Proper RPA acetylation promotes accurate DNA replication and repair

Abstract The single-stranded DNA (ssDNA) binding protein complex RPA plays a critical role in promoting DNA replication and multiple DNA repair pathways. However, how RPA is regulated to achieve its functions precisely in these processes remains elusive. Here, we found that proper acetylation and deacetylation of RPA are required to regulate RPA function in promoting high-fidelity DNA replication and repair. We show that yeast RPA is acetylated on multiple conserved lysines by the acetyltransferase NuA4 upon DNA damage. Mimicking constitutive RPA acetylation or blocking its acetylation causes spontaneous mutations with the signature of micro-homology-mediated large deletions or insertions. In parallel, improper RPA acetylation/deacetylation impairs DNA double-strand break (DSB) repair by the accurate gene conversion or break-induced replication while increasing the error-prone repair by single-strand annealing or alternative end joining. Mechanistically, we show that proper acetylation and deacetylation of RPA ensure its normal nuclear localization and ssDNA binding ability. Importantly, mutation of the equivalent residues in human RPA1 also impairs RPA binding on ssDNA, leading to attenuated RAD51 loading and homologous recombination repair. Thus, timely RPA acetylation and deacetylation likely represent a conserved mechanism promoting high-fidelity replication and repair while discriminating the error-prone repair mechanisms in eukaryotes.

cation, damage response, HR repair, checkpoint signaling, and RPA interaction with its partners ( 8 , 47-51 ). Meanwhile, the ubiquitination of RPA by the E3 ligase RFWD3 facilitates HR repair at stalled forks or interstrand crosslink sites, likely via facilitating the degradation or removal of RPA and RAD51 from the lesion sites ( 52 , 53 ). Notably, recent studies showed that RPA is acetylated in both yeast and human upon DNA damage (54)(55)(56). In yeast, RPA was reported to be acetylated by the acetyltr ansfer ase NuA4 on lysines K259, K427, K463 and K494 of Rfa1 in response to MMS treatment ( 54 ). This acetylation appears to attenuate RPA binding on ssDNA and is r equir ed for the DNA damage response ( 54 ). In parallel, human RPA1 is acetylated on K163 by the acetyltr ansfer ases GCN5 and PCAF, and this acetylation promotes nucleotide excision repair of UV-induced DNA damage ( 55 , 56 ). Howe v er, whether RPA acetyla tion af fects the repair of DSBs, the most deleterious DNA lesions, remains to be determined.
In this study, we investigated the impact of RPA acetylation and deacetylation on spontaneous mutations and DSB repair. We show that proper acetylation and deacetylation of RPA promote high-fidelity DNA replication and recombination while suppressing mutations and low-fidelity repair pathways. We also provide evidence that this regulation plays a similar role in human cells. Thus, our study re v eals a conserved mechanism by which eukaryotic cells ensure the fidelity of replication or repair and extends the understanding of the role of RPA post-translational modifications.

Yeast strains and cell culture
Yeast strains are listed in Supplementary Table S1. HEK293T and HeLa cells wer e cultur ed in Dulbecco s modified essential medium containing 10% fetal bovine serum with 100 units / ml penicillin and 100 g / ml streptomycin. U2OS cells were cultured in 20% fetal bovine serum, and all cells were maintained in an atmosphere containing 5% CO 2 at 37 • C.

DNA damage sensitivity test
Yeast cells were cultured in the YPD medium (1% yeast extract, 2% peptone and 2% dextrose) overnight to saturation. Undiluted cell culture and 1 / 10 serial dilutions of cell cultures were spotted onto YPD plates containing indica ted concentra tions of camptothecin, phleomycin, zeocin, or MMS. Plates were incubated at 30 • C for 2-3 days before taking pictures.

Fluor escence microscop y
Yeast strains carrying the RFA1-YFP and Nup49-mCherry fusion proteins were grown at 30 • C to mid-log before being harvested. Cells wer e r esuspended in 1.7 l sterile deionized water on the glass slide and examined using a ZEISS LSM 880 fluorescence confocal microscope. Images were analyzed by Zen Application Service. Approximately 100 cells were counted for each experiment.
Nucleic Acids Research, 2023, Vol. 51, No. 11 5567 Mutation rate and spectra The rate of accumulation of CanR mutations was determined as previously described ( 57 ). Yeast cells from single fr esh colonies wer e pla ted on SC arginine-dropout pla tes containing 60 mg / L canavanine. The mutation rate was measured by fluctuation analysis using the median method. CanR mutation spectra were characterized by PCR amplification of the CAN1 gene from independent CanR isolates, followed by DNA sequencing.

Yeast gene conversion, single-strand annealing, and alt-EJ assays
To test the viability of DSB repair by gene conversion (tGI354), SSA (yWH378) or alt-EJ (JKM139), we cultured cells in the pre-induction medium (YEP-Raffinose) overnight to the log phase. Cells were diluted and plated on YEPD and YEP-Gal plates, respecti v ely, then incubated at 30 • C f or 3-5 da ys. The survival ra te was calcula ted by dividing the number of colonies grown on YEP-Gal by the number of colonies on YEPD (x100%). While the measurement of DNA end resection or repair kinetics for ectopic recombination by Southern blot was performed as described ( 58 ).
To measure the cell survival in the NA14 system, we cultured cells in the pre-induction medium YEP-Raffinose overnight to the log phase. Next, 2% of galactose was added to the medium, and cells w ere allow ed to continue growing for additional 10 h. Cells were then washed and plated on YEPD and incubated at 30 • C for three days. Colonies were then analyzed by replica plating to selecti v e media containing G418 (300 mg / ml). At least three independent experiments were performed for each strain.

Yeast BIR assay
Allelic BIR assay was performed as previously described ( 59 , 60 ). The frequencies of BIR, gene conversion, half cr ossovers, and chr omosome loss wer e measur ed based on the percentage of colonies carrying markers specific to these repair outcomes, as reported previously ( 59 , 60 ). The repair efficiency was calculated as the percentage of normalized pixel intensity of the BIR product band compared to the normalized parental bands at 0 h. Quantitati v e analysis was completed with ImageQuant TL 5.2 software (GE Healthcare Life Sciences).

Chromatin immunoprecipitation (ChIP)
Log phase yeast cells ( ∼1 × 10 7 cells / ml) grown in YEP-Raffinose medium were induced with 2% galactose to generate DSBs. Samples were collected at 0 or 4hr after DSB induction. Chromatin DNA was sheared to an average size of ∼300 bp using a Diagenode Bioruptor. ChIP and qPCR assays were carried out as previously described ( 34 ). The anti-FLAG and anti-Myc antibodies used for ChIP were ordered from CST (#20E3) and MBL(M192-( 3 ), respecti v ely.

Recombination protein expression and purification
The WT, 4KQ or 4KR RFA1 allele was individually cloned into the vector pGEX-4T-3 and transformed into BL21(DE3) E. coli cells. When the culture was grown to an OD 600 of 0.4-0.6, 0.1mM IPTG was added to induce protein expression. Cells were cultured at 16 • C for 16 h before harvest. The cell pellet was resuspended in 1xPBS, followed by lysing with sonication. After centrifugation at 16 000 g for 15 min, the supernatant was collected and filtered through a 0.45 m sterile syringe filter. The supernatant was then mixed gently with GST agarose beads at 4 • C for 2 h with agitation. The GST agarose w as w ashed e xtensi v ely with the purifica tion buf fer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 5 mM EDTA, 1% Triton X-100), followed by elution with 20 mM reduced glutathione. The purification of the NuA4 complex was carried out as described by Wang et al. ( 61 ).

Immunoprecipitation and Western blot
Yeast cells were grown at 30 • C overnight to log phase. Cells wer e tr eated with or without 0.1% MMS for 1 h before harvest. Cells were then resuspended in IP lysis buffer (100 mM HEPES, 100 mM KAC, 2 mM MgCl 2 , 2 mM Beta-ME, 0.1% NP-40, 1 mM PMSF, 1 × protease Inhibitor Cocktail, 5 mM Na Butyrate , 5 mM Nicotinamide , and 1 g / ml TSA) and mechanically disrupted using glass beads a t 4 • C . The lysa tes were collected and digested with Ultra-Nuclease (YEASEN Biotech) at 37 • C for 20 min to solubilize chromatin-bound proteins. For immunoprecipitation, each sample was incubated with 10 l of anti-acetyl-lysine (Immunechem, ICP0380) or 3 l of anti-FLAG (MBL, M185-3L) or anti-HA (MBL, M180-3) antibody for 4 h or overnight at 4 • C. Afterward, the mixture was added with pr otein G-sephar ose and incubated at 4 • C for 3 h with agitation. The beads were washed with the IP lysis buffer supplemented with 140 mM NaCl four times and resuspended in 50 l of 2 × SDS loading buf fer. Immunoprecipita ted proteins were analyzed by Western blot using an anti-FLAG antibody (Sigma, F1804).
For immunoprecipitations in human cells, HEK293T cells were washed with 1 × PBS and lysed in the NETN buffer (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 100 mM NaCl, 0.5% NP-40, supplemented with protease inhibitors) containing deacetylase inhibitors (5 mM nicotinamide and 1 g / ml TSA) for 30 min on the ice. After centrifugation at 16 000 g for 15 min, the supernatant was collected and pr eclear ed with the pr otein G-agar ose beads (GE Healthcare). The beads were removed by centrifugation, and the supernatant was incubated with an anti-acetyl lysine antibody (Immunechem, ICP0380) at 4 • C for 6 h. Next, 30 l of pr otein G-agar ose beads wer e added to each r eaction, and the mixture was incubated at 4 • C with agitation for additional 3 h. Finally, the beads were washed e xtensi v ely with the NETN buffer supplemented with deacetylase inhibitors, followed by boiling in 2 × SDS loading buffer.
Pr oducts fr om immunoprecipitation, pull down or the whole cell lysates were resolved on an 8.5% SDS-PAGE followed by transferring onto a PVDF (Immobilon-P, Millipore) membrane using the semi-dry method (Bio-Rad). For Western blot analysis, the anti-FLAG (F3165 or F1804) and anti-HA (30701ES60) antibodies were purchased from Sigma and YESEN, respecti v ely. The anti-GAPDH and anti-GST(AE001) antibodies were purchased from Abclonal. The anti-His (No. 66005-1-Ig) antibody was ordered from Proteintech. The human RPA1(ab176467) antibody was purchased from Abcam. The anti-mouse and rabbit IgG HRP-conjugated secondary antibodies were purchased from Santa Cruz Biotechnology. Blots were de v eloped using the western blotting substrate (Bio-Rad).

In vitro acetylation assays
In vitr o acetyla tion assa y was perf ormed as described ( 54 ). Briefly, the acetylation was carried out in a 15 l of reaction system containing 10 nM of the NuA4 complex, 100 nM of the GST-RPA complex (or 40 nM of GST-RFA1 or GST-RFA1-4KR), 0.25 Ci of 3 H-acetyl CoA (Perkin Elmer), 10 mM Na Butyrate, 25 mM KCl, and 3 l of 5 × HAT buffer (250 mM Tris pH 8.0, 25% glycerol, 0.5 mM EDTA, 5 mM DTT, 5 mM PMSF). The reaction was incuba ted a t 30 • C for 30 min. The reaction product was spotted on a PVDF membrane for the liquid assay. After air drying, the membrane was washed three times with 50 mM carbonate buffer (0.5 M Na 2 CO 3-NaHCO 3 , pH 9.2), followed by rinsing with acetone. The membrane was then placed in a scintillation vial, followed by the addition of a scintillation cocktail, and the radioacti v e signals were counted by a liquid scintillation analyzer (Tri-carb 2910TR, PerkinElmer).
To compare the ssDNA binding ability of human RPA1-WT, 3KQ and 3KR mutant protein, HEK293T cells were transfected with the pCDNA5.0 plasmid expressing the FLAG-tagged RPA1-WT, 3KQ or 3KR using the Gene Twin transfection reagent. After 24 h, cells were washed with 1xPBS three times and harvested, and then lysed in the NETN lysis buffer (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, and 0.5% NP-40, 100 mM NaCl, supplemented with protease inhibitors) on ice for 30 min. After centrifugation, the supernatant was collected and incubated with the biotin-ssDNA-streptavidin beads for 30 min at room temperature. Beads were washed with the NETN buffer three times, followed by boiling in 2 ×SDS loading buffer. The eluted proteins were analyzed by Western blot. Analysis of the ssDNA binding ability of yeast RFA1-WT, 4KQ, or 4KR was conducted as described abov e, e xcept that the lysis buffer and wash buffer used were the same as the ones used for acetyll ysine imm unoprecipitation in yeast.

Electrophoretic mobility shift assay (EMSA)
20 nM of 5 -Cy3 labeled ssDNA (30 or 90 nt) was incubated with indicated amounts of the WT, 4KQ or 4KR RPA complex (10, 20 or 30 nM) at room temperature for 30 min in 50 mM Tris-HCl, pH 8.0. The reaction mixture (10 l) was mixed with 3 l of 6 ×DNA loading dye and loaded onto a 6% nati v e bis-acrylamide gel and resolved using cold 0.3 ×TBE buffer for 20 min. The gel was scanned using Typhoon FLA 9500 imager (GE Healthcare), and band intensities were quantified using the ImageJ software.

Single-molecule study
Single-molecule studies monitoring RPA binding on ss-DNA were performed as previously described ( 34 ). The 12.5 k-nt ssDNA was produced by one-sided PCR, and its two ends were labeled with digoxigenin and biotin groups, respecti v ely. First, the digoxigenin-labeled end of a single ssDNA molecule was anchored to the anti-digoxigenincoated glass surface in a flow cell. Then, the biotin-labeled end of the anchored ssDNA molecule was attached to a superparamagnetic microbead (M-270, Dynal beads). A pair of permanent magnets were used to attract the microbead and thus exert a constant force on the anchored ssDNA molecule. The extension of ssDNA was determined to be the separation between the microbead and glass surface. The assembling buffer contained 100 mM NaAc, 10 mM MgAc 2 , 1 mM ATP and 25 mM Tris-Ac, pH 7.5. All experiments were performed at a constant force of 4.3 pN at 21 • C.

Microscale thermophoresis assay
Purified recombinant 6xHis-Rtt105 protein was labeled with a RED-tris-NTA protein labelling kit (Nano Temper) following the manufacturer's instruction. The labeled protein was incubated at a constant concentration (100 nM) with serial dilutions of the WT, 4KQ, or 4KR RPA complex (from 200 to 0.0058 nM) in the MST buffer (1 × PBS with 0.05% Tween 20). Equal volumes of proteins wer e mix ed by pipetting and incubated for 50 min at room temperature. The r eaction mixtur es wer e enclosed in Monolith ™ NT.115 Series Capillaries and loaded into the instrument (Monolith NT.115, Nano Temper, Germany). The measurement procedures and K d value analysis were determined using the Nano Temper analysis tool.

Human HR, BIR, alt-EJ and SSA reporter assays
HR, BIR, alt-EJ and SSA were measured using the reporter assa ys f ollowing the previously described procedures. The DR-GFP ( 62 ) and alt-EJ-EGFP ( 63 ) reporters in U2OS cells were gifted by Dr Xingzhi Xu (Shenzhen Uni v ersity), and the EGFP-BIR-5085 ( 64 ) and SSA-EGFP ( 65 ) reporters in U2OS cells were provided by Dr Hailong Wang (Capital Normal Uni v ersity) and Dr Jun Huang (Zhejiang Uni v ersity), respecti v el y. The siRN A transfection was performed with 100 pm of RPA1 siRNA duplexes using Lipofectamine 2000 reagent following the manufacturer's instruction. Four hours after siRNA transfection, a plasmid expressing the FLAG-tagged siRNA-resistant RPA1 -WT, -3KQ or -3KR allele was transfected into the reporter system. For analyzing HR, BIR or SSA, cells transfected with indicated plasmids were then transiently transfected with the I-Sce I e xpressing v ector pCBAScel (Addgene). After 48 h, cells were harvested, and the expression of EGFP was analyzed by flow cytometry. For alt-EJ, cells transfected with indicated plasmids were then supplemented with 10 g / ml doxy cy cline to induce DSBs. After 24 h, the percentage of GFP-positi v e cells was analyzed by flow cytometry. The relati v e efficiency of HR, BIR, alt-EJ or SSA was normalized to that of control cells. At least 10000 cells were counted for each sample. The values presented are the analysis of three independent experiments.

Immunofluorescence staining
HeLa cells were seeded on coverslips in 12 wells plates 24 h before experiments. For staining ␥ H2AX foci, cells were treated with HU (5 mM) for 4 h and then allowed to recover for 10-24 h. For staining RPA and RAD51 foci, cells were treated with VP16 (5 M) for 1 h before sampling. Cells on coverslips were washed with 1 ×PBS, then fixed with 4% paraf ormaldehyde f or 15 min a t room tempera ture and permeabilized with 0.5% Triton X-100 for an extra 15 min. After being blocked with 5% BSA, cells on coverslips were incubated with the anti-␥ H2AX (ab26350, Abcam), anti-RPA1(ab176467, Abcam) or anti-RAD51(ab133534, Abcam) antibody at 4 • C overnight. Following washing with PBS three times, secondary antibodies were added and incuba ted a t room tempera tur e for 1 h. Cells wer e then stained with DAPI to visualize nuclear DNA. Images were captured using Leica SP8 inverted fluorescent microscope with a 63 ×objecti v e and processed using Leica Application Suite X software. Quantification of the signals was carried out using ImageJ. Statistical analysis was performed using Prism (GraphPad Software). Statistical significance was determined by the two-tailed t -test.

Cell viability assay
Cell viability was measured using the CCK-8 assay (YEASEN Biotech). HeLa cells stably expressing the FLAG-tagged WT, 3KQ or 3KR RPA 1 allele (siRNA resistant) were seeded at a density of 5 × 10 3 cells / well in 96well plates. After 24hrs, cells were transfected with siRNA against RPA1 using the Lipofectamine 2000 reagent. Cells were then incubated for 24 h befor e tr eating with iniparib (MCE) a t indica ted doses (0, 10, 25, 50 and 100 M). After 24hr treatment, 10 l CCK-8 reagent was added to each well, and the plates were incubated for 1 h at 37˚C. Finally, the absorbance was measured at 450 nm using a scanning microplate reader (Cytation3). Cell viability at individual time points was normalized to the untreated group.

RPA is acetylated in vivo and in vitro by NuA4
To characterize the role of RPA acetylation in preserving genome stability, we first immunoprecipitated acetylated pr oteins fr om yeast l ysates with a pan-anti-acetyl-l ysine antibody and tested RPA acetylation by immunoblotting with an anti-FLAG antibody. In line with previous studies, we noted that yeast RPA was acetylated at a low le v el in unperturbed conditions, and the acetylation was enhanced upon methyl methane sulfonate (MMS) treatment that can induce DSBs (Supplementary Figur e S1A) ( 54 ). Furthermor e, we observed tha t RPA acetyla tion was dependent on the acetyltr ansfer ase NuA4 since the depletion of Esa1, the catalytic subunit of the NuA4 complex, by shifting the esa1-L254P temperature-sensiti v e mutant to a non-permissi v e temperatur e, significantly impair ed RPA acetylation (Supplementary Figure S1B), as previously noted ( 54 ). Consistently, purified yeast NuA4 complex directly acetylated the recombinant RPA complex in vitro , as measured by a liquid scintillation spectrometer (Supplementary Figure Figure S2). Thus, the RPA complex is acetylated by NuA4 primarily on Rfa1, and this acetylation could be stimulated by DNA damage.

Proper acetylation and deacetylation of RPA promotes DNA damage response and repair
To test the role of RPA acetylation in the DNA damage response, we constructed the rfa1-4KR ( 4KR ) and rfa1-4KQ ( 4KQ ) point mutants wherein the four lysines (K259, K427, K463 and K494) were sim ultaneousl y m utated to arginine to block the acetylation or glutamine to mimic constituti v e acetylation. Unlike wild-type (WT) cells, blocking RPA acetylation ( 4KR ) caused hypersensitivity to phleomycin or zeocin, which can induce DSBs. Howe v er, mimicking constituti v e RPA acetylation ( 4KQ ) resulted in a more se v ere defect in response to all tested DNA-dama ging a gents, including phleomycin, zeocin, MMS and camptothecin (Figure 1 A).
To determine which residue is the key acetylation site in cells, we mutated these sites individually and carried out immunoprecipitation with the pan-anti-acetyl-lysine antibody. The product was analyzed by immunoblotting with the anti-FLAG antibody. We noted that the K427R or K494R single-point mutation did not reduce RPA acetylation (Supplementary Figure S3A). Howe v er, K463R and, to a lesser extent, K259R mutation impaired RPA acetylation, and a combined mutation of the four sites ( 4KR ) further reduced RPA acetylation. Notabl y, m utation of any of these single residues to glutamine or simultaneous mutation of K259 and K463 to glutamine caused no or only marginal defects in the DNA damage sensitivity, in contrast to that observed in the 4KQ mutant (Supplementary Figure S3B). These results suggest that although K463 and K259 appear to be the key acetylation sites, the acetylation of RPA on the four residues functions collecti v ely to promote the DNA damage response.
The DNA damage sensitivity in the 4KQ or 4KR mutant was not caused by any defects in the DNA damage checkpoint because both mutants can ef ficiently activa te the checkpoint upon DSB induction, as reflected by the status of Rad53 phosphorylation and G2 / M arrestment (Supplementary Figures S4A and B). Together, these results suggest that proper acetylation and deacetylation of RPA are r equir ed to stimulate DNA damage response or repair.

Proper acetylation and deacetylation of RPA promotes DSB repair by gene conversion
Next, we tested the role of proper RPA acetylation in DSB repair by HR. We employed an ectopic recombination system wherein the HO endonuclease generates a single DSB at the MAT a sequence inserted at the ARG5,6 locus on chromosome V. The DSB is r epair ed by HR using the homologous MAT a -inc sequence on chromosome III as a template (Figure 1 B) ( 66 ). ∼80% of the WT cells completed the repair and survi v ed, while the survi val rate was reduced to 52% for the 4KR mutant and 16% for the 4KQ mutant (Figure 1 C). Moreover, both mutants repaired the break with much slower kinetics, as re v ealed by the Southern blot analysis (Figures 1 D and E). In contrast to the 4KQ m utant, m utation of the four residues individually to glutamine has only modest or no defect in the HR repair ( Supplementary Figure S3C), supporting the conclusion that RPA acetylation on these sites functions collecti v ely. These results indicate that proper acetylation and deacetylation of RPA are important for DSB repair by gene conversion.
We then tested the effect of proper RPA acetylation on br eak-induced r eplication (BIR), a unique HR mechanism for repairing one-end DSBs that occur at collapsed replication forks or eroded telomeres ( 67 ). We used a BIR system in w hich onl y one end of the HO-induced DSB has e xtensi v e homology to the template sequence so that ∼70% of cells use BIR to copy over 100 kb of chromosome III to complete the repair ( 59 , 60 ). > 20% of the remaining cells repaired the break via gene conversion by capturing the second end of the DSB. The repair outcome or chromosome loss was determined by following the genetic markers (Figure 1 F) ( 59 , 60 ). Compared to the WT cells, the repair pattern or outcome in the 4KR mutant remains largely unaffected, except that chromosome loss was slightly increased (Figures 1 G and H). Howe v er, the repair outcome is markedly changed in the 4KQ mutant , with gene conversion reduced from 22% to 1.8% and BIR reduced from 75% to 28%, accompanied by a massi v e chromosome loss (from 1.34% to 68.4%, ∼50-fold) (Figures 1 G and H). Consistently, we failed to detect BIR products within 10 hours following DSB induction in the 4KQ mutant cells (Figure 1 I). Notab ly, we observ ed frequent chromosome rearrangements in this mutant (Supplementary Figures S5A-C). Thus, timely deacetylation of RPA is crucial to promote DSB repair by gene conversion or BIR while preventing chromosome loss. These results establish an important role of proper RPA acetyla tion and deacetyla tion in the repair of DSBs by HR.

Proper RPA acetylation favors DSB repair by the accurate HR while discriminating the err or-pr one pathwa ys
We then asked whether RPA acetylation affects the choice of DSB repair pathways. We employed a haploid yeast strain (N A29) w herein a single HO-induced DSB is repaired primarily by the accurate intrachromosomal or ectopic recombination or by the deleterious intrachromosomal single-strand annealing (SSA) that anneals the complementary ssDNA re v ealed by resection and leads to the deletion of the intervening sequence (Figure 2 A) ( 68 , 69 ). In this system, two copies of the URA3 genes (direct repeats) sharing 1.2-kb homology were inserted on chromosome V, separated by a 5.5 kb interval (Figure 2 A). One of them carries a HO cut site, while the second copy is a WT URA3 gene. A third 1.2-kb fragment of the URA3 gene with a m utated HO reco gnition site ( MA T a-inc) is present a t the LYS2 locus on chromosome II. A single DSB can be induced by the HO endonuclease upon galactose induction. By following the genetic markers, we could distinguish the products r epair ed by gene conversion (Ura + G418 R ) or SSA (Ura + G418 S ). 83% of WT cells completed the DSB repair and survi v ed, while the survi val rate was reduced to 50% for the 4KQ mutant and ∼60% for the 4KR mutant (Figure 2 B). As reported, a pproximatel y half of the survivors for WT cells wer e r epair ed by gene conversion, while the other half wer e r epair ed by SSA (deletions) ( 68 ). Notably, the repair by gene conversion was reduced to 7.6% in the 4KQ mutant and 18.5% in the 4KR cells, while the repair by SSA appeared to remain unaffected in both mutants (Figure 2 C). This was surprising because RPA was known to suppress the Rad52-mediated annealing of complementary ssDNA ( 7 , 29 , 34 ). We reasoned that the effect of RPA acetylation on SSA might be minimized by the presence of ectopic recombination in this system.
To directly measure the impact of RPA acetylation on SSA repair, we employed an SSA assay wherein two partial leu2 repeats separated by 25-kb interval on chromosome III can anneal to complete the HO-induced DSB repair (Figure 2 D) ( 70 , 71 ). Rad51, which is r equir ed for typical HR but is dispensable for SSA, was deleted so that SSA is the only pathway to repair the DSB. A pproximatel y 60% of WT cells survi v ed, while ∼76% of 4KR cells successfully completed the repair (Figure 2 E), suggesting that proper RPA acetylation is r equir ed to suppr ess e xcessi v e SSA repair. This is in line with the role of RPA in restraining SSA ( 29 , 34 ). In contrast, we noted that the survival rate was reduced to 37% in the 4KQ mutant (Figure 2 E), implying that constituti v e RPA acetyla tion might af fect additional steps r equir ed for SSA repair.
The binding of RPA on ssDNA suppresses DSB repair by alternati v e end joining (alt-EJ), a Ku-independent err or-pr one pr ocess that r equir es the annealing of short homologies ( 29 , 72 , 73 ). Ther efor e, we tested the effect of RPA acetylation on alt-EJ in the YKU70 -deleted JKM139 strain, wherein the HR pathway is blocked so that the DSB can only be r epair ed by alt-EJ. Notably, the alt-EJ rate is increased by 5-fold in the 4KR mutant and by 12-fold in the 4KQ mutant (Figure 2 F). In addition, we observed that ∼10% of repair products in the WT cells harbor large deletions at the breakpoint, and the proportion was increased to  Figure S6). Together, these results suggest tha t proper RPA acetyla tion and deacetyla tion facilita te the accurate repair by gene conversion while discriminating the aberrant repair by SSA or alt-EJ that leads to deletions.

Proper RPA acetylation and deacetylation ar e r equir ed to suppress spontaneous large deletions, duplications and chromosome loss
One of the key functions of RPA is to suppress mutations or chr omosome catastr ophe ( 7 , 11 , 13 ). We used the CAN1 gene as a mutation reporter to test whether proper RPA acetylation affects the fidelity of DNA replica tion ( 57 ). Muta tions in the CAN1 gene generate canavanine-resistant (CanR) colonies that can be screened on SC arginine-dropout plates containing canavanine. In the WT cells, the spontaneous muta tion ra te is about 1.7 × 10 −7 . The mutation rate remained unchanged in 4KR cells, while it was increased to ∼10 × 10 −7 ( ∼5-fold) in 4KQ cells (Figures 3 A and B). To identify the nature of the mutations, we sequenced the CAN1 gene from individual CanR colonies deri v ed from the WT or mutant cells. We found that ∼ 77% of the mutations in WT cells were base substitutions, while the   Figure S7). Notably, 80% of the multi-base pair deletions in the 4KQ mutant are large deletions occurring between dir ect r epeats with short microhomologies, with sizes ranging from 63 to 135 bp. In contrast, only 12.5% of multi-base pair deletions in the 4KR mutant belong to micro-homology-mediated large deletion e v ents, while the rest are small deletions (Figures 3 D and E and Supplementary Figures S7-S9). We also detected two large micro-homology-mediated insertion e v ents in the 4KR mutant cells. Both insertions are tandem duplications of DNA fragments with sizes ranging from 23 to 49 bps (Figures 3 Figure S7) ( 34 , 74 ). These results suggest that proper acetylation and deacetylation of RPA suppress mutations with the signature of micro-homology-mediated large deletions or insertions.

D and F and Supplementary Figures S7-S9). Notably, these gross deletions or duplications did not occur in WT cells (Figure 3 E and F and Supplementary
Micro-homology-mediated deletions or duplications wer e r eported to associate with polymerase slippage or SSA (74)(75)(76)(77)(78)(79). Indeed, additional deletion of RAD59, which is r equir ed for SSA ( 80 , 81 ), in the 4KQ mutant gr eatly reduced the micro-homology-mediated large deletions, yet it led to an increase of micro-homology-mediated large insertions (Figures 3 E and F). Interestingly, we observed two unique large insertion events in the 4KQ rad59 Δ double mutant (Supplementary Figures S7-S8), i.e. an insertion of a 329-bp Ty1 retrotransposon element YPRCdelta18 at +1398 bp and an insertion of a 342-bp of YCLWdelta15 fragment at +149 bp position. How retrotransposon elements were activated in this mutant remains to be determined.
Finally, we evaluated the effect of the RPA mutations on chromosome loss using a system that carries an extra ∼320-kb yeast artificial chromosome (YAC) ( 82 ). We observed that the 4KQ and 4KR mutations led to a 30-fold and 6-fold incr ease, r especti v ely, in YAC loss compared to the WT cells (Figure 3 G). Together, these results indicate that proper acetylation and deacetylation of RPA are important for suppressing chromosome loss and spontaneous mutations with the signature of micro-homology-mediated large insertions or deletions.

Proper RPA acetylation and deacetylation facilitate the loading of RPA, Rad52 and Rad51 at DSB ends
To explore how proper RPA acetylation may impact HR r epair, we first measur ed DNA end r esection using a system wherein a single HO-cut is induced on the MAT a locus on chromosome III upon the addition of galactose ( 70 ). The homologous sequence HML and HMR were deleted to pre v ent the r epair by HR so that r esection could proceed persistently. Compared to the WT cells, the 4KR mutant resected the DSB ends at a normal rate, while the 4KQ mutant processed the ends faster in the distal region (Supplementary Figures S10A and B). This is likely attributed to reduced binding of the checkpoint adaptor protein Rad9, which is known to suppress DNA end resection (Supplementary Figure S10C) ( 70 ). Thus, the defect of HR repair in the 4KR or 4KQ mutant was not due to any defects in DNA end resection.
Although ssDNA was efficiently produced, RPA loading at DSB ends was significantly impaired in the 4KQ mutant and, to a lesser extent, in the 4KR mutant (Figure 4 A). RPA is r equir ed for efficient loading of R ad52 and R ad51 at DSB ends. As a result, the loading of Rad52 and Rad51 at DSBs was significantly impaired in the 4KQ mutant and modestly decreased in the 4KR mutant (Figures 4 B and C). The more se v ere defect in Rad52 loading might explain the r educed SSA r epair in the 4KQ mutant (Figur e 2 E). Howe v er, both the recombinant WT and mutant Rfa1 proteins can efficiently interact with Rad51 or Rad52 in vitro (Sup-plementary Figures S11A and B), suggesting that the loading defects of Rad52 and Rad51 were caused by impaired RPA loading rather than any defects in the interaction between RPA and Rad52 or Rad51. Together, these results indica te tha t proper acetyla tion and deacetyla tion of RPA are r equir ed for proper loading of RPA, Rad52 and Rad51 at DNA breaks.

Proper RPA acetylation regulates RPA binding to ssDNA
Next, we assessed the ssDNA-binding ability of the mutant RP A proteins. W e incubated biotin-labeled ssDNA (30nt) coupled to magnetic beads with cell lysates from MMStreated yeast cells to attract the endogenous RPA. Compared to the WT cells, both mutant proteins exhibited a reduced affinity to ssDNA, with the 4KQ mutant protein having a more se v ere defect (Figures 4 D and E). It is worth noting that the 4KQ or 4KR mutation did not affect the RPA protein le v el or the assemb ly of the RPA comple x, as the mutant Rfa1 can efficientl y imm unoprecipitate Rfa2 and Rfa3, as WT cells do ( Supplementary Figures S11C and  D). To confirm the result, we purified the WT and mutant RPA complexes from yeast cells and tested their ssDNAbinding abilities using the electrophoretic mobility shift assay (EMSA) (Supplementary Figure S12A). The addition of the WT RPA complex resulted in the shift of ssDNA in a dose-dependent manner, indicating the binding of RPA on ssDNA. In contrast, the 4KR mutant RPA complex exhibited reduced binding ability to ssDNA, and the defect was e v en more pronounced for the 4KQ mutant protein ( Figures  4 F and G). A similar result was obtained with a longer ss-DNA (90 nt) substrate (Supplementary Figures S12B, C). Reduced formation of the RPA-ssDNA complex for the 4KQ and 4KR mutant RPA was not due to any degradation of ssDNA by potential contamination of nucleases since digestion of the RPA-ssDNA complex with protease K completely r estor ed the le v els of free ssDNA (Supplementary Figure S12D).
We then directly measured the impact of acetylation on the ssDNA-binding ability of RPA. Purified GST-tagged RPA complex was first incubated with the 5 -biotinylated ssDNA (90 nt), followed by the addition of the acetyltransferase NuA4 and acetyl-CoA to initiate the acetylation reaction. By streptavidin pull-down assay, we observed that the inclusion of NuA4 in the reaction impaired the binding of RPA on ssDNA in a dose-dependent manner ( Figures  4 H and I, lanes 3-6). This is not due to any potential degradation of ssDNA by the purified protein complex (Supplementary Figure S12E). Thus, these results suggest that direct acetylation of RPA reduces its ssDNA binding ability, as previously noted ( 54 ).
Finally, we employed single-molecule magnetic tweezers (MT) to monitor the ssDNA binding kinetics for the purified WT, 4KR, or 4KQ mutant Rfa1. We labeled 12.5 k nt ssDNA molecules with digoxigenin and biotin at the 5 -and 3 -ends, respecti v el y, and stretched ssDN A molecules using MT at 4.2 pN, 21 • C, and pH 7.5, as previously reported (Figure 4 J) ( 34 ). We calculated the average values from multiple ssDNA molecules over time courses for each condition and plotted them as a time-e xtension curv e. The addition of WT Rfa1 resulted in a striking extension of ssDNA, reflecting that RPA efficiently assembled with ssDNA. Howe v er, the assemb ly of the 4KR and, more se v erely, the 4KQ mutant protein on ssDNA was impaired (Figure 4 K). Thus, we conclude that proper RPA acetylation and deacetylation are required for efficient RPA binding on ssDNA.
Proper RPA acetylation cooperates with Rtt105 to regulate RPA nuclear import and HR repair Next, we fused a yellow fluorescent protein (YFP) to the C-terminus of Rfa1 and examined whether proper acetylation affects RPA nuclear localization. As expected, the WT RPA is fully localized in the nucleus. Surprisingly, the 4KQ mutant protein localized in both cytoplasm and the nucleus ( Figures 5 A and B), as observed in cells lacking Rtt105, an RPA chaperone protein facilitating RPA nuclear import ( 32 , 34 ). Interestingly, ∼17% of 4KR mutant cells also exhibited altered RPA subcellular localization ( Figures 5 A and B and Supplementary Figure S13). These results suggest that proper acetylation and deacetylation of RPA are required for its normal nuclear import.
Previous studies showed that Rtt105 cooperates with the importin Kap95 to regulate RPA nuclear import ( 32 ).
Ther efor e, we tested whether the 4KR or 4KQ mutation affects RPA interaction with Rtt105 or Ka p95. Interestingl y, we found that the 4KQ mutant RPA exhibited an enhanced interaction with Rtt105 and, to a lesser e xtent, K ap95, as compared to the WT RPA ( Figures 5 C and D). In contr ast, these inter actions appear ed to r emain unaffected for the 4KR mutant protein. Consistently, using the microscale thermophoresis (MST) assay, we found that the 4KQ mutant RPA exhibited a higher affinity to 6xHis-Rtt105 ( K d ∼0.51 nM) than the WT RPA ( K d ∼1.06 nM) or the 4KR mutant RPA ( K d ∼1.17 nM) (Supplementary Figure S14). We specula ted tha t the slightly enhanced association might prolong the occupancy of Rtt105 or Kap95 by the 4KQ mutant RPA, impairing the dynamic hand-on and hand-off of RPA upon importing into the nucleus, thus leading to improper RPA nuclear localization.
Rtt105 promotes high-fidelity DNA replication or repair via regulating RP A ( 34 ). W e assessed the relationship between Rtt105 and RPA acetylation in regulating RPA functions. First, disruption of the Rtt105-RPA interaction by the r tt105-E171A L172A ( r tt105 -EL2A ) muta tion impaired DSB repair by ectopic recombination and the resistance to DNA dama ging a gents ( Figures 5 E and F) ( 32 , 34 ), while additional mutation of rtt105 -EL2A in the 4KQ mutant did not further reduce the HR repair rate and the drug resistance of the latter ( Figures 5 E and F), suggesting that Rtt105 and 4KQ act epistatically in the DNA damage response or HR repair. Second, the rtt105-EL2A or 4KQ single mutant and the double mutant exhibited similar phenotypes in mutation frequency and pattern (Figures 3 A-F and Supplementary Figure S7), especially with the signature of micro-homology-mediated large deletions or duplications ( 34 ). Finall y, both 4KQ m utation and R TT105 deletion can lead to improper RPA nuclear localization. These results suggest that proper RPA acetyla tion coopera tes with Rtt105 to facilitate RPA nuclear localization, HR repair, and the response to DNA damage.

Proper acetylation and deacetylation of human RPA promotes RPA and RAD51 loading at DNA breaks
Notably, three of the four acetylated lysines in yeast Rfa1 are conserved across species, which are equivalent to K259, K458 and K489 in human RPA1, the largest subunit of the human RPA complex (Supplementary Figure 1F). Except for the UV-induced acetylation on K163 ( 55 , 56 ), several other residues, including K259, K458 and K489 of human RPA1, were also identified as acetylated sites by mass spectrometry studies ( 54 , 83-85 ). To confirm this result, we immunoprecipita ted acetyla ted proteins with an anti-panacetyl-lysine antibody from HEK293T cells expressing the FLAG-tagged RPA1 and examined RPA acetylation using an anti-FLAG antibody. In line with previous studies, we successfully detected the acetylation of RPA1 in the presence of trichostatin A (TSA) and nicotinamide (NAM), the inhibitors of deacetylases. In addition, we noted that the signal was increased upon treatment with the DNAdama ging a gents MMS or 4NQO but not CPT (Figure 6 A) ( 55 , 56 ). Notably, we found that RPA acetylation was significantly reduced in cells expressing the plasmid-borne RPA1-3KR (K259R, K458R and K489R) mutant allele, suggesting that these residues are indeed acetylated in human cells (Figure 6 B).
Guided by our findings in yeast, we evaluated whether RPA acetyla tion af fects its loading and HR repair in human cells. First, we depleted endogenous RPA1 in Hela cells using a small hairpin interfering RN A (shRN A) and transfected these cells with either an empty vector or a plasmid expressing the shRNA-resistant WT, 3KR -or 3KQ -RPA1 allele (Figure 6 C). We observed that RPA foci formation was impaired in the reconstituted 3KR cells upon VP16 (etoposide) trea tment tha t can induce DSBs, and the defect was more pronounced in the reconstituted 3KQ cells as compared to the corresponding WT cells (Figures 6 D  and E). Howe v er, unlike the yeast 4KQ mutant, the human 3KR or 3KQ mutation did not change RPA nuclear localization (Figure 6 D). To confirm the above result, we incubated biotin-labeled ssDNA (30nt) coupled to magnetic beads with the whole cell extracts derived from VP16treated HEK293T cells expressing the FLAG-tagged WT or mutant RPA1 allele and performed an ssDNA pull-down assay. Notably, compared to the WT RPA, the 3KQ mutant protein exhibited an evident defect in binding ssDNA at both lower and higher concentrations, while the 3KR mutant protein only displayed a defect in binding ssDNA at the higher concentration ( Figures 6 F and G). Consequently, RAD51 f oci f ormation was impaired in both 3KR and 3KQ reconstituted cells upon VP16 treatment ( Figures 6 D  and E).

Proper acetylation and deacetylation of human RPA promote efficient HR repair
We then assessed the effect of RPA acetylation on HR repair using the I-Sce I DR-GFP reporter in U2OS cells ( 62 ). In this r eporter, r epair of the I-Sce I-induced DSB by HR r estor es the fluor escent signal of EGFP that can be monitored by flow cytometry (Supplementary Figure S15A). First, we depleted the endogenous RPA1 with siRNA and complemented the cells with an empty vector or a plasmid;  expr essing a siRNA-r esistant WT, 3KQ , or 3KR RPA1 allele. The reconstituted WT cells were fully proficient in repairing the I-Sce I-induced DSB by HR, while the reconstituted 3KQ or 3KR mutant allele exhibited a significant defect in the HR repair (Figure 7 A). Ne xt, we e valuated the effect of 3KQ or 3KR mutation on BIR using the EGFP-BIR-5085 reporter system in U2OS cells ( 64 ). In this reporter, the I-Sce I-induced DSB on the recipient chromatid can be r epair ed by BIR by copying the sequence on the template chromatid to the end via synthesis-dependent strand annealing (SDSA) or through copying a short segment of the donor template followed by end joining with the other end of the DSB (Supplementary Figure S15B). Compared to the reconstituted WT cells, the reconstituted 3KQ but not 3KR cells exhibited an a ttenua ted BIR repair ef ficiency (Figure  7 B). Thus, proper RPA acetylation and deacetylation are r equir ed for efficient RPA and RAD51 loading and DSB repair by gene conversion or BIR.

Constitutive acetylation of human RPA increases alt-EJ, SSA and genome instability
In parallel, we measured the effect of 3KQ or 3KR mutation on DSB repair by alt-EJ in U2OS cells that carry an EGFP-alt-EJ reporter and stab ly e xpress doxy cy cline (DOX)-inducible Cas9 and sgRN A ( 63 ). sgRN A directs the Cas9 endonuclease to the I-Sce I recognition site to generate a DSB upon induction with DOX ( Supplementary Figure S15C). We noted that the reconstituted 3KQ but not the 3KR cells had a slight increase in alt-EJ efficiency compared to the corresponding WT cells (Figure 7 C). Similarly, using an established SSA-EGFP reporter in U2OS cells ( 65 ), we measured the SSA-mediated repair of the I-Sce I-induced DSB (Supplementary Figure S15D). We found that the reconstituted 3KR cells did not af fect SSA ef ficiency, while the reconstituted 3KQ cells exhibited a significant increase in SSA repair (Figure 7 D). Accordingly, the Hela cells expressing the 3KQ mutant allele had an increased le v el of micronuclei, an indicator of chromosome loss (Figure 7 E).
In line with the role of proper RPA acetylation in DSB repair, cells expressing the 3KQ mutant allele had a severe defect in resistance to Iniparib, an anti-cancer agent, and the 3KR mutant lines also exhibited a r educed r esistance to the drug, albeit to a lesser extent (Figure 7 F). Together, these results suggest that proper acetylation and deacetylation of RPA ar e r equir ed to promote DSB r epair by the accurate gene conversion or BIR while suppressing the repair by the deleterious alt-EJ or SSA mechanism in human cells. Otherwise, it leads to genome instability. Thus, RPA acetylation appears to be a conserved regulatory mechanism in guarding genome stability.
Since replication gaps and the RPA le v el are important determinants impacting the killing efficiency of BRCA deficiency cells by PARP inhibitors( 86 ), we examined the response of the reconstituted WT, 3KQ , and 3KR Hela cells to replication stresses. Both the WT and the 3KR or 3KQ mutant RPA can localize to ssDN A ga ps at stalled forks induced by hydroxyurea (HU, 5 mM) at a comparab le le v el, and all these cell lines can r espond and r ecover from the HU treatment at similar kinetics, as reflected by the formation and removal of ␥ H2AX foci ( Supplementary   Figures S16A-D). Thus, proper acetylation and deacetylation of RPA appear particularly important for DNA repair rather than replication stresses in the BRCA proficient cells.

DISCUSSION
How RPA is finely regulated to ensure accurate DNA replication and repair remains poorly understood. In this study, we found that proper acetylation and deacetylation of RPA suppresses micro-homology-mediated large deletions or duplica tions and facilita tes the accura te DSB repair by gene conversion or BIR while discriminating the err or-pr one SSA or alt-EJ pathway in yeast (Supplementary Figures  S17A, B). In parallel, we showed that proper RPA acetylation or deacetylation also facilitates RPA and RAD51 loading, gene conversion, and BIR while inhibiting alt-EJ and SSA in human cells. Mechanistically, we showed that proper acetyla tion regula tes the nuclear import or ssDNA-binding ability of RPA. Thus, we re v eal a conserv ed function of RPA acetylation in promoting high-fidelity DNA replication or repair and regulating DSB repair pathway choice.
We found that RPA acetylation acts via at least two layers of mechanisms in yeast. First, proper RPA acetylation is r equir ed for normal RP A nuclear localization. RP A is primarily localized in the nucleus. Howe v er, how and why acetyla tion af fects RPA unclear localization is unclear. One possibility is that RPA accompanies a small fraction of damaged DNA that is disassociated from chromatin and has been passi v ely e xported to the cytoplasm. Alternati v ely, RPA acetylation may affect its turnover, as noted for the DNA damage-induced Sae2 acetylation that promotes Sae2 degradation via autophagy ( 87 ). Acetylationdependent control of protein subcellular localization has also been observed for other proteins, such as human MDM2 and E1A ( 88 , 89 ). Howe v er, mutation of the equivalent sites in human RPA1 does not affect RPA nuclear localiza tion, suggesting tha t RPA nuclear transport is differ entially r egulated between yeast and human. Second, we showed that proper acetylation and deacetylation of RPA facilitate its binding to ssDNA. These acetylation sites appear on the surfaces of the RPA structure and are close to the ssDNA binding surfaces (Supplementary Figure S1G), making it possible to regulate RPA ssDNA affinity by acetyla tion. Acetyla tion of human Dna2 and FEN1 also regulates their DNA binding a bilities ( 90 , 91 ). Nota bly, the 4KQ mutant is more defecti v e than the 4KR mutant in all tested phenotypes, suggesting that timely deacetylation of RPA is more critical. This is consistent with the extent of their defects in ssDNA binding or nuclear localization. Howe v er, it should be noted that arginine and glutamine may not always accurately mimic lysine and acetyl-lysine, respecti v ely. Ther efor e, we cannot exclude the possibility that the defects in the extreme 4KQ / 3KQ or 4KR / 3KR mutants are partially caused by the mutation itself. The human 3KR mutation causes a moderate defect in RPA loading and HR repair, while it does not impact BIR, alt-EJ, or SSA. BIR repairs one-ended DSBs, while alt-EJ and SSA join or anneal short complementary sequences, respecti v ely. These repair mechanisms occur at a much lower frequency than typical HR repair and usually involve a lesser amount of ssDNA. This probab ly e xplains w hy the 3KR m utation, w hich has a experiments. Data were analyzed by Students' t-test. n.s., no significance. * P < 0.05, ** P < 0.01. Scale bar: 10 m. ( F ) Survival curves for RPA1 siRNA cells complemented with the plasmid expressing the WT, 3KQ or 3KR plasmid upon exposure to iniparib trea tment. Da ta were analyzed by two-way ANOVA, n = 3. * P < 0.05, ** P < 0.01. moderate defect in RPA loading, does not affect the repair by BIR, alt-EJ or SSA.
Spontaneous micro-homology-mediated deletions or duplications are belie v ed to associa te with replica tion slippage or the repair by SSA. These types of mutations were seen in cells with mutations in DNA replication genes, such as RFA1 , RAD27 , POL3 , POL32 and RTT105 ( 34 , 74-79 ). A potential mechanism was proposed by Tishkoff et al. to explain how this type of e v ent is generated ( 77 ). They proposed that the lagging strand DN A pol ymerase Pol ␦ extends DNA into the downstream Okazaki fragment and dis-places it during DNA synthesis, generating a 5 -flap structure that can lead to a short duplication of the DNA sequence if left unprocessed. The 5 -ssDNA overhang can potentially be filled in by DNA synthesis and subsequently resected to expose the short repea ts a t the 3 -end for SSA ( 77 ). The pairing of the repeats out of the register can cause either duplications or deletions ( 77 ). The nucleases Dna2 and Rad27 cooperate to process the 5 -flaps at Okazaki fragments, while RPA is known to regulate the two enzymes ( 14 ). Ther efor e, in the 4KR or 4KQ mutant, reduced RPA binding on ssDNA may lead to aberrant processing of the displaced flaps or misalignment of short homologies, causing duplications or deletions. Micro-homology-mediated large duplications or deletions often occur in human cancers (2)(3)(4)(5). Our results and previous work suggest that mutations in RPA or its regulator could contribute to such genome rearrangements ( 34 , 74 ).
RPA plays a variety of functions at multiple steps of r eplication, r epair, and r ecombination. Ther efor e, the binding of RPA with ssDNA or proteins must be highly dynamic to allow efficient binding yet timely disassociation.
Howe v er, how these interactions are dynamically regulated in the chromatin context is poorly understood. Here, we found that constituti v e RPA acetylation impaired its ss-DNA binding ability yet enhanced its affinity with Rtt105. Thus, proper RPA acetylation r epr esents an important mechanism for regulating RPA activities or interactions with its partners. This regulation might facilitate passing ssDNA substrates to downstream replication or repair proteins or channeling ssDNA intermediates to a correct repair pathway.
The ssDNA-RPA complex is an important biological inter mediate for med throughout the life of cells ( 9 , 92 ). Due to its broad functions in DNA metabolism, RPA is tightly associated with carcinogenesis ( 9 , 20-23 ). RPA depletion or exhaustion can lead to genome instability, compromised DNA repair, and reduced resistance to radiation or chemotherapeutic agents in cancer cells ( 9 , 93 ). Ther efor e, both the nuclear import and proper binding of RPA on ss-DNA are essential for avoiding mutations or cancer. Notably, both aspects of RPA are affected by its acetylation, underscoring the importance of this regulation in guarding genome integrity and pre v enting cancer.

DA T A A V AILABILITY
The data underlying this article are available in the article and in the online supplementary data.