DNA Damage During the G0/G1 Phase Triggers RNA-Templated

DNA damage during the G0/G1 phase triggers PNAS PLUS RNA-templated, Cockayne syndrome B-dependent homologous recombination

Leizhen Weia,b, Satoshi Nakajimaa,b, Stefanie Böhma,b, Kara A. Bernsteina,b, Zhiyuan Shenc, Michael Tsangd, Arthur S. Levinea,b, and Li Lana,b,1

aUniversity of Pittsburgh Cancer Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA 15213; bDepartment of Microbiology and Molecular Genetics, University of Pittsburgh School of Medicine, Pittsburgh, PA 15219; cDepartment of Radiation Oncology, Rutgers Cancer Institute of New Jersey, Robert Wood Johnson Medical School, New Brunswick, NJ 08903; and dDepartment of Developmental Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA 15201

Edited by Philip C. Hanawalt, Stanford University, Stanford, CA, and approved May 27, 2015 (received for review April 13, 2015)

Damage repair mechanisms at transcriptionally active sites during found in patients, primarily UVC sensitivity and susceptibility to the G0/G1 phase are largely unknown. To elucidate these mecha- skin cancer (12). However, unlike human CSB patients, premature nisms, we introduced genome site-specific oxidative DNA damage aging, growth failure, and neurologic dysfunction are present only and determined the role of transcription in repair factor assembly. in mild forms in mice. Therefore, the function of CSB in radiation- We find that KU and NBS1 are recruited to damage sites independent induced damage repair is unclear. of transcription. However, assembly of RPA1, RAD51C, RAD51, and Given that CSB may be involved in strand break repair, it RAD52 at such sites is strictly governed by active transcription and might regulate the repair pathway choice at DNA strand breaks. requires both wild-type Cockayne syndrome protein B (CSB) function Recently, it was reported that CSB facilitates HR and represses and the presence of RNA in the G0/G1 phase. We show that the NHEJ in S and G2-phase cells (13). A member of the SNF2/SWI2 ATPase activity of CSB is indispensable for loading and binding of the family of ATPases, CSB contains a central ATPase domain recombination factors. CSB counters radiation-induced DNA damage flanked by N-terminal and C-terminal regions. To understand the

mechanism of CSB function in the repair of DNA strand breaks, CELL BIOLOGY in both cells and zebrafish models. Taken together, our results have we developed the damage at RNA transcription site (DART) uncovered a novel, RNA-based recombination mechanism by which approach. This approach allows reactive oxygen species (ROS)- CSB protects genome stability from strand breaks at transcriptionally induced DNA strand breaks to be created at a single defined active sites and may provide insight into the clinical manifestations of genomic locus with either active or inactive transcription in real Cockayne syndrome. time. Here we show that the recombination proteins RAD51, RAD52, RPA1, and RAD51C are significantly enriched at dam- DNA damage | transcription | recombination | CSB | RNA polymerase II age sites where active transcription takes place in G1/G0 phase cells. Recruitment of the HR factors relies on CSB function. NA double strand breaks (DSBs) are a most severe type of Consistently, transcription suppression sensitizes WT cells but not DDNA damage caused by endogenous metabolic processes CSB-deficient cells to IR, indicating that CSB contributes to cell and exogenous exposure to radiation and chemicals. Unrepaired survival through active transcription-induced repair of DNA DSBs induce genomic instability, carcinogenesis, and premature DSBs. The CSB-HR DNA repair pathway that we have identified aging. In mammalian cells, DSBs are repaired by either the non- in cells is also necessary for the normal development of zebrafish homologous end joining (NHEJ) or the homologous recombination (HR) pathway. Although it is a common understanding that HR Significance primarily takes place in response to strand breaks in the S-G2 phases of the cell cycle where the undamaged sister chromatids are Unrepaired DNA strand breaks at transcriptionally active sites are present as donor templates, recent studies have suggested that expected to be more deleterious than elsewhere in the genome homologous pairing also occurs during the G0/G1 phase and is because the integrity of the coding regions is likely to be com- associated with transcription (1), although the mechanisms remain promised. The commonly recognized homologous recombination to be elucidated. At active transcription sites, RNA polymerase II (HR) process occurs in the G2/M phase and depends on the pres- (RNAPOLII)canbypassbasemodificationssuchas8-oxoguanine ence of sister chromatids as a donor template. Our data demon- but not single strand breaks (SSBs) and DSBs (2–5), indicating that strate a Cockayne syndrome protein B- and RNA-dependent unrepaired strand breaks at transcriptionally active (TA) sites can mechanism of transcription-associated HR in the G0/G1 phase and be especially deleterious and may lead to secondary damage. offer insight into double strand break repair at sites of active The Cockayne syndrome B (CSB) gene is defective in ap- transcription. The data suggest that a deficiency in this repair proximately two-thirds of patients with Cockayne syndrome mechanism might explain why neurodegeneration as well as tu- (CS), an autosomal recessive disease with diverse clinical signs morigenesis may be associated with seemingly stable, terminally including severe growth failure, progressive neurodegeneration, differentiated (G0) cell populations. and hypersensitivity to sunlight. CSB has an established role in transcription-coupled nucleotide excision repair (TC-NER) of Author contributions: L.W. and L.L. designed research; A.S.L. helped design experiments; photo lesions. When RNA POLII is stalled at bulky lesions, CSB L.W. and L.L. performed research; S.N. helped perform experiments; L.W. and L.L. con- is loaded to facilitate NER of the transcribed strand (6, 7). As tributed new reagents/analytic tools; S.B. and K.A.B. helped prepare reagents; L.W. and noted, in addition to UV sensitivity, CS patients also manifest L.L. analyzed data; M.T. helped with zebrafish analysis; and L.W., A.S.L., and L.L. wrote severe neurodegeneration (8, 9), suggesting the importance of the paper. CS proteins in maintaining genome stability against a broad The authors declare no conflict of interest. spectrum of DNA damage. For example, CSB-defective cells are This article is a PNAS Direct Submission. also sensitive to ionizing radiation (IR) (10, 11), which is phe- Freely available online through the PNAS open access option. notypically distinctive from classic NER deficiencies and indi- 1To whom correspondence should be addressed. Email: [email protected]. cates that CSB function is not limited to UV-derived photo le- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. sions. In addition, CSB-deficient mice exhibit a subset of symptoms 1073/pnas.1507105112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1507105112 PNAS Early Edition | 1of10 Downloaded by guest on October 1, 2021 upon DNA damage. Taken together, our results reveal a novel locus. Compared with the active form of RNA POLII, YFP-POLII mechanism of transcription-associated HR. maintained its presence at TA-KR over time, indicating that damage leads to dephosphorylation but not degradation of POLII Results (Fig. 1C). To further verify the transcriptional recovery of the TRE DNA Damage at Active Transcription Sites Induces Reversible cassette with or without DNA damage, we analyzed levels of the Transcription Inhibition. The KillerRed protein (KR) is a modi- peroxisome signal tagged-CFP (CFP-SKL) signals in cytosol, which fied red fluorescent protein chromophore, which generates su- serves as a transcriptional readout of the TRE array. As expected, peroxide upon visible light irradiation (14, 15). When positioned CFP-SKLexpressionwasatsimilarlevelsinbothTA-KRand directly onto DNA, KR is capable of inflicting oxidative DNA TA-cherry cells before light induction (16, 18). However, 4 h after damage and strand breaks after white fluorescent light excitation. light irradiation, CFP-SKL expression in TA-KR cells decreased We took advantage of the light-dependent damage inducibility of dramatically and then recovered 24 h after damage induction (Fig. KR and developed the DART system as illustrated (Fig. 1A)to 1D and Fig. S1B). These observations suggest that the reversible study molecular responses at a single genome locus with control- transcription inhibition is a result of efficient repair processes at lable transcription (16, 17). A tandem tetracycline repressive ele- actively transcribed chromatin. ment (TRE) array cassette was integrated at a defined genomic Preferential Enrichment of HR Factors at Transcriptionally Active Sites locus in U2OS TRE cells. A CMV promoter is located after the of DNA Damage. To examine DSB repair factors assembled at KR- TRE repeats to allow the activation of transcription upon ectop- induced strand breaks during active transcription, we measured ical expression of the TRE-fusion proteins. When the U2OS TRE – the enrichment of NHEJ components and recombination proteins cellswereexposedtoa10 20 min light exposure, the tet-repressor RPA1, RAD51, RAD52, and RAD51C, at both tetR-KR and (tetR)- or tet-transcription activator (TA)-tagged KR proteins TA-KR sites after light activation. The recruitment of NHEJ factors (tetR-KR or TA-KR) induced similar numbers of DSBs at tran- Ku70 and DNA Ligase IV is indistinguishable between tetR-KR scriptionally inactive or active sites, respectively (Fig. 1A) (16). We and TA-KR sites regardless of transcription status (Fig. 2A). At used tetR-cherry and TA-cherry fusion proteins as undamaging the onset of DNA strand break formation, the MRN complex, controls and to mark the site of TRE integration at tetR and along with BRCA1, BRCA2, and CtIP, is involved in the initial TA, respectively. signal transduction and DNA end processing in early NHEJ and Upon KR light activation, we analyzed the active form of RNA HR processes (19, 20). NBS1, BRCA1, BRCA2, and CtIP are POLII by using the anti-RNA POLII CTD repeat YSPTSPS equally loaded to tetR-KR and TA-KR–mediated damage (Fig. (phosphor S) antibody (4H8). U2OS TRE cells were transfected S2A). Despite equivalent levels of initial damage at both tetR-KR with TA-KR and exposed to light for 10 min for damage induction and TA-KR loci, endogenous RAD51 is selectively recruited to and then harvested at indicated postlight recovery times to quantify DSBs at the sites of TA-KR (Fig. 2B). TetR-cherry and TA-cherry active transcription via 4H8 staining. We found that the intensity of fusion proteins were used as undamaged controls and to mark the the 4H8 focus at the sites of TA-KR exhibited drastic decay within site of TRE integration at tetR and TA, respectively. The absence 30 min and remained at near background level until 6 h postlight of the HR proteins at sites of TA-cherry excludes the possibility activation, reflecting the transcription suppression caused by DNA that HR factor recruitment is caused by active transcription per se damage induced by TA-KR. Twenty-four hours after light activa- (Fig. 2B). Preferential recruitment of GFP fusions of RPA1, tion, the intensity of the 4H8 focus recovered to 70% of the control RAD51C, and RAD52 was also observed at TA-KR sites but not at (Fig. 1B and Fig. S1A). This most likely indicates reactivation of tetR-KR or tetR/TA-cherry sites after light exposure (Fig. 2 C and D transcription due to the repair and restoration of the damaged and Fig. S2B). This was also seen in a second U2OS cell line (2-6-3)

Fig. 1. DNA damage at active transcription sites induces reversible transcription inhibition by the DART system. (A) Scheme of the DART system for inflicting genome locus-specific damage. KR generates ROS upon visible light exposure and induces localized damage at transcription either off or on sites. (B and C) TA-KR transfected U2OS TRE cells were irradiated with light for 10 min followed by the indicated postlight incubation time. Phosphorylated POLII detected by anti-4H8 (B) or YFP-POLII (C)is shown at the indicated time points. Quantification of the average fold increase of 4H8 foci intensity of 10 cells at TA-KR damage sites is shown, and error bars indicate the SEM of three independent experiments. Dephosphorylation of POLII but not the dissociation of POLII was observed upon damage. (D)The percentage of CKL-CFP positive cells (defined as CFP foci > 10/cell) in tetR-KR or TA-cherry/KR transfected U2OS TRE cells before and after damage induction is shown. n = 50, and error bars indicate the SEM of three independent experiments.

2of10 | www.pnas.org/cgi/doi/10.1073/pnas.1507105112 Wei et al. Downloaded by guest on October 1, 2021 PNAS PLUS

Fig. 2. HR factors are specifically recruited to transcriptionally active damage sites. (A) Recruitment of KU70 and NBS1 at TA-KR and tetR-KR damage sites is equal after damage. (B) RAD51 is specifically recruited to TA-KR but not TA-mCherry or tetR-KR/ mCherry. U2OS TRE cells were illuminated with 15W cool fluorescent white light for 10 min for damage production. The recruitment of RAD51 stained with its Ab is shown 30 min after damage production. (C) GFP tagged RPA1, RAD52, and RAD51C are specifically recruited to TA-KR with the same treatment as in A.(D) Quantification of percentage of cells with positive foci of the HR factor at the indicated spot 30 min after damage production. Error bars indicate the SEM of three independent experiments, n = 100. (E) Percentage of cells with γH2AX and 53BP1 foci (defined as the fold increase of foci at TA-KR sites >2 compared with intensity of background in the same cell) at the indicated time points after damage induction. (F) Quantification of average fold increase of foci intensity (n = 10) of the HR factor at sites of

TA-KR at the indicated recovery time after damage CELL BIOLOGY production. Error bars indicate the SEM of three independent experiments in E and F.

with a similar TRE copy number but a different chromosomal in- (Fig. S3A). Similar to what we observed in the unsynchronized tegration site (1p36) (18) (Fig. S2C). cell population, G0/G1 synchronization did not generate any The transcription recovery 24–48 h after damage indicates the detectable reduction in the percentages of RAD52 and RAD51C successful repair at the TA-KR site. To gain further understanding foci colocalizing with TA-KR (Fig. 3B), showing that recruitment of how DNA damage is resolved, the kinetics of surrogate markers of HR factors at TA-KR damaged sites does not appear to be γH2AX and 53BP1 were monitored in real time. As shown in Fig. confined to the S-G2 phases. The average diameter (DIA) of 2E, the kinetic profile of both γH2AX and 53BP1 foci resolution TA-KR before and after damage was measured; Fig. S3B shows in U2OS-TRE cells exhibited a steady rate of decline, reflecting the array length was not significantly altered as a result of repair. damage removal until 48 h after damage production at the transcriptionally active damaged genome locus. We next determined RNA Polymerase II Inhibition Affects the Recruitment of HR Factors. the kinetics of the recruitment of HR and NHEJ factors at To further confirm that transcription facilitates the recruitment TA-KR damaged sites. The focus intensities of RAD52, RAD51C, of HR factors specifically to TA-KR damaged sites, we treated and RPA1 were retained at the DNA damage site for at least 24 h cells with RNA POLII inhibitors that inhibit transcriptional and gradually returned to basal levels between 48 and 72 h after elongation (DRB) or both initiation and elongation (α-amanitin) light activation (Fig. 2F). This kinetic signature correlates with the (Fig. 3 C and D). Both DRB and α-amanitin treatment signifi- rate of damage removal (Fig. 2E), temporally consistent with the cantly diminished the recruitment of RAD52 and RAD51C at notion that HR is involved in the repair of transcription-blocking the site of TA-KR. In contrast, the ATM inhibitor KU55933 has oxidative damage. In contrast, KU proteins dissociated from TA- no detectable effect on the recruitment of RAD52 and RAD51C KR sites 30 min after damage induction but were present 8 h after to transcriptionally active damage sites. This result suggests that damage induction at tetR-KR sites (Fig. S2D). These observations loss of ATM-mediated transcription silencing (21, 22) is not confirm that the recruitment of HR factors at sites of TA-KR after sufficient to alter the recruitment of HR factors (Fig. 3 C and D), damage is more persistent and prevalent than that of the NHEJ suggesting that unlike canonical HR repair, a factor distinct from factors. Thus, HR proteins are preferentially recruited to active the DNA damage signaling pathway is responsible for the as- transcription sites upon DNA damage. sembly of HR factors at active transcription sites.

Recruitment of Recombinational Proteins at TA-KR Damaged Sites CSB Function Is Necessary for the Recruitment of Recombinational Occurs During the G0/G1 Phase. HR is commonly perceived as Factors. The sensitivity of CSB-defective cells to oxidative DNA occurring within late-S to G2 phases when sister chromatids are damage (10, 11) indicates that CSB deficiency is phenotypically present. We observed, however, that more than 90% of the cell distinct from classic NER deficiencies and indicates that CSB population exhibits HR factor loading at TA-KR damage sites in a function is not limited to monoadduct lesions. Therefore, tran- confluent/G0 cell population (Fig. 2D). Therefore, it is unlikely that scription blockage by DSBs may also elicit CSB function to re- transcription- and damage-induced recombination factor assembly cruit repair factors most suitable for strand break repair at at TA-KR damage sites is an S/G2 event. Consistently, we observed actively transcribed regions. recruitment of RAD52 and RAD51C to the TA-KR damaged sites We tested this hypothesis by analyzing the recruitment of HR in the G1 and early S marker Cdt1-stained cells after light exposure proteins to TA-KR sites in cells devoid of CSB via siRNA in- (Fig. 3A), suggesting that such recruitment does not only occur in hibition (Fig. 3E). In parallel, siRNAs for the NER/TC-NER late S and G2 phase cells. To further validate this observation, factors XPA and XPC (which bind monoadduct lesions), XPG we tested cells synchronized to G1 via double thymidine block (which introduces a nick at a monoadduct lesion), and CSA

Wei et al. PNAS Early Edition | 3of10 Downloaded by guest on October 1, 2021 Fig. 3. The recruitment of HR factors at TA-KR damage sites is dependent on active transcription and CSB. (A) Colocalization of RAD51C or RAD52 (green) and TA-KR (red) in cdt1 (blue) expressing U2OS-TRE cells 30 min after KR activation. (B) Quantification of the percentage of cells with positive foci at TA-KR (n = 100) in double thymidine synchronized G1 cells. Error bars indicates the SEM of three independent experiments, and the P values were determined by using Student’s two-tailed t test. (C and D) α-amanitin and DRB treatment abolish the recruitment of RAD51 and RAD51C at TA-KR. RAD52/RAD51C and TA-KR transfected U2OS TRE cells were pretreated with the polymerase II inhibitor (POLIIi) α-amanitin (100 μg/mL) for 1 h or DRB (20 μM) for 24 h or KU55933 (10 μM) for 1 h, then irradiated with light for 10 min followed by incubation for 30 min. Images and quantification of the fold increase of RAD52/RAD51C foci intensity at TA-KR damage sites are shown. (E)The fold increase of RAD52/RAD51C foci intensity of 10 cells at TA-KR damage sites is shown after siRNA treatment of the indicated protein. Data are repre- sented as mean ± SEM, and the P values were determined by using Student’stwo-tailedt test. (F)RNaseH (15 units) treatment for 15 min abolishes EU incorporation and the recruitment of RAD52 at TA-KR after damage induction. Images with light irradiation for 10 min followed by 30 min incubation after RNaseH treatment are shown. (G) The percentage of cells with RAD52 or RAD51C foci of 100 cells at TA-KR damage sites is shown after RNaseH treatment. Error bars indicate the SEM of three independent experiments.

(a CSB interacting protein at stalled transcription sites) were used induction following the RNaseH treatment, the localization of as controls (Fig. 3E). We found that only knockdown of CSB, RAD52 dramatically decreased at TA-KR damage sites, but that but not any other NER and TC-NER factors, significantly di- of RAD51C did not (Fig. 3 F and G), indicating that persistent minished the recruitment of RAD52 or RAD51C. It has been localization of RAD52 foci at transcriptionally active damage shown that DRB and α-amanitin cause no appreciable changes in sites requires the presence of an RNA template. cell cycle profiles (23). siCSB and RNA POLII inhibition, which inhibits the initiation or elongation of RNA POLII, did not alter CSB Is Localized at TA-KR Damaged Sites and Interacts with RAD52 the expression of RAD52, cell cycle profiles, or the localization and RAD51C After Damage. Loss of CSB leads to IR sensitivity and of RNA POLII itself under our experimental conditions (Fig. S3 attenuates HR protein recruitment. It is plausible that CSB C–E). Thus, recruitment of HR factors to TA-KR damage sites is functions upstream of HR factor assembly at the sites of DNA unlikely the result of cell cycle distribution or reduction of pro- damage. Accordingly, we tested and found that CSB is also tein expression level. These results indicate that both active recruited to TA-KR damage sites, colocalizing precisely with transcription and CSB are necessary for the enrichment of RAD52 and RAD51C (Fig. 4 A and B). The recruitment of CSB RAD52 and RAD51C at the site of TA-KR induced damage. is further described in SI Results and Discussion (Fig. S4). CSB enrichment was significantly diminished by RNA POLII in- Recruitment of RAD52 at TA-KR Damage Sites Requires an RNA hibition (Fig. 4C) but not by depletion of HR factors RAD51, Template. Given that active transcription but not the presence RAD52, and RAD51C (Fig. S4). of a sister chromatid is necessary for the recruitment of re- Because CSB is localized to TA-KR damage sites and serves as combinational factors, the enrichment of recombinational pro- a prerequisite for HR factor recruitment, we next tested whether teins might be involved in transcript-RNA–templated DNA HR protein recruitment is mediated by physical protein–protein recombination repair machinery as shown in a recent study in interactions. To this end, 293 cells stably expressing GFP-RAD52 yeast (1). To evaluate the influence of RNA on the enrichment or RAD51C were treated or mock-treated with IR (5 Gy) and of repair factors, we used RNaseH to degrade synthesized RNA subjected to coimmunoprecipitation. We found that both RAD52 (24) after damage induction. The RNaseH treatment abolished and RAD51C interact with either endogenous CSB (Fig. 4D)or the incorporation of RNA labeled by the uridine analog tagged CSB (Fig. 4E) independently of DNA (Fig. S4F)afterDNA 5-ethynyluridine (EU) at TA-KR sites after light irradiation (Fig. damage. Furthermore, DRB or RNaseH treatment abolished the 3F), indicating that RNA has been digested from the DNA-RNA interaction between RAD51C, RAD52, and CSB (Fig. 4 D and F). hybrid structure at TA-KR. With the same RNaseH treatment, Collectively, these findings suggest that recruitment of HR proteins the protein level of RAD52 was not affected (Fig. S3F). The to damaged TA-KR sites is dependent on active transcription and binding of the RAD51C complex to ssDNA and single-stranded their association with CSB. We also found that the C terminus of gaps in duplex DNA in an early stage of recombinational repair RAD51C is critical for its recruitment to transcriptionally active (25) implies that the initial recruitment of RAD51C to damage damage sites and its interaction with CSB (Fig. 4 G and H). The C might not be dependent on the presence of RNA. After damage terminus of RAD52, but not the N-terminus DNA binding domain

4of10 | www.pnas.org/cgi/doi/10.1073/pnas.1507105112 Wei et al. Downloaded by guest on October 1, 2021 PNAS PLUS CELL BIOLOGY

Fig. 4. CSB localizes at TA-KR damage sites and interacts with RAD52 and RAD51C after DNA damage. (A) Recruitment of CSB at sites of TA-KR after damage. FLAG-CSB and TA-cherry/KR were cotransfected into U2OS TRE cells. Twenty-four hours later, cells were irradiated with light for 10 min followed by 30 min of incubation; the recruitment of CSB at TA-KR sites is shown with FLAG-antibody. (B) Colocalization of CSB, RAD52, or RAD51C at sites of TA-KR after damage. FLAG-CSB, GFP-RAD52/RAD51C, and TA-KR were cotransfected into U2OS TRE cells. Twenty-four hours later, cells were irradiated with light for 10 min followed by 30 min of incubation and stained with FLAG-antibody. (C) The average fold increase of FLAG-CSB at sites of TA-KR after damage (30 min postlight incubation after 10 min of light incubation) and after treatment with α-amanitin (100 μg/mL) for 1 h, DRB (20 μM) for 24 h, or KU55933 (10 μM) for 1 h. Images were quantified by ImageJ (n > 50); SEM indicates three independent experiments. (D and E) GFP-RAD52/GFP-RAD51C stably expressed in 293 cells with or without 5 Gy IR and DRB treatment was analyzed by IP with anti-GFP and 1 h postirradiation incubation. Detection of endogenous CSB or HA-tagged CSB is shown. (F) RNaseH treatment of cells abolishes the interaction between CSB and RAD52. HA-CSB expressed in GFP-RAD52 stably expressing 293 cells treated with 5 Gy IR, analyzed by IP with anti-GFP and 1 h postirradiation incubation with or without RNaseH (15 units) treatment for 15 min. Detection of HA-tagged CSB is shown. (G) The recruitment of the C-terminus RAD52 and RAD51C at TA-KR 30 min after KR activation. (H) CSB interacts with the C terminus of RAD51C. GFP-RAD51C fragments were transiently transfected into 293 cells. Forty-eight hours after transfection, cells were irradiated with 5 Gy IR. Cell lysates were collected 1 h after irradiation and immunoprecipated by GFP/HA antibody. Western blotting by anti-HA, anti-GFP, and anti-CSB for immunoprecipitates and input are shown.

Wei et al. PNAS Early Edition | 5of10 Downloaded by guest on October 1, 2021 Fig. 5. The recruitment of CSB at TA-KR damage sites is via its C terminus and depends upon its ATPase activity. (A) Scheme of CSB mutants and deletions used in the study. (B) CSB aa 337–1493 is recruited to sites of TA-KR after damage. GFP-RAD52 or RAD51C, CSB or the indicated CSB mutant, and TA-KR were transfected into siCSB pretreated U2OS-TRE cells. The recruitment of GFP-RAD52 or RAD51C (green) and CSB (blue) at TA-KR is shown 30 min after KR activation. (C) CSB aa 337–1493 complements the recruitment of RAD51C and RAD52 after damage. The fold increase of foci intensity of GFP-RAD52 or RAD51C at TA-KR damage sites is quantified (n = 10). (D) The recruitment of GFP-RAD52 or RAD51C (green) and CSB K538R or UbΔ (blue) at TA-KR is shown 30 min after KR activation. (E)CSBUbΔ but not K538R complements the recruitment of RAD52 after damage. The fold average increase of foci intensity of GFP-RAD52 at TA-KR damage sites is quantified (n = 10). (F and G) GFP-RAD52/GFP-RAD51C stably expressed in 293 cells with or without 5 Gy IR was immunoprecipitated by anti-GFP with 1 h postirradiation incubation. Detection of CSB deletions (F) or mutants (G) by anti-FLAG, HA, or myc is shown.

6of10 | www.pnas.org/cgi/doi/10.1073/pnas.1507105112 Wei et al. Downloaded by guest on October 1, 2021 of RAD52, responds to TA-KR damage sites (Fig. 4G), consistent creased interaction with RAD52, whereas the UBΔ mutant main- PNAS PLUS with our observation that persistent localization of RAD52 foci at tained its interaction with RAD52 (Fig. 5G). Together, it appears transcriptionally active damage sites requires the presence of an that the ATPase domain but not the UB domain is essential for the RNA template, further indicating the importance of the RNA recruitment of RAD52 to damage sites during active transcription. template in facilitating repair processes at transcriptionally active damage sites. CSB Function Is Important for Efficient Repair of DNA Strand Break Damage at Active Transcription Sites. Recruitment of HR factors to The ATPase of CSB Is Necessary for Its Damage Response as Well as DNA damage at transcriptionally active regions may constitute a the Recruitment of RAD52/RAD51C. CSB harbors a seven helicase- potentially protective mechanism for countering DSBs. This like ATPase motif and a C-terminal ubiquitin-binding (UB) mechanism seems to have a characteristic dependency on CSB domain (Fig. 5A). To further confirm the role of CSB in recruiting function. Indeed, we observed that CSB suppression does lead to HR factors, we transfected full-length (FL) CSB, CSB aa 1–336, decreased HR efficiency as well as suppression of recombination and CSB aa 337–1493 into siCSB pretreated cells. The N-terminus factors but not NHEJ (Fig. S5A), measured by the HR and NHEJ CSB aa 1–336 was unable to complement the recruitment of assays (26). As such, cells devoid of CSB would be expected to RAD52/RAD51C, whereas the C-terminus CSB aa 337–1493 con- exhibit resistance to transcription inhibition-mediated damage taining the helicase and UB domains fully complemented RAD52 sensitization. We tested this premise by exposing the CSB null and RAD51C recruitment (Fig. 5 B and C). To further elucidate mutants UVS1KOSV (27) to the transcription inhibitor DRB and the functional domain requirement of CSB in regulating the re- measured clonogenic survival against IR. As shown in Fig. 6A, cruitment of HR factors, we expressed FL CSB, a CSB ATPase- CSB mutant cells complemented with HA-CSB are rendered diminished mutant (K538R), and a CSB UB deletion (UBΔ,aa more sensitive to IR, as are CSB-proficient HeLa cells and the 1–1400) into CSB siRNA pretreated cells. As shown in Fig. 5 D XPA mutant (XP12RO) (Fig. S5B), indicating that this is an and E, expression of the CSB K538R mutant failed to support the NER-independent mechanism. In contrast, mock-complemented recruitment of RAD52, indicating that the ATPase function of UVS1KOSV cells acquired no additional sensitivity to radiation CSB is essential for RAD52 recruitment. The CSB mutant lacking damage when exposed to DRB (Fig. 6A). We tested the sensitivity the UB domain complemented the recruitment of RAD52 to the in another patient-derived cell line, CS1AN (28, 29). Similar re- same extent as FL CSB did. Furthermore, CSB UBΔ is recruited sults were obtained in CS1AN cells (Fig. 6A). These results sug- to damage sites of TA-KR as well as CSB FL (Fig. 5D and Fig. gest that CSB provides a significant function in countering S4G). These results together indicate that the helicase activity is radiation-induced DNA damage at actively transcribed regions of necessary and the region of CSB (aa 337–1493) is sufficient in the genome, presumably through its role in transcription-associ- CELL BIOLOGY recruiting HR proteins to TA-KR sites. Consistently, the C-terminus ated recruitment of HR factors. CSB aa 337–1493, but not the N-terminal CSB aa 1–337, interacts Next, we measured γH2AX and 53BP1 foci resolution at with the HR proteins as determined by reciprocal co-IP (Fig. 5F). TA-KR sites in G0/G1 cells under the condition of FBS starvation, We also found that the CSB K538R mutant showed a slightly de- which prevents cell cycle progression. Our results (Fig. S5C, Left)

Fig. 6. CSB is important for cell survival and in vivo after damage coupled with active transcription. (A) UVs1KOSV and HA-CSB expressing UVs1KOSV cells (Left) and CS1AN and HA-CSB expressing CS1AN cells (Right) were pretreated with or without DRB (20 μM) for 24 h, then irradiated with IR at the indicated dose. Colony forming assays were performed, and the survival fraction is shown. (B) Persistent γH2AX and 53BP1 signals at sites of TA-KR after damage without HR and/or CSB. U2OS TRE cells were treated with the indicated siRNA, then transfected with TA-KR. Cells were exposed to white light for 10 min followed by 48 h of postlight incubation, then fixed and immunostained with anti-γH2AX and anti- 53BP1. The percentage of γH2AX and 53BP1 foci postive cells (defined as the fold increase of foci at TA-KR sites compared with background intensity being >2) was quantified. (C) Zebrafish embryos injected with CSB-MO/con-MO or WT were irradiated with or without 5 Gy IR at 6 h postfertilization (hpf). Images of three levels of abnormalities (L1, L2, and L3) at 72 hpf are shown. (D) Percentage of embryos in each of the categories. Error bars indicate the SEM of three independent experiments, and the P values were determined by using Student’stwo-tailedt test. (E and F) WesternblotofCSBandγH2AX in embryos in D (E) or in siCSB treated 293 cells with or without irradiation (F).

Wei et al. PNAS Early Edition | 7of10 Downloaded by guest on October 1, 2021 show that these two surrogate markers’ rate of decline is similar to (33, 34), the low abundance of BRCA2 during G1/G0 perhaps that in unsynchronized cells (Fig. 2E), indicating that CSB-medi- may render it insufficient for HR processing (20). Therefore, al- ated repair still occurs in G0/G1 cells. We also measured γH2AX ternative mechanisms may compensate BRCA2 function in the and 53BP1 foci resolution at TA-KR sites in cells depleted of CSB G0/G1 phase and/or as a backup for BRCA2 in S phase. Also, the and/or RAD52 and RAD51C. The induction of γH2AX53BP1 absence of sister chromatids necessitates an alternative template(s) foci intensity at the onset of damage production was comparable for high-fidelity repair of DSBs, which has recently been sug- among single, double, and triple knockdowns (Fig. S5C, Right). gested in yeast (1). In transcript RNA-templated DNA recom- RAD52 depletion resulted in a marked increase (20–45%) of cells γ bination and repair, both yeast and human RAD52 proteins have showing strong residual levels of both H2AX and 53BP1 48 h been shown to be necessary for catalyzing the annealing of RNA after the damage (Fig. 6B). Concurrent double (CSB and RAD52, to a DSB-like DNA end (1), supporting our observations that the CSB and RAD51C) and triple knockdown (CSB, RAD52, and stable localization of RAD52 in transcription-dependent HR RAD51C) resulted in a similar percentage of cells with unrepaired relies on the presence of RNA. How DSBs are channeled into the residual DNA damage similar to that of the RAD52 knockdown (Fig. 6B), suggesting that CSB, RAD52, and RAD51C function in recombination process at active transcription sites in G0/G1 cells the same mechanism and are required for efficient strand break is further discussed in SI Results and Discussion. repair at transcriptionally active damage sites. The Role of CSB Is to Protect Genome Stability at Transcriptionally CSB Deficiency Sensitizes Zebrafish Embryos to DNA Radiation Damage Active Damage Sites, Mediated by HR Machinery. We found that the During the Early Development Stage. To further substantiate the role recruitment of RPA, RAD51, RAD51C, and RAD52 to damage of CSB function in radiation damage repair in vivo, we depleted sites with active transcription is a downstream event of CSB CSB in zebrafish using antisense morpholino oligonucleotides (MO). CSB-MO and control scrambled-MO oligos were injected into 1–2 cell stage embryos and assessed for their morphologic aberrations defined by three levels: L1 embryos exhibited smaller heads; L2 exhibited hydrocephalus and curved tails; and L3 was the most severe with markedly malformed, shortened, and curved body axes. Seventy-two hours postfertilization (hpf), untreated CSB-MO embryos appeared to have moderately increased frequencies of morphologic aberrations (Fig. 6C), which appear to recapitulate the congenital and developmental manifestations of CSB patients. CSB-deficient embryos showed a dramatic increase (∼23-fold) in morphologic aberrations when exposed to 5 Gy of IR (Fig. 6D). This result indicates an indispensable role of CSB in countering radiation DNA damage. To validate the developmental/morphological readouts at the molecular level and to visualize increased DNA DSBs, we analyzed γH2AX levels in both control- or CSB-MO–injected embryos (Fig. 6E) and siCSB treated 293 cells (Fig. 6F). We found that the loss of CSB alone slightly elevated basal γH2AX but substantially increased γH2AX levels after IR treatment (CSB-MO + IR) in embryos (Fig. 6E) and human cells (Fig. 6F). These results suggest that defective repair of radiation damage is a prevalent cause of radiation-induced embryonic aberrations, underscoring the function of CSB in repair of strand breaks. Discussion Cell Cycle Progression-Independent and RNA Template-Based HR at Transcriptionally Active Sites. Our study suggests a CSB-dependent and RNA-templated recombination mechanism of DSB repair at active transcription sites in a G0/G1 cell population. An over- whelming majority of somatic cells reside in the G1/G0 phase of the cell cycle, where canonical homologous recombination, re- lying strictly on the presence of sister chromatids during S phase, is unattainable. However, error-free protection of the somatic cellular genome from the mutagenic and/or deleterious effects of DSBs is vitally important in maintaining somatic cell functions and to prevent activation of oncogenes and unnecessary cell cycle reentry. Increasing evidence demonstrates that the recombination process is initiated and active during G0/G1. For example, homologous chromosomes indeed contact each other at IR- or endonuclease I-PpoI–induced DSB sites in human somatic cells. The chromosomal contact requires short homologous seg- ments as well as formation of a DSB break occurring in an active gene. Furthermore, these contacts do not occur at I-PpoI sites when the DSB is introduced in an intergenic DNA sequence (30, 31). Recently, it was reported that in all cell cycle phases, resection-dependent DSB repair pathways are required to resolve complex DSBs in the G1 phase (32). We have observed that the recruitment of HR factors to TA-KR damage sites occurs during G0/G1 and an RNA-dependent localization of RAD52 at sites of Fig. 7. Model of the ROS damage repair process at transcriptionally active damage (Fig. 3). Although BRCA2 recruits RAD51 in S phase damage sites.

8of10 | www.pnas.org/cgi/doi/10.1073/pnas.1507105112 Wei et al. Downloaded by guest on October 1, 2021 function. This result suggests a CSB-mediated HR pathway. The DSB repair deficiencies are often linked to developmental PNAS PLUS fact that RNA POLII inhibitors sensitize CSB-proficient cells to abnormalities of the brain because nondividing cells do not IR, but not CSB-deficient cells (Fig. 6), indicates that CSB plays possesssisterchromatidsforrepair.Therefore,protecting an important role in enabling cell survival by facilitating tran- genome stability at transcriptionally active sites and coding scription-induced recombination during DNA damage repair. sequences in G0/G1 is critical for neurons. Our data have We have analyzed acetylation of histone H3 lysine 9 (AcK9H3) uncovered a mechanism by which HR factors are preferentially which is preferentially present at euchromatin and found that the recruited to transcriptionally active DNA damage sites through intensity of AcK9H3 signals was not affected by DRB, α-ama- protein–protein interactions with CSB and may subsequently nitin, or CSB depletion (Fig. S5D). These results indicate that facilitate recombinational repair of DNA damage at these chro- the presence of active RNA POLII (Fig. 1) after damage might mosome domains (Fig. 7). Premature aging, growth failure, and − − maintain the histone modification at sites of active transcription neurologic dysfunction are present only in mild forms in CSB / sites to provide an open chromatin structure and to allow the mice. This might be due to the transcriptional differences be- subsequent recruitment of repair proteins to sites of damage. tween rodent and human (42). The phenotype of CSB-depleted The damage-induced physical interaction between CSB and zebrafish exhibiting severe developmental defects after DNA RAD52/RAD51C (Figs. 4 and 5) raised the possibility that CSB radiation damage possibly reflects the deficient repair of damage may recruit HR factors to transcriptionally active damage sites. incurred at regions of active transcription critical in main- In addition to the interaction between CSB and components taining normal embryonic development and differentiation. of the transcription machinery such as RNA POLII, the in- Thus, the CSB-HR repair mechanism may be considered as a volvement of CSB in TC-NER relies on its interaction with NER molecular basis for the developmental phenotypes and IR factors (35, 36), supporting the notion that CSB-mediated repair sensitivity observed in CS patients. process may be modulated via pathway-specific interacting partners. Future studies are needed to identify functional mutant(s) Materials and Methods which can distinguish the role of CSB in HR-associated strand Cell Lines, Plasmids, and Microscopy. U2OS, Flp-in 293, CS1ANSV, and UVS1KOSV break repair from UVC-induced damage. cells were cultured in DMEM with 10% (vol/vol) FBS at 37 °C. pBROAD3 tetR-KR The ATPase domain of CSB is critical in its response to and pBROAD3 tetR-cherry and pBROAD3 TA-KR and pBROAD3 TA-cherry were transcriptionally active damage sites and to the recruitment of constructed as previously described (16). Other plasmids were described HR factors, as shown in Fig. 5. It has been reported that CSB in SI Materials and Methods. siRNA and antibodies used in the study are UBΔ becomes immobilized at UV lesions and impedes tran- shown in Tables S1 and S2, respectively. Olympus FV2000 confocal mi- Δ croscopy was used for imaging studies.

scription reactivation (37, 38). Although UB CSB is effi- CELL BIOLOGY ciently recruited to damage sites, here we found that compared with FL CSB, it does not dissociate from damage sites 24 h Zebrafish Study. Zebrafish (AB) were maintained as standards and described post-light exposure (Fig. S4G). Therefore, the UB domain of in SI Materials and Methods. All animal experiments were approved by and CSB is not required for the initial recruitment but is required conducted in accordance with the guidelines established by the Institutional for the dissociation of CSB from damage sites. CS1AN fibro- Animal Care and Use Committee at the University of Pittsburgh. blast cells possess a CSB allele with A to T transversions at nucleotide 1088, resulting in a stop codon at position 337 (39, ACKNOWLEDGMENTS. We thank Dr. Jesper Q. Svejstrup for providing the – CSB Ub deletion mutant, Dr. Susan Janicki for providing the U2OS 2-6-3 cell 40). N-terminus CSB 1 337 aa, which mimics the expression of line, Dr. Vilhelm A. Bohr for providing the CS1ANSV cell line, Dr. Liang Wei CSB in the CS1AN fibroblast cells, did not respond to damage. for aiding us with ionizing irradiation, and Jacqueline Starr Welty for help Another CS gene, CSA, promotes the ubiquitination and sub- with manuscript editing. This work was supported in part by grants from sequent proteasomal degradation of CSB (41). siCSA did not the National Institutes of Health (AG045545-01 to L.L. and ES024872 to K.A.B.), affect HR factor recruitment to transcriptionally active dam- the Competitive Medical Research Fund of the University of Pittsburgh age sites in our study (Fig. 3F). Because CSB degradation is Medical Center (to S.N.), and the Ellison Medical Foundation (AG-NS-0935-12 essential for the recovery of RNA synthesis after transcrip- to K.A.B.) and the V Foundation (V Scholar Award to K.A.B.). This Δ project used the University of Pittsburgh Cancer Institute (UPCI) Imaging tion-coupled repair, the failed dissociation of CSB UB raises Facility and UPCI Cytometry Facility, supported in part by National Instit- the possibility that CSA may also have an important role in utes of Health Grant P30CA047904. Funding for the open access charge is repair completion. from National Institutes of Health/AG045545-01.

1. Keskin H, et al. (2014) Transcript-RNA-templated DNA recombination and repair. 12. Dollé ME, et al. (2006) Increased genomic instability is not a prerequisite for short- Nature 515(7527):436–439. ened lifespan in DNA repair deficient mice. Mutat Res 596(1-2):22–35. 2. Zhou W, Doetsch PW (1994) Transcription bypass or blockage at single-strand breaks 13. Batenburg NL, Thompson EL, Hendrickson EA, Zhu XD (2015) Cockayne syndrome on the DNA template strand: Effect of different 3′ and 5′ flanking groups on the T7 group B protein regulates DNA double-strand break repair and checkpoint activation. RNA polymerase elongation complex. Biochemistry 33(49):14926–14934. EMBO J 34(10):1399–1416. 3. Kuraoka I, et al. (2007) RNA polymerase II bypasses 8-oxoguanine in the presence of 14. Bulina ME, et al. (2006) A genetically encoded photosensitizer. Nat Biotechnol 24(1): – transcription elongation factor TFIIS. DNA Repair (Amst) 6(6):841–851. 95 99. 4. Tornaletti S, Maeda LS, Kolodner RD, Hanawalt PC (2004) Effect of 8-oxoguanine on 15. Carpentier P, Violot S, Blanchoin L, Bourgeois D (2009) Structural basis for the pho- – transcription elongation by T7 RNA polymerase and mammalian RNA polymerase II. totoxicity of the fluorescent protein KillerRed. FEBS Lett 583(17):2839 2842. 16. Lan L, et al. (2014) Novel method for site-specific induction of oxidative DNA damage DNA Repair (Amst) 3(5):483–494. 5. Charlet-Berguerand N, et al. (2006) RNA polymerase II bypass of oxidative DNA reveals differences in recruitment of repair proteins to heterochromatin and euchromatin. Nucleic Acids Res 42(4):2330–2345. damage is regulated by transcription elongation factors. EMBO J 25(23):5481–5491. 17. Lan L, et al. (2010) The ACF1 complex is required for DNA double-strand break repair 6. Hanawalt PC (1996) Role of transcription-coupled DNA repair in susceptibility to en- in human cells. Mol Cell 40(6):976–987. vironmental carcinogenesis. Environ Health Perspect 104(Suppl 3):547–551. 18. Janicki SM, et al. (2004) From silencing to gene expression: Real-time analysis in single 7. Kitsera N, et al. (2014) Cockayne syndrome: Varied requirement of transcription- cells. Cell 116(5):683–698. coupled nucleotide excision repair for the removal of three structurally different 19. Wang H, et al. (2012) CtIP protein dimerization is critical for its recruitment to adducts from transcribed DNA. PLoS ONE 9(4):e94405. chromosomal DNA double-stranded breaks. J Biol Chem 287(25):21471–21480. 8. Weidenheim KM, Dickson DW, Rapin I (2009) Neuropathology of Cockayne syndrome: 20. Vaughn JP, et al. (1996) Cell cycle control of BRCA2. Cancer Res 56(20):4590–4594. Evidence for impaired development, premature aging, and neurodegeneration. Mech 21. Clerici M, Trovesi C, Galbiati A, Lucchini G, Longhese MP (2014) Mec1/ATR regulates – Ageing Dev 130(9):619 636. the generation of single-stranded DNA that attenuates Tel1/ATM signaling at DNA 9. Tan WH, Baris H, Robson CD, Kimonis VE (2005) Cockayne syndrome: The developing ends. EMBO J 33(3):198–216. phenotype. Am J Med Genet A 135(2):214–216. 22. Smith J, Tho LM, Xu N, Gillespie DA (2010) The ATM-Chk2 and ATR-Chk1 pathways in 10. Cramers P, et al. (2011) Impaired repair of ionizing radiation-induced DNA damage in DNA damage signaling and cancer. Adv Cancer Res 108:73–112. Cockayne syndrome cells. Radiat Res 175(4):432–443. 23. Aymard F, et al. (2014) Transcriptionally active chromatin recruits homologous re- 11. Ropolo M, et al. (2011) Defective resolution of pH2AX foci and enhanced DNA combination at DNA double-strand breaks. Nat Struct Mol Biol 21(4):366–374. breakage in ionizing radiation-treated cockayne syndrome B cells. IUBMB Life 63(4): 24. Schein CH (1997) From housekeeper to microsurgeon: The diagnostic and therapeutic 272–276. potential of ribonucleases. Nat Biotechnol 15(6):529–536.

Wei et al. PNAS Early Edition | 9of10 Downloaded by guest on October 1, 2021 25. Masson JY, et al. (2001) Identification and purification of two distinct complexes 37. Anindya R, et al. (2010) A ubiquitin-binding domain in Cockayne syndrome B required containing the five RAD51 paralogs. Genes Dev 15(24):3296–3307. for transcription-coupled nucleotide excision repair. Mol Cell 38(5):637–648. 26. Gunn A, Stark JM (2012) I-SceI-based assays to examine distinct repair outcomes of 38. Gray LT, Weiner AM (2010) Ubiquitin recognition by the Cockayne syndrome group B – mammalian chromosomal double strand breaks. Methods Mol Biol 920:379 391. protein: Binding will set you free. Mol Cell 38(5):621–622. 27. Horibata K, et al. (2004) Complete absence of Cockayne syndrome group B gene 39. Leadon SA, Cooper PK (1993) Preferential repair of ionizing radiation-induced dam- product gives rise to UV-sensitive syndrome but not Cockayne syndrome. Proc Natl age in the transcribed strand of an active human gene is defective in Cockayne Acad Sci USA 101(43):15410–15415. syndrome. Proc Natl Acad Sci USA 90(22):10499–10503. 28. Scheibye-Knudsen M, et al. (2014) A high-fat diet and NAD(+) activate Sirt1 to rescue 40. Selzer RR, et al. (2002) Differential requirement for the ATPase domain of the premature aging in Cockayne syndrome. Cell Metab 20(5):840–855. 29. Troelstra C, et al. (1992) ERCC6, a member of a subfamily of putative helicases, is Cockayne syndrome group B gene in the processing of UV-induced DNA damage and – involved in Cockayne’s syndrome and preferential repair of active genes. Cell 71(6): 8-oxoguanine lesions in human cells. Nucleic Acids Res 30(3):782 793. 939–953. 41. Groisman R, et al. (2006) CSA-dependent degradation of CSB by the ubiquitin-pro- 30. Gandhi M, et al. (2012) Homologous chromosomes make contact at the sites of teasome pathway establishes a link between complementation factors of the Cock- double-strand breaks in genes in somatic G0/G1-phase human cells. Proc Natl Acad Sci ayne syndrome. Genes Dev 20(11):1429–1434. USA 109(24):9454–9459. 42. Lin S, et al. (2014) Comparison of the transcriptional landscapes between human and 31. Gandhi M, Evdokimova VN, Cuenco KT, Bakkenist CJ, Nikiforov YE (2013) Homolo- mouse tissues. Proc Natl Acad Sci USA 111(48):17224–17229. gous chromosomes move and rapidly initiate contact at the sites of double-strand 43. Wei L, et al. (2011) BRCA1 contributes to transcription-coupled repair of DNA damage ₀ – breaks in genes in G -phase human cells. Cell Cycle 12(4):547 552. through polyubiquitination and degradation of Cockayne syndrome B protein. Can- 32. Averbeck NB, et al. (2014) DNA end resection is needed for the repair of complex cer Sci 102(10):1840–1847. lesions in G1-phase human cells. Cell Cycle 13(16):2509–2516. 44. Gao M, et al. (2012) A novel role of human holliday junction resolvase GEN1 in the 33. Feng Z, et al. (2011) Rad52 inactivation is synthetically lethal with BRCA2 deficiency. maintenance of centrosome integrity. PLoS ONE 7(11):e49687. Proc Natl Acad Sci USA 108(2):686–691. 45. Lan L, et al. (2004) In situ analysis of repair processes for oxidative DNA damage in 34. Chun J, Buechelmaier ES, Powell SN (2013) Rad51 paralog complexes BCDX2 and CX3 mammalian cells. Proc Natl Acad Sci USA 101(38):13738–13743. act at different stages in the BRCA1-BRCA2-dependent homologous recombination 46. Liu J, Meng X, Shen Z (2002) Association of human RAD52 protein with transcription pathway. Mol Cell Biol 33(2):387–395. – 35. Nakatsu Y, et al. (2000) XAB2, a novel tetratricopeptide repeat protein involved factors. Biochem Biophys Res Commun 297(5):1191 1196. in transcription-coupled DNA repair and transcription. J Biol Chem 275(45): 47. Pfister SX, et al. (2014) SETD2-dependent histone H3K36 trimethylation is required for 34931–34937. homologous recombination repair and genome stability. Cell Reports 7(6):2006–2018. 36. Bradsher J, et al. (2002) CSB is a component of RNA pol I transcription. Mol Cell 10(4): 48. Mott C, Symington LS (2011) RAD51-independent inverted-repeat recombination by a 819–829. strand-annealing mechanism. DNA Repair (Amst) 10(4):408–415.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1507105112 Wei et al. Downloaded by guest on October 1, 2021