CRL7^SMU1 E3 ligase complex-driven H2B ubiquitylation functions in sister chromatid cohesion by regulating SMC1 expression

Ubiquitylation is an essential post-translational modification that regulates a wide array of cellular processes in eukaryotes (Hershko and Ciechanover, 1998). Ubiquitin is covalently attached through its C-terminal glycine (G) residue to the ε-amino group of lysine (K) or, occasionally, to the N-terminus of the protein substrate. Proteins can be tagged with either single ubiquitin or with a polyubiquitin chain. Linkage of ubiquitin with substrates occurs in a three-enzyme cascade catalyzed by ubiquitin like modifier activating enzymes (UBAs, hereafter referred to as E1), ubiquitin-conjugating enzymes (UBE2s, hereafter referred to as E2) and ubiquitin ligases (E3 ligases). E3 ligases are the most heterogeneous class of enzymes, which bring together the correct E2 with the right substrate and, thus, are critical for defining substrate specificity during ubiquitylation process (Berndsen and Wolberger, 2014; Hershko and Ciechanover, 1998).

Depending on the domain architecture and on the mechanism of ubiquitin transfer to the substrate, E3 ligases have been divided into three types: homologous to the E6-AP carboxyl-terminus (HECT) E3s, really interesting new gene (RING) E3s and ring between ring fingers (RBR) E3s. The HECT-type E3 ligases, characterized by the presence of a HECT domain, possess intrinsic catalytic activity. HECT-type E3s load ubiquitin on themselves through formation of an ubiquitin-thioester intermediate with the catalytic cysteine residue within the HECT domain and then transfers the ubiquitin to the target protein (Berndsen and Wolberger, 2014; Rotin and Kumar, 2009). HECT-type E3s predominantly function as monomeric enzymes, but the existence of functional multimeric HECT-type E3s in the cell has recently emerged (Maddika and Chen, 2009). For example, the HECT-type E3 UBR5 (also known as and hereafter referred to as EDD) forms a multi-component E3 ligase with proteins DDB1, DYRK2 and VPRBP (officially known as DCAF1) as core components, to regulate G2/M transition. RING-type E3s, characterized by the presence of a RING domain or an U-box domain, are the most abundant type of ubiquitin ligases in the cell (Berndsen and Wolberger, 2014). RING-type E3s do not have intrinsic catalytic activity however, they function as scaffold between E2 and the substrate and thereby mediate direct transfer of ubiquitin to the substrate. RING-type E3s can function as monomers, homodimers, heterodimers and multi-protein complexes, such as in those comprising cullin proteins and RING-type E3s ligases (CRLs) (Francis et al., 2013; Lydeard et al., 2013; Petroski and Deshaies, 2005; Zimmerman et al., 2010). However, RBR-type E3 ligases share common features with both RING-type and HECT-type E3 ligase families. They recruit ubiquitin-linked E2 through their RING domain but, like HECT-type E3s, catalyze the ubiquitin transfer in a two-step mechanism, with the first transfer to the E3 itself and the second transfer to the substrate (Spratt et al., 2014).

CRLs are a superfamily of RING-type E3s responsible for as much as 20% of ubiquitin-dependent protein modification in cells (Bennett et al., 2010; Petroski and Deshaies, 2005). CRLs assemble into multimeric complexes to enhance substrate diversity during ubiquitylation. CRLs consist of two distinct modules: (1) a substrate-targeting unit composed of a substrate-recognition protein and an adaptor protein that links the module to the cullin, and (2) the RING component that is active in recruiting an ubiquitin-linked E2 enzyme (Zimmerman et al., 2010). CRLs ubiquitylate several cellular substrates and thus control broad range of biological processes, including cell growth, development, signal transduction, transcriptional control and tumor suppression (Harper and Tan, 2012; Petroski and Deshaies, 2005). However, no CRL E3 ligase is implicated in sister chromatid cohesion so far. In this study, we identified a new CRL (i.e. the CRL7^SMU1 complex, comprising SMU1, DDB1, CUL7 and RNF40 as core components) that has an essential role in maintenance of chromatid cohesion. We demonstrated that CRL7^SMU1 E3 ligase supports sister chromatid cohesion through H2B ubiquitylation at SMC1a locus and thereby controlling its expression.

Previous studies have demonstrated that VPRBP, a protein with a WD repeat region and a LIS1 homology (LisH) domain, acts as a substrate recognition component of HECT-type as well as RING-type E3 ubiquitin ligase complexes (Maddika and Chen, 2009; Nakagawa et al., 2013). To identify proteins with a similar combination of LisH and WD repeat organization that can assemble E3 ligase complexes, we performed a global search for LisH domain (ID: PS50896) using UniProt database. We retrieved 28 human proteins that contain LisH domain, nine of which exhibit a combination of LisH domain with WD repeats in their architecture (Fig. 1A). To test if these proteins assemble E3 ligase complexes, we isolated protein complexes associated with SMU1, one of the listed proteins with such a domain organization. SMU1 (suppressor of mec-8 and unc-52) is a spliceosome accessory protein known for its role in DNA replication, mitotic spindle assembly and maintenance of genomic stability (Ren et al., 2013; Sugaya et al., 2006, 2005). Tandem affinity purification of S-protein tag, FLAG tag, streptavidin-binding protein tag (SFB)-triple-tagged SMU1 followed by mass spectrometry (MS) analysis revealed several known and unknown interacting partners (Fig. 1B). Interestingly, we found DDB1, CUL7 and RNF40 among SMU1-interacting proteins. CUL7 is a member of the cullin family of proteins that function as scaffold for E3 ubiquitin ligases. Like other cullins, CUL7 also assembles into a CRL by associating with Skp1, Fbx29 and ROC1 (Dias et al., 2002). However, DNA damage-binding protein 1 (DDB1) together with DNA damage-binding protein 2 (DDB2) is known to function in nucleotide excision repair. At the molecular level, DDB1 functions as an adaptor protein for the Cul4 ubiquitin E3 ligase complex, and DYRK2 in complex with the EDD, DDB1, VPRBP (EDVP)-comprising E3 ligase complex (DYRK2–EDVP) to regulate ubiquitin-dependent degradation of various substrates, such as Cdt1, c-Jun, p21 and katanin p60 (Iovine et al., 2011). RNF40 is a RING-type E3 ligase that heterodimerizes with RNF20 to monoubiquitylate histone H2B to H2Bub (Kim et al., 2009). As it is well known that CUL7 acts as a scaffolding protein (Dias et al., 2002) and DDB1 as adaptor (Iovine et al., 2011) for various E3 ligase complexes and, further, RNF40 being an E3 ligase, we thus propose the existence of a new CRL7 E3 ligase complex with SMU1 as the substrate recognition component.

By performing immunoprecipitation using antibody against SMU1, we validated the endogenous association of RNF40, DDB1 and CUL7 with SMU1 (Fig. 1C). Importantly, DYRK2 – a scaffolding subunit of the EDVP E3 ligase complex – does not interact with SMU1, suggesting that CRL7^SMU1 is a distinct complex. Further, exogenously expressed SMU1 efficiently interacted with CUL7, RNF40 and DDB1, but not with DYRK2 (Fig. S1A). Likewise, we found no interaction of SMU1 with VPRBP (Fig. 1D), a known functional subunit of the DYRK2–EDVP complex as well as the CUL4–DDB1 complex, again supporting our conclusion that CRL7^SMU1 is a discrete E3 ligase complex in cells. In fact, SMU1 is not part of any known CRLs because SMU1 neither interacts with SKP1 (Fig. 1E) nor with ROC1 (Fig. 1F), while ROC1 and SKP1 interacted positively. Whereas DDB1 can associate with CUL4A, CUL7, SMU1 and RNF40 (Fig. S1B), we found no interaction of CUL7 and RNF40 with CUL4A (Fig. S1C,D), again supporting CRL7^SMU1 as an independent complex. Previously, RNF40 was known to act as a functional E3 ligase in its RNF40/RNF20 heterodimeric form (Kim et al., 2009). Surprisingly, however, we found that SMU1 does not interact with RNF20 (Fig. 1G), implying that the SMU1–RNF40 E3 ligase complex is formed independently of RNF20.

Furthermore, to investigate whether SMU1 forms a stable E3 ligase complex in vivo, we analyzed these proteins in HEK-293T cell extracts by using size-exclusion chromatography. SMU1, DDB1, CUL7 and RNF40 were co-eluted in similar cell lysate fractions corresponding to a molecular mass of ∼440 kDa (Fig. 1H), suggesting that CRL7^SMU1 complex proteins physically interact with each other and assemble a large complex in cells. It is well established that WD repeats assist in the assembly of E3 ligase complexes, such as DDB1-CRLs. To delineate the regions of SMU1 required for assembly of the E3 ligase complex, we made deletion constructs of SMU1 that either lack the WD repeat region or the N-terminal LisH domain-containing region (Fig. 1I). Interestingly, we found that, while the WD repeat region is dispensable for association with components of the E3 ligase, the LisH domain region is required for assembly of the E3 ligase complex (Fig. 1J). We demonstrate that SMU1, via its LisH domain region, assembles a so-far-unknown E3 ligase complex by associating with DDB1, CUL7 and RNF40 in cells.

In search of substrate(s) of the newly identified E3 ligase complex, we next examined the MS data derived from SMU1 purification. We found histone 2B (H2B) to be one of the interacting partners of SMU1. H2B, together with H2A, H3 and H4, is a part of the core nucleosome and a well-known substrate of the RNF20–RNF40 heterodimer that monoubiquitylates H2B at the K residue at position 120 (K120) in humans (K123 in yeast) (Kim et al., 2009; Wood et al., 2003). However, no substrate recognition subunit for RNF20–RNF40 E3 ligase has yet been reported. As we identified RNF40 as a core component in the CRL7^SMU1 complex, we proposed H2B to be a probable substrate of this complex. By performing the co-immunoprecipitation experiments using streptavidin pull-down followed by immunoblotting with anti-H2B antibody, we confirmed the specific interaction of H2B with CUL7, DDB1, RNF40 and SMU1 (Fig. 2A). To establish the architecture of the complex and further test if SMU1 acts as substrate recognition component in the E3 ligase complex, we performed knockdown experiments. On one hand, interaction between RNF40 and H2B was severely hampered upon depletion of SMU1, whereas DDB1 interaction with RNF40 was intact in these cells (Fig. 2B). On the other hand, interaction between SMU1-H2B and SMU1-DDB1 was unaffected upon knockdown of RNF40 (Fig. 2C). However, depletion of DDB1 resulted in reduced binding of SMU1 to CUL7 (Fig. 2D). Also, depletion of CUL7 led to loss of interaction between RNF40 and DDB1 without having any effect on the interaction between SMU1 and DDB1 (Fig. 2E). These data suggest that H2B binds to SMU1, which associates to CUL7 and, further, RNF40 through DDB1 (Fig. 2F). To substantiate the role of SMU1 as the substrate recognition protein of the complex, we next checked the direct interaction of recombinant bacterially expressed glutathione S-transferase (GST)-tagged SMU1 (GST-SMU1) with maltose-binding protein (MBP)-tagged H2B (MBP-H2B). Indeed, we found that SMU1 interacts with H2B directly under in vitro conditions (Fig. 2G). Together, these experiments suggest that SMU1 functions as a substrate-recognition component that links H2B with DDB1–CUL7–RNF40. Next, to examine the role of the CRL7^SMU1 complex in histone H2B ubiquitylation, we used specific small interfering (si)RNAs or short hairpin (sh)RNAs to deplete complex proteins in HeLa cells. It has been shown that knockdown of RNF20/40 abolishes H2B ubiquitylation (Kim et al., 2009). In agreement with previous studies we found that knockdown of RNF40 led to downregulation of H2B monoubiquitylation at K120 (Fig. 2H). Interestingly, similar to knockdown of RNF40, depletion of SMU1 (Fig. 2I), DDB1 (Fig. 2J) and CUL7 (Fig. 2K) also led to significant downregulation of H2B ubiquitylation, suggesting that these proteins function together to ubiquitylate H2B K120 in vivo. Further, RNF40-mediated ubiquitylation of H2B is known to stimulate H3K4 trimethylation via the complex of proteins associated with Set1 (COMPASS) (Kim et al., 2009). Similar to RNF40 (Fig. 2H), depletion of SMU1, DDB1 and CUL7 resulted in reduction of H3K4 trimethylation (Fig. 2I-K). DDB1 depletion, as well as RNF40 depletion, seems to affect H3K4 trimethylation levels more than H2B ubiquitylation. Given that DDB1 is part of multiple independent E3 ligase complexes, it might be possible that DDB1 regulates H3K4me3 through additional mechanisms that are independent of H2B ubiquitylation. In fact, it has been reported that CUL4-DDB1 ubiquitin E3 ligase interacts with multiple WD repeat-containing proteins and regulates H3 methylations, including that of K4Me1, K4Me3, K9Me3 and K27Me3 (Higa et al., 2006). Likewise, RNF40 might also be forming other protein complexes to directly regulate H3K4me3 independently of H2B ubiquitylation. Nonetheless, our data suggest that SMU1, DDB1, CUL7 and RNF40 are integral components of the functional E3 ligase complex that regulate histone modifications.

Next, we sought to understand the functional role of the CRL7^SMU1 complex in cells. We observed that cells depleted of SMU1 in response to SMU1 siRNA transfection (Fig. S2A) proliferated at a significantly slower rate than control cells transfected with control siRNA (Fig. S2B). Furthermore, we found that depletion of SMU1 led to the accumulation of 4N cells as well as polyploid (>4N) cells (Fig. 3A). This increase in the 4N population is due to arrest in the mitotic phase, as time-lapse imaging analysis confirmed that cells lacking SMU1 spent several hours in mitosis while control siRNA cells took 60 min on average to complete mitosis (Fig. 3B,C). Likewise, we found that knockdown of all individual components of the CRL7^SMU1 complex led to accumulation of phosphorylated H3-positive cells (Fig. 3D,E) and significant accumulation of cells with round morphology in culture (Fig. S2C), typical of mitotically arrested cells. Since we found that CRL7^SMU1 complex proteins are required for normal progression of mitosis, we next tested if loss of SMU1 leads to any mitotic defects. We observed various chromosomal and spindle defects upon SMU1 knockdown (Fig. 4A). Loss of SMU1 resulted in significant increase in cells displaying lagging chromosomes, anaphase/nuclear bridges and multipolar spindles (Fig. 4B). Similarly, we also found that depletion of CUL7, DDB1 or RNF40 individually in cells led to severe mitotic defects (Fig. 4C,D). Since loss of CRL7^SMU1 complex proteins resulted in numerous mitotic defects, we next tested if loss of H2B monoubiquitylation at K120 phenocopies the loss of E3 ligase complex from cells. Expression of the H2B K120R mutant, but not wild type H2B, resulted in accumulation of cells with multiple mitotic defects (Fig. 4E–G), thus suggesting that H2B ubiquitylation at K120 is critical for normal mitotic progression and further prevention of genomic instability.

It is well known that H2B ubiquitylation is enriched at sites of active gene transcription and modulates transcription elongation. Since we found that the CRL7^SMU1 complex promotes H2B ubiquitylation and H3K4 trimethylation (sites of active gene transcription), we hypothesized that this complex controls transcription of a specific set of genes required for mitosis. Previously, a microarray analysis upon SMU1 depletion had revealed alterations in the expression of various mitotic genes (Papasaikas et al., 2015). We tested if SMU1 is enriched on the DNA of this set of mitotic genes by using chromatin immunoprecipitation assay (ChIP). SMU1 was significantly enriched on different mitotic genes, such as those encoding SMC1a, ANAPC12, ANAPC5, Aurora A and CDCA2 (Fig. 5A). Consequently, we tested if depletion of CRL7^SMU1 components affects H2B ubiquitylation at these loci. Although, depletion (Fig. S3) of individual components of the E3 ligase complex reduced H2Bub levels at distinct loci, we found SMC1a to be common locus where H2B ubiquitylation is hampered following the loss of the CRL7^SMU1 complex (Fig. 5B). As H2B ubiquitylation is directly associated with gene transcription, we further tested whether this E3 ligase complex regulates transcript levels of mitotic genes. Again, we observed that expression levels of only SMC1a were reduced upon depletion of CUL7, DDB1, RNF40 or SMU1 (Fig. 5C). Consequently, we also found a significant reduction in SMC1a protein levels upon depletion of individual components of CRL7^SMU1 complex (Fig. 5D). Notably, expression of the H2B K120R mutant also reduced the gene expression of SMC1a (Fig. 5E) consistent with our hypothesis that H2B ubiquitylation is required for expression of this crucial mitotic gene.

Structural maintenance of chromosomes protein 1A (SMC1a) is a central component of the cohesin complex, which is required for the cohesion of sister chromatids and essential for accurate chromosome segregation at the onset of mitosis (Brooker and Berkowitz, 2014; Sumara et al., 2000). To determine if the CRL7^SMU1 complex is required for sister chromatid cohesion we prepared metaphase spreads on HeLa cells that had been depleted of individual components of the E3 ligase complex. Control cells display normal ‘X-shaped’ mitotic chromosomes, with sister chromatids tightly linked at the centromere and chromosome arms separated. However, strikingly, knockdown of CUL7, DDB1, RNF40 or SMU1 led to parallel chromatids and single chromatids in isolation within a large fraction of mitotic cells (Fig. 6A,B) suggesting a premature loss of sister chromatid cohesion in these cells. We next measured the distance between metaphase sister kinetochores in CRL7^SMU1 complex-depleted cells by immunostaining CENP-A. Interkinetochore distances between sister chromatids were significantly increased in aligned metaphases of cells depleted of SMU1, CUL7, DDB1 and RNF40 compared with control metaphases (Fig. 6C,D), further supporting a critical role of CRL7^SMU1 complex in establishing cohesion between sister chromatids. Interestingly, expression of the H2B K120R mutant but not of wild type H2B also resulted in defective chromatid cohesion (Fig. 6E,F) and enhanced interkinetochore distance (Fig. 6G,H), thus suggesting that H2B ubiquitylation by CRL7^SMU1 complex is critical for maintenance of sister chromatid cohesion.

Based on these observations, we hypothesized that expression of H2B-Ub fusion protein, which mimics H2B K120 ubiquitylation (Zhang et al., 2013), rescues the cohesion defects caused by depletion of CRL7^SMU1 E3 ligase from cells. To our surprise, overexpression of H2B-Ub fusion protein could not rescue cells from cohesion defects (Fig. S4A,B) caused by loss of E3 ligase. These data prompted us to reason that dynamic H2B ubiquitylation and deubiquitylation is necessary for maintenance and later dissolution of sister chromatid cohesion during appropriate phases of mitosis. Therefore, constitutive loss or presence of H2B monoubiquitylation might be deleterious for normal chromatid cohesion. This is in agreement with previous data, which showed that loss of USP44, a deubiquitinase for H2B that mimicks constitutive H2B ubiquitylation, also leads to various mitotic abnormalities (Zhang et al., 2012). In support of the dynamic nature of H2B ubiquitylation during chromatid cohesion, we found that transient expression of H2B-Ub fusion protein alone is sufficient to induce sister chromatid cohesion defects (Fig. S4C,D), increased interkinetochore distance (Fig. S4E) as well as mitotic defects in cells (Fig. S4F,G). Nonetheless, we next tested if restoration of SMC1a in cells through plasmid-based expression would rescue the chromatid cohesion defects caused by loss of E3 ligase. Indeed, exogenous expression of SMC1a (Fig. 7A) significantly prevented defective chromatid cohesion in cells due to depletion of E3 ligase components (Fig. 7B). Also, SMC1a expression in CUL7-, DDB1-, RNF40- or SMU1-depleted cells rescued the interkinetochore distance that was enhanced due to loss of E3 ligase complex (Fig. 7C). In conclusion, we identified that SMU1, DDB1, CUL7 and RNF40 assemble a CRL type E3 ligase complex and promotes monoubiquitylation of H2B to drive the expression of SMC1a, which is essential for maintenance of sister chromatid cohesion during mitosis.

H2B ubiquitylation is one of the critical histone modifications associated with gene expression (Cole et al., 2015; Xie et al., 2017). In humans, RNF20/40 is considered as a major E3 ligase for H2B monoubiquitylation (Kim et al., 2009). Although, H2Bub is known to be globally associated with transcribed genes and its levels correlate positively with gene expression (Minsky et al., 2008), knockdown of RNF20 affects transcription of only a subset of genes (Shema et al., 2008). This suggests the presence of multiple E3 ligase complexes that, independently of RNF20, regulate H2B monoubiquitylation in the cell. In this study, we provided multiple lines of evidence that support an essential role of the CRL7^SMU1 E3 ligase complex in mediating H2B monoubiquitylation. Though H2B ubiquitylation has been implicated in a wide range of cellular processes such as transcription initiation and elongation, DNA damage response, replication, stem cell differentiation, RNA processing and export, its role in chromatid cohesion and mitosis is unexplored. We clearly have shown that our newly identified E3 ligase complex promotes H2B ubiquitylation at SMC1a to maintain sister chromatid cohesion during mitosis. Moreover, ectopic expression of the H2Bub1 mutant (H2B K120R) caused an alteration of the transcription of SMC1a and a significant increase in mitotic defects, including cohesion loss, which is perhaps direct evidence that H2B ubiquitylation is critical for mitotic progression.

E3 ligases play an essential role in the final step of the ubiquitylation process to catalyze the transfer of ubiquitin to appropriate substrates. E3 ligases, in particular the CRLs, often assemble into multimeric complexes to enhance substrate diversity and specificity during the process of ubiquitylation (Lydeard et al., 2013; Zimmerman et al., 2010). Several interaction domains, such as F-Box, SOCS-box, β-domains, WD40 repeats, ankyrin, Kelch, WW and RLD motifs are known to assist in the assembly of multi-component E3 ligases (Petroski and Deshaies, 2005). In this study, using SMU1 as an example, we demonstrated that LisH domain proteins participate in the organization of a functional E3 ligase complex. The LisH domain is a highly conserved domain in eukaryotic proteins and proposed to mediate protein–protein interactions. Although proteins harboring LisH domain are known to participate in processes, such as microtubule dynamics and chromosome segregation, and are implicated in pathogenesis (Emes and Ponting, 2001), studies on the molecular function of the LisH domain are limited. There are certain examples to suggest that the LisH domain participates in the organization of multimeric complexes to regulate protein stability. For instance, in S. cerevisiae, LisH domain proteins, such as GID1, GID7, GID8 together with GID2, GID4, GID5 and GID9, assemble a multimeric glucose induced degradation deficient (GID) complex. The GID complex mediates polyubiquitylation of fructose-1,6-bisphosphatase (FBPase) via the E3 ubiquitin ligase activity of GID2, which contains a RING domain (Menssen et al., 2012). However, the role of LisH domain-containing proteins in the assembly of an E3 ligase in humans is unknown. Although we and others have previously shown that VPRBP, which contains a LisH domain, acts as a substrate recognition protein in RING-type E3s as well as in HECT-type E3s (Maddika and Chen, 2009; Nakagawa et al., 2013), whether the LisH domain is required for the assembly of E3 ligases is unexplored. In another example, WDR26 – in complex with Axin1 – controls β-catenin levels and negatively regulates Wnt signaling (Goto et al., 2016). Although, the LisH domain of WDR26 is critical for β-catenin degradation, whether it participates in the organization of E3 ligase in this case is not studied. Thus, our current study provides clues in corroborating the critical role of LisH domain proteins in organizing E3 ligase complexes.

SMC1 is an essential component of cohesin – a multi-protein complex made up of four subunits [Smc1; Smc3; an α-kleisin subunit, i.e. Mcd1/Scc1 (mitosis) or Rec8 (meiosis); and Irr1/Scc3] – that is conserved from yeast to humans. Cohesin has a well-documented role in chromatid cohesion, where it provides stable but reversible connections between sister chromatids during both mitosis and meiosis (Nasmyth and Haering, 2009). However, recent findings indicate that, in higher eukaryotes, sister chromatid cohesion is not the only major function of cohesin, but also regulates other processes including transcriptional regulation, DNA repair, chromosome condensation and morphogenesis (Dorsett, 2011). Indeed, elegant studies in yeast indicate that severe reduction in the level of chromatin-bound cohesin does not entirely affect its function of holding the sister chromatids together but drastically affects the non-canonical functions (Dorsett, 2011; Heidinger-Pauli et al., 2010; Mehta et al., 2013). Thus, although our studies have clearly demonstrated that defective sister chromatid cohesion due to loss of CRL7^SMU1 complex is dependent on SMC1, it still remains to be determined whether these effects are mediated directly through SMC1 localization at centromeres or indirectly through regulation of the expression of unknown genes.

In addition to H2B ubiquitylation, the CRL7^SMU1 complex might also directly regulate ubiquitylation of mitotic proteins. In support of this speculation, we found proteins, such as Adenomatous polyposis coli (APC) in the list of SMU1-associated proteins. Further studies are required to test if APC acts as a substrate of this E3 ligase complex. However, our work revealed that the LisH domain protein SMU1 associates with RNF40 to form a RING-type CRL. Interestingly, the list of SMU1-associated proteins also contains HECT type E3s, EDD and HUWE1. In fact, proteins, such as VPRBP and DDB1, were previously shown to associate with RING-type E3s as well as HECT-type E3s (Maddika and Chen, 2009; Nakagawa et al., 2013). Thus, in future studies it would be interesting to test if SMU1 participates in the assembly of both RING- and HECT-type E3 ligases to regulate different cellular processes by controlling distinct set of substrates.

Full-length SMU1, DDB1, CUL7, RNF40, RNF20, SMC1a, Roc1, VPRBP, Rab7, H2B wild type or K120R mutant were cloned into S-protein tag, FLAG tag, streptavidin-binding protein tag (SFB)-triple-tagged destination vectors using the Gateway cloning system (Invitrogen). H2B and SMU1 were cloned into GST and MBP destination vectors using the same system. SMU1 and DYRK2 were also cloned into Myc destination vectors, and SKP1 was cloned into a HA destination vector by using the Gateway cloning system. The point mutations for H2B were generated by PCR-based site-directed mutagenesis and cloned into SFB- and GST-tagged destination vectors. RNF40 plasmid was a kind gift from Dr Steven Johnsen (University Medical Center Göttingen, Germany). Myc-tagged RNF20 was kindly provided by Dr Jae Bum Kim (Seoul National University, South Korea). The FLAG H2B-Ub fusion construct was a kind gift from Dr Hengbin Wang (University of Alabama at Birmingham, AL). The cMyc-SMC1a plasmid was a gift from Michael Kastan (Addgene plasmid #32363) (Kim et al., 2002), Myc3-CUL7 was a gift from Yue Xiong (Addgene plasmid #20695) (Andrews et al., 2006) and GFP-H2B was a gift from Geoff Wahl (Addgene plasmid #11680) (Kanda et al., 1998).

Antibodies against SMU1 (Abgent #AT3965a; 1:1000); RNF40 (Sigma #R9029; 1:2000); CUL7 (Sigma #C1743; 1:2000); DDB1 (Bethyl #A300-462A; 1:5000); SMC1a (Abcam #ab133643; 1:1000); H2B (Millipore #07-371; 1:5000); histone H2B ubiquitylated at Lys120 (H2Bub; Cell Signaling Technology #5546; 1:1000); cyclin A (BD #611269; 1:1000); CDT1 (Bethyl #A300-786A; 1:1000); histone H3 phosphorylated at Ser10 (pH3; Cell Signaling Technology #9701L; for western blotting 1:1000, for immunofluorescence 1:200); Cul4a (Bethyl #A300-739A; 1:5000); RNF20 (Abcam #ab32629; 1:1000); Myc-tag (Santa Cruz #9E10; 1:1000); FLAG-tag (Sigma, #F3165; 1:10,000); HA (Bethyl, #A190-108A; 1:1000), actin (Sigma, #A5441; 1:10,000) and α-tubulin (Sigma, #T6074; for western blotting 1:5000, for immunofluorescence 1:200) were used in this study. HRP-/FITC-conjugated anti-mouse and anti-rabbit secondary antibodies were obtained from Jackson ImmunoResearch.

HEK-293T, HeLa and BOSC23 cell lines were used in this work. All cell lines were purchased from American Type Culture Collection, and were tested and authenticated by the cell bank using their standard short tandem repeats (STR)-based techniques. Cells were also continuously monitored by microscopy to maintain their original morphology and tested for mycoplasma contamination by using DAPI staining. HEK-293T or HeLa were transfected with various plasmids using PEI (Polysciences) according to the manufacturer's protocol. Briefly, the plasmid was mixed with PEI (1 mg/ml) at a ratio of 1:3 in serum-free RPMI medium. Then, the DNA-PEI mixture was incubated for 20 min at room temperature (RT) and the complexes were added to cells to allow the transfection of plasmid.

SMU1-associated proteins were isolated using tandem affinity purification as described before (Maddika and Chen, 2009). Briefly, HEK-293T cells expressing SFB-triple-tagged SMU1 were lysed with NETN lysis buffer (20 mM Tris-HCl at pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40) containing protease inhibitors [phenylmethylsulfonyl fluoride (PMSF), Pepstatin A and aprotinin] on ice for 20 min. The cell lysates were added to streptavidin–Sepharose beads (Amersham Biosciences) and incubated for 1.5 h at 4°C. Then beads were washed thrice with lysis buffer and the associated proteins were eluted using 2 mg/ml biotin (Sigma) for 1 h at 4°C. The eluates from the first step of purification were then incubated with S-protein–agarose beads (Novagen) for 1 h at 4°C. After clearing the unbound proteins by washing, the proteins associated with S-protein–agarose beads were eluted by boiling in SDS-loading buffer for 10 min at 95°C. Proteins of the eluted lysate were separated by SDS-PAGE. The associated proteins were identified by in-gel trypsin digestion followed by liquid chromatography–mass spectrometry (LC–MS)/MS analysis at the Taplin Mass Spectrometry Facility (Harvard University).

For immunoprecipitation assays, cells were lysed with NETN buffer. The whole-cell lysates obtained by centrifugation were incubated with 2 µg of specified antibody bound to protein G–Sepharose beads, or with streptavidin–sepharose beads (GE) for 1.5 h at 4°C. The immunocomplexes were then washed with NETN buffer three times and loaded onto an SDS polyacrylamide gel. Western blotting was carried out by following standard protocols. Proteins were separated by denaturing SDS–PAGE and then transferred onto polyvinylidene difluoride (PVDF) membrane. The membranes were blocked in 5% non-fat dried milk in Tris-buffered saline (TBS) and then incubated with the primary antibodies overnight at 4°C. Then, the blots were incubated with the corresponding secondary antibodies conjugated with HRP for 1 h at room temperature. Visualization was carried out by enhanced chemiluminescence detection (Thermo Fisher Scientific).

Histones were isolated according to the acid extraction protocol. Briefly, cells were lysed under hypotonic conditions and the intact nuclei were collected. Isolated nuclei were acid extracted using 0.2M H₂SO₄. Histones were precipitated using 33% TCA and residual TCA was removed by ice-cold acetone. The pellets were air dried and Milli-Q water was added to dissolve histones. Histone fractions were subjected to SDS-PAGE followed by western blotting with antibodies of interest.

Bacterially expressed GST and GST-SMU1 bound to glutathione-Sepharose beads were incubated with eluted MBP-H2B for 1 h at 4°C. Beads were washed and proteins eluted by boiling in 2×SDS Laemmli buffer and then separated by SDS-PAGE; the interactions were analysed by western blotting.

Control siRNA and pre validated SMU1 siRNA were purchased from Dharmacon (Catalogue no: J-021129-10). TRC lentiviral SMU1 shRNA (RHS4533-EG55234), DDB1 shRNA (RHS4533-EG1642), CUL7 shRNA (RHS4533-EG9820) and RNF40 shRNA (RHS4533-EG9810) were purchased from Dharmacon. Transfection was performed twice, 24 h apart, with 200 nM siRNA using Oligofectamine reagent in accordance with the manufacturer's protocol (Invitrogen). shRNAs were transfected transiently using PEI (Invitrogen) in BOSC23 packaging cells along with packaging vectors. 48 h post transfection, the viral medium was collected and added to the target cells along with polybrene (8 mg/ml). 48 or 72 h post infection, cells were collected and processed for various assays and immunoblotting was performed with the specific antibodies to check the efficiency of knockdown.

Cells grown on coverslips were fixed with 3% paraformaldehyde solution in phosphate-buffered saline (PBS) containing 50 mM sucrose for 15 min at room temperature. After permeabilization with 0.5% Triton X-100 buffer containing 20 mM HEPES at pH 7.4, 50 mM NaCl, 3 mM MgCl₂ and 300 mM sucrose for 5 min at room temperature, cells were incubated with a 1% BSA for blocking for 30 min at room temperature. After washing with PBS, cells were incubated with primary antibodies for 2–3 h at room temperature followed by three washes with 1×PBS for 5 min each. Then, cells were incubated with FITC- or Rhodamine-conjugated secondary antibodies for 60 min at room temperature followed by three washes with 1×PBS for 5 min each. Nuclei were counterstained with DAPI. After a final wash with PBS, coverslips were mounted with glycerine-containing paraphenylenediamine. Images were taken using a Zeiss confocal microscope (LSM Meta 510 or LSM 700).

HeLa cells transfected with the desired expression vectors and siRNA or shRNA were harvested, washed with phosphate-buffered saline and fixed with ice-cold 70% ethanol for at least 1 h. Cells were washed thrice in PBS and treated for 30 min at 37°C with RNase A (5 μg/ml) and propidium iodide (25 μg/ml), then analysed using a BD Accuri C6 flow cytometer.

HeLa cells were transfected with Control and SMU1 siRNA. 48 h post transfection 1×10⁵ cells were seeded in five different plates and cells counted based on the Trypan Blue dye exclusion test.

A Sephacryl 300 column (GE Healthcare) was used to separate the protein complexes in the range of 10–1500 kDa. The column was calibrated using four gel filtration markers thyroglobulin (669 kDa), ferritin (440 kDa), conalbumin (75 kDa) and ovalbumin (44 kDa). After calibration, lysates (0.8–1 ml, total protein concentration 1–2 mg/ml) were injected and allowed to pass through the column. 1 ml fractions were collected using a Bio-Rad 2110 fraction collector within the molecular weight range of interest and all 1 ml fractions were concentrated separately by using protein concentration columns to the final volume of 80 µl. All the fractions were subjected to SDS-PAGE followed by western blot staining of proteins of interest.

Total RNA was isolated using Trizol reagent (Invitrogen) as per manufacturer's instructions. 2 μg of total RNA was transcribed in the presence of anchored oligo dT using Superscript-III (Invitrogen) as per manufacturer's protocol. Quantitative PCR (qPCR) was then initiated using the SYBR Pre mix Ex Taq (Tli RNaseH plus) kit (Clontech Laboratories) in 7500 real-time (RT) PCR systems (Applied Biosystems) as per manufacturer's protocol. The threshold cycle (Ct) values for particular genes were normalized to GAPDH for each sample. Sequences for primers used for qRT-PCR analysis are included in Table S2.

4 μg mouse IgG (Bethyl Laboratories), 4 μg SMU1 (Abgent #AT3965a) or 4 µl of H2Bub (Cell Signaling Technology #5546; dilution 1:250) antibodies were used for chromatin immunoprecipitation (ChIP) assay. Briefly, HeLa cells were crosslinked by using 1% formaldehyde solution. Crosslinking was allowed to proceed for 10 min at room temperature and was then stopped by addition of glycine to a final concentration of 0.125 M. Cells were washed twice with ice-cold 1×PBS and lysed using cell lysis buffer (5 mM PIPES pH 8, 85 mM KCl, 0.5% NP-40 and protease inhibitors) followed by nuclei lysis buffer (50 mM Tris pH 8.1, 10 mM EDTA, 1% SDS and protease inhibitors). Isolated chromatin was sonicated using a Diagenode Bioruptor at medium power and checked for fragment size. After microcentrifugation, the supernatant was diluted with 1×IP dilution buffer and preclearing was performed by using Protein G beads. Then equal amount of antibody was added and incubated on a rotating platform for 12 to 16 h at 4°C. Next day, protein G beads were added to chromatin. Beads were washed two times with IP Dialysis buffer (2 mM EDTA and 50 mM Tris pH 8) and three times with IP wash buffer (100 mM Tris pH 8, 250 mM LiCl, 1% NP-40 and 1% Na deoxycholate) for 5 min at room temperature. Then beads were quickly washed with 1×TE and bound DNA fragments were eluted using elution buffer (50 mM NaHCO3 and 1% SDS) at 65°C. Input and elution products were kept for decrosslinking at 65°C overnight. Then, samples were treated with RNase A and proteinase K for 3 h at 37°C and DNA was extracted using PCI treatment. qRT-PCRwas then initiated using the SYBR Pre mix Ex Taq (Tli RNaseH plus) kit (Clontech Laboratories) in 7500 real-time PCR systems (Applied Biosystems) as per manufacturer's protocol. Sequences for primers used for ChIP analysis were included in Table S1.

Metaphase chromosome spreads were performed as described before (Maddika et al., 2009). Briefly, cells treated with colcemid for 4 h were collected, washed with PBS and treated with 75 mM KCl for 30 min at room temperature. The treated cells were then fixed in fresh solution of methanol:acetic acid (3:1) and dropped onto glass slides. Cells were allowed to air dry, stained with Giemsa’s solution (5%) and visualized under the microscope.

We thank the Centre for DNA Fingerprinting and Diagnostics (CDFD) core facility as well as Dr Rohit Joshi's lab for their assistance in confocal imaging. We thank Nanci Rani for technical assistance, T. S. Shaffiqu for providing assistance in gel filtration experiment, Sawant Suresh, Zaffer Ullah Zargar and Amit M. Karole for their critical inputs during gene expression and ChIP experiments, Dr Ashwin Dalal and Dr Usha Dutta for their inputs during mitotic spread experiments. We also thank all members of LCDCS for their suggestions and critical inputs at various stages of the project.