Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

SUMO-1 Modification Regulates the DNA Binding Activity of Heat Shock Transcription Factor 2, a Promyelocytic Leukemia Nuclear Body Associated Transcription Factor

2001; Elsevier BV; Volume: 276; Issue: 21 Linguagem: Inglês

10.1074/jbc.m008066200

ISSN

1083-351X

Autores

Michael L. Goodson, Yiling Hong, Richard S. Rogers, Michael J. Matunis, Ok-Kyong Park-Sarge, Kevin D. Sarge,

Tópico(s)

Endoplasmic Reticulum Stress and Disease

Resumo

Heat shock transcription factor 2 (HSF2) is a transcription factor that regulates heat shock protein gene expression, but the mechanisms regulating the function of this factor are unclear. Here we report that HSF2 is a substrate for modification by the ubiquitin-related protein SUMO-1 and that HSF2 colocalizes in cells with SUMO-1 in nuclear granules. Staining with anti-promyelocytic leukemia antibodies indicates that these HSF2-containing nuclear granules are PML bodies. Our results identify lysine 82 as the major site of SUMO-1 modification in HSF2, which is located in a "wing" within the DNA-binding domain of this protein. Interestingly, SUMO-1 modification of HSF2 results in conversion of this factor to the active DNA binding form. This is the first demonstration that SUMO-1 modification can directly alter the DNA binding ability of a transcription factor and reveals a new mechanism by which SUMO-1 modification can regulate protein function. Heat shock transcription factor 2 (HSF2) is a transcription factor that regulates heat shock protein gene expression, but the mechanisms regulating the function of this factor are unclear. Here we report that HSF2 is a substrate for modification by the ubiquitin-related protein SUMO-1 and that HSF2 colocalizes in cells with SUMO-1 in nuclear granules. Staining with anti-promyelocytic leukemia antibodies indicates that these HSF2-containing nuclear granules are PML bodies. Our results identify lysine 82 as the major site of SUMO-1 modification in HSF2, which is located in a "wing" within the DNA-binding domain of this protein. Interestingly, SUMO-1 modification of HSF2 results in conversion of this factor to the active DNA binding form. This is the first demonstration that SUMO-1 modification can directly alter the DNA binding ability of a transcription factor and reveals a new mechanism by which SUMO-1 modification can regulate protein function. Ran GTPase-activating protein 1 heat shock transcription factor 1 heat shock transcription factor 2 promyelocytic leukemia 4′,6-diamidino-2-phenylindole polymerase chain reaction heat shock protein ubiquitin-activating enzyme ubiquitin carrier protein ubiquitin-protein isopeptide ligase green fluorescent protein glutathione S-transferase SUMO-1-activating enzyme 1 SUMO-1-activating enzyme 2 bovine serum albumin fetal bovine serum phosphate-buffered saline Dulbecco's modified Eagle's medium polyacrylamide gel electrophoresis In the past several years, a number of reports have described the covalent attachment of several proteins similar to ubiquitin in their ability to target proteins in the cell. The best studied of these is SUMO-1, a 97-amino acid, 11-kDa polypeptide, which shares 18% amino acid sequence identity with ubiquitin. Originally SUMO-1 was identified as a modifier of the Ran GTPase activating protein (RanGAP1)1 (1Matunis M.J. Coutavas E. Blobel G. J. Cell Biol. 1996; 135: 1457-1470Crossref PubMed Scopus (955) Google Scholar, 2Mahajan R. Delphin C. Guan T. Gerace L. Melchior F. Cell. 1997; 88: 97-107Abstract Full Text Full Text PDF PubMed Scopus (1003) Google Scholar). Unlike ubiquitin, SUMO-1 modification of proteins does not appear to target proteins for degradation and instead appears to have a number of other functions in the cell (3Saitoh H. Pu R.T. Dasso M. Trends Biochem. Sci. 1997; 22: 374-376Abstract Full Text PDF PubMed Scopus (125) Google Scholar, 4Hodges M. Tissot C. Freemont P.S. Curr. Biol. 1998; 8: R749-R752Abstract Full Text Full Text PDF PubMed Google Scholar, 5Kretz-Remy C. Tanguay R.M. Biochem. Cell Biol. 1999; 77: 299-309Crossref PubMed Scopus (31) Google Scholar, 6Yeh E.T. Gong L. Kamitani T. Gene. 2000; 248: 1-14Crossref PubMed Scopus (415) Google Scholar). For RanGAP-1, SUMO-1 modification is required for its association with Nup358 or RanBP2 (Ran-binding proteins in the nuclear pore complex) and localization to the nuclear pore complex (7Mahajan R. Gerace L. Melchior F. J. Cell Biol. 1998; 140: 259-270Crossref PubMed Scopus (238) Google Scholar, 8Matunis M.J. Wu J. Blobel G. J. Cell Biol. 1998; 140: 499-509Crossref PubMed Scopus (377) Google Scholar). SUMO-1 modification of PML is correlated with localization of this protein to nuclear bodies, which are discrete subdomains within the nucleus (9Sternsdorf T. Jensen K. Will H. J. Cell Biol. 1997; 139: 1621-1634Crossref PubMed Scopus (289) Google Scholar, 10Muller S. Matunis M.J. Dejean A. EMBO J. 1998; 17: 61-70Crossref PubMed Scopus (578) Google Scholar, 11Kamitani T. Nguyen H.P. Kito K. Fukuda-Kamitani T. Yeh E.T. J. Biol. Chem. 1998; 273: 3117-3120Abstract Full Text Full Text PDF PubMed Scopus (177) Google Scholar, 12Duprez E. Saurin A.J. Desterro J.M. Lallemand-Breitenbach V. Howe K. Boddy M.N. Solomon E. de The H. Hay R.T. Freemont P.S. J. Cell Sci. 1999; 112: 381-393Crossref PubMed Google Scholar, 13Zhong S. Muller S. Ronchetti S. Freemont P.S. Dejean A. Pandolfi P.P. Blood. 2000; 95: 2748-2752Crossref PubMed Google Scholar). Like ubiquitin, SUMO-1 uses a multi enzyme ligase complex to attach to target proteins, but the specific enzymes are distinct from those involved in ubiquitination. The processed SUMO-1 is a substrate for the SUMO E1 enzyme, which is a heterodimer of two proteins calledSUMO-1-activating enzyme 1 (SAE1) and SUMO-1-activating enzyme 2 (SAE2) (14Desterro J.M. Rodriguez M.S. Kemp G.D. Hay R.T. J. Biol. Chem. 1999; 274: 10618-10624Abstract Full Text Full Text PDF PubMed Scopus (284) Google Scholar, 15Okuma T. Honda R. Ichikawa G. Tsumagari N. Yasuda H. Biochem. Biophys. Res. Commun. 1999; 254: 693-698Crossref PubMed Scopus (183) Google Scholar). Ubc9 is the E2 for SUMO-1 conjugation, receiving the SUMO-1 from SAE1/2 and transferring it to the target protein (15Okuma T. Honda R. Ichikawa G. Tsumagari N. Yasuda H. Biochem. Biophys. Res. Commun. 1999; 254: 693-698Crossref PubMed Scopus (183) Google Scholar, 16Desterro J.M. Thomson J. Hay R.T. FEBS Lett. 1997; 417: 297-300Crossref PubMed Scopus (303) Google Scholar). All SUMO-1-modified proteins characterized to date interact with Ubc9 directly, suggesting that SUMO-1 does not require a separate E3 ligase for specificity (12Duprez E. Saurin A.J. Desterro J.M. Lallemand-Breitenbach V. Howe K. Boddy M.N. Solomon E. de The H. Hay R.T. Freemont P.S. J. Cell Sci. 1999; 112: 381-393Crossref PubMed Google Scholar, 17Lee G.W. Melchior F. Matunis M.J. Mahajan R. Tian Q Anderson P. J. Biol. Chem. 1998; 273: 6503-6507Abstract Full Text Full Text PDF PubMed Scopus (119) Google Scholar, 18Kim Y.H. Choi C.Y. Kim Y. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 12350-12355Crossref PubMed Scopus (143) Google Scholar, 19Poukka H. Aarnisalo P. Karvonen U. Palvimo J.J. Janne O.A. J. Biol. Chem. 1999; 274: 19441-19446Abstract Full Text Full Text PDF PubMed Scopus (156) Google Scholar). Supporting this idea is the finding that SUMO-1 modification can be reconstituted in vitro with only ATP, SUMO-1, SAE1, SAE2, Ubc9, and the target protein (12Duprez E. Saurin A.J. Desterro J.M. Lallemand-Breitenbach V. Howe K. Boddy M.N. Solomon E. de The H. Hay R.T. Freemont P.S. J. Cell Sci. 1999; 112: 381-393Crossref PubMed Google Scholar, 17Lee G.W. Melchior F. Matunis M.J. Mahajan R. Tian Q Anderson P. J. Biol. Chem. 1998; 273: 6503-6507Abstract Full Text Full Text PDF PubMed Scopus (119) Google Scholar). Heat shock transcription factor 1 (HSF1) and heat shock transcription factor 2 (HSF2) are transcription factors that regulate the expression of heat shock protein (hsp) genes (20Morano K.A. Thiele D.J. Gene Expr. 1999; 17: 271-282Google Scholar, 21Cotto J.J. Morimoto R.I. Biochem. Soc. Symp. 1999; 64: 105-118PubMed Google Scholar). HSF1 DNA-binding is activated in response to cell stress, but the signals and mechanisms that regulate HSF2 function are not yet clear. In this study, we show that HSF2 is a substrate for SUMO-1 and SUMO-2 modification in vitro and identify lysine 82 as the primary site of SUMO-1 modification. Furthermore, we show that HSF2 colocalizes with SUMO-1 in nuclear domain structures, and that these HSF2-containing structures are PML bodies. SUMO modification of HSF2 results in a significant increase in DNA binding activity of this protein. Thus, it appears that SUMO-1 modification regulates HSF2 function by modulating the DNA binding activity and possibly also the subcellular localization of this transcription factor. The yeast-two-hybrid vector pGBD-HSF2 was cloned as previously described (22Hong Y. Sarge K.D. J. Biol. Chem. 1999; 274: 12967-12970Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). Polymerase chain reaction (PCR) was used to generate BglII sites immediately before and after the open reading frame of the mouse HSF2β cDNA. TheBglII-digested PCR fragment of HSF2β was cloned into theBamHI site of pQE9 (Qiagen), thus generating pQE9-HSF2β. The pGEX-SUMO-1 plasmid was a generous gift of Dr. Joana Desterro (16Desterro J.M. Thomson J. Hay R.T. FEBS Lett. 1997; 417: 297-300Crossref PubMed Scopus (303) Google Scholar). PCR was used to generate a SalI site and a Kozak consensus sequence (5′-CCACC-3′) immediately before and a ClaI site immediately after the open reading frame of the mouse HSF2β cDNA. This undigested PCR fragment of HSF2β was cloned into theSmaI site of the pGEM-7Z cloning vector (Promega), in which the ClaI site had been destroyed, thus generating the plasmid pGEM-HSF2βSC. The plasmid pcDNA-HSF2β-MH6was cloned by digesting pGEM-HSF2βSC with SalI andHindIII to liberate the majority of the HSF2 ORF and cloning it into pcDNA3.1/MycHisA(−) (Invitrogen) digested withXhoI and HindIII. The remaining portion of the HSF2β open reading frame was cloned by PCR using primers that spanned the endogenous HindIII site in HSF2 and added aHindIII site immediately 5′ to the endogenous stop codon. The HindIII-digested PCR fragment was cloned into theHindIII site of the previous construct, and orientation of the insert was verified using PCR. The insert for pEGFP-HSF2β was generated by digesting pGEM-HSF2βSC with ClaI, filling the resulting ends in with Klenow DNA polymerase (New England Biolabs, Beverly, MA), and digesting with SalI. The insert was then cloned into pEGFP-C1 (CLONTECH) digested withSalI and SmaI to create pEGFP-HSF2β. The Saccharomyces cerevisiae strain PJ 69–4A (MATa trp 1–901 leu2–3, 112 ura3–52 ade2–101 his3–200 gal4Δ gal80Δ LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ) were cotransformed with yeast GAL4 DNA-binding domain fusion plasmids (bait) and VP16 activation domain fusion plasmids and plated onto yeast minimal selective medium as previously described (23James P. Halladay J. Craig E.A. Genomics. 1996; 144: 1425-1436Google Scholar). Colonies were transferred onto plates that contained the yeast minimal selective medium and also onto plates that additionally lacked adenosine or histidine, which are complemented by the two-hybrid assay reporter gene. Cells were lysed in a solution containing 0.15 m Tris-HCl (pH 6.7), 5% SDS, and 30% glycerol, which was then diluted 1:10 in phosphate-buffered saline (PBS)/0.5% Nonidet P-40 plus complete protease inhibitor (Roche). Four microliters of anti-HSF2 polyclonal antibody were added to the lysate, incubated for 1 h at 4 °C with gentle inversion mixing, after which protein-G-Sepharose was added. After incubation for 3 h, the beads were collected, washed four times with ice-cold PBS, 0.5% Nonidet P-40 plus complete protease inhibitor mixture. Immunoprecipitated proteins were analyzed by SDS-PAGE and Western blot using anti-SUMO-1 monoclonal antibodies (21C7) (1Matunis M.J. Coutavas E. Blobel G. J. Cell Biol. 1996; 135: 1457-1470Crossref PubMed Scopus (955) Google Scholar). Full-length HSF2 protein was in vitro translated in a rabbit reticulocyte lysate system and then subjected to in vitro SUMO-1 modification assay essentially as previously described (12Duprez E. Saurin A.J. Desterro J.M. Lallemand-Breitenbach V. Howe K. Boddy M.N. Solomon E. de The H. Hay R.T. Freemont P.S. J. Cell Sci. 1999; 112: 381-393Crossref PubMed Google Scholar). HeLa cells were transfected with pEGFP-C1 or pEGFP-HSF2β independently using LipofectAMINE 2000 (Life Technologies, Inc.). In brief, HeLa cells were seeded on coverslips such that the cells would be ∼80% confluent by the following morning. HeLa cells were grown in DMEM, 10% fetal bovine serum (FBS), 50 μg/ml gentamicin at 37 °C with 5% CO2. For each transfection, 4 μg of DNA and 7.5 μl of LipofectAMINE 2000 was used according to the manufacturer's protocol. The DNA/LipofectAMINE mixture was incubated with HeLa cells for 6 h at 37 °C with 5% CO2. After 6 h, the DNA-containing DMEM was removed and replaced with 3 ml of DMEM, 10% FBS, 50 μg/ml gentamicin. The cells were analyzed 24–36 h later by fluorescence microscopy. For immunofluorescence analysis of HeLa cells transfected with EGFP-HSF constructs, transfected cells on coverslips were fixed with cold methanol and then blocked in cold PBS (137 mm NaCl, 2.7 mm KCl, 8 mmNa2HPO4, 1.5 mmKH2PO4) +2% bovine serum albumin (BSA Fraction V, Sigma). Nontransfected cells on coverslips were fixed using 2% paraformaldehyde in PBS at room temperature as described previously (24Jolly C. Morimoto R. Robert-Nicoud M. Vourc'h C. J. Cell Sci. 1997; 110: 2935-2941Crossref PubMed Google Scholar). Coverslips were then incubated for 60 min with one of the following primary antibodies in PBS containing 2% BSA: HSF2 rat monoclonal antibody from Neomarkers (1:100 dilution), SUMO-1 mouse monoclonal antibody 21C7 (1:1000 dilution) (1Matunis M.J. Coutavas E. Blobel G. J. Cell Biol. 1996; 135: 1457-1470Crossref PubMed Scopus (955) Google Scholar), or PML mouse monoclonal antibody from Santa Cruz Biotechnology, Santa Cruz, CA (1:100 dilution). After washing with PBS + 2% BSA, the coverslips were incubated 30 min with a 1:200 dilution of the appropriate secondary antibody linked to either the Texas Red fluorochrome or fluoroscein isothiocyanate (Vector Laboratories, Burlingame, CA). After washing with PBS + 2% BSA and PBS, some coverslips were also incubated 5 min with 50 ng/ml 4′,6-diamidino-2-phenylindole (DAPI) in PBS. Coverslips were washed briefly in distilled water and mounted on a slide with Vectashield (Vector Laboratories). Immunostaining was visualized using a Nikon fluorescent microscope with a 60× objective and a Nikon Spotcam digital-imaging camera. Point mutants were generated in pcDNA-HSF2β-MH6 that were expected to change the three predicted SUMO-1-modified lysine residues to arginine. The predicted residues are Lys-82, Lys-139, and Lys-151. Site-directed mutagenesis was performed using the QuickChange mutagenesis kit (Stratagene) according to the manufacturer's protocol. Mutations were confirmed by DNA sequencing. In vitro translated HSF2 protein, with or without subsequent in vitro SUMO-1 modification, was incubated with 20 μl of binding buffer (10 mm Tris-HCl (pH 7.4), 50 mm NaCl, 1 mm EDTA, 0.5 mm dithiothreitol, 5% glycerol) containing 0.1 ng of 32P end-labeled DNA probe, 0.5 μg of poly(dI-dC)-poly(dI-dC), and 10 μg of BSA at 20 °C for 10 min. The oligonucleotide probe contains four inverted repeats of the heat shock element consensus sequence 5′-nGAAn-3′. After incubation, binding reactions were subjected to electrophoresis on native 4% polyacrylamide gels in 0.5 × TBE, and HSF2 DNA-binding complexes were visualized by autoradiography. During the course of yeast two-hybrid analysis we discovered the existence of an interaction between HSF2 and the Ubc9 protein. As shown in Fig.1, yeast containing an HSF2 bait plasmid (pGBD-HSF2) and a prey plasmid containing the region of the Ubc9 protein corresponding to amino acids 4–128 (pVP16ΔUbc9) are able to grow on selective media. The relevant negative control plasmid combinations, pGBD-HSF2 + pVP16 and pGBD-C2 + pVP16ΔUbc9, did not confer growth on these selective media, indicating specificity of the interaction. Also shown for comparison is the interaction between HSF2 and a protein called PR65, for which we have previously demonstrated a strong and specific interaction (22Hong Y. Sarge K.D. J. Biol. Chem. 1999; 274: 12967-12970Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar) (Fig. 1). Interaction with Ubc9 is a characteristic of many proteins shown subsequently to be modified by SUMO-1. To test whether this was true for HSF2, we immunoprecipitated HSF2 from cell extracts and then subjected the immunoprecipitate to Western blot using SUMO-1 antibodies. The results, shown in Fig.2 A, reveal the existence of a band of the expected size of ∼80 kDa, suggesting that HSF2 is modified by SUMO-1 in vivo. We also tested whether the HSF2 protein is a substrate for SUMO-1 modification in vitro (Fig. 2 B). This assay contained 35S-labeled in vitro translated HSF2, purified recombinant SUMO-1 and Ubc9, a HeLa cell extract (which contains the SUMO E1 enzyme activity of the SAE1/2 heterodimer), ATP, and an ATP regenerating system. When the HSF2 protein is incubated with the HeLa extracts alone, a faint higher molecular weight protein corresponding in size to the SUMO-1-modified form of HSF2 appears (Fig.2 B, lane 2). This is presumably because of small amounts of endogenous SUMO-1 and Ubc9 in the HeLa extracts. However, when SUMO-1 and Ubc9 are added to the reaction mixture a substantial increase in amount of the higher molecular weight product is observed, corresponding to SUMO-1-modified HSF2 (lane 6). HSF2 appears to also be a substrate for SUMO-2 modification in vitro(lane 7). To verify that the higher molecular weight band appearing in thesein vitro modification reactions does in fact correspond to SUMO-1-modified HSF2, we compared the sizes of HSF2 modification products that result from the use of His6-SUMOversus GST-SUMO-1 as the SUMO-1 substrate for the reaction. As expected, the use of GST-SUMO-1 instead of His6-SUMO-1 results in the appearance of a larger size corresponding to the approximate 26-kDa difference in size because of the GST moiety (Fig.2 C). Numerous SUMO-1-modified proteins localize to discrete nuclear domain structures known as PML nuclear bodies, including PML, HIPK2, and Sp100 (9Sternsdorf T. Jensen K. Will H. J. Cell Biol. 1997; 139: 1621-1634Crossref PubMed Scopus (289) Google Scholar, 10Muller S. Matunis M.J. Dejean A. EMBO J. 1998; 17: 61-70Crossref PubMed Scopus (578) Google Scholar, 11Kamitani T. Nguyen H.P. Kito K. Fukuda-Kamitani T. Yeh E.T. J. Biol. Chem. 1998; 273: 3117-3120Abstract Full Text Full Text PDF PubMed Scopus (177) Google Scholar, 12Duprez E. Saurin A.J. Desterro J.M. Lallemand-Breitenbach V. Howe K. Boddy M.N. Solomon E. de The H. Hay R.T. Freemont P.S. J. Cell Sci. 1999; 112: 381-393Crossref PubMed Google Scholar,18Kim Y.H. Choi C.Y. Kim Y. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 12350-12355Crossref PubMed Scopus (143) Google Scholar, 25Sternsdorf T. Jensen K. Reich B. Will H. J. Biol. Chem. 1999; 274: 12555-12566Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar). Thus, we were interested in determining whether HSF2 was colocalized with SUMO-1 in PML nuclear bodies. To test this possibility we performed double-label immunofluorescence analysis on HeLa cells using antibodies that recognize HSF2 and SUMO-1. The results, shown in Fig. 3, demonstrate colocalization between HSF2 and SUMO-1 staining in nuclear structures in these cells. Preabsorbing the SUMO-1 and HSF2 antibodies with purified recombinant SUMO-1 or HSF2 protein prior to staining completely abolished the nuclear domain structure staining that had already been observed with these antibodies (data not shown). As an additional test of colocalization between HSF2 and SUMO-1, we transfected HeLa cells with a green fluorescent protein (GFP)-HSF2 fusion protein expression plasmid and then stained with the anti-SUMO-1 monoclonal antibody as well as with DAPI for visualization of the nucleus (Fig. 4). Only a few of the cells that were transfected with pEGFP-HSF2β had the punctate nuclear GFP-HSF2 staining (∼7%), with the majority of the cells displaying cytosolic staining. However, in those cells that contained GFP-HSF2 nuclear domain staining, the HSF2 nuclear domain structures did colocalize with SUMO-1. Next, to definitively demonstrate that the nuclear structures observed with the HSF2 and SUMO-1 antibodies are PML nuclear bodies, we performed double-label immunofluorescence analysis on HeLa cells using antibodies that recognize HSF2 and PML. The results, shown in Fig.5, demonstrate a good correlation between staining for HSF2 and PML, consistent with our hypothesis. To identify amino acid residues of HSF2 that are modified by SUMO-1, we analyzed the mouse HSF2α protein sequence for consensus SUMO-1 modification sites (26Goodson M.L. Park-Sarge O.K. Sarge K.D. Mol. Cell. Biol. 1995; 15: 5288-5293Crossref PubMed Scopus (85) Google Scholar, 27Sarge K.D. Zimarino V. Holm K. Wu C. Morimoto R.I Genes Dev. 1991; 5: 1902-1911Crossref PubMed Scopus (299) Google Scholar). Most SUMO-1-modified proteins described to date conform to a consensus modification site of (I/V/L)KX(D/E) (6Yeh E.T. Gong L. Kamitani T. Gene. 2000; 248: 1-14Crossref PubMed Scopus (415) Google Scholar). This sequence analysis revealed that HSF2 contains three matches to this SUMO-1 consensus modification sequence, centered around Lys-82, Lys-139, and Lys-151. Mutations were made in pcDNA-HSF2β-MH6 that changed each of these lysine residues to arginine, K82R, K139R, and K151R. We then tested the ability of each of these mutant HSF2 proteins to undergo SUMO-1 modification relative to wild type HSF2. The results (Fig. 6 A) suggest that among these three sites, lysine 82 is a major site of SUMO-1 modification in HSF2, because this mutant failed to show the appearance of the major HSF2-SUMO band, whereas mutation of Lys-139 and Lys-151 did not affect the ability of HSF2 to undergo SUMO-1 modification in this assay. In this experiment we also noted the existence of a faint band, the mobility of which was slightly reduced relative to the major HSF2-SUMO-1 band, particularly noticeable in lane 4 where the major HSF2-SUMO band is absent. We suspect this band may arise from the use of a minor SUMO-1 modification site. Shown in Fig.6 B is a comparison of the sequence surrounding lysine 82 of HSF2 with characterized SUMO-1 modification sites in other proteins. We also made the K82R mutation in the HSF2-GFP mammalian expression construct used in the experiments above to determine the effect this had on HSF2 localization to PML bodies. Our results revealed that the mutant HSF2-GFP still localized to nuclear bodies (data not shown). We think this may be because of complex formation between the mutant HSF2 and endogenous wild type HSF2 present in these cells, so that the endogenous HSF2 essentially carries the mutant HSF2 to the localization site within PML bodies. This is consistent with our previous results showing spontaneous HSF2 trimer formation in cells transfected with HSF2 expression constructs (28Sarge K.D. Murphy S.P. Morimoto R.I. Mol. Cell. Biol. 1993; 13: 1392-1407Crossref PubMed Scopus (748) Google Scholar). A similar explanation has been suggested for results of experiments on PML SUMO mutants, in which failure of mutant proteins to localize to PML bodies was observed only in cells lacking endogenous PML (13Zhong S. Muller S. Ronchetti S. Freemont P.S. Dejean A. Pandolfi P.P. Blood. 2000; 95: 2748-2752Crossref PubMed Google Scholar, 29Ishov A.M. Sotnikov A.G. Negorev D. Vladimirova O.V. Neff N. Kamitani T. Yeh E.T. Strauss III, J.F. Maul G.G. J. Cell Biol. 1999; 147: 221-234Crossref PubMed Scopus (677) Google Scholar). Lysine 82 is located within the DNA-binding domain of HSF2, and so we hypothesized that SUMO-1 modification at this site might regulate the DNA binding function of this protein in some way. To test this possibility, we compared the DNA binding activity of HSF2 that was SUMO-1-modifiedin vitro to that of unmodified HSF2 using the gel shift assay with a probe that specifically binds HSFs. Fig.7 shows that SUMO-1 modification of wild type HSF2 results in a significant increase in DNA binding activity of this protein (compare lane 1 and lane 2). However, no such increase in DNA binding activity is observed for the HSF2 K82R mutant subjected to the SUMO-1 modification reaction, consistent with the hypothesis that SUMO-1 modification of this residue is responsible for this increase in DNA binding activity. The regulation and function of HSF2 as a transcription factor are still largely a mystery. The only thing that is really clear is that HSF2 can bind to heat shock elements and regulate transcription of heat shock protein genes. However, very little is known about how and under what conditions HSF2 DNA binding and function are regulated. In addition, in contrast to the related family member HSF1 that mediates the stress-induced expression of hsp genes, it is not clear what biological function HSF2 may be performing with regards to regulating hsp expression. Adding to the complexity, recent data from our laboratory suggest that HSF2 also has a role in regulating PP2A activity in cells via its interaction with the PR65 subunit of PP2A (22Hong Y. Sarge K.D. J. Biol. Chem. 1999; 274: 12967-12970Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar, 30Hong Y. Lubert E.J. Rodgers D.W. Sarge K.D. Biochem. Biophys. Res. Commun. 2000; 272: 84-89Crossref PubMed Scopus (14) Google Scholar). The results of this study shed new light on HSF2 by demonstrating that HSF2 is a target for modification by SUMO-1. A key finding of this study is that SUMO-1 modification causes activation of HSF2 DNA binding ability. This is the first demonstration that SUMO-1 modification can directly regulate the DNA-binding ability of a transcription factor, thus expanding our knowledge of the repertoire of mechanisms by which this modification can regulate protein function. Regarding the mechanism by which SUMO-1 modification activates HSF2 DNA-binding, one possibility is that the modification causes a conformational change leading to trimerization and DNA binding ability. Consistent with this hypothesis, lysine 82, the lysine residue of HSF2 indicated by our results to be a site for SUMO-1 modification, is located in a so-called "wing" within the DNA-binding domain of this protein. It is interesting to note that that this wing region has already been suggested to play a role in stabilizing the trimeric DNA binding form of HSF by forming interactions between monomers within the trimer (31Littlefield O. Nelson H.C. Nat. Struct. Biol. 1999; 6: 464-470Crossref PubMed Scopus (137) Google Scholar). Thus, SUMO-1 modification of HSF2 at this lysine residue could induce a conformational change that makes these wings accessible for interaction and stimulates trimer formation. Consistent with the importance of SUMO-1 modification for HSF2 function, sequence analysis reveals that the consensus SUMO-1 modification site found at Lys-82 in mouse HSF2 is conserved in both the human and chicken HSF2 homologs (32Schuetz T.J. Gallo G.J. Sheldon L. Tempst P. Kingston R.E. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 6911-6915Crossref PubMed Scopus (260) Google Scholar, 33Nakai A. Morimoto R.I. Mol. Cell. Biol. 1993; 13: 1983-1997Crossref PubMed Scopus (208) Google Scholar). SUMO-1 modification could also be involved in regulating the stability of the HSF2 protein within cells. Several examples exist which demonstrate that SUMO-1 modification can stabilize proteins against degradation, such as in the case of IκB, Mdm2, and possibly p53 (34Desterro J.M. Rodriguez M.S. Hay R.T. Mol. Cell. 1998; 2: 233-239Abstract Full Text Full Text PDF PubMed Scopus (911) Google Scholar, 35Buschmann T Fuchs S.Y. Lee C.G. Pan Z.Q. Ronai Z. Cell. 2000; 101: 753-762Abstract Full Text Full Text PDF PubMed Scopus (227) Google Scholar, 36Gostissa M. Hengstermann A. Fogal V. Sandy P. Schwarz S.E. Scheffner M. Del Sal G. EMBO J. 1999; 18: 6462-6471Crossref PubMed Scopus (437) Google Scholar, 37Rodriguez M.S. Desterro J.M. Lain S. Midgley C.A. Lane D.P. Hay R.T. EMBO J. 1999; 18: 6455-6461Crossref PubMed Scopus (559) Google Scholar). Measurements indicate that the HSF2 protein has a relatively short half-life within cells, on the order of 60 min (38Mathew A. Mathur S.K. Morimoto R.I. Mol. Cell. Biol. 1998; 18: 5091-5098Crossref PubMed Scopus (174) Google Scholar). Thus, SUMO-1 modification of HSF2 could serve to render a pool of HSF2 resistant to degradation, or conversely, perhaps regulated removal of SUMO-1 is the controlling step that regulates the timing of HSF2 turnover. SUMO-1 modification of HSF2 is also correlated with its localization to PML nuclear bodies, as has been demonstrated for a number of other proteins including PML itself and p53 (9Sternsdorf T. Jensen K. Will H. J. Cell Biol. 1997; 139: 1621-1634Crossref PubMed Scopus (289) Google Scholar, 10Muller S. Matunis M.J. Dejean A. EMBO J. 1998; 17: 61-70Crossref PubMed Scopus (578) Google Scholar, 11Kamitani T. Nguyen H.P. Kito K. Fukuda-Kamitani T. Yeh E.T. J. Biol. Chem. 1998; 273: 3117-3120Abstract Full Text Full Text PDF PubMed Scopus (177) Google Scholar, 12Duprez E. Saurin A.J. Desterro J.M. Lallemand-Breitenbach V. Howe K. Boddy M.N. Solomon E. de The H. Hay R.T. Freemont P.S. J. Cell Sci. 1999; 112: 381-393Crossref PubMed Google Scholar, 13Zhong S. Muller S. Ronchetti S. Freemont P.S. Dejean A. Pandolfi P.P. Blood. 2000; 95: 2748-2752Crossref PubMed Google Scholar, 39Pearson M. Carbone R. Sebastiani C. Cioce M. Fagioli M. Saito S. Higashimoto Y. Appella E. Minucci S. Pandolfi P.P. Pelicci P.G. Nature. 2000; 406: 207-210Crossref PubMed Scopus (1130) Google Scholar). What role might HSF2 localization to PML bodies play in the regulation and function of this transcription factor? There are several possibilities, some of which have been proposed for other transcription factors that have been found to reside within these bodies (40Stein G.S. van Wijnen A.J. Stein J.L. Lian J.B. Montecino M. Choi J Zaidi K Javed A. J. Cell Sci. 2000; 113: 2527-2533Crossref PubMed Google Scholar, 41Zhong S. Salomoni P. Pandolfi P.P. Nat. Cell Biol. 2000; 2: E85-E90Crossref PubMed Scopus (490) Google Scholar). First, one function might be to bring HSF2 into close proximity with other transcriptional co-factors that also localize to PML bodies so it can form complexes with these proteins that are important for the subsequent ability of HSF2 to regulate gene expression. Another possibility is that localization of HSF2 to these bodies serves a sequestering function, tying up this pool of HSF2 so that it can't act to regulate gene expression until some signal triggers its release from this body. A particularly intriguing question to address in future studies is whether the SUMO-1-induced activation of HSF2 DNA binding is unique to this transcription factor or whether it may also play a role in regulating the DNA binding activity of other transcription factors that are SUMO-1 modified. Studies to test these and other hypotheses will likely increase understanding of HSF2 function in cells as well as the biological functions of SUMO modification and PML bodies. We thank Dr. Joana Desterro and Dr. Ron Hay for the generous gift of the GST-SUMO-1 bacterial expression plasmid. We are also very grateful to Dr. Wally Whiteheart for expert advice on the immunofluorescence analysis.

Referência(s)