1   Systematic E2 screening reveals a UBE2D-RNF138-CtIP axis promoting DNA repair Christine K Schmidt1,4, Yaron Galanty1,4,5, Matylda Sczaniecka-Clift1, Julia Coates1, Satpal Jhujh1, Mukerrem Demir1, Matthew Cornwell1, Petra Beli3 and Stephen P Jackson1,2,5 1The Wellcome Trust/Cancer Research UK Gurdon Institute and Department of Biochemistry, University of Cambridge, CB2 1QN Cambridge, UK 2The Wellcome Trust Sanger Institute, Hinxton, CB10 1SA Cambridge, UK 3Institute of Molecular Biology (IMB), 55128 Mainz, Germany 4These authors contributed equally to this work. 5Correspondence: y.galanty@gurdon.cam.ac.uk 5Correspondence: s.jackson@gurdon.cam.ac.uk   2   Ubiquitylation is crucial for proper cellular responses to DNA double-strand breaks (DSBs). If unrepaired, these highly cytotoxic lesions cause genome instability, tumourigenesis, neurodegeneration or premature ageing. Here, we conduct a comprehensive, multilayered screen to systematically profile all human ubiquitin E2- enzymes for impacts on cellular DSB responses. Applying a widely applicable approach, we use an exemplary E2 family, UBE2Ds, to identify ubiquitylation-cascade components downstream of E2s. Thus, we uncover the nuclear E3-ligase RNF138 as a key homologous recombination (HR)-promoting factor that functions with UBE2Ds in cells. Mechanistically, UBE2Ds and RNF138 accumulate at DNA-damage sites and act at early resection stages by promoting CtIP ubiquitylation and accrual. This work supplies insights into regulation of DSB repair by HR. Moreover, it provides a rich information resource on E2s that can be exploited by follow-on studies. In response to DSBs, cells mediate a complex, highly regulated DNA-damage response (DDR) to sense DNA lesions and arrest cell-cycle progression to allow DNA repair, or initiate apoptosis1,2. One principal DSB-repair mechanism is non-homologous end-joining (NHEJ), which is active throughout the cell-cycle and occurs through two pathways: classical NHEJ and alternative/microhomology-mediated end-joining (MMEJ)3. The other, homologous recombination (HR), requires a sister chromatid as template and is limited to S/G2 cell-cycle phases2. Key to initiating HR is DNA-end resection promoted by CtIP (RBBP8) recruitment to DSB sites2,4,5, yielding single-stranded DNA (ssDNA) that is rapidly bound by RPA and subsequently replaced by RAD51, leading to strand invasion and ensuing HR processes2. Precisely how CtIP and early HR events are regulated, however, is not known.   3   Among the earliest DDR events is activation of the protein kinases ATM, ATR and DNA- PKcs6. Activated ATM phosphorylates histone H2AX (H2AFX) to yield γH2AX at DSBs, which in turn recruits numerous DDR modulators into ionizing-radiation induced foci (IRIF)1,2,7. Recently, ubiquitylation has emerged as a key DDR regulator8. Mediated by two E1 activating, ~40 E2 conjugating and >600 E3 ligating enzymes, posttranslational modification by ubiquitin modulates the stability, localization, activity or interaction properties of proteins9,10. E2s, previously regarded as mere basic ubiquitylation components, have recently emerged as key ubiquitylation mediators. E2-E3 pairs exist in various combinations, controlling the switch between ubiquitin-chain initiation and elongation, and determining ubiquitylation processivity and linkage specificity11,12. Despite the above, few systematic analyses of human E2s or E3s have been conducted, although recent proteomic approaches have identified hundreds of DDR-regulated ubiquitylation substrates13, suggesting that E2s and many E3s with DDR roles await discovery. Here, using a three-module, siRNA-based, semi-automated analysis pipeline, we systematically interrogate E2s for DSB-response functions in human cells. In addition to identifying various E2s with previously non-established DDR functions, we show how such data can be used to identify E2-E3-substrate ubiquitylation pathways. Specifically, by applying data-mining and phenotypic-mimicry approaches, we identify the ubiquitin-E3- ligase RNF138 as a DDR factor that cooperates with UBE2Ds to promote HR by stimulating CtIP ubiquitylation and accrual at DSB sites.   4   RESULTS Systematic multi-module screen for DDR E2s To identify E2 DDR components, we performed loss-of-function screens in U2OS cells with siRNA pools targeting 37 E2s, control siRNA (siCTRL) and validated siRNAs against known DDR factors (Supplementary Table 1). We evaluated siRNA-treatments of all known ubiquitin-E2s and the two NEDD8-E2s, UBE2M and UBE2F. To obtain comprehensive DDR “fingerprints” we multiplexed the screen into three modules. Module 1 evaluated impacts of E2 depletions on IRIF kinetics for DDR factors/markers by semi-automated, quantitative high-content, high-throughput (HC/HT) microscopy (Fig. 1a, and Supplementary Fig. 1a-b). 96-well plates containing duplicates for each siE2 were irradiated or not, fixed and assessed for γH2AX IRIF (a DNA-damage marker), 53BP1 (TP53BP1) whose IRIF require ubiquitin8, and conjugated ubiquitin (FK2 antibody). Validating the screening pipeline, depleting the ubiquitin-E3s RNF8 and RNF168, or depleting UBE2N that has known DDR connections, strongly impaired FK2 and 53BP1 but not γH2AX IRIF (Fig. 1b)8. Notably, siRNAs targeting 19 E2s diminished induction of FK2 IRIF 30 minutes after IR, to <60% of siCTRL cells; and surprisingly given established links between conjugated ubiquitin and 53BP1 IRIF8, without markedly affecting 53BP1 or γH2AX foci (Fig. 1b, Supplementary Fig. 1c, PubChem BioAssays and Discussion). Depleting UBE2D, -R, -J, -Q1, -Q2, UBE2O or -K, reduced induction of FK2 IRIF to <25% of siCTRL, while siUBE2D3 reduced FK2 IRIF induction by >97% (Fig. 1b-c; siUBE2D3 is highlighted because we subsequently found that it depleted all UBE2D family members). Collectively, these results suggested that many E2s impact on DDR IRIF, particularly FK2- ubiquitin foci (Fig. 1b-c, Supplementary Fig. 1c and PubChem BioAssays).   5   Module 2 evaluated HR and mutagenic end-joining (mutEJ) by the traffic-light-reporter (TLR) system in U2OS cells14,15 (Fig. 2a; as HR operates only in S/G2, data were corrected to flow-cytometry S/G2 values). Positive controls of ATM inhibition or CtIP depletion significantly decreased HR4,16, while DNA-PK inhibition decreased mutEJ, likely via DNA- PK retention at DSBs impeding access of other factors. Furthermore, the proteasome inhibitor MG132 drastically impaired HR, highlighting the importance of ubiquitylation in HR17. In agreement with this, we found that depleting many E2s markedly inhibited HR (Fig. 2b, Supplementary Table 2 and PubChem BioAssays). Furthermore, depleting several E2s, most notably UBE2O and -R2, strongly impaired mutEJ (Fig. 2c, and Supplementary Table 2). Module 3 measured DDR signalling triggered by IR that generates DSBs throughout the cell- cycle), or camptothecin (CPT), a topoisomerase I poison that yields S-phase DSBs, which are repaired by HR and activate ATR. DDR readouts employed were CHK1 (CHEK1) phosphorylated on Ser-345 (pCHK1, a marker for ATR activation), RPA2 phosphorylated on Ser-4 and Ser-8 (pRPA2, a resection marker) and KAP1 (TRIM28) phosphorylated on S824 (pKAP1, mainly ATM-dependent). γH2AX and Cyclin A (CCAN2) were used as markers for DNA-damage induction and cells in S/G2, respectively (Fig. 3a, Supplementary Fig. 2a-d). As expected4, siCtIP reduced pRPA2 and pCHK1 but not γH2AX or pKAP1 following camptothecin-treatment (Fig. 3b and Supplementary Fig. 2a4). BRCA1 depletion decreased pCHK1 induction and, consistent with recent findings indicating that BRCA1 functions mainly downstream of resection initiation18–22, strongly reduced HR (7.91% ±4.45% SD of siCTRL, n=4) but not camptothecin-induced pRPA2 (Fig. 3b and Supplementary Fig. 2a4).   6   Strikingly, many siE2s affected camptothecin- and IR-induced DDR signalling, with pCHK1 effects following camptothecin generally correlating with reduced pRPA2 (Fig. 3b-c). Of the 37 E2 depletions, many produced >40% reductions in pRPA2 following camptothecin (20 E2s), pCHK1 following camptothecin (21 E2s) or pCHK1 after IR (17 E2s; Fig. 3d; see examples in Fig. 3e). Decreased camptothecin-induced pCHK1 sometimes correlated with impaired pKAP1 and γH2AX induction, possibly reflecting lower S-phase indices (Supplementary Fig. 2a and 2c). Notably, however, 10 siE2s affected all three DDR readouts (Fig. 3d) but not IR-induced pKAP1 and γH2AX (Fig. 3c, Supplementary Fig. 2a1-4, and 2d), indicating that the targeted E2s likely possess DDR functions. These included UBE2N, -O, - V2, -D, -R, -L and -J family members, plus UBE2T and -W linked to the Fanconi anemia (FA) DNA-repair pathway23,24. These results thus highlighted potential roles of many E2s in ATR signalling, consistent with ubiquitylation promoting ATR activation25. By contrast, the main mark affected by siUBE2S or siBIRC6 was pKAP1 (both after camptothecin or IR- treatment; Fig. 3c, 3e, Supplementary Fig. 2a3-4 and 2d), suggesting these E2s may promote ATM signalling. Validating DDR functions for selected E2s To help interpret screen readouts, we represented them in a Circos plot (Fig. 4a). Here, siE2s are plotted clockwise in order of decreasing DDR impact (segment colours and breadths reflect cumulative impacts based on module readouts) and siE2s scoring in the top 25% of all readouts are connected by ribbons to the respective colour-coded readout (further details in Fig. 4a). Thus, it is clear that many siE2 treatments affected HR, with various siE2s also impairing FK2 IRIF. By contrast, mutEJ effects were less pronounced, possibly reflecting the lower complexity of end-joining than HR.   7   To validate the value of our screening data for future studies, we selected E2 families where at least one family member had ≥3 connecting Circos plot ribbons (Fig. 4a). Excluding UBE2N and -V1, due to their defined DDR functions8, this group comprised UBE2D, -R, -J and -L members, and UBE2O. Importantly, these 11 siE2s efficiently depleted their targets and, apart from UBE2Ds, no marked cross-depletions were detected (Supplementary Fig. 3 and 4a). By contrast, several siRNAs designed to target a particular UBE2D significantly co- depleted other UBE2Ds (Supplementary Fig. 4a), reflecting their high sequence homology at the mRNA (77-86% coding-sequence identity) and protein levels (Supplementary Fig. 4b). Indeed, we noted that siRNAs targeting UBE2D3 efficiently depleted all UBE2Ds. Similarly, an siRNA designed to target UBE2D2 depleted both UBE2D2 and -D3 and had some effect on UBE2D1 (Supplementary Fig. 4a). While some siUBE2Ds may have unique DDR functions, based on their high degree of sequence homology, it seems likely that they have largely overlapping functions. To avoid potential redundancy issues of UBE2D proteins and strengthen screening data in follow-up experiments, we established an siRNA mixture (siALL-Ds) that efficiently depleted all UBE2Ds (Supplementary Fig. 4c-d). Importantly, when we tested the 11 siRNAs with strongest DDR impacts for potential effects on RAD51, a key HR factor and a common “off-target” for many siRNAs26, none caused a pronounced reduction in RAD51 (Supplementary Fig. 4e). We next explored DDR functions for selected E2s by live-imaging of cells expressing GFP- tagged E2s. This revealed that various UBE2R, -L and -D family members displayed both nuclear and cytoplasmic localizations, and were rapidly recruited to DSBs induced by laser micro-irradiation (Fig. 4b; recruitments were weak and detected best between 5 and 30 minutes). By contrast, UBE2O, -J1 and -J2 were largely cytoplasmic and not detected at laser tracts. Thus, >70% (8 of 11) of the selected E2s and 100% of the nuclear ones (UBE2R1, -R2,   8   -L3, -L6, -D1, -D2, -D3 and -D4) accumulated at DSB sites (previously, only UBE2N, -A and -B were shown to do this27,28; note that, although overexpressed UBE2J1, -J2 and UBE2O were preferentially cytoplasmic and were not chosen for further analysis, they may impact on the DDR indirectly, or function directly in the DDR requiring associated factors for nuclear localization). We then examined the eight selected DNA-damage-recruited E2s functionally by assessing their impact on proliferation/cell growth after IR-treatment. This revealed that depleting UBE2R1, -R2, -L3 or -L6 with at least two independent siRNAs caused IR hyper-sensitisation, in some cases comparable to siATM (Fig. 4c). This was also so for co-depleting UBE2Ds. Collectively, these results supported there being important DDR functions for UBE2R1, -R2, -L3, -L6, and -Ds. DNA-end resection links UBE2Ds to RNF138 Due to their strong phenotypes in multiple screen readouts, we focused subsequent studies on the UBE2D family, co-depleting them with siALL-Ds. In line with our pRPA2 data, depleting UBE2Ds strongly decreased camptothecin-induced formation of ssDNA and nuclear RPA2 foci (to avoid potential cell cycle effects, these analyses focused on γH2AX- positive S phase cells; Fig. 5a). Similar RPA2 foci and ssDNA formation defects were also observed upon MG132 treatment (Supplementary Fig. 5a), consistent with previous reports following intermediate29 but not extreme17 IR doses. From these and preceding data on HR and DDR-signalling, we concluded that UBE2Ds function upstream of ssDNA formation to promote ATR signalling and DSB repair by HR. To explore the mechanism for these effects, we endeavoured to identify the UBE2D partner E3(s). Thus, we selected UBE2D-interacting E3s retrieved from several databases (UniProtKB, MINT, STRING and I2D). To reduce the E3 hits (96, 109, 106 and 99 for   9   UBE2D1, -D2, -D3 and -D4, respectively; 121 unique hits in total), we prioritised E3s previously identified as potential ATM/ATR targets30. TRIP12, STUB1 (CHIP), RAD18 and BRCA1 were not selected due to their established DDR phenotypes not being consistent with UBE2Ds-depletion1,8,31–34. In particular, BRCA1 depletion did not markedly affect camptothecin-induced pRPA or ssDNA formation in our assays (Fig. 3b and data not shown). Strikingly, when we investigated the resulting six E3s in a semi-automated screen, only RNF138 (RING finger protein 138) depletion substantially impaired ssDNA formation (Fig. 5b). Furthermore, like UBE2Ds-depletion, RNF138 depletion with two independent siRNAs (Supplementary Fig. 5b) markedly reduced RPA2 and ssDNA camptothecin-induced foci (Fig. 5c). Moreover, depleting UBE2Ds and RNF138 simultaneously did not lead to additive effects in BrdU-based ssDNA detection assays (Supplementary Fig. 5c). These results thereby highlighted RNF138 as a DDR factor and potential functional partner of UBE2Ds. RNF138 and UBE2Ds affect similar DDR processes RNF138 is a ubiquitin-E3 ligase, highly conserved in higher eukaryotes, containing a RING- domain, three zinc fingers (ZNFs) and a ubiquitin-interacting motif (UIM; Fig. 6a)35,36. Live imaging of cells transiently expressing wild-type (WT) GFP-RNF138 (endogenous RNF138 depleted) revealed that it rapidly accumulated at laser tracks (within 5 minutes) and persisted for >1 hour (Fig. 6b). Notably, UBE2D depletion or ATM, ATR or DNA-PK inhibition did not appreciably affect RNF138 accrual or retention at DSBs (Supplementary Fig. 6a and data not shown). By contrast, upon MG132 treatment, initial RNF138 recruitment was weaker, while its retention 25 minutes after micro-irradiation was significantly impaired (Fig. 6b), implicating ubiquitylation events in promoting RNF138 retention at DSB sites. Accordingly, deleting its UIM or mutating the RING-domain (RM; C18/C21A) strongly impaired RNF138 retention (Fig. 6b). The most striking phenotype occurred upon deleting the ZNFs, which – in   10   addition to causing a more predominant localisation in the cytoplasm – severely reduced both RNF138 accrual and retention at DSB sites (Fig. 6b). Consistent with ZNFs often binding to nucleic acids, wild-type RNF138 but not RNF138 ΔZNFs bound to streptavidin beads coated with biotinylated DNA (Fig. 6c). Collectively, these data supported a model wherein the ZNFs promote RNF138 recruitment by mediating contacts with exposed DNA at damage sites, and that ubiquitylation events – generated by RNF138 and/or other ubiquitin-E2-E3 pairs – are required for effective RNF138 retention. Together with the database interaction linkages, the similar impacts of UBE2Ds- and RNF138-depletions on resection suggested that they functionally cooperate. Furthermore, RNF138 depletion mimicked UBE2D1-4 depletion, reducing FK2 and BRCA1 IRIF without markedly influencing 53BP1 or γH2AX IRIF (Fig. 6d-e, and Supplementary Fig. 6b-c). Although a previous report connected UBE2D3 with BRCA137, significant differences in phenotypes between depleting UBE2Ds or BRCA1 in resection (Module 3, Fig. 3b, 5a and data not shown) suggested that UBE2Ds also act upstream to, and independently of, BRCA1. RNF138 depletion also phenocopied UBE2Ds depletion in every other DDR readout we tested. Thus, RNF138 depletion severely impaired HR and also affected mutEJ (Fig. 6f and Supplementary Fig. 6d; controls as in Fig. 2b-c). Finally, like UBE2Ds (Fig. 4c), RNF138 was required for cellular resistance to IR in clonogenic survival and cell proliferation/growth- rate assays (Fig. 6g and Supplementary Fig. 6e). Moreover, depleting UBE2Ds and RNF138 simultaneously did not lead to clear additive effects regarding IR-sensitisation or HR efficiency (Supplementary Fig. 6f-g). We therefore concluded that RNF138 acts directly or indirectly with UBE2Ds to promote DSB resection and repair by HR, as well as potentially influencing mutEJ and other DDR processes.   11   UBE2Ds and RNF138 promote CtIP DNA-damage accrual and ubiquitylation Key events leading to DSB resection are accumulation of the MRE11-RAD50-NBS1 (MRN) complex and CtIP at DSBs38. Strikingly, we found that depleting UBE2Ds, or RNF138 with two independent siRNAs, strongly impaired recruitment of endogenous or GFP-CtIP to laser tracks but had no discernible effect on GFP-MRE11 recruitment (Fig. 7a-b rows 1-2 and Supplementary Fig. 7a-b). By contrast, CtIP depletion did not affect RNF138 accrual/retention (Supplementary Fig. 7c). Importantly, the CtIP recruitment defect and the siCtIP-related pRPA2- and pCHK1-induction defects in UBE2Ds-depleted cells (Supplementary Fig. 2b) were partially corrected by stable inducible expression of siALL-Ds resistant wild-type (WT) but not catalytically dead (CD) UBE2D1 (Fig. 7a, rows 3-4, 7c and Supplementary Fig. 7d). Similarly, the CtIP recruitment defect caused by RNF138 depletion was almost entirely corrected in stable cell lines by inducible expression of siRNF138-1 resistant WT-RNF138, but not the RING-domain mutant (RM; Fig. 7b, rows 3-4). The above results and our previous findings supported a model in which the ubiquitin-ligase activities of both UBE2Ds and RNF138 act downstream of MRE11 recruitment to promote CtIP accumulation at DSBs. In line with this and our previous observation that MG132 inhibited DSB resection and HR, MG132 also impaired CtIP accumulation (Supplementary Fig. 7e). As MG132 blocks RNF8-dependent accumulation of RNF168, 53BP1 and BRCA1 at DSBs8, we tested if DNA-damage accumulation of CtIP was impaired by depleting such proteins. In contrast to the impacts of UBE2Ds- or RNF138-depletion, DNA-damage accrual of CtIP occurred effectively in cells depleted of RNF8, RNF168 or BRCA1 (Supplementary Fig. 7f-g).   12   Previous work has established that RNF8, RNF168 and BRCA1 are required for effective FK2 IRIF formation8,37,39. As we had found that UBE2Ds and RNF138 promote CtIP recruitment to DSBs as well as FK2 and BRCA1 IRIF, we tested if CtIP was required for IRIF formation by these and other DDR components. Significantly, this revealed that FK2, BRCA1 and 53BP1 IRIF still formed efficiently in CtIP-depleted cells (Supplementary Fig. 8a-c). Taking these data together, we concluded that cellular DDR functions of UBE2Ds and RNF138 relating to resection and CtIP accrual are specific and largely independent of RNF8, RNF168 or BRCA1. Conversely, RNF8, RNF168 and BRCA1 clearly promote FK2 IRIF formation and HR via BRCA1 recruitment1,2,8 and therefore, operate through mechanisms distinct from the UBE2D-RNF138-CtIP-resection axis. Further strengthening the links between UBE2Ds, RNF138 and CtIP, we found that UBE2D1 co-immunoprecipitated with both RNF138 and CtIP (Fig. 7d). By contrast, we did not detect interactions between RNF138 or CtIP and UBE2K (Supplementary Fig. 8d), another E2 that markedly impaired FK2 IRIF and pRPA2 induction (Fig. 1b, 3b and 4a). Moreover, in reciprocal experiments, RNF138 immunoprecipitation retrieved both UBE2D1 and CtIP (Fig. 7e and Supplementary Fig. 8e). As expected, the GFP-RNF138 RING mutant (RM) did not efficiently co-immunoprecipitate with UBE2D1. However, it efficiently co- immunoprecipitated with CtIP, suggesting RM-RNF138 was not generally misfolded and retained some functionality (Fig. 7e; note: these studies were done in RNF138-depleted cells). Importantly, we found that both ΔUIM and ΔZNFs RNF138 mutants co-immunoprecipitated with UBE2D1 and CtIP to extents comparable to those of wild-type RNF138 (Fig. 7e), suggesting that these two mutants retained some cellular functions.   13   To test whether CtIP is ubiquitylated in a UBE2Ds/RNF138-dependent manner, we co- expressed HA-ubiquitin with either GFP or GFP-CtIP in cells, then treated or mock-treated cells with IR. Next, we prepared cell extracts and assessed GFP immunoprecipitates for HA- ubiquitin staining by immunoblotting. This revealed ubiquitylation bands migrating above GFP-CtIP that were induced upon IR (Fig. 7f). Importantly, while we detected endogenous RNF138 in CtIP immunoprecipitates (Fig. 7g-h), RNF138 ubiquitylation was not increased by IR (Supplementary Fig. 8e), supporting the idea that the IR-induced species in CtIP immunoprecipitates represented ubiquitylated CtIP. Supporting our finding that CtIP depletion did not impair FK2 IRIF, ubiquitylated forms of CtIP were almost undetectable by the FK2 antibody (Fig. 7f). Crucially, we found that depleting UBE2Ds or RNF138 abolished IR-induced CtIP ubiquitylation (Fig. 7g-h), suggesting that UBE2Ds/RNF138 might mediate ubiquitylation as a functional pair (note that the increase in ubiquitylation in siALL-Ds- over siCTRL-treated cells in the absence of DNA damage is due to experimental variation rather than biological significance). Supporting this hypothesis, UBE2D1, but not UBE2T – another E2 enzyme that scored highly in our screen – supported RNF138-auto-ubiquitylation in in vitro assays (Supplementary Fig. 8f). Furthermore, UBE2D1 mediated in vitro ubiquitylation of purified CtIP in an RNF138-dependent manner (Supplementary Fig. 8g). These data thus indicated that UBE2Ds and RNF138 can selectively act as a functional pair and play important roles early on in the DDR to promote IR-induced CtIP ubiquitylation. Finally, to gain deeper mechanistic insights into the importance of UBE2Ds-/RNF138- dependent CtIP ubiquitylation for DNA-end resection and CtIP recruitment, we identified 13 ubiquitylated lysines on CtIP by mass spectrometry, immunoprecipitated from irradiated cells (Fig. 8a). One of these lysines does not play a role in CtIP recruitment40, so was excluded from further analyses. In addition to mutating all 12 potentially relevant lysines to arginines   14   (CtIP 12KR), we generated CtIP mutants with lysine-to-arginine mutations clustered in the N- or C-terminal CtIP regions (CtIP 5KR and 6KR, respectively; Fig. 8a). Immunoprecipitation experiments showed that IR-induced ubiquitylation observed with wild- type CtIP was significantly reduced in the context of these mutants, especially CtIP 12KR and 5KR (Fig. 8b). These results suggested that CtIP-N-terminal ubiquitylation may be important for DNA-end resection and CtIP accrual at DSBs. Furthermore, we found that cells stably expressing siRNA-resistant CtIP 5KR failed to complement camptothecin-induced pRPA2 and pCHK1 in cells depleted of endogenous CtIP (Fig. 8c). Moreover, compared with wild-type CtIP, the 5KR mutant displayed significant laser micro-irradiation recruitment defects, implying that IR-induced CtIP ubiquitylation is important for CtIP accumulation at DSBs (Fig. 8d). Discussion Although ubiquitylation affects virtually every aspects of eukaryotic cell biology, defining the precise nature of such events, and the ubiquitylation components mediating them, has been difficult because of technical challenges and the large number of ubiquitylation factors (>1,000 in human cells). Here, we provide a paradigm for how key aspects of ubiquitin biology can be defined through focused, in-depth screens for E2-enzyme functions, leveraging this and other knowledge to define E2-E3-target-protein cascades. Accordingly, we have established UBE2D-family proteins as important DDR enzymes, and identified RNF138 as a resection/HR promoting factor that functionally interacts with UBE2Ds and CtIP. Our data are consistent with a model in which UBE2Ds and RNF138 constitute a functional E2-E3 ubiquitylation axis that, either directly or in co-operation with additional factors, promotes IR-induced CtIP ubiquitylations, including on five key lysines towards the CtIP N-terminus. As CtIP’s N-terminus is important for CtIP dimerization, tetramerization,   15   recruitment to DSBs and repair by HR15,41, we speculate that CtIP ubiquitylation in this region may promote these functions by affecting its multimeric state and/or its interactions with other factors. Alternatively or in addition, IR-induced ubiquitylation could expose a recently identified internal DNA binding motif in CtIP38 to facilitate its DSB recruitment/functions. UBE2Ds/RNF138 could also generate ubiquitylation moieties recognized by CtIP via its proposed ubiquitin binding function42. Our data thereby suggest further mechanistic studies to determine precisely how ubiquitylation of these N-terminal CtIP lysines promotes resection and HR. We note however, that UBE2Ds and RNF138 – either as a functional pair or in combination with other ubiquitylation components – likely target additional proteins to affect other pathways within and beyond the DDR. Indeed, the accompanying paper by Ismail et al.43 establishes that RNF138 mediates ubiquitylation of the DSB repair protein, Ku, promoting Ku removal from DSBs to facilitate CtIP access, resection and HR. Many additional avenues await exploration through exploiting our E2 screening data. For instance, these studies have highlighted links between various understudied E2 proteins and diverse aspects of the DDR, including DSB repair by HR and mutEJ, IRIF formation and ATM/ATR mediated signalling. For example, we found that depletion of various E2s reduced FK2 IRIF, often without affecting 53BP1 IRIF. Since FK2 antibodies likely recognise many conjugated ubiquitin species, including different chain linkages and/or lengths, we speculate that these ubiquitylation events require multiple E2s, serving as initiators or elongators, and/or ubiquitylating the same protein at multiple sites. Moreover, we speculate that FK2 IRIF reduction upon RNF8/RNF168 depletion may partly reflect the reduction of ubiquitylation events promoting BRCA1 recruitment, thought to further amplify FK2 IRIF37,39. In line with this hypothesis, previous studies have reported an uncoupling between   16   FK2 and 53BP1 IRIF44,45. Collectively, our results suggest that various ubiquitylation events are required for the accumulation of various DDR proteins at DSBs. We also note that FK2 IRIF reduction in ALL-Ds- or RNF138-depleted cells coincided with reduced BRCA1 IRIF. Thus, it will be interesting to explore additional roles of UBE2Ds and RNF138 distinct from the CtIP ubiquitylation-resection axis defined here. It will also be interesting to further study the complex requirements for RNF138 accrual at DSBs, which may have evolved to ensure optimal spatial and temporal control of DSB processing, their downstream signalling and HR. We speculate that follow-on studies emanating from our E2 screen will be greatly facilitated by the use of orthogonal datasets, generated from other functional screens together with proteomic, mutational and gene-expression resources13,30,46–48. Finally, it is noteworthy that such work may have medical applications because small-molecule targeting of DDR- enzymes and ubiquitin system components is providing opportunities for cancer therapies49,50. Indeed, our findings highlight how certain ubiquitin E2 enzymes with DDR functions may represent attractive therapeutic targets.   17   METHODS Methods and any associated references are available in the online version of the paper. Note: Supplementary Information is available in the online version of the paper ACKNOWLEDGEMENTS We are grateful to all S.P.J lab members for support and comments. We thank Tobias Oelschlaegel for polyclonal GFP-Flag-MRE11 U2OS cells, Josep Forment for stable GFP- CtIP wild-type U2OS cells, Carlos le Sage for helping to generate U2OS cells stably expressing GFP-CtIP variants (12KR, 6KR and 5KR), Jon Travers for the pEGFP-C1/TO plasmid and for helping establish inducible GFP-RNF138 U2OS cells, the Shiloh laboratory for the RPA2 mouse hybridoma, the Shiloh and Oren laboratories for HA-ubiquitin plasmid, the Philip Cohen laboratory for UBE2T plasmid, Luca  Pellegrini    and  Neil  Rzechorzek  for  providing   baculovirus-­‐purified   CtIP,  Rafael Carazo-Salas for access to the Opera, Andy Riddell for flow-cytometry cell sorting support, Mareike Herzog and Nigel Smith for advice on Circos plots, the Gurdon Institute bioinformatics core facility, in particular Charles Bradshaw and George Allen, and the Gurdon Institute imaging facility, in particular Alex Sossick, Nicola Lawrence and Richard Butler. Research in the S.P.J. lab is funded by Cancer Research UK Program Grant C6/A11224, the European Research Council (DDREAM), the European Community Seventh Framework Programme grant agreement no. HEALTH-F2- 2010-259893 (DDResponse). Core infrastructure funding was provided by Cancer Research UK Grant C6946/A14492 and Wellcome Trust Grant WT092096. S.P.J. receives a salary from the University of Cambridge, supplemented by Cancer Research UK. C.K.S. was funded by a FEBS Return-to-Europe fellowship. P.B. is supported by the Emmy Noether Programme of the German Research Foundation (DFG, BE 5342/1-1).   18   AUTHOR CONTRIBUTIONS C.K.S./Y.G. and S.P.J. conceived the project. C.K.S./Y.G. performed the experiments with help from M.S. and J.C. C.K.S./Y.G. analysed data. LC-MS/MS was by P.B.. M.D., M.C. and S.J. generated reagents. C.K.S./Y.G. and S.P.J. wrote the manuscript. All authors made suggestions and commented on the manuscript. COMPETING  FINANCIAL  INTERESTS The authors declare no competing financial interests. Reprints and permissions information is available at www.nature.com/reprints. 1. Jackson, S. P. & Bartek, J. The DNA-damage response in human biology and disease. Nature 461, 1071–8 (2009). 2. Ciccia, A. & Elledge, S. J. The DNA damage response: making it safe to play with knives. Mol. Cell 40, 179–204 (2010). 3. Deriano, L. & Roth, D. B. Modernizing the nonhomologous end-joining repertoire: alternative and classical NHEJ share the stage. Annu. Rev. Genet. 47, 433–55 (2013). 4. Sartori, A. A. et al. Human CtIP promotes DNA end resection. Nature 450, 509–14 (2007). 5. Greenberg, R. A. et al. Multifactorial contributions to an acute DNA damage response by BRCA1/BARD1-containing complexes. Genes Dev. 20, 34–46 (2006). 6. Shiloh, Y. & Ziv, Y. The ATM protein kinase: regulating the cellular response to genotoxic stress, and more. Nat. Rev. Mol. Cell Biol. 14, 197–210 (2013). 7. Polo, S. E. & Jackson, S. P. Dynamics of DNA damage response proteins at DNA breaks: a focus on protein modifications. Genes Dev. 25, 409–33 (2011). 8. Jackson, S. P. & Durocher, D. Regulation of DNA damage responses by ubiquitin and SUMO. Mol. Cell 49, 795–807 (2013). 9. Kulathu, Y. & Komander, D. Atypical ubiquitylation - the unexplored world of polyubiquitin beyond Lys48 and Lys63 linkages. Nat. Rev. Mol. Cell Biol. 13, 508–23 (2012).   19   10. Husnjak, K. & Dikic, I. Ubiquitin-binding proteins: decoders of ubiquitin-mediated cellular functions. Annu. Rev. Biochem. 81, 291–322 (2012). 11. Wenzel, D. M., Stoll, K. E. & Klevit, R. E. E2s: structurally economical and functionally replete. Biochem. J. 433, 31–42 (2011). 12. Ye, Y. & Rape, M. Building ubiquitin chains: E2 enzymes at work. Nat. Rev. Mol. Cell Biol. 10, 755–64 (2009). 13. Povlsen, L. K. et al. Systems-wide analysis of ubiquitylation dynamics reveals a key role for PAF15 ubiquitylation in DNA-damage bypass. Nat. Cell Biol. 14, 1089–98 (2012). 14. Certo, M. T. et al. Tracking genome engineering outcome at individual DNA breakpoints. Nat. Methods 8, 671–6 (2011). 15. Davies, O. R. et al. CtIP tetramer assembly is required for DNA-end resection and repair. Nat. Struct. Mol. Biol. 22, 150–157 (2015). 16. Jazayeri, A. et al. ATM- and cell cycle-dependent regulation of ATR in response to DNA double-strand breaks. Nat. Cell Biol. 8, 37–45 (2006). 17. Murakawa, Y. et al. Inhibitors of the proteasome suppress homologous DNA recombination in mammalian cells. Cancer Res. 67, 8536–43 (2007). 18. Zhou, Y., Caron, P., Legube, G. & Paull, T. T. Quantitation of DNA double-strand break resection intermediates in human cells. Nucleic Acids Res. 42, e19 (2014). 19. Polato, F. et al. CtIP-mediated resection is essential for viability and can operate independently of BRCA1. J. Exp. Med. 211, 1027–36 (2014). 20. Kakarougkas, A. et al. Co-operation of BRCA1 and POH1 relieves the barriers posed by 53BP1 and RAP80 to resection. Nucleic Acids Res. 41, 10298–311 (2013). 21. Long, D. T., Joukov, V., Budzowska, M. & Walter, J. C. BRCA1 Promotes Unloading of the CMG Helicase from a Stalled DNA Replication Fork. Mol. Cell 1–12 (2014). doi:10.1016/j.molcel.2014.08.012 22. Cruz-García, A., López-Saavedra, A. & Huertas, P. BRCA1 Accelerates CtIP- Mediated DNA-End Resection. Cell Rep. 451–459 (2014). doi:10.1016/j.celrep.2014.08.076 23. Machida, Y. J. et al. UBE2T is the E2 in the Fanconi anemia pathway and undergoes negative autoregulation. Mol. Cell 23, 589–96 (2006). 24. Zhang, Y. et al. UBE2W interacts with FANCL and regulates the monoubiquitination of Fanconi anemia protein FANCD2. Mol. Cells 31, 113–22 (2011).   20   25. Maréchal, A. et al. PRP19 transforms into a sensor of RPA-ssDNA after DNA damage and drives ATR activation via a ubiquitin-mediated circuitry. Mol. Cell 53, 235–46 (2014). 26. Adamson, B., Smogorzewska, A., Sigoillot, F. D., King, R. W. & Elledge, S. J. A genome-wide homologous recombination screen identifies the RNA-binding protein RBMX as a component of the DNA-damage response. Nat. Cell Biol. 14, 318–28 (2012). 27. Ikura, T. et al. DNA damage-dependent acetylation and ubiquitination of H2AX enhances chromatin dynamics. Mol. Cell. Biol. 27, 7028–40 (2007). 28. Liu, C. et al. RNF168 forms a functional complex with RAD6 during the DNA damage response. J. Cell Sci. 126, 2042–51 (2013). 29. Jacquemont, C. & Taniguchi, T. Proteasome function is required for DNA damage response and fanconi anemia pathway activation. Cancer Res. 67, 7395–405 (2007). 30. Matsuoka, S. et al. ATM and ATR substrate analysis reveals extensive protein networks responsive to DNA damage. Science 316, 1160–6 (2007). 31. Gudjonsson, T. et al. TRIP12 and UBR5 suppress spreading of chromatin ubiquitylation at damaged chromosomes. Cell 150, 697–709 (2012). 32. Parsons, J. L. et al. CHIP-mediated degradation and DNA damage-dependent stabilization regulate base excision repair proteins. Mol. Cell 29, 477–87 (2008). 33. Inagaki, A. et al. Human RAD18 interacts with ubiquitylated chromatin components and facilitates RAD9 recruitment to DNA double strand breaks. PLoS One 6, e23155 (2011). 34. Watanabe, K. et al. RAD18 promotes DNA double-strand break repair during G1 phase through chromatin retention of 53BP1. Nucleic Acids Res. 37, 2176–93 (2009). 35. Giannini, A. L., Gao, Y. & Bijlmakers, M.-J. T-cell regulator RNF125/TRAC-1 belongs to a novel family of ubiquitin ligases with zinc fingers and a ubiquitin-binding domain. Biochem. J. 410, 101–11 (2008). 36. Yamada, M. et al. NARF, an nemo-like kinase (NLK)-associated ring finger protein regulates the ubiquitylation and degradation of T cell factor/lymphoid enhancer factor (TCF/LEF). J. Biol. Chem. 281, 20749–60 (2006). 37. Polanowska, J., Martin, J. S., Garcia-Muse, T., Petalcorin, M. I. R. & Boulton, S. J. A conserved pathway to activate BRCA1-dependent ubiquitylation at DNA damage sites. EMBO J. 25, 2178–88 (2006). 38. You, Z. et al. CtIP links DNA double-strand break sensing to resection. Mol. Cell 36, 954–69 (2009).   21   39. Morris, J. R. & Solomon, E. BRCA1  : BARD1 induces the formation of conjugated ubiquitin structures, dependent on K6 of ubiquitin, in cells during DNA replication and repair. Hum. Mol. Genet. 13, 807–17 (2004). 40. Kaidi, A., Weinert, B. T., Choudhary, C. & Jackson, S. P. Human SIRT6 promotes DNA end resection through CtIP deacetylation. Science 329, 1348–53 (2010). 41. Andres, S. N. et al. Tetrameric Ctp1 coordinates DNA binding and DNA bridging in DNA double-strand-break repair. Nat. Struct. Mol. Biol. 22, 158–166 (2015). 42. Murina, O. et al. FANCD2 and CtIP cooperate to repair DNA interstrand crosslinks. Cell Rep. 7, 1030–8 (2014). 43. Ismail, I. H. et al. RNF138 is an E3 ubiquitin ligase that displaces Ku to promote DNA end resection and regulate DNA repair pathway choice. Nat. Cell Biol. 44. Galanty, Y., Belotserkovskaya, R., Coates, J. & Jackson, S. P. RNF4, a SUMO- targeted ubiquitin E3 ligase, promotes DNA double-strand break repair. Genes Dev. 26, 1179–95 (2012). 45. Smeenk, G. et al. Poly(ADP-ribosyl)ation links the chromatin remodeler SMARCA5/SNF2H to RNF168-dependent DNA damage signaling. J. Cell Sci. 126, 889–903 (2013). 46. Markson, G. et al. Analysis of the human E2 ubiquitin conjugating enzyme protein interaction network. Genome Res. 19, 1905–11 (2009). 47. Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1 (2013). 48. Paulsen, R. D. et al. A genome-wide siRNA screen reveals diverse cellular processes and pathways that mediate genome stability. Mol. Cell 35, 228–39 (2009). 49. Curtin, N. J. DNA repair dysregulation from cancer driver to therapeutic target. Nat. Rev. Cancer 12, 801–17 (2012). 50. Mattern, M. R., Wu, J. & Nicholson, B. Ubiquitin-based anticancer therapy: carpet bombing with proteasome inhibitors vs surgical strikes with E1, E2, E3, or DUB inhibitors. Biochim. Biophys. Acta 1823, 2014–21 (2012). 51. Krzywinski, M. et al. Circos: an information aesthetic for comparative genomics. Genome Res. 19, 1639–45 (2009). 52. You, Z. & Bailis, J. M. DNA damage and decisions: CtIP coordinates DNA repair and cell cycle checkpoints. Trends Cell Biol. 20, 402–9 (2010).   22   Figure legends Figure 1. Screening E2s for IRIF-kinetics. a, Experimental pipeline for module 1 encompassing automated 96-plate format high-throughput/high-content quantitative imaging (spinning disk Opera platform) and automated IRIF detection (Acapella spot detection) of γH2AX (a DNA-damage marker), 53BP1 and conjugated ubiquitin (FK2 antibody) in a total of >1 million cells.  b, IRIF induction per cell and average nuclear area, 30 minutes post- irradiation. Negative control siRNA targeting luciferase (siCTRL) and positive control siRNA mix targeting RNF8 plus RNF168 are marked in red and blue, respectively.   Plots represent medians of n=80 (siCTRL and siRNF8+168) or n=20 (all other siRNAs) 96-plate- well image fields (lines in boxes), 25-75 percentiles (boxes) and 10-90 percentiles (whiskers) of, on average, 704 imaged cells per siRNA based on one screening experiment. c, Full IRIF- kinetics of selected siE2s. Data represent averages ± range of n=8 (siCTRL and siRNF8+168) or n=2 (all other siRNAs) 96-microplate-wells of, on average, 704 imaged cells per siRNA based on one screening experiment. Note that siD3-2 co-depletes all UBE2Ds and siD2-1 co- depletes UBE2D2, UBE2D3 and, to a lesser extent, UBE2D1 (Supplementary Fig. 4a). See Supplementary Figure 1c and PubChem BioAssays for a comprehensive overview of screening data. siG2=siUBE2G2. Figure 2. Screening E2s for DSB repair. a, Schematic of pipeline for module 2 TLR assay14,15 b, TLR results for HR. Note that, as HR can only occur in S/G2, data were corrected to flow-cytometry S/G2 values. Grey boxes highlight control siRNA oligonucleotides, dotted line represents arbitrary cut-off at 60%. Data represent means ± SDs for, on average, n=5 biological experiments for controls and a minimum of n=3 for siE2s. See   Supplementary   Table   2   for   precise   n   values   for   each   condition.   c, TLR results for mutEJ. Grey boxes highlight control siRNAs, dotted line represents arbitrary cut-off at 60%.     23   Data represent means ± SDs for, on average, n=5 biological experiments for controls and a minimum of n=3 for siE2s. See   Supplementary   Table   2   for   precise   n   values   for   each  condition.  Note: depleting the NHEJ factor DNA-PKcs or MMEJ factor ligase III3 did not affected mutEJ significantly, suggesting that both pathways act redundantly in this assay. Note that siD3-2 co-depletes all UBE2Ds and siD2-1 co-depletes UBE2D2, UBE2D3 and, to a lesser extent, UBE2D1 (Supplementary Fig. 4a). See PubChem BioAssays for a comprehensive overview of screening data. Figure 3. Screening E2s for DDR signalling. a, Schematic of module 3 encompassing immunoblotting of DDR signalling events after camptothecin or IR-treatment. b, c, Quantification of key module 3 readouts following camptothecin (CPT) or IR, respectively, based on immunoblots shown in Supplementary Fig. 2a. Results represent ratios of phosphorylated over total protein levels e.g. phospho-CHK1/total CHK1, which were normalised to the corresponding ratio of siCTRL-treated cells and from which the equivalent ratios in undamaged cells were subtracted. In b, negative control siRNA (siCTRL) and positive controls are highlighted in red and blue, respectively. Grey background shows regions of >40% reductions. In c, a grey box highlights controls and the dotted red line indicates arbitrary cut-off at 60%.   d, Overlaps of siE2s scoring with >40% reductions in indicated module 3 readouts. e, Whole cell extract (WCE) immunoblot examples for selected siE2 pools based on one screening experiment. Red frames highlight key readouts that the indicated siE2 pools score in, based on their quantification by densitometry as indicated above in b and c. Dotted red frame indicates that only one of the two siE2 pools shows a clear phenotype in the indicated readout. Note: siD3-2 co-depletes all UBE2Ds and siD2-1 co- depletes UBE2D2, UBE2D3 and, to a lesser extent, UBE2D1 (Supplementary Fig. 4a). See PubChem BioAssays for a comprehensive overview of quantitative screening data.   24   Figure 4. DDR validation of selected E2s. a, Circos plot51; see legend on top right of the panel: (I) Legend for siE2-segments (arranged clockwise underneath the grey arrow on the right half of the Circos plot according to decreasing overall DDR effects). Outer bar: colour indicates readouts that this siE2 scores in; elements are ordered clockwise for decreasing contributions of individual readouts giving a cumulative impact for each siE2; width of element indicates strength of phenotype. siE2: colour-coded; segment width reflects cumulative impact of all readouts that the respective siE2 scores in; ordered clockwise for siE2s with decreasing cumulative impact. Ribbons: colours according to read-out, ordered according to readout-segments on the left half of the circle; only for siE2s scoring in top 25 percentile (siE2 values pooled from all modules; siCTRL set to 100%); ribbon width constant for all siE2s. (II) Legend for key DDR readout-segments located underneath orange rim on the left (arranged on left half of Circos plot). Outer bar: colour indicates siE2s scoring in this module readout; ordered clockwise corresponding to decreasing strength of individual siE2 effect; width indicates strength of phenotype. Ribbons: ordered according to siE2 segment list on other half of circle (see also above for siE2s). Read-out: colour-coded, segment width reflects cumulative impact of all siE2s scoring in this module. b, Live-cell laser micro- irradiation analyses of transiently expressed GFP-E2s. Data represent fluorescence intensity means of n=9 (UBE2D1), n=6 (UBE2D2), n=3 (UBE2D3), n=3 (UBE2D4), n=6 (UBE2R1), n=9 (UBE2R2), n=12 (UBE2L3) and n=18 (UBE2L6) measurements ± SEM based on three measured sites across each laser tract. c, Incucyte cell proliferation/growth rate assays. Data represent medians ± SEM of n=10 experiments for controls and n=8 for siE2s. Statistics source data for this panel can be found in Supplementary Table 5. Scale bar=10 µm. AUs: arbitrary units; see PubChem BioAssays for a comprehensive overview of screening data.   25   Figure 5. DNA-end resection links UBE2Ds with RNF138. a, Defects in camptothecin (CPT)-induced RPA2 foci and ssDNA-generation (BrdU) in siALL-Ds treated cells. Dotted outlines represent nuclei according to DAPI staining. Quantifications were performed exclusively in γH2AX-positive nuclei (see arrowheads), representing actively replicating S phase cells that encountered CPT-trapped TopI. Negative control siRNA (siCTRL) is shown in grey. Plots represent RPA2 or BrdU fluorescence intensity medians (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=161 (siCTRL) and 222 (siALL-Ds) γH2AX-positive cells for RPA2, and n=249 (siCTRL) and 173 (siALL-Ds) γH2AX-positive cells for BrdU, accumulated over two independent experiments. P-values are based on Mann-Whitney analyses. b, Semi-automated Opera/Acapella resection screen for selected UBE2D-interacting E3s. Negative control siRNA (siCTRL) and positive controls are highlighted in red and blue, respectively. Plot represents image field medians (lines in boxes), 25-75 percentiles (boxes) and 10-90 percentiles (whiskers) of, on average, n=203 cells per condition based on one screening experiment (only nuclei with strong γH2AX staining (yellow outlines) were selected. c, Defects in CPT-induced RPA2 foci and ssDNA-generation (BrdU) in siRNF138-treated cells. Dotted outlines represent nuclei according to DAPI staining. Quantifications were performed in nuclei showing strong γH2AX staining (see arrowheads), representing actively replicating S phase cells that encountered CPT-trapped TopI. Negative control siRNA (siCTRL) is shown in grey. Plots represent RPA2 or BrdU fluorescence intensity medians (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=73 (siCTRL), n=74 (siRNF138-1), n=72 (siRNF138-2) γH2AX positive cells for RPA2, and n=72 (siCTRL, siRNF138-1 and siRNF138-2) γH2AX positive cells for BrdU, accumulated over two independent experiments. P-values are based on Mann-Whitney analyses. All scale bars=10 µm.   26   Figure 6. Phenotypic mimicry between UBE2Ds and RNF138. a, Schematic of RNF138 domain architecture. b, Laser micro-irradiation experiments to determine accrual/retention requirements of siRNA-resistant GFP-RNF138 transiently expressed in U2OS cells depleted for endogenous RNF138. Data represent fluorescence intensity means of, on average, n=24 measurements ± SEM based on three measured sites across each laser tract accumulated over two independent experiments. c, Streptavidin pull-down experiments of biotinylated DNA in the presence of RNF138-WT or –ΔZNFs.   Blots  with  dotted   lines   and   framed  by   a   solid  line   are   from   the   same   gel   and   the   same   exposure.   d, Depleting UBE2Ds or RNF138 impairs FK2 and BRCA1 but not 53BP1 or γH2AX IRIF. e, Quantification of FK2, 53BP1 and BRCA1 IRIF and mean γH2AX nuclear intensities (Opera/Acapella platform). Data represent means ± SDs of n=5 experiments for siCTRL, siRNF8+168 and siALL-Ds and n=4 experiments for siRNF138-1 (FK2 and 53BP1 foci and γH2AX nuclear intensity) and n=4 for all siRNAs (BRCA1 foci), merged from, on average, 13,971 cells per condition. Statistics source data for this panel can be found in Supplementary Table 5. f, TLR results for HR efficiency in siRNF138-treated U2OS cells. Note that, as HR only occurs in S/G2, data were corrected to flow-cytometry S/G2 values. Dotted line represents arbitrary cut-off at 60%. Data represent means ± SDs for n=5 independent experiments for siCTRL and n=3 for siRNF138. Statistics source data for this panel can be found in Supplementary Table 5. g, Clonogenic survival assays. Data represent means ± SDs of 4 independent experiments. Statistics source data for this panel can be found in Supplementary Table 5. AUs: arbitrary units. All scale bars=10 µm. Figure 7. UBE2Ds- and RNF138-dependent CtIP-accrual and its IR-induced ubiquitylation. a, siALL-Ds resistant WT- but not CD-GFP-UBE2D1 partially rescues CtIP   27   accrual defects in siALL-Ds treated Cyclin A (CycA)-positive cells. Plots represent median CtIP fluorescence intensity ratios (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=151 (1), 74 (2), 154 (3) and 75 (4) Cyclin A positive cells, accumulated over two independent experiments. P-values are based on Mann- Whitney analyses. ****p<0.0001. Statistics source data for this panel can be found in Supplementary Table 5. b, siRNF138-1 resistant WT- but not RM-GFP-RNF138 rescues CtIP accrual defects in siRNF138-1-treated Cyclin A (CycA)-positive cells. Plots represent median CtIP fluorescence intensity ratios (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=74 (1), 80 (2), 94 (3) and 76 (4) Cyclin A positive cells, accumulated over two independent experiments. P-values are based on Mann-Whitney analyses. ****p<0.0001; ns: not significant (p≥0.05). Statistics source data for this panel can be found in Supplementary Table 5. c, siALL-Ds resistant WT- but not CD- GFP-UBE2D1 partially rescues pCHK1 and pRPA2 induction in siALL-Ds-treated U2OS cells following IR- or camptothecin (CPT) treatment. d, GFP-UBE2D1 co-imunoprecipitates with CtIP and RNF138 in cells irradiated or not (30 min, 15 Gy). e, GFP-RNF138-RING mutant (RM) does not efficiently bind UBE2D1 in cells depleted of endogenous RNF138. Comparable levels of CtIP co-imunoprecipitate with WT GFP-RNF138 and all variants. Note that this experiment was conducted under stringent conditions (500 mM NaCl). f, Ubiquitylation kinetics of CtIP following IR (15 Gy) at indicated times. Note that HA- ubiquitin is detected in GFP-CtIP imunoprecipitates by HA- but not FK2-antibodies. g, IR- induced CtIP ubiquitylation in cells depends on UBE2Ds (30 min, 15 Gy). Note that the increase in siALL-Ds-treated cells in the absence of DNA damage is due to experimental variation rather than biological significance. h, IR-induced CtIP ubiquitylation in cells depends on RNF138 (30 min, 15 Gy). Molecular weight markers are in kilodaltons. Reco.:   28   recombinant; Endo.: endogenous. Experiments shown in d-h were conducted in HEK293 cells. All scale bars=10 µm. Figure 8. N-terminal, IR-induced CtIP ubiquitylation promotes its recruitment and function. a, Schematic of CtIP showing regions/sites of interest15,41,52. The 13 ubiquitylated lysines (K) identified by LC-MS/MS of CtIP peptides retrieved from irradiated HEK293 cells are highlighted. K604, known to not be important for CtIP recruitment40, is shown in black; other ubiquitylated lysines are highlighted in red. Red lysines were mutated to arginines (R), as indicated, leading to CtIP mutants 12KR, 5KR and 6KR. b, Defects in IR-induced CtIP ubiquitylation in HEK293 cells, especially for the CtIP-12KR and -5KR lysine to arginine mutants. c, siRNA-resistant GFP-CtIP-WT, but not -5KR, rescue pRPA2 and pCHK1 induction in siCtIP-treated U2OS cells, one hour after CPT-treatment. d, Recruitment of siRNA-resistant GFP-CtIP-5KR to sites of DNA damage induced by laser micro-irradiation in siCtIP-treated Cyclin A (CycA)-positive U2OS cells is reduced compared to wild-type GFP-CtIP. Plots represent median GFP-CtIP fluorescence intensity ratios (lines in boxes; numbers above boxes), 25-75 percentiles (boxes) and 10-90 percentiles (whiskers) of n=135 (WT) and 112 (5KR) Cyclin A positive cells, accumulated over two independent experiments. Dots represent remaining values outside 10-90 percentile. P-values are based on Mann- Whitney analyses. Scale bar=10 µm. Statistics source data for this panel are available in Supplementary Table 5. Molecular weight markers are in kilodaltons. Figure 1 15 si Z 20 si T 36 si F 1 siC TR L 2 siR NF 8/ 16 8 3 siD 3 (D 1- D4 ) 4 siN 5 siO 6 siD 2 (D 2/ D3 ) 7 siR 2 8 siJ 1 9 siK 10 s iL 6 11 s iQ 2 12 si R1 13 s iV 1 14 s iG 1 16 si A 17 s iG 2 18 si W 19 s iQ 1 21 si NL 22 si U 23 s iU EV 3 24 s iT SG 10 1 25 si C 26 s iV 2 27 si BI RC 6 28 si M 29 si S 30 s iA KT IP 31 si N* 32 si D1 33 si H 34 s iL 3 35 s iE 2 37 s iE 3 38 si B 39 s iE 1 40 s iJ2 41 si D4 0 2 3 4 Nu cle ar ar ea (a rb itra ry un its ) 0 5 10 15 FK 2 0 10 20 30 53 BP 1 30 min, 2 Gy 0 10 20 γH 2A X IR (2 Gy) non-treated 2 h 0.5 h 8 h 24 h + E2 siRNAs Human U2OS cells Automated HT/HC microscopy & Acapella analysis Ubiquitin γH2AX 53BP1 IRIF- kinetics Module 1 a (FK2) b 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 3130 32 33 34 35 36 37 38 39 40 41 c siCTRL siRNF8/168 siD3 (D1-D4) siG2 0 5 10 15 Fo ci/ ce ll ( IR in du cti on ) NT 0.5 2 8 24 Time after IR (h) Ubiquitin (FK2) 53BP1 0 5 10 15 20 25 Fo ci/ ce ll ( IR in du cti on ) NT 0.5 2 8 24 Time after IR (h) 0 5 10 15 20 25 Fo ci/ ce ll ( IR in du cti on ) NT 0.5 2 8 24 Time after IR (h) γH2AX *Inefficient depletion of UBE2N Fo ci/c ell (IR in du ctio n) Error bars in C: range HR (cell-cycle corrected) 0 20 40 60 80 100 120 140 160 HR (uncorrected) b Figure 2 mutagenic EJ siC TR L AT M i DN AP Ki M G1 32 siD NA PK siC tIP siL IG -II I-1 siL IG -II I-2 siT siD 3* siR 2 siV 2 siD 1 siJ 1 siO siL 6 siR 1 siNsiHsiUsiV 1 siL 3 siFsiQ 2 siB IR C6siN L siA siK siT SG 10 1 siE 3 siG 1 siA KT IP siG 2 siD 2* siU EV 3 siB siQ 1 siN ** siW siE 2 siE 1siZsiC siD 4 siJ 2 siMsi S0 20 40 60 80 100 120 140 160 c Cell-cycle correction to S/G2 phases HR & mutagenic EJ efficiencies Four-colour flow-cytometry a Stable human U2OS TLR cells + E2 siRNAs + HR donor plasmid + I-SceI plasmid Module 2 siC TR L AT M i DN AP Ki M G1 32 siD NA PK siC tIP siL IG -II I-1 siL IG -II I-2 siD 3* siO siR 2 siJ 1 siD 1 siV 2 siR 1 siL 6 siL 3 siE 3 siV 1 siN L siB IR C6 siT SG 10 1 siN ** siQ 2 siQ 1 siG 1 siD 4 siA KT IP siG 2 siJ 2 siE 2 siW siE 1 siD 2* siU EV 3 siMsi T siN siH siU si F siK siA siB siZ siC siS eGFP (+1) T2A (+3) mCherry (+3) I-SceI reco- gnition site eGFP (+1) T2A (+3) mCherry (+3) DSBstably integrated TLR + I-SceI plasmid eGFP (+1) T2A (+3) mCherry (+3) Gibberish FP (+3) T2A (+1) mCherry (+1) gene targeting HR mutEJ eGFP fragment eGFP (+1) T2A (+3) mCherry (+3) + HR donor plasmid (eGFP fragment) T2A (+3) mCherry (+3)eGFP (+1) gene disruption 2 bp frameshift + HR donor plasmid *siD3 depletes all UBE2Ds siD2 depletes UBE2D2 and -D3 (and some effect on -D1) **Inefficient depletion of UBE2N Re pa ir ef fic ien cy (% ) Re pa ir ef fic ien cy (% ) b c 0 50 100 150 200 250 siC TR L siC tIP siB RC A1 M G1 32 siD 3* siN siD 2* siL 6 siV 2 siR 1 siN L siT siL 3 siZ siQ 2 siD 1 siH siO siW siA KT IP siJ 1 siD 4 siR 2 siN ** siG 2 siE 3 siF siJ 2 siV 1 siC siG 1 siS siU EV 3 siT SG 10 1 siB IR C6 siQ 1 siM si U siE 2 siE 1 siK siA siBPh os ph o- o ve r t ot al p ro te in lev els a fte r I R (% ) pCHK1/CHK1 (IR)pKAP1/KAP1 (IR) Module IIIpCHK1 pRPA2 γH2AX pKAP1 IR (10 Gy) non-treated 2 h SDS-PAGE/ immunoblots of WCEs signalling a Cyclin A monitoring Module 3 + E2 siRNAs Human U2OS cells + Ionizing radiation CtIP 0 50 10 0 15 0 0 50 100 150 % pRPA2/RPA2 (CPT) % p CH K1 /C HK 1 (C PT ) BRCA1 MG 132 D3* N D2* L6 V2 R1 NL T L3 Z Q2D1H O W AK TIP J1 D4 R2 N** G2 E3 F J2 V1 C G1 S UEV3 TSG 101 BIRC6Q1 M U E2 E1 K A B CTRL + Camptothecin CPT (1 μM) 1 h overlap Figure 3 d pC HK 1 ( CP T) pCHK1 (IR) pRPA (CPT) 10 E2s 4 E2s 4 E2s 3 E2s 0 E 2s 6 E2s 3 E2s D4 G1 R2 NL L6 D3* S U K Z Q2 N C V1 A AKTIP D1 V2 Q1 N** H D2* F O E3 R1 L3 J1 W T Circle segments e siC TR L siU BE 2R 1 CPT (1 μM) 1 h siU BE 2R 2 Cyclin A γH2AX pS4/S8 RPA2 RPA2 pS345 CHK1 CHK1 pS824 KAP1 KAP1 Cyclin A γH2AX pS345 CHK1 CHK1 siC TR L siU BE 2L 3 siU BE 2L 6 IR (10 Gy) 2 h siC TR L siU BE 2S pS824 KAP1 KAP1 pS345 CHK1 CHK1 Cyclin A γH2AX IR (10 Gy) 2 h 55 55 55 130 130 17 siC TR L siU BE 2K CPT (1 μM) 1 h pS824 KAP1 KAP1 pS4/S8 RPA2 RPA2 Cyclin A γH2AX 55 17 36 36 28 130 130 pS4/S8 RPA2 RPA2 Cyclin A γH2AX pS345 CHK1 CHK1 siC TR L siC tIP CPT (1 μM) 1 h siB RC A1 55 55 36 36 28 55 17 55 55 36 36 28 55 17 55 55 55 130 130 17 *siD3 depletes all UBE2Ds siD2 depletes UBE2D2 and -D3 (and some effect on -D1) **Inefficient depletion of UBE2N siRNAs E2 HR m utEJ FK 2 (IR ) pC HK 1 ( IR ) pCH K1 (CP T) pRPA (CP T) M od ul e- 1 Module-2 D3* N O R1 V1 D2* R2 J1 T L6 W L3 Q2 G1 D1 U H K NL A G2Q1UEV3N**E 3ZBIR C6FD4 V 2 CJ 2 SAKTIP TSG101 E 2 BM E 1 Mo dul e-3 a c Secondary screen 2: IR hyper-sensitivity Figure 4 GFP-J2 GFP-J1 GFP-O -> No recruitment 0 1 2 3 4 10 100 IR (Gy) Gr ow th ra te (% ) CTRL R1-1 R1-2 R1-3 ATM siRNA: ATM 0 1 2 3 4 10 100 IR (Gy) Gr ow th ra te (% ) CTRL ALL-Ds siRNA: 0 1 2 3 4 10 100 IR (Gy) Gr ow th ra te (% ) CTRL R2-1 R2-2 R2-3 ATM siRNA: ATM 0 1 2 3 4 10 100 IR (Gy) Gr ow th ra te (% ) CTRL L3-1 L3-2 siRNA: ATM 0 1 2 3 4 10 100 IR (Gy) Gr ow th ra te (% ) CTRL L6-1 L6-2 siRNA: b Secondary screen 1: DNA-damage recruitment GFP-R1 GFP-R2 GFP-L3 GFP-L6 GFP-D2GFP-D1 GFP-D4GFP-D3 -> Recruitment Cross- section 0 1 2 Position along cross-section (AUs) Flu or es ce nc e i nte ns ity alo ng la se r li ne (A Us ) GFP-D1GFP-D2 GFP-D3 GFP-D4 GFP-R1 GFP-R2 GFP-L3 GFP-L6 0 1 2 3 4 (I) siE2-segments (II) DDR readout-segments Outerbar siE2 Ribbons D3* m utEJ siE2 indicator Outerbar Ribbons Read-out *siD3 depletes all UBE2Ds siD2 depletes UBE2D2 and -D3 (and some effect on -D1) **Inefficient depletion of UBE2N Figure 5 a γH2AX RPA2 siC TR L siA LL -D s + CPT γH2AX BrdU (ssDNA) siC TR L siA LL -D s + CPT b siE3 pools siC TR L siC tIP siA LL- Ds RN F1 38 AR IH2 HU WE 1 RN F1 81 RF WD 3 UB E3 A 0.0 1.5 2.0 1.0 0.5 Fl uo re sc en ce in te ns ity siCTRL γH 2A X siCtIP siRNF138siALL-Ds siARIH2 siHUWE1 siRNF181 siRFWD3 siUBE3A Br dU (s sD NA ) c Flu or es en ce in te ns ity si CTRL siRNF138 21 0.0 0.5 1.0 1.5 1.00 0.36 p<10-4 0.29 p<10-4 2.0 1.00 0.35 p<10-4 0.23 p<10-4 si CTRL siRNF138 21 + CPT γH2AX RPA2 siC TR L siR NF 13 8- 1 siR NF 13 8- 2 γH2AX BrdU (ssDNA) siC TR L siR NF 13 8- 1 siR NF 13 8- 2 + CPT 0.0 0.5 1.0 1.5 2.0 1.00 p<10-4 Fl uo re sc en ce in te ns ity si CTRL siALL- Ds 1.00 p<10-4 si CTRL siALL- Ds RPA2 BrdU RPA2 BrdU + CPT BrdU 0.37 0.31 0 1 2 3 4 1 2 3 40 5 min 25 min 0 1 2 3 4 1 2 3 40 5 min 25 min 0 1 2 3 4 1 2 3 40 5 min 25 min 0 1 2 3 4 1 2 3 40 5 min 25 min 0 1 2 3 4 1 2 3 40 a RING C2HC C2H2 C2H2 UIM 18 1 245 58 86 105 159 180189 215229 243 zinc fingers Ubiquitin- interaction motif e FK2 siA LL -D s 53BP1 Merge siC TR L Inset 30 min, 2 Gy siR NF 13 8- 1 BRCA1 siA LL -D s CycA Merge siC TR L Inset 30 min, 2 Gy siR NF 13 8- 1 10 0% 10 0% 0 40 80 120 f HR (uncorrected) HR (cell-cycle corrected) siCTRL siRNF138 2 23 % 10 %20 % 12 % 1 100 60 20 Figure 6 d RNF 138 Flu or es ce nc e i nte ns ity ac ros s l as er mi cro irra dia ted lin es (A Us ) 5 min 25 min Cross- section 5 min 25 min GFP-RNF138 W T M G1 32 (1 0 μM ) RM (C 18 /C 21 A) Δ UI M W T b siCTRL siRNF8/168 siALL-Ds siRNF138-1 Position along cross-section (AUs) IR (Gy) g 0 1 2 3 4 Δ 86 -2 15 Δ ZN Fs c Streptavidin pull-downs InputsBi ot in- DN A Un ta gg ed DN A RNF138 WT ΔZNFsRNF138 WT RNF138 ΔZNFs Bi ot in- DN A Un ta gg ed DN A 50 30 30 50 0 50 100 150 Nu cle ar in ten sit y ( % ) 0 5 10 15 20 25 Fo ci/ ce ll 0 3 6 9 12 Fo ci/ ce ll 0 5 10 15 Fo ci/ ce ll γH2AX 53BP1 Ubiquitin (FK2) BRCA1 1 10 100 Su rv iva l ( % ) siCTRL siRNF138-1 siATM 30 min, 2 Gy Re pa ir ef fic ien cy (% ) IR Figure 7a CtIPγH2AX/CycA 1= ve cto r siC TR L 2= ve cto r siA LL -D s 1 2 3= siR D1 W T siA LL -D s 4= siR D1 C D siA LL -D s 3 4 γH2AX/CycA CtIP 1= ve cto r siC TR L 2= ve cto r siR NF 13 8- 1 1 2 3= siR RN F1 38 W T siR NF 13 8- 1 3 4= siR RN F1 38 R M siR NF 13 8- 1 4 b 1 2 3 4 Ct IP flu or es ce nc e int en sit y ra tio (l ine /n uc leu s) 1.95 1.23 1.58 1.19 2 3 1 **** **** **** ns 1 2 3 4 1 2 3 2.11 1.39 1.85 1.40 Ct IP flu or es ce nc e int en sit y ra tio (l ine /n uc leu s) **** **** **** ns f c α CtIP α CtIP α HA (Ubiquitin) α FK2 α pS345 CHK1 IR (min) GFP - - 30 60 GFP-CtIP HA-Ub Input (2%) 30 60 GFP - - 30 60 GFP-CtIP HA-Ub GFP-IP 30 60 55 250 130 250 130 130 h GFP-IP - + - + - + CTRL RNF138 GFP-CtIP HA-Ub GFP siRNA 1.00 : 1.91 1.00 : 1.02 α HA (Ubi- quitin) - + - + - + CTRL RNF138 GFP-CtIP HA-Ub GFP α CtIP α RNF138 α RNF138 Input (2%) 36 36 250 130 130 g GFP + - + GFP-CtIP α RNF138 α RNF138 α CtIP - + CTRL ALL-Ds GFP + - + GFP-CtIP HA-Ub GFP-IP - + CTRL ALL-Ds IR siRNA α HA (Ubi- quitin) 1.00 : 2.19 1.00 : 0.86 HA-Ub Input (2%) 36 36 130 250 130 α CtIP α HA (RNF138) α GFP (UBE2D1) α RNF138 IR GFP - - + + UBE2D1 HA-RNF138 Input (2%) GFP- UBE2D1 GFP Reco. Endo. - - + + UBE2D1 HA-RNF138 GFP-IP 130 36 36 28 36 55 a d α CtIP α GFP (RNF138) α UBE2D1 GFP-IP GFP-RNF138 WT RM ΔUIM ΔZNFGFP siRNF138-2 Input GFP-RNF138 WT RM ΔUIM ΔZNFGFP siRNF138-2 NT --- - 130 36 28 72 55 17 e IR 10 Gy, 2 h CPT 1 μM, 1 h CT RL AL L- Ds AL L- Ds CT RL AL L- Ds AL L- Ds + + + + + +- - - - - - pS345 CHK1 CHK1 pS824 KAP1 γH2AX GFP GFP-D1 WT siRNA Cyclin A KAP1 pS4/S8 RPA2 RPA2 pS345 CHK1 CHK1 pS824 KAP1 γH2AX GFP GFP-D1 CD siRNA Cyclin A KAP1 pS4/S8 RPA2 RPA2 IR 10 Gy, 2 h CPT 1 μM, 1 h CT RL AL L- Ds AL L- Ds CT RL AL L- Ds AL L- Ds + + + + + +- - - - - - 55 55 36 28 130 130 17 55 Figure 8 GFP-CtIP CTRL CtIP Camptothecin (1 h, 1 μM) siRNA GFP only WT 5KR pS345 CHK1 CHK1 γH2AX GFP-CtIP (WT/5KR) pS4/S8 RPA2 RPA2 1 897 K62 K78 K115 K132 K133 K404 K572 K578 K604 K640 K759 K760 K782 CtIP MRN binding MRN bindingPCNA binding 153 157 pRb binding 165 509 557 22 45 650 897 S327 P BRCA1 binding 515 537 P P P P PDimerisation DNA binding SAE2/Ctp1-like CDK site ATM site S664 S745 T847490 494 793 CtBP binding 5KR 6KR 12KRa c Tetramerisation 18 31 d GF P- Ct IP W T GFPγH2AX/CycA GF P- Ct IP 5 KR S/G2 S/G2 S/G2 S/G2 WT 5KR 1 2 3 GFP-CtIP GF P- Ct IP flu or es ce nc e i nte ns ity ra tio (li ne /nu cle us ) p<10-4 1.80 1.46 b α CtIP α CtIP α HA (Ubiquitin) IR GFP - GFP-CtIP HA-Ub GFP-IP + - + - + - ++ WT 12KR 6KR 5KRGFP - GFP-CtIP HA-Ub Input + - + - + - ++ WT 12KR 6KR 5KR 250 130 250 130 36 28 17 55 55 130   1   Supplementary Figure and Table legends Supplementary Figure 1. Opera and Acapella pipeline and IRIF screen results. a, Images illustrating selected steps of an optimised spot detection script pipeline operated by the Acapella software package (Perkin Elmer). b, IRIF results for well (top; all siRNAs) and plate (bottom; control siRNA) replicates. Each data point represents a 96-microplate well average. High correlation coefficients of R2>0.97 for replicates within and between plates highlight accuracy and reproducibility. c, IRIF screen results for module 2: Automated HT/HC confocal microscopy quantification of the kinetics of FK2, 53BP1 and γH2AX IRIF after IR-treatment (2 Gy) in siE2 treated cells. Data are arranged according to phylogenetic families as reported previously70. siCTRL and siRNF8/168 (co-depletion of RNF8 and RNF168)-treated cells were used as a negative and positive control, respectively. Note: siCTRL and siRNF8/168 curves are identical in each graph being part of the same experimental pipeline. Data represent averages ± range of n=8 (siCTRL and siRNF8+168) or n=2 (all other siRNAs) 96-microplate-wells of, on average, 704 imaged cells per siRNA based on one screening experiment. See PubChem Bioassays for a comprehensive overview of the screening data. Supplementary Figure 2. DDR signalling for selected siE2s. a, Whole cell extract immunoblots representing one screening experiment. U2OS cells were treated with the indicated siE2 pools at the indicated time points following IR or treatment with camptothecin (CPT) and probed with the indicated antibodies. Black arrowheads indicate CHK1-specific bands. Positive controls used in this module were siCtIP, siBRCA1 and MG132 treatment (1 hour, 20 µM). b, Whole cell extract immunoblots of U2OS cells using the indicated antibodies at the indicated time points following IR or CPT-treatment. Cells were treated with siCTRL or siALL-Ds, a 1:1:1 siRNA mixture of siD1-2, siD2-1 and siD4-2, which – as   2   we found later (Supplementary Fig. 4c) – efficiently depletes all UBE2Ds with less cytotoxicity than siD3-2. c, Schematic illustrating positions of molecular weight markers, shown in kilodaltons, with respect to where the indicated proteins run on a 4-20% Tris/glycine SDS-PAGE gel. d, Quantification of pKAP1 and γH2AX signalling following CPT-treatment based on immunoblots in a (see also PubChem BioAssays). Data represent ratios of phosphorylated KAP1 over total KAP1 levels, or γH2AX levels, which were normalised to siCTRL-treated cells and from which the equivalent ratios in undamaged cells were subtracted. Negative control siRNA (siCTRL) and positive controls are highlighted in red and blue, respectively. Grey background shows regions of >40% reductions. Supplementary Figure 3. Validation of selected DDR E2s. siRNA depletion efficiencies for deconvoluted siRNA oligonucleotides used for screening modules 1-3 targeting selected candidate E2s. Protein levels of GFP-tagged E2s and mRNA levels of target genes and related E2 family members relative to GAPDH levels are indicated. As expected, siRNAs targeting 3' untranslated regions (UTRs; siD1-1, siD2-2 and siD3-1) did not deplete the corresponding, exogenously expressed GFP-UBE2D constructs, which only contained the respective coding sequences. However, they efficiently depleted the on-target endogenous mRNAs. Histogram data represent averages of two technical replicates. Supplementary Figure 4. Additional validation regarding co-depletion of UBE2Ds. a siRNA depletion efficiencies for deconvoluted siRNA oligonucleotides used for screening modules 1-3 targeting UBE2Ds. GFP-tagged UBE2Ds and mRNA levels of target genes relative to GAPDH levels are indicated. On-target depletion efficiencies and off-target depletion of other members of the same family are analysed. siRNAs targeting 3' untranslated regions (UTRs; siD1-1, siD2-2 and siD3-1) did not deplete the corresponding, exogenously   3   expressed GFP-UBE2D constructs, which only contained the respective coding sequences (Supplementary Fig. 3), but efficiently depleted the on-target endogenous mRNAs. Histogram data represent averages of two technical replicates. b, Amino acid differences between UBE2D1, -D2, -D3, -D4, UBE2R1, -R2, UBE2L3, -L6 and UBE2J1, -J2. c, Efficient simultaneous depletion of all UBE2Ds measured on protein level (left) and by qRT- PCR (right), using a 1:1:1 siRNA oligonucleotide mix containing siD1-1, siD2-1 and siD4-2 (siALL-Ds). Histogram data represent averages of two technical replicates. d, siALL-Ds treatment is specific for depletion of all UBE2Ds but not for other E2s such as UBE2K and - J2, as measured by qRT-PCR. Histogram data represent averages of two technical replicates. e, No marked RAD51 off-target depletion effects in U2OS cells treated with the indicated siRNA oligonucleotides. Supplementary Figure 5. DNA-end resection is linked to ubiquitylation, validation of siRNF138 oligonucleotides and siALL-Ds plus siRNF138 co-depletion effects in BrdU assays a, Defects in camptothecin (CPT)-induced RPA2 foci and ssDNA generation (BrdU staining) in cells treated for 1 hour with 20 µM MG132 followed by a 1 hour CPT-treatment (1 µM). Results are normalised to siCTRL. Dotted outlines represent nuclei according to DAPI staining. Quantifications were performed exclusively in γH2AX-positive nuclei (see arrowheads), representing actively replicating S phase cells that have encountered CPT- trapped TopI. Negative control siRNA (siCTRL) targeting luciferase is shown in grey. Plots represent medians (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=79 (DMSO) and n=80 (MG132) γH2AX-positive cells for RPA, and n=110 (DMSO) and n=106 (MG132) γH2AX positive cells for BrdU, accumulated over two independent experiments. b,  siRNF138-1 and -2 depletion efficiencies determined by qRT-PCR in U2OS cells. Histogram data represent averages of two technical replicates. c,   4   No significant additive  effects in CPT-induced ssDNA-generation (BrdU) in siALL-Ds- plus siRNF138-treated cells compared to cells treated individually with siALL-Ds or siRNF138 oligonucleotides. Quantifications were performed exclusively in γH2AX-positive nuclei (see also Fig. 5), representing actively replicating S phase cells that have encountered CPT- trapped TopI. Plot represents BrdU fluorescence intensity medians (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=132 (siCTRL), n=108 (siALL-Ds), n=112 (siRNF138-1 and siRNF138-2), n=102 (siALL-Ds + siRNF138-1) and n=106 (siALL-Ds + siRNF138-2) γH2AX positive cells, accumulated over two independent experiments. P-values are based on Mann-Whitney analyses; ns: not significant (p≥0.05). All scale bars=10 µm. Supplementary Figure 6. Additional characterisation of RNF138 in the DDR, validation of siBRCA1 and BRCA1 antibody and RNF138-UBE2Ds epistasis assays. a, Transiently expressed siRNA-resistant GFP-RNF138 is recruited to, and retained at, laser micro- irradiation-induced sites of DNA damage independently of UBE2Ds (endogenous RNF138 depleted). b, c, Specificity of BRCA1 antibody and siBRCA1 depletion efficiency, as determined by immunofluorescence images (left; non-treated (NT) for DNA damage) (b) and automated Acapella spot detection of non-targeting siCTRL- and siBRCA1-treated cells (right; 30 min, 2 Gy) (c). Histogram data represent means ± SDs of n=4 independent experiments merged from, on average, 17,129 cells per condition. d, TLR results for mutEJ in siRNF138-treated cells. Controls are as in Figure 2c being part of the same pipeline (only siCTRL is shown). Data represent means ± SDs for n=8 experiments for siCTRL and n=3 for siRNF138. Statistics source data for this panel are available in Supplementary Table 5. e, Incucyte proliferation/growth rate analyses of U2OS cells treated with the indicated siRNA oligonucleotides. siCTRL and siATM were used as a negative and positive control,   5   respectively. Controls are as in Figure 4c, being part of the same pipeline. Data represent medians ± SEM of n=10 experiments for controls and n=8 experiments for siRNF138. Statistics source data for this panel are available in Supplementary Table 5. f, Incucyte proliferation/growth rate analyses of U2OS cells treated with the indicated siRNA oligonucleotides. siCTRL and siATM were used as a negative and positive control, respectively. Controls are as in Figure 4c, being part of the same pipeline. Data represent medians ± SEM of n=10 experiments for controls and n=3 experiments for all others. Statistics source data for this panel are available in Supplementary Table 5. g, TLR-HR results for the indicated siRNA oligonucleotides. Data represent means ± SDs of n=4 independent experiments. Statistics source data for this panel are available in Supplementary Table 5. All scale bars=10 µm. Supplementary Figure 7. CtIP accrual depends on RNF138 and ubiquitylation events but is independent of RNF8, RNF168 and BRCA1; MRE11 accrual is independent of UBE2Ds and RNF138; RNF138 accrual is independent of CtIP; expression levels of WT- and CD-GFP-UBE2D1. a, Stably overexpressed GFP-CtIP efficiently accumulates at laser micro-irradiated DNA damage lines in Cyclin A positive cells (GFP-CtIP+/CycA+ cells) treated with non-targeting siCTRL, but not with two independent siRNF138 oligonucleotides. Data represent medians ± SDs for n=3 experiments, scoring, on average, 183 Cyclin A positive cells for each condition. Statistics source data for this panel can be found in Supplementary Table 5. b, GFP-Flag-MRE11 accrual to sites of laser micro- irradiation-induced DNA damage is intact in cells depleted of all UBE2Ds or RNF138. c, Transiently expressed siRNA-resistant GFP-RNF138 is recruited to, and retained at, laser micro-irradiation-induced sites of DNA damage independently of CtIP (endogenous RNF138 depleted). d, Expression levels of siALL-Ds resistant GFP-UBE2D1-WT and -CD   6   (catalytically-dead), in stable doxycycline-inducible U2OS cell lines compared to endogenous levels of UBE2D1. e, CtIP fails to efficiently accumulate at laser micro- irradiation induced sites of DNA damage in MG132-treated cells. Plots represent median CtIP fluorescence intensity ratios (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=99 (DMSO) and n=97 (MG132) Cyclin A positive cells, merged from two independent experiments. Statistics source data for this panel can be found in Supplementary Table 5. f, g, CtIP accrual to sites of laser micro-irradiation induced DNA damage is intact in cells depleted of RNF8 or RNF168 (f) or BRCA1 (g). Plots represent median CtIP fluorescence intensity ratios (lines in boxes; numbers above whiskers), 25-75 percentiles (boxes) and overall ranges (whiskers) of n=86 (siCTRL, siRNF8) and n=83 (siRNF168) Cyclin A positive cells in f, and n=76 (siCTRL) and n=75 (siBRCA1 and siBRCA2) Cyclin A positive cells in g, merged from two independent experiments. Molecular weight markers are shown in kilodaltons. Endo.: endogenous. All scale bars=10 µm. Supplementary Figure 8. FK2 IRIF are independent of CtIP; CtIP siRNA and antibody validation; GFP-UBE2K pull-downs; Cellular and in vitro ubiquitylation of RNF138; in vitro ubiquitylation of CtIP. a, Immunofluorescence images showing that depletion of CtIP has no marked effect on focal accumulation of ubiquitin (FK2 antibody) and BRCA1 at DNA damage sites 30 minutes after 2 Gy (the latter in Cyclin A positive cells) compared to siCTRL treated cells. b, Automated quantification of 53BP1, FK2 and 53BP1 IRIF and mean γH2AX nuclear intensities in cells transfected with siCTRL or siCtIP oligonucleotides after the indicated IR-treatment based on Opera/Acapella image acquisition/analysis. Data are represented as 96-plate well averages ± SDs of n=4 experiments based on >15,922 cells per condition. Statistics source data for this panel can be found in Supplementary Table 5. c,   7   Specificity of CtIP antibody and siCtIP efficiency, illustrated by CtIP immunofluorescence staining (left) and immunoblotting of whole cell extracts (right) of siCTRL- or siCtIP-treated cells. NT, non-treated for DNA-damage. α-tubulin was used as a loading control. d, CtIP and RNF138 do not co-immunoprecipitate with UBE2K, irradiated or not (15 Gy). e, Ubiquitylation levels of GFP-RNF138 are unchanged following IR-treatment (30 min, 15 Gy). Note that endogenous CtIP and UBE2D1 co-immunoprecipitate with GFP-RNF138 under stringent conditions (500 mM NaCl). f, In vitro ubiquitylation of His-SUMO-RNF138 WT in the presence of UBE2D1, but not UBE2T. All samples contained ATP and biotinylated ubiquitin. g, In vitro ubiquitylation of baculovirus-purified CtIP by His-SUMO- RNF138 WT in the presence of UBE2D1. Other components in the reactions are as in f. Molecular weight markers are shown in kilodaltons. Reco.: recombinant; Endo.: endogenous. Immunoprecipitation assays shown in d-e were conducted in HEK 293 cells.   Blots   with  dotted  lines  and  framed  by  a  solid  line  are  from  the  same  gel  and  the  same  exposure.  All scale bars=10 µm. Supplementary Table 1. Sequences of siRNA oligonucleotides used in this study. Oligonucleotides were purchased from MWG and contain dTdT at the end. siRNA pools containing siRNA-1 and -2 for the indicated E2s and E3s were used in the primary E2 and mini E3 resection screen, respectively. *Inefficient depletion. Supplementary Table 2. HR and mutEJ repair efficiency measured by TLR assay. Two algorithms, Watson Pragmatic (WP) and Dean-Jett-Fox (DJF) were used to calculate cell- cycle phase percentages. SD: standard deviation formed from at least 3 independent biological replicates. *Inefficient depletion. See also PubChem BioAssays for a comprehensive overview and individual values of the screening data.   8   Supplementary Table 3. Sequences of primers used in this study. All primers were purchased from Sigma Aldrich. Supplementary Table 4. Antibodies used in this study. IF: immunofluorescence; WB: Western blotting. Supplementary Table 5. Statistics source data. Individual values are listed for Figures 6e-g, 7a-b, 8d and Supplementary Figures 6d-g, 7a, 7e and 8b.   Supplementary Figure 1 30 20 10 0 W ell re pli ca te 2 FK2 γH2AX 53BP1 Well replicate 1 3020100 R2=0.9722 R2=0.9893 R2=0.9918 30 20 10 0 Pl at e re pli ca te 2 Plate replicate 1 3020100 FK2 γH2AX 53BP1 R2=0.9896 R2=0.9933 R2=0.9972 a bNuclei segmentation Mask transferral Spot detection Foci #/cell Foci #/cell 0 5 10 15 20 CTRL RNF8/168 K siRNA: 0 5 10 15 20 0 5 10 Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) UBE2K (E2 family 1) CTRL RNF8/168 A B siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2A, -B (E2 family 2) 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2G1, -G2 (E2 family 3-1) siRNA: CTRL RNF8/168 G1 G2 siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2R1, -R2 (E2 family 3-2) CTRL RNF8/168 R1 R2 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2D1-D4 (E2 family 4-1) siRNA:CTRL RNF8+168 D1 D2* D3* D4 FK2 FK2 FK2 FK2 FK2 53BP1 53BP1 53BP1 53BP1 53BP1 γH2AX γH2AX γH2AX γH2AX γH2AX Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) c 2525 siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2J1, -J2 (E2 family 5) siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 CTRL RNF8/168 E1 E2 E3 UBE2E1-E3 (E2 family 4-2) siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2U (E2 family 4-3) CTRL RNF8/168 U CTRL RNF8/168 J1 J2 siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2H (E2 family 6**) and AKTIP CTRL RNF8/168 H siRNA: 0 5 10 15 20 NT 30 min 2 h 8 h 24 hNT 30 min 2 h 8 h 24 h 0 5 10 15 20 0 5 10 NT 30 min 2 h 8 h 24 h UBE2F, -M (E2 family 8**) CTRL RNF8/168 F M Time after IR (2 Gy) FK2 FK2 FK2 FK2 FK2 53BP1 53BP1 53BP1 53BP1 53BP1 γH2AX γH2AX γH2AX γH2AX γH2AX Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2N, -NL (E2 family 9-1) CTRL RNF8/168 N*** N NL siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2T (E2 family 9-2) CTRL RNF8/168 T siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2V1-V2 (E2 family 10) CTRL RNF8/168 V1 V2 siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2S (E2 family 11) CTRL RNF8/168 S siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2C (E2 family 12) CTRL RNF8/168 C FK2 FK2 FK2 FK2 FK2 53BP1 53BP1 53BP1 53BP1 53BP1 γH2AX γH2AX γH2AX γH2AX γH2AX Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) CTRL RNF8/168 TSG101 UEV3 siRNA: 0 5 10 15 0 5 10 15 0 5 10 UBE2W (E2 family 13) 2020 CTRL RNF8/168 W siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 BIRC6, UBE2O, -Z (E2 family 14) CTRL RNF8/168 BIRC6 O Z siRNA: 0 5 10 15 20 0 5 10 15 20 0 5 10 UBE2L3, -L6 (E2 family 15) CTRL RNF8/168 L3 L6 siRNA: 0 5 10 15 20 0 5 10 15 20 TSG101 (E2 family 16) and UEV3 0 5 10 siRNA: 0 5 10 15 20 NT 30 min 2 h 8 h 24 hNT 30 min 2 h 8 h 24 h 0 5 10 15 20 0 5 10 NT 30 min 2 h 8 h 24 h UBE2Q1-Q2 (E2 family 17) CTRL RNF8/168 Q1 Q2 Time after IR (2 Gy) FK2 FK2 FK2 FK2 FK2 53BP1 53BP1 53BP1 53BP1 53BP1 γH2AX γH2AX γH2AX γH2AX γH2AX Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) Fo ci/ ce ll (IR in du cti on ) E2 families according to Michelle et al., 2009 siD3 depletes all UBE2Ds siD2 depletes UBE2D2 and -D3 (and some effect on -D1) * E2 family 7: UBE2I (SUMO E2)** Inefficient depletion of UBE2N*** siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2D 1 siU BE 2D 2* siU BE 2D 1 siU BE 2D 2* siU BE 2D 1 siU BE 2D 2* siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2D 3* siU BE 2D 4 siU BE 2D 3* siU BE 2D 4 siU BE 2D 3* siU BE 2D 4 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siUBE2D1 siUBE2D2* siUBE2D3* siUBE2D4 Supplementary Figure 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2G 1 siU BE 2G 2 siU BE 2G 1 siU BE 2G 2 siU BE 2G 1 siU BE 2G 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2Q 1 siU BE 2Q 2 siU BE 2Q 1 siU BE 2Q 2 siU BE 2Q 1 siU BE 2Q 2 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2A siU BE 2B siU BE 2A siU BE 2B siU BE 2A siU BE 2B siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2K siU BE 2C siU BE 2K siU BE 2C siU BE 2K siU BE 2C γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siUBE2G1 siUBE2G2 siUBE2Q1 siUBE2Q2 siUBE2A siUBE2B siUBE2K siUBE2C a-1 *siD3 depletes all UBE2Ds siD2 depletes UBE2D2 and -D3 (and some effect on -D1) Supplementary Figure 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2N L siU BE 2Z siU BE 2N L siU BE 2Z siU BE 2N L siU BE 2Z siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siA KT IP siT SG 10 1 siA KT IP siT SG 10 1 siA KT IP siT SG 10 1 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siUBE2NL siUBE2Z siAKTIP siTSG101 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2R 1 siU BE 2R 2 siU BE 2R 1 siU BE 2R 2 siU BE 2R 1 siU BE 2R 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2L 3 siU BE 2L 6 siU BE 2L 3 siU BE 2L 6 siU BE 2L 3 siU BE 2L 6 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 IR 10 Gy 2 hNon-treated CPT 1 μM 1 h siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2V 1 siU BE 2V 2 siU BE 2V 1 siU BE 2V 2 siU BE 2V 1 siU BE 2V 2 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siC TR L siU BE 2N * siU BE 2N siC TR L siU BE 2N * siU BE 2N siC TR L siU BE 2N * siU BE 2N siUBE2R1 siUBE2R2 siUBE2L3 siUBE2L6 siUBE2N* siUBE2N siUBE2V1 siUBE2V2 a-2 *Inefficient depletion of UBE2N Supplementary Figure 2 siUBE2H siUBE2U siUBE2T siUBE2W siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2H siU BE 2U siU BE 2H siU BE 2U siU BE 2H siU BE 2U siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2T siU BE 2W siU BE 2T siU BE 2W siU BE 2T siU BE 2W γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siUBE2E1 siUBE2E2 siUBE2E3 siUEV3 siUBE2J1 siUBE2J2 siUBE2O siBIRC6 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2E 1 siU BE 2E 2 siU BE 2E 1 siU BE 2E 2 siU BE 2E 1 siU BE 2E 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2E 3 siU EV 3 siU BE 2E 3 siU EV 3 siU BE 2E 3 siU EV 3 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2J 1 siU BE 2J 2 siU BE 2J 1 siU BE 2J 2 siU BE 2J 1 siU BE 2J 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2O siB IR C6 siU BE 2O siB IR C6 siU BE 2O siB IR C6 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 a-3 Supplementary Figure 2 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2F siU BE 2M siU BE 2F siU BE 2M siU BE 2F siU BE 2M siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siU BE 2S M G1 32 siU BE 2S M G1 32 siU BE 2S M G1 32 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siUBE2F siUBE2M siUBE2S MG132 siC TR L IR 10 Gy 2 hNon-treated siC TR L siC TR L CPT 1 μM 1 h siC tIP siB RC A1 siC tIP siB RC A1 siC tIP siB RC A1 γH2AX pS4/S8 RPA2 RPA2 pS824 KAP1 KAP1 Cyclin A pS345 CHK1 CHK1 siCtIP siBRCA1 c 0 50 100 150 0 50 10 0 15 0 % pKAP1/KAP1 (CPT) % γ H2 AX (C PT ) 200 20 0 CTRL CtIPBRCA1 MG 132 N D3* V2 D2* O R1 G1 L6 L3 Q2 D1 J1 W T U H D4 V1 S Q1 CBIRC6 G2 R2 J2 B Z A N** AKTIP E3 NL K F E1E2 UEV3 M TSG 101 + Camptothecin a-4 siC TR L siA LL -D s siC TR L siA LL -D s siC TR L siA LL -D s Non- treated IR 10 Gy 2 h CPT 1 μM 1 h pS4/S8 RPA2 RPA2 KAP1 pS824 KAP1 pS345 CHK1 CHK1 Cyclin A γH2AX b Page Ruler Plus 4-20% Tris/Glycine pS4/S8 RPA2 RPA2 pS345 CHK1/CHK1 γH2AX pKAP1/KAP1 Cyclin A d 250 130 100 72 55 36 28 17 10 *siD3 depletes all UBE2Ds siD2 depletes UBE2D2 and -D3 (and some effect on -D1) **Inefficient depletion of UBE2N CTRL J1-1 J1-2 GFP-J1 65 KAP1 CTRL L3-1 L3-2 GFP-L3 KAP1 50 CTRL L6-1 L6-2 GFP-L6 KAP1 50 GFP-R1 KAP1 CTRL R1-1 R1-2 65 GFP-R2 65 CTRL R2-1 R2-2 KAP1 siC TR L siJ 1-1 siJ 1-2 siJ 2-1 siJ 2-2 siC TR L siL 3-1 siL 3-2 siL 6-1 siL 6-2 siC TR L siR 1-1 siR 1-2 siR 2-1 siR 2-2 0 100 200 J1 J2 E2s m RN A (% ) 0 50 100 L3 L6 E2s m RN A (% ) 0 100 200 R1 R2 E2s m RN A (% ) siRNAs siRNAs siRNAs siRNAs siRNAs siRNAs GFP-O 185 CTRL O-1 O-2 α-tub siRNAs Supplementary Figure 3 GFP-J2 65 CTRL J2-1 J2-2 KAP1 115 115 115 115 115 115 50 0 50 100 siC TR L siD 1- 1 siD 1- 2 UB E2 D1 m RN A (% ) 0 50 100 siC TR L siD 2- 1 siD 2- 2 UB E2 D2 m RN A (% ) 0 50 100 siC TR L siD 3- 1 siD 3- 2 UB E2 D3 m RN A (% ) 0 50 100 siC TR L siD 4- 1 siD 4- 2 UB E2 D4 m RN A (% ) CTRL D1-1* D1-2 KAP1 GFP- UBE2D1 50 siRNAs CTRL D2-1 D2-2* KAP1 GFP- UBE2D2 50 siRNAs KAP1 GFP- UBE2D3 50 CTRL D3-1* D3-2 siRNAs CTRL D4-1 D4-2 KAP1 GFP- UBE2D4 50 siRNAs 115 115 115 115 *siRNAs targeting 3’ UTRs that are not included in GFP-UBE2D constructs. a b 50 CTRL D1-2 D1-3 D2-1 D2-3 D3-2 D3-3 D3-4 D4-1 D4-2 GFP- UBE2D1 siRNA: KAP1 GFP- UBE2D2 50 KAP1 GFP- UBE2D3 KAP1 50 GFP- UBE2D4 KAP1 50 0 50 100 D1 D2 D3 D4 siCTRL siD1-1* 0 50 100 D1 D2 D3 D4 siCTRL siD2-2* 0 50 100 D1 D2 D3 D4 siCTRL siD3-1* UBE2s UBE2s UBE2s m RN A (% ) m RN A (% ) m RN A (% ) Supplementary Figure 4 *siRNAs targeting 3’ UTRs that are not included in GFP-UBE2D constructs. co-depletion of all UBE2Ds co-depletion of UBE2D2 and -D3 (and some effect on -D1) 115 115 115 115 a 0 50 100 D1 D2 D3 D4 siALL-DssiCTRL siALL-DssiCTRL 0 50 100 K J2 UBE2s UBE2s m RN A (% ) m RN A (% ) KAP1 CTRL ALL-Ds CTRL ALL-Ds CTRL ALL-Ds CTRL ALL-DssiRNA: GFP-D1 GFP-D2 GFP-D3 GFP-D4c d D1 D2 D3 D4 D1 89 88 92 D2 97 93 D3 92 D4 protein sequence identity (%) UBE2D1-4 homologies protein sequence identity (%) R1 R2 R1 R2 80 UBE2R1/-R2 homology protein sequence identity (%) L3 L6 L3 L6 55 UBE2L3/-L6 homology protein sequence identity (%) J1 J2 J1 J2 26 UBE2J1/-J2 homology b 50 115 e α-tubulin siR AD 51 siC TR L siA LL -D s siR 1 siR 2 siJ 1 siJ 2 siC TR L siR AD 51 siL 3 siL 6 siO siC TR L siR NF 13 8 (1 +2 ) 50 RAD51 30 25 Supplementary Figure 5 0 20 40 60 80 100 120 RN F1 38 m RN A (% ) si CTRL siRNF138 21 Flu or es ce nc e int en sit y DMSO MG 132 DMSO MG 132 γH2AX RPA2 DM SO M G1 32 (2 0 μM ) γH2AX BrdU (ssDNA) 1.00 0.47 p<10-4 BrdU 0 1 2 1.00 0.53 p<10-4 RPA2+ CPT DM SO M G1 32 (2 0 μM ) + CPT a b c p<10-4 CT RL AL L-D s RN F1 38 -1 RN F1 38 -2 AL L-D s + RN F1 38- 1 AL L-D s + RN F1 38- 2 0.0 0.5 1.0 1.5 2.0 Br dU flu or es ce nc e int en sit y 1.00 0.33 0.43 0.41 0.36 0.32 ns ns siRNA: γH 2A X Br dU (s sD NA ) siCTRL siALL-Ds siRNF138-1 siRNF138-2 siALL-Ds + RNF138-1 siALL-Ds + RNF138-2 γH 2A X Br dU (s sD NA ) siR NF 138 -2 47% siR NF 138 -1 51% Supplementary Figure 6 Gr ow th ra te (% ) 0 10 100 e 0 4 8 12 16 Fo ci/ ce ll 30 min 2 Gy si CTRL si BRCA1 b siC TR L siB RC A1 CycABRCA1 DAPI NT c mutEJ 120 80 40 0 siC TR L 100% d BRCA1 IRIF a 5 min 10 min 20 min 30 min siC TR L siA ll-D s g 100% 11% 22% 17% 21% 12% 14% 0 20 40 60 80 100 120 siC TR L siC tIP siR NF 138 -1 siR NF 138 -2 siA ll-D s siR NF 138 -1 + s iAll -Ds siR NF 138 -2 + s iAll -Ds 0 1 2 3 4 IR (Gy) ATM ALL-Ds + RNF138-2 RNF138-2 ALL-Ds CTRL siRNA: f HR siRNA: CTRL ATM RNF138-1 RNF138-2 RNF138-3 0 1 2 3 4 IR (Gy) Gr ow th ra te (% ) 0 10 100 GFP-RNF138 Re pa ir ef fic ien cy (% ) Re pa ir ef fic ien cy (% ) 30 min Supplementary Figure 7 97% 48% 54% 0 20 40 60 80 100 si CTRL siRNF 138-1 siRNF 138-2 Ct IP +/ Cy cA + ce lls (% ) γH2AX/CycA GFP-CtIP siC TR L siR NF 13 8- 1 siR NF 13 8- 2 a GFP-D1 WT/CD Endo. D1 Loading siC TR L siA LL -D s siA LL -D s - + + +- - - - - siA LL -D s GFP GFP-D1 CD GFP-D1 WT + - - 25 15 30 50 d α UBE2D1 b GF P- Fl ag -M RE 11 siCTRL siRNF138-1 siRNF138-2 siAll-Ds γH2AX/CycA CtIP p<10-4DM SO M G1 32 (2 0 μM ) 2.13 1.32 1 2 3 4 Ct IP flu or es ce nc e int en sit y ra tio (l ine /n uc leu s) DM SO MG 132 e f siC TR L siR NF 8 siR NF 16 8 γH2AX/CycA CtIP g γH2AX/CycA CtIP siC TR L siB RC A1 -1 siB RC A1 -2 4 Ct IP flu or es en ce in te ns ity ra tio (l ine /n uc leu s) CTRL RNF 8 RNF 168 1 2 3 siCTRL 1 2 2 4 Ct IP flu or es en ce in te ns ity ra tio (l ine /n uc leu s) 1 3 5 siRNA siBRCA1 5 min siC TR L siC tIP c GFP-RNF138 10 min siC TR L siC tIP siC TR L siC tIP + - Supplementary Figure 8 d - - + + UBE2K HA-RNF138 Input (2%) - - + + UBE2K HA-RNF138 GFP-IP IR GFP GFP UBE2K GFP Reco. Endo. α CtIP α HA (RNF138) α GFP (UBE2K) α RNF138 130 36 36 28 36 55 250 130 100 72 55 130α CtIP α UBE2D1 17 72 55 α GFP (RNF138) α HA (Ub) - IR IR - -- IR IR GFP GFP-138 GFP GFP-138 HA-Ub GFP-IP HA-Ub Inputea FK2 53BP1 merge DAPIzoom BRCA1 CycA merge DAPIzoom 30 min 2 Gy siCTRL siCtIP 30 min, 2 Gy b siC TR L siC tIP siC TR L siC tIP 30 min 2 Gy c siC TR L siC tIP DAPIγH2AX/CycACtIP NT 185 CtIP α-tub- ulin 50 115 80 65 siC TR L siC tIP Ub- UBE2T α RNF138 - + - - + + + +UBE2T RNF138 E1 + RNF138 1 2 3 80 25 65 50 30 α Biotin (Ub) - + - - + + + ++ 1 2 3 f α RNF138 RNF138 Ub- RNF138 + + - - + - - + + - + + - ++ 1 2 3 54 α Biotin (Ub) 80 25 65 50 30 + + - - + - - + + - + + - ++ 1 2 3 54 UBE2D1 RNF138 E1 Ub- UBE2D1 α CtIP 205 120 + + - - - - + ++ + UBE2D1 RNF138 (ng) E1 CtIP g + + + + + + 1000 + - +- + 50 100 200 + UBE2D1 RNF138 E1 UBE2T RNF138 E1 0 50 100 Nu cle ar in ten sit y ( % ) γH2AX 0 5 10 15 20 25 Fo ci/ ce ll 53BP1 0 3 6 9 12 Fo ci/ ce ll Ubiquitin (FK2) 0 5 10 15 Fo ci/ ce ll BRCA1