SMC5/6 acts jointly with Fanconi anemia factors to support DNA repair and genome stability

Abstract SMC5/6 function in genome integrity remains elusive. Here, we show that SMC5 dysfunction in avian DT40 B cells causes mitotic delay and hypersensitivity toward DNA intra‐ and inter‐strand crosslinkers (ICLs), with smc5 mutants being epistatic to FANCC and FANCM mutations affecting the Fanconi anemia (FA) pathway. Mutations in the checkpoint clamp loader RAD17 and the DNA helicase DDX11, acting in an FA‐like pathway, do not aggravate the damage sensitivity caused by SMC5 dysfunction in DT40 cells. SMC5/6 knockdown in HeLa cells causes MMC sensitivity, increases nuclear bridges, micronuclei, and mitotic catastrophes in a manner similar and non‐additive to FANCD2 knockdown. In both DT40 and HeLa systems, SMC5/6 deficiency does not affect FANCD2 ubiquitylation and, unlike FANCD2 depletion, RAD51 focus formation. SMC5/6 components further physically interact with FANCD2‐I in human cells. Altogether, our data suggest that SMC5/6 functions jointly with the FA pathway to support genome integrity and DNA repair and may be implicated in FA or FA‐related human disorders.


Introduction
Genomic integrity is safeguarded by multiple genome caretakers that are often implicated in chromosome metabolism reactions induced by various types of replication stress. The structural maintenance of chromosomes (SMC) complexes, including cohesin (SMC1/3), condensin (SMC2/4), and SMC5/6, are critical for chromosome transactions that ensure normal proliferation and genome integrity. Structurally, SMC complexes form molecular rings that can entrap genomic DNA [1]. The SMC5/6 complex ring is composed of two coiled-coil SMC heterodimers SMC5 and SMC6, which associate with a kleisin component, NSMCE4 [2]. In addition, SMC5/6 contains several peripheral subunits. The NSMCE1-NSMCE3 heterodimer has ubiquitin ligase activity and interacts with NSMCE4 and SMC6, while NSMCE2 has SUMO ligase activity and interacts with SMC5 [3]. The NSMCE1-NSMCE3-NSMCE4 sub-complex presents doublestranded (ds) DNA-binding activity, without preference for structured DNA [4]. SMC5/6 also interacts with SLF1 and SLF2, which are functional orthologs of budding and fission yeast Nse5 and Nse6, respectively [5]. The budding yeast Nse5/6 heterodimer is a component of the Smc5/6 complex (essential for proliferation in budding yeast, but not in fission yeast) and has single-stranded (ss) DNA-binding activity [6]. In human cells, SLF1 and SLF2 physically link RAD18 to the SMC5/6 complex, defining a pathway for SMC5/6 recruitment to sites of DNA damage [5].
In terms of molecular functions, SMC5/6 contributes to DNA repair [7], facilitates DNA topological transitions [8], and promotes timely resolution of DNA cruciform structures arising during homologous recombination repair [9,10]. In unperturbed conditions, budding yeast Smc5/6 is required for the organization and segregation of repeat elements [1,11,12] and promotes replication through difficult to replicate regions known as natural pausing sites [13].
Understanding the functions of SMC5/6 in genome integrity and DNA repair is important, because mutations in SMC5/6 have been linked to different human disease conditions that may derive from impaired DNA metabolism reactions. Specifically, mutations in SMC5/6 components increase overall breast cancer risk [14], promote brain metastasis development [15], and result in debilitating diseases associated with severe developmental defects. To date, two genetic disorders caused by mutations in SMC5/6 have been reported: the NSMCE2-associated syndrome featuring primordial dwarfism and deregulation of glucose metabolism [16], and the NSMCE3-associated disorder also known as LICS, characterized by increased chromosome breakage and defective T-and B-cell function [17]. At the cellular level, the two syndromes are characterized by increased replication stress, micronuclei formation, and defects in homologous recombination (HR). While the involvement of SMC5/6 in preventing replication-stress accumulation is consistent with the results obtained in other model systems and likely linked to its function in DNA repair, the molecular roles of SMC5/6 in DNA repair are incompletely understood.
Here, we set out to investigate the roles of SMC5/6 that assist genome integrity in vertebrate and mammalian cells, by studying the consequences of SMC5/6 depletion in avian DT40 cells and HeLa cells. Both knockout and conditional depletion of SMC5 in DT40 cells allowed cellular proliferation, as previously reported [18]. smc5 mutant cells showed sensitivity toward DNA intra-and inter-strand crosslinkers (ICLs). Notably, the repair defect of smc5 DT40 cells toward cisplatin was genetically epistatic with mutations in the Fanconi anemia (FA) components, FANCC and FANCM, and with mutations in the FA-like pathway defined by the DNA damage checkpoint clamp loader, RAD17, and the DDX11 helicase [19,20]. Moreover, smc5 mutants were additive to mutations in KU70, mediating non-homologous end-joining (NHEJ) repair of double-strand breaks (DSBs). SMC5/6 knockdown in human cells also caused ICL sensitivity, micronuclei, and aberrant mitoses, in a manner similar and non-additive with FANCD2 knockdown [21]. However, SMC5/6 dysfunction did not interfere with FANCD2 ubiquitylation, and unlike FANCD2 depletion, it did not affect RAD51 focus formation in unperturbed or replication-stress conditions. Combining coimmunoprecipitation and mass-spectrometry approaches, we detected physical interaction between several SMC5/6 components and FANCD2-I in human cells, both when SMC5/6 was overexpressed and with endogenous proteins. The results support the notion that vertebrate SMC5/6 functions jointly with the FA pathway in the repair of DNA lesions to prevent genome instability. We propose that mutations in SMC5/6 in humans may be implicated in FA or FA-like disorders.

SMC5/6 function is required for normal proliferation in DT40 cells
Here, we applied the auxin-inducible degron (AID) system, which enables rapid degradation of target proteins by the proteasome [22,23] to establish conditional depletion of SMC5. We used the DT40 avian B-cell line that stably expresses TIR1, an essential component in the auxin degron system [24] in which we C-terminally tagged the endogenous SMC5 gene with the 3AID-6FLAG tag, using the Flip-In system for insertion of epitope tags [25] ( Fig EV1A). As SMC5 is present on chromosome Z, with only one replacement we generated SMC5 /3AID6FLAG cells expressing TIR1 (hereafter referred to as smc5-aid) cells. The functionality of the AID tag inserted to the carboxyl-terminus of SMC5 was confirmed by Western blotting (Fig 1A). After auxin addition, the SMC5-3AID-6FLAG protein was strongly reduced within 24 h (Fig 1A). In regard to proliferation, smc5-aid cells behaved similarly with the wild-type (WT) control in the absence of auxin, but addition of auxin caused slower proliferation (Fig 1B).
To rule out that remaining low levels of SMC5-AID are sufficient for proliferation, albeit with lower speed, we established constitutive knockout of SMC5, employing a construct that causes complete disruption of the open-reading frame ( Fig EV1B). SMC5 knockout cells (hereafter referred to as smc5) proliferated slower than WT cells and had a longer doubling time (Fig 1C), showing results similar to the ones observed upon SMC5-AID depletion.
We further investigated whether the observed slower proliferation phenotype of smc5 cells is complemented by expressing WT copies of chicken SMC5 (cSMC5) tagged with HA. To estimate the level of introduced cSMC5-HA vs. endogenous SMC5 present in WT, we also established an smc5-HA cell line using the Flip-In system and used both Western blotting against HA and quantification of mRNA levels ( Fig EV1C). The results revealed that cSMC5-HA introduced for complementation was expressed 3-4 times higher than endogenous SMC5 and fully complemented the growth defect of smc5 cells ( Fig EV1C). Next, we assessed the cell cycle distribution profile, and found a statistically significant increase in the G2/M population in smc5 mutants (about 25% in smc5 compared with 18% in WT), which was associated with a concomitant decrease in the S-phase population ( Fig 1D). We considered the scenario in which smc5 mutants endure replication stress and as a result accumulate lesions that cause G2/M arrest and/or problems in completing mitosis. Supportive of this notion, addition of caffeine, which inhibits damage checkpoint kinases, reduced the population of smc5 cells arrested in G2/M ( Fig EV1D). Thus, SMC5 is dispensable for viability in avian DT40 cells [18], but is required for physiological levels of fast proliferation. A WB analysis of smc5-aid clones in the absence or presence of auxin using a-FLAG. Tubulin is used as loading control and WT cells are used as a negative control for the FLAG tag. B Proliferation curve of WT and smc5-aid cells in the presence or absence of auxin. The data represent means AE SD of three independent experiments. C Proliferation curves of WT and two independent smc5 clones grown at 39.5°C, with the estimated doubling time on the right. Data represent means AE SD of four experiments. Asterisks indicate P value ≤ 0.05, as derived from unpaired t-test type 3, two-sample unequal variance, carried out to check statistical significance between smc5 clones and WT. smc5 clone 1 vs. WT, P = 0.016, smc5 clone 2 vs. WT, P = 0.04. D Bidimensional FACS analysis shows significant accumulation in G2/M and decrease in the S phase in smc5 mutants. The data represent means AE SD of three independent experiments for smc5 clones 2 and 3, and 4 independent experiments for WT and smc5 clone 1. Asterisks indicate P value ≤ 0.05, detailed below, using paired t-test. S phase (smc5 clones 1, 2, 3 vs. WT: P = 0.0031; P = 0.0002; P = 0.004, respectively). G2/M (smc5 clones 1, 2, 3 vs. WT: P = 0.029; P = 0.0014; P = 0.02, respectively).
Source data are available online for this figure.   [9,10,18,[26][27][28][29]. In chicken DT40 cells, smc5 mutants are sensitive to IR (ionizing radiation) and MMS (methyl methanesulfonate) [18]. In regard to IR sensitivity, smc5 cells are epistatic to rad54, defective in HR, but additive to ku70, defective in the non-homologous end-joining (NHEJ) repair of double-strand breaks (DSBs) [18]. However, more information on the roles of SMC5/6 in DNA repair in vertebrate cells is currently missing, especially in regard to lesions other than DSBs. We investigated the sensitivity of DT40 smc5 mutants to several DNA damaging agents that cause replication-associated lesions, using Celltiter-Glo assay that measures the amount of ATP levels in living cells. We uncovered hypersensitivity of smc5 cells toward cisplatin, a DNA damaging agent causing ICLs (Fig 2A). In addition, we observed intermediate levels of sensitivity of smc5 cells toward the PARP inhibitor olaparib (Fig EV2A), which creates substrates relying on HR-mediated functions for fork protection and/or fork restart [30,31]. The observed cisplatin sensitivity of smc5 knockout cells was fully recovered by the expression of the cSMC5-HA  Fig EV2B). Similarly to cisplatin, endogenous formaldehyde causes ICLs [32]. Thus, we next examined the sensitivity of smc5 cells to this chemical. We found smc5 mutants to be sensitive to formaldehyde (Fig EV2C), although to a lesser extent than fancc mutants defective in the FA pathway.
Previous reports found smc5 mutants to be additive to mutations in ku70, defective in NHEJ, in regard to repair of lesions induced by IR [18]. We found additivity between smc5 and ku70 knockout mutations also in regard to cisplatin (Fig 2C), while eliminating NHEJ suppresses the sensitivity of FA mutants [33,34]. Thus, SMC5 facilitates damage tolerance of bulky lesions and ICLs, and may be an important player, along with the FA pathway in mediating ICL repair.
Fanconi anemia is a multigenic syndrome characterized by bone marrow failure, developmental abnormalities, and predisposition to cancer, with FA patient cells showing a characteristic cellular sensitivity to agents that induce ICLs. So far, 22 complementation groups have been delineated in FA, with the causal genes shown to act in a common repair pathway [35][36][37]. In addition, several FA components have also FA-independent roles in DNA replication, fork protection, and checkpoint activation [38][39][40][41].
The central components of the FA pathway, FANCD2 and FANCI, interact with each other [42], and are monoubiquitylated by the FANCL ubiquitin ligase component of the FA core complex, comprising several FA subunits (FA-A, FA-B, FA-C, FA-E, FA-F, FA-G, FA-L, FA-M, FA-T) and associated proteins FAAP20 and FAAP100. FANCD2-FANCI ubiquitylation promotes lesion unhooking, causing the formation of a gapped DNA molecule, with the unhooked lesion in the gapped part, and a double-strand break (DSB) on the other. The gap and DSB are subsequently repaired by TLS and HR [43]. We investigated potential genetic relationships between SMC5 and FANCC, which is essential for FANCD2-I ubiquitylation and required for both HR and TLS-mediated pathways of DNA repair in DT40 cells [44]. Both fancc and smc5 mutants show mild proliferation delays with the double-mutant fancc smc5 showing additive growth defects ( Fig EV2D). Double-mutant smc5 fancc did not show additive sensitivity (Fig 2D), suggesting that SMC5 and FANCC do not work in redundant pathways. The apparent improved survival of the double mutant when exposed to cisplatin might stem from the poorer proliferation of these cells (Fig EV2D), which may cause them not to be effectively damaged by cisplatin that functions during replication.
Because FANCC is required for FANCD2 ubiquitylation, which also relies on an intact ATR checkpoint response [42,45], we next ◀ Figure 2. SMC5 contributes to the repair of intra-and inter-strand crosslinks, independently of KU70 and jointly with FANCC.   analyzed whether these signaling pathways were functional in smc5 mutants. We found that both CHK1 phosphorylation, used as readout for ATR activity, and FANCD2 ubiquitylation, used as readout of FA pathway functionality, proceeded unaffected in smc5 mutants ( Fig 2E and F). Thus, SMC5 functions downstream of FANCD2 ubiquitylation in a manner non-redundant with FANCC, potentially to support the DNA repair function of the FA pathway.
SMC5 acts jointly with FANCM and the FA-like pathway involving the RAD17 checkpoint clamp loader and the DDX11 DNA helicase FANCM is a component of the FA core presenting DNA translocase and branch-migrating activities and a degenerate C-terminal ERCC4 nuclease domain that binds to DNA structures in vitro [46]. FANCM has functions within the FA pathway related to the recruitment of the FA core complex to chromatin and in triggering checkpoint activation, as well as functions outside the FA pathway related to lesion bypass potentially by fork reversal (reviewed in Ref. [47]). We established double mutants between fancm and smc5 knockouts and found these cells to be proliferating similarly with the smc5 single mutants (Fig 3A). In regard to cisplatin sensitivity measured again by the Celltiter-Glo assay, smc5 mutants are more sensitive than fancm, while smc5 fancm mutants showed reduced cisplatin sensitivity compared with smc5 ( Fig 3A), a situation recapitulating the phenotype observed with fancc ( Fig 2D), but in a situation in which smc5 and smc5 fancm mutants proliferated equally well (although worse than WT). Moreover, long-term clonogenic assays revealed a similar epistatic relationship between fancm and smc5 mutations in regard to cisplatin sensitivity (Fig 3B).
A backup pathway for the FA pathway involves the conserved DDX11 helicase and the checkpoint clamp loader RAD17, which act jointly to facilitate recombination-mediated repair of bulky lesions in vertebrate cells [19,20]. In human cells, RAD17 also affects FANCD2 ubiquitylation [48]. To study the relationship between this FA-related pathway and SMC5 function, we established double mutants between smc5 (and smc5-aid) and rad17, ddx11 mutations. The double mutants proliferated similarly with smc5 single mutants, and both rad17 and ddx11 showed epistasis to smc5 mutants toward cisplatin (Fig 3C and D). ddx11 showed similar non-additive relationship with depletion of SMC5-AID in regard to proliferation and repair (Fig EV3A and B). Thus, SMC5 functions jointly with FANCM and FA-related pathway involving DDX11 and RAD17 in DNA repair.

SMC5/6 acts jointly with FANCD2 to mediate DNA repair and prevent genomic instability in human cells
To validate and extend our previous observations made in the chicken DT40 cell system, we analyzed the consequence of SMC5/6 depletion in HeLa cells. siSMC5 depletion and siSMC6 depletion were very efficient by 48 h post-transfection and validated by using different siRNA sequences to avoid off-target consequences of the siRNA approach. SMC5 or SMC6 depletion strongly affected the stability of the other partner ( Fig 4A). Notably, their depletion was associated with an increased level of FANCD2, and conversely, FANCD2 depletion was associated with increased SMC5/SMC6 levels ( Fig 4A).
Cell survival in response to mitomycin C (MMC) exposure, measured using crystal violet staining, revealed that SMC6-depleted HeLa cells are slightly more sensitive than control cells but significantly less sensitive than FANCD2-or FANCJ-depleted cells. The simultaneous depletion of SMC6 and FANCJ had minimal effect on the MMC sensitivity of FANCJ-depleted cells, whereas the downregulation of SMC6 in FANCD2-depleted cells increased their resistance to MMC (Fig 4B). These results were obtained using smart siRNA pools, but qualitatively similar results were obtained using individual different siRNAs for SMC6 and a single siRNA for FANCD2 and FANCJ ( Fig EV4). Thus, also in human cells, SMC6 acts in a DNA repair pathway non-redundant with the one mediated by the FA proteins FANCJ and FANCD2. In each individual experiment, we first normalized SMC5, SMC6, or FANCD2 expression to that of Vinculin (internal control) and calculated its relative level in each cell line in relation to the target protein/Vinculin ratio of the siLacZ-transfected cells, which was set as one in each experiment. Data present the mean AE SEM and were analyzed by unpaired t-test. SMC5 in siSMC6: n = 7 and P = 0.0002; SMC5 in siFANCD2: n = 7 and P = 0.0016; SMC6 in siSMC5: n = 8 and P < 0.0001; SMC6 in siFANCD2: n = 24 and P < 0.0001; FANCD2 in siSMC6: n = 30 and P < 0.0001; and FANCD2 in siSMC5: n = 10 and P = 0.0008.   SMC6 depletion using different siRNAs did not affect FANCD2 monoubiquitylation in either untreated or replication-stress conditions involving hydroxyurea (HU), MMC, or aphidicolin (APH) treatment (Figs 4A and 5A, and EV5) nor FANCD2 foci assembling in S/ G2 phases of the cell cycle (Fig EV5), supporting the notion that SMC5/6 acts independently or downstream of FANCD2 activation.
The FANCD2 monoubiquitylation is important for optimal RAD51 focus formation [49][50][51], and DDX11 deletion in DT40 that does not affect FANCD2 monoubiquitylation also reduces RAD51 focus formation [19]. We assessed the consequences of SMC6 depletion on the formation of RAD51 foci in MMC-or HU-treated HeLa cells. Differently from FANCD2 depletion, siSMC6 did not alter RAD51 foci per se and did not modify the frequency of RAD51 focipositive cells observed in the absence of FANCD2 (Fig 5B). Thus, our data suggest that SMC5/6 acts independently or downstream of RAD51 focus assembly, and downstream of FANCD2 and DDX11 that facilitate efficient RAD51 focus assembly.
It was proposed that FANCD2 foci assembled during S phase at stalled/delayed replication forks persist into G2/M and up to the end of mitosis to rescue under-replicated or untangled replicated regions, with this action serving to avoid anaphase bridges and genomic instability in daughter cells [21]. Indeed, FANCD2 loss of function is associated with an increased frequency of mitotic catastrophes and nuclear bridges at anaphase-telophase as well as with increased frequency of post-mitotic cells presenting micronuclei [21]. Notably, SMC6-depleted cells revealed a similar level of mitotic abnormalities with FANCD2-depleted cells, and the frequency of these abnormalities remains unchanged in SMC6 and FANCD2 double-depleted cells (Fig 6A and B). The results thus suggest joint action between SMC6 and FANCD2 in preventing genomic instability in mitotic and post-mitotic cells.

SMC5/6 physically interacts with FANCD2-I in human cells
To examine potentially informative physical interactions, we transiently expressed FLAG-tagged SMC5, SMC6, and NSMCE2 in HEK293 cells and immunoprecipitated the complexes with an anti-FLAG antibody. Mass spectrometry identified the other SMC5/6 components in the immunoprecipitate of SMC5, along with a few other proteins of interest, among which FANCI (Table 1). Immunoblotting further validated that FANCI was present in the immunoprecipitates of SMC5/6 and NSMCE2 (Fig 7A), even in the presence of ethidium bromide (EtBr), which binds DNA and disrupts DNA-protein interactions.
Because in the above experiments SMC5/6 and NSMCE2 were overexpressed, we examined whether the interaction was observed among endogenous proteins. As we experienced low IP efficiency with commercial SMC5 antibodies, we tagged endogenous SMC5 alleles C-terminally with the mAID-EGFP tag in human TK6 cells using CRISPR-Cas9, using the strategy reported in Ref. [52]. Using SMC5-mAID-EGFP TK6 cells, we found that SMC5-EGFP pull-down led to co-immunoprecipitation of FANCD2, and vice versa, independently of induced DNA damage (Fig 7B and C). These results suggest that SMC5/6 physically interacts with FANCD2-I in vivo.

Discussion
Studies of the Smc5/6 complex in budding and fission yeast are consistent with the involvement of this complex in dynamic DNA processes, such as stabilization of stalled replication forks and regulation of recombination intermediate resolution [9][10][11]13]. The essential functions of the SMC5/6 complex, at least in budding yeast, primarily manifest after the bulk of replication is complete, and involve processes mediating replication termination and DNA synthesis through difficult to replicate regions [11,13], such as ribosomal DNA, rDNA [12,53]. The postreplicative or late replication nature of the essential processes requiring SMC5/6 suggests that this complex may be especially important during late replication, in a subset of DNA metabolism reactions with repercussions on the mitotic chromosome structure and its compaction [54]. In budding yeast, those processes may be particularly relevant to rDNA condensation and segregation [12], whereas in vertebrates they may encompass several classes of fragile sites or structures that remain ◀ Figure 6. SMC6 acts jointly with FANCD2 to prevent micronuclei and mitotic abnormalities.   to be identified. Nevertheless, the partners of SMC5/6 in these repair processes remain poorly understood, especially in vertebrate cells.
In this study, we undertook a dual approach to better understand the functions of vertebrate/mammalian SMC5/6 in DNA repair and genome stability. One approach aimed at identifying new interacting partners of SMC5/6 in mammalian cells, while the other addressed the molecular functions of SMC5/6 in DNA repair and genome integrity by examining several mutant combinations in DT40 and upon knockdown in HeLa cells. Remarkably, both our approaches intersected on the FA pathway and particularly FANCD2-I, already reported to play roles in facilitating replication through common fragile sites (CFSs) [21,[55][56][57] and for preventing fragile site expression in the presence or absence of replication stress [45,[58][59][60].
Here, we find that SMC5/6 components interact physically with FANCD2-I and function jointly with FA proteins (FANCC, FANCM, FANCJ, FANCD2) and FA-related pathways (RAD17, DDX11) in the repair of ICLs created by cisplatin and mitomycin C. Considering that SMC5/6 depletion does not impair FANCD2 ubiquitylation and either FANCD2 or RAD51 focus formation, these results indicate roles of SMC5/6 in the late steps of HR, perhaps by resolving emerging recombination intermediates (Fig 7D). This could explain the effects of SMC5/6 depletion on anaphase bridge and micronuclei formation, which resemble in extent and are non-additive with the ones of FANCD2 depletion. FANCD2 counteracts NHEJ, and inhibition of NHEJ in FA rescues survival and genetic instability in FA [33,34]. Our new results suggest that this rescue happens because cells become able to funnel lesions in the HR pathway dependent on DDX11 and SMC5/6 ( Fig 7D). Because ICL repair resembles replication termination [61] and termination regions are predisposed to fragility [62,63] and protected by budding yeast Smc5/6 [13], we propose joint roles of SMC5/6 and FANCD2-I at these regions and possibly at other subsets of CFSs.
Our findings can be accommodated by a model envisaging two functional interaction points between the FA pathway and SMC5/6: one in which SMC5/6 function/recruitment to the DNA lesion or terminal sites of replication is facilitated by FA components such as FANCM, and a subsequent one in which SMC5/6 facilitates the repair function of the FA pathway (Fig 7D). The role of SMC5/6 in facilitating DNA repair may be manifested via recruitment and/or Sumoylation of BLM, already reported to act jointly with FANCD2 and SMC5/6/NSMCE2 in certain conditions of replication stress [21,64]. Alternatively, SMC5/6 may downregulate the fork transversal function of FANCM and/or activate the degradation/mobilization of other replication, repair, or structural factors to facilitate lesion bypass and subsequently replication completion. This latter envisaged function may involve the ubiquitin and SUMO ligase activity of the SMC5/6 complex and potentially facilitate replisome disassembly during replication termination [65] or cohesin removal to facilitate repair completion and normal chromosome structure in prophase [66]. We thus envisage that there remain undiscovered genetic disorders caused by mutations in genes of the SMC5/6 complex that lead to FA or FA-like disorders and in which the physiological role SMC5/6 in mediating DNA repair and replication termination becomes more apparent.

Cell lines and genotoxic treatments
All DT40 cell lines used in the study were derived from the DT40 WT clone 18 [67] and reported in Table 2.

DNA restriction enzymes and DNA ligase
All restriction enzymes and T4 DNA ligase were purchased from New England Biolabs (Frankfurt, Germany) and used according to the manufacturer's specification. GeneArt â Seamless Cloning and Assembly enzyme mix (Cat. No. A14606, Thermo Fisher Scientific) and GeneArt â Seamless Plus Cloning and Assembly enzyme mix (Cat. No. A14603, Thermo Fisher Scientific) were used to clone the homologous recombination arms for SMC5 C-terminus tagging in DT40 cells.

Plasmids
The vectors used for DT40 cloning were previously reported [68,69]. Vectors used for SMC5 tagging with HA and AID were ◀ Figure 7. SMC5/6 physically interacts with FANCD2-I in human cells.

A
Immunoprecipitation of FLAG-SMC6 and FLAG-NSMCE2 with FANCI in HEK293T cells overexpressing the SMC6 or NSMCE2, respectively, in the presence or absence of ethidium bromide (EtBr). The experiment was performed three times, and the blot shows a representative experiment. The band under the one indicated as FANCI may represent a degradation product. B, C Co-immunoprecipitation of endogenous SMC5-mAID-EGFP and FANCD2 in TK6 cells in the presence of benzonase. Both experiments were performed twice independently, with the note that the cisplatin condition was performed only in one of the experiments, and the FANCD2 pull-down in the experiment not shown was performed without benzonase. D Model schematizing the proposed view of SMC5/6 action in genome integrity and DNA repair. SMC5/6 acts differently from non-homologous end-joining and downstream of RAD51 filament formation and of homologous recombination branches mediated by DDX11, RAD17, and FANCD2, RAD17.
Source data are available online for this figure.
ª 2019 The Authors EMBO reports 21: e48222 | 2020 modified in the laboratory starting from a backbone received from the Hirota Lab [25]. The Px458 CRISPR/Cas9 vector reported in Ref. [70] was used in this study to establish fancc, and fancm knockouts in WT and smc5 backgrounds. Guide DNA targeting for fancc and fancm was purchased from Sigma-Aldrich and cloned in the px458 vector (https://www.addgene.org/48138/), generating vectors FR18# (for fancc) and FR20# (for fancm), respectively. FR18# was transiently co-transfected with vector #122 [71] in smc5 and WT, respectively, in order to induce double-strand breaks to enhance HR and the insertion of the resistance marker contained in the vector in order to KO fancc. FR20# was transiently co-transfected with vectors #12-31 and #12-29 in smc5 or WT in order to induce double-strand breaks to enhance HR and the insertion of the resistance marker contained in vector in order to KO fancm.
SMC5-3xmAID-6xFLAG Flip-In was generated from genomic PCR products combined with p3xmAID-6xFLAG containing histidinol D selection marker cassette. Genomic DNA sequences were amplified using primers 5 0 -ATAAAAGTCGACGGAGAGCTTAATTTTGTGTGA C-3 0 (containing SalI restriction enzyme site) and 5 0 -ACCATTGCTA GCCTGTTCATCCATTCTTCC-3 0 (containing NheI restriction enzyme site). Amplified PCR product was purified by gel extraction and digested by SalI high-fidelity and NheI high-fidelity restriction enzymes, then purified again, and ligated in p9xMyc vector [72]. Then, they were cloned with the same strategy in 3xmAID-6xFLAG vector [24]. The Flip-In vector was then linearized at one restriction enzyme site in the middle of the homology region and transfected to DT40 cells as previously described [25].
SMC5-HA Flip-In was generated from genomic PCR products combined with p3xmAID-6xFLAG containing Ecogpt selection marker cassette. Genomic DNA sequences were amplified using primers 5 0 -ATAAAAGTCGACGGAGAGCTTAATTTTGTGTGAC-3 0 (containing SalI restriction enzyme site) and 5 0 -ATGCGGCGCGCCTCAAGCGTAATCT GGAACGTCATATGGATACTGTTCATCCATTCTTCCAAGTCGC-3 0 (containing sequence coding HA-tag and AscI restriction enzyme site). Amplified PCR product was digested by SalI and AscI restriction enzymes and purified by gel extraction, and then ligated into p3xmAID-6xFLAG vector. The Flip-In vector was then linearized by SnaBI restriction enzyme before transfection.
SMC5 knockout construct was generated by modifying the SMC5 KO-Neo vector published in Ref. [18], which was digested with BamHI in order to cut exon 1 from the 5 0 arm, purified by gel extraction, and ligated with loxp (histidinol D) selection marker cassette. Then, the vector was digested with PvuI and BamHI restriction enzymes, purified by gel extraction, and ligated with loxP (puromycin) selection marker cassette. SMC5-KO-Puro vector was then linearized with NotI before being transfected to DT40 cells. rad17 and ddx11 mutations were established as previously reported [19].
TK6 cells expressing TIR1 were electroporated using Neon â Transfection System MPK5000 (voltage/width/number of pulses: 1,350 V/10 ms/3), with medium containing no PBS. Cells were selected 24 h after electroporation, using the drugs corresponding to the resistance marker(s) used, and then, the cells were plated in 96 wells to get single-cell clonal population. Cells were grown in medium until a single colony/well was visible, then picked up with a filter tip, and moved to a 24-well dish. Then, genomic DNA was extracted, the genotype was established by PCR, and ultimately confirmed by WB.

Cell culture and transfection procedures
All DT40 cells were derived from DT40 WT clone 18 [67]. For drug selection during knockout/knock-in generation, cells were selected 16-24 h after electroporation, using the drug corresponding to the employed resistance marker, and then the cells were plated in 96 wells to get single-cell clonal population. Cells were grown in medium until a single colony per well was visible, then it was picked up with a filter tip, and moved to a 24well dish. Then, genomic DNA was extracted and the genotype was checked by PCR, and eventually, mRNA expression was checked with reverse transcriptase-PCR (RT-PCR) for the indicated clones. For DT40 cells, electroporation was carried out using Gene Pulser II â Electroporation System (Cat. No. 165-2105, Bio-Rad Laboratories, Hercules, CA, USA). Methods for DNA transfection and selection of clones were performed as described previously [67].
HeLa cells were plated at 2 × 10 5 cells/well in a 6-well plate and then transfected 24 h later with a mix of 190 ll of Opti-MEM â (Invitrogen TM ), 8.5 ll of INTERFERin â (Polyplus), and 2.2 ll of siRNA (20 nM final) incubated 10 min at RT before use. siLacZ transfection was used as a control.

Growth curve
For proliferation curves, cells were plated with a starting concentration of 10 5 cells/ml in triplicate. They were counted at the indicated time points with the Burker chamber. Before counting, they were stained with erythrosine (1 volume of medium containing cells and 1 volume of erythrosine) to mark viable cells. Raw data were plotted in Excel to derive growth curves and to calculate the doubling time of individual cell lines.

Auxin treatment
When indicated, auxin was added to the medium to a final concentration of 500 lM. Cells were treated with auxin for the indicated time, dependent on the assay. In order to check the depletion or the recovery of the expression of the target protein, 1 × 10 6 cells were pelleted, then washed 1× with PBS 1×, and lysed with sample buffer 1×. Since auxin is reported to be degraded in the media, during proliferation experiments, auxin concentration was maintained by diluting the culture with fresh medium containing auxin. In Celltiter-Glo ATP sensitivity assays, cells were supplemented with auxin after 24 h of treatment. Quantitative PCR for mRNA Total RNA was isolated from DT40 cells by using TRIzol RNA Isolation Reagents (Invitrogen) according to the manufacture's protocol. cDNA library was prepared from total RNA by using SuperScript TM III Reverse Transcriptase (Invitrogen) according to the manufacture's protocol. Quantitative PCR was performed using GoTaq â qPCR Master Mix (Promega) and LightCycler â 96 Instrument (Roche) according to the manufacture's protocols. ACTB was used as house-keeping gene. Used primers are described below. ACTB fw: 5 0 -CGTGCTGTGTTCCCATCTATCGTG-3 0 ACTB rv: 5 0 -TACCTCTTTTGCTCTGGGCTTCATC-3 0 SMC5 fw: 5 0 -GCAGGGCTGCTTGAAAGGTCTAGTG-3 0 SMC5 rv: 5 0 -GCTTCCTGTTGATATGCCAGGTTGAC-3 0 Drug sensitivity assays (Celltiter-Glo, colony survival, and crystal violet assays) Sensitivity assays in DT40 were carried out treating 2 × 10 4 cells/ml with the indicated drug concentration for 48 h. To assess cellular sensitivity, we measured the amount of ATP levels using Celltiter-Glo â assay (Cat. No. G7571, Promega). Luminescence levels were detected using Victor3 TM (PerkinElmer). Control cells were treated with the vehicle diluted at the same concentration of the drug used in the assay. We have carried out each biological experiment in technical triplicates, and in certain occasions, eliminated outliers. We defined as outlier the measurement having a ratio bigger than 25% of each of the other two measurements and with their averages, dividing the smaller data with the bigger data. The other two data should not have a ratio bigger than 25% among themselves. Each experiment was conducted in at least three independent biological experiments.
To determine cisplatin sensitivity in DT40 by colony survival assay, an appropriate number of DT40 cells were inoculated in a medium supplemented with 1.5% (w/v) methylcellulose, 15% fetal bovine serum, and 1.5% chicken serum. Indicated concentration of cisplatin was added and mixed for 6 h before inoculating cells. Colonies were counted after 14 days, and the percent survival was determined relative to the number of colonies of untreated cells. Two independent biological experiments and two independent clones for smc5 fancm were used in the analysis shown in Fig 3B. For cell viability assays in HeLa (Figs 4B and EV4), 2-3 × 10 3 cells/well were seeded in 48-well plates and allowed to adhere for 8-10 h. Chronic drug treatment was given to cells with indicated concentrations for 5 days. The cell viability was measured by staining the cells using 0.5% crystal violet containing 20% methanol, and the graph was plotted by normalizing with untreated cells. Three independent biological experiments were performed.

SDS-PAGE and Western blot
DT40 cells were harvested by trypsin treatment and lysed or fractionated according to the experimental aim. Protein quantification was obtained using the Bradford reagent assay (Cat N., Bio-Rad Laboratories, Hercules, CA, USA), according to the manufacturer's instructions. Samples were boiled in sample buffer (50 mM Tris-HCl pH 6.8, 10% glycerol, 1.6% SDS, 0.1 M DTT, 0.1% bromophenol blue). Whole-cell lysates (WCLs) were separated on 4-12% Blot TM Bis-Tris Plus (NW04125BOX, Thermo Fisher Scientific) or Criterion TGX Stain-Free (#5678094, Bio-Rad). Resolved samples were transferred to Hybond-ECL nitrocellulose membrane (Amersham Bioscience, Glattbrugg, Switzerland), 1 h at 100 V or overnight at 30 V (4°C). Then, the membrane was saturated with 5% non-fat milk Tris-buffered saline (TBS 10 mM Tris-HCl PH 7.4, 100 mM NaCl) supplemented with 0.1% Tween-20 (TBS-T). The membranes were incubated with primary antibodies for 1 h at room temperature or overnight at 4°C. Then, the membranes were washed as mentioned above and incubated with the horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h at room temperature.

FACS analysis
Bidimensional FACS analysis was carried out pulsing the cells with BrdU, final concentration 20 lM for 15 min at 39.5°C (10 mM stock, Cat. No. B9285, Sigma-Aldrich). After pelleting 1 × 10 6 cells, cells were resuspended in 50 ll ice-cold PBS, and then 1 ml of 70% ethanol (ice cold) was added. Cells were washed once in 1 ml PBS 1% BSA (4°C in all the steps if not differently indicated), then resuspended in 1 ml denaturing solution (2N HCl), and incubated at room temperature for 25 min.
Note: If not differently indicated, all the solutions were ice-cold. Cells were spinned down at 300 × g at 4°C, for 5 min.

Immunofluorescence
Cells were seeded at 2 × 10 5 cells/well in a 6-well plate on glass slides. After treatment, if any, cells were washed with PBS and then incubated in pre-extraction buffer CSK100 (100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 , 1 mM EGTA, 0.2% Triton X-100, PIPES pH 6.8 10 mM, anti-protease) for 5 min at room temperature and then fixed for 10 min with PBS + 4% formaldehyde, and finally permeabilized with PBS + 0.1% Triton X-100 for 10 min at RT. The cells were incubated 1 h with PBS + 5% BSA and then 1 h at 37°C with the primary antibodies, washed in PBS, and incubated again for 1 h at 37°C in the dark with the secondary 16 of 19 EMBO reports 21: e48222 | 2020 ª 2019 The Authors antibodies coupled to Alexa 488 fluorophore (green) at 1/1,000. After a final wash with PBS, the slides were mounted with the DAPI-DAKO solution. The labeling was revealed using an epifluorescence microscope (Zeiss Axio Observer Z1). The following antibodies and conditions were used: rabbit anti-FANCD2 (ab180928 Abcam, 1/1,000); rabbit anti-RAD51 (PC.130 Calbiochem, 1/500); and mouse anti-a-tubulin (T5168 Sigma 1/2,000). A minimum of 200 cells for each condition was analyzed to calculate the percentage of RAD51-positive cells. Detection and scoring of micronuclei were performed using an adapted cytokinesis-block assay. Cells were incubated with cytochalasin-B (3 lg/ml; Sigma) for 18 h before fixation. Micronuclei were scored in 40-80 bi-nucleated cells for each condition.
Mitotic catastrophes, anaphase bridges, and post-mitotic cells with nucleocytoplasmic bridges were scored on cells prepared for immunofluorescence analysis looking at nuclei stained with DAPI. A minimum of 10 fields (100-300 cells) were recorded for each condition to score the percentage of abnormal mitotic figures in relation to the totality of the mitoses in the microscopic fields.
Data are available upon request or uploaded as Source Data.
Expanded View for this article is available online.