The DNA-binding activity of USP1-associated factor 1 is required for efficient RAD51-mediated homologous DNA pairing and homology-directed DNA repair

USP1-associated factor 1 (UAF1) is an integral component of the RAD51-associated protein 1 (RAD51AP1)–UAF1-ubiquitin-specific peptidase 1 (USP1) trimeric deubiquitinase complex. This complex acts on DNA-bound, monoubiquitinated Fanconi anemia complementation group D2 (FANCD2) protein in the Fanconi anemia pathway of the DNA damage response. Moreover, RAD51AP1 and UAF1 cooperate to enhance homologous DNA pairing mediated by the recombinase RAD51 in DNA repair via the homologous recombination (HR) pathway. However, whereas the DNA-binding activity of RAD51AP1 has been shown to be important for RAD51-mediated homologous DNA pairing and HR-mediated DNA repair, the role of DNA binding by UAF1 in these processes is unclear. We have isolated mutant UAF1 variants that are impaired in DNA binding and tested them together with RAD51AP1 in RAD51-mediated HR. This biochemical analysis revealed that the DNA-binding activity of UAF1 is indispensable for enhanced RAD51 recombinase activity within the context of the UAF1–RAD51AP1 complex. In cells, DNA-binding deficiency of UAF1 increased DNA damage sensitivity and impaired HR efficiency, suggesting that UAF1 and RAD51AP1 have coordinated roles in DNA binding during HR and DNA damage repair. Our findings show that even though UAF1's DNA-binding activity is redundant with that of RAD51AP1 in FANCD2 deubiquitination, it is required for efficient HR-mediated chromosome damage repair.


USP1-associated factor 1 (UAF1) is an integral component of the RAD51-associated protein 1 (RAD51AP1)-UAF1-ubiquitin-specific peptidase 1 (USP1) trimeric deubiquitinase complex. This complex acts on DNA-bound, monoubiquitinated
Fanconi anemia complementation group D2 (FANCD2) protein in the Fanconi anemia pathway of the DNA damage response. Moreover, RAD51AP1 and UAF1 cooperate to enhance homologous DNA pairing mediated by the recombinase RAD51 in DNA repair via the homologous recombination (HR) pathway. However, whereas the DNA-binding activity of RAD51AP1 has been shown to be important for RAD51-mediated homologous DNA pairing and HR-mediated DNA repair, the role of DNA binding by UAF1 in these processes is unclear. We have isolated mutant UAF1 variants that are impaired in DNA binding and tested them together with RAD51AP1 in RAD51-mediated HR. This biochemical analysis revealed that the DNA-binding activity of UAF1 is indispensable for enhanced RAD51 recombinase activity within the context of the UAF1-RAD51AP1 complex. In cells, DNA-binding deficiency of UAF1 increased DNA damage sensitivity and impaired HR efficiency, suggesting that UAF1 and RAD51AP1 have coordinated roles in DNA binding during HR and DNA damage repair. Our findings show that even though UAF1's DNA-binding activity is redundant with that of RAD51AP1 in FANCD2 deubiquitination, it is required for efficient HR-mediated chromosome damage repair.
Fanconi anemia (FA) is a multigenic, autosomal disorder characterized by developmental defects, bone marrow failure, and strong cancer disposition (1,2). To date, 23 FA complementation groups and their corresponding genes (FANCA-FANCW) have been identified (3,4). FA cells exhibit elevated sensitivity to DNA interstrand cross-linking (ICL) agents such as mitomycin C (MMC) and cisplatin (5). Within the FA DNA damage response pathway, the monoubiquitination of the FANCD2 protein is a crucial early step that serves to license downstream events. FANCD2 modification occurs on DNA and within the context of the FANCI-FANCD2 (ID2) het-erodimer. The reaction is catalyzed by the multisubunit FA core complex that harbors the E3 ubiquitin ligase FANCL and a number of other factors (1,(6)(7)(8)(9)(10).
ICL repair occurs mostly during DNA replication (11,12). Herein, the monoubiquitinated ID2 complex associates with damaged DNA and helps recruit several nucleases for "lesion unhooking," leading to an intermediate amenable to replicative lesion bypass by DNA polymerase and also a DNA doublestrand break (DSB) intermediate that is eliminated via homologous recombination (HR) catalyzed by the recombinase RAD51 (FANCR) in conjunction with the tumor suppressors BRCA1 (FANCS) and BRCA2 (FANCD1) and a number of other HR factors (13)(14)(15). Upon the completion of DNA lesion removal and replication fork restart, timely deubiquitination of FANCD2 regenerates the unmodified ID2 complex and helps extinguish the DNA damage response (16,17).
The deubiquitination of FANCD2 is catalyzed by the deubiquitinase (DUB) USP1, which forms a trimeric complex with the UAF1 and RAD51AP1 proteins (16,18,19). UAF1 associates with USP1, up-regulates DUB activity, and helps target this activity to the monoubiquitinated ID2 complex via a physical interaction with FANCI. Consistent with the presence of DNAbinding activity in the heterotrimeric RAD51AP1-UAF1-USP1 complex, we have shown recently that the deubiquitination of FANCD2 occurs on DNA, and, in this regard, that efficient deubiquitination requires the DNA-binding activity of either UAF1 or RAD51AP1 (19). Moreover, evidence has been furnished by the D'Andrea laboratory (20) that the DNA-binding activity of USP1 is germane for the replication fork association of the trimeric DUB complex and deubiquitination of the DNA polymerase processivity clamp PCNA in the replicative bypass of DNA lesions.
Importantly, in addition to its role in FANCD2 deubiquitination, the dimeric RAD51AP1-UAF1 complex also enhances the activity of the recombinase RAD51 to catalyze homologous DNA pairing between recombining ssDNA and dsDNA molecules to form a DNA joint called the displacement loop, or D-loop. As such, D-loop formation represents the central step in HR responsible for initiating DNA information transfer from a donor DNA molecule (21). In this regard, the RAD51AP1-UAF1 complex enhances the ability of the RAD51-ssDNA nucleoprotein filament to capture duplex DNA to assemble the "synaptic complex," in which the recombining ssDNA and dsDNA molecules are homologously aligned and which is the precursor to the D-loop product formed by the DNA strand exchange activity of RAD51 (21,22). Even though the DNAbinding attribute of UAF1 and RAD51AP1 serves a redundant role in FANCD2 deubiquitination (19), ablation of the DNAbinding activity of RAD51AP1 impairs the ability of the RAD51AP1-UAF1 complex to enhance RAD51-mediated homologous DNA pairing in vitro and to conduct HR-mediated DSB repair in cells (22,23). Here, we present evidence that the DNA-binding activity of UAF1 is likewise indispensable for RAD51-mediated HR both biochemically and in cells. Our results help clarify the role of UAF1 DNA binding in HR-mediated genome repair.

Differential association of UAF1 with the D-loop structure
We have previously shown that UAF1 binds ssDNA, dsDNA, and the D-loop (19,22). Here, we further investigated the relative affinity of UAF1 for various DNA substrates (oligo(83dT), 90-mer ssDNA, 80-bp dsDNA, and a D-loop structure constructed with oligonucleotides that are between 80 and 90 nucleotides in length; Table 1 and Fig. 1A) by examining their electrophoretic mobility shift. The strategy entailed coincubation of the substrates in pairs and quantification of the amount of substrates shifted by highly purified UAF1 in each case. The results from this analysis revealed the following order of substrate preference: D-loop Ͼ Ͼ ssDNA/dsDNA Ͼ oligo(dT) (Fig.  1, B-D).

Role of UAF1 DNA binding in RAD51 recombinase enhancement
The DNA-binding domain of UAF1 resides within its N-terminal portion that harbors a number of WD40 repeats (22). In a recently published study, we described the construction of two DNA-binding mutants, termed 3A and 11A (Table 2), of UAF1 that are impaired for binding ssDNA, dsDNA, and the D-loop structure (19). Both the UAF1 3A and UAF1 11A retain the ability to interact with RAD51AP1 and USP1 (19). Because the UAF1 3A , but not the UAF1 11A mutant retains residue DNAbinding activity under lower-salt conditions, we used both the UAF1 3A and UAF1 11A mutants in the following experiments. RAD51AP1 physically interacts with RAD51 and enhances the RAD51-mediated D-loop reaction (24,25). This attribute of RAD51AP1 depends on its DNA-binding activity and interaction with RAD51. Even though UAF1 does not interact with RAD51, it synergizes with RAD51AP1 in RAD51 enhancement (22) in a manner that is contingent upon its complex formation with RAD51AP1. Here, we employed our well-established D-loop assay (22,25) to interrogate whether the DNA-binding activity of UAF1 is needed for the synergistic action of the RAD51AP1-UAF1 complex in RAD51 enhancement ( Fig. 2A). In these reactions, we compared the ability of WT UAF1 and the DNA binding-deficient mutant UAF1 3A and UAF1 11A in conjunction with WT RAD51AP1 (Fig. 2). As expected, RAD51AP1 alone enhanced the efficiency of the RAD51-mediated D-loop reaction (Fig. 2, B, lane 4, and C, lane 3), whereas RAD51AP1 in combination with WT UAF1 enhanced D-loop formation to a much greater extent (Fig. 2, B, lanes 6 and 7, and C, lanes 4 and 5). Importantly, relative to WT UAF1, we observed a marked decrease in D-loop formation with the UAF1 3A mutant (Fig. 2B, lanes 10 and 11), whereas the UAF1 11A mutant was devoid of any ability to stimulate the reaction (Fig. 2C, lanes 7 and 8). Thus, the DNA-binding activity of UAF1 is critical for its functional synergy with RAD51AP1 in enhancing the recombinase activity of RAD51.

Role of UAF1 DNA binding in synaptic complex formation
Our published work has revealed a role of RAD51AP1-UAF1 in the assembly of the synaptic complex during RAD51-mediated homologous DNA pairing (22). We applied two different approaches to interrogate whether DNA binding by UAF1 is relevant for synaptic complex assembly. First, we tested the UAF1 3A mutant alongside WT UAF1 for the ability to synergize with RAD51AP1 to enhance duplex DNA capture that is mediated by the RAD51 presynaptic filament. For this, we assembled RAD51 presynaptic filaments on ssDNA that was immobilized on magnetic resin and then incubated these filaments with radiolabeled dsDNA without or with RAD51AP1 and the WT or mutant variant of UAF1. Nucleoprotein complexes were magnetically captured, followed by their elution with SDS, PAGE, and phosphorimaging analysis to quantify the amount of radiolabeled dsDNA retained in each case (Fig. 3A). As reported previously (22,26), RAD51AP1 alone stimulated duplex capture, and the addition of WT UAF1 led to further  6 -8). However, the UAF1 3A mutant was much less effective than WT UAF1 in this reaction (Fig. 3B, lanes 10 and 11).
Then, to further test the role of DNA binding by UAF1 in synaptic complex assembly, we employed an assay in which synaptic complex formation leads to the protection of the duplex DNA target against the action of the restriction enzyme SspI (Fig. 3C) (27). We showed previously (22) and confirmed here that RAD51AP1 enhanced synaptic complex formation and that RAD51AP1 and UAF1 synergize in this reaction (Fig.  3, D, lanes 4, 6, and 7, and E, lanes 4 -6). However, functional synergy was not observed for either the UAF1 3A or UAF1 11A mutant (Fig. 3, D, lanes 11 and 12, and E, lanes 8 and 9). Taken together, these results indicate that the DNA-binding activity of UAF1 is indispensable for the synaptic stage of the HR process.

Role of the UAF1 DNA-binding activity in HR and DNA damage repair
Recently, we showed that HeLa cells expressing the UAF1 3A and UAF1 11A mutants display hypersensitivity to the MMC, a DNA cross-linking agent, and olaparib, a poly(ADP-ribose) polymerase inhibitor (19). Here, we inquired whether HR proficiency is likewise reliant on the DNA-binding activity of UAF1. To this end, we used a U2OS cell line that harbors an integrated DR-GFP reporter suitable for examining DSB repair by HR (28, Figure 1. Preferential binding of the D-loop structure by UAF1. A, schematic depicting DNA substrates used. B, UAF1 (100, 200, 300, 400, 500, 750, and 1000 nM) was incubated with ssDNA and oligo(83dT) (10 nM each). The mobility shift of the DNA was analyzed, and the results were quantified and plotted. Treatment with SDS and proteinase K (SDS/PK) released the DNA substrates from nucleoprotein complexes. Error bars represent mean values Ϯ S.D. of data from at least three independent experiments. C, UAF1 (100, 200, 300, 400, 500, 750, and 1000 nM) was incubated with ssDNA and dsDNA (10 nM each). The ability of UAF1 to bind the indicated radiolabeled DNA substrate was analyzed as in B. D, UAF1 (20, 40, 80, 160, and 320 nM) was incubated with ssDNA, dsDNA, and D-loop substrates (10 nM each). The ability of UAF1 to bind different DNA substrates was analyzed as in B.  29). Endogenous UAF1 was knocked down by siRNA, and cDNAs encoding WT or a mutant form of UAF1 (UAF1 3A or UAF1 11A ) were introduced into cells. Immunoblotting of cell extracts confirmed comparable expression of all the UAF1 species and the expected restoration of FANCD2 deubiquitination in each case (Fig. 4A). Then the cells were transfected with a plasmid expressing the I-SceI endonuclease to generate a DSB within the reporter. Successful DSB repair leads to GFP expression, the frequency of which is determined by fluorescent cell sorting (28). As reported previously (22,30), UAF1 knockdown caused a 3-4-fold decrease in HR efficiency (Fig.  4B, lanes 1 and 2). We found that although WT UAF1 can fully complement the HR deficiency of these cells, the two UAF1 DNA-binding mutants are significantly less effective in this regard (Fig. 4B, lanes 3-5). Moreover, the UAF1 mutants are much less adept than WT UAF1 in complementing the hypersensitivity of U2OS cells depleted for endogenous UAF1 to MMC and camptothecin (CPT, a DNA topoisomerase I inhibitor) (Fig. 4C). Thus, the DNA-binding activity of UAF1 is required for HR and HR-mediated DNA damage repair. UAF1 deficiency does not affect DNA damage-induced RAD51 focus formation in DT40 chicken or HeLa cells (18,22). Consistent with this finding, we have found that HeLa cells depleted for endogenous UAF1 but expressing either of the two UAF1 DNA-binding mutants remain competent for RAD51 focus formation after MMC treatment (Fig. 4D).

Discussion
UAF1-deficient cells are impaired for HR (18,30). Previously, we demonstrated that UAF1 synergizes with RAD51AP1 in RAD51-mediated D-loop formation, and this synergistic activity requires complex formation between RAD51AP1 and UAF1 and also the DNA-binding attribute of RAD51AP1 (22). We have isolated UAF1 mutants that are specifically compromised for DNA binding, because these mutants (UAF1 3A and UAF1 11A ) still retain the ability to form a stable complex with RAD51AP1, FANCI, and USP1 to stimulate USP1 deubiquitinase activity (19). In this study, we provide evidence that the DNA-binding activity of UAF1 is also indispensable for HR and DNA damage repair (Fig. 5). Specifically, our results show that: 1) UAF1 preferentially binds an HR intermediate-like DNA substrate (D-loop); 2) the two DNA binding-deficient mutants of UAF1 that we have constructed are unable to stimulate RAD51-mediated D-loop formation; 3) the UAF1 DNA-binding activity is specifically required for synaptic complex in HR; and 4) cells depleted of endogenous UAF1 and expressing the DNA-binding mutants are impaired for DNA damage repair and HR.
RAD51AP1-UAF1 is associated in a stable complex with the DUB USP1, and the resulting heterotrimeric complex catalyzes the deubiquitination of DNA-bound Ub-FANCD2 (19). Importantly, USP1 also binds DNA, with specificity toward replication fork-like DNA structures (20). This USP1 attribute is believed to be germane for targeting the RAD51AP1-UAF1-USP1 complex to stressed and damaged replication forks, to act on substrates such as the ubiquitinated form of DNA polymerase clamp PCNA (20). We note, however, that we have not detected any effect of USP1 on homologous DNA pairing mediated by RAD51 in conjunction with RAD51AP1-UAF1 in vitro (22), suggesting that DNA binding by USP1 may be dispensable for the DNA strand invasion step of RAD51-dependent HR. A recent report (31) has implicated RAD51AP1 in the alternate

UAF1 DNA binding in HR DNA repair
lengthening of telomere (ALT), such that its deficiency leads to generational telomere shortening in ALTϩ cancer cells. It will be of considerable interest to determine whether UAF1 and USP1 are likewise involved in ALT and to ascertain the DNAbinding attributes of the RAD51AP1-UAF1-USP1 complex therein.
The FA DNA damage response pathway has been associated with the avoidance or clearance of R-loops, a three-stranded nucleic acid structure that harbors a RNA-hybrid and a displaced DNA strand, that arise upon perturbations in transcription or transcription-associated processes, such as mRNA splicing. Persistent R-loops can lead to replication fork collapse and thus pose a threat to the maintenance of genome stability (32)(33)(34). Increased R-loops have been found in FA mutant cells (35)(36)(37). Recently, we found that the ID2 complex directly associates with R-loops in cells and that various RNA species and R-loops are just as capable as DNA in supporting the monoubiquitination of FANCD2 (7). It will be of particular interest to determine whether RAD51AP1-UAF1-USP1 also binds RNA and whether this activity is relevant for the deubiquitination of RNA-or R-loop-bound Ub-FANCD2 and for the functional integrity of the FA pathway in R-loop avoidance and clearance. , and their combination were tested for the ability to enhance dsDNA capture by the RAD51-ssDNA nucleoprotein filament. The results were quantified and plotted, and error bars represent means Ϯ S.D. of data from three independent experiments. C, schematic of the synaptic complex assembly assay. D, UAF1, UAF1 3A , RAD51AP1, and their combinations were tested for the ability to enhance synaptic complex assembly catalyzed by the RAD51-ssDNA nucleoprotein filament. The control reaction without SspI treatment is shown in lane 1. The results were quantified and plotted, and the percentages of protected DNA are shown as the means Ϯ S.D. from three independent experiments. E, UAF1 11A was tested for the ability to influence synaptic complex assembly as in D.

Mutant construction
The cDNA for human UAF1 (isoform 1) was subcloned into the pFastBac-1 (Thermo Fisher) vector containing a C-terminal Strep-II tag. Mapping of the DNA-binding domain and UAF1 and construction of DNA-binding mutants (UAF1 11A and UAF1 3A ; Table 2) have been described (19). The UAF1 bacmids were generated in DH10Bac Escherichia coli cells.

Purification of UAF1 and other proteins
Recombinant baculoviruses were generated in Sf9 insect cells (Invitrogen), and HighFive insect cells (Invitrogen) were used for protein expression. UAF1, UAF1 11A , and UAF1 3A were

UAF1 DNA binding in HR DNA repair
purified by a procedure entailing ion exchange, affinity chromatography, and gel filtration (19). RAD51AP1 (isoform 2) were expressed in E. coli and purified by affinity chromatography and ion-exchange chromatography (25).

DNA mobility shift assay
The ssDNA substrates (dT 83 and the 90-mer oligonucleotide; Table 1) were labeled with 32 P at their 5Ј terminus. The dsDNA substrate was generated by hybridizing 5Ј 32 P-labeled oligonucleotide P1 to its complement P2 ( Table 1). The D-loop substrate was generated by hybridizing 5Ј 32 P-labeled oligonucleotide D3 to oligonucleotides D1 and D2 ( Table 1). Purification of the dsDNA and D-loop substrates followed our published procedure (25). UAF1 or mutant UAF1 was incubated with the indicated radiolabeled DNA substrates (10 nM each) in 10 l of reaction buffer (15 mM HEPES, pH 7.5, 50 mM KCl, 1 mM MgCl 2 , 2% glycerol, and 100 g/ml BSA) at 37°C for 10 min, placed on ice, and mixed with 3 l of gel loading buffer (20 mM Tris, pH 7.5, 50% glycerol, Orange G). The reaction mixtures were resolved in 3-12% gradient polyacrylamide gels in Tris borate-EDTA (0.5ϫ TBE) buffer (20 mM Tris-HCl, pH 8.3, 25 mM boric acid, and 0.5 mM EDTA) on ice at 50 V for 20 min and then 200 V for 40 min. The gels were dried onto 3-mm Whatman filter paper (GE Healthcare) for 1 h under heat and vacuum and then visualized in a Bio-Rad personal molecular imager FX phosphorimager (22).

D-loop assay
The 32 P-labeled 90-mer oligonucleotide (2.5 M nucleotides; Table 1) and RAD51 (0.8 M) were incubated for 5 min at 37°C in 11.5 l of reaction buffer (35 mM Tris, pH 7.5, 1 mM DTT, 5 mM MgCl 2 , and 50 mM KCl containing 2 mM ATP). Then RAD51AP1 (100 nM), UAF1, UAF1 3A , UAF1 11A (100 or 200 nM), or the indicated combination of these factors was added to the reaction in 1 l, followed by a 5-min incubation at 37°C. Finally, pBlueScript replicative form I DNA (35 M base pairs) was added in 1 l. After 10 min of incubation at 37°C, reaction mixtures were deproteinized by treatment with SDS and proteinase K before being subjected to agarose gel electrophoresis. The gel was dried and analyzed by phosphorimaging as previously described (22).

Duplex capture assay
Magnetic resin (Roche Molecular Biochemicals) containing 80 ng of 5Ј-biotinylated 83-mer oligo(dT) immobilized via streptavidin (4 l) was incubated with RAD51 (3 M) in 20 l of reaction buffer (35 mM Tris-HCl, pH 7.5, 50 mM KCl, 2 mM ATP, 4 mM MgCl 2 , 100 g/ml BSA, and 1 mM DTT) at 37°C for 5 min. The resin was captured magnetically and washed once with 20 l of reaction buffer. The wash was removed, and 18 l of fresh reaction buffer was added. RAD51AP1 (200 nM), UAF1, and its DNA-binding mutant (200 or 400 nM) were added either individually or in combination in 1 l. After a 5-min incubation, the resin was captured and washed as before and then resuspended in 19 l of reaction buffer. Radiolabeled 80-mer dsDNA (4 M base pairs, oligonucleotides P1/P2 in Table 1) was added in 1 l, and the reaction mixture was gently mixed for 10 min at 37°C. After capturing the resin and applying two separate washes as before, bound proteins and radiolabeled DNA were eluted with 20 l of 2% SDS buffer. The eluate was resolved in a 10% polyacrylamide gel in TBE buffer without any DNA denaturant. The gel was dried and subjected to phosphorimaging analysis to reveal and quantify the radiolabeled dsDNA (22).

Synaptic complex assembly assay
The 60-mer oligonucleotide S1 that is complementary to the SspI restriction site in the pUC19 plasmid (12 M nucleotides; Table 1) was incubated with 4 M of RAD51 in 8 l of reaction buffer (35 mM Tris-HCl, pH 7.5, 50 mM KCl, 2 mM ATP, 4 mM MgCl 2 , 100 g/ml BSA, 1 mM DTT) for 5 min at 37°C, followed by the addition of RAD51AP1 (300 nM), UAF1, its DNA-bind- In the presence of DNA damage (ICL), ubiquitinated ID2 recruits DNA repair proteins including RAD51, RAD51AP1, and UAF1 to the lesion. The DNA-binding activity of UAF1 and RAD51AP1 is indispensable for RAD51-dependent synaptic complex assembly and D-loop formation in the repair phase. Afterward, DNA-bound RAD51AP1-UAF1-USP1 removes ubiquitin from Ub-FANCD2, a step that is critical for extinguishing the FA DNA damage response and recycling of the ID2 complex. Our studies described herein and earlier (19,23) have shown that although the DNAbinding attributes of both RAD51AP1 and UAF1 are indispensable for HR, they serve a redundant role in the deubiquitination of DNA-bound FANCD2.

UAF1 DNA binding in HR DNA repair
ing mutants (200 or 400 nM), or the indicated combination of these factors in 1 l and a 5-min incubation at 37°C. Then linear pUC19 plasmid DNA (85 M nucleotides) was added in 0.5 l, and the reaction mixture was incubated for 5 min at 37°C. Finally, SspI (2.5 units) was added in 0.5 l to complete the reaction mixture. After a 10-min incubation at 37°C, the reaction mixtures were deproteinized with SDS and proteinase K and resolved in a 1% agarose gel in TAE buffer (40 mM Tris acetate, pH 7.4, and 0.5 mM EDTA). DNA was stained with ethidium bromide, recorded, and quantified in a gel documentation station (Bio-Rad) (22). Oligonucleotide S2 that bears no homology to pUC19 DNA (Table 1) was included as the heterologous control.

Cell survival assay
The cells were seeded into 6-well plates (1 ϫ 10 4 cells/well) and incubated at 37°C for 24 h. After the addition of MMC (Sigma) or CPT (Sigma), the cells were incubated at 37°C for 5 days. The surviving cells were fixed and stained with crystal violet dye, and the absorbance was measured as described (22).

Measurement of HR frequency
The exponentially growing U2OS-DR-GFP cells were seeded in 6-well plates (3 ϫ 10 5 /well) and transfected with 4 g of I-SceI expression vector (pCBASce) using 10 l of Lipofectamine 2000 and cultured for 3 days. The cells were processed for flow cytometric analysis of GFP in a BD LSRII flow cytometer (BD Bioscience), as described (22).

Immunofluorescence to visualize RAD51 foci
HeLa cells treated with siUAF1 were transfected with the cDNA coding for the WT or a mutant form of UAF1 in the pCMV (delta4) vector using Lipofectamine 2000. The cells were seeded the next day at 1 ϫ 10 4 cells/well in 4-well chamber slides. After 24 h, the cells were treated with 1 M MMC for 16 h. The cells were then washed in PBS buffer and fixed in 4% paraformaldehyde. After washing and permeabilizing in PBS with 0.3% Triton X-100 (Sigma), the cells were incubated in PBS buffer with 0.1% Nonidet P-40 ϩ 10% normal goat serum (ab7481, Abcam) for 1 h at 25°C. Then anti-RAD51 antibody (H-92, sc-8349, Santa Cruz) diluted 1/50 in PBS buffer with 0.1% Nonidet P-40 and 0.5% BSA was added to the cells for 1 h at 25°C. After washing in PBS with 0.1% Tween, the cells were incubated with the secondary antibody (1:50, goat anti-rabbit IgG-CFL 594, sc-362282, Santa Cruz) for 1 h at 25°C. After being stained with 4Ј,6Ј-diamino-2-phenylindole (1:1000, catalog no. 62248, Thermo Fisher) for 10 min at 25°C, the cells were washed with PBS and mounted with Vectashield antifade mounting medium (H-1000, Vector). Image acquisition of RAD51 foci was performed in an inverted microscope (Nikon ECLIPSE TE2000) with the Velocity analysis software.

Data availability
All data are included in the manuscript. The raw data sets are available upon request. Funding and additional information-This work was supported by CPRIT REI Award RR180029 (to P. S.), a Gray Foundation Team Science Grant, and National Institutes of Health Grants RO1 CA168635, R35 CA241801, U54 DK106857, RO1 CA205224, RO1 ES007061, RO1 GM109645, PO1 CA92584, and P30 CA054174 (to P. S. and G. M. K.). This work was also supported in part by a Robert A. Welch Distinguished Chair in Chemistry Grant AQ-0012 (to P. S.) and a V Foundation Scholarship (to W. Z.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.