UNC3866

The HDAC inhibitor SAHA regulates CBX2 stability via a SUMO- triggered ubiquitin-mediated pathway in leukemia

Abstract
Polycomb group (PcG) proteins regulate transcription, playing a key role in stemness and differentiation. Deregulation of PcG members is known to be involved in cancer pathogenesis. Emerging evidence suggests that CBX2, a member of the PcG protein family, is overexpressed in several human tumors, correlating with lower overall survival. Unraveling the mechanisms regulating CBX2 expression may thus provide a promising new target for anticancer strategies. Here we show that the HDAC inhibitor SAHA regulates CBX2 stability via a SUMO-triggered ubiquitin-mediated pathway in leukemia. We identify CBX4 and RNF4 as the E3 SUMO and E3 ubiquitin ligase, respectively, and describe the specific molecular mechanism regulating CBX2 protein stability. Finally, we show that CBX2-depleted leukemic cells display impaired proliferation, underscoring its critical role in regulating leukemia cell tumorogenicity. Our results show that SAHA affects CBX2 stability, revealing a potential SAHA-mediated anti-leukemic activity though SUMO2/3 pathway.

Introduction
SUMOylation is a post-translational modification (PTM) that regulates target protein function, playing a critical role in cellular processes such as DNA damage response, cell cycle progression, apoptosis, and cellular stress response [1–3]. Small ubiquitin-like modifier (SUMO) proteins are involved in several cancers, including leukemia [4], functioning as either oncogenes or oncosuppressors in a cell context-dependent manner [5–7]. Leukemias are character- ized by bone marrow failure due to oncogenic mutations of hematopoietic stem cells (HSC) or blood precursor cells. HSC differentiation and self-renewal properties are tightly regulated by Polycomb group (PcG) proteins, a well- characterized family of transcriptional epigenetic regulators [8]. PcG proteins form two canonical complexes: Polycomb repressive complex 1 (PRC1), which mediates ubiquitination of H2A at lysine 119 (H2AK119ub), and Polycombrepressive complex 2 (PRC2), which trimethylates H3 at lysine 27 (H3K27me3) [9]. Non-canonical PRC1 com- plexes have also been described, and are emerging as reg- ulators of gene transcription [10]. Mechanistically, the hierarchical model of PcG-mediated gene silencing requires H3K27 trimethylation by PRC2 followed by binding of PRC1 via one of the five chromobox proteins (CBX2, 4, 6, 7, 8), which in turns triggers H2AK119ub, eventually leading to transcriptional repression [11, 12]. Unsurpris- ingly, as regulators of stem cell properties and blood cell differentiation, PcG proteins are involved in leukemia and other solid cancers [13–15].CBX proteins link the activity of PRC1 with PRC2, serving as critical regulators of PcG-mediating activity. While the functional role of some CBX proteins in cancer has been largely described [15–17], recent reports supportprimary blasts derived from AML patients untreated (−) or treated with 5 μM SAHA at 24 h. d Real-time qPCR analysis of CBX2 expression levels (relative to ctr) in three primary blasts derived from AML patients untreated or treated with 5 μM SAHA at 24 h. Results show CBX2 relative fold change in SAHA-treated blasts compared tountreated counterpart.

Error bars represent STD of three technical replicates. p-value is not significantthe specific role of CBX2 in human tumors. CBX2 is overexpressed in several human cancers. Genotran- scriptomic meta-analysis of CBX2 revealed its amplifica- tion and upregulation in breast, lung, colorectal, prostate, brain, and hematopoietic tumors compared to normal tissue highlighting its potential oncogenic role [18]. Increased CBX2 expression has also been correlated with lower overall survival, whereas CBX2 depletion negatively affects prostate tumor proliferation and progression [18, 19]. CBX2 may thus represent a promising new target for anticancer strategies, warranting a better understanding of the mechanisms regulating CBX2 stability and biological activity. To date, chromodomain inhibitors have been identified for CBX7 [20, 21], but no molecules inhibiting CBX2 have been described. Nevertheless, different chromatin-modulating drugs such as histone deacetylase inhibitors (HDACi) are reported to regulate CBX2 targets on chromatin, suggesting that HDACi might be used to indirectly modulate aberrant effects of CBX2 in cancer [22]. Furthermore, the well-known pan-HDACi SAHA was recently shown to alter the profile of the whole proteome, modulating several PTM pathways such as ubiquitination and acetylation [23]. However, the precise role of HDACi in regulating CBX2 remains to be elucidated.Here we describe a novel SAHA-mediated mechanism of CBX2 post-translational regulation. We found that CBX2 undergoes SAHA-induced SUMO2/3 modification and thatCBX2 SUMOylation promotes its ubiquitination and proteasome-dependent degradation. We also identified the specific molecular pathway and key players regulating CBX2 stability, demonstrating that CBX4 and RNF4 act as the E3 SUMO and E3 ubiquitin ligase, respectively. Additionally, CBX2-depleted leukemic cells display impaired proliferation, showing that CBX2 is required for leukemia cell clonogenicity. Our study provides the first evidence of a non-canonical SAHA-mediated anti- tumorigenic activity via CBX2 SUMOylation and degradation.

Results
HDACi regulate CBX2 targets on chromatin [22], sug- gesting that they might indirectly modulate CBX2 in leu- kemia. To investigate the effect of SAHA on CBX2 expression, we treated K562, U937 and HL-60 cells with SAHA (5 µM) at different times. Western blot analysis showed CBX2 downregulation in all cell lines tested in a time-dependent manner (Fig. 1a). qRT-PCR experiments showed that SAHA does not exert its effect transcriptionally (Fig. 1b), as previously described for many SAHA targetgenes [24], suggesting that SAHA acts via post-translational mechanisms. Similarly, CBX2 destabilization was also observed in SAHA-treated ex vivo primary AML blasts at protein (Fig. 1c) but not RNA level (Fig. 1d). To investigate the mechanisms underlying CBX2 destabilization, we per- formed western blot analysis of K562 and U937 cells treated with the proteasome inhibitor MG132 (Fig. 2a). Our results showed that SAHA promotes CBX2 downregulation via a proteasome-dependent pathway. Interestingly, in addition to CBX2 degradation, SAHA treatment increased endogenous expression of SUMO2/3 (but not SUMO1) and its conjugates in a time-dependent manner (Fig. 2b). We therefore speculated that CBX2 SUMOylation is respon- sible for SAHA-mediated CBX2 degradation.To study the involvement of SUMO2/3 pathway in CBX2 degradation after SAHA treatment, we knocked down endogenous SUMO2/3 in K562 cells. SUMO2/3- depleted K562 cells were treated with 5 µM SAHA for 6 and 24 h. Western blot analysis revealed that SUMO2/3 silencing limits SAHA-mediated CBX2 degradation compared to shSCR negative control (Fig. 2c). Moreover, treatment of leukemic cells with other epi-drugs, including also non-HDACi, showed that HDACi-mediated induction of SUMO 2/3 pathway seems to be a required step to promote CBX2 protein degradation (Supplementary Fig. 1a).

Together, these findings suggest that, in leukemic cells, SAHA promotes activation of SUMO2/3 modificationdown assaying CBX2 polyubiquitination and SUMOylation in HEK293-FT cells upon transfection of GFP/CBX2 alone or in com- bination with His/SUMO1-, His/SUMO2-, His/SUMO3- encoding plasmids (3 µg) in presence of 25 µM MG132. IP experiments were performed with anti-CBX2 antibody and immunoblotted with anti- ubiquitin and anti-HIS antibodies in HEK293-FT cells. e Endogenous CBX2 SUMOylation and polyubiquitination status in SAHA-treated K562 cells at indicated times. Total extracts were immunoprecipitated with anti-CBX2 antibody. Immunoblotting was performed with indi- cated antibodiespathway, which seems to play a functional role in SAHA- mediated CBX2 destabilization.To investigate whether SUMO-mediated CBX2 modifica- tions are involved in SAHA-induced CBX2 degradation, we first analyzed CBX2 sequence to identify putative SUMOylation sites. CBX2 sequence analysis using SUMOsp 2.0 software revealed that CBX2 is endowed with several potential SUMOylation sites, including K153, which lies in a canonical SUMO consensus sequence [25] (Supplementary Fig. 1b). To determine whether CBX2 is endogenously SUMOylated in leukemic cells, weperformed western blot analysis on U937 and K562 cell extracts. We observed three different forms of CBX2: the predicted form at 70 kDa and two slower migrating forms at about 90 and 130 kDa (Supplementary Fig. 2a). CBX2 knockdown led to a reduction in both slower migrating bands, confirming that they are specific forms of CBX2 (Supplementary Fig. 2b). Moreover, SUMO2/3 depletion in K562 cells confirmed that CBX2 undergoes SUMO2/3 covalent modifications (Supplementary Fig. 2c). Note- worthy, SUMO2/3 knockdown causes CBX2 stabilization. Taken together, these results indicate that CBX2 is endogenously SUMOylated by SUMO2/3 in leukemic cells. To corroborate that CBX2 undergoes SUMO2/3 modifica- tions, we performed IP experiments of endogenous CBX2and pull-down of His/SUMO2/3-associated proteins with Ni2+ beads. HEK293-FT cells were transiently transfected with His/SUMO2 expression plasmid with or without MG132 to overcome CBX2 destabilization.

IP followed byimmunoblotting analysis revealed that endogenous CBX2 undergoes polySUMOylation upon SUMO2 over- expression, and that its SUMOylation status increases with MG132 proteasome inhibitor treatment (Fig. 3a). Similarly,IP experiments upon SUMO2/3 depletion showed a reduction in CBX2-SUMO2/3 conjugates (Fig. 3b). Nickel pull-down assay also confirmed that CBX2 is SUMOylated by SUMO2 (Supplementary Fig. 3a).Together, these findings demonstrate that CBX2 is SUMOylated by SUMO2/3 in vivo.To investigate the molecular function of SUMO-mediated CBX2 modification, we transfected HEK293-FT cells with increasing amounts of YFP-SUMO1, 2 or 3. In line with our results, overexpression of either SUMO2 or SUMO3 but not SUMO1 reduced CBX2 expression in a dose-dependent manner (Fig. 3c). Since SUMOylation regulates gene expression and chromatin dynamics, we performed real- time qPCR experiments to investigate whether SUMO2 transcriptionally modifies CBX2 expression. Our results showed that SUMO2 alters CBX2 expression at post- translational level only, since no effect was observed on its mRNA (Supplementary Fig. 3b). SUMO2/3 conjugation and the ubiquitin-proteasome system cooperate to regulate the stability of a subset of SUMO2/3-conjugated proteins [26]. Thus, to determine whether the ubiquitin-proteasome system is involved in SUMO2/3-mediated CBX2 degrada- tion, we transfected HEK293-FT cells with increasing amounts of His/SUMO2 plasmid with or without MG132. Western blot analysis revealed that MG132 blocks SUMO- dependent CBX2 degradation, providing evidence that this phenomenon occurs through a ubiquitin-proteasome path- way (Supplementary Fig. 4a). To examine the possible interplay between SUMOylation and ubiquitination in reg- ulating CBX2 turnover, we performed GFP/Pull-down on exogenous GFP/CBX2 to assay its ubiquitination rates following SUMO1, SUMO2 and SUMO3 overexpression. Western blotting using anti-ubiquitin antibody showed that CBX2 was polyubiquitinated exclusively by SUMO2 and SUMO 3. Although SUMO1 seems to slightly modifyCBX2, its modification is not responsible for CBX2 poly- ubiquitination (Fig. 3d). Moreover, SUMO2-mediated ubi- quitination was also observed upon IP experiments of endogenous CBX2 (Supplementary Fig. 4b).Likewise, SUMO2/3 knockdown decreased CBX2 polyubiquitination compared to non-targeting shRNA- expressing cells (Supplementary Fig. 4c).These findings indicate that CBX2 SUMOylation pro- motes its ubiquitination and proteasome-mediated degra- dation.

As SAHA treatment promotes both CBX2 degradation and SUMO2/3 induction, we assayed CBX2 SUMOylation and ubiquitination status after SAHA treat- ment. K562 cells were treated with SAHA (5 µM) at 1 and 3 h instead of 6 and 24 h to avoid degradation of CBX2 and its SUMOylated forms. IP of endogenous CBX2 after SAHA treatment followed by immunoblotting with anti- SUMO2/3 and anti-ubiquitin antibodies, respectively, showed that SAHA promotes CBX2 polySUMOylation and polyubiquitination (Fig. 3e).These results provide evidence that SAHA affects CBX2 expression by inducing SUMO2/3 pathway. CBX2 SUMOylation by SUMO2 and SUMO3 causes its poly- ubiquitination and proteasome-dependent degradation, revealing a novel and non-canonical mechanism by which SAHA regulates CBX2 expression.SUMOylation commonly occurs on specific lysine residues falling in the so-called SUMO-acceptor site ΨKxE (where Ψ is an aliphatic branched amino acid and x is any amino acid), although many proteins are SUMOylated throughnon-canonical lysine residues [27]. CBX2 sequence analy- sis using SUMOsp 2.0 software identified several canonical and non-canonical lysine (K) residues as potential SUMOylation sites, with K153 lying in a canonical SUMO consensus site. To identify CBX2-SUMO2/3 acceptor site (s), we replaced, by point mutagenesis, several K residues with an arginine (R), inserting CBX2 K/R mt in GFP- tagged vectors. wt and mt GFP/CBX2 were transfected with His/SUMO2 in HEK293-FT cells. GFP pull-down followed by western blot using anti-SUMO2/3 antibody identified three K residues, K60, K153, and K410, which partially impair CBX2 SUMOylation when replaced by R (Supple- mentary Fig. 5a). We therefore constructed a triple CBX2- 3K/R mt in which all three lysines were replaced. MG132- treated HEK293-FT cells were co-transfected with SUMO2/3 and either wt CBX2- or CBX2-3K/R-expressing plas- mids. IP assays followed by western blot with anti-SUMO2/3 and anti-ubiquitin antibodies showed that CBX2 SUMOylation was completely abolished in CBX2-3K/RK562 stably expressing shRNF4 construct #1 and shSCR control treated with 5 µM SAHA at 3 cells Total extracts were immunopre- cipitated with anti-CBX2 antibody. CBX2 polyubiquitination rate in SAHA-treated shRNF4#1-transduced K562 cells compared to shSCR control are shown. e Western blot analysis of endogenous CBX2 in K562 cells stably expressing two different shRNA construct targeting shRNF4 (#1 and #2) or shSCR control. ERK1/2 was used as loading control(Fig. 4a).

As expected, GFP pull-down and western blot of CBX2-3K/R showed impairment of its polyubiquitination (Fig. 4a). Accordingly, 3 K/R mutant also showed lower SUMOylation and ubiquitination rate upon SAHA treat- ment (Fig. 4b).To evaluate the effect of SUMO2/3 on CBX2-3K/R stability, we measured the turnover rate of wt and triple mt CBX2 upon SUMO2/3 overexpression using CHX. HEK293-FT cells were transfected with GFP-tagged wt CBX2 and CBX2-3K/R with or without a fixed amount of SUMO2/3. After transfection (24 h), cells were treated with CHX at different times. While the protein half-life of wtCBX2 was around 9 h, CBX2-3K/R was much more stable. Furthermore, SUMO2/3 overexpression almost completely eliminated wt CBX2 expression after around 3 h, while CBX2-3K/R retained its stability (Fig. 4c). Next, we tested the effect of SAHA treatment on CBX2-3K/R. K562 cells were nucleofected with GFP-tagged wt and triple mt CBX2. After nucleofection (24 h), cells were treated with SAHA at the indicated times. Western blot analysis followed by immunoblotting with anti-GFP antibody showed that SAHA robustly promotes degradation of wt CBX2, but has a much reduced effect on CBX2-3K/R (Supplementary Fig. 5b). Furthermore, wt and CBX2-3K/R half-life was alsomeasured upon SAHA treatment using CHX. HEK293-FT cells were transfected with CBX2 wt or 3 K/R mutant and then treated with 5 µM SAHA. 24 h upon SAHA treatment, CBX2 wt and 3 K/R mutant protein decay were assayed adding CHX at different time points. Results showed that SAHA strongly reduces CBX2 wt protein expression. Conversely, 3 K/R mutant is much more stable showing a differential rate of decay compared to CBX2 wt protein (Fig. 4d).These findings indicate that K60, K153 and K410 are key residues for SAHA-mediated CBX2 degradation.Taken together, our data demonstrate that these three lysines are responsible for CBX2 SUMOylation by SUMO2/3, governing its stability.Human RNF4 protein is one of the most extensively studied SUMO-targeted ubiquitin ligases. To determine whether RNF4 is involved in CBX2 degradation, we transfected HEK293-FT cells with increasing amounts of FLAG-tagged RNF4 and its mutant form, RNF4-CS, lacking E3 ubiquitin ligase activity.

Western blot analysis revealed that RNF4 mediates CBX2 degradation, while RNF4-CS does not affect CBX2 stability, suggesting that RNF4 E3 ubiquitin ligase activity is required to promote CBX2 degradation (Fig. 5a).To verify whether RNF4 and CBX2 interact, we immunoprecipitated CBX2 following FLAG-tagged RNF4 overexpression. Our results show that CBX2 physically interacts with RNF4 (Fig. 5b).We then investigated the involvement of RNF4 in SUMO-triggered CBX2 polyubiquitination. HEK293-FT cells were transfected with His/SUMO2 and FLAG-tagged RNF4 plasmids, and treated with MG132.Following CBX2 IP, immunoblotting with anti-ubiquitin antibody showed that RNF4 promotes CBX2 poly- ubiquitination. Moreover, RNF4 and SUMO2 co- expression synergistically caused a strong increase in CBX2 polyubiquitinated forms compared to either SUMO2 or RNF4 alone (Fig. 5c). To investigate the role of RNF4 in SAHA-mediated CBX2 polyubiquitination HEK293-FT cells were cotransfected with GFP/CBX2, FLAG-tagged RNF4 or RNF4-CS mutant plasmids and treated with SAHA and MG132. GFP/Pull-down assay of CBX2 fol- lowed by immunoblotting with anti-Ubiquitin showed that RNF4, but not its catalytic mutant, promotes CBX2 ubi- quitination upon SAHA treatment (Supplementary Fig. 6a). In parallel, SAHA-treated K562 cells knocked down for RNF4 showed an impairment of CBX2 polyubiquitination compared to SCR control cells (Fig. 5d). Accordingly, silencing of RNF4 promote CBX2 protein stabilization (Fig. 5e).Taken together, these results indicate that RNF4 acts as the ubiquitin ligase involved in SAHA-mediated CBX2 ubiquitination and degradation of SUMOylated CBX2.To identify novel CBX2 interactors, we performed MS- based quantitative interaction proteomics. We generated a transgenic inducible Tet-On HeLa stable cell line over- expressing N-terminal GFP-tagged CBX2. After doxycy- cline induction (24 h), nuclear extract was obtained from transgenic (Tet-On inducible GFP/CBX2) and wt HeLa cells. A single step GFP-affinity enrichment (GFP-Trap) of lysates was performed in triplicate. Affinity purification of GFP/CBX2 followed by label-free LC-MS/MS analysis revealed the interaction of CBX2 with several proteins including CBX4 (Fig. 6a and Supplementary Table 1). Human CBX4 is the only member of the canonical PRC1 complex that possesses SUMO E3 ligase activity [28].

A limited number of substrates for CBX4 are reported [29– 31]. To validate LC-MS/MS results, we performed IP experiments in HEK293-FT cells after transfection of GFP- tagged CBX2 alone or with FLAG-tagged CBX4-encoding vector. IP experiments confirmed that CBX2 interacts with CBX4 (Fig. 6b). To investigate whether CBX4 has a functional role in CBX2 SUMO2/3-mediated modifications, IP experiments were conducted. HEK293-FT cells were transfected with GFP-tagged CBX2 alone, with His-tagged SUMO2, or with both SUMO2 and FLAG-tagged CBX4 plasmids. Our results show that CBX4 overexpression increases CBX2 SUMOylation, highlighting that CBX4 is required for SUMO moiety catalyzation (Fig. 6c). To test the effect of CBX4 overexpression on CBX2 stability, we transfected HEK293-FT cells with a fixed quantity of SUMO2 and SUMO3 either alone or together to increasing amounts of CBX4. Western blot analysis showed that, in presence of either SUMO2 or SUMO3, CBX4 promotes CBX2 polySUMOylation and, subsequently, its degradation in a dose-dependent manner (Fig. 6d). Our results identify CBX4 as the E3 SUMO ligase of CBX2, highlighting its involvement in SUMO2/3-dependent CBX2 degradation.The involvement of CBX2 in maintaining hematopoietic stem and progenitor self-renewal was recently described [32]. To investigate the functional role of CBX2 in hema- topoietic malignancies, we analyzed the effect of CBX2 depletion in leukemic cells. To mimic SAHA-associated CBX2 downregulation, U937 and K562 cells were trans- duced with lentivirus expressing non-targeting shRNA (shSCR) and shCBX2. Following CBX2 knockdown, weexamined colony-formation unit (CFU) efficiency to eval- uate tumorigenic properties and self-renewal (clonogenic) capability of leukemic cells. CBX2 depletion resulted in a marked reduction in CFU efficiency (Fig. 7a, b).

Silencing of CBX2 also affected cell viability and proliferation of U937 and K562 cells (Fig. 7c). We also assayed the effect on leukemic cell viability and proliferation of 3 K/R mutant following SAHA treatment by MTT assay. CBX2 and 3K/R mutant were overexpressed in K562 cells and then treated (or left untreated) with SAHA for 24 h. SAHA-mediated impairment of K562 cell viability was significantly reduced by 3K/R mutant, but not by CBX2wt, highlighting theFLAG/CBX4 in HEK293-FT cells. c GFP pull-down and immunoblot of CBX2-SUMO2/3 conjugates upon His/SUMO2 and FLAG/CBX4 overexpression. d Western blot of endogenous CBX2 in HEK293-FT cells upon overexpression of a fixed amount (0.5 µg) of YFP/SUMO2 or YFP/SUMO3 alone or together to increasing amounts (1 and 2 µg) of FLAG/CBX4. Upper panel: CBX2 high exposure. Poly-S is referred to polySUMOylated forms and mono-S to monoSUMOylated form of CBX2, respectivelypossibility that reduced sensitivity of 3K/R mutant to SAHA-mediated degradation counteracts the impairment of cell proliferation by SAHA (Supplementary Fig. 7).In addition, we performed Giemsa staining and Immu- nophenotypic assays to verify whether silencing of CBX2 affects cell differentiation. shCBX2-transduced U937 cells display morphological changes in cell size, more distinct nuclei and nucleoli, and polarized cytoplasm compared to SCR (Fig. 7d). Moreover, cellular margins show cyto- plasmic protrusion and filaments, a phenotype reminiscent of monocyte/macrophage differentiated cells. In addition, FACS analysis of CBX2 knocked down cells showed anGiemsa Digital images acquired in color brightfield microscopy by Cytation 5 Cell Imaging Multi-Mode Reader. Arrows indicate polar- ized cytoplasm, cytoplasmatic protrusions and filaments. Magnifica- tions ×40. e FACS analysis of CD11b expression U937 CBX2- depleted cells. Error bars represent STD three independent experi- ments conducted in triplicate (*P < 0.05)increase of CD11b differentiation marker (Fig. 7e). These data indicate that CBX2 silencing causes a reduction of leukemic cell tumorigenicity as revealed by cell prolifera- tion impairment and higher leukemic cell differentiation potential.To investigate the potential interplay of CBX2 down- regulation and SAHA-mediated anti proliferative effects, we analyzed microarray gene expression profile data of U937 and K562 SAHA-treated cells [33] to identify com- mon targets in SAHA-treated cells (Supplementary Fig. 8a). Gene Ontology analysis of common deregulated genes between K562 and U937 cells was performed (Supple- mentary Table 2). Since CBX2 silencing affects leukemic cell proliferation, we focused on genes related to SAHA- altered cell proliferation biological processes. In particular, we selected genes differentially regulated by SAHA (up- and down-regulated in both cell lines; Supplementary Fig. 8b and c) including CDKN1A, a well-known in cis CBX2 target [32] and analyzed by qRT-PCR their expression upon CBX2 depletion. The selected SAHA-deregulated geneswere also similarly regulated upon CBX2 silencing (Sup- plementary Fig. 9).Collectively, these findings reveal a potential role for CBX2 in sustaining tumorigenic properties of leukemic cells, highlighting its possible oncogenic role in hematolo- gical malignancies. Moreover, data suggest that SAHA can also exert its anti-proliferative effects reducing CBX2 pro- tein levels and consequently its transcriptional activity. Discussion Several PcG proteins are deregulated in cancer [15] and their dysfunction is responsible for proliferation, inhibition of apoptosis, and increase in cancer stem cell population [34–36]. The critical interplay of PcG members in deter- mining cellular behavior gives rise to a dynamic regulation of their biological function. However, the molecular mechanisms underlying the finely tuned control of PcG activity are poorly understood. PTMs are an effective way to regulate and possibly modify biological activities and properties. The increasing number of studies demonstrating that several PcG proteins are targets for ubiquitination, SUMOylation and phosphorylation [30, 37, 38] are pro- viding a better understanding of PcG functional regulation by post-translational mechanisms. For example, CBX2 is phosphorylated at Ser42 residue and its phosphorylation changes the binding specificity of CBX2 for methylated H3 [39]. However, no other PTMs regulating CBX2 biological activity have as yet been reported. Identifying novel PcG PTMs could be an important step forward that may help to better understand the regulatory mechanisms of PcG protein activity. Here we provide the first evidence that the pan- HDACi SAHA induces CBX2 SUMOylation and degra- dation, the latter resulting in cell proliferation arrest and loss of self-renewal in leukemia cell lines. Our identification of a novel mechanism by which SAHA directly regulates CBX2 stability via a SUMO- triggered ubiquitin-mediated pathway may have important implications. Besides their well-established effects on HDACs, HDACi seem to have a much broader and more complex range of action. The destabilization of CBX2 may contribute to chromatin reset at specific loci. Since these loci are known, the use of targeted strategies might apply to leukemias where chromatin alterations are reported [22]. In agreement, CBX2 is potentially a tumorigenic target in hematological malignancies, since it regulates stemness and self-renewal of leukemic cells. Given that SAHA targets CBX2 destabilization, its anti- tumorigenic activity might be attributed not only to reg- ulation of the histone and non-histone acetylome, but also to PTM pathway modulation. Interestingly, our findings indi- cate that SAHA acts specifically on SUMO2/3 and not SUMO1 modification pathway. As SUMO conjugations can be induced by cell stress responses [40], and SAHA is reported to induce cell stress [41], it is tempting to speculate that SAHA regulation of PTM pathways might be linked to cell stress response. However, this hypothesis requires further investigation. Similarly, as other SUMO2/3 targets such as PML- RARα [5] and NR4A1 [42] have been described, the effects exerted by SAHA on CBX2 might extend to a plethora of targets, likely explaining the broad effects reported on proteome modulation. Although SUMOylation of other PcG proteins has been described [30, 37, 38], here we clearly demonstrate that CBX2 SUMOylation is responsible for its degradation. Consequently, several intriguing hypotheses might be advanced: i) SUMO-mediated CBX2 degradation may lead to a rearrangement of the canonical PRC1 complex, affecting selective chromatin bindings; ii) PcG proteins may be able to inter-regulate their biological activity (and cell fate) via CBX4 E3 ligase action; iii) since CBX2 is the only human CBX family member able to induce chromatin compaction [43], it may exert a crucial function in PcG- mediated transcriptional repression. Therefore, SAHA- mediated CBX2 degradation suggests that HDACi induce an open chromatin conformation status not only by mod- ulating global acetylation levels, but also by regulating key proteins involved in chromatin compaction. Targeting ubiquitin ligases such as RNF4 and/or SUMO ligases such as the newly identified CBX4 may represent a novel way through which PcG members can be regulated. This could potentially represent an attractive alternative therapeutic strategy to affect and ‘drug’ chromatin UNC3866 complexes.