CoREST family of transcriptional co-repressors regulates gene expression and cell fate determination during development. CoREST co-repressors recruit with different affinity the histone demethylase LSD1 (KDM1A) and the deacetylases HDAC1/2 to repress with variable strength the expression of target genes. CoREST protein levels are differentially regulated during cell fate determination and in mature tissues. However, regulatory mechanisms of CoREST co-repressors at the protein level have not been studied. Here, we report that CoREST (CoREST1, RCOR1) and its homologs CoREST2 (RCOR2) and CoREST3 (RCOR3) interact with PIASγ (protein inhibitor of activated STAT), a SUMO (small ubiquitin-like modifier)-E3-ligase. PIASγ increases the stability of CoREST proteins and facilitates their SUMOylation by SUMO-2. Interestingly, the SUMO-conjugating enzyme, Ubc9 also facilitates the SUMOylation of CoREST proteins. However, it does not change their protein levels. Specificity was shown using the null enzymatic form of PIASγ (PIASγ-C342A) and the SUMO protease SENP-1, which reversed SUMOylation and the increment of CoREST protein levels induced by PIASγ. The major SUMO acceptor lysines are different and are localized in nonconserved sequences among CoREST proteins. SUMOylation-deficient CoREST1 and CoREST3 mutants maintain a similar interaction profile with LSD1 and HDAC1/2, and consequently maintain similar repressor capacity compared with wild-type counterparts. In conclusion, CoREST co-repressors form protein complexes with PIASγ, which acts both as SUMO E3-ligase and as a protein stabilizer for CoREST proteins. This novel regulation of CoREST by PIASγ interaction and SUMOylation may serve to control cell fate determination during development.

Introduction

CoREST family of transcriptional co-repressors that include CoREST1, CoREST2, and CoREST3 has been functionally associated with cell fate programs [16]. The first member described was CoREST1, initially named CoREST for its role as a co-repressor of the transcription factor REST/NRSF [repressor element 1 (RE1)-silencing transcription factor/neuron-restrictive silencer factor] [7]. CoREST1 through its interaction with REST/NRSF silences the expression of neuronal genes in non-neuronal cells and neuronal stem cells [8,9]. Moreover, CoREST1 is necessary for the correct development of cortical pyramidal neurons [10]. Out of the central nervous system, CoREST1 is the repressive partner of Gfi-1 (growth factor independence-1) and Gfi-1b transcription factors to regulate hematopoietic differentiation [1,4]. Several years after CoREST1 description, it was shown that CoREST2 (RCOR2) maintains pluripotency of embryonic stem cells, a role not shared with CoREST1 [11]. Recently, it was shown that CoREST2 is essential for early events during cortical neurogenesis [6], with some functions that could be shared with CoREST1 during cortex development [5]. On the other hand, CoREST2 seems to play a redundant role to CoREST1 during blood cell fate differentiation [2]. Specific functions for CoREST3 (RCOR3) have not been described, so far. Interestingly, CoREST3 is rich in splicing variants, 4 in human, and 5 in mice genomes [12]. The shortest splice variant of CoREST3 opposes CoREST1- and CoREST2-induced blood cell differentiation [2], suggesting a dominant-negative role.

Direct targeting of each CoREST to DNA showed that they have different transcriptional repressive capacity. CoREST1 is a more efficient repressor compared with CoREST2 and CoREST3 [13]. All CoRESTs share a similar structure that stands out for three highly conserved domains: one ELM2 (Egl-27 and MTA homology 2) and two SANT (SWI/SNF, ADA, NCoR, and TFIIIB) domains [13]. Through the arrangement of the ELM2 domain contiguous to the first SANT domain, CoRESTs recruit HDAC1 and HDAC2. However, we showed that CoREST2 has a weaker capacity to bind HDAC1/2 and consequently lower associated deacetylase activity [13]. Through the second SANT domain, all CoRESTs recruit the histone demethylase LSD1 (also known as KDM1A) [8,1316]. Recent data showed that the second SANT domain of CoREST1 and CoREST3 also allows the direct binding to nucleosomal DNA, and the detaching of the tail of histone 3 to make it accessible for HDAC1/2 and LSD1 action [17]. The associated enzymatic activities of LSD1–CoREST–HDAC1/2 (LCH) complexes permit modifying post-translationally the tails of histones to establish a repressive chromatin [15,16]. Therefore, the distinct repressive capacity of each CoREST is related to a differential capacity to recruit LSD1 and HDAC1/2 and to modulate their enzymatic activity.

All CoRESTs are co-expressed in the same cells during development and adulthood [12,13], raising several questions like how their function is specifically regulated during development fate programs and the role they play in mature cells. In this regard, we have shown that protein levels of CoREST are differentially regulated during neuronal maturation [12]. While CoREST1 and CoREST2 mRNA and protein decrease significantly during cortical neuron maturation, CoREST3 does not change [12], allowing one to hypothesize that a way of controlling the function of CoRESTs is by regulating their protein levels.

SUMOylation is a posttranslational modification that regulates transcription, half-life, and subcellular location of transcription factors and co-regulators, among several other processes in the cell (reviewed in ref. [18]). The SUMOylation process consists in the covalent linking of a small ubiquitin-like modifier (SUMO) peptide to target proteins. The SUMO acceptor lysines (K) often are in the consensus motif ΨKXE [19], in which Ψ is a large hydrophobic residue, X is any aminoacid, and E is glutamate. In mammals, there are three functional SUMO peptides: SUMO-1, and the almost identical peptides SUMO-2 and SUMO-3 (97% of amino acid identity [18]). SUMO peptides are conjugated to a target lysine by a three-step enzymatic process. In the first step, the SAE1/SAE2 (SUMO-activating enzyme subunits 1/2) dimer induces SUMO peptide activation. Then, the unique Ubc9 enzyme conjugates SUMO peptides to all substrates. In the third step, E3-ligases facilitate the SUMOylation process, giving substrate specificity and regulation [18]. PIAS (protein inhibitor of activated STAT) family of proteins composed of four members: PIAS1, PIAS2 (PIASxα, PIASxβ), PIAS3, and PIASγ (PIAS4), is a prominent group of SUMO E3-ligases involved in gene expression regulation [20]. SUMOylation is a reversible process, in which the removal of SUMO peptides from SUMOylated proteins is carried out by SUMO proteases that, in mammals, include the SENP (SUMO/Sentrin-specific protease) family of proteins (SENP1–2–3–5–6–7), also important for SUMO peptide maturation [21].

Previously, it was shown that SUMO-1 is conjugated to CoREST1 [22]. However, whether CoREST1 and the other CoRESTs are targets of SUMO-2 is unknown. In the present study, we show that all CoRESTs are targets of SUMO-2, which is conjugated by the SUMO E3-ligase PIASγ and by the SUMO-conjugating enzyme Ubc9. Lysine acceptors for SUMO-2 peptide in CoREST proteins are different and lie in nonconserved domains. Replacement of SUMO acceptor lysines by arginines in CoREST1 and CoREST3 did not change their capacity to interact with HDAC1/2 and LSD1, and consequently did not modify their transcriptional repressive function. PIASγ, but not Ubc9, increased CoREST protein levels. We conclude that PIASγ regulates CoREST stability and facilitates their SUMOylation.

Experimental procedures

Cell culture and reporter assays

HEK293-T cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM, from Gibco) supplemented with 10% fetal bovine serum (FBS, from Gibco) plus 1% penicillin/streptomycin and maintained at 37°C in an atmosphere of 5% CO2. Transcriptional repressor activity of CoRESTs and mutants was quantified using a reporter gene assay as described previously [13].

Plasmids and transfection

The following recombinant plasmids were already described: Gal4-CoREST1 [7]; Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, Gal4-CoREST2, and Gal4-CoREST3, G5S4tkLuc [13]; pCDNA-PIASγ, HA (hemagglutinin)-PIASγ, and pCDNA3.1-PIASγC342A [23]; pBTM-CoREST1 [24]. Site-directed mutagenesis of Myc-CoREST1(K297R), Myc-CoREST1(K148R), Myc-CoREST2(K60R), Myc-CoREST2(K88R), Myc-CoREST2(K223R), Myc-CoREST2(K240R), Myc-CoREST3(K98R), Myc-CoREST3(K232R), Myc-CoREST3(K236R), Gal4-CoREST1(K297R), and Gal4-CoREST3(K236R) was performed by overlapping PCR using primers encoding the point mutation followed by DpnI treatment to destroy parental DNA [25]. Cloning of Gal4-CoREST1(K297R) and Gal4-CoREST3(K236R) was carried out by PCR using Myc-CoREST1(K297R) and Myc-CoREST3(K236R) as templates, respectively. All recombinant plasmids were sequenced throughout to prove that the specific mutagenesis was successful, and no other mutations were introduced.

6xHis-SUMO-2 was kindly donated by Dr Ron Hay (Center for Biomolecular Sciences, U.K.); pcDNA-PIASγ was kindly donated by Dr Fletcher White (Indiana University School of Medicine, Indianapolis, U.S.A.); and pCDNA3.1-Ubc9 was kindly donated by Dr Eduardo Arzt (Universidad de Buenos Aires, Argentina). Expression plasmids for SENP1 and SENP1-dominant-negative (SENP1-DN) were kindly donated by Dr Peter O'Hare (Marie Curie Research Institute, U.K.). All transfections were carried out with Lipofectamine 2000® (Invitrogen) according to the manufacturer's instructions. The DNA : Lipofectamine 2000® ratio was 1 : 3 (1 µg of DNA for 3 µl of Lipofectamine 2000®) and, after appropriate incubation time, was added on the cells. All experiments were carried out 48 h after transfection.

Antibodies

Anti-Myc (Abcam, ab9106), anti-Myc (Santa Cruz Biotechnology, sc-40), β-actin (Sigma, A2228), anti-CoREST1 (Neuromab, 75-039), anti-CoREST2 (Sigma, HPA021638); anti-CoREST3 (Abcam, ab76921), anti-HA (Covance, MMS-101P), anti-HA (Santa Cruz Biotechnology, sc-805), anti-HDAC1 (Abcam, ab46985), anti- LSD1 (Abcam, ab17721); anti-PIASγ (Santa Cruz Biotechnology, sc-30875); anti-SUMO-2 (Sigma, S9571).

Yeast two-hybrid assays

Interaction assays were performed using the L-40 yeast strain co-transformed with pBMT-hCoREST1 [24] and pGAD-hPIASγ [26]. A positive interaction was identified by growth on selective medium lacking histidine and qualitatively confirmed by assaying for β-galactosidase activity, essentially as previously described [7,24].

Protein extraction, co-immunoprecipitation, and western blot

HEK293-T cells were washed and scraped out in ice-cold phosphate-buffered saline. Total protein extracts were obtained by passing cells through a 1 ml syringe in RIPA buffer plus protease inhibitor cocktail and N-ethylmaleimide for SUMOylation assays. Homogenates were centrifuged at 19 500 × g for 20 min at 4°C and supernatants were saved for further analysis. Immunoprecipitation was performed essentially as we have described [13]. Protein samples were resolved on SDS–PAGE and proteins were detected by western blots. Quantification of specific bands was performed with the ImageJ software as described previously [24] using β-actin as a loading control.

Nickel–Nitrilotriacetic acid affinity pull-down assays

Ni-NTA (Nickel–Nitrilotriacetic acid) pull-down assays were performed essentially as described in ref. [27]. HEK293-T cells were transfected with empty vectors (mock) or PIASγ plus 6xHis-SUMO-2 expression vectors. Twenty-four hours after transfection, cells were lysed in a buffer containing 10 mM Tris (pH 8.0), 0.2 M Na2HPO4/NaH2PO4 (pH 8.0), 6 M guanidine hydrochloride, 5 mM β-mercaptoethanol, and 5 mM imidazole. Then, lysates were sonicated during 30 s on ice and centrifuged at 3 000 × g during 15 min. The supernatants were incubated during 3 h at room temperature with 50 µl of packed Ni-NTA agarose beads. Beads were washed twice with lysis buffer containing 0.1% Triton X-100, once with 0.2 M Na2HPO4/NaH2PO4 (pH 8.0), and then with 0.2 M Na2HPO4/NaH2PO4 (pH 6.0). Finally, beads were boiled in 1× Laemmli sample buffer and the soluble material was loaded on SDS–PAGE for western blot assays.

Animals

Two adult female Sprague–Dawley rats were obtained from the animal care facility of the Faculty of Biological Sciences of Pontificia Universidad Católica de Chile. Rats were anesthetized with isoflurane and immediately decapitated. Brains were dissected out, cells were obtained by treatment with 0.05% trypsin and suspended in hypotonic buffer, and then nuclear extracts were obtained by the Dignam fractionation method [28]. One milligrams of brain nuclear extracts were dialyzed against IP buffer [50 mM Tris (pH 7.9), 10% (v/v) glycerol, 150 mM NaCl, 0.5 mM EDTA, 0.5% (v/v) Triton X-100, 0.5 mM PMSF, and 0.5 mM DTT] during 4 h at 4°C. Then, dialyzed proteins were pre-cleared with 0.75 µg of rabbit IgG in IP buffer, and then incubated with 1 µg of anti-PIASγ antibody (Santa Cruz Biotech, sc50348) or anti-CoREST1 (Upstate 07-755), or 1 µg of rabbit IgG overnight at 4°C. Immunocomplexes bound to protein A/G were extensively washed, and fractionated in SDS–PAGE. Western blots were revealed with anti-CoREST1 (NeuroMab 75-039) and anti-PIASγ antibody (Santa Cruz Biotech, sc50348).

The procedure was conducted to reduce the number of animals and their level of pain and discomfort. The experimental protocols were approved by the Bioethical Committee of the Pontificia Universidad Católica de Chile and the Bioethical Committee of the Comisión Nacional de Ciencia y Tecnología de Chile (CONICYT).

Protein half-life measurement

HEK293-T cells were transfected with empty vectors or vectors expressing recombinant PIASγ and His-SUMO-2 using Lipofectamine 2000 (Invitrogen, #52887). Twenty-four hours after transfection, cells were treated with 100 µg/ml cycloheximide (Sigma–Aldrich, #C4859) and harvested to extract proteins with RIPA buffer at 0, 2, 4, and 8 h. Western blots were performed measuring GAPDH as a loading control and densitometry analyses were performed using ImageJ.

Results

CoREST proteins interact with PIASγ

Our and others' previous data showed that CoREST1 is part of a transcriptional complex with the nuclear receptor Nurr1 and PIASγ [23,26,29]. These data prompted us to test whether the CoREST family of transcriptional co-repressors are regulated by PIASγ. To test this hypothesis, we performed a two-hybrid assay with yeast transformed with pBTM-CoREST1 and pGAD-PIASγ or control empty vectors. Only yeast co-transformed with pBTM-CoREST1 and pGAD-PIASγ expressed the reporter β-galactosidase and grew in triple selective media, indicating the interaction between CoREST1 and PIASγ (Supplementary Figure S1). This interaction was confirmed by co-immunoprecipitation assays in HEK293-T cells overexpressing Myc-CoREST1 and HA-PIASγ (Figure 1A,B). Immunoprecipitation with anti-Myc antibody brought HA-PIASγ into the same immunocomplex (Figure 1A), and, in the reciprocal way, immunoprecipitated HA-PIASγ showed a strong interaction with Myc-CoREST1 (Figure 1B). Strenthening the idea that PIASγ, which is higly abundant in brain tissue (Figure 2A), forms complexes with CoREST transcriptional co-repressors, we showed that immunoprecipitation of endogenous PIASγ from adult rat brain extracts precipitated significant amounts of endogenous CoREST1 (Figure 2B) and, reciprocally, the immunoprecipitation of CoREST1 precipitated PIASγ from brain extracts (Figure 2C). We extended the present study to CoREST2 and CoREST3 homologs. We used the longest splice variant of CoREST3 (CoREST3a) [13] for all assays. Myc-CoREST2 and Myc-CoREST3 also precipitated recombinant HA-PIASγ (Figure 1A), revealing that PIASγ can form protein complexes with each member of the CoREST family.

CoREST proteins interact with the SUMO E3-ligase PIASγ.

Figure 1.
CoREST proteins interact with the SUMO E3-ligase PIASγ.

(A) Whole extracts of HEK293-T cells co-transfected with Myc-CoREST1, Myc-CoREST2, or Myc-CoREST3 and HA-PIASγ were immunoprecipitated with anti-Myc antibody (A) or anti-HA (B). Protein complexes were fractionated in SDS–PAGE and western blots probed with anti-HA antibody to detect HA-PIASγ (A) or anti-Myc antibody to detect Myc-CoRESTs. Protein extracts from cells transfected with empty plasmid (Mock) and immunoprecipitation with preimmune serum (IgG) were used as negative controls.

Figure 1.
CoREST proteins interact with the SUMO E3-ligase PIASγ.

(A) Whole extracts of HEK293-T cells co-transfected with Myc-CoREST1, Myc-CoREST2, or Myc-CoREST3 and HA-PIASγ were immunoprecipitated with anti-Myc antibody (A) or anti-HA (B). Protein complexes were fractionated in SDS–PAGE and western blots probed with anti-HA antibody to detect HA-PIASγ (A) or anti-Myc antibody to detect Myc-CoRESTs. Protein extracts from cells transfected with empty plasmid (Mock) and immunoprecipitation with preimmune serum (IgG) were used as negative controls.

PIASγ and CoREST1 interact in the adult rat brain.

Figure 2.
PIASγ and CoREST1 interact in the adult rat brain.

(A) Increasing amounts of proteins (25, 50, and 75 μg) from rat brain and HEK293-T total extracts were fractionated in SDS–PAGE. Western blots were revealed with anti-PIASγ and anti-β-actin (loading control). (B) One milligram of nuclear extracts of rat brain was incubated with 1 µg of anti-PIASγ antibody or 1 µg of rabbit IgG as control overnight at 4°C. Immunocomplexes were resolved by SDS–PAGE and western blots revealed with anti-PIASγ (upper panel) or anti-CoREST1 (lower panel). (C) One milligram of rat brain nuclear extract was incubated with 1 µg of anti-CoREST1 antibody or 1 µg of rabbit IgG as control overnight at 4°C. Immunocomplexes were resolved by SDS–PAGE and western blots were revealed with anti-CoREST1 (upper panel) or anti-PIASγ (lower panel).

Figure 2.
PIASγ and CoREST1 interact in the adult rat brain.

(A) Increasing amounts of proteins (25, 50, and 75 μg) from rat brain and HEK293-T total extracts were fractionated in SDS–PAGE. Western blots were revealed with anti-PIASγ and anti-β-actin (loading control). (B) One milligram of nuclear extracts of rat brain was incubated with 1 µg of anti-PIASγ antibody or 1 µg of rabbit IgG as control overnight at 4°C. Immunocomplexes were resolved by SDS–PAGE and western blots revealed with anti-PIASγ (upper panel) or anti-CoREST1 (lower panel). (C) One milligram of rat brain nuclear extract was incubated with 1 µg of anti-CoREST1 antibody or 1 µg of rabbit IgG as control overnight at 4°C. Immunocomplexes were resolved by SDS–PAGE and western blots were revealed with anti-CoREST1 (upper panel) or anti-PIASγ (lower panel).

PIASγ facilitates SUMO-2 conjugation to CoRESTs

As PIASγ is a SUMO E3-ligase, we studied whether CoREST proteins are SUMO-2 targets and whether PIASγ plays a role in their SUMOylation. Slower migrating bands characteristic of SUMOylated proteins were observed in western blots for Myc-CoREST1 and Myc-CoREST3, when PIASγ and SUMO-2 were overexpressed. However, a very weak band was observed in western blots for Myc-CoREST2 in the same conditions (Figure 3A). These results were confirmed in western blots of immunoprecipitated extracts of HEK293-T cells overexpressing PIASγ, SUMO-2, and Myc-CoREST proteins (Figure 3B). To test the specificity of SUMO-2 conjugation to CoRESTs, similar experiments were performed using a catalytic inactive mutant of PIASγ (PIASγC342A) [30]. No slower migrating bands were observed in western blots for Myc-CoRESTs when the mutant PIASγC342A was co-expressed (Figure 3A), demonstrating that PIASγ facilitates SUMO-2 conjugation to CoREST proteins. A western blot using a specific antibody against CoREST1 showed several slower migrating bands (Supplementary Figure S2) when Myc-CoREST1 is co-expressed with SUMO-2 and PIASγ, suggesting poly-SUMOylation or multi-SUMOylation of the protein.

CoRESTs are targets of SUMO-2.

Figure 3.
CoRESTs are targets of SUMO-2.

(A) Whole extracts of HEK293-T cells co-transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, PIASγ, and PIASγC342A as indicated were fractionated in SDS–PAGE. (B) Whole extracts of HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, and PIASγ as indicated were immunoprecipitated with anti-Myc antibody, and immunocomplexes were fractionated in SDS–PAGE. (C) HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, and Ubc9 as indicated were fractionated in SDS–PAGE. Protein extracts from cells transfected with empty plasmids (Mock) and immunoprecipitation with preimmune serum (IgG) were used as negative controls. Anti-Myc antibody was used to detect Myc-CoRESTs in the western blots.

Figure 3.
CoRESTs are targets of SUMO-2.

(A) Whole extracts of HEK293-T cells co-transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, PIASγ, and PIASγC342A as indicated were fractionated in SDS–PAGE. (B) Whole extracts of HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, and PIASγ as indicated were immunoprecipitated with anti-Myc antibody, and immunocomplexes were fractionated in SDS–PAGE. (C) HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, and Ubc9 as indicated were fractionated in SDS–PAGE. Protein extracts from cells transfected with empty plasmids (Mock) and immunoprecipitation with preimmune serum (IgG) were used as negative controls. Anti-Myc antibody was used to detect Myc-CoRESTs in the western blots.

SUMOylation is a three-enzymatic-step process, but either the E2 enzyme Ubc9 or the SUMO E3-ligases can carry out SUMO ligation to target proteins. To test whether CoREST2 is just a poor substrate for PIASγ, but can be SUMOylated in other conditions, Ubc9 was overexpressed instead PIASγ. As shown in Figure 3C, the characteristic slower migrating band attributable to the SUMO-2 modified CoREST2 protein was observed in western blots assays of cell extracts overexpressing Ubc9. SUMO-2 was also conjugated to CoREST1 and CoREST3 in the presence of Ubc9 (Figure 3C).

The above data strongly indicated that CoRESTs are the target of SUMO-2, prompting us to test with endogenous CoRESTs. To this end, we performed pull-down assays using affinity Ni-NTA agarose columns capable of retaining proteins modified with 6xHis-SUMO-2 [19]. The protocol of the experiments is depicted in Figure 4A and consists of passing guanidine-treated HEK293-T total extracts through the column, which will retain poli-His proteins. Then, proteins retained in the column are recognized by specific antibodies in western blots. Overexpression of PIASγ and 6xHis-SUMO-2 significantly increased SUMOylated proteins as observed by the large high molecular mass banding in the western blots revealed with anti-His antibody (Figure 4B,C, left panels). High molecular mass bands in extracts that overexpress PIASγ and 6xHis-SUMO-2 were also detected by specific antibodies against CoREST1 (Figure 4C, middle panel) and CoREST2 (Figure 4C, right panel), demonstrating that endogenous CoRESTs are substrates of SUMO-2. Observing more than one band suggests that CoREST1 and CoREST2 can be poly- or multi-SUMOylated (Figure 4C).

Endogenous CoREST1 and CoREST2 are SUMOylated by SUMO-2.

Figure 4.
Endogenous CoREST1 and CoREST2 are SUMOylated by SUMO-2.

(A) Experimental design for the detection of endogenous SUMOylated proteins by pull-down of 6xHis-SUMO-2 targets by nickel-coated agarose beads. (B) Western blot analysis of the presence of 6xHis-modified proteins, CoREST1, CoREST2, PIASγ, and GAPDH in the 2% input fractions. (C) Western blot of 6xHis (left), CoREST1 (middle), and CoREST2 (right) after nickel affinity purifications from HEK293-T cells. Molecular mass markers are shown in kDa units.

Figure 4.
Endogenous CoREST1 and CoREST2 are SUMOylated by SUMO-2.

(A) Experimental design for the detection of endogenous SUMOylated proteins by pull-down of 6xHis-SUMO-2 targets by nickel-coated agarose beads. (B) Western blot analysis of the presence of 6xHis-modified proteins, CoREST1, CoREST2, PIASγ, and GAPDH in the 2% input fractions. (C) Western blot of 6xHis (left), CoREST1 (middle), and CoREST2 (right) after nickel affinity purifications from HEK293-T cells. Molecular mass markers are shown in kDa units.

Specificity of PIASγ-induced SUMOylation of recombinant Myc-CoREST proteins was further supported by showing that the slower migrating bands were no longer observed when SENP-1, but not a dominant negative form of SENP-1 (SENP-1-DN), was overexpressed (Figure 5A and Supplementary Figure S3). Similarly, SENP-1 reversed Ubc9-dependent CoREST1 and CoREST2 SUMOylation (Figure 5B). In conclusion, all CoREST proteins are the target for SUMO-2, but different efficacy is exhibited by PIASγ conjugating SUMO-2 to CoREST proteins.

CoRESTs SUMOylation mediated by Ubc9 and PIASγ is prevented by SENP-1 presence.

Figure 5.
CoRESTs SUMOylation mediated by Ubc9 and PIASγ is prevented by SENP-1 presence.

(A) Whole extracts of HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, PIASγ, SENP-1, and SENP-1-DN as indicated were fractionated in SDS–PAGE. (B) Whole extracts of HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Ubc9, SUMO-2, and SENP-1 as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect Myc-CoRESTs in the western blots.

Figure 5.
CoRESTs SUMOylation mediated by Ubc9 and PIASγ is prevented by SENP-1 presence.

(A) Whole extracts of HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Myc-CoREST3, SUMO-2, PIASγ, SENP-1, and SENP-1-DN as indicated were fractionated in SDS–PAGE. (B) Whole extracts of HEK293-T cells transfected with Myc-CoREST1, Myc-CoREST2, Ubc9, SUMO-2, and SENP-1 as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect Myc-CoRESTs in the western blots.

Lysine 297 is the main target of SUMO-2 in CoREST1 and lysine 236 is in CoREST3

CoREST sequences were examined using the SUMOplot and the SUMOsp software (available online) to identify a potential lysine target of SUMOylation. Table 1 shows the list of selected lysines based on higher scores obtained with both types of software, to be further analyzed in SUMOylation assays. A striking conserved lysine in CoREST1(K148), CoREST2(K88), and CoREST3(K98) appeared as a good candidate, since this lysine is located in equivalent positions of the ELM2 domain of all CoRESTs. SUMOylation assays using the corresponding KΔR mutants showed no significant changes compared with wild-type proteins (Supplementary Figure S4A,C), discarding the possibility of a common SUMOylation of the ELM2 domain of CoREST proteins. Next, we tested lysine 297 in CoREST1 (lysine 294 in outdated sequence), which was previously shown to be the target for SUMO-1 conjugation [22]. Indeed, this lysine is in a SUMO consensus motif according to our in silico analysis (Table 1). The replacement K297R in CoREST1 eliminated the main band of SUMOylation induced by either Ubc9 or PIASγ (Figure 6 and Supplementary Figure S4A). On the other hand, minor bands were still detected in the mutant CoREST1(K297R), suggesting other lysine targets for SUMOylation (Figure 6A).

Determination of SUMO-2 lysine targets of CoREST1 and CoREST3.

Figure 6.
Determination of SUMO-2 lysine targets of CoREST1 and CoREST3.

(A) Whole extracts of HEK293-T cells transfected with Myc-CoREST1 wild-type, Myc-CoREST1 (K297R), SUMO-2, Ubc9, and PIASγ as indicated were fractionated in SDS–PAGE. (B) Whole extracts of HEK293-T cells transfected with Myc-CoREST3 wild-type, Myc-CoREST3 (K236R), SUMO-2, and PIASγ as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect Myc-CoREST recombinant proteins in the western blots.

Figure 6.
Determination of SUMO-2 lysine targets of CoREST1 and CoREST3.

(A) Whole extracts of HEK293-T cells transfected with Myc-CoREST1 wild-type, Myc-CoREST1 (K297R), SUMO-2, Ubc9, and PIASγ as indicated were fractionated in SDS–PAGE. (B) Whole extracts of HEK293-T cells transfected with Myc-CoREST3 wild-type, Myc-CoREST3 (K236R), SUMO-2, and PIASγ as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect Myc-CoREST recombinant proteins in the western blots.

Table 1
Predicting potential lysines as SUMO targets in CoREST proteins using SUMOsp and SUMOplot

Potential SUMOylation sites containing the predicted acceptor lysine are highlighted in bold.

SUMOspSUMOplot
CoREST1 Position Peptide Score Cutoff Type Position Peptide Score 
148 SEAKLD2.853 2.26 Type II: Non-consensus K297 ETVPQ VKKE KHSTQ 0.93 
297 PQVKKE4.735 0.1 Type I: ψ-K-X-E K148 QNLSE AKLD EYIAI 0.79 
CoREST2 Position Peptide Score Cutoff Type Position Peptide Score 
60 PECKPE2.441 2.26 Type II: Non-consensus K88 HCVSD AKLD KYIAM 0.79 
88 SDAKLD2.926 2.26 Type II: Non-consensus K240 ARPGP GKKE IQVSQ 0.67 
223 GDPKRE1.123 0.1 Type I: ψ-K-X-E K223 PDTGD PKRE PLPSR 0.61 
CoREST3 Position Peptide Score Cutoff Type Position Peptide Score 
98 PDAKLD2.779 2.26 Type II: Non-consensus K236 DPKKE AKKE GNTEQ 0.79 
232 YDPKKE0.46 0.1 Type I: ψ-K-X-E K98 HSIPD AKLD EYIAI 0.79 
236 KEAKKE0.882 0.1 Type I: ψ-K-X-E K232 DSDYD PKKE AKREG 0.61 
SUMOspSUMOplot
CoREST1 Position Peptide Score Cutoff Type Position Peptide Score 
148 SEAKLD2.853 2.26 Type II: Non-consensus K297 ETVPQ VKKE KHSTQ 0.93 
297 PQVKKE4.735 0.1 Type I: ψ-K-X-E K148 QNLSE AKLD EYIAI 0.79 
CoREST2 Position Peptide Score Cutoff Type Position Peptide Score 
60 PECKPE2.441 2.26 Type II: Non-consensus K88 HCVSD AKLD KYIAM 0.79 
88 SDAKLD2.926 2.26 Type II: Non-consensus K240 ARPGP GKKE IQVSQ 0.67 
223 GDPKRE1.123 0.1 Type I: ψ-K-X-E K223 PDTGD PKRE PLPSR 0.61 
CoREST3 Position Peptide Score Cutoff Type Position Peptide Score 
98 PDAKLD2.779 2.26 Type II: Non-consensus K236 DPKKE AKKE GNTEQ 0.79 
232 YDPKKE0.46 0.1 Type I: ψ-K-X-E K98 HSIPD AKLD EYIAI 0.79 
236 KEAKKE0.882 0.1 Type I: ψ-K-X-E K232 DSDYD PKKE AKREG 0.61 

The amino acid sequence of the longest variant of CoREST3 is highly similar to CoREST1 [13]. However, in the equivalent position of SUMOylable K297 of CoREST1, there is an arginine in CoREST3 (Supplementary Figure S5), which is a natural mutation to avoid SUMOylation in that position. Thus, we tested other potential lysines in CoREST3. Two close lysines (K232 and K236) in consensus SUMOylation motifs of CoREST3 (Table 1) appeared as possible candidates. The major band of SUMOylated CoREST3 decreased slightly in the mutant CoREST3(K232R) (Supplementary Figure S4C), but was no longer observed in the mutant CoREST3(K236R) (Figure 6), indicating that K236 is the main SUMO-2 lysine acceptor in CoREST3, even though other lysines (K98 and K232) appear as secondary lysine acceptors (Supplementary Figure S4).

CoREST2 lacks consensus sequences for SUMOylation, but it conserves a lysine (K240) in the equivalent position of the K297 SUMO target of CoREST1 (Supplementary Figure S5). Replacement of K240 for an arginine did not abolish SUMOylation of CoREST2 (Supplementary Figure S4B). Then, we tested the lysine 223 of CoREST2, which is located in the equivalent position to lysine 236 of CoREST3. CoREST2(K223R) was still SUMOylated (Supplementary Figure S4B). At the moment when we were carrying out these experiments, a proteomic analysis published by Hendriks and Vertegaal [31] showed that lysine 60 is a SUMO acceptor of CoREST2. Experiments carried out with the mutant CoREST2(K60R) showed that K60 is not the SUMOylation target of CoREST2, under our experimental conditions (Supplementary Figure S4B). These data suggest that the SUMO-2 target in CoREST2 could be another lysine in an unexpected sequence or, alternatively, several lysines of CoREST2 could be targets of SUMO-2.

PIASγ controls the stability of CoREST proteins

As noted in previous western blots, PIASγ overexpression induced significant augmentation of recombinant Myc-CoREST protein levels, an effect not observed when Ubc9 is overexpressed (Figures 3 and 5). To further investigate this observation, quantitative analyses comparing recombinant CoRESTs in the presence of PIASγ and Ubc9 were performed. As shown in Figure 7A, PIASγ overexpression increased Myc-CoREST1, Myc-CoREST2, and Myc-CoREST3 protein levels significantly. This effect is specific for CoREST proteins because recombinant HA-LSD1 levels remained similar to controls when PIASγ was overexpressed (Figure 7B). Strikingly, the augmentation of CoREST proteins induced by PIASγ seems to be independent of CoRESTs’ own SUMOylation, since co-expression of Ubc9 instead of PIASγ did not induce augmentation of recombinant Myc-CoRESTs (Figure 7C). This is further supported by the fact that the SUMO mutant CoREST1(K297R) protein levels were significantly increased in the presence of PIASγ (Figure 8A). These data support the idea that PIASγ has two effects on the CoRESTs that are independent of each other. On the one hand, it facilitates the SUMOylation of CoREST and, on the other hand, it induces their stabilization, but this stabilization does not require that PIASγ directly SUMOylate the CoRESTs. Interestingly, increased levels of CoREST proteins required a competent enzymatic PIASγ, since the inactive mutant, PIASγ-C342A, did not increase CoREST protein levels as wild-type PIASγ (Supplementary Figure S6). Furthermore, SENP-1 prevented the increase of CoREST1 and CoREST3 induced by PIASγ (Figure 8A and Supplementary Figure S6). To further test the stabilizing effect of PIASγ, we quantified endogenous CoREST1 in the presence of cycloheximide (CHX). As shown in Figure 8B, the half-life of endogenous CoREST1 increased when SUMO-2 and PIASγ are overexpressed. Altogether, the data indicate that the SUMO E3-ligase PIASγ regulates the stability of CoREST transcriptional co-repressors.

PIASγ increases CoREST protein levels.

Figure 7.
PIASγ increases CoREST protein levels.

(A) Whole extracts of HEK293-T cells co-transfected with Myc-CoREST1, Myc-CoREST2, and Myc-CoREST3 along with PIASγ or mock empty plasmid (−) were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect the Myc-CoRESTs in western blots. β-Actin was used as a loading control. Left: representative western blots. Right: values correspond to the mean ± SEM of four independent experiments; *P < 0.05; **P < 0.01. Statistical analysis was performed by the Mann–Whitney U-test. (B) Whole extracts of HEK293-T cells co-transfected with HA-LSD1 along with PIASγ or mock empty plasmid (−) were fractionated in SDS–PAGE. Anti-HA antibody was used to detect the HA-LSD1 in western blots. β-Actin was used as a loading control. Left: representative western blot. Right: values correspond to the mean ± SEM of three independent experiments. (C) Whole extracts of HEK293-T cells transfected with Myc-CoREST(1,2,3), Ubc9, and SUMO-2 as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect Myc-CoRESTs in western blots. β-Actin was used as a loading control. Top: representative western blot. Below: values are expressed as fold of change over control (Myc-CoREST transfection only) of three independent experiments.

Figure 7.
PIASγ increases CoREST protein levels.

(A) Whole extracts of HEK293-T cells co-transfected with Myc-CoREST1, Myc-CoREST2, and Myc-CoREST3 along with PIASγ or mock empty plasmid (−) were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect the Myc-CoRESTs in western blots. β-Actin was used as a loading control. Left: representative western blots. Right: values correspond to the mean ± SEM of four independent experiments; *P < 0.05; **P < 0.01. Statistical analysis was performed by the Mann–Whitney U-test. (B) Whole extracts of HEK293-T cells co-transfected with HA-LSD1 along with PIASγ or mock empty plasmid (−) were fractionated in SDS–PAGE. Anti-HA antibody was used to detect the HA-LSD1 in western blots. β-Actin was used as a loading control. Left: representative western blot. Right: values correspond to the mean ± SEM of three independent experiments. (C) Whole extracts of HEK293-T cells transfected with Myc-CoREST(1,2,3), Ubc9, and SUMO-2 as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect Myc-CoRESTs in western blots. β-Actin was used as a loading control. Top: representative western blot. Below: values are expressed as fold of change over control (Myc-CoREST transfection only) of three independent experiments.

PIASγ-induced increase of CoREST1 is independent of its SUMOylation.

Figure 8.
PIASγ-induced increase of CoREST1 is independent of its SUMOylation.

(A) Whole extracts of HEK293-T cells transfected with wild-type Myc-CoREST1, Myc-CoREST1-K297R, SUMO-2, PIASγ, and SENP-1 as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect wild-type Myc-CoREST1 and mutant Myc-CoREST1 (K297R) in western blots. β-Actin was used as a loading control. (B) Quantitative analisis of protein levels of Myc-CoREST1 (WT and K297R mutant) as indicated on letter (A). Values correspond to the mean ± SEM of three independent experiments. (C) HEK293-T cells transfected as indicated were treated with 100 μg/ml CHX for the indicated times. Western blots were performed to identify endogenous CoREST1 and GAPDH was used as a loading control. Densitometry analyses were performed using ImageJ. Top: representative western blot. Bottom: values correspond to the mean ± SEM of three independent experiments.

Figure 8.
PIASγ-induced increase of CoREST1 is independent of its SUMOylation.

(A) Whole extracts of HEK293-T cells transfected with wild-type Myc-CoREST1, Myc-CoREST1-K297R, SUMO-2, PIASγ, and SENP-1 as indicated were fractionated in SDS–PAGE. Anti-Myc antibody was used to detect wild-type Myc-CoREST1 and mutant Myc-CoREST1 (K297R) in western blots. β-Actin was used as a loading control. (B) Quantitative analisis of protein levels of Myc-CoREST1 (WT and K297R mutant) as indicated on letter (A). Values correspond to the mean ± SEM of three independent experiments. (C) HEK293-T cells transfected as indicated were treated with 100 μg/ml CHX for the indicated times. Western blots were performed to identify endogenous CoREST1 and GAPDH was used as a loading control. Densitometry analyses were performed using ImageJ. Top: representative western blot. Bottom: values correspond to the mean ± SEM of three independent experiments.

CoREST1 and CoREST3 SUMO-deficient mutants maintain transcriptional repressor function

SUMOylation of proteins that regulate transcription has been associated with control of gene expression, and thus, we tested whether SUMOylation of CoRESTs could regulate their transcriptional repressor capacity. Gene reporter assays using the Gal4-UAS system were carried out to evaluate CoREST1 and CoREST3 SUMO-deficient mutants in transcriptional repressor activity. The SUMO-deficient CoREST1(K297R) and CoREST3(K236R) showed no significant differences with their wild-type counterparts repressing the reporter (Figure 9A,C). The SUMO-2 acceptor lysines of CoREST1 and CoREST3 lie in the linker region between the two SANT domains. The ELM2 and first SANT domains of CoREST proteins are essential for HDAC1/2 interaction [8,13,14], while the linker region and second SANT domain of CoRESTs play an important role in LSD1 and nucleosome DNA interactions [17]. Thus, we wondered if SUMOylation could regulate either HDAC1/2 or LSD1 interaction. To this end, co-immunoprecipitation assays were performed to compare the interaction capacity of CoREST1 and CoREST1(K297R) with LSD1 and HDAC1. As shown in Figure 9B, CoREST1(K297R) displayed a similar interaction with LSD1 and HDAC1 compared with wild-type CoREST1, indicating that CoREST1 SUMOylation is not required for LCH complex constitution. Similar to CoREST1, the SUMOylation-deficient mutant CoREST3(K236R) showed the same interaction profile with HDAC1 and LSD1 as that of wild-type CoREST3 (Figure 9D).

Mutation of main SUMO-2 acceptor lysines of CoREST1 and CoREST3 does not affect their transcriptional repressor function and interaction profile with HDAC1 and LSD1

Figure 9.
Mutation of main SUMO-2 acceptor lysines of CoREST1 and CoREST3 does not affect their transcriptional repressor function and interaction profile with HDAC1 and LSD1

(A,C) Recombinant Gal4-DBDs fused to full-length wild-type CoREST1 or CoREST3, and SUMO-deficient mutants CoREST1(K297R) and CoREST3 (K236R) were assayed for their ability to repress the luciferase reporter gene (G5S4tkLuc) in HEK293-T cells transfected in a reporter/repressor molar ratio of 1 : 0.5. Values are expressed as fold of luciferase activity over control (reporter transfection only) and correspond to the mean ± SEM of three independent experiments. **P < 0.01. Statistical analysis was performed by the Mann–Whitney U-test. (B,D) Whole extracts of HEK293-T cells transfected with the indicated plasmids were immunoprecipitated with anti-Myc antibody, and complexes were fractionated in SDS–PAGE. Anti-LSD1 and anti-HDAC1 antibodies were used to detect the specific proteins in western blots. Protein extracts from cells transfected with empty plasmids (Mock) and immunoprecipitation with preimmune serum (IgG) were used as negative controls.

Figure 9.
Mutation of main SUMO-2 acceptor lysines of CoREST1 and CoREST3 does not affect their transcriptional repressor function and interaction profile with HDAC1 and LSD1

(A,C) Recombinant Gal4-DBDs fused to full-length wild-type CoREST1 or CoREST3, and SUMO-deficient mutants CoREST1(K297R) and CoREST3 (K236R) were assayed for their ability to repress the luciferase reporter gene (G5S4tkLuc) in HEK293-T cells transfected in a reporter/repressor molar ratio of 1 : 0.5. Values are expressed as fold of luciferase activity over control (reporter transfection only) and correspond to the mean ± SEM of three independent experiments. **P < 0.01. Statistical analysis was performed by the Mann–Whitney U-test. (B,D) Whole extracts of HEK293-T cells transfected with the indicated plasmids were immunoprecipitated with anti-Myc antibody, and complexes were fractionated in SDS–PAGE. Anti-LSD1 and anti-HDAC1 antibodies were used to detect the specific proteins in western blots. Protein extracts from cells transfected with empty plasmids (Mock) and immunoprecipitation with preimmune serum (IgG) were used as negative controls.

Discussion

The coexistence of transcriptional regulators highly similar, which play diverse and sometimes opposite functional roles in the same cells, requires that their action be confined to a specific cellular context. Posttranslational modifications of transcription factors and co-regulators provide a mechanism to maintain the specificity of transcriptional regulator functions. SUMOylation of transcription factors and co-regulators plays key roles in controlling transcriptional activity, half-life, localization, and subunit composition of the complexes they belong to. Here, we have described similarities and differences of SUMOylation of the CoREST family of transcriptional co-repressors. Our data show that all CoRESTs are SUMOylated. However, the lysine targets differ among them. The SUMO E3-ligase PIASγ was identified as an interacting partner for all CoRESTs that facilitates their SUMOylation with different efficacy. PIASγ, but not Ubc9, increases the stability of CoREST proteins.

The SUMO-2 acceptor lysines in CoREST proteins are different and lie in divergent sequences of these transcriptional co-repressors. The lysine 297 of CoREST1, previously shown to be the target of SUMO-1 [22], is also the SUMO-2 acceptor, as our data show. Lysine 297 of CoREST1 localizes on the linker region connecting the two SANT domains. The SUMO-2 acceptor lysine (K236) of CoREST3 also lies in the linker region. The linker regions have dissimilar sequences among the three CoRESTs [13]. CoREST3 lacks the lysine target of SUMOylation present in CoREST1 (Supplementary Figure S5). In that position, instead of VK297KE present in CoREST1, CoREST3 has GRRE, a sequence that impedes CoREST3 SUMOylation in the equivalent position to CoREST1. On other hand, CoREST2 harbors GK240KE in the equivalent position to the SUMO site of CoREST1. However, K240 of CoREST2 is not as efficient a SUMO-2 target as is K297 in CoREST1. The glycine preceding the lysine does not fulfill the requirement of a large hydrophobic amino acid needed for efficient SUMOylation [19]. Similarly, the lysine acceptor in CoREST3 AK236KE is not conserved in CoREST2 and a serine instead of the alanine preceding the lysine precludes its SUMOylation in CoREST1. It is worth mentioning that the lysine 60 in CoREST2 (CK60PE), located in the ELM2 domain, which is not conserved in CoREST1 and CoREST3, was found to be the SUMO-2 target in a proteomic study using HeLa cells [31,32]. Hendriks et al. [32] showed that CoREST2 was found SUMOylated on K60 in response to heat shock or by SUMO protease and proteasome inhibition [32]. We did not observe differences between CoREST2(K60R) and wild-type CoREST2 in our SUMOylation assays, strengthening the idea that CoREST2 SUMOylation in lysine 60 responds to a particular cellular context and stimulus.

An important consequence of SUMOylation is the regulation of transactivating or transrepressing functions of transcriptional regulators. Here, we show that CoREST1 and CoREST3 SUMO-defective mutants displayed no significant difference in transcriptional repression action compared with their wild-type counterparts. Moreover, co-immunoprecipitation assays showed that CoREST1(K297R) and CoREST3(K236R) have similar ability to interact with LSD1 and HDAC1. Altogether, the data suggest that SUMOylation of CoREST1 and CoREST3 is not required for them to organize the formation of LCH complexes, and does not regulate their repression capacity. The fact that the SUMO-2 acceptor sites lie in domains different from those important for recruiting the enzymes required for histone tail modification supports the idea that SUMOylation of CoRESTs serves other purposes in these co-regulators.

SUMOylation regulates also the stability of target proteins. For instance, conjugation of SUMO-1 to HDAC1 decreases its stability, while conjugation of SUMO-2 exerts the opposite effect [33]. Our data suggest that SUMOylation regulates CoRESTs' stability in an indirect manner. On one hand, PIASγ increased protein levels of all CoRESTs. However, this effect appears to be independent of CoRESTs' SUMOylation, as shown by the similar increase in the SUMO-defective CoREST1(K297R) compared with wild-type CoREST1. On the other hand, Ubc9 induced SUMO-2 conjugation to all CoRESTs. However, it did not increase their protein levels. Nevertheless, the effect of PIASγ for increasing CoREST levels depends on its E3-SUMO ligase activity and is prevented by SENP-1 presence. Therefore, the data suggest that the increase in CoRESTs induced by PIASγ is due to an indirect SUMOylation mechanism. Supporting this suggestion, recent evidence showed that SUMOylation of the translation initiation factor 4E promotes the translation of a selected group of mRNAs [34], which could explain the increase in a specific set of proteins. It is interesting to think that the presence or absence PIASγ could regulate the abundance of CoREST proteins and other components of the complex during specific events, such as the neuronal differentiation process, in which a reduction of the LCH complex is evidenced [12].

The multiple banding observed for endogenous CoREST1 and CoREST2, and overexpressed CoREST1, suggests either multi- or poly-SUMOylation. The results with the mutant CoREST1(K297R) allow us to suggest that these are multi-SUMOylation events, since other bands are maintained for the SUMO mutant CoREST1(K297), indicative of secondary lysine acceptors of SUMO-2. This SUMOylation output is markedly different from those observed when CoRESTs are SUMOylated in the presence of Ubc9 that showed only one defined band, suggesting that PIASγ-induced SUMOylation of CoREST1 may play a different role than SUMO conjugation by Ubc9.

SUMOylation has been associated with a plethora of physiological and pathological processes. For instance, an increase in global SUMOylation and induction of the SUMO machinery is observed under heat-shock conditions and ischemic damage in the brain [35]. On the other hand, it is known that CoREST1 plays an important role in regulating the heat-shock response [24], and it has been associated with the regulation of ischemia-induced neuronal damage [36]. The regulation of these events by SUMOylation of the CoREST proteins is a concrete possibility.

In summary, the SUMO E3-ligase PIASγ facilitates the SUMOylation of CoRESTs and increases their stability. Evolution has favored a particular SUMOylation site in each CoREST, which could be related to different functions. Future studies are needed to find the consequences, functional implications, and impact of the specific SUMOylation of CoREST transcriptional co-repressors.

Abbreviations

     
  • CHX

    cycloheximide

  •  
  • Gfi-1

    growth factor independence-1

  •  
  • HA

    hemagglutinin

  •  
  • LCH

    LSD1–CoREST–HDAC1/2

  •  
  • Ni-NTA

    Nickel–Nitrilotriacetic acid

  •  
  • PIASγ

    protein inhibitor of activated STAT

  •  
  • SENP

    SUMO/Sentrin-specific protease

  •  
  • SUMO

    small ubiquitin-like modifier

Author Contribution

J.E.S. conducted most of the experiments, analyzed the results, and wrote most of the paper. C.A. set up the conditions for SUMOylation and half-life experiments. C.R. performed co-immunoprecipitations and SUMOylation assays with endogenous proteins. M.E.A. conceived the idea for the project, directed the experiments, and wrote the paper with J.E.S.

Funding

This work was supported by Fondecyt grant 1150200 to M.E.A. J.E.S. was recipient of a Conicyt doctoral fellowship # 21110058.

Acknowledgments

We thank Dr Eduardo Arzt (Universidad de Buenos Aires, Argentina), Dr Ron Hay (Center for Biomolecular Sciences, U.K.), Dr Peter O'Hare (Marie Curie Research Institute, U.K.), and Dr Fletcher White (Indiana University School of Medicine, Indianapolis, U.S.A.) for kind donations.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Saleque,
S.
,
Kim,
J.
,
Rooke,
H.M.
and
Orkin,
S.H.
(
2007
)
Epigenetic regulation of hematopoietic differentiation by Gfi-1 and Gfi-1b is mediated by the cofactors CoREST and LSD1
.
Mol. Cell
27
,
562
572
2
Upadhyay,
G.
,
Chowdhury,
A.H.
,
Vaidyanathan,
B.
,
Kim,
D.
and
Saleque,
S.
(
2014
)
Antagonistic actions of Rcor proteins regulate LSD1 activity and cellular differentiation
.
Proc. Natl Acad. Sci. U.S.A.
111
,
8071
8076
3
Yao,
H.
,
Goldman,
D.C.
,
Nechiporuk,
T.
,
Kawane,
S.
,
McWeeney,
S.K.
,
Tyner,
J.W.
et al
(
2014
)
Corepressor Rcor1 is essential for murine erythropoiesis
.
Blood
123
,
3175
3184
4
Yao,
H.
,
Goldman,
D.C.
,
Fan,
G.
,
Mandel,
G.
and
Fleming,
W.H.
(
2015
)
The corepressor Rcor1 is essential for normal myeloerythroid lineage differentiation
.
Stem Cells
33
,
3304
3314
5
Monaghan,
C.E.
,
Nechiporuk,
T.
,
Jeng,
S.
,
McWeeney,
S.K.
,
Wang,
J.
,
Rosenfeld,
M.G.
et al
(
2017
)
REST corepressors RCOR1 and RCOR2 and the repressor INSM1 regulate the proliferation-differentiation balance in the developing brain
.
Proc. Natl Acad. Sci. U.S.A.
114
,
E406
E415
6
Wang,
Y.
,
Wu,
Q.
,
Yang,
P.
,
Wang,
C.
,
Liu,
J.
,
Ding,
W.
, et al (
2016
)
LSD1 co-repressor Rcor2 orchestrates neurogenesis in the developing mouse brain
.
Nat. Commun.
7
,
10481
7
Andrés,
M.E.
,
Burger,
C.
,
Peral-Rubio,
M.J.
,
Battaglioli,
E.
,
Anderson,
M.E.
,
Grimes
,
J.
et al
(
1999
)
CoREST: a functional corepressor required for regulation of neural-specific gene expression
.
Proc. Natl Acad. Sci. U.S.A.
96
,
9873
9878
8
Ballas,
N.
,
Battaglioli,
E.
,
Atouf,
F.
,
Andres,
M.E.
,
Chenoweth,
J.
,
Anderson
,
M.E.
et al
(
2001
)
Regulation of neuronal traits by a novel transcriptional complex
.
Neuron
31
,
353
365
9
Ballas,
N.
,
Grunseich,
C.
,
Lu,
D.D.
,
Speh,
J.C.
and
Mandel,
G.
(
2005
)
REST and its corepressors mediate plasticity of neuronal gene chromatin throughout neurogenesis
.
Cell
121
,
645
657
10
Fuentes,
P.
,
Cánovas,
J.
,
Berndt,
F.A.
,
Noctor,
S.C.
and
Kukuljan,
M.
(
2012
)
CoREST/LSD1 control the development of pyramidal cortical neurons
.
Cereb. Cortex
22
,
1431
1441
11
Yang,
P.
,
Wang,
Y.
,
Chen,
J.
,
Li,
H.
,
Kang,
L.
,
Zhang
,
Y.
et al
(
2011
)
RCOR2 is a subunit of the LSD1 complex that regulates ESC property and substitutes for SOX2 in reprogramming somatic cells to pluripotency
.
Stem Cells
29
,
791
801
12
Sáez,
J.E.
,
Gómez,
A.V.
,
Barrios,
Á.P.
,
Parada,
G.E.
,
Galdames,
L.
,
González,
M.
et al
(
2015
)
Decreased expression of CoREST1 and CoREST2 together with LSD1 and HDAC1/2 during neuronal differentiation
.
PLOS ONE
10
,
e0131760
13
Barrios,
A.P.
,
Gómez,
A.V.
,
Sáez,
J.E.
,
Ciossani,
G.
,
Toffolo,
E.
,
Battaglioli
,
E.
et al
(
2014
)
Differential properties of transcriptional complexes formed by the CoREST family
.
Mol. Cell Biol.
34
,
2760
2770
14
Shi,
Y.
,
Lan,
F.
,
Matson,
C.
,
Mulligan,
P.
,
Whetstine,
J.R.
,
Cole
,
P.A.
et al
(
2004
)
Histone demethylation mediated by the nuclear amine oxidase homolog LSD1
.
Cell
119
,
941
953
15
Lee,
M.G.
,
Wynder,
C.
,
Cooch,
N.
and
Shiekhattar,
R.
(
2005
)
An essential role for CoREST in nucleosomal histone 3 lysine 4 demethylation
.
Nature
437
,
432
435
16
Shi,
Y-.J
,
Matson,
C.
,
Lan,
F.
,
Iwase,
S.
,
Baba,
T.
and
Shi,
Y.
(
2005
)
Regulation of LSD1 histone demethylase activity by its associated factors
.
Mol. Cell
19
,
857
864
17
Pilotto,
S.
,
Speranzini,
V.
,
Tortorici,
M.
,
Durand,
D.
,
Fish,
A.
,
Valente
,
S.
et al
(
2015
)
Interplay among nucleosomal DNA, histone tails, and corepressor CoREST underlies LSD1-mediated H3 demethylation
.
Proc. Natl Acad. Sci. U.S.A.
112
,
2752
2757
18
Flotho,
A.
and
Melchior,
F.
(
2013
)
Sumoylation: a regulatory protein modification in health and disease
.
Annu. Rev. Biochem.
82
,
357
385
19
Rodriguez,
M.S.
,
Dargemont,
C.
and
Hay,
R.T.
(
2001
)
SUMO-1 conjugation in vivo requires both a consensus modification motif and nuclear targeting
.
J. Biol. Chem.
276
,
12654
12659
20
Rytinki,
M.M.
,
Kaikkonen,
S.
,
Pehkonen,
P.
,
Jääskeläinen,
T.
and
Palvimo,
J.J.
(
2009
)
PIAS proteins: pleiotropic interactors associated with SUMO
.
Cell Mol. Life Sci.
66
,
3029
3041
21
Kim,
J.H.
and
Baek,
S.H.
(
2009
)
Emerging roles of desumoylating enzymes
.
Biochim. Biophys. Acta
1792
,
155
162
22
Muraoka,
A.
,
Maeda,
A.
,
Nakahara,
N.
,
Yokota,
M.
,
Nishida,
T.
,
Maruyama
,
T.
et al
(
2008
)
Sumoylation of CoREST modulates its function as a transcriptional repressor
.
Biochem. Biophys. Res. Commun.
377
,
1031
1035
23
Arredondo,
C.
,
Orellana,
M.
,
Vecchiola,
A.
,
Pereira,
L.A.
,
Galdames,
L.
and
Andrés,
M.E.
(
2013
)
PIASγ enhanced SUMO-2 modification of Nurr1 activation-function-1 domain limits Nurr1 transcriptional synergy
.
PLoS ONE
8
,
e5503
24
Gómez,
A.V.
,
Galleguillos,
D.
,
Maass,
J.C.
,
Battaglioli,
E.
,
Kukuljan,
M.
,
Andrés
,
M.S.
et al
(
2008
)
CoREST represses the heat shock response mediated by HSF1
.
Mol. Cell
31
,
222
231
25
Heckman,
K.L.
and
Pease,
L.R.
(
2007
)
Gene splicing and mutagenesis by PCR driven overlap extension
.
Nat. Protoc.
2
,
924
932
26
Galleguillos,
D.
,
Vecchiola,
A.
,
Fuentealba,
J.A.
,
Ojeda,
V.
,
Alvarez,
K.
,
Gómez
,
A.
et al
(
2004
)
PIASgamma represses the transcriptional activation induced by the nuclear receptor Nurr1
.
J. Biol. Chem.
279
,
2005
2011
27
Tatham,
M.H.
,
Rodriguez,
M.S.
,
Xirodimas,
D.P.
and
Hay,
R.T.
(
2009
)
Detection of protein SUMOylation in vivo
.
Nat. Protoc.
4
,
1363
1371
28
Dignam,
J.D.
,
Lebovitz,
R.M.
and
Roeder,
R.G.
(
1983
)
Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei
.
Nucleic Acids Res.
11
,
1475
1489
29
Saijo,
K.
,
Winner,
B.
,
Carson,
C.T.
,
Collier,
J.G.
,
Boyer,
L.
,
Rosenfeld,
M.G.
et al
(
2009
)
A Nurr1/CoREST pathway in microglia and astrocytes protects dopaminergic neurons from inflammation-induced death
.
Cell
137
,
47
59
.   
30
Nishida,
T.
,
Terashima,
M.
and
Fukami,
K.
(
2006
)
PIASγ-mediated repression of the Ets-1 is independent of its sumoylation
.
Biochem. Biophys. Res. Commun.
345
,
1536
1546
31
Hendriks,
I.A.
and
Vertegaal,
A.C.O.
(
2016
)
A comprehensive compilation of SUMO proteomics
.
Nat. Rev. Mol. Cell Biol.
17
,
581
595
32
Hendriks,
I.A.
,
D'Souza,
R.C.J.
,
Yang,
B.
,
Verlaan-de Vries,
M.
,
Mann,
M.
and
Vertegaal,
A.C.
(
2014
)
Uncovering global SUMOylation signaling networks in a site-specific manner
.
Nat. Struct. Mol. Biol.
21
,
927
936
33
Citro,
S.
,
Jaffray,
E.
,
Hay,
R.T.
,
Seiser,
C.
and
Chiocca,
S.
(
2013
)
A role for paralog-specific sumoylation in histone deacetylase 1 stability
.
J. Mol. Cell Biol.
5
,
416
427
34
Chen,
L.Z.
,
Li,
X.Y.
,
Huang,
H.
,
Xing,
W.
,
Guo,
W.
,
He,
J.
et al
(
2014
)
SUMO-2 promotes mRNA translation by enhancing interaction between eIF4E and eIF4G
.
PLoS ONE
9
,
e100457
35
Henley,
J.M.
,
Craig,
T.J.
and
Wilkinson,
K.A.
(
2014
)
Neuronal SUMOylation: mechanisms, physiology, and roles in neuronal dysfunction
.
Physiol. Rev.
94
,
1249
1285
36
Noh,
K.M.
,
Hwang,
J.Y.
,
Follenzi,
A.
,
Athanasiadou,
R.
,
Miyawaki,
T.
,
Greally
,
J.M.
et al
(
2012
)
Repressor element-1 silencing transcription factor (REST)-dependent epigenetic remodeling is critical to ischemia-induced neuronal death
.
Proc. Natl Acad. Sci. U.S.A.
109
,
E962
E971

Supplementary data