MAD2L2 dimerization and TRIP13 control shieldin activity in DNA repair

MAD2L2 (REV7) plays an important role in DNA double-strand break repair. As a member of the shieldin complex, consisting of MAD2L2, SHLD1, SHLD2 and SHLD3, it controls DNA repair pathway choice by counteracting DNA end-resection. Here we investigated the requirements for shieldin complex assembly and activity. Besides a dimerization-surface, HORMA-domain protein MAD2L2 has the extraordinary ability to wrap its C-terminus around SHLD3, likely creating a very stable complex. We show that appropriate function of MAD2L2 within shieldin requires its dimerization, mediated by SHLD2 and accelerating MAD2L2-SHLD3 interaction. Dimerization-defective MAD2L2 impairs shieldin assembly and fails to promote NHEJ. Moreover, MAD2L2 dimerization, along with the presence of SHLD3, allows shieldin to interact with the TRIP13 ATPase, known to drive topological switches in HORMA-domain proteins. We find that appropriate levels of TRIP13 are important for proper shieldin (dis)assembly and activity in DNA repair. Together our data provide important insights in the dependencies for shieldin activity.

The mode of action that MAD2L2 employs in these different protein complexes is remarkably similar. MAD2L2 is a relatively small protein and consists mainly of a HORMA domain, originally defined by sequence conservation among the yeast proteins Hop1, Rev7, and Mad2 16 . HORMA binding to an interacting motif in its client protein promotes a striking topological remodeling of the HORMA domain from an "open" to a "closed" conformation of its "safety-belt", a structurally mobile element comprising the last~50 residues of the protein. This process is best understood for spindle assembly checkpoint (SAC) protein MAD2, where the considerable activation energy to be invested in the structural changes results in the creation of the very stable, albeit temporary, signaling complex. The structural conversion of MAD2 is catalyzed, both at the assembly and the disassembly level, by specialized protein machinery, allowing dynamic control of signaling. As part of that machinery, AAA + -ATPase TRIP13 has been shown to convert "closed" MAD2 to "open" MAD2 [17][18][19][20] . Core properties of MAD2 signaling are conserved in MAD2L2, including the capture of peptides in a closed conformation and dimerization. Indeed, MAD2L2 was found to entrap peptides of different interactors, including REV3, CHAMP1 and RAN GTP, through its safety-belt 3,21,22 . These proteins interact with MAD2L2 through a REV7 Binding Motif (RBM), a consensus sequence defined as ϕϕxPxxxpPSR 21,23,24 . REV3 contains two of these RBMs, that both are important for the function of Polζ in ICLrepair 24 . Interestingly, MAD2L2 is able to bind to both RBMs on REV3 simultaneously and forms a REV3-mediated homodimer, which is important for TLS 25,26 .
Within the shieldin complex, MAD2L2 directly interacts with SHLD3 12,27 . This interaction relies on an RBM motif in SHLD3 and is abolished by mutations in the MAD2L2 safety-belt region 12,28 . Hence it resembles the interaction between MAD2L2 and REV3. Interestingly, and again in analogy to the REV3-interaction, SHLD3 contains two RBM motifs 12 . However, different from REV3, only one of the SHLD3 RBMs (RBM2) was found important for shieldin function, raising uncertainty on whether MAD2L2 dimerizes in shieldin. We therefore set out to study the importance of MAD2L2 dimerization in the assembly and function of shieldin. Interestingly, we find that abolishing the ability of MAD2L2 to dimerize prevents MAD2L2 from interacting with SHLD2 and impairs its function in NHEJ. Indeed, a recent report showing the crystal structure of a MAD2L2-SHLD3 1-64 -SHLD2 1-52 complex revealed a MAD2L2 dimer within shieldin 29 . Interestingly, analogous to MAD2, the AAA + family ATPase TRIP13 was recently shown to facilitate the opening of the MAD2L2 safety-belt, thereby affecting shieldin complex formation 30 . Indeed, we find that TRIP13 interacts with MAD2L2 in the shieldin complex. We show that this critically depends on SHLD2-driven MAD2L2 dimerization and SHLD3, and that TRIP13 affects shieldin (dis)assembly and DNA repair. Altogether, our work reveals that appropriate activity of shieldin depends on MAD2L2 dimerization to drive appropriate assembly and disassembly of shieldin, facilitated by regulatory activity by TRIP13.

Results
MAD2L2 forms a dimer mediated by SHLD2 and important for shieldin assembly. To understand to what extent MAD2L2's ability to dimerize affects its function in shieldin we introduced dimer-breaking mutations in a flag-tagged RNAi-resistant (RR) MAD2L2 cDNA expression construct. We introduced an R124A amino acid substitution, originally used to disrupt homodimerization of MAD2L2 in crystallography studies and known to maintain MAD2L2 safety-belt interactions 3,12,21,27,31 . In addition, we changed two additional residues, K44A and A135D, that were recently identified to localize to a defined MAD2L2 dimerization-surface 25 . We thereby obtained expression constructs encoding for MAD2L2 harboring a single amino acid substitution (MAD2L2 R124A ; 1xMut) or three (MAD2L2 K44A/ R124A/A135D ; 3xMut) ( Fig. 1a and Supplementary Fig. 1a). We complemented MAD2L2-depleted cells with the different shRNA-resistant wild-type or mutant MAD2L2 expression constructs and confirmed their expression. As expected based on previous reports 25,31 , the mutations did not impair MAD2L2 protein stability (Fig. 1b). To address whether MAD2L2 dimerization affects shieldin complex formation we focused on the two shieldin proteins that directly interact with MAD2L2: SHLD3 and SHLD2. SHLD3 interacts with MAD2L2 through the MAD2L2 safety-belt, and SHLD2 interacts with MAD2L2-SHLD3 through its N-terminal region 9,10,12,27 . SHLD1, on the other hand, participates in shieldin by interacting with the C-terminal domain of SHLD2 10,27 . Thus, any impact on the interaction between MAD2L2 and SHLD2 would also affect the participation of SHLD1 in shieldin. We co-transfected 293T cells with epitope-tagged SHLD2, SHLD3, and wild-type MAD2L2 or the different MAD2L2-mutants and immunoprecipitated MAD2L2 in order to assess its interaction with SHLD2 and SHLD3. Interestingly, a single R124A amino acid substitution in MAD2L2 (1xMut) strongly reduced its ability to interact with SHLD2, but in line with previous in vitro data on MAD2L2 R124A -SHLD3 interaction 27,28 , did not disturb the interaction of MAD2L2 with SHLD3 (Fig. 1c). Moreover, three single amino acid changes in the dimerization domain of MAD2L2 completely disrupted its ability to interact with SHLD2, while only slightly affecting its interaction with SHLD3 (Fig. 1c). This indicates that in vivo MAD2L2 dimerization is necessary for efficient interaction of MAD2L2 with SHLD2, but is largely dispensable for MAD2L2-SHLD3 interaction. Indeed, size-exclusion chromatography (SEC) on purified shieldin proteins indicated that SHLD2 1-95 induces the dimerization of MAD2L2 in vitro, suggesting that a dimer of MAD2L2 is present in a full-length shieldin complex interacting with single molecules of SHLD3 and SHLD2 ( Fig. 1d and Supplementary Fig. 1b, c). Of note, both the wild-type and MAD2L2 (1xMut) appear as monomers in the SEC experiment shown in Fig. 1d. This is expected as the reported dimerization affinity of MAD2L2 is weak 25 , and proteins were applied in low concentrations.
In MAD2, dimerization is known to accelerate the slow, but spontaneous structural conversion from the open to closed topology required for MAD2 incorporation into the Mitotic Checkpoint Complex MCC 32 . In vitro, MAD2 conversion and MCC assembly requires several hours [33][34][35][36] . Indeed, when mixing MAD2L2 with SHLD3 we observed similar slow complex assembly. The kinetics of complex assembly were further attenuated when using a dimerization deficient MAD2L2 (1xMut, or 3xMut) (Fig. 1e, quantified in f). This could subsequently be partially rescued by forcing dimerization of MAD2L2 (1xMut) through a GST fusion (Fig. 1e, f, see Supplementary Fig. 1d for the GST-MAD2L2 dimer). Altogether, these results indicate that within shieldin, MAD2L2 interacts with SHLD3 as a dimer, and that MAD2L2 dimerization is induced by binding of SHLD2. This is in line with recent structural data showing MAD2L2 as a dimer in the shieldin complex 29 .
Efficient NHEJ at telomeres and in CSR requires MAD2L2 dimerization. To address the functional importance of MAD2L2 dimerization in DNA repair, we tested the ability of the different MAD2L2 dimerization-mutants to promote NHEJ. For this we made use of Trf2 −/− ;p53 −/− mouse embryonic fibroblasts (MEFs) harboring a temperature-sensitive allele of the telomeric shelterin component TRF2 (TRF2ts) that provide a system for assessing ligase 4-dependent classical NHEJ at natural chromosome ends following TRF2-inactivation mediated telomere deprotection 37  Triple mutation of residues K44, R124, and A135 (3xMut) in the dimerization interface of MAD2L2 has the same effect as 1xMut on binding kinetics. Inducing dimerization by creating GST-fusion partially rescues the assembly kinetics. Representative of n = 4 (MAD2L2 WT /MAD2L2 1xMut ) or n = 2 (MAD2L2 3xMut /GST-MAD2L2 1xMut ). f Graphical representation of data in e. Red curve represents binding kinetics between MAD2L2 WT and SHLD3  peptide. Light blue and green curves represent the reduction in binding kinetics of MAD2L2 1xMut and MAD2L2 3xMut with SHLD3 28-83 . Dark blue curve represents partial restoration of binding kinetics between MAD2L2 1xMut and SHLD3  induced by dimerization of the GST-tag. Graphs represent mean ± s.e.m, n = 4 for MAD2L2 WT /MAD2L2 1xMut , n = 2 for MAD2L2 3xMut /GST-MAD2L2 1xMut . Source data are provided as a Source Data file. telomeres, leaving telomeric chromosome ends deprotected and causing them to elicit a DNA damage response. This results in processing of the deprotected telomeres through NHEJ, giving rise to chromosome end-to-end fusions, and eventually cell death due to genomic crisis 38 (Fig. 2a). Depletion of factors critical for NHEJ, including MAD2L2 and other shieldin components, prevents chromosomal fusions and cell death 5,9,27,38 . To assess the impact of disrupting MAD2L2 dimerization on NHEJ, we restored expression of MAD2L2 in MAD2L2-depleted TRF2ts MEFs with wildtype or dimerization mutant MAD2L2 cDNAs and quantified chromosome end-to-end fusions upon 24 h of telomere uncapping at the non-permissive temperature (Fig. 2b). As demonstrated before 5 , chromosomal fusions upon telomere deprotection in MAD2L2-depleted cells were strongly diminished, indicative of defective NHEJ (Fig. 2c, d). Complementation of MAD2L2depleted cells with exogenous full-length wild-type MAD2L2 restored chromosome end-to-end fusion (Fig. 2c, d). Interestingly however, the single MAD2L2 R124A mutant (1xMut), that partially lost SHLD2 interaction, was significantly impaired in its ability to support chromosomal fusion upon telomere deprotection. Moreover, in line with the complete loss of MAD2L2-SHLD2 interaction, the triple MAD2L2 K44A/R124A/A135D mutant (3xMut) was completely defective in promoting chromosome end fusion in a background of endogenous MAD2L2 depletion (Fig. 2c, d), indicating that this dimerization mutant cannot support NHEJ at telomeres. These findings were recapitulated in an assay for cell survival upon prolonged telomere uncapping (Fig. 2b, e and Supplementary Fig. 1e, f). In line with our previous work 5 , control cells expressing endogenous MAD2L2 died of crisis upon prolonged telomere uncapping, while MAD2L2 depletion allowed these cells to survive. Consistent with their ability to promote telomere fusion, this survival was abolished by re-introducing full-length wild-type MAD2L2, but not by the single and triple MAD2L2-mutants ( Fig. 2e and Supplementary Fig. 1e, f).    Fig. 2 MAD2L2 dimerization is needed for MAD2L2-mediated control of NHEJ at deprotected telomeres and MAD2L2-mediated class switch recombination. a Experimental approach to address telomere deprotection induced NHEJ. b Schematic overview of the experimental setup used in c-e. c Chromosomal fusions in TRF2ts MEFs transduced with indicated shRNAs and expression constructs (n = 3 independent experiments, mean ± s.e.m., a minimum of 1500 chromosomes were counted per condition, per experiment). One-way ANOVA with Dunnett's multiple comparisons test (**p = 0.0046; ***p = 0.0002). d Representative images of metaphases counted in c. Scale bar, 10 μm. e Survival assay of TRF2ts MEFs used in c-d Cells were cultured at 32°C to exclude toxicity of the infected cells, or at 39°C for 12 days to induce telomere deprotection. Cells were returned to 32°C for the indicated times, and stained to assess survival. Plates are representative of n = 2 independent biological duplicates, the second biological duplicate is shown in Supplementary Fig. 1e, f. f CSR defect in Mad2l2 −/− cells is rescued after complementation with WT MAD2L2 and 1xMut but not with 3xMut. Bars represent mean ± s.e.m., two-sided Wilcoxon-Mann -Whitney test. Significance; ns, not significant (p ≥ 0.05); *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.
In addition, we addressed the ability of the different MAD2L2mutants to promote NHEJ during immunoglobulin class switch recombination (CSR), a physiological process that relies on classical NHEJ-mediated joining of activation-induced cytidine deaminase-mediated DSBs ( Supplementary Fig. 2a for experimental setup). As we demonstrated before 5,9 , MAD2L2-deficient cells are strongly impaired in CSR. This CSR defect could be fully restored by complementing MAD2L2-deficient cells with exogenous wild-type MAD2L2 (WT) or with MAD2L2 R124A (1xMut), but not with MAD2L2 K44A/R124A/A135D (3xMut) that lost the capability to interact with SHLD2 in the shieldin complex ( Fig. 2f and Supplementary Fig. 2b-d). These findings were further recapitulated by assessing the ability of MAD2L2 to prevent Igh breaks and translocations in CH12F3 cells. While Igh breaks and translocations were strongly increased in MAD2L2deficient CH12F3 cells, these were restored to normal low levels by complementation with wild-type MAD2L2 (WT) or MAD2L2 R124A (1xMut), but not with MAD2L2 K44A/R124A/ A135D (3xMut), indicating that these residues are critical for proper MAD2L2 function at DNA breaks at the Igh locus (Fig. 2g, h). Interestingly, our finding that NHEJ during CSR, but not telomere-NHEJ, can still be fully supported by the dimerizationdefective MAD2L2 (1xMut), indicates that CSR and telomere-NHEJ have different requirements. These might relate to differences in the underlying biology. We hypothesize that for CSR, in which AID induces a DSB at a defined locus, the low level of shieldin assembly that still takes place with the 1xMut is sufficient to maintain functional AID-DSB repair. CSR might therefore be less sensitive to partial defective shieldin complex, such as with the MAD2L2 (1xMut), compared to telomere-NHEJ in which multiple telomeres get deprotected at the same time. Moreover, as 30-50% of CSR can still occur independently of ligase-4/XRCC4-dependent NHEJ, it is possible that in a situation with partially defective shieldin complex some of the breaks are repaired through alternative end-joining.
Altogether this indicates that the interactions mediated by the K44, R124 and A135 residues on MAD2L2, involved in MAD2L2 dimerization, are critical for the ability of MAD2L2 to function in NHEJ.
The ATP-ase TRIP13 interacts with shieldin in a MAD2L2 dimerization and SHLD3-dependent manner. To obtain more insight in shieldin complex assembly we set out to identify MAD2L2 interacting proteins with a potential role in MAD2L2 complex formation. For this, we immunoprecipitated MAD2L2 from HeLa cells depleted for endogenous MAD2L2 and stably expressing GFPtagged MAD2L2, and analyzed co-precipitating peptides by mass spectrometry. Cells were either left untreated or harvested 3 h after inducing DNA damage by irradiation (IR) with 25 Gy. Aside from SHLD2 and SHLD3, we identified multiple additional interacting proteins ( Supplementary Fig. 3a). Among these, the AAA + family ATPase TRIP13 in particular caught our attention. One reason for this being its known interaction with MAD2 and other HORMAdomain containing proteins 39 . The other reason being that TRIP13 was among positive hits in the same functional genetic screen for regulators of telomere-induced genomic instability, that originated our identification of MAD2L2 as regulator of NHEJ, suggesting that TRIP13 acts in similar processes as MAD2L2 ( Supplementary  Fig. 3b, c) 5 . Indeed, we could confirm an interaction between MAD2L2 and TRIP13 in co-immunoprecipitation assays with exogenous epitope-tagged MAD2L2 and TRIP13, as well as with endogenous MAD2L2 in cells expressing GFP-tagged ATPase-dead TRIP13 EQ (Fig. 3a, b). Interestingly, MAD2L2 dimerization appears critical for interaction with TRIP13. In vitro, MAD2L2 directly interacts with TRIP13, but only when dimerization was induced by a GST-tag on MAD2L2 or by the presence of SHLD2 1-95 (Fig. 3c, d, f, also see Supplementary Fig. 1d). In addition, interaction of TRIP13 with MAD2L2 requires full-length SHLD3 to be part of the complex. In absence of full-length SHLD3 the interaction of a MAD2L2 dimer with TRIP13 was only very weak or non-detectable (Fig. 3c, compare lanes 7 and 9). Furthermore, SEC revealed that, analogous to MAD2, a hexamer of ATPase activity deficient TRI-P13 E253Q forms a stable complex with MAD2L2-SHLD3 and MAD2L2-SHLD3-SHLD2  in the presence of ATP in vitro (Fig. 3e, f). Interestingly, while TRIP13 association with MAD2 requires p31 comet as a cofactor 39 , the formation of a stable TRIP13shieldin complex apparently does not need a cofactor. However, p31 comet does seem important for the activity of TRIP13 towards shieldin 40 .
Altogether, our in vivo and in vitro analysis indicate that TRIP13 interacts with shieldin in a manner requiring MAD2L2 dimerization supported by SHLD2, and SHLD3.
TRIP13 affects DNA repair. We next addressed the functional relevance of the interaction between TRIP13 and shieldin. We hypothesized that similar to what is known for MAD2 17-20 , TRIP13 might facilitate a conformational switch in MAD2L2, providing the energy to open up MAD2L2 and release its interacting partner(s). This would allow MAD2L2 to shuttle between different protein complexes and activities, or facilitate release of shieldin activity after DNA repair has advanced or completed. Moreover, by promoting shieldin-disassembly, TRIP13 could promote HR and inhibit NHEJ at the right place and time. Hence, disturbing such a role by TRIP13 might impact DNA repair. To address this we overexpressed exogenous wild-type (WT) or catalytic inactive (ATPase-dead) TRIP13 (TRIP13 EQ ), or depleted endogenous TRIP13 in different cell lines. We confirmed effective ablation of TRIP13 and functionality of exogenous wild-type or EQ-mutant TRIP13 expression by making use of the notion that a functional SAC requires TRIP13 ATPase-activity 41 (Supplementary Fig. 4a-e). Indeed, unlike TRIP13 proficient cells, cells completely deficient for TRIP13 failed to arrest in mitosis upon mitotic spindle disruption with nocodazole, indicating a dysfunctional SAC. Exogenous expression of wild-type TRIP13, but not TRIP13 EQ , restored the ability of TRIP13 knockout cells to arrest in mitosis.
We first investigated if TRIP13 affects DNA end-resection, as MAD2L2 and shieldin factors are known to promote NHEJ and inhibit HR by counteracting DNA end-resection. Interestingly, stable overexpression of wild-type TRIP13, in both Hela and U2OS cells, resulted in increased IR-induced phosphorylation of the single-stranded DNA binding protein RPA on Ser4/8 (p-RPA), indicative of elevated end-resection ( Fig. 4a, b). On the other hand, overexpression of ATPase-dead TRIP13 EQ did not, or to much lesser extent, increase p-RPA. As TRIP13 functions as a hexamer, we hypothesize that the overexpression of TRIP13 EQ in the presence of endogenous TRIP13 can contribute to the formation of TRIP13-hexamers with partial activity. In addition, IR-induced foci formation by RPA appeared slightly elevated (not reaching statistical significance) upon overexpression of wild-type TRIP13, but not TRIP13 EQ , in U2OS cells (Fig. 4c). As TRIP13 enhanced DNA end-resection in our assays, we hypothesized that overexpression of TRIP13, by facilitating end-resection, would render cells more efficient in HR. To address this, we treated U2OS cells expressing WT or EQ-mutant TRIP13 with the PARP-inhibitor olaparib and addressed survival, as resistance to PARP inhibition is dependent on HR. Indeed, overexpression of WT TRIP13, but not the EQ mutant, caused enhanced resistance to treatment with olaparib ( Fig. 4d and Supplementary Fig. 5a). In line with this, cells overexpressing TRIP13 showed enhanced HR efficiency in a DR-GFP reporter assay ( Fig. 4e and Supplementary  Fig. 5b). Finally, we addressed whether the ability of TRIP13 to enhance end-resection depends on MAD2L2. As expected from earlier work 5,6 , depletion of MAD2L2 resulted in elevated IRinduced p-RPA at 3 h post IR (Fig. 4f). While overexpression of TRIP13 enhanced p-RPA in MAD2L2-proficient cells, it did not further enhance end-resection in a MAD2L2 KO background, indicating that TRIP13 controls end-resection in a MAD2L2dependent manner. Thus, opposite of the activity of MAD2L2 that counteracts resection and HR, TRIP13 promoted DNA endresection and HR when overexpressed. This is in line with the interaction we observed between MAD2L2 and TRIP13, with a regulatory activity of TRIP13 on MAD2L2 that is related to the topological switches it induces in HORMA proteins and with a recent report showing that TRIP13 overexpression promotes shieldin-disassembly 30 .
As we reasoned that also absence of TRIP13 might interfere with proper MAD2L2 function we addressed the consequences of CRISPR/Cas9-mediated TRIP13-deficiency for DNA endresection in stable polyclonal TRIP13 KO U2OS cells, in clonal TRIP13 KO Hela and U2OS cell lines, and upon short-term polyclonal depletion of TRIP13 with lentiviral sgRNAs (Fig. 4g-j,  Supplementary Figs. 4f and 5c, d, f, g). Interestingly, upon different conditions of long-term TRIP13 inactivation in U2OS and HeLa cells, IR-induced p-RPA levels were increased ( Fig. 4g and Supplementary Fig. 5c). IR-induced RPA foci were also increased in both MAD2L2 and TRIP13 KO lines (Supplementary Figs. 4f and 5d). Of note, shRNA-mediated depletion of MAD2L2 could still further enhance p-RPA levels in a TRIP13 KO background ( Supplementary Fig. 5e). While this could indicate that shieldin-independent roles of TRIP13 contribute to the increased p-RPA in TRIP13-KO lines, it could also reflect more efficient interference with shieldin function that is  Relative protein amount (normalised on SHLD3) potentially not complete with either shRNA-mediated depletion of MAD2L2 (despite being very efficient) or TRIP13 KO alone. Conversely, and in line with previous reported findings 30 , efficient but not full disruption of TRIP13 with independent lentiviral sgRNAs was generally associated with signs of reduced end-resection, showing a slight decrease in p-RPA levels and RPA foci in MEFs and U2OS at 3 or 24 hrs post IR (Fig. 4h, i and Supplementary Fig. 5f, g). As RPA responses were mild and for RPA foci not reaching statistical significance over limited number of replicates, we further addressed the effect of TRIP13 depletion on resection by using a PCR-based resection assay 42,43 . As expected, the amount of ssDNA was enhanced in absence of MAD2L2. TRIP13 depletion had the opposite effect and decreased the amount of resection ( Fig. 4j and Supplementary  Fig. 5h). In line with this, TRIP13 depletion also decreased HRefficiency in a DR-GFP reporter assay (Fig. 4e). So while longterm complete TRIP13 disruption appears associated with signs of elevated end-resection ( Fig. 4g and Supplementary Figs. 4f and 5c-e), short-term TRIP13 depletion seems associated with decreased resection (Fig. 4h-j and Supplementary Fig. 5f, g). We hypothesize that the observed increase in RPA or p-RPA in long-term TRIP13 KO cells reflects disturbed shieldin complex dynamics, or might be a consequence of shieldin-independent roles of TRIP13, as further eluted to in the discussion below.
Overall, the effects of TRIP13 on resection were less prominent than of MAD2L2, which could well be a reflection of manipulation of a factor that affects the kinetics of shieldin (dis)assembly being less severe than removing a component of shieldin itself. Also, this would be consistent with MAD2L2 spontaneously but slowly disassembling from its interactors over time without the need for TRIP13, in line with the known spontaneous but very slow disassembly of MAD2L2-SHLD3 and other HORMA interactions 28,34 .
Given the disturbed end-resection control in TRIP13-depleted cells, we assessed sensitivity to olaparib, both in wild-type cells and in BRCA1-deficient cells that are hypersensitive to olaparib due to HR-deficiency. Efficient depletion of TRIP13 in U2OS or RPE-WT cells did not clearly affect the sensitivity to olaparib ( Supplementary Fig. 6a, b). However, both shRNA-and sgRNAmediated depletion of TRIP13 in BRCA1-deficient RPE cells rendered these cells less sensitive to olaparib ( Supplementary  Fig. 6c, d). Although less pronounced, this effect of TRIP13 depletion resembled that of MAD2L2 depletion in BRCA1deficient cells, observed in the same assays and as reported before 6 . As PARP inhibitor resistance upon depletion of MAD2L2 in BRCA1-deficient cells is known to be mediated by restoration of end-resection and HR 6 , this could indicate that HR is also partially restored in TRIP13-depleted BRCA1-deficient cells. This is in contrast to a recent report showing enhanced sensitivity to PARP inhibitors in TRIP13-depleted BRCA1deficient SUM149PT cells 30 and to the decreased HR and resection in BRCA1-proficient U2OS cells we observed upon TRIP13 depletion. These differences might be due to different cell types used, as cells differ in the amount of free MAD2L2 to engage in different protein complexes, further eluted to in the discussion. They may also relate to the signs of elevated resection we observe with long-term depletion of TRIP13 that might come through in these survival assays of longer duration and might result from impaired shieldin complex dynamics or shieldinindependent effects. Furthermore, as MAD2L2 has a prominent role in ICL-repair, we addressed the sensitivity of TRIP13 KO cells to cisplatin induced ICLs. Interestingly, TRIP13 KO U2OS cells displayed modest sensitivity to cisplatin ( Supplementary  Fig. 7a). Although this did not reach statistical significance over 3 replicates, it may indicate that TRIP13 to some extent also impacts other MAD2L2-functions, such as in ICL-repair.
Finally, we studied the role of TRIP13 at the level of NHEJ. Overexpression of TRIP13 in TRF2ts MEFs did not significantly affect the amount of chromosome end-to-end fusions upon telomere uncapping (Supplementary Fig. 7b, c), suggesting that it did not interfere with the ability of MAD2L2 to promote telomere-NHEJ. However, depletion of TRIP13 using two different shRNAs significantly reduced the amount of chromosomal fusions, albeit to lesser extent than the depletion of MAD2L2 did (Fig. 5a and Supplementary Fig. 7d). This is in line with TRIP13 being among the hits of the same shRNA screen that originally identified MAD2L2 as positive regulator of NHEJ and relies on cell survival upon prolonged telomere uncapping as proxy for impaired NHEJ at telomeres, as extensive fusion of uncapped telomeres causes p53-deficient cells to die of genomic crisis 5 (Supplementary Fig. 3b, c). Interestingly, opposite from the effect on telomere-NHEJ, depletion of TRIP13 raised NHEJ during CSR (Fig. 5b and Supplementary Fig. 7e-h). Moreover, the increased CSR-efficiency was dependent on SHLD1 and MAD2L2 (Fig. 5b), indicating that TRIP13 depletion increases CSRefficiency through the shieldin complex. Thus, while this clearly indicates that in absence of TRIP13, normal control of NHEJ is perturbed, NHEJ at telomeres and at immunoglobulin loci are affected differently. We hypothesize that this might be due to differences in the biology underlying telomere-NHEJ and CSR, or might be related to additional roles for TRIP13 in the context of telomere-NHEJ, as we will further discuss below.
Together our data indicate that aberrant TRIP13 levels, both too much or too little, interfere with normal control of DNA repair, with the outcome for DNA repair being context dependent.
TRIP13 controls the assembly of SHLD2/3 at DNA breaks. As DNA repair related processes under control by shieldin were affected by TRIP13, we hypothesized that aberrant expression of TRIP13 impairs shieldin complex dynamics. To address this we studied shieldin complex formation in the absence of TRIP13. We expressed 3xflag-tagged MAD2L2 in 293T cells depleted for TRIP13 or in control cells (sgNT), followed by MAD2L2immunoprecipitation and mass-spectrometry analysis of immunoprecipitated complexes (Fig. 6a). Interestingly, while the interaction of MAD2L2 with SHLD1 and SHLD2 was not affected, the interaction of MAD2L2 with SHLD3 seemed enhanced in TRIP13 KO cells, despite the difference not being statistically significant over 3 replicate experiments (Fig. 6a). Since SHLD3 is interacting with MAD2L2 through the MAD2L2 safety-belt, the SHLD3-MAD2L2 interaction is the most likely interaction in shieldin to be affected by TRIP13, that would be expected to be stabilized in the absence of TRIP13. To address this we studied the recruitment of the different shieldin complex members to sites of broken DNA in the presence or absence of TRIP13. For this we expressed GFP-tagged SHLD2 or SHLD3 in U2OS-DSB reporter cells, in which DNA breaks are induced with 4-OHT and shield-1 inducible mCherry-LacI-Fok1 nuclease 44 , and depleted TRIP13 (Fig. 6b-e and Supplementary Fig. 8a-c). As expected, as shieldin localization to DSBs depends on ATM kinase activity 5,7 , ATM-inhibitor treatment strongly decreased GFP-SHLD2 and GFP-SHLD3 foci intensity at mCherry-marked DSBs. Interestingly, cells depleted for TRIP13 displayed increased GFP-SHLD3 and GFP-SHLD2 accumulation at DSBs. This illustrates that in absence of TRIP13 accumulation of shieldin at DNA DSBs is increased and indicates that TRIP13 affects the recruitment or the stability of the shieldin complex at DNA breaks.

Discussion
MAD2L2 engages in multiple complexes to coordinate essential cellular processes, including TLS by participating in Polζ and DSB repair by being a central component of shieldin. By doing so MAD2L2 affects the development and treatment of human disease, such as Fanconi Anemia and cancer, in which it determines the sensitivity of BRCA1-deficient cancers to treatment with Parp inhibitors. Given these critical roles, understanding the requirements for proper MAD2L2 activity is important. Here, we show how dimerization of MAD2L2 is critical for shieldin complex formation and activity in DNA repair and is promoted by SHLD2. Moreover, MAD2L2 dimerization and the presence of SHLD3 allow the shieldin complex to interact with the ATPase TRIP13, which is important for the appropriate dynamics of shieldin complex assembly and appropriate control of DNA repair reactions.
For the function of MAD2L2 in TLS the formation of a REV3tethered MAD2L2-homodimer, facilitated by two RBM motifs in REV3, is essential 24,25 . SHLD3 also contains two putative RBM motifs, however only one was found important for the function of shieldin 12,28 . Whether the formation of a functional shieldin complex would involve and require dimerization of MAD2L2 was therefore not immediately obvious. Here we show that MAD2L2 indeed forms a dimer within the shieldin complex. This dimer formation is important for the assembly of shieldin and is stabilized by the SHLD2 component. We find that in absence of MAD2L2 dimerization, SHLD2 cannot participate in the shieldin complex in vivo. This finding explains a previous reported result on a K129A-mutation in MAD2L2 that reduced the interaction of SHLD3-MAD2L2 to SHLD1 and SHLD2 12 , as this mutation is located in the MAD2L2 dimerization domain. In line with its necessity for productive shieldin assembly, we show that MAD2L2 dimerization is essential for the function of MAD2L2 in the repair of deprotected telomeres through NHEJ, as well as the appropriate NHEJ-mediated joining of DSBs in immunoglobulin loci during CSR. While completing this manuscript, the dimerization of MAD2L2 was also revealed in the crystal structure of MAD2L2-SHLD3 1-64 -SHLD2 1-52 , showing that a dimer of MAD2L2 molecules is embraced by a peptide of SHLD3 29 . At the C-terminus of the SHLD3 1-64 peptide, the RBM forms the canonical safety-belt with the MAD2L2 molecule, while the N-terminus of the peptide forms a novel safety-belt interaction with the other MAD2L2 molecule that is mediated by the SHLD2 peptide. This clarifies why previously only one of the SHLD3 RBMs was found important for shieldin function, while we find here that dimerization of MAD2L2 is critical for its in vivo function in shieldin. In addition, our finding that SHLD2 supports the dimerization of MAD2L2 in vitro, is in line with the proposal from the structural data that SHLD2 primes the non-canonical interaction between the SHLD3 N-terminus and MAD2L2.
Importantly, we show that MAD2L2 dimerization not only promotes interactions within shieldin but also brings MAD2L2/ shieldin the opportunity to interact with the TRIP13 ATPase. TRIP13 was recently reported to induce the dissociation of MAD2L2-SHLD3 through opening the MAD2L2 safety-belt and subsequent release of SHLD3 30 , indicating that in analogy to MAD2, TRIP13 regulates MAD2L2 complex disassembly in the context of shieldin. Our initial biochemical reconstitution efforts highlighted surprising differences in the molecular mechanism of TRIP13 on shieldin, when compared to the MCC. In the MCC, p31 comet , itself in a permanent closed conformation, is an essential cofactor for MCC binding by bridging MAD2 to TRIP13 18,45 . This is in contrast to shieldin, where we observed no necessity for p31 comet in binding to TRIP13. p31 comet , however, does seem important in mediating TRIP13-activity towards the shieldin complex 40 . It seems likely that the closed MAD2L2, that interacts with RBM2 in SHLD3 via the canonical safety-belt interaction, will be the equivalent of the closed MAD2. Interestingly, the second MAD2L2, of which the seatbelt is invisible in the recent structure 29 , but the core fold suggests that it too is in a closed conformation, might therefore take over the role of (closed) p31 comet in bridging TRIP13 towards the shieldin complex. Additionally, we observed that SHLD3 contributes to the interaction to TRIP13 outside of RBM2. It is currently unclear if SHLD3 provides direct binding to TRIP13, or whether longer segments are needed to fully stabilize the dimerization and thus interaction with TRIP13. Future detailed reconstitution efforts and structural studies will be necessary to explain the functional relevance of these differences.
Indeed, we find that the assembly of shieldin at DSBs and DNA repair reactions are sensitive to the presence and activity of TRIP13. Overexpression of TRIP13 WT but not the ATP-ase dead TRIP13 EQ increased the amount of ssDNA, increased HR-efficiency, and rendered cells resistant to PARP-inhibition. Moreover, depletion of TRIP13 stabilized SHLD3-MAD2L2 interaction, increased SHLD2 and SHLD3 accumulation at DSBs, decreased HR-efficiency and ssDNA formation, and enhanced NHEJ during CSR. This is in line with recent work 30 and supports that in analogy to the activity of TRIP13 towards other HORMA interactions, enhanced TRIP13 activity dissociates shieldin, thereby dismantling its inhibitory activity to end-resection, while lack of TRIP13 activity stabilizes shieldin. However, we also observed seemingly opposite effects of TRIP13 depletion on DNA repair, such as signs of elevated end-resection in longterm TRIP13 KO cells, olaparib resistance in BRCA1-/-RPE cells and, in line with an earlier notion of impaired NHEJ in TRIP13deficient cells 46 , impaired telomere-NHEJ. These observations indicate that the consequences of TRIP13 depletion are less straightforward and might reflect a more widespread effect of TRIP13 depletion on MAD2L2 interactions that may enhance shieldin function but may also negatively affect shieldin function, depending on context and read-out. We hypothesize that this is a reflection of MAD2L2 being a bi-stable system in which TRIP13 deficiency can yield two results that differentially impact reactions involving MAD2L2, depending on context and specific requirements of those reactions. On one hand the absence of TRIP13 slows down disassembly of shieldin and might therefore improve certain DNA repair reactions as it gives a more constitutive shieldin complex. However, in absence of TRIP13 promoting disassembly of MAD2L2 from the different complexes it engages in, MAD2L2 would also be locked in its interaction with partner proteins, such as shieldin, Polζ, CHAMP1 and others. We envision that this would strongly impair the ability of MAD2L2 to shuttle between the different complexes it participates in, reduce or ablate the free pool of MAD2L2 available to engage in specific complexes and activities when needed, and interfere with the temporal nature of MAD2L2 complex activity that potentially underlies normal control of the biological processes it operates in. Hence, by interfering with normal release and temporary action of MAD2L2 containing complexes, TRIP13 deficiency could both enhance and impair MAD2L2 controlled processes. The outcome might also be cell type-dependent as cells differ in amounts of free MAD2L2 and the distribution of MAD2L2 over different complexes. It might also be process dependent as different processes have different requirements due to the underlying biology. For instance, telomere end-joining relies heavily on ligase 4/XRCC4 and on the joining of distant deprotected chromosome ends that need to find each other, while end-joining in CSR can still occur for 30-50% in absence of ligase 4/XRCC4 and involves cis-joining events of switch-junctions brought together within CSR centers [47][48][49] . Affecting the MAD2L2-SHLD1/2/3 complex dynamics through modulation of TRIP13 levels might have different outcomes based on these differences. Cell cycle dependencies might also factor in the different outcomes; as NHEJ is preferred in G1, it is possible that the activity of TRIP13 is mostly important in S/G2 to promote efficient HR by inducing shieldin complex disassembly. In addition, it is also possible that some of the effects observed with TRIP13 deficiency are a consequence of other roles of TRIP13, outside of regulatory activity towards MAD2L2.
Overall, our data support a model in which dimerization of MAD2L2 is critical for shieldin assembly and allows TRIP13 to control appropriate dynamics and activity of the shieldin complex (Fig. 7). The modulation of the reversible topology switch in the MAD2L2 core structure by TRIP13 is expected to regulate its protein-protein interaction potential and kinetics. We therefore speculate that TRIP13 is part of a larger regulating network that will control that shieldin is assembled and disassembled in the right place and at the right time. TRIP13, as an enzymatic regulator of shieldin, has been proposed as an interesting candidate to explore as drug target to modulate Parp inhibitor responses 30 . However, the different outcomes we observed with TRIP13-deficiency and the regulation of multiple MAD2L2-or HORMA-complexes by TRIP13, indicate that this might not be straightforward and warrant thorough further exploration of effectiveness and potentially unwanted consequences.  37,38 were cultured in Dulbecco's modified Eagle medium with 10% FBS (Sigma), supplemented with 100 U penicillin, 100 μg ml −1 streptomycin and 2 mM L-Glutamine (Gibco, Life Technologies). CH12F3 (CH12) 53 cell lines were cultured in RPMI 1640 supplemented with 10% FBS, 100 U ml −1 penicillin, 100 μg ml −1 streptomycin, 50 μM 2-mercaptoethanol, 1x MEM non-essential amino acids, 1 mM sodium pyruvate and 10 mM HEPES. HeLa, U2OS, RPE, 293Ts, Phoenix or CH12 cells were cultured at 37°C, TRF2ts MEFs were cultured at the permissive temperature of 32°C, all with 5% CO 2 . BRCA1 −/− RPE cells were maintained in a low oxygen (3%) incubator. To induce GFP-TRIP13 expression in HeLa cells, cells were treated with 0.5 μg ml −1 doxycycline for 24 h. To evaluate long-term survival upon prolonged telomere deprotection, TRF2ts MEFs were seeded at 40,000 cells/10 cm plate and from the next day cultured at the non-permissive temperature (39°C) for 12 days. Cells were then returned to 32°C for the indicated times, fixed with 4% formaldehyde and stained with 0.1% crystal violet. Crystal violet was extracted with 10% acetic acid and absorbance at 595 nm was measured in a Tecan microplate reader (Infinite M200pro, TECAN) for quantification.
For experiments in FokI DSB reporter cells, SHLD2 and SHLD3 expression constructs were cloned into a pMSCVblast-eGFP retroviral backbone using Gateway technology (Invitrogen).
Immunoblotting. Immunoblotting was done according to standard protocols and as described before 5 . Briefly, cells were washed with PBS and collected by scraping in 2xSDS sample buffer followed by boiling at 95°C. Alternatively, cell extracts were made using an Urea-SDS buffer ( Immunoprecipitations. For flag-immunoprecipitation experiments, 293T cells were seeded at 5 × 10 6 cells/10 cm plate and transfected with 10 μg of the indicated combinations of tagged-cDNA constructs using standard CaPO 4 transfection. Cells were collected 48 h post transfection and resuspended in 1 ml lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA and 1% triton X-100) supplemented with protease-and phosphatase-inhibitors (cOmplete Tablets EDTA-free protease cocktail tablets, PhosStop, Roche), 10 mM iodoacetamide (I1149, Sigma) and 5 mM Sodium Butyrate (B5887, Sigma). Lysates were kept on ice for a total of 30 min while pipetting extensively every 10 min followed by centrifugation (20,000 × g) after which an input sample was taken. Lysates were added to anti-Flag M2 Magnetic beads (M8823, Sigma-Aldrich) pre-washed in TBS (50 mM Tris HCl, 150 nM NaCl pH 7.4) and incubated at 4°C overnight. After washing with TBS, beads were resuspended in 2xSDS sample buffer and boiled at 95°C for 5 min to elute immunocomplexes. For GFP-immunoprecipitation assays, cells expressing the indicated constructs were plated in 15 cm plates, irradiated or left untreated and collected in ice-cold PBS by scraping after washing three times with PBS. Immunoprecipitation was done using GFP-Trap beads (chromotek) essentially according to the manufacturers protocol. Cells were resuspended in ice-cold lysis buffer (10 mM TrisHCl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5% NP-40), supplemented as above with protease-and phosphatase-inhibitors, iodoacetamide and Sodium Butyrate, and incubated on ice for 30 min while pipetting extensively every 10 min. Cell lysates were centrifuged (20,000 × g for 10 min), the pellet was discarded and the lysate was supplemented with dilution buffer (10 mM TrisHCl pH 7.5, 150 nM NaCl, 0.5 mM EDTA) supplemented with the inhibitors indicated above and an aliquot was taken as input sample. Beads were equilibrated by washing three times with dilution buffer, and added to the lysates followed by rotation overnight at 4°C. After collecting the flow-through, beads were washed three times with dilution buffer. For elution of protein complexes, beads are resuspended in 100 μl 2xSDS-sample buffer and boiled for 10 min at 95°C. For analysis by mass spectrometry, MAD2L2-depleted HeLa cells were used complemented with GFP-MAD2L2 RR or a GFP-control. One sub-confluent 15 cm plate (~1 × 10 7 cells) was used per condition per experiment.
Mass spectrometry for MAD2L2 interacting proteins. For mass spectrometry, samples were run short-distance on a 4-12% SDS-PAGE gel and stained with Coomassie Blue. The lane was excised from the gel after which proteins were reduced with dithiothreitol and alkylated with iodoacetamide. Proteins were digested with trypsin (mass spec grade, Promega) overnight at 37°C and peptides were extracted with acetonitrile. Digests were dried in a vacuum centrifuge and reconstituted in 10% formic acid for MS analysis. Peptide mixtures (10% of total digest) were loaded directly on the analytical column and analyzed by nanoLC-MS/ MS on an Orbitrap Fusion Tribrid mass spectrometer equipped with a Proxeon nLC1000 system (Thermo Scientific).
Mass-spectrometry data analysis. Raw data were analyzed by Proteome Discoverer (PD) (version 2.3.0.523, Thermo Scientific) using standard settings. MS/MS data were searched against the Human Swissprot database (20,367 entries, release 2020_02) using Sequest HT. The maximum allowed precursor mass tolerance was 50 ppm and 0.6 Da for fragment ion masses. Trypsin was chosen as cleavage specificity allowing two missed cleavages. Carbamidomethylation (C) was set as a fixed modification, while oxidation (M) was used as variable modifications. False discovery rates for peptide and protein identification were set to 5% and as additional filter Sequest HT XCorr>1 was set. For the heatmap, samples were analyzed using the free R scripting language. The PD output file containing the PSM counts was loaded into R (version 4.0.2). Spectral counts of the biological replicates were averaged and proteins were filtered for zero counts in the control and for at least one count (average of two replicates) in one of the other two conditions. The counts were represented in a heat map with a color code ranging from white (0 PSM(s)) to red (≥5 PSM(s)). For the volcano plot, the PD output file containing the abundances was loaded into Perseus (version 1.6.14.0) 57 . LFQ intensities were Log2-transformed and the proteins were filtered for at least two out of three valid values. Missing values were replaced by imputation based on the standard settings of Perseus, i.e., a normal distribution using a width of 0.3 and a downshift of 1.8. Differentially expressed proteins were determined using a t-test Immunofluorescence. For immunofluorescence, 40,000-60,000 U2OS cells/well were seeded in 8-well chamber slides (Millicell EZ slide, Millipore) and irradiated as indicated in the individual experiments. For RPA foci, cells were pre-extracted for 5 min by adding ice-cold 0.5% Triton/PBS on ice prior to fixing. Then, cells were washed with PBS, fixed for 10 min with 2% paraformaldehyde followed by 10 min permeabilization in 0.5% Triton/PBS. After 1 h of blocking (0.02% Triton, 5% NGS, 5% FCS in PBS), cells were incubated with primary antibody in blocking solution overnight at 4°C. Primary antibodies used were RPA (RPA34-20, NA18, Calbiochem, 1:500). Next, cells were washed three times with 0.02% Triton/PBS and incubated with Alexa Fluor 488 or 568 goat anti-mouse or anti-rabbit IgG secondary antibodies (Invitrogen, 1:500) in blocking solution for 1 h. After washing three times with 0.02% Trition/PBS, slides were mounted using DAPI-containing Vectashield (Vector Laboratories).
For FokI DSB reporter experiments, U2OS ER-mCherry-LacI-FokI-DD cells 44 were stably transduced with pMSCVbl-eGFP-SHLD2 or pMSCVbl-eGFP-SHLD3 prior to transduction with the indicated sgRNA constructs. Cells were treated with shield-1 (1 μM) and 4-OHT (0.5 μM) for 3 h and harvested as described above. In ATM inhibitor conditions, cells were treated 10 min prior to addition of shield-1 and 4-OHT with 10 μM of ATM Kinase inhibitor (KU55933, cat#118500, Calbiochem). Cells were stained with a primary antibody against GFP (A11122, Life Technologies, 1:500) and further processed as described above. Slides were imaged using a confocal Leica SP5 system with X63 oil objective and LAS-AF software (version 2.7.4). Image analysis was performed with a customized macro on ImageJ (version 1.52p) software and data was processed using Microsoft Excel (version 16.16.27).
Metaphase chromosome analysis for telomere fusions. Cell harvesting, preparation of metaphases spreads and image acquisition was done as described before 5 . Briefly, Trf2 −/− ;p53 −/− ; TRF2ts MEFs were placed at the non-permissive temperature of 39°C for 24 h to induce telomere deprotection. To enrich for mitotic cells, 10 μg ml −1 colcemid (Karyomax) was added to the cells 2 h prior to harvesting. Cells were collected by trypsinization, resuspended in 0.075 M KCl and incubated for 7 min at 37°C followed by spinning down (5 min at 194 × g) and fixing in Methanol-Acidic Acid fixative. Telomere FISH was performed on chromatin spreads with an Alexa488-labeled (CCCTAA)3 (TelC) PNA probe (PN-TC060-005, Panagene/Eurogentec). Digital images of metaphases were captured using a Metafer4 (MetaSystems) equipped with an AxioImager Z2 microscope (Carl Zeiss) 5 .
DR-GFP assay. Stable DR-GFP U2OS cells were a kind gift from Jeremy Stark 50 . To address HR-efficiency, 2 × 10 5 cells were seeded in 12-well plates, 24 h prior to transfection with BFP and I-SceI using Lipofectamine 2000 (ThermoFisher). Cells were refreshed 3 h after transfection and harvested 72 h post transfection by trypsinization, followed by resuspension in PBS and flow cytometry analysis. Cells were gated for single cells and BFP, and the percentage of GFP-positive cells was quantified (Attune TM NxT Acoustic Focusing Cytometer, Thermo Fisher Scientific).
PCR end-resection assay. The procedure was carried out as described previously 42,43 . Briefly, 1.5 × 10 6 AID-AsiSI-ER-U2OS (AID-DIvA) cells 51 were seeded in 10 cm plates 24 h before treatment with 0.3 µM 4-hydroxytamoxifen for 4 h. Genomic DNA was isolated using DNeasy Blood & Tissue kit with RNAse A addition (Qiagen), treated with RNAse H and digested or mock digested with BanI (New England Biolabs) overnight. BanI was then heat inactivated and 12.5 ng of the gDNA was used in a 10 μl quantitative PCR reaction (QuantStudio 6 Flex Real-Time PCR System, Applied Biosystems) in triplicate with SensiFAST™ SYBR® (Bioline) using primers flanking the BanI site 200 bp from the inducible AsiSI cut site (see Supplementary Table 1). The percentage of ssDNA was calculated as previously described 43 .
Olaparib and cisplatin sensitivity assays. RPE or U2OS cells transduced with the indicated constructs were seeded at 1000-5000 cells/well (RPE or BRCA1 −/− RPE) or 800 cells/well (U2OS) into 6-well plates. For RPE, cells were seeded with either DMSO or the PARP inhibitor olaparib at the indicated concentrations. The medium was refreshed every 5 days, every condition in each experiment was seeded as a technical duplicate. For U2OS cells, olaparib was added at the indicated concentrations the day after seeding. For cisplatin sensitivity assays, 400 cells/well (U2OS) were seeded into 12-well plates as a technical duplicate for each experiment, and cisplatin was added at the indicated concentrations the day after seeding. After 10-12 days, cells were washed with PBS, fixed with 4% formaldehyde and stained using 0.1% crystal violet. The crystal violet was extracted using 10% acetic acid and quantified by measuring absorbance at 595 nm at a plate reader (Infinite M200pro, TECAN). For every condition, crystal violet intensity was normalized to the untreated (DMSO) control.
RT-PCR analysis. Igh α germ-line transcripts (αGLT) and Aid mRNA were quantified as previously described 9 . Briefly, total RNA was isolated from activated CH12 cells using the RNeasy mini kit (Qiagen) and cDNA was synthesized using the high capacity reverse transcription kit (Applied Biosystems). Semi-quantitative RT-PCR using 2.5-fold serial dilutions of cDNA was made from CH12 cells stimulated for 72 h with anti-CD40, IL-4, and TGF-β. Hprt was used as a control for transcript expression. Transcripts were amplified using oligonucleotides described in Supplementary Table 1. DNA FISH on metaphase spreads for Igh translocations. DNA FISH on metaphases spreads was performed as previously described 58 using BAC probes RP24-134G24 (5' Igh C) and RP24-386J17 (3' Igh V) and XCyting Mouse Chromosome 12 (Orange) paint from MetaSystems. Metaphases were imaged using a ZEISS AxioImager.Z2 microscope and the Metafer automated capture system (MetaSystems), and counted manually.
Expression of recombinant proteins and purification. All recombinant proteins used in biochemical reconstitution of complexes in this study were of human origin. MAD2L2 WT , MAD2L2 1xMut , MAD2L2 3xMut and SHLD3 28-83 were expressed with an N-terminal GST/hexahistidine-MBP fusion-tag from pLIB at 16°C in E.coli Rosetta strain for 16 h after induction with 0.5 mM IPTG. Cells were lysed by sonication in buffer A containing 25 mM HEPES (pH 7.5), 0.3 M NaCl, 5% glycerol, 0.5 mM TCEP (VWR life sciences) and 1 mM PMSF (Roche). After clearing, the lysate was loaded on a Hi-Trap metal chelating column (GE Healthcare). Bound proteins were eluted with an imidazole gradient. The fusiontag was cleaved from MAD2L2 WT , MAD2L2 1xMut and MAD2L2 3xMut using Pre-Scission protease. Protein containing fractions were pooled, and concentrated in 10 kDa MWCO concentrator (Merck) and loaded onto a Superdex-75 column (GE Healthcare) in tandem with HiTrap Talon Crude 1 ml column (GE Healthcare) equilibrated with buffer B (10 mM HEPES pH 7.5, 0.15 M NaCl, 5% glycerol and 0.5 mM TCEP) for size-exclusion chromatography (SEC). Fractions containing purified MAD2L2 and SHLD3 28-83 were concentrated, flash-frozen and stored at −80°C until use. TRIP13 E253Q , MAD2L2 1xMut -SHLD3 and MAD2L2 WT/1xMut -SHLD3-SHLD2 1-95 complexes were purified as His 6 -EGFP, His-MBP fusion and GST-fusion constructs respectively from insect cell using biGBac Expression system 59 . Bacmid was then produced from DH10-beta cells and subsequently used to transfect Sf9 cells and produce baculovirus. Baculovirus was amplified through three rounds of amplification and used to infect Hi5 cells. Cells infected with the viruses were cultured for 72 h before harvesting. Purification of protein complexes was carried out using above protocol while buffer C (25 mM Tris (pH 8), 0.3 M NaCl, 10% glycerol, 0.5 mM TCEP, 10 mM MgCl 2 and 1 mM PMSF) was used for His 6 -EGFP-TRIP13 E253Q . Further polishing of MAD2L2 containing complexes prior SEC and after tag cleavage with PreScission protease was carried out by loading onto a cation-exchange (CE) Resource-S column (GE Healthcare) equilibrated in 10 mM HEPES (pH 7.5), 50 mM NaCl, 5% glycerol and 0.5 mM TCEP. Elution was carried out using NaCl gradient. Fractions containing purified His 6 -EGFP-TRIP13 E253Q , MAD2L2 1xMut -SHLD3 and MAD2L2 WT/1xMut -SHLD3-SHLD2 1-95 were concentrated, flash-frozen and stored at −80°C until use.
Size-exclusion chromatography. SEC runs were performed on Superdex-200 column (GE Healthcare) equilibrated with buffer C + 0.1 mM ATP. Prior run proteins were incubated for 2 h on ice in buffer C + 2 mM ATP. For GST induced REV7 dimerization profiles, SEC runs were performed on Superdex-200 column equilibrated with buffer B. Fractions collected from SEC runs were analyzed by 12% SDS-PAGE.