Report

Ku70 Corrupts DNA Repair in the Absence of the Fanconi Anemia Pathway

See allHide authors and affiliations

Science  09 Jul 2010:
Vol. 329, Issue 5988, pp. 219-223
DOI: 10.1126/science.1192277

Righting Repair Pathways

The genetic disease Fanconi anemia (FA) results from mutations in a series of genes involved in a DNA repair pathway that helps process the damage caused by erroneous chemical cross-links between the two strands of the DNA double helix. The double-stranded breaks in DNA that arise from such cross-links can be repaired in an error-free manner or through an error-prone repair pathway. Pace et al. (p. 219, published online 10 June) show that the FA pathway can drive repair through the error-free pathway. The FA FANCC gene shows a genetic interaction with a component of the error-prone repair pathway, Ku70, inhibiting its action and thereby promoting the error-free pathway.

Abstract

A conserved DNA repair response is defective in the human genetic illness Fanconi anemia (FA). Mutation of some FA genes impairs homologous recombination and error-prone DNA repair, rendering FA cells sensitive to DNA cross-linking agents. We found a genetic interaction between the FA gene FANCC and the nonhomologous end joining (NHEJ) factor Ku70. Disruption of both FANCC and Ku70 suppresses sensitivity to cross-linking agents, diminishes chromosome breaks, and reverses defective homologous recombination. Ku70 binds directly to free DNA ends, committing them to NHEJ repair. We show that purified FANCD2, a downstream effector of the FA pathway, might antagonize Ku70 activity by modifying such DNA substrates. These results reveal a function for the FA pathway in processing DNA ends, thereby diverting double-strand break repair away from abortive NHEJ and toward homologous recombination.

Fanconi anemia (FA) proteins are components of a conserved metazoan DNA repair pathway that facilitates homologous recombination (HR) and DNA translesion synthesis (TLS) (1, 2). Genetic and biochemical evidence point to an upstream role for the FA pathway in interstrand cross-link repair during replication (3, 4). Consequently, FA knockout cell lines accumulate large numbers of chromatid breaks after exposure to cross-linking agents, indicating a crucial role for the FA proteins in resolving DNA double-strand breaks (DSBs) created at cross-links. DSBs are preferentially repaired by nonhomologous end joining (NHEJ) in the G1 phase of the cell cycle, and by HR during replication (5). Although the selective use of either pathway shows clear associations with different phases of the cell cycle, little is known about the regulation of this specificity in cross-link repair.

The FA pathway has been implicated in the repair of some DSBs by HR but also in NHEJ fidelity (6, 7). These apparent contradictions led us to examine the nature of the relationship between the FA pathway and NHEJ. We compared DT40 chicken B cells with combined disruption of the FA nuclear complex gene FANCC and the NHEJ genes Ku70 (∆Ku70), DNA-PKCS (∆DNA-PKCS), or Ligase IV (∆Lig4) (8). Strains were tested for their sensitivity to the DNA cross-linking agent cisplatin (Fig. 1, A to C) and to x-rays (fig. S1, B and C). ∆FANCC cells were very sensitive to cross-links but not to x-rays, in contrast to ∆DNA-PKCS or ∆Ku70 strains. However, combined ablation of FANCC with Ku70, but not with DNA-PKCS or Ligase IV, caused suppression of the cisplatin sensitivity seen in the ∆FANCC strain [Fig. 1A; dose resulting in 37% population survival (D37) values: ∆FANCC = 45 nM, ∆Ku70∆FANCC = 95 nM; dose resulting in 10% population survival (D10) values: ∆FANCC = 90 nM, ∆Ku70∆FANCC = 400 nM]. The ∆Ku70∆FANCC cells were complemented with Ku70 cDNA and the marked sensitivity to cisplatin was restored (Fig. 1, D and E; Ku70-complemented ∆Ku70FANCC lines ∆Ku70∆C+Ku70-1 and ∆Ku70∆C+Ku70-2; D37 = 15 nM, D10 = 25 nM). To extend our finding to other vertebrates, we knocked down Ku70/80 in human FANCC-deficient cells, facilitating cross-linker resistance (fig. S1, D and E; D37 values: mock siRNA = 3 μM, Ku80 siRNA = 6 μM).

Fig. 1

Suppression of cross-linker sensitivity and chromosome breakage in ∆Ku70FANCC DT40 cells. (A to C) Cisplatin sensitivity of NHEJ DT40 mutants (∆Ku70, ∆DNAPKcs, ∆Lig4) combined with FANCC (∆FANCC) disruption, analyzed by modified proliferation assay (3). Each point represents the mean from three independent experiments ± SE. (D) Cisplatin sensitivity of ∆Ku70FANCC complemented with Ku70 cDNA (∆Ku70∆C+Ku70-1 ∆Ku70∆C+Ku70-2). (E) Immunoblot analysis for Ku70 expression in the indicated cell lines. (F) Chromosome breakage analysis of strains after exposure to a single dose of the cross-linking agent mitomycin C (MMC). Log-phase cultures were exposed to MMC (50 ng/ml) for 12 hours; bars represent means ± SE.

After exposure to cross-linking agents, ∆FANCC cells display higher levels of chromosome breakage than wild-type DT40 cells (3). Because Ku70 appears to be required for increased cellular sensitivity to cross-linkers in ∆FANCC cells, we investigated whether Ku70 ablation affects levels of chromosome breakage. Exposure to a single dose of the cross-linking agent mitomycin C (MMC) caused much less chromosome breakage in ∆Ku70FANCC cells than observed in the FANCC strain (Fig. 1F).

DT40 cells express immunoglobulin M on their surface (sIgM). The variable (V) loci of IgM undergo constant diversification by gene conversion (HR) or point mutation (TLS) (9, 10); both processes are impaired in the FANCC strain (3). We therefore used sIgM fluctuation analysis to determine whether Ku70 contributes toward defective gene conversions and point mutations in the FANCC strain (fig. S2A). The prevalence of sIgM loss was reduced in the FANCC cells relative to the Ku70 and DT40 strains, yet this defect was largely reversed in the Ku70FANCC double-deficient strain (Fig. 2A). It is noteworthy that sIgM negative frequency in Ku70FANCC cells is lower than that reported for HR-deficient XRCC2 strains (in which increased TLS accounts for locus diversification) (9, 10). To confirm that gene conversions, and not point mutations, are rescued at the IgM locus in Ku70FANCC cells, we used fluorescence-activated cell sorting (FACS) to separate the sIgM-negative populations from multiple clones and then amplified and sequenced their V gene loci. Gene conversion and point mutation events were identified and their frequency compared to our published database for DT40 and the FANCC strain. These data clearly show that the number of gene conversion tracts per individual sequence was increased in the ∆Ku70∆FANCC strain relative to the FANCC strain (Fig. 2B). In contrast, no increase in point mutations was observed after Ku70 ablation (∆FANCC, 1/85 changes; ∆Ku70∆FANCC, 0/30 changes). Two additional HR events were analyzed to confirm the rescue of HR-dependent repair: sister chromatid exchanges (SCEs) and gene targeting. DT40 and ∆Ku70 cells induced SCEs when exposed to cisplatin, but this response did not occur in ∆FANCC cells. However, ∆Ku70∆FANCC was clearly capable of inducing SCEs in response to cisplatin (Fig. 2, C and D). The ∆FANCC strain also showed a defect in specific gene targeting, which was also reversed in the ∆Ku70∆FANCC strain (fig. S2B).

Fig. 2

Improved HR repair of endogenous and cisplatin-induced DNA damage in the ∆Ku70FANCC strain. (A) sIgM+ to sIgM fluctuation assay to assess Ig gene conversion efficacy. sIgM+ clones (each symbol denotes a single clone) were expanded from single cells up to 55 doublings and then assessed for sIgM by FACS. sIgM loss was expressed as a percentage of the total per clone and the median calculated. (B) Sequence analysis of amplified V genes obtained from sorted sIgM cells. Pie charts depict the proportion of the total number of sequences (number at center) containing one to five distinct gene conversion tracts per locus. Gene conversion events (solid bars) are represented as 10 separate diversified V gene loci for each strain. (C) Inducible sister chromatid exchanges in response to cisplatin, scored as described previously (3). Each symbol depicts a single scored metaphase. (D) Statistical analysis of SCE data.

These results indicate a pivotal role for the Ku70 gene in inhibiting homologous recombination repair after inactivation of the FA pathway, and this role is not due to the NHEJ genes DNA-PKcs or Ligase IV (11). Ku70 binds DNA ends at DSBs (12) and can interfere with HR repair (8, 13). DSBs generated as a by-product of replication-coupled repair could therefore be trapped by Ku70, causing them to be diverted from HR repair. Our data suggest that the FA pathway intervenes in this process. The FANCC protein regulates monoubiquitination of FANCD2, a key downstream effector of the FA pathway (14, 15). It was therefore logical to test whether the purified FANCD2 protein (Fig. 3, A and B, Fig. 4B, and fig. S3C) might counteract Ku70 access to DNA. Both human (16) and chicken FANCD2 bound single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) in the absence of divalent cations (Fig. 3C). Addition of Mg2+/Mn2+ resulted in degradation of ssDNA and 3′ overhangs (fig. S3, A and B). Degradation occurred with 3′→5′ polarity and was common to both human and chicken FANCD2 (Fig. 3, D and E). Despite extensive analysis of the FANCD2 primary sequence, an exonuclease domain remained elusive. We therefore established the specificity of this activity by several distinct approaches. First, the activity fractionated with FANCD2 (fig. S4A). Second, adsorption of purified FANCD2 by means of anti-FANCD2 columns removed exonucleolytic activity (Fig. 4A). Third, human FANCD2 expressed and purified from a different host specifically copurified with exonuclease activity (Fig. 4B). Furthermore, exonuclease activity was coincident with purified FANCD2 renatured in SDS–polyacrylamide gel electrophoresis (PAGE) activity gels, whereas truncations abolished activity (fig. S4, B and C). Finally, we crystallized chicken FANCD2, washed the crystals, and reconstituted them in nuclease buffer (Fig. 4C). These preparations retained 3′→5′ exonuclease activity. These converging lines of evidence indicate that exonuclease activity is intrinsic to the FANCD2 polypeptide.

Fig. 3

Purified human and chicken FANCD2 possess 3′→5′ exonuclease activity. (A) Purification scheme (top) and Coomassie-stained SDS-PAGE analysis (bottom) for the purification of human FANCD2 (hFANCD2) containing a C-terminal His tag. M, marker; S, supernatant; FT, flowthrough; W, wash; E, eluate(d, dialyzed; c, concentrated). (B) Purification scheme (top) and Coomassie-stained SDS-PAGE analysis (bottom) for chicken FANCD2 (cFANCD2) carrying an N-terminal tandem affinity purification tag. T, total lysate; B, bound fractions. (C) Electrophoretic mobility shift assay to determine binding of cFANCD2 (ChD2) to radiolabeled ssDNA and dsDNA. (D) hFANCD2 (200 nM) was reacted with 5′- or 3′-radiolabeled single-stranded T15 oligonucleotide for increasing times (5 to 80 min at 37°C). +, mung bean nuclease was used as a positive control (10 min). (E) Purified equimolar hFANCD2 and cFANCD2 were compared for exonuclease activity (0, 5, 10, 25, and 50 nM protein reacted with 5′-labeled polyT 36-mer (T36) for 15 min at 37°C).

Fig. 4

Multiple approaches indicate that exonuclease activity is intrinsic to both human and chicken FANCD2 polypeptide. (A) Adsorption of recombinant hFANCD2 to anti-FANCD2 or anti-His antibodies, but not a control antibody, depletes exonuclease activity. Activity was assayed on 3′-labeled ssDNA substrate, incubated with FANCD2 fractions from one to four rounds of depletion. Anti-FANCD2 immunoblot shows the level of FANCD2 depletion. (B) C-terminal His-tagged hFANCD2 was expressed and purified from yeast cells. A parallel mock purification of His-tagged topoisomerase II (TopoII) demonstrates that activity is coincident with FANCD2. The relevant fraction corresponding to the FANCD2 elution fraction was then tested for purity (left panel) and exonuclease activity on a polyT 15-mer (T15) radiolabeled substrate (right panel). (C) Crystallized TAP-tagged cFANCD2, amino acids 48 to 1440, has nuclease activity. Purified protein was crystallized (bottom left, representative crystal next to a 100-μm scale bar) and multiple larger crystals were removed and washed (middle, Coomassie-stained SDS-PAGE analysis). Crystals were dissolved and tested for activity on 5′-radiolabeled ssDNA substrate (right panel). The controls were recombinant cFANCD2; Klenow fragment; and a clear drop from a well containing FANCD2, which did not crystallize.

Our results indicate that the FA pathway promotes HR repair of DSBs created at cross-links or abasic sites by counteracting Ku70. We suggest that this is achieved by direct modification of DSBs. Recent work using Xenopus extracts showed that a DNA cross-link causes bidirectional replication fork arrest (fig. S5). Repair is initiated when one fork advances up to the cross-link; dual incisions combined with TLS then generate an intact chromatid, thereby enabling HR repair of DSBs, with the repaired chromatid acting as a template. FANCD2 depletion reduces incisions and TLS, although these steps do still occur inefficiently (4).

In addition to promoting incisions and TLS, we suggest that FANCD2 modifies the resulting DSBs to prevent Ku70 from binding and subverting HR. It is noteworthy that the polarity of the FANCD2 exonuclease (3′ to 5′) is predicted to generate 5′ ssDNA tails, contrary to a general view that only 3′ ssDNA tails stimulate recombination. However, recent biochemical work shows that the fungal ortholog of FANCD1/BRCA2 is able to promote HR by using 5′ ssDNA tails (1719). We therefore speculate that FANCD2 may function with BRCA2 in a similar manner to repair DSBs generated during cross-link repair.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1192277/DC1

Materials and Methods

Figs. S1 to S5

References

References and Notes

  1. We thank M. Takata for the anti-Ku70 antibody and the Ku70 cDNA expression vector; H. Koyama and S. Takeda for ∆Lig4, ∆Ku70, and ∆DNA-PKcs DT40 cell lines; H. Joenje and J. DeWinter for human FANCC cell lines; and W. Niedzwiedz and J. Sale for help with IgM sequence analysis. Supported by the Children’s Leukaemia Trust (G.M.), the Leukaemia and Lymphoma Research Fund (M.R.H.), and a FEBS fellowship (I.V.R.).
View Abstract

Navigate This Article