCC, and FANCG proteins suppresses HR. These FA proteins are known to exist within the cell in a complex that also contains the FANCE and FANCF proteins (65)(66)(67), and the complex is absent from cells lacking any of these proteins. We therefore propose that this intact complex of FA proteins generates a signal that suppresses HR activity. The FANCD2 protein cannot play an essential role in this suppression since HR levels are not aberrantly elevated in FANCD2 cells. However, the FANCD2 protein nevertheless does play a critical role in regulating cellular DNA doublestrand break repair. We propose that monoubiquitination of this protein, which requires the presence of the intact FA protein complex and which occurs in response to a variety of triggers including induced DNA damage and cell cycle progression (9,27), generates a separate signal that overcomes the suppression of HR mediated by the FA protein complex, thereby activating cellular HR. A role in activating cellular HR is consistent with the fact that the FANCD2 protein localizes to discrete loci within the nucleus in response to DNA damage  and during S phase of the cell cycle (68). The recombination protein RAD51 also localizes to these loci under these conditions, as does the BRCA1 protein, which is thought to play an important role in regulating cellular HR (27,68). It is somewhat paradoxical that elevated levels of the recombinational repair protein RAD51 are associated with enhanced sensitivity to DNA cross-linking agents in FA cells. It is unclear at this time whether the enhanced sensitivity of these cells results from the elevated HR activity or is instead due to the deficient DNA end joining activity within these cells. We do not favor the former explanation, particularly in light of reports demonstrating increased radioresistance in RAD51-overexpressing cell lines (52). Rather, the latter explanation is supported by a recent report showing that overexpression of the human RAD51 protein is associated with reduced levels of doublestrand break-induced chromosomal HR (69). The authors speculated that imbalanced levels of HR proteins may interfere with the proper functioning of these proteins during doublestrand break repair. Clearly, additional experimentation will be required to resolve this issue.
Our previous results combined with the antibody inhibition data presented herein indicate that the FANCA, FANCC, FANCD, and FANCG proteins are required for normal DNA end joining activity. We propose that the FANCA, FANCC, and FANCG proteins function in this capacity with the essential cooperation of the other members of the FA complex, i.e. FANCE and FANCF. Therefore, an additional mechanism through which the FA proteins function to regulate cellular DNA double-strand break repair is to activate DNA end joining. A schematic representation of this model is presented in Fig. 8, where the arrows indicate activation or suppression of the end joining or HR pathway, respectively, in response to signals received from the FA proteins.
It is presently not clear precisely which enzymatic machinery is activated or repressed by the FA proteins. However, the BRCA1-associated complex, a large complex of DNA repair enzymes containing, among others, BRCA1 and the RAD50-MRE11-NBS1 complex (70), is an attractive candidate. The RAD50, MRE11, and NBS1 proteins are of particular interest since the available evidence suggests that they can participate in both recombinational and non-recombinational repair of DNA double-strand breaks (62,71). It is therefore tempting to speculate that the RAD50-MRE11-NBS1 complex is the target for the signal transduction events we have described above. The very recent findings that the FANCD2 and NBS1 proteins co-localize following treatment with a DNA cross-linking agent (72) and that FANCC is minimally needed for proper MRE11 focus formation (73) are consistent with this hypothesis.