Analytical Biochemistry 423 (2012) 171–177

Contents lists available at SciVerse ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Detecting S-adenosyl-L-methionine-induced conformational changeof a histone methyltransferase using a homogeneous time-resolvedfluorescence-based binding assay

Ying Lin a,1, Hong Fan a,1, Mathias Frederiksen b, Kehao Zhao a, Lei Jiang a, Zhaofu Wang a, Shaolian Zhou a,Weihui Guo a, Jun Gao a, Shu Li c, Edmund Harrington c, Peter Meier b, Clemens Scheufler b, Yao-Chang Xu a,Peter Atadja a, Chris Lu a, En Li a, X. Justin Gu a,⇑a China Novartis Institute for Biomedical Research, Pudong New Area, Shanghai 201203, Chinab Novartis Institute for Biomedical Research, CH-4002 Basel, Switzerlandc Novartis Institute for Biomedical Research, Cambridge, MA 02139, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 14 September 2011Received in revised form 16 January 2012Accepted 19 January 2012Available online 27 January 2012

Keywords:HMTUncompetitive inhibitorG9aBIX-01294SAMSinefunginConformational changeSET domainCJP702

0003-2697/$ - see front matter � 2012 Elsevier Inc. Adoi:10.1016/j.ab.2012.01.019

⇑ Corresponding author.E-mail address: justin.gu@novartis.com (X. Justin G

1 These authors contributed equally to this article.2 Abbreviations used: HMT, histone methyltransferas

nine; KMT, lysine methyltransferase; SAH, S-ademechanism of action; HTRF, homogeneous time-restime-resolved fluorescence resonance energy transfeGSH, glutathione; DTT, dithiothreitol; LC, liquid chrotrometry; IS, internal standard; FA, formic acid; GST,streptavidin; APC, allophycocyanin.

A homogeneous time-resolved fluorescence (HTRF)-based binding assay has been established to measurethe binding of the histone methyltransferase (HMT) G9a to its inhibitor CJP702 (a biotin analog of theknown peptide–pocket inhibitor, BIX-01294). This assay was used to characterize G9a inhibitors. Asexpected, the peptide–pocket inhibitors decreased the G9a–CJP702 binding signal in a concentration-dependent manner. In contrast, the S-adenosyl-L-methionine (SAM)–pocket compounds, SAM and sine-fungin, significantly increased the G9a–CJP702 binding signal, whereas S-adenosyl-L-homocysteine(SAH) showed minimal effect. Enzyme kinetic studies showed that CJP702 is an uncompetitive inhibitor(vs. SAM) that has a strong preference for the E:SAM form of the enzyme. Other data presented suggestthat the SAM/sinefungin-induced increase in the HTRF signal is secondary to an increased E:SAM orE:sinefungin concentration. Thus, the G9a–CJP702 binding assay not only can be used to characterizethe peptide–pocket inhibitors but also can detect the subtle conformational differences induced by thebinding of different SAM–pocket compounds. To our knowledge, this is the first demonstration of usingan uncompetitive inhibitor as a probe to monitor the conformational change induced by compound bind-ing with an HTRF assay.

� 2012 Elsevier Inc. All rights reserved.

Histone methyltransferases (HMTs)2 belong to a family ofenzymes that catalyze the transfer of a methyl group from the cofac-tor S-adenosyl-L-methionine (SAM) to either the lysine or arginineresidues of their target proteins [1–4]. Together with other histone-and DNA-modifying enzymes, such as histone demethylases (HDMs),histone acetyltransferases (HATs), histone deacetylases (HDACs),and DNA methyltransferases (DNMTs), HMTs play critical roles inwriting the ‘‘epigenetic code’’ that regulates gene expression [5–8].

ll rights reserved.

u).

e; SAM, S-adenosyl-l-methio-nosyl-l-homocysteine; MoA,olved fluorescence; TR–FRET,r; b-ME, b-mercaptoethanol;matography; MS, mass spec-glutathione S-transferase; SA,

More than 50 lysine methyltransferases (KMTs) have been iden-tified in humans and other mammals. With the exception ofDOT1L, all KMTs contain a conserved SET domain that, togetherwith the flanking sequences, forms the catalytic core [9,10]. Inthe SET domain-containing HMTs, the peptide binding pocketand the SAM binding pocket are well separated and located onthe opposite sides of the catalytic domain. These two pockets areconnected by a channel through which the alkyl-amine side chainof the target lysine passes to position its epsilon amine moiety intothe catalytic center for the methyltransfer reaction [9,10].

The catalytic mechanism of the HMT-mediated reaction isbelieved to be the SN2 type. The epsilon amine of the target lysinefunctions as the nucleophile that attacks the electron deficit carboncenter of the SAM methyl group [9,10]. In this mechanism, peptideand SAM need to bind to the enzyme simultaneously before thenucleophilic substitution reaction occurs. Thus, this catalytic mech-anism dictates that the kinetic mechanism for the HMT-mediatedreaction is sequential, and the formation of an E:SAM:peptideternary complex is required for catalysis. Depending on how the

172 Detecting HMT conformational change using HTRF assay / Y. Lin et al. / Anal. Biochem. 423 (2012) 171–177

ternary complex is formed, the sequential mechanism can befurther classified into either an ordered mechanism or a randommechanism. Published data plus our unpublished studies suggestthat HMT-mediated reactions range from a purely random mecha-nism, where either substrate can bind first to the enzyme, to astrictly ordered mechanism, where SAM binds first (Refs. [11,12]and unpublished data).

Significant progress has been made during recent years in iden-tifying KMT inhibitors [13–17]. In the SAM–pocket, the reactionproduct S-adenosyl-L-homocysteine (SAH) and its analog sinefun-gin have been identified as broad-spectrum ‘‘pan’’ HMT inhibitors.Selective inhibitors designed on the basis of SAH/SAM/sinefunginhave recently been reported for DOT1L [16]. A fungal metabolite,chaetocin, was also reported as a SAM-competitive inhibitor forSuv39H1 [18], although the mechanism of action (MoA) for thiscompound is not very clear yet. In the peptide binding site, BIX-01294 is the first non-peptidomimetic compound reported thatselectively inhibits G9a and GLP1 [13]. It behaved as a noncompet-itive inhibitor versus SAM [13]. Further optimization of this com-pound yielded more potent analogs with good cellular activityand reduced off-target toxicity [19]. One of these analogs,UNC0638, has been declared as a chemical probe for G9a/GLP bythe Structural Genomics Consortium (SGC, http://www.thesgc.org/chemical_probes/UNC0638/#overview). Another peptide–pocketinhibitor, AZ505, has recently been reported to selectively inhibitSmyD2, a p53K370 methylase [17]. For arginine methyltransfer-ases, the indole- and pyrazole-based CARM1 inhibitors have re-cently been shown to target the arginine binding cavity and thesurrounding pocket [20].

A number of enzymatic and binding assays (e.g., enzyme–substrate and/or enzyme–inhibitor binding) can be used to charac-terize enzyme–inhibitor interaction. One of the commonly usedbinding assays for high-throughput inhibitor identification andcharacterization is the homogeneous time-resolved fluorescence(HTRF) (or time-resolved fluorescence resonance energy transfer[TR–FRET]) assay. TR–FRET-based binding assays for kinases usingthe inhibitor probes (e.g., biotinylated staurosporine) have been re-ported [21,22] and are being used to characterize ATP–pocketinhibitors. Another example is an HTRF assay for measuring thebinding of biotin-tagged geldanamycin to HSP90, a chaperone pro-tein with an ATP binding pocket, to screen the HSP90 inhibitors[23].

In this article, we report the development of an HTRF assay formeasuring the binding of CJP702 (a biotin-tagged BIX-01294 ana-log) to G9a. This assay has been optimized into a robust and sensi-tive assay suitable for the identification and characterization ofnew peptide–pocket inhibitors of G9a. In addition, our datashowed that CJP702 is also an uncompetitive inhibitor (vs. SAM)that binds preferentially to the E:SAM form of the enzyme. Further-more, the affinity of CJP702 to G9a can be affected by the binding ofother SAM–pocket compounds (e.g., sinefungin) besides SAM.Therefore, the developed G9a–CJP702 binding assay not only canbe used to characterize the peptide–pocket compounds but alsocan detect the conformational change induced by the binding ofSAM–pocket compounds.

Materials and methods

G9a protein expression and purification

Human G9a complementary DNA (cDNA) encoding amino acidresidues 913 to 1193 was synthesized by GenScript. The fragmentwas subcloned to pGEX-KG vector using restriction sites EcoRI andXhoI and verified by sequencing. The construct was transformedinto Escherichia coli BL21(DE3). A fresh single colony was

inoculated into Luria–Bertani (LB) medium and grows to anOD600 of approximately 0.4, and protein expression was inducedwith 0.5 mM isopropyl b-D-1-thiogalactopyranoside (IPTG)overnight at 16 �C. The cells were collected by centrifugation. Forpurification, the cell pellet was resuspended in buffer A (50 mMTris [pH 7.5], 200 mM NaCl, 5% glycerol, 5 mM b-mercaptoethanol[b-ME], 1 mM phenylmethanesulfonyl fluoride [PMSF], and 1%[v/v] Triton X-100). Cells were sonicated for 15 min on ice (5 son/5 s off), and cell debris was then removed by centrifugation.The supernatant was loaded to a HiTrap GST column preequilibrat-ed with buffer B (buffer A minus Triton X-100). The column waswashed at least 30 times of the column volume or until no proteincould be detected in the wash solution. The protein was eluted bybuffer C (50 mM Tris [pH 7.5], 200 mM NaCl, 5% glycerol, 5 mMb-ME, and 20 mM glutathione [GSH]). The most pure fractionsbased on sodium dodecyl sulfate–polyacrylamide gel electrophore-sis (SDS–PAGE) results were collected and subjected to overnightdialysis against a large volume of buffer D (50 mM Tris [pH 7.5],200 mM NaCl, 5% glycerol, and 5 mM b-ME) to remove GSH. Theprotein was further purified by gel filtration with a Superdex 200column in buffer E (50 mM Tris [pH 7.5], 150 mM NaCl, and5 mM b-ME), concentrated, aliquoted, and stored at �80 �C in afreezer. The purity of the protein was approximately 95%.

LC/MS-based G9a enzymatic assay

The enzymatic reactions (16 ll) were performed in a white,384-well ProxiPlate (PerkinElmer) at room temperature for 2 h inassay buffer (20 mM Tris–HCl [pH 8.0], 0.01% Tween 20, 10 mMMgCl2, and 1 mM dithiothreitol [DTT]) containing the indicatedamounts of SAM, peptide, and enzyme. The reactions were stoppedby the addition of 4 ll quench solution (2.5% trifluoroacetic acid[TFA] and 320 nM D4–SAH).

The amount of SAH produced from the reactions was measuredusing an API4000 liquid chromatography/tandem mass spectrom-etry (LC/MS/MS) system. D4–SAH was used as an internal standard(IS) for SAH detection and normalization. LC was performed on aChromolith FastGradient HPLC column (RP-18e, 25–2 mm, Merck).The column was connected to the mass spectrometer through asix-port valve. The turbo ion electrospray was operated in thepositive ion mode. The m/z values for the parent ions of SAH andD4–SAH are 385.1 and 389.1, respectively, and the m/z values forthe daughter ions of SAH and D4–SAH both are 136.1. Mobile phaseA is 0.02% formic acid (FA) and 2% methanol in water, and B is 0.1%FA in acetonitrile. The injection volume was 4 ll, and the autosam-pler was kept at 4 �C. The eluants between 0.4 and 1.0 min were di-verted to the mass spectrometer for analysis. The plot of SAH peakarea/IS peak area versus SAH concentration was used to generatethe normalization factor of SAH. The production of SAH from realenzymatic reaction was derived from the standard curve of SAH.The down-limit of our system for the detection of SAH is approxi-mately 1 to 2 nM, and the linear range can reach up to 400 nM.

In the case of peptide detection, the peptide conversion ratiowas calculated by dividing the peak area of the corresponding pep-tide by the total peak area of all the peptides on the assumptionthat the ionization efficiency of the peptides with a different meth-ylation state is the same as that of their unmethylated counterpart.

Initial velocity studies in the presence of inhibitor and data analysis

The initial velocity studies in the presence of the inhibitor wereperformed using LC/MS assay. The initial rates were measured withone reactant (either SAM or peptide) fixed at Km, whereas the con-centration of the other reactant and inhibitor vary. The data werefit into either Eq. (1) for noncompetitive inhibition or Eq. (2) foruncompetitive inhibition using the nonlinear regression fitting

Detecting HMT conformational change using HTRF assay / Y. Lin et al. / Anal. Biochem. 423 (2012) 171–177 173

method in SigmaPlot 11.0, where v and Vmax are initial and maxi-mum velocities, respectively, A is substrate concentration, I isinhibitor concentration, Ka is Michaelis constant for substrate A,and Kis and Kii represent inhibition constants for the slope andintercept, respectively:

v ¼ VmaxA=½Kað1þ I=KisÞ þ Að1þ I=KiiÞ� ð1Þ

v ¼ VmaxA=½Ka þ Að1þ I=KiiÞ� ð2Þ

G9a–CJP702 binding assay

The binding assay was performed in a white, 384-well Proxi-Plate (PerkinElmer). Assay mixture 1 (typical 6 ll) containing theindicated concentrations of glutathione S-transferase (GST)–G9a,CJP702, and compounds in the assay buffer (20 mM Tris–HCl [pH8.0], 0.01% Tween 20, 10 mM MgCl2, and 1 mM DTT) were first pre-pared. The mixture was then added to an equal volume of the HTRFdetection solution containing 14 lg/ml PhycoLink streptavidin–allophycocyanin (SA–APC, product code PJ25S, Prozyme) and2 nM anti-GST-Eu (cat. No. 61GSTKLA, Cisbio) in detection buffer(20 mM Tris–HCl [pH 7.0], 0.01% Tween 20, 150 mM NaCl, 0.2% bo-vine serum albumin [BSA], and 0.8 M KF). After 2 h of incubation,the plate was read in the Envision reader using the HTRF protocol(Lance/Delfia dual enhancer mirror, excitation at 320 UV2(TRF),emission at APC665, second emission at Europium 615).

For kinetic reading, assay mixture 1 (containing GST–G9a andCJP702 in assay buffer) and the HTRF detection solution (contain-ing SA–APC and anti-GST–Eu in the detection buffer) were preincu-bated for 2 h. SAM, SAH, or sinefungin was then added, and theplate was read immediately and for every 20 s thereafter for theindicated time.

Quantifying the amount of SAM and SAH in G9a prep

The LC/MS method was used to detect the enzyme-associatedSAM and SAH. For SAH detection, the protocol is the same as usedin the enzymatic assay. For SAM detection/quantification, D3–SAMis used as the IS. The m/z values for the parent ions of SAM andD3–SAM are 399.1and 402.1, respectively. The m/z values for thedaughter ions of SAM and D3–SAM both are 136.1. The down-limitof our system for the detection of SAM is approximately 100 nM,and the linear range can reach up to 50 lM.

Results

HTRF assay for measuring binding of BIX-01294 to GST–G9a

BIX-01294 is a known inhibitor of G9a and GLP-1 [13]. Deter-mined crystal structures clearly showed that this compound andits close analogs bind to the peptide–pocket of G9a and/or GLP-1[24,25]. To establish an HTRF binding assay between G9a and thiscompound, we synthesized a biotinylated analog of BIX-01294called CJP702. CJP702 contains a biotin tag with an ethylene glycollinker attached to the piperidine nitrogen (Fig. 1A). Thus, the tagreplaces the benzyl group of BIX-01294. This modification didnot affect the activity of the compound. As shown in Fig. 1B,CJP702 and BIX-01294 showed similar potency in the G9a enzy-matic assay with IC50 values of approximately 1 lM (Fig. 1B).

The binding of CJP702 to GST-tagged G9a was then tested usingan HTRF assay in the presence of anti-GST–Eu and SA–APC. BothGST–G9a and CJP702 increased the HTRF signal in a concentra-tion-dependent manner (Fig. 1C). The signal reached saturationat approximately 500 nM for both CJP702 and GST–G9a. The

reasons why such a high concentration of GST–G9a is required togenerate a signal are discussed later.

To make sure that the HTRF signal is indeed caused by the spe-cific interaction of G9a and CJP702, we chose 500 nM GST–G9a and500 nM CJP702 and tested a few peptide–pocket inhibitors:BIX-01294, UNC0224, and UNC0638 (structures are shown inFigs. 1A and 2A) in the competition mode. These compounds havebeen reported to have different potencies in inhibiting G9a enzymeactivity [19]. In our enzymatic assay, their IC50 values range from0.05 lM for UNC0638 to 1 lM for BIX-01294 (Fig. 2B). In theHTRF-based competition assay, all three compounds concentra-tion-dependently reduced the G9a–CJP702 binding signal(Fig. 2C). The IC50 values for all three compounds in the HTRF-based competition assay are higher than those in the enzymaticassay, but the ranking order is the same. Thus, the data stronglysuggest that this HTRF binding assay indeed measures the specificinteraction between CJP702 and G9a. JAB415, a negative controlcompound that did not inhibit G9a in the enzymatic assay, showedno signal reduction in the G9a–CJP702 binding assay.

Effect of SAM–pocket compounds on binding of G9a to CJP702

We also tested a few compounds binding to the SAM–pocket,including SAM, sinefungin, and SAH, in the G9a–CJP702 binding as-say. SAM is the universal methyl donor used by all HMTs. Sinefun-gin and SAH are closely related analogs of SAM. They inhibited theG9a HMT activity with IC50 values of 30 and 1 lM, respectively, inour enzymatic assay (Fig. 2B). In the G9a–CJP702 binding assay, theaddition of SAM and sinefungin increased the HTRF signal, whereasthe addition of SAH had little effect (Fig. 2C).

The effects of these SAM–pocket compounds on the CJP702–G9a binding signal are even more dramatic if low concentrationsof GST–G9a were used in the binding assay. Under these conditions(60 nM GST–G9a and 250 nM CJP702), the HTRF signal is very lowwithout the addition of SAM (Fig. 3; also see Fig. 1C). These signalscannot be reduced by BIX-01294 and its analogs (Fig. 3), suggestingthat these are nonspecific signals. After the addition of SAM andsinefungin, a dramatic increase of HTRF signal was observed. Onthe other hand, SAH slightly increases the signal only at the highestconcentrations (Fig. 3).

Identification of CJP702 as an uncompetitive inhibitor that bindspreferentially to E:SAM and the reason(s) for SAM-induced increasein CJP702–G9a binding signal

The increase of HTRF signal after the addition of SAM isexpected because CJP702’s parental compound, BIX-01294, wasdemonstrated to inhibit G9a in an uncompetitive manner [13].To confirm that CJP702 is also an uncompetitive inhibitor (vs.SAM), we did a detailed CJP702/SAM matrix study using a quanti-tative LC/MS-based enzymatic assay we established for G9a. Thedata obtained were fitted to either an uncompetitive model or anoncompetitive model using the nonlinear regression method inSigmaPlot 11.0. As shown in Table 1, the fitting for the uncompet-itive inhibition model with Eq. (2) gives a Kii of 0.8 lM. On theother hand, the fitting for the noncompetitive model with Eq. (1)gives the same Kii value and an undefined Kis value. Fig. 4 alsoshows that CJP702 follows the uncompetitive kinetics that reducesboth the apparent Vmax and Km values with the increase in CJP702concentration. Thus, all of the data suggest that CJP702 is anuncompetitive inhibitor (vs. SAM) that interacts preferably withthe E:SAM form, as compared with the E form, of the enzyme. Theyalso suggest that binding of SAM to G9a causes a conformationalchange in the peptide–pocket that allows CJP702 binding.

If CJP702 indeed binds mainly to the E:SAM form, then the datashown in Figs. 2C and 3 would predict that both enzyme forms

Fig.1. Binding of GST–G9a to CJP702 in the HTRF-based assay. (A) Chemical structure of BIX-01294 and CJP702. (B) Inhibition of G9a’s enzymatic activity by BIX-01294 andCJP702. (C) Binding of GST–G9a to CJP702 detected in the HTRF assay.

Fig.2. Peptide–pocket inhibitors compete with CJP702 for G9a binding. (A) Chemical structure of UNC0224 and UNC0638. (B) Activities of the compounds in the LC/MS-basedG9a enzymatic assay. The inset shows the IC50 values (in lM) of the compounds in the enzymatic assay. (C) Activities of the compounds in the HTRF-based G9a–CJP702binding assay (competition mode). In this experiment, 500 nM GST–G9a and 500 nM CJP702 were used. The peptide–pocket inhibitors reduced the HTRF signal, and their IC50

values (in lM) in the binding assay are shown in the inset. On the other hand, SAM and sinefungin increased the HTRF signal. JAB451 is a negative control compound that didnot show any activity in either the enzymatic assay or the binding assay.

174 Detecting HMT conformational change using HTRF assay / Y. Lin et al. / Anal. Biochem. 423 (2012) 171–177

(E and E:SAM) should be present in the GST–G9a prep used in thebinding assay. To see whether this is the case, we then quantifiedthe enzyme-associated SAM and/or SAH in the purified GST–G9apreps. Indeed, we found approximately 0.1 to 0.2 mol of SAM for

every 1 mol of GST–G9a, whereas no SAH was detected. Thus, allof the data explain our observation nicely. The specific signal de-tected in the HTRF assay originated from the binding of CJP702to E:SAM. Without the addition of SAM, the level of E:SAM was

Fig.3. Effect of SAM–pocket compounds in the G9a–CJP702 binding assay. In thisexperiment, 250 nM CJP702 and 60 nM GST–G9a were used. Both SAM andsinefungin dramatically increase the HTRF signal, whereas SAH has minimal effect.In this experiment, the basal signal (without the addition of SAM) is likely to benonspecific because it cannot be reduced by the addition of the peptide–pocketinhibitors.

Table 1Inhibition constants obtained from nonlinear regression analysis of CJP702/SAMmatrix study data fitted to noncompetitive and uncompetitive models.

Variedsubstrate

Fixedsubstrate

Inhibitor Kis (lM) Kii (lM) Inhibitionpattern

SAM H3K9 (1–21)

CJP702 (7.7 � 106)± (1.2 � 1012)

2.5 ± 0.2 NC

2.5 ± 0.1(0.80 ± 0.03)a

UC

a The value in parentheses is the Ki corrected with the concentration of the fixedsubstrate.

Fig.4. Inhibition of G9a activity by CJP702 at different concentrations of SAM. Initialvelocities were measured at different CJP702 concentrations [0 lM (d), 1 lM (j),2 lM (N), 3 lM (.), 5 lM (�), and 8 lM (s)] and SAM concentrations (0.7, 1, 1.5, 3,and 5 lM), with peptide H3K9 at its Kapp

m . The points are experimental, whereas thelines are theoretical based on the nonlinear regression fitting to the uncompetitivemodel (Eq. (2)). For clarity, the data for 15 lM SAM are not shown here.

Fig.5. Time course of SAM- and SAH-induced changes (or lack of them) in the G9a–CJP702 binding signal. Here, 50 nM GST–G9a and 200 nM CJP702 were preincu-bated with 1 nM anti-GST–Eu and 7 lg/ml SA–APC for 2 h. After that, 1, 10, or100 nM SAM or SAH was added to the binding mixture, and the HTRF signals weremonitored every 20 s for 50 min.

Detecting HMT conformational change using HTRF assay / Y. Lin et al. / Anal. Biochem. 423 (2012) 171–177 175

low and so was the HTRF signal. After the addition of SAM, the freeE form was then converted to the E:SAM form, which led to in-creased CJP702 binding to G9a and, thus, a higher HTRF signal. Be-cause sinefungin behaved similarly to SAM in the CJP702–G9abinding assay, it is very likely that binding of sinefungin increasesthe affinity of CJP702 to G9a as well and CJP702 prefers E:sinefun-gin. Thus, the HTRF binding assay not only can be used to charac-terize peptide–pocket inhibitors but also is sensitive to(conformational) changes induced by the binding of SAM and otherSAM–pocket compounds.

The lack of effect for SAH in the CJP702–G9a binding assay is asurprise to us. SAH is a close analog of SAM. In the enzymaticassays, SAH can potently inhibit the G9a activity with an IC50 ofapproximately 1 lM (Fig. 2B). Thus, the lack of effect of SAH inthe G9a–CJP702 binding assay cannot be attributed to the lack ofbinding of SAH to all G9a enzyme forms. It is more likely thatthe conformational changes induced by SAH differ from those in-duced by SAM or sinefungin. It is also possible that SAH binds toa different enzyme form(s) from the one(s) that SAM, sinefungin,and CJP702 bind to.

Binding kinetics of SAM–pocket compounds to G9a

The above-mentioned experiments were done after 2 h of incu-bation of the compounds with the other components of the bindingreaction (including GST–G9a). To see how fast the HTRF signalschange after the addition of the SAM–pocket compounds, wetested the HTRF binding assay in the kinetic mode. Here, 50 nMGST–G9a and 200 nM CJP702 were incubated with SA–APC andanti-GST–Eu for 2 h to allow the formation of the two binary com-plexes: Eu–ab:GST–G9 and SA–APC:CJP702. SAM, sinefungin, orSAH was then added to the mixtures, and the HTRF signals weremonitored continuously for 50 min. Rapid increase in the HTRF sig-nals was observed after the addition of SAM (Fig. 5). The timeneeded to reach 50% maximum signal (tapp

1=2 for association) isapproximately 2.5 min for 10 lM SAM and 4 min for 1 lM SAM.Sinefungin showed similar kinetics as SAM (data not shown). ForSAH, there is no increase at all concentrations tested (Fig. 5).

The ‘‘binding’’ kinetics observed is likely a result of the com-bined events of SAM binding to E, conformational change afterSAM binding (E:SAM to E�:SAM), and the subsequent CJP702 bind-ing to E�:SAM. At this time, we do not know what the rate-limitingstep is. Nevertheless, the tapp

1=2 measured sets the lower limit for therate constant of the forward reactions in the three events men-tioned above.

Optimization of G9a–CJP702 binding assay for characterization ofpeptide–pocket inhibitors

The original goal of this study was to establish a binding assayto identify and characterize new peptide–pocket compounds ofG9a. With all of the knowledge gained from the studies regarding

Fig.6. Optimization of G9a–CJP702 binding assay in the presence of 100 lM SAM.Signal saturation was reach at approximately 50 nM GST–G9a and 250 nM CJP702.

Fig.7. Activities of peptide–pocket inhibitors in optimized CJP702–G9a bindingassay. The assay contained 250 nM CJP702, 10 nM GST–G9a, and 100 lM SAM.Peptide–pocket inhibitors potently inhibited the binding of CJP702 to GST–G9a(IC50 values [in lM] are listed in the inset). Note that SAM and sinefungin did notfurther increase the signal under this condition, as expected.

176 Detecting HMT conformational change using HTRF assay / Y. Lin et al. / Anal. Biochem. 423 (2012) 171–177

the effect of SAM and other SAM–pocket compounds, we wentback and optimized the HTRF binding assay with the addition of100 lM SAM. In the presence of SAM, a significantly lower GST–G9a concentration (�50 nM vs. �500 nM in the absence of SAM)was needed to reach signal saturation (Fig. 6). We then chose10 nM GST–G9a, 250 nM CJP702, and 100 lM SAM as the new‘‘optimized’’ conditions to retest all of the compounds. As shownin Fig. 7, under this new condition, the IC50 values for all of the pep-tide–pocket inhibitors are much smaller compared with the dataobtained from the previous condition (cf. Fig. 2C). They are now al-most identical to their IC50 values in the enzymatic assay (Fig. 2B).Under the optimized conditions (with the presence of 100 lMSAM), no further increase in signal was observed after the additionof SAM or sinefungin, as expected (Fig. 7).

In summary, we have developed and optimized an HTRF bind-ing assay measuring the interaction between G9a and its inhibitor,CJP702. This assay is sensitive and robust in identifying peptide–pocket inhibitors of G9a. More important, we also found thatCJP702 is an uncompetitive inhibitor (vs. SAM) for G9a and thatit binds mainly to the E:SAM form of the enzyme. As a result, theG9a–CJP702 HTRF assay was also capable of detecting the bindingof SAM and some SAM–pocket inhibitors (e.g., sinefungin). To ourknowledge, this is the first report of using an uncompetitive inhib-itor as a probe to detect the conformational changes induced by thebinding of compounds/inhibitors using an HTRF assay.

Discussion

HTRF- (or TR–FRET)-based assays measuring the binding of aninhibitor probe to its target are commonly used for the identifica-tion and characterization of new inhibitors that compete directlywith the probe. This method has been successfully demonstratedfor protein kinases [21,22], HSP90 [23], and other targets. Thereport here is the first demonstration of such an approach beingapplied to an SET domain-containing HMT. Even though CJP702used in this study is too specific to be used broadly besides G9aand GLP-1, the general principle is applicable for other HMTs.

Another interesting observation in this report is that SAM andsinefungin increased the CJP702–G9a binding signal, whereasSAH had little effect. This suggests that the binding assay is capableof detecting the conformational changes induced by the binding ofSAM and SAM–pocket inhibitors. The identification of BIX-01294[13] and CJP702 as uncompetitive (vs. SAM) inhibitors might be aharbinger of more SAM uncompetitive HMT inhibitors consideringthat the kinetic mechanisms for a number of HMTs might beordered with SAM binding first (data not shown). If this is the case,the HMTs will be more like the dehydrogenase and reductasefamily of enzymes and different from protein kinases in terms ofkinetic mechanism and MoA of inhibitors. For the dehydrogenasesand reductases, an ordered mechanism with nicotinamide-containing cofactors binding first is common, and a number ofuncompetitive inhibitors have been successfully developed andare being used in the clinic. For protein kinases, a random mecha-nism is likely to be prevalent [26], and an uncompetitive inhibitor(vs. ATP) is very rare. Even for the type 3 inhibitors that do not oc-cupy the ATP pocket, most of them seem to be noncompetitive(rather than uncompetitive) versus ATP. For example, the type 3MEK inhibitor, CI-1040, behaves as a ‘‘pure’’ noncompetitive inhib-itor that binds to E:ATP and E form of the enzyme with similaraffinity in the MEK enzymatic assay [27].

The dramatic increase in HTRF signal induced by the addition ofSAM in the G9a–CJP702 binding assay is another reminder of theimportance of choosing the right enzyme form when screeningfor HMT inhibitors. This is particularly important when affinity-based biophysical methods (e.g., nuclear magnetic resonance[NMR], surface plasmon resonance [SPR], MS based) are being usedas the screening methods. Choosing the right enzyme form basedon the knowledge and understanding of the kinetic mechanismof the target enzyme will be critical for the success of such ascreening campaign.

Acknowledgments

We thank Martin Klumpp, Zhengtian Yu, Wen Xiong, JianyongShou, and Jingquan Dai for their suggestions and help, and wethank Zhui Chen for critical reading of this manuscript.

References

[1] M. Albert, K. Helin, Histone methyltransferases in cancer, Sem. Cell Dev. Biol.21 (2010) 209–220.

[2] C. Qian, M.M. Zhou, SET domain protein lysine methyltransferases: structure,specificity, and catalysis, Cell. Mol. Life Sci. 63 (2006) 2755–2763.

[3] S.C. Dillon, X. Zhang, R.C. Trievel, X. Cheng, The SET-domain proteinsuperfamily: protein lysine methyltransferases, Genome Biol. 6 (2005) 227.

[4] V.M. Richon, D. Johnston, C.J. Sneeringer, L. Jin, C.R. Majer, K. Elliston, L.F. Jerva,M.P. Scott, R.A. Copeland, Chemogenetic analysis of human proteinmethyltransferases, Chem. Biol. Drug Des. 78 (2011) 199–210.

[5] E. Li, Chromatin modification and epigenetic reprogramming in mammaliandevelopment, Nat. Rev. Genet. 3 (2002) 662–673.

[6] P. Chi, C.D. Allis, G.G. Wang, Covalent histone modifications—miswritten,misinterpreted and mis-erased in human cancers, Nat. Rev. Cancer 10 (2010)457–469.

[7] C. Martin, Y. Zhang, The diverse functions of histone lysine methylation, Nat.Rev. Mol. Cell Biol. 6 (2005) 838–849.

Detecting HMT conformational change using HTRF assay / Y. Lin et al. / Anal. Biochem. 423 (2012) 171–177 177

[8] J.F. Couture, R.C. Trievel, Histone-modifying enzymes: encrypting an enigmaticepigenetic code, Curr. Opin. Struct. Biol. 16 (2006) 753–760.

[9] J.R. Wilson, C. Jing, P.A. Walker, S.R. Martin, S.A. Howell, G.M. Blackburn, S.J.Gamblin, B. Xiao, Crystal structure and functional analysis of the histonemethyltransferase SET7/9, Cell 111 (2002) 105–115.

[10] B. Xiao, C. Jing, J.R. Wilson, P.A. Walker, N. Vasisht, G. Kelly, S. Howell, I.A.Taylor, G.M. Blackburn, S.J. Gamblin, Structure and catalytic mechanism of thehuman histone methyltransferase SET7/9, Nature 421 (2003) 652–656.

[11] D. Patnaik, H.G. Chin, P.O. Estève, J. Benner, S.E. Jacobsen, S. Pradhan, Substratespecificity and kinetic mechanism of mammalian G9a histone H3methyltransferase, J. Biol. Chem. 279 (2004) 53248–53258.

[12] H.G. Chin, D. Patnaik, P.O. Estève, S.E. Jacobsen, S. Pradhan, Catalytic propertiesand kinetic mechanism of human recombinant Lys-9 histone H3methyltransferase SUV39H1: participation of the chromodomain inenzymatic catalysis, Biochemistry 45 (2006) 3272–3284.

[13] S. Kubicek, R.J. O’Sullivan, E.M. August, E.R. Hickey, Q. Zhang, M.L. Teodoro, S.Rea, K. Mechtler, J.A. Kowalski, C.A. Homon, T.A. Kelly, T. Jenuwein, Reversal ofH3K9me2 by a small-molecule inhibitor for the G9a histonemethyltransferase, Mol. Cell 25 (2007) 473–481.

[14] R.A. Copeland, M.E. Solomon, V.M. Richon, Protein methyltransferases as atarget class for drug discovery, Nat. Rev. Drug Discov. 8 (2009) 724–732.

[15] R.A. Copeland, E.J. Olhava, M.P. Scott, Targeting epigenetic enzymes for drugdiscovery, Curr. Opin. Chem. Biol. 14 (2010) 505–510.

[16] S.R. Daigle, E.J. Olhava, C.A. Therkelsen, C.R. Majer, C.J. Sneeringer, J. Song, L.D.Johnston, M.P. Scott, J.J. Smith, Y. Xiao, L. Jin, K.W. Kuntz, R. Chesworth, M.P.Moyer, K.M. Bernt, J.C. Tseng, A.L. Kung, S.A. Armstrong, R.A. Copeland, V.M.Richon, R.M. Pollock, Selective killing of mixed lineage leukemia cells by apotent small-molecule DOT1L inhibitor, Cancer Cell 20 (2011) 53–65.

[17] A.D. Ferguson, N.A. Larsen, T. Howard, H. Pollard, I. Green, C. Grande, T.Cheung, R. Garcia-Arenas, S. Cowen, J. Wu, R. Godin, H. Chen, N. Keen,Structural basis of substrate methylation and inhibition of SMYD2, Structure19 (2011) 1262–1273.

[18] D. Greiner, T. Bonaldi, R. Eskeland, E. Roemer, A. Imhof, Identification of aspecific inhibitor of the histone methyltransferase SU(VAR)3–9, Nat. Chem.Biol. 1 (2005) 143–145.

[19] F. Liu, X. Chen, A. Allali-Hassani, A.M. Quinn, T.J. Wigle, G.A. Wasney, A. Dong,G. Senisterra, I. Chau, A. Siarheyeva, J.L. Norris, D.B. Kireev, A. Jadhav, J.M.

Herold, W.P. Janzen, C.H. Arrowsmith, S.V. Frye, P.J. Brown, A. Simeonov, M.Vedadi, J. Jin, Protein lysine methyltransferase G9a inhibitors: design,synthesis, and structure activity relationships of 2,4-diamino-7-aminoalkoxy-quinazolines, J. Med. Chem. 53 (2010) 5844–5857.

[20] J.S. Sack, S. Thieffine, T. Bandiera, M. Fasolini, G.J. Duke, L. Jayaraman, K.F. Kish,H.E. Klei, A.V. Purandare, P. Rosettani, S. Troiani, D. Xie, J.A. Bertrand, Structuralbasis for CARM1 inhibition by indole and pyrazole inhibitors, Biochem. J. 436(2011) 331–339.

[21] C.S. Lebakken, S.M. Riddle, U. Singh, W.J. Frazee, H.C. Eliason, Y. Gao, L.J.Reichling, B.D. Marks, K.W. Vogel, Development and applications of a broad-coverage, TR–FRET-based kinase binding assay platform, J. Biomol. Screen. 14(2009) 924–935.

[22] J. Kwan, A. Ling, E. Papp, D. Shaw, J.M. Bradshaw, A fluorescence resonanceenergy transfer-based binding assay for characterizing kinase inhibitors:important role for C-terminal biotin tagging of the kinase, Anal. Biochem. 395(2009) 256–262.

[23] V. Zhou, S. Han, A. Brinker, H. Klock, J. Caldwell, X.J. Gu, A time-resolvedfluorescence resonance energy transfer-based HTS assay and a surfaceplasmon resonance-based binding assay for heat shock protein 90 inhibitors,Anal. Biochem. 331 (2004) 349–357.

[24] Y. Chang, X. Zhang, J.R. Horton, A.K. Upadhyay, A. Spannhoff, J. Liu, J.P. Snyder,M.T. Bedford, X. Cheng, Structural basis for G9a-like protein lysinemethyltransferase inhibition by BIX-01294, Nat. Struct. Mol. Biol. 16 (2009)312–317.

[25] F. Liu, X. Chen, A. Allali-Hassani, A.M. Quinn, G.A. Wasney, A. Dong, D. Barsyte,I. Kozieradzki, G. Senisterra, I. Chau, A. Siarheyeva, D.B. Kireev, A. Jadhav, J.M.Herold, S.V. Frye, C.H. Arrowsmith, P.J. Brown, A. Simeonov, M. Vedadi, J. Jin,Discovery of a 2,4-diamino-7-aminoalkoxyquinazoline as a potent andselective inhibitor of histone lysine methyltransferase G9a, J. Med. Chem. 52(2009) 7950–7953.

[26] A.E. Szafranska, K.N. Dalby, Kinetic mechanism for p38 MAP kinase a: apartial rapid-equilibrium random-order ternary-complex mechanism forthe phosphorylation of a protein substrate, FEBS J. 272 (2005) 4631–4645.

[27] W.S. Van Scyoc, G.A. Holdgate, J.E. Sullivan, W.H. Ward, Enzyme kinetics andbinding studies on inhibitors of MEK protein kinase, Biochemistry 47 (2008)5017–5027.