Biochimica et Biophysica Acta 1833 (2013) 2357–2368

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbamcr

SUMO1 negatively regulates the transcriptional activity of EVI1 andsignificantly increases its co-localization with EVI1 after treatmentwith arsenic trioxide

Sneha Singh, Anjan Kumar Pradhan, Soumen Chakraborty ⁎Department of Gene Function and Regulation, Institute of Life Sciences, Nalco Square, Bhubaneswar, Orissa, India

⁎ Corresponding author at: Institute of Life Sciences, Nal751023, India. Tel.: +91 674 2302420; fax: +91 674 230

E-mail addresses: soumen_ils@yahoo.co.in, soumen@

0167-4889/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.bbamcr.2013.06.003

a b s t r a c t

a r t i c l e i n f o

Article history:Received 3 January 2013Received in revised form 31 May 2013Accepted 4 June 2013Available online 13 June 2013

Keywords:EVI1SumoylationBcl-xLPIASySIRT1Arsenic trioxide (ATO)

Aberrant expression of the proto-oncogene EVI1 (ecotropic virus integration site1) has been implicated notonly in myeloid or lymphoid malignancies but also in colon, ovarian and breast cancers. Despite its impor-tance in oncogenesis, the regulatory factors and mechanisms that potentiate the function of EVI1 and its con-sequences are partially known. Here we demonstrated that EVI1 is post-translationally modified by SUMO1at lysine residues 533, 698 and 874. Although both EVI1 and SUMO1 were found to co-localize in nuclearspeckles, the sumoylation mutant of EVI1 failed to co-localize with SUMO1. Sumoylation abrogated theDNA binding efficiency of EVI1 and also affected EVI1 mediated transactivation. The SUMO ligase PIASywas found to play a bi-directional role on EVI1, PIASy enhanced EVI1 sumoylation and augmentedsumoylated EVI1 mediated repression. PIASy was also found to interact with EVI1 and impaired EVI1 tran-scriptional activity independent of its ligase activity. Arsenic trioxide (ATO) known to act as an antileukemicagent for acute promyelocytic leukemia (APL) not only enhanced EVI1 sumoylation but also enhanced theco-localization of EVI1 and SUMO1 in nuclear bodies distinct from PML nuclear bodies. ATO treatment alsoaffected the Bcl-xL protein expression in EVI1 positive cell line. Thus, the results showed that arsenic treat-ment enhanced EVI1 sumoylation, deregulated Bcl-xL, which eventually may induce apoptosis in EVI1 posi-tive cancer cells. The study for the first time explores and reports sumoylation of EVI1, which plays anessential role in regulating its function.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Transcriptional activation of human EVI1 gene occurs in myeloidneoplasm as a consequence of chromosomal translocation and inser-tion involving chromosome band 3q26, where the gene is located.There are seven splice variants of EVI1 (Homo sapiens) reported inNCBI which differs in 5′UTR region of the nucleotide sequence. Ab-normal expression of EVI1 has also been demonstrated in patientswith a normal karyotype, suggesting that activation of the EVI1 genemay occur via other mechanisms as well. Studies reveal that EVI1may be over expressed in solid tumors which can affect the diseaseprogression or sensitivity to chemotherapy [1–3]. Ectopic expressionof EVI1 in bone marrow (BM) progenitors lead to an alteration ingranulopoiesis and a block in erythropoiesis in response to cytokinesin vitro whereas in ovarian epithelial cells, only short term expressionof EVI1 increased proliferation however long term expression strong-ly reduced growth and/or survival [4,5]. Recent studies point to thefact that EVI1 controls the function of a number of genes some of which

co Square, Bhubaneswar, Orissa0728.ils.res.in (S. Chakraborty).

rights reserved.

are deciphered and/or the function of EVI1 is dictated by a group of spe-cific regulators that post-translationally modify EVI1. Post-translationalmodifications on a protein constitute complex regulatory program thattransduces molecular information through signaling pathways. Similarcascade of eventsmay occur in the process of post-translationalmodifica-tion of EVI1 which needs to be investigated.

SUMO (small ubiquitin-likemodifier)modification is a covalent pro-tein modification mediated by a set of sumoylation pathway enzymes.SUMO covalently attaches to a wide variety of proteins, through an en-zymatic cascade on lysine residues at consensus sequence ΨKXE/D(where Ψ is a large hydrophobic amino acid, K is the target lysine, Xis any amino acid and E/D is glutamic/aspartic acid). Sumoylationincreases molecular weight of its target protein by 15–18 kDa inSDS-PAGE gel. In proteins having more than one sumoylation site, anincrease in the molecular mass is expected depending on the numberof the sites. Multiple bands are also observed which corresponds todifferent sumoylation states of the protein. Sumoylation can modulateprotein function by stabilizing it, changing its subcellular localization,helping it with protein–protein and/or protein–DNA interaction [6–13].Indeed several transcription factors, such as p53, SP3, c-Jun, c-Myb,MEL1S, transcriptional co-activator such as GRIP1, viral protein IE2-p86and various nuclear receptors like AR have recently been shown to be a

2358 S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

subject of sumoylation, and this modification can have a positive or anegative influence on the transcriptional activity of several targets[14–18]. Pathologies including prostate cancer, breast cancer and severalneurological disorders like Parkinson's, Alzheimer's, and Huntington'sdiseases have been reported to be associated with irregular patterns ofsumoylation [19,20]. Arsenic trioxide (ATO), a potent anti-cancer agentdirectly binds to PML and degrades PML-RARα as well as PML throughSUMO dependent polyubiquitination leading to apoptosis of APL cells[21]. Low doses of ATO induce partial differentiation whereas higherdose causes apoptosis in APL cells. ATO inhibits cell growth and inducesapoptosis in several cancer cell lines [22,23].

Previously we and others have reported that acetylated anddeacetylated form of EVI1 can divergently regulate various functionsof EVI1 by influencing Bcl-xL and GATA-2 transcriptional activity[24,25]. Our present observation reveals that EVI1 is modified bySUMO1 at lysine residues 533, 698 and 874. Sumoylation was foundto negatively regulate the transactivation function and DNA bindingaffinity of EVI1. PIASy was found to be a new interacting partner anda negative regulator of EVI1 which also enhanced EVI1 sumoylation.ATO treatment in ovarian carcinoma (SKOV3) cells not only increasedsumoylation of EVI1 but also enhanced the co-localization rate of EVI1and SUMO1 in nuclear speckles other than PML nuclear bodies. Thestudy reveals the role of ATO in enhancing EVI1 sumoylation, and thata marked decrease in Bcl-xL expression supports the mechanism in-volved in ATO induced apoptosis in cancer cells.

2. Materials and methods

2.1. Plasmids

CMV-flag-EVI1 was cloned from EVI1-myc as described previously(the first methionine of EVI1 was disrupted and the construct utilizesthe start codon of flag tag) [24]. Full length EVI1 was cloned aspGEX-flag-EVI1 from CMV-flag-EVI1 into pGEX-6P-1-flag (pGEX-6P-1-flag was constructed by inserting the flag tag epitope sequence inBamHI and SalI sites within the multiple cloning site of pGEX-6P-1).CMV-Ubc9-flag was cloned from pTET23a-Ubc9 (provided by Dr. F.Melchior, Heidelberg University, Germany) into EcoRV and SalI ofCMV-flag-NLS. pCDNA3-SUMO1-HA was a kind gift from Dr. R. Hay(University of Dundee, Scotland, UK). PIASy-flag was provided by Dr.R. Grosschedl (Max Planck Institute for Immunobiology and Epige-netics, Germany). Reporter plasmid; pGL2-Bcl-xL promoter (fragmentcontaining the EVI1 binding site cloned in pGL2 is from −3102 to +98taking transcription start site as 1) or pGL4.10-SIRT1 promoter (covers2.4 kb upstream from the start codon of SIRT1) and Renilla luciferaseplasmid; pRL-TK are described previously [24,26]. SUMO–Ubc9 (Su Ub)fusion plasmid was provided by Dr. Jin-Hyun Ahn (Samsung BiomedicalResearch Institute, Korea) [27]. Su–Ub fusion plasmid was generated bycloning SUMO and Ubc9 together with a linker. As per the literaturethe construct enhanced the sumoylation efficiency of proteins havingSIM (SUMO interacting motif). Although EVI1 doesn't have a knownSIM sequence in its coding region, we found it to be equally efficient toUbc9 to promote sumoylation of EVI1.

2.2. Site directed mutagenesis

The point mutants used in the study were generated by insertingmutation into CMV-flag-EVI1 and pGEX-flag-EVI1 by PCR cloningmethod using primers designed for introducing specified mutations(lysine to arginine). All the plasmids were sequenced to validate thepresence of desired mutations.

2.3. Cell culture and transfection

HEK293T (human embryonic kidney) and SKOV3 (ovarian cancer)cell lines were maintained in Dulbecco's modified Eagles medium

(DMEM, Invitrogen, USA) supplemented with 10% heat inactivatedfetal bovine serum (Invitrogen, USA) in 37 °C, 5% CO2 incubator.HEK293T cells were transfected by calcium phosphate precipitationmethod with various plasmid combinations as indicated in the figures.Control-siRNA and SUMO1-siRNA (Santa Cruz, CA) transfection inSKOV3 cell line was done using Icafectin siRNA transfection reagent(Eurogentec, Belgium).

2.4. Antibodies

Antibodies that were used in the assays include, mouse monoclonalanti-flag (M2) antibody (Sigma, USA), anti-EVI1 and anti-EVI1 (C50E12)(Cell Signaling Technology, USA), anti-HA (3F10) (Roche Applied Science,Germany), anti-SUMO1 (from Active Motif kit, USA and Santa Cruz, CA),anti-SUMO1 (D-11) and anti-PML (PG-M3) (Santa Cruz, CA) and anti-Bcl-xL (Imgenex, India). Anti-mouse, anti-rabbit and anti-rat secondaryantibodies were purchased from Santa Cruz Biotechnology (Santa Cruz,CA). Alexa-488 and alexa-594 conjugated anti-rabbit, anti-mouse andanti-rat were purchased from Invitrogen, USA.

2.5. Purification and expression of pGEX-EVI1-flag and mutants inEscherichia coli

For production of EVI1-flag and mutant proteins by GST pull downassay, BL21 competent cells were transformed with pGEX-flag-EVI1and themutants. Transformed cells were grown and protein expressionwas induced with IPTG (Sigma, USA). Cells were harvested and lysed inphosphate buffer saline supplemented with Triton-X (Sigma, USA) andprotease inhibitor (Sigma, USA). Lysates were incubated with 50 μl ofGST-tagged sepharose beads (GE Healthcare, UK) for 2 h at 4 °C. Pro-teins were eluted by incubating the beads in cleavage buffer withpreScission protease (GE healthcare, UK).

2.6. In vitro sumoylation assay

The assay was performed using GST purified protein and in vitrosumoylation assay kit (Active Motif, USA) which consists of E1 activat-ing enzyme, E2 conjugating enzyme, SUMO1/SUMO1-mut proteins,protein buffer and sumoylation buffer.

2.7. Immunoblot analysis and immunoprecipitation

For in vivo sumoylation assays, HEK293T cells seeded in 6 cm cellculture dish were transfected with 3 μg of CMV-flag-EVI1, 5 μg ofpCDNA3-SUMO1-HA, 3 μg of Su Ub (SUMO–Ubc9 fusion plasmid)/CMV-Ubc9-flag, with or without 3 μg of PIASy-flag. The DNA concen-tration was equalized by adding equal concentration of CMV-flag vec-tor. EVI1 positive SKOV3 cells or transiently transfected HEK293Tcells were washed with phosphate buffer saline, lysed in 2× samplebuffer containing 20 mM N-ethylmaleimide (NEM; sumoylation pro-tease inhibitor) and boiled for 10 min. After centrifugation superna-tants were subjected to SDS-PAGE and analyzed by immunoblottingusing indicated antibodies.

For co-immunoprecipitation assay, HEK293T cells grown in 10 cmcell culture plates were transfected with 7 μg each of EVI1-myc andPIASy-flag. For immunoprecipitation experiments, HEK293T cellstransfected with indicated plasmids mentioned in figure were lysedin lysis buffer (25 mM Tris–Cl; pH 7.5, 150 mM NaCl, 1% NP-40)supplemented with protease inhibitor (Sigma, USA) and cleared bycentrifugation. Lysates were precleared with protein A/G beads (GEHealthcare, UK). Precleared lysates were incubated with indicated an-tibodies for 2 h at 4 °C followed by overnight incubation with proteinA/G beads. Immunoprecipitates collected on the beads were washedthree times with IP wash buffer (50 mM Tris–Cl; pH 8.0, 150 mMNaCl, 0.1% NP-40, 1 mM EDTA) supplemented with protease inhibitor

2359S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

cocktail (Sigma, USA). Bound proteins were extracted with 2× sam-ple buffer and analyzed by SDS-PAGE and immunoblotting.

2.8. Indirect immunofluorescence assay

SKOV3 or HEK293T cells were cultured on gelatin coated coverslips.HEK293T cells were transiently transfected with 3 μg EVI1-flag/EVI1(3KR)-flag (lysine to arginine mutation at position 533, 698 and 874in EVI1) and 5 μg SUMO1-HA. After 36 h, cells were washed in cold1× PBS, fixed with 4% paraformaldehyde (Sigma, USA) and perme-abilized with methanol:acetone (Qualigen, USA). Cells were stainedwith primary antibodies for 1 h and Alexa fluor tagged secondary anti-bodies for 40 min. Cells were washed and stained with DAPI (Sigma,USA), mounted on glass slides with an antifade mounting reagent(Invitrogen, USA). The slides were visualized under a laser scanningconfocal microscope (Leica, Germany) using Argon-488, He–Ne-594and 405 solid state laser.

2.9. Reporter gene assay

HEK293T cells plated in sixwell platewere co-transfectedwith indi-cated plasmids in the figures at a concentration of 1 μg of pGL2-Bcl-xL/pGL4.10-SIRT1 promoter, 1 μg each of CMV-flag, EVI1-flag/EVI1 (3KR)-flag, PIASy-flag and 2 μg of SUMO1-HAalongwith 0.2 μg of Renilla lucif-erase plasmid; pRL-TK per well. After 24 h, cells were assayed with aDual luciferase assay system (Promega, USA) and luciferase activitywas determined by luminometer (Glomax 20/20, Promega, USA) asper manufacturer's instruction.

2.10. Chromatin immunoprecipitation (ChIP) assay

ChIP assaywas performed in HEK293T cells transfectedwith EVI1 andSu Ub (SUMO–Ubc9 fusion plasmid) or SKOV3 cells with control-siRNAand SUMO1-siRNA using Quik ChIP kit (Imgenex, India) as described pre-viously [24]. EVI1 protein was immunoprecipitated with EVI1 (C50E12)antibody (Cell Signaling Technology, USA). The promoter-specific primersflanking the binding site used for the PCR were: Bcl-xL-F AGGGTAAATGGCATGCATATTAA and Bcl-xL-R TTATAATAGGGATGGGCTCAACCA(amplicon size-170 bp).

2.11. Real-time quantitative PCR

Total RNA was extracted from control-siRNA and SUMO1-siRNAtransfected SKOV3 cells by using TRIzol reagent (Invitrogen, CA). SYBRgreen master mix (Roche Applied Science, Germany) was used for thereal time PCR (RQ-PCR) on Opticon 2.0 (MJ Research, USA). Fold changeof target gene relative to β-actin was calculated by 2¯ΔΔCt whereΔΔCt = (Ct Bcl-xL- Ct β-actin for SKOV3 cells) − (Ct Bcl-xL- Ctβ-actin for SUMO1-siRNA transfected SKOV3 cells). PCR primers weredesigned from the Bcl-xL mRNA reference sequence, NM_138578.1:Bcl-xL forward-GCCACTTACCTGAATGACCAC, Bcl-xL reverse-TGGATGGTCAGTGTCTGGT (amplicon size-210 bp). P value was determined byusing the GraphPad Prism software, version 5.01.

2.12. Arsenic treatment

ATO (Sigma, USA) was dissolved in 1.65 M NaOH to make 50 mMstock solution. 1 μM solution was prepared by serial dilution inDMEM media and cells were treated for 1 h.

2.13. Statistical analysis

All the statistical comparisons were conducted by using GraphPadPrism version 5.01 (Graph Pad software). Data presented as meanvalue ± s.e. for experiments performed in replicates. A P value lessthan 0.05 is considered to be significant. Statistical analysis for all

the reporter gene assays was done by one way ANOVA. RQ-PCR anal-ysis and densitometric analysis was done by unpaired t-test.

3. Results

3.1. EVI1 is sumoylated both in vitro and in vivo by SUMO1

Sumoylation is a very dynamic process and it is very difficult to de-tect the modification, firstly because only a small fraction of the totalprotein is sumoylated and secondly in the cellular system several cyste-ine proteases cleave SUMO from its target [28]. Bioinformatically, it waspredicted earlier that EVI1 is a potent target of sumoylation [14].We ini-tially checked whether EVI1 is sumoylated by in vitro analysis. GST pu-rified EVI1 was incubated with E1 activating enzyme, E2 conjugatingenzyme and SUMO1 or SUMO1 (mut). The samples were run in 6%SDS-PAGE and analyzed by immunoblotting with anti-flag antibody.Sumoylation of a protein on a western blot is inferred by the presenceof a slowmigrating higher size band. A band of slower mobility was ob-served in EVI1-flag when incubated with SUMO1 (Fig. 1A, lane 2,marked by an arrow) and was absent in others. To determine whethersumoylation of EVI1 also occurs in vivo, HEK293T cells transfectedwith plasmids as indicated in the figure were lysed in 2× sample buffercontaining NEM. The sample was separated in SDS-PAGE and subjectedto immunoblotting with anti-flag antibody. We observed higher sizebands (marked by arrows) along with the 145 kDa band of EVI1, inlanes where EVI1 was co-transfected with Su Ub and SUMO1 (Fig. 1B,lane 2). Lower panel shows the sumoylated EVI1 band in the sameblot, stripped and reprobed with anti-HA antibody (Fig. 1B, lane 2). En-dogenous sumoylation of EVI1was checked in SKOV3 cell line by silenc-ing SUMO1. Immunoblotting analysis was performed with anti-EVI1antibody for the samples lysed in 2× sample buffer and loaded in twodifferent volumes in a 6% SDS-PAGE gel for 2 h in order to get a betterresolution of the separated bands and to get a proper demarcationbetween sumoylated EVI1 and MDS1-EVI1 (band present abovesumoylated EVI1). An additional slow migrating band was observed inboth the lanes, but the band intensity of lane 2 was less in comparisonto lane 1, as the amount of protein lysates loaded in lane 2 was lessthan lane 1 (Fig. 1C, lane1 and 2, upper panel). The higher sizesumoylated EVI1 band disappeared when the cells were transfectedwith SUMO1 siRNA (Fig. 1C, lane 3, upper panel) which was furtherconfirmed with anti-SUMO1 antibody (Fig. 1C, lower panel). Fig. 1Dshows the endogenous SUMO1 expression (upper panel) and actin(lower panel). Thus we show that EVI1 is sumoylated both in vitro aswell as in vivo.

3.2. Lysines-533, 698 and 874 are the major sumoylation sites in EVI1

Sumoylation is known to occur at lysine(s) embedded in a con-sensus motif ΨKXE/D [29]. Recent study also defines extended mo-tifs for sumoylation termed as PDSM (phosphorylation dependentsumoylation motif) and NDSM (negatively charged amino acid-dependent sumoylationmotif) [30,31]. Examination of protein sequenceof EVI1 by SUMOplot (http://www.abgent.com/tools/sumoplot) andSUMOsp 2.0 (http://sumosp.biocuckoo.org/online.php) revealed thepresence of core, conserved and extended SUMO motifs in EVI1 atlys-533, 698 and874with scores equal to or above 0.85 or high thresholdand were found to be evolutionarily conserved among different species(Fig. 2A). There are seven variants of EVI1 (H. sapiens) which differ intheir 5′UTR region. All the three sumoylation sites were also conservedamong all the seven variants of EVI1 (Fig. 2B). Interestingly, all thethree sites come under NDSM and lys-533 is a predicted PDSM site.To determine whether the predicted sites in EVI1 are the sites ofsumoylation we generated sumoylation deficient point mutants of EVI1and checked for the sumoylation status in vitro. Single mutants ofK533R or K698R or K874R failed to completely block sumoylation, dou-ble mutant of K533R and K698R termed as EVI1 (2KR)-flag partially

Fig. 1. SUMO1 modifies EVI1. (A) Purified EVI1-flag protein produced from pGEX-flag-EVI1 transformed BL21 competent cells by GST pull down method was subjected tosumoylation assay in the presence of E1, E2 and SUMO1/SUMO1-mut. Proteins were resolved in 6% SDS-PAGE and visualized by immunoblotting with anti-flag antibody. The mod-ified band is indicated with an arrow. (B) HEK293T cells transiently transfected with CMV-flag-EVI1, Su Ub (SUMO–Ubc9 fusion plasmid) and pCDNA3-SUMO1-HA was lyseddirectly in SDS gel loading buffer containing NEM. Proteins were separated in 6% SDS-PAGE gel and visualized by immunoblotting with anti-flag antibody/anti-HA antibody. Themodified bands are indicated with arrows. (C) For analyzing endogenous sumoylation of EVI1, SKOV3 cells with and without SUMO1-siRNA was lysed directly in 2× sample buffer,20 μl and 10 μl for SKOV3 and 20 μl for SUMO1 siRNA transfected SKOV3 samples were subjected to 6% SDS-PAGE gel and probed with anti-EVI1/anti-SUMO1 antibodies. (D) Thesamples were re-run in 14% SDS PAGE and visualized by immunoblotting with anti-SUMO1 (for SUMO1 expression) and anti-actin antibodies (loading control).

2360 S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

blocked the sumoylation (data not shown). Mutation of all the three ly-sines 533, 698 and 874 to arginine, termed as EVI1 (3KR)-flag abolishedthe slow migrating band (Fig. 2C, lane 4) which was present in thewild type EVI1 (lane 2). Thus, EVI1 was found to be a substrate forsumoylation at three positions, two in repression domain, as found inseveral other transcription factors and one in the acidic domain [14].

3.3. EVI1 and SUMO1 co-localizes in nuclear speckles

EVI1 existsmostly in a diffused patternwith approximately 5–10% ofcells showing nuclear speckles [32]. In transiently transfected HEK293Tcells and EVI1 positive SKOV3 cellswe observed approximately 5–10%ofcells with EVI1 in speckles. We found by indirect immunofluorescenceassay that EVI1 co-localized with SUMO1 in nuclear speckles whenoverexpressed in HEK293T cells (Fig. 3A, upper panel) as well as endog-enously in SKOV3 cells (Fig 3A, lower panel). Scatter plot of the mergedimages revealed co-localization rate equal to or above 85% and Pearson'scoefficient showed a positive association between EVI1 and SUMO1.We also checked whether sumoylation affects the localization pat-tern of sumoylation deficient EVI1 by indirect immunofluorescenceassay in HEK293T cells transfected with EVI1 (3KR)-flag and SUMO1-HA. The mutant was also found to be localized in nucleus but the co-localization of mutant EVI1 and SUMO1 was appreciably disrupted.Scatter plot result showed a co-localization rate of 20.15% andPearson's coefficient of 0.7636 (Fig. 3B). Thus it may be possiblethat sumoylation plays a role in translocating EVI1 from diffusedform to nuclear speckles.

3.4. Sumoylation of EVI1 interferes with its transcriptional activity andDNA binding ability

Sumoylation is known to bring a diverse effect on the activity ofseveral transcription factors, such as; it decreases the transcriptionalactivity of AR, c-Jun whereas it helps in the transactivation of Oct-4,PAX-6, DLX3 [6,18,33–35]. EVI1 activates Bcl-xL and SIRT1 promoteractivities [24,26]. In order to test the effect of sumoylated EVI1 onBcl-xL promoter, reporter gene assay was performed in HEK293T cellstransfected with plasmids indicated in figure along with reporter geneand Renilla constructs. A significant impairment in EVI1 mediated acti-vation was observed on Bcl-xL promoter when SUMO1 was added(Fig. 4A, bar 3 compared to bar 2). Negative regulatory effect imposedby sumoylation on EVI1 activitywas also observedwith SIRT1 promoter(Fig. 4B). Expressions of EVI1 and SUMO1 are shown by westernblotting. Results obtained from reporter gene assay of EVI1 withcontrol-siRNA and SUMO1-siRNA showed an increase in the activity ofEVI1 when HEK293T cells were transfected with SUMO1-siRNA(Fig. 4C, bar 4 compared to bar 2). Expressions of EVI1 and SUMO1 areshown by immunoblotting and actin is taken as the loading control.The results thus obtained show that sumoylation acts as a negative reg-ulator for EVI1. In order to further validate our data, transcriptional ac-tivity of wild type EVI1 was compared with the triple mutant EVI1(3KR) taking Bcl-xL promoter into account. The data revealed a minorincrease in promoter activity in presence of EVI1 (3KR) (Fig. 4D, bar4) in comparison to wild type EVI1 (Fig. 4D, bar 2). Reporter geneassay was also performed with empty promoter vector and pGL2-Bcl-xL together with EVI1 and EVI1 (3KR). No role of EVI1 or EVI1

Fig. 2. Lysines 533, 698 and 874 are themain acceptor sites for SUMO1 and are conserved among species. (A) The schematic diagram represents the three putative sumoylation sites presentin EVI1, two in repression domain and one in acidic domainswhich were found to be evolutionarily conserved among the different species. Table shows the scores equal to or above 0.85 orhigh threshold in EVI1, generated from the prediction softwares (SUMOplot and SUMOsp 2.0). (B) Bioinformatically deduced sumoylation sites are also conserved in all the variants of EVI1.The EVI1 sequences were obtained from theNCBI reference sequence database. (C) In vitro sumoylation assay was done by takingGST purified proteins of pGEX-flag-EVI1 and triplemutantpGEX-flag-EVI1 (3KR), incubated with kit reagents and checked by immunoblotting with anti-EVI1 antibody. A line diagram represents the site of mutation in EVI1.

2361S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

(3KR) was observed on the empty promoter vector (SupplementaryFig. 1). Since there was no significant difference in the activity of EVI1and EVI1 (3KR), we also checked the above by co-expressing SUMO1.As shown earlier, SUMO1 negatively regulated the activity of wildtype EVI1 (Fig. 4D, bar 3), however it failed to deactivate EVI1-3KR ac-tivity (Fig. 4D, bar 5) suggesting that the absence of sumoylation sitesprevent SUMO1 tomodify EVI1 and regulate its function. Similar resultswere obtained with SIRT1 promoter (Fig. 4E). Protein expressions ofEVI1 and SUMO1 are shown below the bar diagrams. The above data in-dicates that mutating sumoylation sites can modify the functions ofEVI1 and eventually can influence several downstream pathways con-trolled by EVI1. Further, to understand the role of sumoylated EVI1 weassessed the possible reason for sumoylation mediated negative reg-ulation. To this end, we checked the effect of sumoylation on the DNAbinding ability of EVI1 by ChIP assay in HEK293T cells transfected withEVI1 and Su Ub or SKOV3 cell line transfected with control-siRNA/SUMO1-siRNA. The binding specificity of EVI1 to Bcl-xL was previouslyreported by us [24]. In case of HEK293T cells there was a decrease inDNA binding on Bcl-xL promoter when EVI1 was co-transfected withSu Ub in comparison to only EVI1 (Fig. 4F, lane 3 and 2, left upperpanel) whereas no amplification was observed in the only vector andSu Ub. Right panel shows the expression of EVI1 and SuUb. EVI1 positiveSKOV3 cells were transfected with control-siRNA and SUMO1-siRNA.Cells from the same set of experiment were taken for ChIP, real time

PCR and immunoblotting experiments. In case of SKOV3, an increase inbinding of EVI1 to Bcl-xL promoter was observed in SUMO1-siRNAtransfected SKOV3 cells in comparison to control-siRNA transfectedSKOV3 cells (Fig. 4G, lanes 4 and 2, left upper panel). Left lower paneldepicts the input control. In order to accurately quantify the Bcl-xL ex-pression at mRNA level we performed real time PCRwith RNA extractedfrom the control and SUMO1-siRNA transfected cells. By down regulatingSUMO1 expression we observed almost three fold-up regulation inBcl-xL activity (Fig. 4H). Data was normalized to the endogenousβ-actin expression level. The protein expression of Bcl-xL in SUMO1-siRNA transfected cells was also found to be comparatively more thanin control-siRNA transfected SKOV3 cells (right panel). Protein expres-sions of EVI1, SUMO1 and actin as loading control are shown in theright panel. Thus, negative regulation imposed by sumoylated EVI1 onBcl-xL promoter is due to the decrease in DNA binding of EVI1 in thepresence of SUMO1. Non-sumoylated EVI1 indeed exhibits higher tran-scriptional activity compared to its sumoylated counterpart, correlatingwith its enhanced binding to the Bcl-xL promoter.

3.5. PIASy enhances sumoylation of EVI1 and regulates its function

SUMO E3 ligases function to promote the efficiency and substratespecificity of sumoylation. SUMO E3 ligases include the proteininhibitor of activated STAT family proteins (PIAS, which include

Fig. 3. EVI1 and SUMO1 reside in nuclear speckles. (A) Indirect immunofluorescence assay was performed in HEK293T cells transfected with CMV-flag-EVI1 andpCDNA3-SUMO1-HA, probed with anti-flag/anti-HA antibodies and EVI1 positive SKOV3cells with anti-EVI1/anti-SUMO1 antibodies and its corresponding Alexa fluor 488 and594 secondary antibodies. Merged picture reflects co-localization of EVI1 and SUMO1 in nuclear speckles. The images represent reconstruction of 10 optical sections (Z stack)obtained by confocal microscopy (Leica, Germany). Scatter plot data showed a significant co-localization which is above 85% (panel 5) in both the cell line. P. C, Pearson's coefficient;C. R, co-localization rate. (B) Indirect immunofluorescence assay were performed in HEK293T cells co-transfected with CMV-flag-EVI1 (3KR) with pCDNA3-SUMO1-HA by anti-flag/anti-HA antibodies and corresponding Alexa fluor 488 and 594 secondary antibodies. The images represent 10 optical sections (Z stack) reconstructed and obtained from confocalmicroscopy. Scatter plot data reflects the co-localization rate (C. R) and Pearson's coefficient (P. C) of EVI1 (3KR) with SUMO1. 630× (63× ∗ 10×) magnification was used for cap-turing all the images with confocal microscopy.

2362 S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

PIAS1, PIAS3, PIASXα, PIASXβ, and PIASy), polycomb group proteinPc2, and Ran binding protein 2 (RanBP2) [36]. PIAS proteins regulatethe transcriptional activity of several proteins through differentmechanisms [37,38]. To this end we found that EVI1 interactswith PIASy by co-immunoprecipitation assay. HEK293T cellstransfected with EVI1-myc and PIASy-flag showed that PIASy effi-ciently co-immunoprecipitated with EVI1 protein (Fig. 5A, panel 1,lane 3). We also found that EVI1 interacts with PIASy by reverseco-immunoprecipitation i.e. immunoprecipitated with anti-flag(PIASy) and western blotting with anti-EVI1 antibody (Fig. 5A,panel 3, lane 3). Panels 2 and 4 show the immunoprecipitated EVI1and PIASy proteins and panels 5 and 6 depict the input cell lysates.As PIAS protein family acts as SUMO ligase and helps in recruitingSUMO to target proteins, we checked whether PIASy has any role inEVI1 sumoylation. There was a comparatively more intense slow mi-grating band (indicated by an arrow) observed when EVI1-flag wasco-transfected with PIASy-flag, SUMO1-HA and Ubc9-flag (Fig. 5B,lane 3) in comparison to EVI1-flag co-transfected with SUMO1-HAand Ubc9-flag (Fig. 5B, lane 2). Lower panels showed the expressionof SUMO1, Ubc9 and PIASy. To check the functional relevance of the in-teraction between EVI1 and PIASy, reporter gene assay was performedwhich revealed that PIASy when co-expressed with EVI1 repressedthe activity of EVI1 on Bcl-xL promoter (Fig. 5C, bar 4). Since, SUMO1and PIASy both had a negative effect on EVI1 transcriptional activity;we checked the extent of negative regulation of PIASy and SUMO1 on

EVI1 activity. There was an enhanced impairment in promoter activitywhen PIASy and SUMO1 were co-transfected with EVI1 in comparisonto only EVI1 (Fig. 5C, bar 5 and 2). Thus our data shows that PIASy canaffect EVI1 activity both in presence or absence of SUMO1. In order tofurther elucidatewhether PIASy as a SUMO ligase is promoting the neg-ative regulation of EVI1 or it is a sumoylation independent function weperformed reporter gene assay by co-transfecting Bcl-xL promoter con-struct and PIASy-flag with EVI1-flag and EVI1(3KR)-flag. The luciferaseactivity showed that PIASy negatively regulated EVI1 (3KR) (Fig. 5D, bar5) similar to wild type EVI1 (Fig. 5D, bar 3) which suggest that PIASy isnot only a SUMO ligase for EVI1, but also acts as an interacting partnerwhich regulates EVI1 independent of its SUMO ligase activity. The pro-tein expressions of EVI1, EVI1 (3KR), SUMO1 and PIASy for reportergene assays in Fig. 5C and D are shown through western blotting anddepicted below the bar diagrams.

3.6. Arsenic enhances SUMO conjugation of EVI1 and negatively regulatesBcl-xL expression

ATO treatment iswell studied for curing APL, inwhich arsenic inducesthe formation of PML bodies through SUMOmodification which leads tothe degradation of PML and PML/RARα fusion protein. SKOV3 cells treat-ed with 1 μM ATO for 1 h were lysed and subjected to immunoblottingwith anti-EVI1 antibody. We found that the modified form of EVI1 wasenhanced after 1 h of ATO treatment to SKOV3 cells (Fig. 6A, lane 2,

2363S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

upper panel) which was confirmed with SUMO1 antibody (Fig. 6A, mid-dle panel). Also arsenic exposure enhanced EVI1–SUMO1 co-localizationto large nuclear bodies (Fig. 6B). It was strongly suggested that ATO me-diated PML sumoylation was found to be critical for PML targeting ontothe nuclear bodies and recruitment of NB-associated proteins, which inturn contribute to apoptosis induction or transcriptional regulation[39–41]. Co-localization study of EVI1 and PML in SKOV3 cells after 1 hof ATO treatment revealed that ATO treatment didn't recruit EVI1 toPML bodies (Fig. 6C) which suggests that EVI1 mediated functions or its

Fig. 4. Sumoylation negatively regulates EVI1 mediated transcriptional activation and DNAmoter plasmid pGL2-Bcl-xL/pGL4.10-SIRT1 promoter and pRL-TK (internal control) togethtions. After 24 h cells were lysed and luciferase activity was measured with luminometerwere analyzed by immunoblotting with anti-EVI1/anti-HA antibodies. Statistical significanceP value b0.01 (**) for EVI1 compared with EVI1 + SUMO1 in subpanel A; P value b0.001 (**assay performed by silencing endogenous SUMO1 in HEK293T cells to assess the activity of Eanalyzed by immunoblot with anti-EVI1/anti-SUMO1 antibodies. P value b0.05 (*) for EVtransfected with CMV-flag-NLS, CMV-flag-EVI1/CMV-flag-EVI1 (3KR) with or without ppRL-TK were lysed after 24 h and luciferase activity was measured. Expression levels of EP value b0.01 (**) for EVI1 compared with vector and EVI1 + SUMO1 compared with EVISu Ub (SUMO–Ubc9 fusion plasmid) and CMV-flag-EVI1 with Su Ub (SUMO–Ubc9 fusion plwas performed. Western blotting with anti-EVI1 and anti-SUMO1 shows the expression otransfected with control-siRNA or SUMO1-siRNA was subjected to ChIP, real time PCR andsamples. (H) RQ-PCR was performed with RNA isolated from SKOV3 cells transfected with cwere used for normalization. Statistical significance was assessed by unpaired t-test; P valuewith SUMO1-siRNA. The expression of EVI1, Bcl-xL, and SUMO1 in control-siRNA and SUMO1Actin is taken as loading control.

degradation is not directly related to PML nuclear bodies. As sumoylationnegatively regulated EVI1 mediated transactivation of Bcl-xL promoter,immunoblotting assay performed with the lysates of SKOV3 cells treatedwith 1 μMATO for 1 h showed a decrease in Bcl-xL protein expression incomparison to untreated SKOV3 cells (Fig. 6D, upper panel). Lower paneldepicts the ratio of densitometric analysis of Bcl-xL and actin proteinsplotted in graphical form. Thus, there are possibilities that the decreasein the Bcl-xL protein expression may be an effect imposed by enhancedsumoylated form of EVI1.

binding activity. (A–B) HEK293T cells were transiently transfected with luciferase pro-er with CMV-flag-NLS, CMV-flag-EVI1 and pCDNA3-SUMO1-HA in different combina-(GLOMAX 20/20, Promega, USA). Expression levels of transfected EVI1 and SUMO1was assessed by one way ANOVA; P value b0.001 (***) for EVI1 compared with vector,*) for EVI1 compared with vector and EVI1 + SUMO1 in subpanel B. (C) Reporter geneVI1 on Bcl-xL promoter. Expression levels of transfected EVI1 and silenced SUMO1 wereI1 with control siRNA compared to EVI1 with SUMO1-siRNA. (D–E) HEK293T cellsCNDA3-SUMO1-HA along with pGL2-Bcl-xL/pGL4.10-SIRT1 promoter construct andVI1 and SUMO1 were checked by immunoblotting with anti-EVI1/anti-HA antibodies.1 (3KR) + SUMO1. (F) HEK293T cells transfected with CMV-flag-NLS, CMV-flag-EVI1,asmid) was processed for ChIP assay with anti-EVI1 antibody. PCR for Bcl-xL promoterf EVI1 and Su Ub (SUMO–Ubc9 fusion protein) proteins. (G) Same set of SKOV3 cellsimmunoblotting. ChIP PCR for Bcl-xL promoter was performed with input and elutedontrol-siRNA or SUMO1-siRNA to check the relative Bcl-xL mRNA level. β-actin primersb0.01 (**) for SKOV3 cells transfected with control-siRNA compared to cells transfected-siRNA transfected with SKOV3 cells for ChIP and RQ-PCR is shown by western blotting.

Fig. 4 (continued).

2364 S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

Fig. 5. PIASy interacts with and enhances EVI1 sumoylation. (A) HEK293T cells were transfected with EVI1-myc, PIASy-flag and EVI1-myc together with PIASy-flag. Cells were lysedand immunoprecipitated with anti-EVI1/anti-flag antibodies followed by western blotting. (B) HEK293T cells were transfected with only CMV-flag-EVI1 or CMV-flag-EVI1 withSUMO1-HA, CMV-Ubc9-flag and with or without PIASy-flag. After 36 h cells were lysed and subjected to immunoblotting with anti-flag and anti-HA antibodies. (C–D) HEK293Tcells were transfected with pGL-2-Bcl-xL promoter construct, pRL-TK in different combination with CMV-flag-NLS, CMV-flag-EVI1/CMV-flag-EVI1 (3KR), PIASy-flag andSUMO1-HA. After 24 h of transfection cells were lysed and luciferase activity was measured. Protein expression for the transfected plasmids was checked by immunoblottingwith anti-EVI1, anti-flag and anti-HA antibodies. Statistical significance was assessed by one way ANOVA; P value b0.001(***) for EVI1 compared with vector and EVI1 comparedwith EVI1 + SUMO1 + PIASy in subpanel C; P value b0.01 (**) for EVI1 (3KR) compared with EVI1 (3KR) + PIASy in subpanel D. All the results represent the mean ± s.e. of ex-periments performed in triplicates.

2365S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

Fig. 6. ATO enhances sumoylation and co-localization of EVI1 with SUMO1. (A) SKOV3 cells treated with 1 μM of ATO for 1 h were subjected to immunoblotting with anti-EVI1 andanti-SUMO1 antibodies. Arrow shows the sumoylated band of EVI1. Actin is taken as loading control. (B) Indirect immunofluorescence assay was performed in untreated and ATO(1 μM, 1 h) treated SKOV3 cells with anti-EVI1/anti-SUMO1 antibodies and its corresponding Alexa fluor-488 and 594 secondary antibodies. The images were obtained by confocalmicroscopy. The magnification used for capturing the images is 630× (63× ∗ 10×). (C) Indirect immunofluorescence assay performed in untreated and ATO (1 μM, 1 h) treatedSKOV3 cells with anti-EVI1/anti-PML antibodies and its corresponding Alexa fluor-488 and 594 secondary antibodies. The images were obtained by confocal microscopy. The mag-nifications used for capturing the images are 1000× (100× ∗ 10×) for upper panel and 630× (63× ∗ 10×) for lower panel. (D) ATO treated SKOV3 cells were lysed and immuno-blotting was performed with anti-Bcl-xL antibody. Actin expression is taken as loading control. Bar diagram shows the graphical representation of western blotting by densitometricanalysis. Error bar represents ± s.e., asterisk (**) indicates P value b0.01 obtained by t-test.

2366 S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

4. Discussion

For proper functioning of cellular pathways and to dictate biologicalfunctions a fine tuning of the protein is strictly required. Several covalentmodifiers involved in controlling various cellular processes can offer asimple way to control the activity that can eventually regulate or deregu-late protein functions. Sumoylation is one of the modifications whichhave emerged as an important module regulating the function of variousproteins. EVI1 consists of activation/repression domain present in be-tween two sets of DNA binding domain which activates as well as re-presses several downstream targets. Bcl-xL, SIRT1, PLZF, GATA-2 areactivated by EVI1 whereas calreticulin and PTEN were found to be re-pressed [24,26,42–45]. Aberrant function of EVI1 imposed on down-stream targets is related to several forms of leukemia and otherdisorders. Thus, unveiling the role of sumoylation on EVI1 transcriptionalactivitymay help to understand the basic biology of the protein. The pres-ent study demonstrates that EVI1 is sumoylated at lysine residues 533,698 & 874 and modification of the sites can abrogate EVI1 sumoylation.

SUMOmodification iswell known for regulating sub-cellular distribu-tion and nuclear trafficking of proteins and is associated with nuclearspeckles which are known to be hot spots for functionally active proteins

[32,46]. It is already known that PML requires sumoylation for its nuclearbody formation which is disrupted in acute promyelocytic leukemia[47,48]. Confocal imaging revealed that EVI1 was co-localized withSUMO1. However, co-localizationwas disruptedwhen the lysine residuesthat underwent sumoylation were modified to arginine. This shows thatlike acetylation, sumoylation is indispensible for discrete cellular localiza-tion and also plays a role in regulating the function of EVI1 by controllingtranslocation from diffused to speckle pattern within the nucleus.

Bcl-xL was found to be activated by EVI1. Alteration in the cellularlevel of pro and anti-apoptotic Bcl-2 family proteins is related to tumori-genesis in both transgenicmodels and clinical cases [49]. As inmost of thecases sumoylation acts as a repressor command for the protein, we ob-served the same phenomenon with EVI1. Sumoylated EVI1 significantlyimpaired Bcl-xL promoter activity, whereas when the sumoylation siteswere mutated or SUMO1 was silenced, the EVI1 mediated activationwas restored. Thus the results obtained clearly show that sumoylatedform of EVI1 affects the extent of activation imposed by unmodifiedEVI1. Reporter gene assay performed with sumoylation mutants showedthat the three putative sumoylation sites found in EVI1 are important forthe proper functioning of EVI1 protein in the cellular system. In severalproteins like SOX2, TEL sumoylation affects the DNA binding ability of

2367S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

protein thereby imposing its effect on transcriptional activity [12,13]. Ourdata also shows that sumoylation decreased the DNA binding ability ofEVI1 which may be a possible reason for negative regulation ofsumoylation on EVI1 function. As the sumoylation sites were found in re-pression domain and acidic domain and not in DNA binding domain, it islikely that sumoylation causes some conformational changes in EVI1, asfound in thymine DNA glycosylase (TDG) which affects its DNA bindingability [50]. Itmay be possible that by recruitingHDAC or bringing confor-mational changes, sumoylated EVI1 can favor transcriptional repression.It was shown earlier that EVI1 regulates apoptosis by activating Bcl-xL,an anti-apoptotic gene, and the effect is impaired by acetylation. Regula-tion of apoptosis is a reflection of the changes in the level of pro andanti-apoptotic genes. Since sumoylation regulates the EVI1 mediated ac-tivity on Bcl-xL, we hypothesize that sumoylation can regulate apoptosisby regulating EVI1.

Taking SUMO ligases into consideration, in context of enhancementof EVI1 sumoylation we found PIASy to be a new interacting partnerwhich not only interacts with EVI1 but also enhanced its sumoylation.Although PIASy acted as a ligase and enhanced sumoylation of EVI1,the major limiting factor for modification to occur was SUMO1. Ourdata also revealed that PIASy itself negatively regulated EVI1 mediatedtransactivation both with and without SUMO1. We showed that PIASyefficiently repressed the transcriptional activity of mutant form similarto wild type EVI1, which implies that the negative regulation imposedby PIASy is independent of its ligase function. Thus like LEF1[51], wefound the dual regulatory role of PIASy, first, by acting as a SUMO ligaseand enhancing EVI1 sumoylation; and second as a co-regulator inde-pendent of its ligase activity.

ATO affects cells through various mechanisms, influencing severalpathwayswhich results in vast range of cellular processes like apoptosisinduction, angiogenesis inhibition, autophagy and growth inhibition.The response varies fromone cell type to the other. ATO is known to tar-get AME (AML1/MDS1/EVI1) via MDS1 and EVI1 and degrades EVI1through ubiquitin–proteosomal pathway [52]. Three out of five MDSpatients with high EVI1 expression showed a good response whentreated with ATO and thalidomide [53]. We found that ATO treatmentnot only enhanced sumoylated EVI1 but also drives nucleoplasmic frac-tion of EVI1 to large nuclear dots distinct from PML bodieswith SUMO1.As Bcl-xL downregulation was also observed after ATO treatment, itshows that depending on the ratio of modified and unmodified formof EVI1 in the cellular pool the cells' fate is determined.

4.1. Conclusion

Various protein modification and cellular pathways deregulate thecell which leads to neoplastic transformation. Despite its role in severalsolid tumors and hematopoietic malignancies, the regulators thatgovern the function of EVI1 are partially known. Our study for thefirst time shows that EVI1 is post-translationally modified by SUMO1which negatively regulates the function of EVI1. Present study suggeststhat unmodified form of EVI1 protects cells from apoptosis whereas itssumoylated counterpart significantly impairs the activation imposed byEVI1 on Bcl-xL which may interfere with the EVI1 mediated regulationof apoptosis. Thus, the study revealed sumoylation as a new regulatorof EVI1 and shows that the sumoylated and desumoylated form ofEVI1 behave differently in their transcriptional activity. It will be inter-esting to further understand the role of sumoylated/desumoylated EVI1in context of cellular transformation.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.bbamcr.2013.06.003.

Acknowledgement

The authors would like to thank Dr. Rudolf Grosschedl, Max PlanckInstitute for Immunobiology and Epigenetics, Germany for the expres-sion plasmids for PIASy-flag, Dr. F. Melchior, Heidelberg University,

Germany for the pTET23a-Ubc9 plasmid, Dr. R. Hay, University ofDundee, Scotland, UK for the SUMO1-HA expression plasmid and Dr.Jin-Hyun Ahn, Samsung Biomedical Research Institute, Korea for theSUMO-Ubc9 (Su Ub) fusion plasmid. The authors also extend theirthanks to Ms. Nivedita Kuila and Mr. Bhabani Shankar Sahoo for theirtechnical support. This work was supported by grant-in aid from theDepartment of Biotechnology (DBT), Govt. of India to SC and also partlyby the Institutional core grant. SS is supported by fellowship from theUniversity Grants Commission, India. AKP is supported by fellowshipfrom Council of Scientific and Industrial Research, India. The fundershave no role in the designing of the study, data collection and analysis,and decision to publish the work or manuscript preparation.

References

[1] Y. Liu, L. Chen, T.C. Ko, A.P. Fields, E.A. Thompson, Evi1 is a survival factor whichconveys resistance to both TGFbeta- and taxol-mediated cell death via PI3K/AKT,Oncogene 25 (2006) 3565–3575.

[2] D.M. Smith, S. Patel, F. Raffoul, E. Haller, G.B. Mills, M. Nanjundan, Arsenic trioxideinduces a beclin-1-independent autophagic pathway via modulation of SnoN/SkiLexpression in ovarian carcinoma cells, Cell Death Differ. 17 (2010) 1867–1881.

[3] N. Yamakawa, K. Kaneda, Y. Saito, E. Ichihara, K. Morishita, The increased expres-sion of integrin alpha6 (ITGA6) enhances drug resistance in EVI1(high) leukemia,PLoS One 7 (2012) e30706.

[4] L. Laricchia-Robbio, K. Premanand, C.R. Rinaldi, G. Nucifora, EVI1 impairsmyelopoiesis by deregulation of PU.1 function, Cancer Res. 69 (2009) 1633–1642.

[5] M. Nanjundan, Y. Nakayama, K.W. Cheng, J. Lahad, J. Liu, K. Lu, W.L. Kuo, K.Smith-McCune, D. Fishman, J.W. Gray, G.B. Mills, Amplification of MDS1/EVI1and EVI1, located in the 3q26.2 amplicon, is associated with favorable patientprognosis in ovarian cancer, Cancer Res. 67 (2007) 3074–3084.

[6] F. Wei, H.R. Scholer, M.L. Atchison, Sumoylation of Oct4 enhances its stability,DNA binding, and transactivation, J. Biol. Chem. 282 (2007) 21551–21560.

[7] C. Kho, A. Lee, D. Jeong, J.G. Oh, A.H. Chaanine, E. Kizana, W.J. Park, R.J. Hajjar,SUMO1-dependent modulation of SERCA2a in heart failure, Nature 477 (2011)601–605.

[8] V.G. Wilson, D. Rangasamy, Intracellular targeting of proteins by sumoylation,Exp. Cell Res. 271 (2001) 57–65.

[9] Y. Morita, C. Kanei-Ishii, T. Nomura, S. Ishii, TRAF7 sequesters c-Myb to the cyto-plasm by stimulating its sumoylation, Mol. Biol. Cell 16 (2005) 5433–5444.

[10] A. Pichler, F. Melchior, Ubiquitin-related modifier SUMO1 and nucleocytoplasmictransport, Traffic 3 (2002) 381–387.

[11] J. Song, L.K. Durrin, T.A. Wilkinson, T.G. Krontiris, Y. Chen, Identification of aSUMO-binding motif that recognizes SUMO-modified proteins, Proc. Natl. Acad.Sci. U. S. A. 101 (2004) 14373–14378.

[12] S. Tsuruzoe, K. Ishihara, Y. Uchimura, S. Watanabe, Y. Sekita, T. Aoto, H. Saitoh, Y.Yuasa, H. Niwa, M. Kawasuji, H. Baba, M. Nakao, Inhibition of DNA binding of Sox2by the SUMO conjugation, Biochem. Biophys. Res. Commun. 351 (2006) 920–926.

[13] M.G. Roukens, M. Alloul-Ramdhani, A.C. Vertegaal, Z. Anvarian, C.I. Balog, A.M.Deelder, P.J. Hensbergen, D.A. Baker, Identification of a new site of sumoylationon Tel (ETV6) uncovers a PIAS-dependent mode of regulating Tel function, Mol.Cell. Biol. 28 (2008) 2342–2357.

[14] A. Verger, J. Perdomo, M. Crossley, Modification with SUMO. A role in transcrip-tional regulation, EMBO Rep. 4 (2003) 137–142.

[15] S.S. Kwek, J. Derry, A.L. Tyner, Z. Shen, A.V. Gudkov, Functional analysis and intracel-lular localization of p53 modified by SUMO-1, Oncogene 20 (2001) 2587–2599.

[16] A. Valin, G. Gill, Regulation of the dual-function transcription factor Sp3 by SUMO,Biochem. Soc. Trans. 35 (2007) 1393–1396.

[17] I. Nishikata, S. Nakahata, Y. Saito, K. Kaneda, E. Ichihara, N. Yamakawa, K.Morishita, Sumoylation of MEL1S at lysine 568 and its interaction with CtBP facil-itates its repressor activity and the blockade of G-CSF-induced myeloid differen-tiation, Oncogene 30 (2011) 4194–4207.

[18] H. Poukka, U. Karvonen, O.A. Janne, J.J. Palvimo, Covalent modification of the an-drogen receptor by small ubiquitin-like modifier 1 (SUMO-1), Proc. Natl. Acad.Sci. U. S. A. 97 (2000) 14145–14150.

[19] K. Bettermann, M. Benesch, S. Weis, J. Haybaeck, SUMOylation in carcinogenesis,Cancer Lett. 316 (2012) 113–125.

[20] V. Dorval, P.E. Fraser, SUMO on the road to neurodegeneration, Biochim. Biophys.Acta 1773 (2007) 694–706.

[21] X.W. Zhang, X.J. Yan, Z.R. Zhou, F.F. Yang, Z.Y. Wu, H.B. Sun, W.X. Liang, A.X. Song,V. Lallemand-Breitenbach, M. Jeanne, Q.Y. Zhang, H.Y. Yang, Q.H. Huang, G.B.Zhou, J.H. Tong, Y. Zhang, J.H. Wu, H.Y. Hu, H. de The, S.J. Chen, Z. Chen, Arsenictrioxide controls the fate of the PML-RARalpha oncoprotein by directly bindingPML, Science 328 (2010) 240–243.

[22] B. Kong, S. Huang, W. Wang, D. Ma, X. Qu, J. Jiang, X. Yang, Y. Zhang, B. Wang, B.Cui, Q. Yang, Arsenic trioxide induces apoptosis in cisplatin-sensitive and-resistant ovarian cancer cell lines, Int. J. Gynecol. Cancer 15 (2005) 872–877.

[23] M.J. Shim, H.J. Kim, S.J. Yang, I.S. Lee, H.I. Choi, T. Kim, Arsenic trioxide induces ap-optosis in chronic myelogenous leukemia K562 cells: possible involvement of p38MAP kinase, J. Biochem. Mol. Biol. 35 (2002) 377–383.

[24] A.K. Pradhan, A.D. Mohapatra, K.B. Nayak, S. Chakraborty, Acetylation of theproto-oncogene EVI1 abrogates Bcl-xL promoter binding and induces apoptosis,PLoS One 6 (2011) e25370.

2368 S. Singh et al. / Biochimica et Biophysica Acta 1833 (2013) 2357–2368

[25] A. Shimahara, N. Yamakawa, I. Nishikata, K. Morishita, Acetylation of lysine 564adjacent to the C-terminal binding protein-binding motif in EVI1 is crucial fortranscriptional activation of GATA2, J. Biol. Chem. 285 (2010) 16967–16977.

[26] A.K. Pradhan, N. Kuila, S. Singh, S. Chakraborty, EVI1 up-regulates the stress re-sponsive gene SIRT1 which triggers deacetylation and degradation of EVI1,Biochim. Biophys. Acta 1809 (2011) 269–275.

[27] E.T. Kim, K.K. Kim, M.J. Matunis, J.H. Ahn, Enhanced SUMOylation of proteinscontaining a SUMO-interacting motif by SUMO-Ubc9 fusion, Biochem. Biophys.Res. Commun. 388 (2009) 41–45.

[28] F. Melchior, M. Schergaut, A. Pichler, SUMO: ligases, isopeptidases and nuclearpores, Trends Biochem. Sci. 28 (2003) 612–618.

[29] R.T. Hay, SUMO: a history of modification, Mol. Cell 18 (2005) 1–12.[30] V. Hietakangas, J. Anckar, H.A. Blomster, M. Fujimoto, J.J. Palvimo, A. Nakai, L.

Sistonen, PDSM, a motif for phosphorylation-dependent SUMO modification,Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 45–50.

[31] S.H. Yang, A. Galanis, J.Witty, A.D. Sharrocks, An extended consensusmotif enhancesthe specificity of substrate modification by SUMO, EMBO J. 25 (2006) 5083–5093.

[32] S. Chakraborty, V. Senyuk, S. Sitailo, Y. Chi, G. Nucifora, Interaction of EVI1 withcAMP-responsive element-binding protein-binding protein (CBP) and p300/CBP-associated factor (P/CAF) results in reversible acetylation of EVI1 and in co-localization in nuclear speckles, J. Biol. Chem. 276 (2001) 44936–44943.

[33] Q. Yan, L. Gong, M. Deng, L. Zhang, S. Sun, J. Liu, H. Ma, D. Yuan, P.C. Chen, X. Hu, J.Liu, J. Qin, L. Xiao, X.Q. Huang, J. Zhang, D.W. Li, Sumoylation activates the tran-scriptional activity of Pax-6, an important transcription factor for eye and braindevelopment, Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 21034–21039.

[34] O. Duverger, S.X. Chen, D. Lee, T. Li, P.B. Chock, M.I. Morasso, SUMOylation ofDLX3 by SUMO1 promotes its transcriptional activity, J. Cell. Biochem. 112(2011) 445–452.

[35] S.Muller, M. Berger, F. Lehembre, J.S. Seeler, Y. Haupt, A. Dejean, c-Jun and p53 activ-ity is modulated by SUMO-1 modification, J. Biol. Chem. 275 (2000) 13321–13329.

[36] R. Geiss-Friedlander, F. Melchior, Concepts in sumoylation: a decade on, Nat. Rev.Mol. Cell Biol. 8 (2007) 947–956.

[37] T. Kubota, M. Matsuoka, S. Xu, N. Otsuki, M. Takeda, A. Kato, K. Ozato, PIASy in-hibits virus-induced and interferon-stimulated transcription through distinctmechanisms, J. Biol. Chem. 286 (2011) 8165–8175.

[38] K. Shuai, B. Liu, Regulation of gene-activation pathways by PIAS proteins in theimmune system, Nat. Rev. Immunol. 5 (2005) 593–605.

[39] V. Lallemand-Breitenbach, J. Zhu, F. Puvion, M. Koken, N. Honore, A.Doubeikovsky, E. Duprez, P.P. Pandolfi, E. Puvion, P. Freemont, H. de The, Roleof promyelocytic leukemia (PML) sumolation in nuclear body formation, 11Sproteasome recruitment, and As2O3-induced PML or PML/retinoic acid receptoralpha degradation, J. Exp. Med. 193 (2001) 1361–1371.

[40] J. Zhu, M.H. Koken, F. Quignon, M.K. Chelbi-Alix, L. Degos, Z.Y. Wang, Z. Chen, H.de The, Arsenic-induced PML targeting onto nuclear bodies: implications for the

treatment of acute promyelocytic leukemia, Proc. Natl. Acad. Sci. U. S. A. 94(1997) 3978–3983.

[41] G.J. Block, C.H. Eskiw, G. Dellaire, D.P. Bazett-Jones, Transcriptional regulation isaffected by subnuclear targeting of reporter plasmids to PML nuclear bodies,Mol. Cell. Biol. 26 (2006) 8814–8825.

[42] S. Takahashi, J.D. Licht, The human promyelocytic leukemia zinc finger gene isregulated by the Evi-1 oncoprotein and a novel guanine-rich site binding protein,Leukemia 16 (2002) 1755–1762.

[43] T. Sato, S. Goyama, E. Nitta, M. Takeshita, M. Yoshimi, M. Nakagawa, M. Kawazu,M. Ichikawa, M. Kurokawa, Evi-1 promotes para-aortic splanchnopleural hemato-poiesis through up-regulation of GATA-2 and repression of TGF-b signaling,Cancer Sci. 99 (2008) 1407–1413.

[44] Y. Qiu, J. Lynch, L. Guo, B. Yatsula, A.S. Perkins, M. Michalak, Regulation of thecalreticulin gene by GATA6 and Evi-1 transcription factors, Biochemistry 47(2008) 3697–3704.

[45] A. Yoshimi, S. Goyama, N. Watanabe-Okochi, Y. Yoshiki, Y. Nannya, E. Nitta, S.Arai, T. Sato, M. Shimabe, M. Nakagawa, Y. Imai, T. Kitamura, M. Kurokawa, Evi1represses PTEN expression and activates PI3K/AKT/mTOR via interactions withpolycomb proteins, Blood 117 (2011) 3617–3628.

[46] Y. Wu, L. Wang, P. Zhou, G. Wang, Y. Zeng, Y. Wang, J. Liu, B. Zhang, S. Liu, H. Luo,X. Li, Regulation of REGgamma cellular distribution and function by SUMO mod-ification, Cell Res. 21 (2011) 807–816.

[47] T. Sternsdorf, K. Jensen, B. Reich, H. Will, The nuclear dot protein sp100, characteri-zation of domains necessary for dimerization, subcellular localization, andmodifica-tion by small ubiquitin-like modifiers, J. Biol. Chem. 274 (1999) 12555–12566.

[48] S. Zhong, S. Muller, S. Ronchetti, P.S. Freemont, A. Dejean, P.P. Pandolfi, Role ofSUMO-1-modified PML in nuclear body formation, Blood 95 (2000) 2748–2752.

[49] J.C. Reed, Bcl-2-family proteins and hematologic malignancies: history and futureprospects, Blood 111 (2008) 3322–3330.

[50] C. Smet-Nocca, J.M. Wieruszeski, H. Leger, S. Eilebrecht, A. Benecke, SUMO-1regulates the conformational dynamics of thymine-DNA glycosylase regulatorydomain and competes with its DNA binding activity, BMC Biochem. 12 (2011) 4.

[51] S. Sachdev, L. Bruhn, H. Sieber, A. Pichler, F. Melchior, R. Grosschedl, PIASy, a nu-clear matrix-associated SUMO E3 ligase, represses LEF1 activity by sequestrationinto nuclear bodies, Genes Dev. 15 (2001) 3088–3103.

[52] D. Shackelford, C. Kenific, A. Blusztajn, S. Waxman, R. Ren, Targeted degradationof the AML1/MDS1/EVI1 oncoprotein by arsenic trioxide, Cancer Res. 66 (2006)11360–11369.

[53] A. Raza, S. Buonamici, L. Lisak, S. Tahir, D. Li, M. Imran, N.I. Chaudary, H. Pervaiz,J.A. Gallegos, M.I. Alvi, M. Mumtaz, S. Gezer, P. Venugopal, P. Reddy, N. Galili, A.Candoni, J. Singer, G. Nucifora, Arsenic trioxide and thalidomide combination pro-duces multi-lineage hematological responses in myelodysplastic syndromes pa-tients, particularly in those with high pre-therapy EVI1 expression, Leuk. Res.28 (2004) 791–803.