Long Residence Time Inhibition of EZH2 in Activated Polycomb

Letters pubs.acs.org/acschemicalbiology

Long Residence Time Inhibition of EZH2 in Activated Polycomb Repressive Complex 2 Glenn S. Van Aller,*,† Melissa Baker Pappalardi,† Heidi M. Ott,† Elsie Diaz,‡ Martin Brandt,‡ Benjamin J. Schwartz,‡ William H. Miller,† Dashyant Dhanak,† Michael T. McCabe,† Sharad K. Verma,† Caretha L. Creasy,† Peter J. Tummino,† and Ryan G. Kruger† †

Cancer Epigenetics Discovery Performance Unit, Cancer Research, Oncology R&D and ‡Platform Technology and Sciences, GlaxoSmithKline, 1250 S. Collegeville Road, Collegeville, Pennsylvania 19426, United States S Supporting Information *

ABSTRACT: EZH2/PRC2 catalyzes transcriptionally repressive methylation at lysine 27 of histone H3 and has been associated with numerous cancer types. Point mutations in EZH2 at Tyr641 and Ala677 identified in non-Hodgkin lymphomas alter substrate specificity and result in increased trimethylation at histone H3K27. Interestingly, EZH2/PRC2 is activated by binding H3K27me3 marks on histones, and this activation is proposed as a mechanism for self-propagation of gene silencing. Recent work has identified GSK126 as a potent, selective, SAM-competitive inhibitor of EZH2 capable of globally decreasing H3K27 trimethylation in cells. Here we show that activation of PRC2 by an H3 peptide trimethylated at K27 is primarily an effect on the rate-limiting step (kcat) with no effect on substrate binding (Km). Additionally, GSK126 is shown to have a significantly longer residence time of inhibition on the activated form of EZH2/PRC2 as compared to unactivated EZH2/PRC2. Overall inhibition constant (Ki*) values for GSK126 were determined to be as low as 93 pM and appear to be driven by slow dissociation of inhibitor from the activated enzyme. The data suggest that activation of EZH2 allows the enzyme to adopt a conformation that possesses greater affinity for GSK126. The long residence time of GSK126 may be beneficial in vivo and may result in durable target inhibition after drug systemic clearance.

E

binding of H3K27me3 to the methyltransferase activity of EZH2,7 presumably through the transduction of a conformation change into the active site of EZH2. In this study, we use steady state kinetics to examine the allosteric activation of wild type and mutant EZH2/PRC2. We report the kinetic parameters for GSK1263 (Figure 2A, compound 1), a novel inhibitor of EZH2 methyltransferase activity, using allosterically activated wild type and mutant EZH2/PRC2. GSK126 exhibits longer residence times on activated EZH2/PRC2 than on the unactivated form, and this phenomenon occurs in both wild type and mutant EZH2. Structure−activity relationships demonstrate that specific chemical features contribute to these longer residence times. These data suggest that longer residence times can be designed into EZH2 inhibitors and may contribute to the GSK126induced antiproliferative effects observed in lymphoma cells. EZH2/PRC2 Methyltransferase Activity Is Enhanced by Histone H3 Lysine 27 Trimethylation. The effect of histone H3K27me3 peptide on EZH2/PRC2 methyltransferase activity was quantified under steady state conditions using

nhancer of zeste homologue 2 (EZH2) is the catalytic component of polycomb repressive complex 2 (PRC2), which represses gene expression through methylation of lysine 27 on histone H3. High frequency heterozygous point mutations in the SET domain of EZH2 have been identified in diffuse large B-cell lymphomas (DLBCL) and follicular lymphomas.1,2 These mutations (A677G and Y641x where x can be F, N, S, H, or C) increase the activity of EZH2 toward dimethylated substrates, resulting in higher levels of H3K27 trimethylation in EZH2 mutant lymphoma cells. Reported data have provided evidence that inhibition of EZH2 methyltransferase activity may be a viable strategy for the treatment of DLBCL and follicular lymphoma harboring mutations in EZH2.3−5 In addition to EZH2, at least two other proteins, SUZ12 and EED, are required for the histone methyltransferase activity of PRC2. Trimethylated lysines, in particular H3K27me3, have been shown to bind EED and allosterically activate EZH2/ PRC2.6 The recognition of pre-existing H3K27me3 is proposed as a mechanism to maintain repressed chromatin domains by re-establishing H3K27 trimethylation onto nascent nucleosomes being incorporated during DNA synthesis. Electron microscopy and cross-linking studies suggest that the SANT domains of EZH2 may be responsible for coupling EED © XXXX American Chemical Society

Received: April 16, 2013 Accepted: December 4, 2013

A

dx.doi.org/10.1021/cb4008748 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Letters

Figure 1. Activation of EZH2/PRC2 by a H3K27 trimethylated peptide. (A) EZH2 activity was quantified using 2 μM peptide substrate over a range of SAM and H3K27me3 activator peptide concentrations (○ 1.3, ● 0.63, □ 0.31, ■ 0.16, Δ 0.08, ▲ 0.04, ▽0.02, ▼ 0.01 μM, and ◇ control) and fit to a model of nonessential activation.8 No effect was observed on SAM or peptide Km (α = 1.1). An increase in kcat is indicated by the higher turnover (β = 7.4). (B) A diagram describing the nonessential activation of EZH2/PRC2 where kcat is the rate-limiting step for the unactivated enzyme and βkcat is the rate in the presence of a saturating concentration of activator peptide. S is the substrate concentration, Ks is the Michaelis constant for substrate, A is the activator peptide concentration, and Ka is the activation constant. The factor α is the factor by which Ks changes when activator peptide is bound to EZH2/PRC2. (C) EZH2 reaction velocity measured at fixed SAM (0.6 μM) and the indicated concentrations of reconstituted nucleosome core particles (NCP) without pre-existing epigenetic marks across a range of activator peptide concentrations (○ 50, ● 25, □ 12.5, ■ 6.2, Δ 3.1 ▲ 1.6, ▽ 0.8 μM and ◇ control). (D) A parallel experiment to panel C using reconstituted NCP heterodimers with preexisting trimethylation on histone H3 at Kc27 across a range of activator peptide concentrations (○ 50, ● 25, □ 12.5, ■ 6.2, Δ 3.1 ▲ 1.6, ▽ 0.8, ▼ 0.4 μM, and ◇ control).

tritiated S-adenosyl-methionine (SAM) and unmodified H3 peptide as substrates (Supplementary Figure 1A). Product formation as a function of enzyme concentration increased 5fold with the addition of 100 nM H3K27me3 activator peptide. There was no product formation detected in the absence of the unmethylated substrate, confirming that the trimethylated peptide could not be used by EZH2 as a substrate. To understand the activation mechanism in more detail, double titration experiments were carried out in which activator peptide was titrated against either SAM or peptide substrate in two separate experiments and the resulting data were fit to a kinetic model of nonessential enzyme activation (Figure 1B, ref 8). Activator peptide and SAM were tested across the concentration ranges indicated in Figure 1A, while peptide substrate was fixed at 2 μM. Comparable results were obtained when titrating peptide substrate and activator and fixing SAM at 0.25 μM (Supplementary Figure 1B). The activation constant (Ka) determined from replicate experiments suggests that PRC2 recognizes H3K27me3 peptide with high affinity (86 ± 40 nM, mean ± SD, n = 4). By comparison, the Km value for unmethylated peptide was reported to be 5.4 μM,1 more than 60-fold higher than the Ka determined here for the H3K27me3 peptide with the same amino acid sequence. The data suggest that activation is a kcat effect and not an effect on substrate Km (α = 1.1, β = 7.4, Figure 1B). Activation of PRC2 was also observed when purified HeLa nucleosomes were used as the EZH2 substrate (Supplementary Figure 1C), although to a lesser degree (β = 1.9). HeLa purified nucleosomes would

already possess a mixture of epigenetics marks including H3K4 and H3K36 methylation,9 and these transcriptionally active chromatin marks have been shown to inhibit the histone methylation activity of EZH2/PRC2.10 Therefore we decided to test whether EZH2/PRC2 could be activated using reconstituted nucleosomes as the substrate. Allosteric activation of EZH2/PRC2 in cells could occur inter- or intranucleosomally. Intranucleosomal activation would occur if EED were to bind pre-existing methylation on one of the H3 tails within a nucleosome while EZH2 methylates the other H3 tail within the same nucleosome. Internucleosomal activation could also be possible if EED binds H3K27me3 from one nucleosome and EZH2 methylates H3K27 on a separate nucleosome. To examine whether an inter- or intramolecular mechanism best describes the activation of EZH2, we tested reconstituted nucleosome core particles (NCPs) with no preexisting methylation and NCPs with a mixture of trimethylated and unmethylated H3K27 (Figure 1C and D, Supplementary Table 3, and Supplementary Figure 4). NCPs trimethylated at position 27 of histone H3 were generated using a chemical alkylation reaction that substitutes a methylated analogue of lysine (aminoethylcysteine, Kc) in place of the existing lysine at K27.11 The data show that the unmarked NCPs and heterodimer H3Kc27me3 NCPs have similar catalytic efficiency (kcat/Km = 0.00083 and 0.0013 min−1 nM−1, respectively) and are activated to a similar extent by the H3K27me3 peptide (β = 8.4 and 11, respectively). Alpha values for both NCPs were very close to 2, indicating that the primary B



Letters

Figure 2. Inhibition of unactivated and activated EZH2 by GSK126. (A) Chemical structure of GSK126 compound (1). (B) Inhibitor (I) potency measured ± enzyme (E) preincubation with and without 1 μM H3K37me3 activator (A) peptide. Time-dependent inhibition is only observed using allosterically activated EZH2/PRC2. (C) Diagram showing the activation and inhibition of EZH2/PRC2 by GSK126.

effect of activation is on kcat, not on Km. These data show that activation of EZH2/PRC2 can be observed to a similar extent with either a nucleosome or peptide substrate. The partially premethylated NCPs were generated by mixing a 1:1 ratio of unmarked recombinant histone H3 with histone H3Kc27me3 prior to the addition of nucleosomal DNA. The partially premethylated NCPs would be expected to randomly assemble into a mixture of 25% unmarked, 25% doubly marked and 50% singly marked nucleosomes. In theory this mixture would be capable of either inter- or intranucleosomal activation, however the catalytic efficiency of these NCPs (Figure 1C and D) was similar to that of the unmarked NCPs. In addition, we made fully methylated NCPs with Kc27me3 on both H3 tails to probe intranucleosomal activation, but these NCPs were unable to activate EZH2 (data not shown). Given the lack of activation with these NCPs, we cannot conclusively say whether the activation observed with peptide activator is intra- or internucleosomal with the data in hand. Several possibilities exist that could explain the lack of activation observed using Kc27me3 NCPs. Although others have successfully used methyl-lysine analogues in their studies, it is possible that they may not bind to EED with the same affinity as native trimethylated lysine.6,12,13 Additionally, it is possible that in a nucleosome context specific epigenetic marks are required for proper positioning of the methylated lysine to activate EZH2/ PRC2 and these marks are missing in the reconstituted trimethylated NCPs we tested. Also, the DNA sequence

selected would have an impact on histone octamer positioning within the nucleosome and could also influence the binding of K27me3 to EED. It has been reported that oligonucleosomes with histone H1 are better substrates than mononucleosomes and oligonucleosomes without histone H1.14 It is possible that an oligonucleosome, possibly including histone H1, could be required to provide the proper steric orientation to observe activation by K27me3 in a nucleosomal context. While activation was not observed using methylated NCPs, the selective and potent binding of the trimethylated lysine peptide is consistent with a biologically relevant mechanism for EED in controlling the activation of EZH2 and intracellular H3K27me3 levels. The increase in catalytic turnover induced by activator binding to EED is likely due to a conformational shift in the PRC2 complex. Insights into the nature of this conformational shift have been elusive as the structure of EZH2 has not yet been reported. EZH2 mutations at Tyr641 and Ala677 identified in nonHodgkin lymphomas have been described that result in increased trimethylation at histone H3K27.1,2,15 We tested whether the allosteric activation mechanism was maintained in mutant EZH2 using the H3K27me3 activator peptide. The methyltransferase activity of A677G and Y641F EZH2/PRC2 was measured over a range of activator concentrations while substrate concentrations were held constant. Apparent dissociation constants for the activator peptide (Kaapp) were determined for the mutant enzymes by fitting initial velocity C



Letters

Figure 3. EED occupancy leads to increased residence time inhibition of EZH2/PRC2. (A) 15 nM wild type EZH2/PRC2 or (B) A677G EZH2/ PRC2, inhibitor, and 1 μM activator were preincubated at 10 times the IC50 for 1 h followed by 100-fold dilution into assay buffer containing 1.5 μM SAM and 10 μM peptide substrates, and the recovery of enzyme activity was monitored over time. GSK126’s inhibition of activated EZH2/PRC2 is slowly reversible with off rates of 0.016 and 0.013 min −1, respectively (t1/2 = 44 and 54 min). In contrast with activated PRC2, GSK126’s inhibition of unactivated PRC2 (C) wild type EZH2/PRC2 or (D) A677G/PRC2 is more rapidly reversible with off rates of 0.11 min −1 (t1/2 = 6 min).

with GSK126 and unactivated EZH2/PRC2 were very similar (Figure 2B). To confirm reversibility of inhibition and to determine the off-rate (koff) of GSK126 binding, rapid dilution experiments were conducted where enzyme and inhibitor were preincubated in the presence of 1 μM activator peptide (Figure 3A). The enzyme/inhibitor/activator complex was then diluted 100-fold into assay buffer containing activator peptide, SAM, and peptide substrate. The activator concentration was held constant from the preincubation to the reaction in order to maintain fully active EZH2/PRC2. Reaction progress curve data show that EZH2/PRC2 enzyme activity is slowly recovered over time (t1/2 = 44 min). Parallel experiments were conducted using A677G and Y641F mutant EZH2, and the results are shown in Figure 3B and Supplementary Table 2. In control experiments using unactivated EZH2/PRC2, there was a significantly less pronounced delay in the recovery of activity with estimated t1/2 values on the order of a few minutes (Figure 3C and D). Given the long residence time of inhibition observed, we used progress curve analysis to determine the true potency (Ki*) of GSK126 against activated EZH2/PRC2. Activator and wild type or mutant EZH2/PRC2 were pre-equilibrated and then added to substrates and GSK126 simultaneously. In the absence of GSK126, product formation was linear over time for almost 2 h under the conditions used (Supplementary Figure

data to eq 1, where it was assumed that binding of the activator peptide did not have an influence on substrate Km (α = 1) as was shown for the WT enzyme. Although some differences in Kaapp values were observed across the enzymes, high affinity for the activator peptide was observed in all cases (Supplementary Table 1). β values (representing the effect on kcat) were determined to be 3.1 using A677G and 4.5 using Y641F. These results indicate that mutant EZH2/PRC2, like wild type, is also subject to allosteric activation upon EED binding to trimethylated H3K27. Inhibitor Residence Time Is Increased When EZH2/ PRC2 Is Activated. The inhibitory potency was determined for GSK126, a recently described selective EZH2 inhibitor,3 in the presence and absence of activator peptide following preincubation with EZH2/PRC2 (Figure 2A and B). As the measured potency of GSK126 is very close to the theoretical tight binding limit of the assay (2.5 nM), these experiments were performed in the presence of a saturating concentration of SAM (7.5 μM, 24-fold above Km). A time-dependent decrease in IC50 was evident following inhibitor preincubation with activated EZH2/PRC2. This decrease in IC50 after enzyme plus inhibitor preincubation indicates there is a slow approach to the equilibrium of GSK126 binding to the activated EZH2/PRC2 complex. This slow binding behavior was only evident with the activated form of EZH2/PRC2, as IC50 values ± preincubation D



Letters

Table 1. Structure−Activity Relationships for GSK126 (1) Compared to Related 4-Substituted Pyridones (2−4) and Sinefungin (5)a

compd

R1

IC50 (nM)

aIC50 (nM)

Ki* (nM)

koff (min‑1)

residence time (min)

KARPAS-422 H3K27me3 IC50 (nM)

KARPAS-422 gIC50 (nM)

Pfeiffer gIC50 (nM)

1 2 3 4 5

N/A propyl isopropyl methyl N/A

12 16 46 154 35000

3.3 2.2 14 91 28000

0.37 0.55 3.5 ND ND

0.016 0.023 0.085 ND ND

62 43 15 ND ND

21 170 407 1995 >38000

230 487 2067 6798 >23770

64 132 596 2881 >36657

N/A = not applicable; there are multiple points of divergence between compounds 1 and (2−4). IC50 values were determined at 6.2 μM SAM (20fold over Km). aIC50s are IC50 values determined using the activated form of EZH2/PRC2 following a preincubation with inhibitor. ND = not determined, inhibition was rapidly reversible. Residence time = 1/koff, assuming a two-step model of enzyme inhibition where k4 ≪ k5 and k6, then koff ≈ k6. KARPAS-422 H3K27me3 IC50 values were determined using H3K27me3 and total histone H3 ELISAs following a 3-day exposure. KARPAS-422 and Pfeiffer growth IC50 values were determined after 6 days of compound exposure. a

inhibition whereas 4 and 5 are rapidly reversible (t1/2 < 2 min) (Table 1). Since the addition of only a few carbon atoms significantly increased the residence time of inhibition it seems reasonable that the pyridone ring is structurally important for the slow release of inhibitor from activated EZH2/PRC2. The structure activity relationship between compounds 1 and 2 indicate that chemical features in addition to the pyridone could enable longer residence times. Cellular data from assays that measure levels of H3K27me3 and proliferation correlate with changes in residence time (Table 1). An increase in inhibitor residence time may occur by a combination of factors including enzyme−inhibitor ground state stabilization and transition state destabilization.18 Understanding which of these mechanisms are contributing to increased residence time could inform further medicinal chemistry optimization. We calculated the free-energy changes associated with the ground and transition states for compounds 1 and 2 in comparison to compound 3 using eqs 8 and 9. A comparison of 1 to shorter residence time inhibitor 3 revealed that similar free-energy changes occur as a result of transition state destabilization and ground state stabilization (ΔGGS = 1.3, ΔGTS = 2.3). Similar results were obtained for compound 2 as compared to 3 (ΔGGS = 1.1, ΔGTS = 1.9). These data suggest that both transition state destabilization and ground state stabilization are viable strategies for further inhibitor optimization. Different mechanisms of slow binding inhibition could potentially describe the kinetic data for EZH2 inhibitors, namely conformational selection and induced fit.19,20 Conformational selection occurs when free enzyme slowly isomerizes between conformations E and E* and the inhibitor selectively binds to E* and rapidly forms a binary E*I complex. A conformational selection mechanism is indicated when the observed rate constant (kobs) decreases with increasing inhibitor concentration.20 Slow binding inhibition could also occur by means of an induced-fit mechanism, which can be further characterized as one-step or two-step dissociation processes. A one-step induced-fit mechanism is indicated when kobs increases

2). In the presence of GSK126 progress curves exhibited concentration-dependent curvature consistent with a long residence time of inhibition. The data were fit globally to a two-step model of time-dependent inhibition as described in the Methods section and summarized in Supplementary Table 2. Off-rates determined from the progress curve analysis were consistent with the results observed from rapid dilution experiments. The results show that GSK126 has a Ki* = 0.37 ± 0.07 nM against wild type EZH2, an order of magnitude lower than the Ki = 4.7 ± 0.5 nM determined from the forward progress curve data analysis. Similar results were observed using mutant EZH2 complexes (Supplementary Table 2). The inhibition kinetics displayed by GSK126 suggest that upon activation EZH2/PRC2 undergoes a conformational shift from which the inhibitor is slowly dissociated. Slow binding inhibition of this type has been described in detail before16 but appears to be unique in that allosteric activation is required to observe the effects of time-dependent inhibition. We next evaluated if time-dependent inhibition of activated EZH2/PRC2 was specific to GSK126 or could be observed with other inhibitors of EZH2. As the pyridone moiety is critical for inhibition of EZH2,17 we hypothesized that this group might contribute to the long residence times. IC50 values were determined with and without enzyme preincubation for inhibitors structurally related to GSK126 but containing modifications to the pyridone (Table 1). IC 50 values determined ± activator peptide in the enzyme + inhibitor preincubation were shifted for some, but not all, of the inhibitors tested. Interestingly, IC50 values for inhibitor sinefungin (5), a natural product SAM analogue, did not shift following preincubation with activated EZH2. In contrast, the molecules exemplified in Table 1 show that small modifications at position 4 of the pyridone, i.e., (4) methyl, (3) isopropyl, and (2) n-propyl, lead to time-dependent IC50 shifts of 1.7-, 3.3-, and 7.1-fold, respectively. Progress curve analysis and rapid dilution data were generated for compounds 2−5, using the same methods as described for 1. The results indicate that compounds 2 and 3 possess measurable residence times of E



Letters

linearly as a function of inhibitor concentration.20 A two-step induced-fit mechanism occurs when enzyme and inhibitor form an initial EI encounter complex with moderate affinity. The EI complex undergoes isomerization to form EI*, which is slowly reversible. A two-step induced fit is indicated when kobs increases as a function of inhibitor concentration yielding a hyperbolic curve.20 To distinguish between the different mechanisms, we plotted the observed rate constants (kobs) as a function of inhibitor concentration using activated EZH2/ PRC2 and GSK126 (Supplementary Figure 3). The increase in kobs values with increasing concentrations of GSK126 appeared to be saturable, suggesting that GSK126 inhibits activated EZH2/PRC2 through a two-step induced-fit mechanism. The results presented here describe mechanisms of allosteric activation by a biologically relevant peptide and of timedependent inhibition by a small molecule. Both mechanisms likely involve conformational shifts suggesting that the EZH2 active site is highly dynamic (see Figure 2 diagram). Interestingly, since significantly longer off rates are observed with the activated form of the enzyme relative to the unactivated form, the conformation of the activated form appears more readily able to adopt the second conformational change required to access the higher affinity inhibitor binding pocket. GSK126 and analogues are SAM-competitive inhibitors of EZH2/PRC2, suggesting that the second conformational change required for high affinity binding may occur in the SAM binding pocket. In contrast, activation of EZH2 with H3K27me3 does not alter the SAM Km or the Ki value of SAM-analogue sinefungin. There are two potential explanations for the different behavior of substrate and inhibitor: (1) binding interactions unique to GSK126 are changed, while those important for SAM are not, or (2) even though GSK126 is SAM-competitive, it does not physically occupy the SAM site and is therefore influenced differently by the conformational changes induced upon activation. Regardless of the exact binding site, the amide linked pyridone ring of GSK126 is important for potency and appears to play a role in the time-dependent inhibition observed. Three recently reported EZH2/PRC2 inhibitors contain the identical pyridone ring and amide linker emphasizing the importance of this chemical feature.3−5 Others have reported that SAMcompetitive small molecule inhibitors can induce conformational shifts in the lysine methyltransferase DOT1L.21 Analogously, we hypothesize that GSK126 induces a conformational shift in EZH2/PRC2 that picks up novel interactions with the pyridone ring. A high-resolution protein structure of EZH2 alone or in the PRC2 complex has not been determined to date, and therefore it is unclear what changes in tertiary structure may be concurrent with the long residence time of inhibition and the activation mechanism of EZH2/PRC2. Activating point mutations in EZH2 at Tyr641 and Ala677 identified in non-Hodgkin lymphoma result in enhanced trimethylation at H3K27.1,2,15 Since H3K27me3 allosterically activates EZH2/PRC2, it seems reasonable that more of the complex is activated in mutant cells as compared to wild type. We demonstrate in this paper that GSK126 is a more potent inhibitor of activated EZH2/PRC2 with a longer residence time. The improved affinity and residence time of GSK126 as compared to inhibitor analogues 2−4 for activated EZH2/ PRC2 most likely contributes to improved potency in cells.19 Here we show that activation of EZH2/PRC2 by an H3K27me3 peptide is primarily an effect on kcat with no effect on substrate binding (Km). Additionally, GSK126 is shown to

have a long residence time on the activated form of wild type and mutant EZH2/PRC2. We propose a binding model that explains the enzyme inhibitor kinetic data for GSK126 and may contribute to the high degree of potency and selectivity over other lysine methyltransferases. Further analysis of inhibitor residence times should provide an additional parameter for compound optimization.20,22 The allosteric activation and long residence times of inhibition highlight the conformational flexibility of EZH2/PRC2, and this may have important implications for the ongoing development of EZH2/PRC2 inhibitors.

■

METHODS

EZH2/PRC2 Enzyme Assay. Five-member PRC2 complexes (EED, SUZ12, AEBP2, RbAp48) containing wild-type EZH2 or mutant (A677G, Y641F) were prepared as previously described.1 Activator peptide, H3K27me3 21−44, (ATKAAR[K-Me3]SAPATGGVKKPHRYRPGGK-amide) was obtained from 21st Century Biochemicals, USA. Substrate peptides (also from 21st Century Biochemicals, USA) with K27me0 (wild type EZH2 and A677G) or K27me2 (Y641F EZH2) and HeLa purified nucleosomes were used as methyltransferase substrates in accord with their catalytic efficiencies as described.1 Reconstituted mononucleosomes were prepared as previously described.23 Basically, recombinant histones were purified from E. coli and reconstituted with a 146-bp DNA sequence using a dialysis method to form nucleosome core particles (NCPs). A trimethyl-lysine analogue (Kc27me3) was prepared as previously described11 using a chemical alkylation reaction that substitutes a methylated analogue of lysine, trimethyl-aminoethyl-cysteine, for H3K27 and was obtained from Active Motif USA (cat. no. 31000). Histones H2A, H2B, and H4 and 146mer DNA were prepared at XTAL Biostructures, USA. Individual histones were mixed in the proper molar ratios to produce a histone octamer with either a homodimer or heterodimer of H3Kc27me3. The histones were refolded to form histone octamers and subsequently purified by SEC. DNA was added to the histone octamers at 2 M KCl, and the ionic strength was gradually dialyzed out to reconstitute the NCPs. The integrity of the NCPs was confirmed by a gel electrophoretic mobility shift assay as shown in Supplementary Figure 4. Enzyme activity was measured in 50 mM Tris pH 8, 2 mM MgCl2, 4 mM DTT and initiated by the addition of [3H]-SAM. Reactions were quenched with the addition of 500-fold excess unlabeled SAM, and the [3H]-methylated product was captured on phosphocellulose (peptides) or diethylaminoethanol (nucleosomes) filter membranes according to the vendor-supplied protocol for Multiscreen plates (EMD Millipore, USA). Plates were read on a TopCount after adding 20 μL of Microscint-20 cocktail (both from PerkinElmer). A small volume of the reaction mixture containing a known amount of [3H]SAM was spotted into an empty well to calculate the specific activity of [3H]-SAM, which enabled the conversion of radioactivity (counts per minute) into moles of product captured. Determination of Activation Constants for Nonessential Activation of EZH2/PRC2 by H3K27me3. Double titration experiments were carried out using activator and substrate in 50 mM Tris pH 8, 2 mM MgCl2, 4 mM DTT, and 9 nM EZH2/ PRC2.The resulting data were fit to a kinetic model of nonessential enzyme activation (Figure 1B, eq 1)

v = (VmaxS /K s + βvmax AS /(αK aK s))/(1 + S /K s + A /K a + AS /(αK aK s))

(1)

where vmax is the maximal velocity of the unactivated reaction, and βvmax is maximal velocity in the presence of a saturating concentration of activator peptide. S is the substrate concentration, Ks is the Michaelis constant for substrate, A is the activator peptide concentration, and Ka is the activation constant. The factor α is the factor by which Ks changes when activator peptide is bound to EZH2/ PRC2. For Ka apparent determinations using A677G and Y641F F



Letters

mutants, both substrates were fixed at Km concentrations and α was held constant at α = 1 for the data fitting. Determination of Dissociation Rates of GSK126 from EZH2/ PRC2. Rates of dissociation were determined for GSK126 using wild type and mutant EZH2 by monitoring the recovery of activity following a 1-h preincubation that included 10× the KIapp of inhibitor and 2 nM EZH2/PRC2 plus or minus 10× the Ka (1 μM) for activator peptide. Preincubated samples were diluted 100-fold into assay buffer containing 5× Km SAM (1.5 μM) and 2× Km peptide (10 μM) substrates and quenched at various time points. [3H]-Methylated peptide product was captured using arginine binding SPA beads in 200 mM NH4HCO3 solution and read on a Microbeta (both from PerkinElmer, USA). Progress curve data was fit to eq 2 to generate koff: v − vs y = vst + i (1 − e−kobst ) + background kobs (2)

where koff is the dissociation rate constant for compound x and 3. H3K27me3 Cell Mechanistic and Proliferation Assays. These were performed as previously described.3 Compound Synthesis. Compounds 1, 2 and 4 were prepared as described.3,17 Compound 3 was prepared as described in patent application WO2011140325(A1). Sinefungin was purchased from Sigma.

where vi and vs represent the initial and steady state velocities respectively, and kobs is the apparent first-order rate constant for the transition from vi to vs, and t is time. Under the experimental conditions used, kobs approximates the dissociation rate constant (koff) of the enzyme−inhibitor complex. The enzyme−inhibitor dissociation half-life (t1/2) was calculated using the formula t1/2 = 0.693/koff. And the residence time (τ) was calculated using the equation τ = 1/koff Ki* and On-Rate Determination by Progress Curve Analysis. Wild type (230 pM) and mutant EZH2/PRC2 (150 and 220 pM) were preactivated using 1 μM H3K27me3 peptide and then mixed with various concentrations of inhibitor, 10 μM peptide, and [3H]SAM (0.3 μM) at the apparent Km for substrate. The formation of product over time at each inhibitor concentration was monitored by quenching with 500-fold excess unlabeled SAM at various time points. To generate Ki, Ki*, koff, and kon values, the reaction progress curves at various inhibitor concentrations were fitted globally to eq 2 for timedependent inhibition using GraFit 5. The data can be described by a two-step model of time-dependent inhibition where E represents the activated form of EZH2/PRC2, which undergoes isomerization to EI* when bound to inhibitor, which is then slowly reversible.

Corresponding Author

■

* Supporting Information

Supplementary figures and tables. This material is available free of charge via the Internet at http://pubs.acs.org.

■

(4)

vs = VmaxS /K m(1 + I /K i*) + S

(5)

kobs = k6 + k5I /(K i(1 + S /K m) + I )

(6)

Notes

The authors declare the following competing financial interest(s): Authors of this work were employed by GlaxoSmithKline.

■

ACKNOWLEDGMENTS The authors respectfully acknowledge helpful discussions with A. Graves and invaluable assistance from A. Taylor, Y. Jiang, P. McDevitt, B. Kirkpatrick, and D. Fisher.

■

k6 k5 + k 6

(7)

Calculation of Free-Energy Changes. Changes in the free energy of the enzyme−inhibitor complex ground state (GS) were calculated at 298 K using the following equation:

ΔGGS = − RT ln(K i(x)/K i(3))

(8)

where Ki is the inhibition constant for compound x and 3, and R = 1.986 cal K−1 mol−1. Changes in the free energy of the transition state (TS) were calculated using the following equation:

ΔGTS = − RT ln(K i(x)/K i(3)) − ln((koff(3)/koff(x))

REFERENCES

(1) McCabe, M. T., Graves, A. P., Ganji, G., Diaz, E., Halsey, W. S., Jiang, Y., Smitheman, K. N., Ott, H. M., Pappalardi, M. B., Allen, K. E., Chen, S. B., la Pietra, A., Dul, E., Hughes, A. M., Gilbert, S. A., Thrall, S. H., Tummino, P. J., Kruger, R. G., Brandt, M., Schwartz, B., and Creasy, C. L. (2012) Mutation of A677 in histone methyltransferase EZH2 in human B-cell lymphoma promotes hypertrimethylation of histone H3 on lysine 27 (H3K27). Proc. Natl. Acad. Sci. U.S.A. 109, 2989−2994. (2) Yap, D. B., Chu, J., Berg, T., Schapira, M., Cheng, S.-W. G., Moradian, A., Morin, R. D., Mungall, A. J., Meissner, B., Boyle, M., Marquez, V. E., Marra, M. A., Gascoyne, R. D., Humphries, R. K., Arrowsmith, C. H., Morin, G. B., and Aparicio, S. A. J. R. (2011) Somatic mutations at EZH2 Y641 act dominantly through a mechanism of selectively altered PRC2 catalytic activity, to increase H3K27 trimethylation. Blood 117, 2451−2459. (3) McCabe, M. T., Ott, H. M., Ganji, G., Korenchuk, S., Thompson, C., Van Aller, G. S., Liu, Y., Graves, A. P., III, A. D. P., Diaz, E., LaFrance, L. V., Mellinger, M., Duquenne, C., Tian, X., Kruger, R. G., McHugh, C. F., Brandt, M., Miller, W. H., Dhanak, D., Verma, S. K., Tummino, P. J., and Creasy, C. L. (2012) EZH2 inhibition as a therapeutic strategy for lymphoma with EZH2-activating mutations. Nature 492, 108−112. (4) Qi, W., Chan, H., Teng, L., Li, L., Chuai, S., Zhang, R., Zeng, J., Li, M., Fan, H., Lin, Y., Gu, J., Ardayfio, O., Zhang, J. H., Yan, X., Fang, J., Mi, Y., Zhang, M., Zhou, T., Feng, G., Chen, Z., Li, G., Yang, T., Zhao, K., Liu, X., Yu, Z., Lu, C. X., Atadja, P., and Li, E. (2012) Selective inhibition of Ezh2 by a small molecule inhibitor blocks tumor cells proliferation. Proc. Natl. Acad. Sci. U.S.A. 109, 21360−21365. (5) Knutson, S. K., Wigle, T. J., Warholic, N. M., Sneeringer, C. J., Allain, C. J., Klaus, C. R., Sacks, J. D., Raimondi, A., Majer, C. R., Song, J., Scott, M. P., Jin, L., Smith, J. J., Olhava, E. J., Chesworth, R., Moyer, M. P., Richon, V. M., Copeland, R. A., Keilhack, H., Pollock, R. M., and Kuntz, K. W. (2012) A selective inhibitor of EZH2 blocks H3K27 methylation and kills mutant lymphoma cells. Nat. Chem. Biol. 8, 890− 896. (6) Margueron, R., Justin, N., Ohno, K., Sharpe, M. L., Son, J., Drury, W. J., III, Voigt, P., Martin, S. R., Taylor, W. R., De Marco, V., Pirrotta, V., Reinberg, D., and Gamblin, S. J. (2009) Role of the polycomb

where Vmax is the maximal velocity of the reaction, S is the SAM concentration, Km is the Michaelis constant for SAM, I is the GSK126 concentration, koff is the dissociation rate constant of the enzyme inhibitor complex, and Ki and Ki* represent the initial and final inhibition constants, respectively. The relationship between Ki and Ki* is described by the equation

K i* = K i

AUTHOR INFORMATION

*E-mail: [email protected].

The parameters vs, vi, and kobs are described by the following equations for competitive inhibition:16 vi = VmaxS /K m(1 + I /K i) + S

ASSOCIATED CONTENT

S

(9) G



Letters

protein EED in the propagation of repressive histone marks. Nature 461, 762−767. (7) Claudio, C., Gabriel, C. L., Alessio, M., Franz, H., Ruedi, A., and Eva, N. (2012) Molecular architecture of human polycomb repressive complex 2. eLife 1, No. e00005. (8) Segel, I. H. (1975) Enzyme Kinetics: Behavior and Analysis of Rapid Equilibrium and Steady-State Enzyme Systems, John Wiley & Sons, New York. (9) Young, N. L., DiMaggio, P. A., Plazas-Mayorca, M. D., Baliban, R. C., Floudas, C. A., and Garcia, B. A. (2009) High throughput characterization of combinatorial histone codes. Mol. Cell. Proteomics 8, 2266−2284. (10) Schmitges, F., Prusty, A., Faty, M., Stützer, A., Lingaraju, G., Aiwazian, J., Sack, R., Hess, D., Li, L., Zhou, S., Bunker, R., Wirth, U., Bouwmeester, T., Bauer, A., Ly-Hartig, N., Zhao, K., Chan, H., Gu, J., Gut, H., Fischle, W., Müller, J., and Thomä, N. (2011) Histone methylation by PRC2 is inhibited by active chromatin marks. Mol. Cell 42, 330−341. (11) Simon, M. D., Chu, F., Racki, L. R., de la Cruz, C. C., Burlingame, A. L., Panning, B., Narlikar, G. J., and Shokat, K. M. (2007) The site-specific installation of methyl-lysine analogs into recombinant histones. Cell 128, 1003−1012. (12) Jia, G., Wang, W., Li, H., Mao, Z., Cai, G., Sun, J., Wu, H., Xu, M., Yang, P., Yuan, W., Chen, S., and Zhu, B. (2009) A systematic evaluation of the compatibility of histones containing methyl-lysine analogues with biochemical reactions. Cell Res. 19, 1217−1220. (13) Seeliger, D., Soeroes, S., Klingberg, R., Schwarzer, D., Grubmüller, H., and Fischle, W. (2011) Quantitative assessment of protein interaction with methyl-lysine analogues by hybrid computational and experimental approaches. ACS Chem. Biol. 7, 150−154. (14) Martin, C., Cao, R., and Zhang, Y. (2006) Substrate preferences of the EZH2 histone methyltransferase complex. J. Biol. Chem. 281, 8365−8370. (15) Sneeringer, C. J., Scott, M. P., Kuntz, K. W., Knutson, S. K., Pollock, R. M., Richon, V. M., and Copeland, R. A. (2010) Coordinated activities of wild-type plus mutant EZH2 drive tumorassociated hypertrimethylation of lysine 27 on histone H3 (H3K27) in human B-cell lymphomas. Proc. Natl. Acad. Sci. U.S.A. 107, 20980− 20985. (16) Morrison, J. F. J. (1988) The behavior and significance of slowbinding enzyme inhibitors. Adv. Enzymol. Relat. Areas Mol. Biol. 61, 201−301. (17) Verma, S. K., Tian, X., LaFrance, L. V., Duquenne, C., Suarez, D. P., Newlander, K. A., Romeril, S. P., Burgess, J. L., Grant, S. W., Brackley, J. A., Graves, A. P., Scherzer, D. A., Shu, A., Thompson, C., Ott, H. M., Aller, G. S. V., Machutta, C. A., Diaz, E., Jiang, Y., Johnson, N. W., Knight, S. D., Kruger, R. G., McCabe, M. T., Dhanak, D., Tummino, P. J., Creasy, C. L., and Miller, W. H. (2012) Identification of potent, selective, cell-active inhibitors of the histone lysine methyltransferase EZH2. ACS Med. Chem. Lett. 3, 1091−1096. (18) Lu, H., and Tonge, P. J. (2010) Drug-target residence time: critical information for lead optimization. Curr. Opin. Chem. Biol. 14, 467−474. (19) Copeland, R. A. R. (2011) Conformational adaptation in drugtarget interactions and residence time. Future Med. Chem. 3, 1491− 1501. (20) Tummino, P. J., and Copeland, R. A. (2008) Residence Time of Receptor−Ligand Complexes and Its Effect on Biological Function. Biochemistry 47, 5481−5492. (21) Basavapathruni, A., Jin, L., Daigle, S. R., Majer, C. R. A., Therkelsen, C. A., Wigle, T. J., Kuntz, K. W., Chesworth, R., Pollock, R. M., Scott, M. P., Moyer, M. P., Richon, V. M., Copeland, R. A., and Olhava, E. J. (2012) Conformational adaptation drives potent, selective and durable inhibition of the human protein methyltransferase DOT1L. Chem. Biol. Drug Des. 80, 971−980. (22) Copeland, R. A., Pompliano, D. L., and Meek, T. D. (2006) Drug-target residence time and its implications for lead optimization. Nat. Rev. Drug Discovery 5, 730−739.

(23) Luger, K.; Rechsteiner, T. J.; Richmond, T. J. (1999) Expression and purification of recombinant histones and nucleosome reconstitution, in Chromatin Protocols, pp 1−16, Springer, New York.

H


Long Residence Time Inhibition of EZH2 in Activated Polycomb

Recommend Documents