Human Fanconi anemia monoubiquitination pathway promotes homologous DNA repair

Materials and Methods

Cell Lines. SV40-transformed FA fibroblasts (GM6914 FA-A, EUFA326 FA-G, and PD20 FA-D2) were grown in DMEM supplemented with 15% FBS. The previously described DR-GFP reporter (21, 22) was modified to contain a hygromycin resistance marker in place of a puromycin resistance marker, and then the vector DRGFPhygro was integrated into the genome of PD20 cells and hprtDRGFPhygro was integrated into the genome of GM6914 and EUFA326 cells. Clones with a single copy of DR-GFP were identified by Southern blotting. These clones were then used for infection with either an “empty” pMMP-puro retrovirus (vector) or a pMMP-puro retrovirus into which a FANC cDNA was inserted. Infected cells were selected in puromycin (generally 2–3 weeks) before subsequent manipulation. FANC expression was verified by Western blotting, immunofluorescence, and resistance to mitomycin C, as described (23). Two independent clones (C8 and H1) of mouse ES cells containing a targeted mutation in Fanca were constructed by sequentially deleting exons 37–39 (Y.-G.Y., M.D., and Z.-Q.W., unpublished results). The Fanca^–/– cells, which contain no detectable Fanca protein, were subsequently targeted at the hprt locus with DR-GFP (22).

GFP Assays. To measure the repair of an I-SceI-generated DSB, 30 μg of the I-SceI expression vector pCBASce (21) or an empty vector was mixed with 5 × 10⁶ cells suspended in 650 μl of opti-MEM medium (Invitrogen) in a 0.4-cm cuvette, followed by pulsing the cells at 270 V, 975 μF. To specifically determine the amount of HDR, the percentage of GFP-positive cells was quantitated by flow cytometric analysis 3 days after electroporation on a Becton Dickinson FACScan, as described (21).

PCR Assays. To determine the percentage of I-SceI site loss for each electroporation, genomic DNA was isolated 7 days after transfection. Genomic DNA (0.4 μg) was used as the template for PCR with primers in a reaction volume of 50 μl. The sequences of primers were as follows: DRGFP-F, CTGCTAACCATGTTCATGCC; DRGFP-R, AAGTCGTGCTGCTTCATGTG. PCRs were performed by using the PCR SuperMix (Invitrogen) with Mastercycler (Eppendorf). Amplification was done for 27 cycles with a 1-min elongation time. After amplification, PCR products were digested overnight with 10 units of I-SceI (Roche Molecular Biochemicals). After I-SceI digestion, products were purified by using GFX PCR DNA and Gel Band Purification Kit (Amersham Pharmacia Biosciences), and then half volume of products were digested with BcgI (NEB, Beverly, MA). The products digested with I-SceI and both I-SceI and BcgI were separated on a 1.2% agarose gel. The gel was stained with ethidium bromide, and the ethidium signals for the enzyme-resistant and enzyme-cleaved band were quantified by using nih image software.

To investigate SSA in the FA cell lines, we used a combined PCR-Southern blotting method. After electroporation, genomic DNA was used as PCR template for two pairs of PCR primers, either SA-F and SA-R1, or SA-F and SA-R2 (SA-F, TTTGGCAAAGAATTCAGATCC; SA-R1, CAAATGTGGTATGGCTGATTATG; SA-R2, ATGACCATGATTACGCCAAG. Amplification was for 20 cycles, which was determined to be in the linear range. After running the products on an 0.8% gel, DNA was transferred to nylon membrane and probed with ³²P-labeled fragment from DR-GFP, which was cut with Hin-dIII/BamHI.

Results

Upstream FANC Pathway Components Promote HDR. To determine whether the FA pathway has a role in HDR in human cells, we established patient-derived human cell lines from the FA-A (GM6914) and FA-G (EUFA326) complementation groups with an integrated copy of the DR-GFP reporter (21) (Fig. 1a). With this reporter, a DSB is introduced into the chromosome by expressing the I-SceI endonuclease, and, if HDR occurs, GFP is expressed, which is quantifiable by flow cytometry. After establishment of cell lines carrying the DR-GFP reporter, a retrovirus expression system was used to complement the specific FANC defect in the cells. Complementation was verified by Western blot analysis (data not shown) and increased survival after mitomycin C treatment (Fig. 6 a and b, which is published as supporting information on the PNAS web site), when cells infected with the FANC-expressing virus were compared with cells infected with a vector control virus.

In the absence of I-SceI expression, very few GFP⁺ cells were detected in any of the cell lines (Fig. 1 b and c). Transfection with the I-SceI expression vector into the uncomplemented GM6914 and EUFA326 cells resulted in ≈2.5–3% GFP⁺ cells (Fig. 1 b and c). By contrast, I-SceI expression in cell lines complemented with each wild-type cDNA resulted in an increased number of GFP⁺ cells (Fig. 1 b and c), i.e., a 2.7- and 2.2-fold increase, respectively (Fig. 1d). Transfection with an intact GFP expression vector gave a similar number of GFP⁺ cells for each of the cell lines, indicating that the increased number of GFP⁺ cells after I-SceI expression was not due to an increased transfection efficiency. These results imply that both the GM6914 and EUFA326 cell lines are deficient in the precise HDR pathway, indicating a role for both FANCA and FANCG in this pathway.

To rule out that the increased HDR was due to nonspecific effects rather that complementation of the specific FANC mutation, we also infected wild-type and noncognate FA cells with FANC-expressing retroviruses. For wild-type cells, either HEK-293 or U2OS cells were infected with FANCA- or FANCG-expressing retroviruses, respectively. No increase in HDR was observed in either cell line (Fig. 7a, which is published as supporting information on the PNAS web site, and data not shown), indicating that infection with the FANC-expressing retroviruses does not itself increase HDR. In addition, GM6914 cells were infected with the complementing FANCA-expressing retrovirus and the noncognate FANCG-expressing retrovirus. Infection with the noncognate FANCG-expressing retrovirus had no effect on HDR, whereas, as before, the FANCA-expressing retrovirus promoted HDR (Fig. 7b). Thus, the increase in HDR in the patient-derived FA cell lines upon infection with the cognate retrovirus is due to complementation of the FANC defect in the cells.

We also tested HDR in mouse ES cells containing a targeted mutation in Fanca (Y.-G.Y., M.D., and Z.-Q.W., unpublished results). Similar to the patient-derived FA-A cells, HDR is reduced in the Fanca^–/– cells (Fig. 1e); comparing the two independently targeted Fanca^–/– cell lines (C8 and H1) with the parental wild-type cell line, the reduction is 2-fold (2.8% vs. 5.5%, respectively). Thus, a mild reduction in HDR in cells with disrupted upstream FANC components is observed for both transformed human cell lines and mouse cell lines containing a newly derived mutation.

Nonhomologous End-Joining (NHEJ) Levels Are Not Altered in the FA Cell Lines. To determine whether overall DSB repair was substantially reduced in the patient-derived cell lines, we physically analyzed DSB repair products using a PCR assay. For this assay, we amplified the genomic region surrounding the I-SceI site after expression of I-SceI endonuclease with PCR primers that would amplify products from several types of repair: HDR, SSA, and NHEJ (Fig. 2a). To determine how much of the amplified fragment consisted of detectable DSB repair products, we cleaved the fragment with I-SceI, and, to specifically detect NHEJ products, we cleaved with I-SceI and with BcgI, which cleaves only the homologous repair product.

In cells that were not transfected with the I-SceI expression vector, the amplified product is completely cleaved by I-SceI (Fig. 2b) whereas, in cells that expressed I-SceI, a substantial portion of the amplified product was uncleaved (Fig. 2b), indicating in vivo cleavage by I-SceI followed by DSB repair, which results in I-SceI site loss (Fig. 2a). For the uncomplemented GM6914 and EUFA326 cells, I-SceI site loss was ≈30–35% of the amplified fragment, indicating that the chromosomal I-SceI site was cleaved and repaired in a substantial fraction of the cells (Fig. 2 b and c). Interestingly, in the FANCA- and FANCG-complemented cell lines, I-SceI site loss was ≈10% higher (Fig. 2 b and c), which is suggestive of enhanced DSB repair by one or more pathways.

To differentiate NHEJ from homologous repair, the amplified products were additionally treated with BcgI, because HDR and SSA products are both cleavable by this enzyme but NHEJ products are not (Fig. 2a). Thus, the amount of uncleaved DNA after digestion with both BcgI and I-SceI indicates the level of NHEJ. In all of the cell lines, whether uncomplemented or FANC-complemented, the BcgI/I-SceI noncleavable product was 30–35%. This result clearly demonstrates that the level of repair of a chromosomal DSB by NHEJ is not reduced by either FANCA or FANCG deficiency.

SSA Is Reduced in the FA Cell Lines. We noted in these experiments that the amount of the BcgI/I-SceI cleavable product differed in level in the uncomplemented versus complemented cells to a greater degree than expected from a decrease in the HDR product alone. This finding suggested that the other BcgI-cleavable product, i.e., deletions arising from SSA, may also be reduced by FANC deficiency. To examine SSA levels, we used a primer set that would specifically amplify the deletion product derived from SSA (Fig. 3a). This same deletion product would be obtained from an HDR pathway involving crossing over; however, crossovers have been measured in several systems to be rare outcomes of HDR, i.e., <2% of HDR events, so a significant contribution of crossovers to deletion events is unlikely (24–26). We also ruled out that long-tract gene conversions coupled to NHEJ contribute significantly to the PCR product (Fig. 8, which is published as supporting information on the PNAS web site).

We compared the level of the PCR product arising from SSA with the level of the PCR product obtained from a structurally intact DR-GFP reporter (Fig. 3a). Although the amount of this control product was similar in all of the cell lines (Fig. 3b), the FANCA-complemented GFM6914 and FANCG-complemented EUFA326 cells both demonstrated an ≈2-fold higher level of the SSA product relative to their uncomplemented counterparts (Fig. 3 b and c). This result suggests that these FANC proteins participate in both the HDR and SSA homologous repair pathways. Interestingly, this finding contrasts with a BRCA2 mutant cell line, in which SSA is increased rather than reduced (Fig. 3 b and c), even though HDR is also decreased in BRCA2-deficient cells (9).

The Downstream FA Component, FANCD2, Participates in Homologous Repair: Role for FANCD2 Monoubiquitination. The FANC complex consists of the A, B, C, E, F, G, and L proteins, which apparently act upstream of FANCD2 (15, 27) and are necessary for FANCD2 monoubiquitination (28). To determine whether the results we obtained with FANC complex members are relevant to FANCD2 function, we performed similar experiments in the PD20 cell line, the reference FA-D2 patient-derived cell line (29). The baseline level of HDR in the uncomplemented PD20 cell line was higher than that obtained in either of the other FANC-deficient lines; however, complementation by expression of FANCD2 led to a similar 2-fold increased level of HDR (Fig. 4a). These results indicate a role for both upstream (FANCA, G) and downstream (FANCD2) FA pathway components in HDR.

FANCD2 is monoubiquitinated on lysine 561 in response to DNA damage and during S phase, and monoubiquitination is necessary for normal levels of resistance to DNA damaging agents (28) (Fig. 6c). Moreover, FANCD2 is phosphorylated by the ATM kinase on serine 222, which is responsible for the S-phase checkpoint after DNA damage but is not essential for DNA repair (23). We tested the role of these modifications on FANCD2 function in the HDR assay. We found that an intact FANCD2 monoubiquitination site was necessary to promote HDR, but that an intact ATM phosphorylation site was not (Fig. 4a).

We also performed the I-SceI site-loss assay in the various PD20 derivatives and found that levels of the NHEJ product were similar in all cell lines (Fig. 4b), indicating that, as with the upstream components of the FA pathway, NHEJ levels are not affected by the FANCD2 status of the cells. Finally, we also examined SSA in the PD20 cells and found that it was increased nearly 2-fold when cells were complemented by FANCD2 (Fig. 4 c and d). As with HDR, the complementation depended upon an intact monoubiquitination site, because the FANCD2-K561R mutant showed similar levels of SSA as the vector control. Like wild-type FANCD2, SSA was increased by expression of FANCD2-S222A, indicating that, as for HDR, ATM phosphorylation at S222 is not essential for normal levels of homologous repair, such that its role may be limited to S-phase checkpoint control (Fig. 4d) (23).

Discussion

In this article, we determined that upstream and downstream components of the FANC pathway promote HDR in human cells (Fig. 5). An intact monoubiquitination site in FAND2 is required to promote HDR, although the FANCD2-S222 ATM phosphorylation site is not, which parallels the requirement for normal levels of resistance to crosslinking agents (23). Importantly, the level of HDR impairment in FA patient-derived cell lines is mild, especially when compared with the severe recombination defects found in BRCA1 (30), BRCA2 (9), and RAD51 mutants (6), or even RAD51 paralog mutants (21, 31). We obtained a similarly mild defect in HDR in mouse ES cells containing a targeted mutation in Fanca, suggesting this is a general feature of mammalian FANC mutants. Thus, our results imply that, although FANC proteins promote HDR, they are not essential components of the HDR machinery. It still remains possible, however, that they have a crucial role in the repair of a particular subset of DNA lesions.

This mild HDR impairment found with FANC deficiency is consistent with the viability of FA patients and mice, but distinct from the more severe defect observed in fancg mutant chicken DT40 cells (19). Although the recombinogenic DT40 B cells may magnify repair defects that are substantially milder in mammalian cells, it should be noted that fancc mutant DT40 cells also show only a mild HDR defect (20). Nevertheless, the mild impairment could certainly be causative for the increased frequencies of chromosome aberrations, spontaneous HPRT deletions (32), and loss of heterozygosity (33) observed in FA cells, as well as for the tumor predisposition of patients (14, 15). Mice with weak hypomorphic alleles of Brca1 (8) and Brca2 (9) have only small reductions in HDR (i.e., 5-fold), yet they manifest chromosomal instability, developmental defects, and cancer predispositon (34–36).

Unlike HDR, we observed that FA cells have levels of NHEJ that are comparable with those of complemented cells. Preliminary analysis of NHEJ junctions in FA-A cells also indicates that the junction sequences are comparable (K.N. and M.J., unpublished results). Previous reports have been contradictory about a role for FANC proteins in NHEJ. One group reported that NHEJ of a plasmid DSB containing cohesive overhangs, which are also present after I-SceI cleavage, is not impaired in FA cells (37), although another group reported an NHEJ defect in FA cells using a similar plasmid assay (38). It is not clear what accounts for the different results obtained in these plasmid assays. However, because plasmids are susceptible to degradation and other processing events (39), the chromosomal assays using I-SceI may better reflect DSB repair.

Unexpectedly, we observed that the FA patient-derived cell lines from the three FA groups tested have a mild defect in another DSB repair pathway, SSA. By contrast to what we observed in the FA cells, mutations in BRCA2, which has been considered to be an FANC gene (40), increase levels of SSA in cells (6, 10). The results obtained with the FA cells, however, parallel those reported recently when BRCA1 is disrupted, in which both homologous repair pathways are disrupted (6) (Fig. 5). They also oppose the results obtained with the NHEJ mutant Ku70, in which both HDR and SSA are elevated (6). Although the biological relevance of SSA is uncertain, the effect of HDR in relation to SSA in mutants provides insight into the role of homologous repair pathway components. Thus, whereas RAD51 and BRCA2 are both key components of the strand invasion steps of HDR (11, 41), BRCA1 and the FANC proteins may regulates step(s) common to both HDR and SSA pathways; Ku70 may oppose this step.

One step common to HDR and SSA is DNA strand resection, which generates the single strands important for strand invasion and strand annealing, respectively. An effect on resection could be direct or indirect, for example, by antagonizing Ku (6, 22, 42). Alternatively, the FANC proteins may affect some other step common to the homologous repair pathways. It is interesting to note that the chicken FANCC has been implicated recently in an additional DNA repair pathway, that of translesion synthesis (20). This observation, together with our results, may again indicate a role for the FANC proteins at a common step in DNA damage repair pathways, perhaps in processing or stabilizing intermediates for multiple pathways, including homologous repair and translesion synthesis. Further speculation will require confirmation of a role for the mammalian proteins in translesion synthesis.

In summary, our results provide evidence that the FA pathway promotes homologous repair in mammalian cells. The consistently mild defect we observed in FANC mutants implies a role for these components that is distinct from central HDR components, yet which may be sufficient to account for the chromosomal abnormalities observed in FA cells.

Supplementary Material