Introduction

The Structural Maintenance of Chromosomes (SMC) complexes, present in both prokaryotes and eukaryotes, are core modulators of chromosome architecture and organization, supporting the development and homeostasis of all living organisms (1,2). The SMC1/3 (cohesin) and SMC2/4 (condensin) complexes ensure accurate genome propagation by folding chromosomes into distinct structures. Meanwhile, SMC5/6 plays a key role in maintaining genome stability and integrity, contributing to processes such as homologous recombination (HR), alternative lengthening of telomeres (ALT), and ribosomal DNA (rDNA) replication (3–7). Recently, SMC5/6 has also been recognized as a viral restriction factor, highlighting its broader role in managing extrachromosomal DNAs (ecDNAs, (8–10)).

Beyond viral restriction, the repression of transcription from simple plasmid DNA has also been added to the activities of the human SMC5/6 “Swiss Army knife” of DNA manipulation (8–12). SMC5/6 selectively represses the transcription of circular but not linear plasmid DNA (12), which is likely related to their topological states. That is, DNA that accumulates superhelical stress during transcription is the preferred binding substrate for SMC5/6, whereas linear DNA that dissipates such tension is not stably bound by the complex (12–15). Once DNA bound, SMC5/6 may further compact and thereby restrict transcriptional activity on the ecDNA.

All SMC complexes share a fundamental structure consisting of a SMC heterodimer connected by a kleisin protein. These complexes form a ring-like structure that traps DNA. The SMC proteins have a bent-arm shape and can open to encircle DNA when their ATPase head domains bind ATP, aiding DNA movement through the ring. Kleisin plays a critical role for connecting and locking the SMC heads together (6,16–22). The SMC5/6 complex is particularly adept at recognizing and compacting special DNA structures, such as HR intermediates and supercoiled DNA (13,15,23,24).

The SMC5/6 complex is composed of a highly conserved “core” hexamer: the SMC5/6 heterodimer and non-SMC-elements (NSE) 1-4 (9). A Kleisin (Nse4 in yeast; NSMCE4 in human) and two interacting partners (Nse1 and Nse3 in yeast; NSMCE1 and NSMCE3 in human) form a trimer that joins the SMC5/6 head domains. The SUMO ligase Nse2 (NSMCE2 in human) attaches to the SMC5 arm. The complex’s functional specificity is further defined by non-core subunits, the Nse5/6 heterodimer (9,18,25).

Yeast Nse5/6 plays key roles in regulating Smc5/6’s genome stability functions, facilitating its chromatin loading, inhibiting its ATPase activity, and enhancing Nse2 SUMO ligase activity (19,26–32). Despite its critical regulatory roles, the identification of a human Nse5/6-like complex lagged, largely due to the lack of cross-species protein sequence conservation. Nevertheless, identified in proteomic approaches, the human SLF1 (SMC5-SMC6 Complex Localization Factor 1)-SLF2 heterodimer (SLF1/2) was shown to fulfill some genome stability roles performed by Nse5/6 in yeast (7). Interestingly, however, with the possible exception of HIV-1 proviral DNA, the repression of viral DNA transcription by SMC5/6 depends on SLF2 but not SLF1 (8,9,11). Therefore, SLF2 either functions as a monomer in viral repression, or there is a cofactor that replaces SLF1 in this role. Towards resolving this issue, we recently discovered SIMC1 as a missing link in SMC5/6 regulation and demonstrated that the SIMC1-SLF2 complex is a human counterpart of yeast Nse5/6 (33).

Our structural and mutational analyses revealed that SIMC1 and SLF2 contain Nse5 and Nse6-like domains, respectively, which interact through a conserved interface. SIMC1-SLF2 interacts with and directs SMC5/6 to subnuclear compartments at PML nuclear bodies (PML NBs) induced by the SV40 polyomavirus large T antigen (LT). Further, we showed that SIMC1-SLF2 and SLF1/2 are mutually exclusive subcomplexes of SMC5/6 that likely support the distinct antiviral and DNA repair activities of the holocomplex. Thus, SIMC1-SLF2 and SLF1/2 act as Nse5/6-like regulatory elements that modulate the broad activities of the SMC5/6 complex (33).

Despite the foregoing advances, the specific roles of the SLF2-based SIMC1 and SLF1 subcomplexes in SMC5/6 regulation remained unclear. Although we previously proposed distinct functions for SIMC1-SLF2 and SLF1/2 in ecDNA silencing and DNA repair, respectively, these mechanisms had yet to be demonstrated. Additionally, SUMOylation, PML NBs, and SV40 LT were implicated in the regulation of SMC5/6-based ecDNA silencing, but their contributions were not fully understood. Here, we address these gaps directly. In line with our prior model, we show that SIMC1-SLF2, but not SLF1/2, mediates plasmid silencing through a direct and evolutionarily conserved interaction with SMC6. Conversely, SLF1/2 specifically recruits SMC5/6 to DNA lesions. Furthermore, we demonstrate that plasmid silencing is facilitated by the SUMO pathway but, unlike the SMC5/6-mediated silencing of HBV (34), does not require PML NBs. Lastly, we reveal that SV40 LT binds SMC5/6 and antagonizes plasmid silencing, thereby adding polyomaviruses to the growing list of pathogenic viruses that counteract SMC5/6-based transcriptional repression.

Results

SIMC1-SLF2 associates with SMC6 through a conserved patch

Our previous structural analysis revealed that the Nse5-like domain of SIMC1 (SIMC1^Nse5) and the Nse6-like domain of SLF2 (SLF2^Nse6) both adopt ɑ-solenoid-like structures (33). These domains interact in a head-to-tail orientation, forming an ellipsoid-shaped dimer (Figure 1A). This dimerization clusters conserved residues from the N-terminus of SIMC1^Nse5 and the C-terminus of SLF2^Nse6 at one edge of the ellipsoid (Figure 1B). We previously used a SIMC1 mutant harboring four substitutions in ɑ1 of SIMC1 (SIMC1^combo, Figure 1B) to demonstrate that this conserved patch of SIMC1 is essential for binding to SMC6 and for the localization of the SMC5/6 complex at LT-containing PML NBs (33).

Figure 1:

In this study, we extended our analysis to SLF2 by introducing two sets of mutations into the Nse6-like region of SLF2 (SLF2^CTD; residues 635-1173): mut 1 (E1121A/K1122A/K1125D/C1126R/E1130K in α16) and mut 2 (T1138R/K1141E/D1142R/A1145R/G1149E in the subsequent loop and α17). These mutations were designed to disrupt charge interactions, hydrogen bonds, and van der Waals contacts, following the same strategy used with SIMC1^combo to disrupt binding (33). Using HEK293 cells co-expressing FLAG-tagged SLF2^CTD mutants, GFP-tagged SIMC1 and Myc-tagged SMC6, we performed GFP pull-downs followed by Western blot analysis (Figure 1C). Wild-type SLF2^CTD interacted with both SIMC1 and SMC6 (Lane 1), whereas both mut 1 and mut 2 disrupted SMC6 binding while preserving the interaction with SIMC1 (Lanes 2, 3).

We also examined several control mutants. To confirm the specificity of the SIMC1-SMC6 interaction, we introduced a new SIMC1 control mutant containing four substitutions (Q842A/H846A/K849E/D857R) in α17, located at the opposite end of the ellipsoid (Figure 1A). This mutant retained binding to both SLF2 and SMC6, indicating that the residues altered in SIMC1^combo are specifically required for SMC6 interaction (Figure 1—figure supplement 1A). We additionally evaluated the most C-terminal residues of SLF2 (1161–1173), which although conserved, are unstructured; however, deleting these residues (SLF2^635-1160) did not affect SMC6 binding (Figure 1—figure supplement 1B). Collectively, these results confirm that the conserved surface patch mediating the SIMC1-SLF2 interaction is essential for binding to SMC6.

SIMC1-SLF2 subcomplex contacts the neck region of SMC6

We used AlphaFold-Multimer (36) to predict the structure of the SIMC1-SLF2-SMC6 complex (Figure 2A, Figure 2—figure supplement 1A). The resulting high-confidence model, as assessed by the predicted aligned error (PAE) plot (Figure 2—figure supplement 1B), aligns well with mutational data and highlights the conserved surface patch of SIMC1-SLF2 as the interface with SMC6 (Figure 2A). SMC6 is a long, hairpin-like polypeptide in which its N- and C-termini fold together to form the ATPase “head” domain at one end. The “hinge” domain at the bending end mediates heterodimerization with SMC5 (Figure 2—figure supplement 1A). Connecting the head and hinge is a flexible coiled-coil; the portion adjacent to the head is referred to as the “neck” (Figure 2A).

Figure 2:

In the model, the conserved SIMC1-SLF2 surface patch interacts with C-terminal residues of SMC6, centered on the neck and opposite the ATPase active site where it dimerizes with SMC5’s ATPase domain (21,30). This interface shows excellent shape complementarity—SMC6’s convex neck helix fits snugly into the concave surface formed by α1 and α2 of SIMC1 and α17 of SLF2 (Figure 2A). Additional interactions arise from SLF2’s α16 and the α16-α17 loop with SMC6’s head residues, spanning a total buried surface area of 1516 Ų.

This model explains the roles of the amino acids altered in our binding studies (Figure 2B; Figure 2—figure supplement 1C, D). The four residues modified in SIMC1^combo all make direct contact with SMC6’s neck helix, including three salt bridges: R473^SIMC1-D946^SMC6, E480^SIMC1-K957^SMC6/K942^SMC6, and E481^SIMC1-K942^SMC6. The SLF2 mutations also engage SMC6’s neck region, forming salt bridges (E1130^SLF2-R940^SMC6 and K1141^SLF2-E1008^SMC6), a cation-π interaction (K1125^SLF2-Y944^SMC6), and multiple van der Waals contacts (Figure 2A; Figure 2—figure supplement 1C, D). Thus, the model and mutational data are consistent.

Next, we validated these SMC6’s interface contacts using alanine substitutions. We introduced three sets of mutations: 1) SIMC1-only (K942A/D946A/K957A), whose wild-type residues form salt bridges with SIMC1; 2) SLF2-only (R940A/Y944A/N947A), where R940 and Y944 interact with SLF2 while N947 forms salt bridges with SIMC1; 3) Mixed (L939A/R940A/K942A/L943A), which alters two leucines that insert into a hydrophobic cleft between α1 of SIMC1 and α17 of SLF2, along with two other residues involved in salt bridges (Figure 2B). When these SMC6 mutants were co-expressed with SIMC1 and SLF2, pull-down assays showed significantly reduced or abolished binding (Figure 2C, lane 2-4). Thus, our mutational data strongly support the AlphaFold model.

We conclude that SIMC1-SLF2 associates with the neck region of SMC6, opposite the ATPase active site used for heterodimerization with SMC5 (Figure 2D; Figure 2—figure supplement 1A). A recent cryo-EM study on the S. cerevisiae Smc5/6 complex similarly demonstrated that Nse5/6 engages with the neck region of Smc6 (30), reinforcing the idea that this mode of interaction is evolutionarily conserved (Figure 2D). This has implications for the function of SIMC1-SLF2 in the regulation of SMC5/6 activity, based on the findings for yeast Nse5/6 (see discussion).

SMC5/6 mediated plasmid silencing requires the Nse5/6-like SIMC1-SLF2 complex

To examine the role of SIMC1 and SLF2 in SMC5/6-mediated episomal DNA silencing, we used CRISPR-Cas9 to create U2OS SIMC1^-/- or SLF2^-/- cells. We compared the expression of a transfected GFP reporter plasmid between wild-type (WT) and null cells. GFP transcripts, quantified by qPCR, showed a strong increase over WT in both SIMC1^-/- and SLF2^-/- cells (Figure 3A). We also used fluorescence activated cell sorting (FACS) to monitor GFP expression (37) and found that both SIMC1^-/- and SLF2^-/- cells expressed significantly more GFP than WT (Figure 3B). In addition to U2OS cells, SIMC1^-/- HEK293 cells also showed increased GFP reporter expression (Figure 3—figure supplement 1). Importantly, expressing WT SIMC1 or SLF2 in the corresponding null cells reduced reporter expression to levels observed in WT cells (Figure 3B), demonstrating the involvement of SIMC1-SLF2 in plasmid silencing. Notably, however, SLF2^mut1 that abolishes SIMC1-SLF2 interaction with SMC6 (Figure 1C) did not reduce reporter expression in SLF2^-/- cells (Figure 3B). These findings demonstrate that SIMC1-SLF2, and their interaction with SMC6, are required for plasmid repression.

Figure 3:

SIMC1 and SLF1 direct SMC5/6 to plasmid silencing and DNA repair, respectively

We previously showed that SLF1/2 and SIMC1-SLF2 form mutually exclusive complexes with SMC5/6 and recruit it to distinct cellular locations (33). Specifically, epitope-tagged SLF1 colocalized with gamma-H2AX at laser stripes, whereas SIMC1 did not. Here we more directly examined the roles of SLF1 and SIMC1 in recruiting SMC5/6 to DNA damage sites. To this end, we induced localized DNA lesions using laser microirradiation in WT, SLF1^-/-, and SIMC1^-/- U2OS cells. In WT and SIMC1^-/- cells SMC5/6 accumulated at laser-induced lesions marked by gamma-H2AX, whereas no such accumulation was seen in SLF1^-/- cells (Figure 4A). Thus, SLF1 but not SIMC1 plays a critical role in directing SMC5/6 to DNA damage sites. Conversely, when comparing GFP reporter expression in WT and SLF1 null cells, we found no change in GFP levels in the absence of SLF1 (Figure 4B). Thus, unlike SIMC1, SLF1 does not play a role in SMC5/6-mediated plasmid silencing, which is consistent with what has been observed in SLF1^-/- HepG2 cells (10).

Figure 4:

SUMOylation supports plasmid repression

Given that SIMC1 contains SIMs, and the SMC5/6 complex is a SUMO ligase that contains other SUMO-interacting interfaces, we also tested for a potential involvement of the SUMO pathway in plasmid silencing (33,38,39). Cells were treated with the SUMO pathway inhibitor (SUMOi), TAK981 (40). We observed a strong dose-dependent increase in reporter gene expression in both U2OS and RPE cells (Figure 5—figure supplement 2), supporting an involvement of the SUMO pathway in plasmid silencing. This is consistent with the broadly observed transcriptionally repressive activity of SUMOylation (41,42). Interestingly, SUMOi treatment does not significantly increase the level of GFP reporter expression in SIMC1^-/- and SLF2^-/- over its effect on WT, despite the pre-existing defect in plasmid silencing in SIMC1 and SLF2 null cell lines (Figure 5A). The lack of additivity between SUMOi and SIMC1^-/- or SLF2^-/- suggests that SMC5/6-mediated plasmid silencing operates in large part under the broad “umbrella” of SUMOylation-based transcriptional repression.

Figure 5:

PML NBs are not required for plasmid silencing

We previously showed that SIMC1-SLF2 directs SMC5/6 to LT antigen-containing SV40 replication centers at PML NBs (33). Moreover, SMC5/6-mediated restriction of HBV occurs at PML NBs (34,43) and, the localization of SMC5/6 at PML NBs is SLF2-dependent in HepG2 and PHH cells (10,43). We also now find that SLF2 is required for the localization of SMC5/6 to PML NBs in U2OS cells (Figure 5—figure supplement 1). Based on the foregoing, PML NBs have been proposed, but not directly shown, to support simple plasmid repression (10). Therefore, we tested the impact of PML NBs on plasmid silencing by measuring GFP reporter expression in WT and PML^-/- U2OS cells (44). Compared to SIMC1 and SLF2 null cells, we observed only a minor increase in GFP expression in PML^-/- versus WT cells (Figure 5B). Moreover, the stable re-expression of PML in PML^-/- cells did not significantly reduce the level of GFP expression, as compared to vector control, suggesting the slight increase observed is due to clonal variation from CRISPR editing (Figure 5B). Thus, we conclude that PML NBs are not needed for the SMC5/6-mediated silencing of plasmids.

SV40 Large T antigen (LT) overcomes plasmid silencing

Considering that HBx and Vpr antagonize SMC5/6 to induce HBV and HIV-1 transcription, respectively (45–47), we tested if SV40 LT similarly impacts plasmid silencing. In this regard, we previously found that SMC5/6 colocalizes with LT-induced nuclear foci at PML NBs in a SIMC1-SLF2-dependent manner (33). Here, we show that ectopically expressed LT coimmunoprecipitates with GFP-SMC5/6 but not a GFP control (Figure 5C). Therefore, LT, like HBx and Vpr, may contact SMC5/6 to antagonize its ecDNA repression activity. Indeed, the stable expression of LT in WT cells promoted GFP reporter plasmid expression, as compared to WT cells with an empty vector control (Figure 5D; Figure 5–figure supplement 3). The host range defective mutant LT-HR684 had a similar effect as WT LT, indicating that the C-terminus of LT, which inhibits the SV40 host range restriction factor FAM111A (48), is not involved in overcoming plasmid repression (Figure 5D). Notably, the enhancement of plasmid DNA transcription by LT is not additive with that already present in SIMC1^-/- and SLF2^-/- cells (Figure 5D). The epistatic relationship between the positive effects of SIMC1^-/-, SLF2^-/-, and LT expression on plasmid transcription indicates that LT overcomes SMC5/6-mediated repression.

Discussion

Our structural analyses indicate that SIMC1-SLF2 engages the SMC5/6 core at the neck region of SMC6, situated opposite the ATPase active site. These findings, together with recent cryo-EM analyses of the S. cerevisiae Smc5/6 complex (20,21,31), highlight a conserved mode of interaction in which Nse5/6 or Nse5/6-like subunits bind to the backside of SMC6’s head domain. This binding site is ideal for allosteric regulation of the ATPase and DNA-binding functions of the SMC5/6 family. Indeed, yeast Nse5/6 reduces Smc5/6 ATPase activity and disrupts the Nse4 (kleisin)-Smc6 interface, which may open the gate to enable topological DNA binding by Smc5/6 (Figure 6, (20,30,31)). The clear overlap between our validated SIMC1-SLF2-SMC6 structure and that of yeast Nse5/6-Smc6 suggests that SIMC1-SLF2, and its orthologue SLF1/2, may similarly modulate SMC5/6 ATPase and DNA binding activities, albeit at distinct DNA targets.

Figure 6:

The distinct targeting functions of the SIMC1 and SLF1-based SLF2 complexes is underscored by our current analysis of microirradiated cells that lack each complex (Figure 4). While the deletion of SLF1 abrogates SMC5/6 localization at DNA lesions, the loss of SIMC1 has no effect (Figure 4A). On the other hand, SIMC1 is required for the recruitment of SMC5/6 to LT-containing PML NBs (33). Lastly, we found that SLF1 does not impact plasmid silencing (Figure 4B), consistent with several analyses of HIV-1, rAAV, and HBV silencing (10,43,45,49) but contrasting with another study (50). It remains unclear why this latter discrepancy exists. While there may still be overlapping functions of SIMC1 and SLF1 to uncover at the organismal level, it is clear they support distinct cellular responses to ecDNAs and DNA lesions (Figure 6).

In keeping with the SUMO-binding ability of SIMC1 (Figure 6, (39)), we found that the inhibition of SUMOylation by TAK-981 strongly enhanced plasmid transcription (Figure 5–figure supplement 2). This is likely a broad effect that includes the loss of repressive modifications of transcription factors (42); as well as the lifting of SMC5/6-mediated repression to enhance ecDNA transcription. A similar effect of TAK-981 was seen on HIV-1 proviral DNA gene expression, which was elevated by SUMO inhibition to a similar degree as that observed in SMC5Δ cells (50). The authors concluded that the effect of TAK-981 was entirely through the inhibition of SMC5/6-mediated SUMOylation (NSE2 subunit). This is different to our results, which show that SUMOylation inhibition has a larger effect than the loss of the SIMC1-SLF2 complex on plasmid gene expression. The distinct natures of the ecDNAs under study, plasmid versus HIV-1 proviral DNA, and the genetic backgrounds may contribute to this difference. Nevertheless, our data support a key role of the SUMO pathway in plasmid repression, and that SMC5/6 works within it.

Despite the fact that SIMC1-SLF2 localizes SMC5/6 to PML NBs that contain SV40 LT (31), and HBV is restricted at PML NBs (34), they are dispensable for plasmid silencing. While the reason for the different impact of PML NBs in HBV versus simple plasmid silencing remain unknown, it may be related to the natural localization of each ecDNA. That is, HBV localizes to PML NBs as part of the viral life cycle, in a manner dependent on the SUMOylation of its core protein (51). In contrast, the direct visualization of plasmids revealed they are pan-nuclear and apparently excluded from PML NBs (52). It is also noteworthy that a recent genome-scale CRISPR screen for repressors of rAAV transcription identified all components of the SMC5/6 complex, including SIMC1 and SLF2, but not SLF1 or PML (49). This raises interesting questions about the mechanism of SMC5/6-mediated transcriptional silencing, since it can occur outside PML NBs. In this regard, SUMOylation was recently shown to support the recombination-based alternative lengthening of telomeres (ALT) pathway in the absence of PML NBs (53), which are normally key components of the fully active pathway (44). Therefore, the SUMOylation of plasmid associated proteins, and or SMC5/6, may support transcriptional repression of plasmids in the absence of PML NBs, consistent with our findings using TAK-981.

Interestingly, the expression of LT enhances plasmid transcription in a non-additive manner with the deletion of SIMC1 or SLF2, suggesting an epistatic relationship. While LT is known to be a promiscuous transcriptional activator (54,55) that does not rule out a co-existing role in antagonizing SMC5/6. Indeed, these findings are reminiscent of HBx from HBV and Vpr of HIV-1, both of which are known promiscuous transcriptional activators that also directly antagonize SMC5/6 to relieve transcriptional repression (12,45–47,56–58). Unlike HBx and Vpr, the interaction of LT with SMC5/6 does not appear to induce the degradation of its subunits (33). Nevertheless, LT binds and sterically inhibits several cellular proteins instead of inducing their degradation, including Rb and p53 (59). Further analysis will be needed to define LT-SMC5/6 contacts, and its potential mode of inhibition. This will have important ramifications for the lifecycle of the pathogenic polyomaviruses since their LT proteins share functions with SV40 LT (60).

In conclusion, our structure-function analyses delineate a specific function for the SIMC1-SLF2 complex in SMC5/6-mediated plasmid repression. This resolves a discrepancy in the field, wherein SLF2 was thought to act alone in ecDNA repression, since SLF1 is not required. While further studies are needed, SIMC1-SLF2 is most likely responsible for supporting SMC5/6-mediated ecDNA silencing, including that of multiple pathogenic viruses. Although HBV silencing requires PML NBs, we find that plasmid silencing does not. This represents a departure from current dogma, suggesting that the SMC5/6-mediated repression mechanism does not always rely on PML NB-resident factors. In some cases, the SUMOylation of ecDNA-associated proteins may be sufficient to support SMC5/6-based repression. Finally, because the SV40 LT antigen interacts with SMC5/6 and antagonizes its repression of plasmids, the orthologous polyomaviral LT antigens are likely to block SMC5/6-mediated restriction of pathogenic polyomaviruses.

Material and Methods

Construction of recombinant plasmids

Construction of plasmid DNA was described in (33). Mutations in SIMC1 (combo ctrl), SLF2 (mut1, mut2) and SMC6 (mixed, SIMC-only, SLF2-only) were either introduced in primers or a gBlock sequence containing desired mutations was purchased (IDT). All plasmids created in this study have been verified by sequencing service provided by Plasmidsaurus. Plasmid used in this study are listed in Appendix 1. Additional details of plasmid construction are available upon request.

Cell culture, transfection, stable line generation

All cell lines and their derivatives (see Appendix 1) were cultured in DMEM (Gibco) supplemented with 10% (v/v) fetal bovine serum (Omega Scientific), 1% (v/v) antibiotic-antimycotic (Gibco) and maintained at 37 °C in humidified air with 5% CO₂. Transient plasmid transfections and generation of stable cell lines was detailed in (33). A complete list of stable mammalian cell lines generated, and reagents used in this study is provided (Appendix 1).

Cell line generation by CRISPR/Cas9

SIMC1, SLF1 and SLF2 knockout clones were generated via reverse transfection of CRISPR/Cas9 Ribonucleoprotein. A mixture of three sgRNAs specific for each gene (Synthego; 24 nM) was combined with 6 nM Cas9-NLS (Synthego) in reduced-serum medium OptiMEM and incubated for 10 min followed by a 15 min incubation with TransIT-X2 (Mirus). 6 x 10⁴ cells were added and plated into 12-well plates. After 72 hours, cells were diluted and plated to isolate single cell clones that were genotyped to confirm successful editing. sgRNA sequences and diagnostic primers used in this study are listed in Appendix 1.

Co-immunoprecipitation and Western blotting

GFP-labeled proteins were immunoprecipitated using the Nano-Trap magnetic agarose (Chromotek) following the manufacturer’s instructions. The protocol as well as whole cell lysate preparation for SDS-PAGE and western blot are described in detail in (33). Antibodies used for immunoblotting are listed in Appendix 1.

RNA extraction and quantitative PCR

For monitoring GFP expression by qPCR, cells were harvested for RNA isolation 72 hours after GFP vector transient transfection. Total RNA precipitated using RNeasy Plus Mini kit (Quiagen) was treated with DNaseI kit (Invitrogen) before cDNA synthesis by SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen). SensiFAST SYBR No-ROX kit (Meridian Bioscience) was used for qPCR and gene expression was calculated as 2^-ΔCt using beta-actin as a reference. Oligonucleotides are listed in Appendix 1.

Flow cytometry

GFP fluorescence measurements were based on (37). GFP reporter vector was transiently transfected into cells using TransIT-LT1 (Mirus). After 72 hours, cells were washed in PBS, dissociated using 0.25% trypsin and resuspended in DMEM. For SUMOi experiments, cells were treated with 100 nM TAK-981 (Cayman Chemical) or 0.1% DMSO in control plates at the time of GFP vector transfection. Median GFP intensity was measured on Cytoflex S (Beckman Coulter) using the CytExpert software. A minimum of 1100 GFP-positive cells were analyzed.

DNA damage by laser microirradiation

Protocol adapted from (61). U2OS cells (1.5 x 10⁵) were plated in a 35 mm μ-Grid dish (Ibidi) a day before microirradiation. Cells were treated with 50 uM angelicin (ThermoFisher) and 1 μg/ml Hoechst 33342 (Invitrogen) for 30 min before transferring to the stage incubator of LSM880 Airyscan confocal laser scanning microscope (Zeiss). Using the bleaching mode in ZEN software, DNA damage was induced by irradiation of a 5-pixel wide region with a 405 nm diode laser (15% power, 1 iteration, zoom 1, averaging 1, pixel dwell time 0.27 μs = speed 7). One hour after the irradiation, cells were treated with extraction buffer (0.5% Triton X-100, 20 mM Hepes, 50 mM NaCl, 3 mM MgCl₂, 300 mM sucrose) for 2 min, washed twice in PBS, fixed in 4% formaldehyde in PBS for 15 min and stained with anti-phospho-Histone H2A.X (Ser139) and anti-SMC6 antibodies following the immunofluorescence protocol.

Immunofluorescence and microscopy

Immunofluorescence and confocal microscopy were done as described (33). For optimal detection of SMC6, we pre-extracted the chromatin with extraction buffer (0.5% Triton X-100, 20 mM Hepes, 50 mM NaCl, 3 mM MgCl₂, 300 mM sucrose) for 2 min on ice. Antibodies used in this study are listed in Appendix 1.

Luciferase and MTT assays

U2OS cells were seeded at 4 x 10³ cells per well of a 96-well plate and transfected with 20 ng of luciferase reporter plasmid (expresses secreted Gaussian luciferase under EF1alpha promoter) using TransIT-LT1 (Mirus) by reverse transfection. TAK-981 in indicated concentrations or DMSO were added at the time of transfection and Luciferase activities were measured 3 days later. Ten microliters of culture media were mixed with 50 ul Gaussia Luciferase Glow Assay buffer that contains freshly diluted coelenterazine substrate (Pierce^TM Gaussia Luciferase Glow Assay Kit). Luminescence was measured after ten minutes on a Tecan Infinite M200 plate reader (Tecan Life Sciences). The number of viable cells were estimated using MTT assay kit (Sigma) following manufacturer’s instructions.

AlphaFold-Multimer Structural modeling

AlphaFold-Multimer modeling (36) was performed using default settings on the following protein sequences: SIMC1^Nse5(425–872), SLF2^Nse6 (733–1173), and SMC6 (1–1091).

Material Availability

All materials produced during this work are available upon written request, in keeping with the requirements of the journal, funding agencies, and The Scripps Research Institute.

Author information

Contributions

Conceptualization, M.N.B., T.O. Design of Methodology, M.O., M.N., M.N.B., T.O. Writing – Original Draft, M.N.B., M.N., T.O. Writing – Review & Editing, M.O., M.N., T.O., M.N.B. Formal analysis/Investigation, M.O., M.N., M.N.B., T.O. Validation, M.O., M.N. Resources, M.O., M.N. Visualization, M.O., M.N., T.O., M.N.B. Funding acquisition, M.N.B., T.O. Supervision, M.N.B., T.O. Project administration, M.N.B.

Supplementary figures

Figure 1 - figure supplement 1.

Figure 2 - figure supplement 1

Figure 3 - figure supplement 1

Figure 5 - figure supplement 1

Figure 5 - figure supplement 2

Figure 5 - figure supplement 3

Acknowledgements

This work was supported by NIH grants R35 GM136273 to M.N.B. and GM092740 to T.O.