Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Med Chem. Author manuscript; available in PMC Nov 10, 2012.
Published in final edited form as:
PMCID: PMC3326379
Structure-Guided Design of a Methyl Donor Cofactor that Controls a Viral Histone H3 Lysine 27 Methyltransferase Activity
Jiaojie Li, Hua Wei, and Ming-Ming Zhou*
Department of Structural and Chemical Biology, Mount Sinai School of Medicine, 1425 Madison Avenue, Box 1677, New York, NY 10029, USA
* Corresponding Author. Phone, 212-659-8652; fax, 212- 849-2456; ming-ming.zhou/at/
These authors contributed equally to this work.
vSET (a viral SET domain protein) is an attractive Polycomb Repressive Complex 2 (PRC2) surrogate to study the effect of histone H3 lysine 27 (H3K27) methylation on gene transcription as both catalyze histone H3K27 tri-methylation. In order to control the enzymatic activity of vSET in vivo with an engineered S-adenosyl-L-methionine (SAM) analogue as methyl donor cofactor, we have carried out structure-guided design, synthesis and characterization of orthogonal vSET methyltransferase mutant/SAM analogue pairs using a “bump-and-hole” strategy.
Epigenetic control of gene expression is responsible for fundamental cellular processes such as genetic imprinting, stem cell self-renewal and differentiation, and it involves many types of chemical modifications on chromatin, including DNA methylation and histone methylation, acetylation, phosphorylation, ubiquitination and ADP-ribosylation. 1,2 The high complexity of this regulatory system is made possible by that different histone modifications function in a context-dependent manner. It is difficult to predict how different modifications work in concert in gene regulation. A means to turn on or off histone modifications on demand would be an invaluable tool.
The Polycomb repressive complex 2 (PRC2) methylates histone H3 lysine 27 (H3K27), a mark for gene transcriptional silencing.3 The PRC2 consists of three components: EZH2, SUZ12 and EED in human.4,5 EZH2 of 751 amino acids contains the catalytic SET domain lysine methyltransferase activity; its optimal methyltransferase activity requires the other two subunits and other partners. It has proven very challenging to express a functional PRC2 in situ. Notably, our recent studies show that a SET protein (vSET) of 119 amino acids encoded by the Paramecium Bursaria chlorella virus can specifically methylate H3K27 in eukaryotic cells.68 vSET is active on its own, making it much easier to manipulate than the PRC2 complex.
We envision that vSET activity can be controlled in a variant whose activity is dependent upon a chemically engineered S-adenosyl-L-methionine (SAM) derivative. The generation of orthogonal vSET/SAM pairing may be accomplished using a “bump-and-hole” strategy, pioneered by Shokat and colleagues in the kinase field.9 Specifically, one would change a bulky residue in vSET to alanine or glycine, and add a substituent to SAM to fill the space created in the protein. Given its function as a common methyl donor cofactor, surprisingly, there are only a few reports on the utilization of SAM analogues by protein methyltransferases. In one study, N6-substituted S-adenosyl-L-homo-cysteine (SAH) analogues were designed as variant specific RMT1 (a yeast arginine methyltransferase) inhibitors.10 Other studies reported that SAM analogues, in which the sulfonium methyl group is replaced with a substituent alkynyl group, are useful for identifying enzyme substrates.11,12 However, no one has yet reported a modified SAM analogue as an effective methyl donor cofactor for a protein methyltransferase. Here, we describe design and synthesis of several substituted SAM molecules, and characterization of their methyl donor capability with vSET H3K27 methylation activity. We succeeded in developing an orthogonal vSET/SAM analogue pair that is very active, thereby demonstrating the feasibility of this strategy for the protein methyltransferases.
We focused on chemical modifications of SAM at three positions, the N6-amino of the adenine ring that is a site used in other studies,10 and the 2′- or 3′-hydroxyl in the ribose moiety that has yet been explored. Close inspection of our crystal structure of the vSET/SAH/H3K27me1 peptide complex (Figure 1)8 reveals that the adenine ring of SAH forms aromatic and hydrophobic interactions with Trp110 and the side-chain methyls of Leu13, and the N6-amino group is hydrogen bonded to the carboxylate of the C-terminal residue Asn119 and the main chain carbonyl oxygen of His70. This set of interactions is bolstered by a hydrogen bond formed between the side-chain amino of Asn119 and the backbone carbonyl oxygen of Leu13. While pointing outward, the 2′-hydroxyl is packed next to Leu116. Further, the phenoxyl of Tyr109 forms two hydrogen bonds, one with the 3′-hydroxyl of SAH and another with the main chain carbonyl of Asp49. Thus, we hypothesized that change of Leu13, Asn119, Tyr109 or Leu116 to Ala or Gly would create a cavity that could accommodate a substituent group from a SAM analog.
Figure 1
Figure 1
(A) Crystal structure of vSET complex (PDB ID: 3KMT). SAH and key vSET residues are shown as sticks, and H3 peptide in orange. Top, the stereo view. Bottom, the mesh and solid form of vSET surface. (B) Pymol modeled surfaces of Y109G and L116A variants (more ...)
We first synthesized (±)-N6-benzyl-S-adenosyl-L-methionine 4 in a four-step synthetic route (Scheme 1). The reaction started with condensation of 6- chloropurine riboside with benzyl amine in ethanol,10 followed by replacement of the 5′-hydroxyl of 1 with chloride, and subsequent coupling with L-homocysteine thiolactone and methylating the sulfur, affording the product of 4 with an overall yield of 6%. The intermediates and final product were confirmed by mass spectrometry and NMR spectra.
Scheme 1a
Scheme 1a
a Reagents and conditions: (a) benzylamine, EtOH, 70°C, 16 h; (b) SOCl2, Py, CH3CN, 0->25°C, 6 h; (c) L-homocysteine thiolactone/ HCl, NaOH, KI, 100°C, O/N; (d) MeI, HCOOH, 25°C, 4 days.
For the synthesis of 2′- and/or 3′-benzyl SAM, we set to generate adenosine derivatives with dialkyl disulfides under Hata’s condition.13,14 Our initial attempt at direct benzylation of unprotected adenosine with phenyldiazomethane 7,15 prepared in three steps from benzylamine in 44% yield (Scheme 2), resulted in a very low yield (< 10%).16 We found that protection of the amino at the 4- position of adenosine with N, N-dimethylformamide dimethyl acetal improved the yield to 55%,17 producing the two regioisomers 9a and 9b (Scheme 2) that were separated by HPLC in a ratio of 3:2. The regiochemistry assignment of 9a and 9b was made by comparison of the 1H NMR spectra with those reported previously.18 The disulfide moiety, N, N-Bis(trifluoroacetyl)-L-homocysteine dimethyl ester 1117 was obtained in 59% yield after two steps from L-homocystine (Scheme 2).
Scheme 2a
Scheme 2a
a Reagents and conditions: (a) tosyl chloride, Py, 25°C, 1 h; (b) NaNO2, CH3COOH: Ac2O = 1:4, 0 -> 25°C, overnight; (c) NaOMe, MeOH: Et2O=1:6, reflux, 30 min; (d) N, N-dimethyl formamide dimethyl acetal, DMF, 25°C, overnight; (more ...)
However, treatment of 9a or 9b with N, N-Bis( trifluoroacetyl)-L-homocysteine dimethyl ester 11 under Hata’s condition, in the presence of tri-n-butylphosphine in pyridine, did not give the desired product. The coupling reaction did not proceed in several solvent conditions tested (Supplemental Table 1). We noticed that the reaction of free adenosine with 11 under the same condition gave the desired product in 46% yield after five days (Supplemental Table 1, entry 3). The 1H NMR spectrum of the product matches well to that reported previously.19 Thus, we reasoned that introduction of the bulky benzyl substituent near the coupling site likely interferes with the reaction, and dialkyl disulfides are generally much less reactive than diaryl disulfides in the phosphine-mediated coupling reaction, which may also contribute to the failure of the reaction.14,19
We next introduced a good leaving group at the 5′-position to install the alkylthio substituent (Scheme 3). Selective conversion of adenosine to 5′-chloro derivative 12 was performed with thionyl chloride in acetonitrile,20 followed by protection of the amino group with N, N-dimethylformamide dimethyl acetal.17 Subsequent benzylation of 13 with phenyldiazomethane 7 afforded two regioisomers 14a and 14b in a ratio of 3:2. Their regiochemistry assignment was done by comparison of the 1H NMR spectra with those obtained from chlorination of 9a and 9b. In addition, the di-benzylated adenosine analogue 14c was synthesized by increasing the amount of phenyldiazomethane in the reaction (Scheme 3).
Scheme 3a
Scheme 3a
a Reagents and conditions: (a) SOCl2, Py, CH3CN, 0->25°C, 6 h; (b) N, N-dimethyl formamide dimethyl acetal, DMF, 25°C, overnight; (c) (i) SnCl2·2H2O, comp. 7, 1,2-dimethoxyethane, 25°C, overnight; (ii) NH4OH/MeOH, (more ...)
As expected (Scheme 3), coupling of 14a, 14b or 14c with L-homocysteine thiolactone proceeded successfully and led to 15a, 15b and 15c, respectively.21 Due to a resulting sulfonium chiral center,22 a diastereomeric mixture of (±)-3′-benzyl-S-adenosyl-L-methionine 16a, (±)- 2′-benzyl-S-adenosyl-L-methionine 16b, and (±)-2′, 3′-dibenzyl-S-adenosyl-L-methionine 16c were synthesized by methylating the sulfur atom of 15a, 15b and 15c with excess amount of methyl iodide in formic acid in the dark for four days. The final products of the SAM derivatives 16a-c were obtained as diastereomeric mixtures.
We first tested the idea of adding a bulky group to the N6-position of the adenine ring of SAM. To generate necessary extra space, either Leu13 was substituted with Ala, or the C-terminal Asn119 was removed (Figure 1A). Almost no activity was detected after an overnight reaction for both vSET variants in the presence of N6-benzyl-SAM 4 (Figure 2B vs. 2A; Table 1). Similar results were obtained when N6-methyl-SAM or N6-isopropyl-SAM was used (data not shown). Note that the previously reported yeast RMT1-E117G/N6-benzyl-SAM pair yielded only 0.2% arginine methyltransferase activity as compared to the WT-RMT1/SAM pair.10 Given that the adenine ring forms a network of hydrogen bonds with vSET residues Leu13, His70 and Asn119, our results combined with those from the RMT1-E117G/N6-benzyl-SAM pair argue that the N6-position of the adenine is not suitable for generating SAM derivatives that would retain significant methylation activity with a paired vSET variant.
Figure 2
Figure 2
Mass spectrometry data showing the activity of different enzyme/cofactor pairs.
Table 1
Table 1
vSET/cofactor pairs with no activity
We next characterized Tyr109 variants with SAM or its analogues. Mass spectrometry analysis shows that Y109G has very little activity in methylating H3K27 with SAM compared to the wild-type vSET (Figure 2A and C). The same is true with Y109A (data not shown). We then tested the activity of Tyr109 variants in the presence of 16a or 16b. The Y109G/16a pair was much less active than wild-type/SAM pair as judged by the amount of methylation signal generated in the mass spectrometry study (Figure 2D). However, it was slightly more active than Y109G/SAM and Y109G/16b pairs (Figure 2E). The activity of Y109A was lower in the presence of SAM analogues than Y109G (data not shown). Therefore, our further studies were carried out with the Y109G variant. To better compare different enzyme/cofactor pairs we performed Michaelis-Menton enzyme kinetic study to obtain the kcat and Km values (Table 2; Supplemental Figure 1).
Table 2
Table 2
kcat and Km of the vSET/cofactor pairs
When SAM is used, change of Tyr109 to Gly causes a 8,125-fold drop in kcat but little change to Km (Table 2). It suggests the substitution at Tyr109 is most likely to directly interfere with its possible function in substrate lysine binding. For Y109G, changing the cofactor from SAM to 16a results in a small increase of kcat but a much higher Km. The same cofactor change for the wild-type vSET causes a 28-fold decrease of catalytic efficiency but surprisingly a 1.5-fold decrease of Km. The increase in Km could be explained by the contribution of fortuitous interaction between the benzyl group and some protein residues. The extremely high Km value for Y109G/16a might be associated with the entropic penalty of fixing the benzyl group at a specific orientation for its insertion into the artificial cavity. Thus, we postulated that a second benzyl group at the 2′-position might restrict the conformation of 3′-benzyl, securing its insertion into the cavity. An additional 2′-benzyl group in SAM (i.e. 2′,3′-dibenzyl-SAM 16c) brings little change of the kcat/Km value to the wild-type vSET with an increase both in kcat and Km. But, a very different scenario was observed for Y109G - a slight increase of kcat but a much-improved Km (Table 2).
To mimic the hydrogen bond between Tyr109 and the 3′-hydroxyl of SAM, we synthesized another SAM analogue 19,23 which bears a carbamoylmethyl substituent at both 2′- and 3′-oxygen atoms, and tested its activity with Y109H, Y109D or Y109N (Supplemental Scheme 1). We postulated that the amide moiety of the carbamoylmethyl substituent of SAM could form a hydrogen bond with the side chain on the residue replacing Tyr109 in these variants. However, no activity was detected for any of these vSET variants after an overnight incubation, neither was any of these variants active with SAM or the wild-type vSET with 19. These results emphasize the importance of Tyr109 in substrate binding, and thus change of Tyr109 to His, Asp, or Asn is detrimental to vSET activity.
Inspection of residues around the ribose moiety of SAH in the crystal structure suggests that changing Leu116 could create a docking site for a benzyl group at the 2′- or 3′-position (Figure 1). Indeed, most of the H3 peptide was converted to mono- or di-methylated species after 19 h of reaction for L116A/16a (Figure 2F). The variant was even more active with 16c as almost all the H3 peptide was converted to tri-methylated species (Figure 2G). Kinetic studies further revealed kcat of 7.4 min−1 and Km of 60 μM for L116A/SAM, which notably is even better than the wild-type vSET/SAM pair (Table 2). When 16a was used, the kcat of L116A was dropped to 0.15 min−1, accompanied by a slight increase of Km. Importantly, L116A/16c was more active than WT/16c, with a 3.5-fold higher kcat and a 2.4-fold lower Km. Comparison of the kcat/Km value of L116A/16c with that of L116A/16a argues that the presence of the benzyl groups at both 2′- and 3′-positions have a synergistic effect on L116A catalytic efficiency when compared to the single benzyl group analogues.
Since the presence of the 3′-benzyl in SAM resulted in a reduction in the enzymatic activity of both the wild-type vSET and L116A variant as compared to SAM, we investigated whether 16b would be more potent for L116A. The results show that for the wild-type vSET, the removal of the 3′-benzyl in 16c causes a small decrease in both kcat and Km with a net result of a 1.4-fold decrease in the kcat/Km value. However, for L116A, a 13-fold decrease of kcat was seen with an overall 19-fold decrease in kcat/Km. Interestingly, while adding a single benzyl to the 2′- or 3′-position in SAM causes an ~120-fold decrease in kcat/Km for L116A, these two benzyl groups in 16c likely act synergistically, leading to only a 7-fold decrease of kcat/Km.
Histone lysine methylation is a fundamental mechanism for epigenetic control of gene expression in chromatin. Chemical modulators capable of controlling the enzymatic activity of specific lysine methyltransferases are extremely valuable tools. In this study, we illustrated our approach with an H3K27-specific methyltransferase vSET and SAM. We selected the common N6-position of the adenine ring as well as the novel 2′- and 3′-hydroxyl groups of the ribose as the modification sites. We synthesized seven SAM analogues and characterized their activity with the wild-type and variant vSET proteins. We succeeded in engineering a paired vSET-L116A/16c, which is only slightly less active than vSET-L116A/SAM but more active than the wild-type vSET/16c. Given the presence of SAM in cells, further optimization via chemical modifications of the enzyme and the cofactor is needed to generate a paired vSET variant/modified SAM with activity better than that of the variant with SAM. In summary, our study demonstrates the feasibility of developing a methyl donor cofactor that controls a histone lysine methyltransferase, which could be a useful tool to study the effects of histone H3K27 methylation in vivo.
Chemical Synthesis
All compounds were synthesized using commercially available starting materials without further purification unless otherwise stated. 1H and 13C NMR spectra were recorded on a Bruker 600 MHz NMR spectrometer using the residual signal of the deuterated solvent as internal standard. Chemical shifts (δ) are reported in ppm, coupling constants (J) in hertz. MS (ESI) analysis for new compounds was performed on an Agilent G1969A high-resolution mass spectrometer. All compounds that were tested in the biological assays were analyzed by HPLC and LC-Mass to confirm the purity ≥ 95%. R and S indicate relative configurations.
(±)-2′, 3′-dibenzyl-S-adenosylmethionine iodide (16c)
MeI (0.03 mL, 0.5 mmol) was added to a solution of 15c (22.6 mg, 0.04 mmol) in HCOOH (1 mL). After the addition was completed, the reaction mixture was stirred at room temperature in the dark for 4 days. Isolation of 16c (6.0 mg, 21.3% yield) was achieved by HPLC on an Agilent Eclipse XDB-C18 9.4 x 250 mm column. 1H NMR (600 MHz, D2O) δ 2.22–2.26 (m, 1H), 2.31–2.35 (m, 1H), 2.83 (s, 3H), 2.95 (s, 2H), 3.45 (t, J = 9.0 Hz, 1H), 3.51–3.56 (m, 2H), 3.60–3.67 (m, 1H), 3.70–3.74 (m, 1H), 3.80 (t, J = 9.0 Hz, 1H), 3.82 (d, J = 12.0 Hz, 1H), 3.88 (d, J = 13.6 Hz, 1H), 4.14–4.23 (m, 2H), 4.34 (d, J = 13.6 Hz, 1H), 4.41 (s, 1H), 4.56–4.57 (m, 1H), 5.89 (d, J = 8.8 Hz, 1H), 6.72–6.78 (m, 1H), 6.86 (t, J = 7.9 Hz, 2H), 6.94 (d, J = 5.6 Hz, 2H), 7.42 (t, J = 5.3 Hz, 1H), 7.46 (t, J = 7.53 Hz, 2H), 7.50 (d, J = 3.9 Hz, 2H), 8.00 (s, 1H), 8.04 (s, 1H); HRMS (ESI) calculated for C29H35N6O5S (M)+ 579.2384, found 579.2338.
Cloning, expression and purification of vSET
vSET was cloned into pET-22b and expressed as a non-fusion protein in BL21 (DE3) cells.8 The protein was purified with a heparin column and then a Superdex 75 column.
Site-directed mutagenesis
Variants of vSET were generated with the QuikChange mutagenesis kit (Stratagene). The presence of appropriate amino acid change was confirmed by DNA sequencing. All variants were purified as described for the wild type vSET.
Lysine Methyltransferase Reaction
The activity of a paired vSET/cofactor was measured in a solution containing 1 μM enzyme, 100 μM H3 peptide (residues 13–33) and 1 mM methyl donor cofactor. The reaction buffer consisted of 20 mM Tris at pH 8, 20 mM potassium chloride and 10 mM magnesium chloride. The reaction was performed at room temperature and stopped at specified time by the addition of trifluoroacetic acid (to 1%). The reaction products were analyzed by mass spectrometry.
Supplementary Material
We thank L. Zeng for his technical support of NMR analysis of chemical compounds, and M. Ohlmeyer for helpful discussion. This work was supported by the grants from the National Institutes of Health (to M.-M.Z.).
PRC2Polycomb Repressive Complex 2

Supporting Information. Synthesis procedures, spectroscopic details of the compounds, and enzyme kinetic analysis. This material is available free of charge via the Internet at
1. Jenuwein T, Allis CD. Translating the histone code. Science. 2001;293:1074–1080. [PubMed]
2. Kouzarides T. Chromatin modifications and their function. Cell. 2007;128:693–705. [PubMed]
3. Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P, Jones RS, Zhang Y. Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science. 2002;298:1039–1043. [PubMed]
4. Kuzmichev A, Nishioka K, Erdjument-Bromage H, Tempst P, Reinberg D. Histone methyltransferase activity associated with a human multiprotein complex containing the Enhancer of Zeste protein. Genes Dev. 2002;16:2893–2905. [PubMed]
5. Montgomery ND, Yee D, Chen A, Kalantry S, Chamberlain SJ, Otte AP, Magnuson T. The murine polycomb group protein Eed is required for global histone H3 lysine-27 methylation. Curr Biol. 2005;15:942–947. [PubMed]
6. Manzur KL, Farooq A, Zeng L, Plotnikova O, Koch AW, Sachchidanand, Zhou MM. A dimeric viral SET domain methyltransferase specific to Lys27 of histone H3. Nat Struct Biol. 2003;10:187–196. [PubMed]
7. Mujtaba S, Manzur KL, Gurnon JR, Kang M, Van Etten JL, Zhou MM. Epigenetic transcriptional repression of cellular genes by a viral SET protein. Nat Cell Biol. 2008;10:1114–1122. [PMC free article] [PubMed]
8. Wei H, Zhou MM. Dimerization of a viral SET protein endows its function. Proc Natl Acad Sci U S A. 2010;107:18433–18438. [PubMed]
9. Knight ZA, Shokat KM. Chemical genetics: where genetics and pharmacology meet. Cell. 2007;128:425–430. [PubMed]
10. Lin Q, Jiang F, Schultz PG, Gray NS. Design of allele-specific protein methyltransferase inhibitors. J Am Chem Soc. 2001;123:11608–11613. [PubMed]
11. Islam K, Zheng W, Yu H, Deng H, Luo M. Expanding Cofactor Repertoire of Protein Lysine Methyltransferase for Substrate Labeling. ACS Chem Biol. 2011;6:679–684. [PMC free article] [PubMed]
12. Wang R, Zheng W, Yu H, Deng H, Luo M. Labeling substrates of protein arginine methyltransferase with engineered enzymes and matched S-adenosyl-L-methionine analogues. J Am Chem Soc. 2011;133:7648–7651. [PMC free article] [PubMed]
13. Nakagawa I, Hata T. A convenient method for the synthesis of 5′-S-alkylthio-5′-deoxyribonucleosides. Tetrahedron Lett. 1975;17:1409–1412.
14. Nakagawa I, Aki K, Hata T. Synthesis of 5′-alkylthio-5′-deoxynecleosides from nucleosides in a one-pot reaction. J Chem Soc Perkin Trans I. 1983:1315–1318.
15. Overberger CG, Anselme J. A convenient synthesis of phenyldiazomethane. J Org Chem. 1963;28:592–593.
16. Christensen LF, Broom AD. Specific chemical synthesis of ribonucleoside O-benzyl ethers. J Org Chem. 1972;37:3398–3401. [PubMed]
17. Chien TC, Chen CS, Yu FH, Chern JW. Nucleosides XI. Synthesis and antiviral evaluation of 5′-alkylthio-5′-deoxy quinazolinone nucleoside derivatives as S-adenosyl- L-homocysteine analogs. Chem Pharm Bull (Tokyo) 2004;52:1422–1426. [PubMed]
18. Kozai S, Fuzikawa T, Harumoto K, Maruyama T. A new method for the synthesis of 2′-O-benzyladenosine using Mitsunobu reaction. Nucleosides Nucleotides Nucleic Acids. 2003;22:145–151. [PubMed]
19. Serafinowski P. A convenient preparation of S-adenosylhomocysteine and its analogues. Synthesis. 1985:926–928.
20. van Tilburg EW, von Frijtag Drabbe Kunzel J, de Groote M, Vollinga RC, Lorenzen A, APIJ N6,5′-Disubstituted adenosine derivatives as partial agonists for the human adenosine A3 receptor. J Med Chem. 1999;42:1393–1400. [PubMed]
21. Miles RW, Nielsen LP, Ewing GJ, Yin D, Borchardt RT, Robins MJ. S-homoadenosyl-L-cysteine and S-homoadenosyl-L-homocysteine. Synthesis and binding studies of hon-hydrolyzed substrate analogues with S-adenosyl- L-homocysteine hydrolase. J Org Chem. 2002;67:8258–8260. [PubMed]
22. Samejima K, Nakazawa Y, Matsunaga I. Improved synthesis of decarboxylated S-adenosylmethionine and related sulfonium compounds. Chem Pharm Bull (Tokyo) 1978;26:1480–1485.
23. Milton S, Yeheskiely RE, Stromberg R. Synthesis of a 2′-O-(carbomoylmethyl)ribonucleoside H-phosphonate building block and a model dinucleotide. Nucleosides Nucleotides Nucleic Acids. 2007;26:1495–1499. [PubMed]