Histone H3 lysine 9 methylation: A signature for chromatin funciton

In most eukaryotes, the histone methyltransferase SU(VAR)3-9 and its orthologues play amajor role in the function of centromeric heterochromatin. Although the methyltransferase domain isrequired for the formation of a fully functional centromere, mutations within other regions of the genesuch as the N-terminus also have a strong impact on its in vivo function. To analyze the contribution ofthe N-terminus on the methyltransferase activity, we have expressed the full-length DrosophilaSU(VAR)3-9 (dSU(VAR)3-9) together with various N-terminal deletions in Escherichia coli and analyzedthe structural and enzymatic properties of the purified recombinant enzymes. Full-length dSU(VAR)3-9specifically methylates lysine 9 within histone H3 on peptides, on intact histones, and, to a lesser extent,on nucleosomes. A detailed analysis of the reaction products shows that dSU(VAR)3-9 adds two methylgroups to an unmethylated H3 tail peptide in a nonprocessive manner. The full-length enzyme elutes withan apparent molecular weight of 160 kDa from a gel filtration column, which indicates the formation ofa dimer. This property is dependent on an intact N-terminus. In contrast to the full-length enzymes, proteinslacking the N-terminus fail to dimerize, and show a 10-fold lower specific activity and a linear dependenceof methyltransferase activity on enzyme concentration. A N-terminal peptide containing amino acids 1-152of dSU(VAR)3-9 is sufficient to mediate this interaction in vitro. The dimerization of dSU(VAR)3-9and the subsequent increase of its methyltransferase activity provide a starting point to understand themolecular details of the formation of heterochromatic structures in vivo. Centromeres are conserved structures of eukaryotic chro-mosomes, which ensure their proper segregation duringmitotic divisions (1, 2). Most centromers are formed byassociation of the centromeric DNA with specific proteinssuch as CENP-A, -B, and -C (3, 4) and have been shown toform clusters within interphase chromatin (5, 6). Althoughcentromeres as well as pericentromeric regions are typicallyrich in repetitive DNA, the main determinant of centromersseems to be of an epigenetic nature as they can also formectopically within euchromatic arms (7, 8). Centromericregions of chromosomes are generally transcriptionallyquiescent, a property that can “spread” into neighboringregions (2, 9). The variable transcriptional activity ofnormally active genes, after their translocation close toheterochromatin, has been termed position effect variegation(PEV)1 (10, 11) (9).Histones within centromeric regions are usually hypo-acteylated (12, 13) and methylated at lysine 9 within the H3tail (14). The enzyme that is critical for this modification isthe histone methyltransferase SU(VAR)3-9 (15). DrosophilaSU(VAR)3-9 (dSU(VAR)3-9) has been initially identifiedin a genetic screen for suppressors of PEV (for a review seeref 16). SU(VAR)3-9 or its orthologue CLR4 are requiredfor heterochromatin-mediated gene silencing in Drosophilaand Schizosaccharomyces pombe (17). Mouse SUV39 en-zyme is important to maintain genome stability (18).Furthermore, it acts as a transcriptional repressor in transientas well as in stable transfection experiments in tissue culturecells (19, 20).Drosophila SU(VAR)3-9 belongs to a large class ofproteins containing a SET domain. The SET domain confersmethyltransferase activity and is crucial for the in vivofunction of most SET-containing proteins. Although the SETdomain is relatively well conserved, SET-methyltransferasesshow a remarkable structural and functional variability. SETdomain proteins can exist as monomers (21-25) or dimers(26). In addition, some need auxiliary factors (27-30) orrequire nucleic acids for full activity (31, 32).In addition to the well-characterized SET domain,dSU(VAR)3-9 contains a chromo domain (17, 19, 33), a † Funding for this work has been provided by a grant (IM23/4-1)from the Deutsche Forschungsgemeinschaft (DFG).* Correspondence should be sent to Axel Imhof, Adolf-ButenandtInstitute, Ludwig-Maximillians University of Munich, Schillerstr. 44,80336 Muenchen, Germany. Tel.: -49 89 5996 435 Fax: -49 895996 425. E-mail: [email protected].§ Both authors contributed equally to this work.# Department of Molecular Biology.‡ Histone Modifications Group and Protein Analysis Unit.1 PEV, position effect variegation; PCR, polymerase chain reaction;PMSF, phenylmethanesulfonyl fluoride; HEPES, N-(2-hydroxyethyl)-piperazine-N′-(2-ethanesulfonic acid); BSA, bovine serum albumin;EGTA, ethylene glycol-bis(2-aminoethyl ether)-N,N,N′,N′-tetraaceticacid; EDTA, ethylenediaminetetraacetic acid; DTT, DL-dithiothreitol;MTase, methyltransferase; HMT, histone methyltransferase; ACN,acetonitrile; TFA, trifluoroacetic acid; MALDI-TOF, matrix assistedlaser desorption ionization-time-of-flight mass spectrometry; SAM,S-adenosyl methionine; SAH, S-adenosyl homocystein.3740Biochemistry 2004, 43, 3740-3749 10.1021/bi035964s CCC: $27.50 © 2004 American Chemical SocietyPublished on Web 03/06/2004 GTPase domain that is derived from a common exon usedby dSU(VAR)3-9 and the eukaryotic translation initiationfactor 2 and a relatively ill-defined N-terminal domain (15).The N-terminus is moderately conserved in humans, mice,and flies and interacts with at least two additional suppressorsof PEV, HP1, and SU(VAR) 3-7 (14, 34). A Structure-function analysis of the human SUV39H1 indicated that adeletion of the N-terminus leads to a failure of SUV39 tobind chromatin in vivo (33). A fragment of SUV39H1containing just the N-terminus and the chromo domain bindsefficiently to heterochromatin when expressed in tissueculture cells. This binding is thought to be mediated by HP1and has been suggested to be a main component in themaintenance mechanism of histone methylation as thechromo domain of HP1 binds strongly to a H3 tail, which ismethylated at lysine 9. In the absence of HP1, SU(VAR)3-9is found at multiple sites along the chromosome arms,suggesting a role for HP1 in restricting SU(VAR)3-9 tocentromeric heterochromatin. However, the HP1 interactingregion is not sufficient to confer chromatin binding to anoverexpressed fusion protein (33), suggesting an additionalstructural component in addition to HP1 binding, which isimportant for heterochromatin association. This hypotheticalstructural component may well be the conformation ofdSU(VAR)3-9 itself. SUV39H1 forms distinct nucleardomains when overexpressed in vivo (19), a phenomenonthat has already been described for certain ring finger proteinswhich form higher order aggregates in vivo and in vitro (35,36). In flies, overexpression of dSU(VAR)3-9 or othersuppressors of PEV leads to a strong dosage-dependentenhancement of PEV independent of the dosage of the otherpartners. This strong dosage dependence argues against asingle limiting factor regulating PEV but rather favors amodel in which several factors participate in the assemblyof a specific heterochromatin scaffold (16, 37).In this report, we show that full-length dSU(VAR)3-9 isa very active methyltransferase when expressed in bacteria.It adds two methyl groups to lysine 9 within the H3 tail ina nonprocessive manner. Moreover, the N-terminus ofdSU(VAR)3-9 mediates an interaction between twoSU(VAR)3-9 molecules, thereby increasing its ability tomethylate H3. The N-terminus of dSU(VAR)3-9 has abipartite interaction domain, which allows the formation ofmultimers of SU(VAR)3-9 proteins. This interaction be-tween dSU(VAR)3-9 molecules may contribute to theclustering of centromers within living cells and the formationof nuclear substructures in vivo after expressing the N-terminus of SU(VAR)3-9 alone and may therefore explainthe strong dosage dependence of the dSU(VAR)3-9 medi-ated enhancement of PEV. MATERIALS AND METHODS Cloning of dSU(VAR)3-9. Full-length dSU(VAR)3-9,deletion mutants, and point mutants were cloned into pET15b(Novagen) via NdeI and XhoI. The pET15b plasmid adds a6× his-tag on the N-terminus. All inserts were created byPCR from a plasmid carrying the dSU(VAR)3-9 cDNA(kind gift of G. Reuter, Halle, Germany) and verified byDNA sequencing. GST fusion proteins were cloned byinserting a PCR generated EcoRI-XhoI fragment into apGEX 4T-2 vector (Amersham).Protein Purification. His-tagged dSU(VAR)3-9 anddSU(VAR)3-9 mutant polypeptides were expressed in E.coli BL21(DE3)pLys, and purified with Talon (Clontech)resin according to the manufacturer’s instructions. For themolecular weight analysis, the Talon-purified proteins wereloaded on a Superdex 200 column (HR 10/30, Amersham)or on a 5-20% sucrose gradient. The column was runisocratically in BC100 buffer (25 mM HEPES/KOH (pH7.3), 100 mM NaCl, 1 mM MgCl2, 0.5 mM EGTA, 0.1 mMEDTA, 10% glycerol (v/v), 1 mM DTT, and 0.2 mM PMSF)for 1.4 CV. 0.5 mL fractions were collected and 15 μL ofeach fraction were analyzed on a 10% SDS PAA gel. 5-20%(w/v) sucrose gradients were prepared in BC100 buffer withor without 3 M urea. The gradient was prepared using aGradient Master 105/106 (BioComp) set at 2.40 min/81.5 deg/15 rpm. A 500 μL sample containing 10 μg ofdSU(VAR)3-9 wild type or ∆213 or 20 μg of BSA wasloaded on top of the gradient. Centrifugation was performedusing a SW41 rotor (Beckman) at 41 000 rpm for 28 h at 4°C. 0.5 mL fractions were collected and analyzed by 10%SDS-PAGE.For activity assays, the enzyme was stabilized by additionof BSA to a final concentration of 100 ng/μL followedby dialysis against BC100 buffer. All recombinantdSU(VAR)3-9 were quantified by Coomassie staining withthe ImageMaster 1D Elite v3.01 software package (Amer-sham) using BSA as a standard. Bacterially expressed HP1was purified according to ref 38 and dialyzed against BC100.Methyltransferase ActiVity. H3 peptides used containedamino acid 1-19 plus a C-terminal cysteine (ARTKQTARK-STGGKAPRKQC) and were either unmodified, dimethylatedat K4, monomethylated at K9, acetylated at K9, or dimethy-lated at K9 (Peptide Speciality Laboratories, Heidelberg).Recombinant Drosophila histones were expressed and puri-fied from E. coli, and reconstituted into octamers as describedpreviously (39). Nucleosomes were reconstituted by saltdialysis. Recombinant histone H3 carrying the mutationsK9A, K27A, or both, were expressed in bacteria fromplasmids kindly provided by D. Reinberg. HMT assays weredone inddH2O using H3-peptide, histone H3, octamer, ornucleosomes as substrates and S-adenosyl-[methyl-3H]-L-methionine (25 μCi/mL) and/or S-adenosylmethionine (NewEngland BioLabs) as methyl donor. Reactions were per-formed at 30 °C for 1-80 min. To stop the reaction, aceticacid was added to a final concentration of 5-10% (v/v).Kinetic assays were carried out in triplicates by varying theconcentration of the H3 peptide (1-16 μM) or SAM (5-60μM) at saturating amounts of SAM/peptide and analyzed bydouble reciprocal plots.GST Pulldown. dSU(VAR)3-9 and mutants were trans-lated in vitro from pET15b vectors, [35S]methionine-labeled.GST-pulldowns were carried out as described earlier (40).In vitro translated proteins and HP1 (1 μg) were incubatedin a total volume of 200 μL containing NTEN100 (20 mMTrisHCl (pH 8), 100 mM NaCl, 1 mM EDTA, 0.5% (v/v)Nonidet P-40), 100 μg of BSA, and 5 μg of ethidiumbromide. Washes were performed two times in NTEN100 andfour times in theNTEN200. The bound proteins were elutedwith SDS sample buffer and analyzed using a phosphoimageror a specific antibody in the case of HP1.MALDI-TOF Analysis. To purify the methylated peptidesfrom contaminating salts, the peptide solution was passedEnzymatic Characterization of dSU(VAR)3-9Biochemistry, Vol. 43, No. 12, 2004 3741