Development and Implementation of a High-Throughput AlphaLISA Assay for Identifying Inhibitors of EZH2 Methyltransferase

Abstract

The methylation state of lysine residues within histone H3 is a major determinant of active and inactive regions of the genome. Enhancer of Zeste homolog 2 (EZH2) is a histone lysine methyltransferase that is part of the polycomb repressive complex 2 (PRC2). Elevated EZH2 expression levels have been linked to hypertrimethylation of histone H3 lysine 27 (H3K27), repression of tumor repressor genes, and the onset of several types of cancers. We used the AlphaLISA technology to develop a high-throughput assay for identifying small molecule inhibitors of EZH2. AlphaLISA Acceptor Beads coated with antibodies directed against methylated H3K27 provided a sensitive method of detecting EZH2 activity through measurement of K27 methylation of a biotinylated H3-based peptide substrate. Optimized assay conditions resulted in a robust assay (Z′>0.7) which was successfully implemented in a high-throughput screening campaign. Small molecule inhibitors identified by this method may serve as powerful tools to further elucidate the potential importance of EZH2 in the development and treatment of cancer.

Introduction

Our past understanding and efforts to fight cancer have largely focused on neutralizing the effects of disregulated proteins which result from oncogene transcription. Moreover, drug research and development efforts are hampered by the evolution of mutant variants of these proteins, some of which are resistant to current therapeutics.^1,2 More recently, it has become evident that abnormal changes to the epigenome can influence the process of tumorigenesis by modulating oncogene and tumor suppressor gene transcription, thus highlighting a pharmaceutical need to address emerging epigenetic targets.^3
–5 In contrast to mutations, epigenetic modifications have the capacity to be reversibly modulated—for example, by targeting the enzymes that are responsible for controlling the methylation state of histones. Histones serve as the gatekeeper to the genome by either tightly packing or relaxing DNA, thereby controlling the accessibility of genes to transcription factors.

Lysine 27 of histone H3 (H3K27) is methylated by the polycomb repressive complex 2 (PRC2), a multi-subunit protein complex whose catalytic subunit is the methyltransferase enhancer of Zeste homolog 2 (EZH2). Other members of PRC2 include the extra sex combs protein (EED), suppressor of Zeste 12 (SUZ12), RbAp48, and AEBP2 [Drosophia homologs are E(z), Esc, Su(z)12, and NURF55/CAF-1].^6

–11 EZH2 catalyzes the transfer of a methyl group from S-adenosyl-L-methionine (SAM) to the terminal nitrogen atom of a H3K27 side chain, releasing S-adenosyl-L-homocysteine (SAH) as a reaction by-product. Hypertrimethylation of H3K27 (H3K27me3) due to elevated EZH2 expression levels has been associated with a concomitant decrease of tumor repressor gene expression and with several types of cancer, including prostate, breast, and colon.^12

–16 EZH2 also appears to work in concert with a mutated form of the enzyme to more efficiently catalyze H3K27me3 production in follicular lymphoma as well as some subtypes of diffuse large B-cell lymphoma, the most common form of non-Hodgkin lymphoma.^17,18 Moreover, structural studies suggest that these elevated H3K27me3 levels can in turn enhance EZH2 activity, thereby potentially leading to an auto-magnification of aberrant epigenetic modifications.¹⁹ Collectively, these discoveries have garnered interest in the development of small molecule inhibitors of EZH2 as novel anti-cancer therapeutics.^20,21

The discovery of small molecule EZH2 inhibitors should be facilitated and hastened by high-throughput screening (HTS). Assays performed in an HTS setting need to be sensitive, robust, cost-effective, and scalable. Currently available radiometric assays for EZH2^18,22 meet these standards, but are accompanied by additional handling requirements as well as safety precautions associated with the use and disposal of large volumes of radioactive waste. Alternatively, nonradioactive homogenous assays have been reported, but these assays rely heavily on coupled enzymatic reactions, tend to be less robust, and require counterscreens with additional assays to ensure that identified hit compounds are acting directly on the target.^23,24 For the development of an EZH2 activity assay that better meets ideal HTS criteria, we explored the use of AlphaLISA™ technology to detect the methylation of a biotinylated, unmethylated 24-mer H3-based peptide (H3K27me0) substrate. This detection technology is widely used due to its adaptability to a variety of targets and substrates, and has been successfully applied to develop high-throughput assays for histone methyltransferases (SET7/9, G9a),^25,26 histone demethylases (Jumonji JMJD3, LSD1),^25,27 and other binding proteins which recognize methylated histones.²⁸ In the case of EZH2, AlphaLISA donor beads coated with streptavidin bind to the biotin tag of the peptide, while H3K27 methylation is bound by AlphaLISA acceptor beads coated with antibodies which recognize and bind to H3K27me1 and H3K27me2. A gain of signal indicates methylation of the peptide substrate at K27 by EZH2. Here we describe the principle, development, and performance of our AlphaLISA EZH2 assay in an HTS setting.

Materials and Methods

Reagents, Recombinant Proteins, and Peptides

Fluorescein-labeled SAH “tracer” and histone methyltransferase (HMT) Buffer 2 (with and without added detergent) were obtained from BPS Biosciences (San Diego, CA). Recombinant PRC2 (>90% purity) was prepared by co-expression of N-terminal His-tagged human EZH2 (residues 2–end), N-terminal His-tagged human EED (amino acid residues 2–end), N-terminal His-tagged human RbAp48 (residues 2–end), N-terminal His-tagged human AEBP2 (amino acid residues 2–end), and N-terminal Flag-tagged human SUZ12 (residues 2–end) in a baculovirus expression system and obtained from BPS Biosciences. Anti-SAH monoclonal antibody was purchased from Axis Shield (Dundee, United Kingdom). Bicine (1 M solution), SAM, SAH, and poly-L-Lysine (0.1% solution) were purchased from Sigma-Aldrich (St. Louis, MO). Biotinylated peptide substrates used for assay development were purchased from AnaSpec (Freemont, CA) and consisted of residues 21–44 of histone H3 and contained various methylation states at the K27 position (H3K27me0, H3K27me1, H3K27me2, H3K27me3) of the K27 marker. The sequences of the peptides were: unmethylated H3K27me0, AT KAA RKS APA TGG VKK PHR YRP GGK-biotin-OH; mono-methylated H3K27me1, AT KAA RK(me1)S APA TGG VKK PHR YRP GGK-biotin-OH; di-methylated H3K27me2, AT KAA RK(me2)S APA TGG VKK PHR YRP GGK-biotin-OH; tri-methylated H3K27me3, AT KAA RK(me3)S APA TGG VKK PHR YRP GGK-biotin-OH. An analogous nonbiotinylated mutant H3K27M peptide (AT KAA RMS APA TGG VKK PHR YRP G-OH) and a shorter nonbiotinylated H3K27me3 peptide consisting of residues 23–34 of histone H3 [KAA RK(me3)S APA TGG-OH] were also purchased from AnaSpec as potential inhibitors or potentiators, respectively, of PRC2 activity. Streptavidin-coated Donor Beads, anti-H3K27me2/1 AlphaLISA^® Acceptor Beads, and Epigenetics Buffer #1 kit were purchased from PerkinElmer (Shelton, CT).

AlphaLISA H3K27me2/1 Detection Assay

Detection bead mix was prepared immediately before use by diluting beads in 1× Epigenetics Buffer, prepared according to the manufacturer's instructions, and keeping light exposure to a minimum. The sensitivity of the AlphaLISA detection assay was initially investigated by testing a range of concentrations (2–20 μg/mL) of streptavidin-coated Donor Beads and anti-H3K27me2/1 AlphaLISA Acceptor Beads, always mixed in a 1:1 ratio, in an attempt to optimize signal–background ratio (S/B) and selectivity for the various methylated peptides. Experiments were carried out by adding 5 μL of 2× detection bead mix to white 384-well low-volume ProxiPlates containing 5 μL of biotinylated H3K27me0, H3K27me1, H3K27me2, or H3K27me3 peptides serially diluted in optimized EZH2 reaction buffer. Assay plates were then covered with Costar (Corning Inc., Corning, NY) adhesive aluminum seals and incubated for 1 h. Plates were measured on an EnVision instrument (PerkinElmer) using the AlphaScreen^® setting adapted for optimal signal detection with a 615 nm filter, after sample excitation at 680 nm. The emission signal at 615 nm was used to quantify lysine methylation.

For the HTS and potency determination of inhibitors, an optimized AlphaLISA H3K27me1 detection assay was used in which 14 μL of 1.4× detection bead mix was added to white 384-well low-volume ProxiPlates containing 6 μL of EZH2 enzyme reaction. The final concentration of Donor and Acceptor beads was optimized to 4 μg/mL per well. Plates were then sealed, incubated for 1 h at room temperature (RT), and measured as described above. It should be noted that a two-step method of AlphaLISA bead addition may also be used, which can result in modest S/B improvement. In this case, the Acceptor beads are first incubated with completed enzyme reaction for 1 h, followed by an additional 1-h incubation with Donor beads. The final bead concentration is the same for both methods.

EZH2 Enzymatic Activity Assays

Enzyme reactions during assay development were carried out in 5-μL volumes in white 384-well low-volume white assay plates. Using HMT Buffer 2, the effects of various buffer additives were tested using a standard set of assay conditions in all experiments, including 7.5 nM PRC2, 200 nM H3K27me0 peptide, and 700–750 nM SAM and incubation times of 120–180 min. The allosteric modulation of EZH2 as a method of improving S/B was investigated by running enzyme reactions in the presence of 0–10 μM nonbiotinylated H3K27me3 peptide. Similarly, experiments to assess the effects of detergent (0%–0.1% Triton X-100 or Tween 20) were performed in the presence of 0.005% BSA before screening compound libraries in high throughput. Experiments to evaluate different types of buffers (pH 8.0) were conducted applying the same standard assay conditions to 50 mM 3-(N-morpholino)propanesulfonic acid (MOPS), Tris, 2-(N-morpholino)ethanesulfonic acid (MES), Bicine, or Hepes buffers (±0.01% Triton X-100), and comparing results with those obtained in the standard HMT Buffer 2 (±0.01% Triton X-100). Experiments to test pH-dependence of EZH2 activity were subsequently conducted with 50 mM Bicine at pH 6.8, 7.2, 7.6, 8.0, 8.4, and 8.8.

All subsequent optimization experiments and biochemical characterization of EZH2 were carried out in an optimized EZH2 reaction buffer (50 mM Bicine, pH 7.6, 0.005% BSA, 0.002% Triton X-100, and 1 mM DTT). The apparent Michaelis constant (K_m ) values for SAM and peptide were determined by conducting time courses using 7.5 nM PRC2, different concentration ranges of the substrate of interest, and fixed saturating concentrations of the other substrate (2 μM H3K27me0 peptide for SAM K_m and 10 μM SAM for peptide K_m ). Reactions were stopped at indicated incubation times with 1× Epigenetics Buffer and diluted to ensure that the top final peptide concentration corresponded to a signal that was within the linear limits for detecting the methylated peptide product.

Peptide K_m was determined using an orthogonal fluorescence polarization (FP) assay that monitors the generation of SAH in the enzyme reaction. Time courses were performed using 25 nM PRC2, 10 μM SAM, and a range of H3K27me0 peptide concentrations in black low-volume 384-well plates. Enzyme reactions were run in 6-μL volumes and were stopped at indicated incubation times by dilution in an FP detection buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 0.05% BSA, 0.1% Tween 20, and 3 mM EDTA).

FP SAH Detection Assay

To 6 μL of enzyme reaction, 3 μL of 4× anti-SAH mAb in FP buffer was added, and the assay plate was incubated for 1 h at RT. Next, 4× fluorescein-labeled SAH in the FP buffer was added, and the assay plate was incubated for 30 min at RT. In the final detection reaction, anti-SAH mAb was diluted 1:500, and the concentration of fluorescein-labeled SAH was 5 nM. Assay plates were measured with a Tecan Safire II (Morrisville, NC) using the FP mode and measuring emission at 520 nm, following sample excitation at 470 nm. By reporting on the conversion of SAM to SAH by methyltransferases, the measured ΔmP could be used as an indirect method of quantifying lysine methylation.

High-Throughput Screening

High-throughput screens for the identification of inhibitors were performed by combining 2 μL of 3× concentrated compound, 2 μL of 3× concentrated PRC2, and 2 μL of 3× concentrated mixture of peptide and SAM, each prepared in optimized EZH2 reaction buffer. Test compounds were initially dissolved and serially diluted in 100% DMSO before predilution to 3× in buffer. Reactions for assessing compound potency against EZH2 contained final concentrations of 3 nM PRC2, 100 nM H3K27me0, 500 nM SAM, and 1% (v/v) DMSO. Enzyme reactions were incubated at RT for 180 min before stopping and measuring the amount of lysine methylation with the AlphaLISA anti-H3K27me2/1 detection assay. The complete assay protocol is shown in Table 1.

Table 1.

High-Throughput Screening Protocol Table for the EZH2 AlphaLISA Assay

Step	Parameter	Value	Description
1	Dispense compounds	2 μL	3× compound [3% (v/v) DMSO]
2	Dispense enzyme	2 μL	3× PRC2 Complex
3	Dispense substrate & co-factor	2 μL	3× mix of SAM/peptide
4	Incubation time	3 h	RT, sealed
5	AlphaLISA detection beads	14 μL	1.4× mix of Donor & Acceptor Beads
6	Incubation time	1 h	RT, sealed
7	Assay readout	615 nm	EnVision plate reader, Alpha Screen mode

Step Notes

1. Final concentrations are 10 μM for 1-pt., 50 μM to 0.024 nM for 22-pt. dilution series, columns 23–24 vehicle only, prepared in enzyme reaction buffer.

2. Final concentration is 1 μg/mL (3 nM), column 24 no enzyme, prepared in enzyme reaction buffer.

3. Final concentrations are 500 nM SAM/100 nM H3K27me0 peptide, prepared in enzyme reaction buffer.

4. Plates sealed with adhesive aluminum foils or placed into humidity chamber.

5. Final concentration is 4 μg/mL of each bead, prepared in the dark using 1× Epigenetics Buffer.

6. Keep dark, and do not expose to light before measurement.

7. Alpha Screen emission filter (570/100 nm), 180 ms excitation time, 550 ms total measurement time.

EZH2, enhancer of Zeste homolog 2; H3K27, lysine 27 of histone H3; PRC2, polycomb repressive complex 2; SAM, S-adenosyl-L-methionine.

In each compound assay plate, two columns of wells (16 wells per column) were reserved for assay controls. One column was used for high controls and consisted of enzyme reactions that were exposed to vehicle only [1% (v/v) DMSO]. The other column was used for low controls containing only vehicle and peptide/SAM mix, but no enzyme. Primary screens were first performed at a single concentration (10 μM) of compound. Hit confirmation was then performed in dose–response format (starting from 50 μM). Mechanism of action was investigated using SAM competition experiments in which the dose–response format was applied to enzyme reactions containing SAM at 1× appK_m and at 20× appK_m . To ensure that the linearity of the enzyme reaction was maintained when using high SAM concentrations, incubation times did not exceed 120 min before the addition of AlphaLISA anti-H3K27me2/1 detection reagents.

AlphaLISA Interference Assay

A bulk reaction containing 3 nM PRC2 Complex, 100 nM H3K27me0, 500 nM SAM, and 1% (v/v) DMSO was incubated at RT for 180 min. Next, 6 μL per well of bulk reaction was dispensed into assay plates containing a prediluted compound. The detection bead mix was then quickly added to a final volume of 20 μL per well and a final concentration of 4 μg/mL per well as described previously. The complete protocol for the AlphaLISA interference assay is shown in Table 2.

Table 2.

High-Throughput Screening Protocol Table for the AlphaLISA Artifact Assay

Step	Parameter	Value	Description
1	Dispense enzyme	2 μL	2× PRC2 Complex
2	Dispense substrate and cofactor	2 μL	2× mix of SAM/peptide [2% (v/v) DMSO]
3	Incubation Time	3 h	RT, sealed
4	Dispense compounds	2 μL	3× compound [1% (v/v) DMSO]
5	AlphaLISA detection beads	14 μL	1.4× mix of Donor & Acceptor Beads
6	Incubation time	1 h	RT, sealed
7	Assay readout	615 nm	EnVision plate reader, Alpha Screen mode

Step Notes

1. Final concentration is 1 μg/mL (3 nM), prepared in enzyme reaction buffer.

2. Final concentrations are 500 nM SAM/100 nM H3K27me0 peptide, prepared in enzyme reaction buffer.

3. Plates sealed with adhesive aluminum foils or placed into humidity chamber.

4. Final concentrations are 10 μM for 1-pt., columns 23–24 vehicle only, prepared in enzyme reaction buffer.

5. Final concentration is 4 μg/mL of each bead, prepared in the dark using 1× Epigenetics Buffer.

6. Keep dark and do not expose to light before measurement

RT, room temperature.

Data Analysis

The amount of signal generated in the presence of compounds versus that in the presence of DMSO vehicle alone (high control) was calculated using the formula: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*}{\rm POC} = {\hbox{compound - average low} \over\hbox{average high- average low}} \times 100 \% \end{align*} \end{document}

where POC is percent of control. The raw values from the high and low controls were used to calculate plate S/B and Z′. Single-compound concentration POC, linear regression analysis plots, goodness of fit (R ²), S/B, and Z′ values were determined with Screener (Genedata AG, Basel, Switzerland) data analysis software. Linear regression analysis plots, goodness of fit (R ²), and correlation (Pearson r) of IC₅₀ values were performed with Prism^® version 5 (GraphPad Software, San Diego, CA). For IC₅₀ determinations, data was fit to a 4-parameter equation ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes {10} {9} {7} {6} \begin{document}$$y = A + \frac { B - A } { 1 + ( x / C ) ^D } $$ \end{document} , where A is the minimum y [POC] value, B is the maximum y [POC], C is the x (compound concentration) at the point of inflection, and D is the slope factor) using Prism. For K_m determinations, the slope (which is equal to the initial velocity [v ₀]) was calculated from each progress curve, and nonlinear regression analysis using a rectangular hyperbola model was performed on plots of substrate concentration versus v ₀. Apparent K_m values were interpolated from this curve as the substrate or cofactor concentration that corresponded to one-half the maximum velocity determined in the presence of saturating amounts of peptide. All analyses for K_m determinations were also performed using Prism.

Results and Discussion

Over the past decade, there has been accumulating evidence linking abnormal methylation of lysine residues on the amino-terminal tail of histone H3 to tumorigenesis, highlighting an unmet potential for the discovery and development of novel anticancer therapeutics against a wide range of epigenetic targets.^3
–5 EZH2 is a histone lysine methyltransferase that serves as the catalytic subunit of the PRC2 and has been shown to hypertrimethylate H3K27, an event that has been shown to silence tumor suppressor genes and has been associated with several types of cancer.^12

–18 The discovery of EZH2 inhibitors would be facilitated by assays that enable the HTS of large compound libraries. For our high-throughput screen, we considered a variety of methods for the measurement of methyltransferase activity. In addition to the use of traditional radiometric assays to study methyltransferase activity,^18,22 a number of coupled enzymatic assays have been described in which the assay readout is dependent on the activity of secondary enzymes that generate other detectable analytes.^23,24 Although these assays are homogenous and do not use radioactivity, they require the development of orthogonal assays to ensure that identified hit compounds are only acting on the desired enzymatic target. Another type of methyltransferase assay involves the use of thiol-reactive fluorescent reagents for detecting homocysteine. However, the homocysteine signal may be complicated by the nonspecific labeling of free thiols on the enzymes involved in the coupled enzymatic reaction.^29,30 To circumvent these issues, we developed a homogenous EZH2 assay that is HTS-amenable and uses the AlphaLISA technology to detect the methylation of a peptide derived from H3 (Fig. 1A). This format is nonradiometric and does not require a coupled enzyme reaction. Signal is generated when streptavidin-coated AlphaLISA Donor Beads bind to the biotin tag at the C-terminal end of the unmethylated peptide substrate, while AlphaLISA Acceptor Beads coated with antibodies against H3K27me2/1 recognize and bind to monomethylated H3K27 (H3K27me1) on the same peptide. These interactions bring both beads into close proximity, allowing the efficient transfer of ¹O₂ from the Donor Bead to the Acceptor Bead needed to trigger a detectable emission signal. Since the antibodies on the Acceptor Beads do not recognize or bind to H3K27me0, a gain of signal in this assay indicates methylation of the H3K27 peptide by EZH2.

Fig. 1.

EZH2 AlphaLISA assay overview. (A) Schematic diagram of the AlphaLISA detection assay for biotinylated H3K27 peptides. EZH2 methyltransferase transfers methyl groups from SAM to lysine side chains on the peptide, releasing SAH as a byproduct of the reaction. Methylation of the peptide is measured by adding AlphaLISA Acceptor Beads coated with anti-H3K27me2/1 antibodies (left; white bead). Streptavidin-coated Donor Beads (right; gray bead) bind strongly to the biotin tag (triangle) of the peptide, bringing Donor and Acceptor Beads into close proximity. Excitation at 680 nm triggers the release and transfer of singlet oxygen (¹O₂) to Acceptor Beads, which produce measureable fluorescence emission at 615 nm. (B) Standard curves for assessing peptide selectivity and assay window. In this example, biotinylated peptides of H3K27me0 (○), H3K27me1 (□), H3K27me2 (▿), and H3K27me3 (▵) were 2-fold serially diluted in an optimized EZH2 reaction buffer and incubated for 1 h with 20 μg/mL AlphaLISA Donor and Acceptor Beads (top) and 4 μg/mL (bottom) mixed at a 1:1 ratio. S/B represents the AlphaLISA signal observed in the absence of peptide as background. Error bars represent±standard error of the mean of triplicate determinations. EZH2, enhancer of Zeste homolog 2; H3K27, lysine 27 of histone H3; SAM, S-adenosyl-L-methionine; SAH, S-adenosyl-L-homocysteine; S/B, signal–background ratio.

AlphaLISA Detection Assay

Before exploring conditions for the enzymatic activity of EZH2, we first optimized the AlphaLISA signal detection to enable the reliable and robust detection of methylated peptide. We started by examining AlphaLISA signal sensitivity and the specificity of the detection reagents to multiple methylation states of the H3K27 peptide (Fig. 1B). When using a mixture of AlphaLISA Donor and Acceptor Beads to compare the detection of several biotinylated H3K27 peptides, a strong increase in the AlphaLISA signal was detected in the presence of both H3K27me1 and H3K27me2. Maximum signal was reached at a concentration of 100–150 nM peptide, with H3K27me2 producing ∼2.5-fold higher S/B compared to H3K27me1. Under the same conditions, neither H3K27me3 peptide nor H3K27me0 was detected above background. Interestingly, we found that a 5-fold reduction in the manufacturer's suggested concentration of each bead improved S/B for H3K27me1 and increased specificity of detection for H3K27me2. Under these conditions, we achieved a 4-fold higher S/B for H3K27me2 relative to H3K27me1 (Fig. 1B). More importantly, reduction of the bead concentration to 4 μg/mL increased the sensitivity of peptide detection such that maximum S/B was reached with only 60 nM peptide. Due to the saturation of signal with lower peptide concentration, appropriate dilution of the enzyme reaction was required to ensure that the peptide remained within the linear range of detection.

EZH2 Reaction Buffer

Using the optimized parameters for AlphaLISA detection, several experiments were then performed to establish favorable EZH2 reaction buffer conditions. Initially, a commercial buffer (HMT Buffer 2; contains 0.01% Triton X-100) was used for the enzymatic reaction, but different buffer components and conditions were explored in an attempt to improve the methyltransferase activity of EZH2. We also explored the inclusion of an H3K27me3 peptide, which has been reported to enhance the activity of EZH2.³¹ Published data suggested that there may be an allosteric mechanism of EZH2 activation in the presence of high levels of H3K27me3. Preferential interaction of EED, another subunit of PRC2, with nucleosomes containing high levels of H3K27me3 has been reported.¹⁹ EED contains several repeating WD40 domains that are abundant in proteins involved in scaffolding functions, including protein–protein and protein–DNA interactions.³² Protein X-ray crystallography confirmed the binding of the H3K27me3 peptide to EED and revealed that the binding was stabilized with an aromatic cage on a side of the EED protein that is opposite of that which interacts with EZH2 within PRC2.^22,31 We therefore investigated the addition of a nonbiotinylated H3K27me3 peptide as a useful method of enhancing EZH2 activity and S/B. We found that adding 5–10 μM of H3K27me3 peptide boosted the S/B of the assay ∼250% (Fig. 2). We also observed a modest 2-fold increase in V _max and a 2-fold stronger affinity (K_m ) for SAM when enzyme reactions were performed in the presence of 10 μM H3K27me3 peptide (data not shown). Both of these effects reflect an increase in EZH2 activity in the presence of the H3K27me3 peptide and suggest that the observed 2.5-fold increase in S/B is due to increased turnover of H3K27me0 to H3K27me1 under these conditions. It is interesting to note that previous studies found the H3K27me3 peptide to cause a marked increase in V _max with no effect on SAM K_m .¹⁹ However, these alternative assays were performed using much more complex substrates (i.e., oligonucleomes).

Fig. 2.

Effect of potentiating peptide on EZH2 activity. Representative experiment to assess the allosteric modulation of EZH2 activity by nonbiotinylated H3K27me3 peptide. Enzyme reactions were performed using 7.5 nM PRC2 Complex, 200 nM biotinylated H3K27me0 peptide, and 750 nM SAM in HMT Buffer 2, incubating for 2 h at RT. Reactions were carried out in the presence of 0–10 μM H3K27me3 peptide. S/B was determined by using signal observed in the absence of enzyme as background. For samples containing H3K27me3, data are plotted as a percentage increase in S/B when compared to S/B measured in the absence of potentiating peptide. Error bars represent standard error of the mean of duplicate determinations. HMT, histone methyltransferase; PRC2, polycomb repressive complex 2; RT, room temperature.

Although the use of H3K27me3 peptide in our EZH2 assay might be an attractive approach for identifying allosteric inhibitors of EZH2 activity, it could be accompanied by complicated analysis, and interpretation of primary HTS data aimed at identifying inhibitors that bind directly to EZH2. For example, compounds that neutralize only the effects of boosting peptide would look like partial inhibitors. Therefore, we chose to focus instead on further optimizing EZH2 activity by examining other components and characteristics of the reaction buffer, such as detergent and pH.

It is well known that detergent can dramatically affect enzyme activity as well as compound potency (IC₅₀) or affinity (K_d ) in biochemical assays.^33
–35 We added 0.005% BSA to detergent-free HMT Buffer 2 and examined the effect of Triton X-100 concentration (0%–0.1%) on EZH2 activity (Fig. 3A). We found that increasing detergent concentration had a parabolic effect on EZH2 activity with a maximum S/B of ∼95 occurring in the presence of 0.002% Triton X-100. The effect of Tween 20 was also examined, and a maximum S/B of ∼75 was achieved in the presence of 0.0008% Tween 20 (data not shown). Interestingly, no assay window was obtained in detergent-free HMT Buffer 2 (Fig. 3B). This suggests that the activity of EZH2 is enhanced in the presence of an optimal concentration of detergent.

Fig. 3.

Effect of detergent, buffer type, and pH on EZH2 activity. (A) Representative assay matrix experiment to determine the effect Triton X-100 on EZH2 activity. Enzyme reactions were performed using 7.5 nM PRC2 Complex, 200 nM biotinylated H3K27me0 peptide, and 700 nM SAM in HMT Buffer 2 supplemented with 0.005% BSA and incubating for 3 h at RT. Triton X-100 concentration ranged 0%–0.1%. S/B was determined using background signal obtained for each condition in the absence of enzyme. Error bars represent±standard error of the mean of triplicate determinations. (B) Effect of buffer type on EZH2 activity and S/B. Enzyme reactions were performed using 7.5 nM PRC2 Complex, 200 nM biotinylated H3K27me0 peptide, and 700 nM SAM and incubating for 3 h at RT. Several buffer types supplemented with or without 0.01% Triton X-100 were examined. S/B for each buffer condition was determined using background signal from terminated reactions (time=0 h) prepared in the same buffer. Error bars represent±standard error of triplicate determinations. (C) Effect of pH on EZH2 activity and S/B. Enzyme reactions were performed using 7.5 nM PRC2 Complex, 200 nM biotinylated H3K27me0 peptide, and 700 nM SAM and incubating for 3 h at RT. About 50 mM Bicine buffer (pH 6.8–8.8) supplemented with 0.01% Triton X-100 was used to investigate the pH-dependence of EZH2 activity. S/B for each buffer condition was determined using background signal from terminated reactions (time=0 h) prepared in the same buffer. Error bars represent standard error of the mean of triplicates. MES, 2-(N-morpholino)ethanesulfonic acid; MOPS, 3-(N-morpholino)propanesulfonic acid.

Next, a panel of different buffers, including Tris, Bicine, Hepes, MES, and MOPS, was evaluated at pH 8.0 to see if improved activity could be achieved compared to the commercial HMT Buffer 2 for EZH2 (Fig. 3B). Since we already found that EZH2 is not active in detergent-free HMT Buffer 2, each type of buffer was tested with and without a fixed amount (0.01%) of Triton X-100. The standard HMT Buffer 2 (±0.01% Triton X-100) was also included as a reference. After a 3-h reaction, none of the buffers resulted in an S/B higher than 2 in the absence of detergent. However, the addition of 0.01% Triton X-100 to HMT Buffer 2 restored enzyme activity and resulted in an S/B of ∼20. This effect was even more pronounced when Triton X-100 was added to Bicine buffer, which resulted in an S/B of ∼80. All other buffers tested performed worse than the standard HMT Buffer 2. We thus used the Bicine buffer with 0.01% Triton X-100 to examine the pH-dependence of EZH2 (Fig. 3C). We observed that enzyme activity and S/B were optimal at pH 7.6 with an S/B of ∼200 while appreciable enzyme activity remained at pH 8.0 with a S/B of ∼100. There was a sharp loss of S/B below pH 7.6 and above pH 8.0. Finally, we reassessed the effects of Triton X-100 on assay performance using the Bicine buffer at pH 7.6 and found that the optimal conditions were comparable to those originally observed with HMT Buffer 2. Thus, the final key elements of the optimized enzyme reaction buffer included Bicine, pH 7.6, and 0.002% Triton X-100. This optimized EZH2 buffer was used in all subsequent studies to biochemically characterize EZH2 and for the detection and potency determination of inhibitors by HTS.

EZH2 Enzyme Reaction

To determine conditions for the detection of SAM and peptide-competitive inhibitors, substrate affinity and reaction linearity experiments were performed. SAM K_m was determined in the presence and absence of 1% (v/v) DMSO, since our test compounds are solubilized in DMSO, and our compound-screening concentration of 10 μM would correspond to 1% (v/v) DMSO in the EZH2 enzyme reaction (Fig. 4A). In the presence of DMSO, we observed a noticeable lag phase at the start of these progress curves, which lasted for ∼40 min. Interestingly, this lag phase was absent when the same experiment was carried out in the presence of 10 μM H3K27me3 peptide (data not shown). Starting at 50 min, signal increased linearly for up to 3 h, and the initial velocities (v ₀) obtained from a linear regression fit of these progress curves allowed us to determine an apparent SAM K_m of 560 nM (Fig. 4B). Interestingly, when using a concentration of SAM similar to that used in the pH-dependence study, there was approximately a 50% reduction in S/B after the same 3-h incubation time in this experiment. We suspect that this loss of S/B is due to the addition of both DMSO and BSA to the reaction buffer, both of which were absent in the pH-dependence study. Although the reason for this initial lag phase is unclear, we can only postulate that the level of H3K27me1 generated under these conditions is under the detection limit of the assay during this initial time period. It is interesting to note that this lag phase was not observed when the reaction progress was monitored by more-sensitive mass spectroscopic methods (data not shown), indicating that it was specific to the AlphaLISA assay.

Fig. 4.

SAM K_m and linearity of the optimized EZH2 enzyme reaction using the AlphaLISA assay. (A) Representative biochemical characterization of EZH2 and determination of the apparent SAM K_m . Experiment was carried out using the optimized EZH2 reaction buffer [supplemented with 1% (v/v) DMSO to mimic high-throughput screening conditions] with 7.5 nM PRC2, 2 μM H3K27me0 peptide, and a range of SAM concentrations (0–10 μM). Enzyme reactions were diluted 40-fold in the detection assay to ensure that the final peptide concentration remained within the linear detection range. Progress curves were plotted, and linear regression analysis was performed for each concentration of SAM to determine initial velocities (v ₀) of each reaction condition. S/B for each condition was determined using background signal from terminated reactions (time=0 h). Points represent mean values of duplicates, and error bars represent±standard error of duplicates. (B) Apparent K_m values were determined by performing nonlinear regression on the derived v ₀ values versus concentration of SAM. (C) Representative enzyme titration and timecourse for the EZH2 AlphaLISA assay. Optimized EZH2 assay buffer containing 500 nM SAM, 100 nM H3K27me0 peptide, and increasing amounts of EZH2 (0–15 nM) were stopped intermittently over a period of 180 min. S/B for each condition was determined using background signal from terminated reactions (time=0 h). Linear regression analysis was performed on progress curves for each enzyme concentration. Points represent mean values of duplicates, and error bars represent standard error of the mean.

Next, we performed linear regression analysis on curves generated using different concentrations of PRC2 (containing the EZH2 catalytic subunit; Fig. 4C). For these experiments, we used 500 nM SAM (at apparent K_m ) together with 100 nM H3K27me0 peptide (below its published K_m )¹⁸ to determine the lower limit of enzyme needed to maintain an S/B that is suitable for HTS. We found that the PRC2 concentration in the enzyme reaction could be reduced to 3 nM (i.e., 0.7 nM EZH2) and still yield S/B of ∼25 under these assay conditions. For the determination of H3K27me0 peptide K_m , enzyme reaction progress curves were determined using a range of peptide concentrations in the presence of saturating amounts of SAM. Samples were subsequently diluted to ensure that the peptide concentration was within the previously determined linear range for detection. However, using the v ₀ values obtained from progress curves obtained at increasing peptide concentrations, we were unable to fit the resulting plot using nonlinear regression to determine the peptide K_m (Fig. 5A). As peptide concentration increased, the initial velocities increased as expected, but then decreased again at the highest concentrations of peptide used in the enzyme reaction. This pattern is consistent with the documented AlphaLISA hook effect caused by unoptimized ratios of detection reagents relative to analyte.³⁶ However, when the samples were serially diluted to ensure that analyte concentrations were within the linear range of detection, the pattern persisted. Since peptides can modulate the activity of EZH2 (Fig. 2) by binding elsewhere on the PRC2,²² we next investigated whether the H3K27me0 peptide may have inhibited EZH2 activity at concentrations >200 nM by an unknown mechanism. We employed an orthogonal FP assay that measures the concentration of SAH generated in the enzyme reaction. Using this assay, we obtained a typical saturation curve when plotting v ₀ values and successfully determined a peptide K_m of 290 nM for H3K27me0 (Fig. 5B). A peptide K_m of 300 nM was also validated by mass spectroscopic methods (data not shown). Thus, the AlphaLISA format was not suitable for peptide K_m determinations, and we relied instead on the K_m value measured with the orthogonal assays. Furthermore, the v ₀ values obtained from the orthogonal formats rule out inhibition of catalytic activity in the presence of high peptide concentrations.

Fig. 5.

Determination of peptide K_m using AlphaLISA and FP assays. (A) Representative determination of the apparent peptide K_m of EZH2. For the AlphaLISA detection assay, all enzyme reactions were performed in duplicate using 5 nM PRC2, 10 μM SAM, and a range of H3K27me0 peptide concentrations (0–2 μM) and were stopped intermittently over a period of 150 min. Enzyme reactions were diluted 16-fold in the detection assay to ensure that the final peptide concentration remained the linear detection range. S/B for each condition was determined using background signal from terminated reactions (time=0 h). For each concentration of peptide, progress curves were plotted to determine the initial velocities (v ₀) of each reaction condition. A plot of the derived v ₀ values versus concentration of peptide in the enzyme reaction could not be fit using nonlinear regression to determine an apparent K_m for the peptide. Points represent mean values of duplicates. (B) For the FP anti-SAH detection assay, enzyme reactions were carried out in an optimized EZH2 reaction buffer [supplemented with 1% (v/v) DMSO] using 10 μM SAM and a range of H3K27me0 peptide concentrations (0–5 μM) and were stopped intermittently over a period of 150 min. Enzyme reactions were diluted 26-fold in the detection assay to avoid nonspecific detection of SAM in the presence of increasing concentrations of SAH. S/B was determined for each time point by using the FP signal in the no-reaction control (stopped at t=0 h) as background. Progress curves were plotted to determine initial velocity (v ₀) values as described above. Apparent K_m values were determined by performing nonlinear regression on the derived v ₀ values versus concentration of H3K27me0 peptide. Points represent mean values of duplicates, and error bars represent standard error of the mean. FP, fluorescence polarization.

Validation of the EZH2 AlphaLISA Assay with Reference Compounds

We determined the potency of several reference compounds with different mechanisms of action to validate the use of our optimized EZH2 AlphaLISA assay. We first tested poly-l-lysine and determined the average IC₅₀ to occur at 0.000008%±0.000001% poly-L-lysine, making it a very effective inhibitor of EZH2 activity (Fig. 6A). Sinefungin, a known homolog of SAM, weakly inhibited EZH2 activity with an IC₅₀ of 11.6±1.2 μM (Fig. 6B). The poor potency of sinefungin against EZH2 has been reported in literature.³⁷ Using SAH as an SAM-competitive inhibitor,³⁷ we determined IC₅₀ values in the presence of 500 nM and 10 μM SAM (1× and 20× SAM appK_m , respectively) (Fig. 6C). Under our assay conditions, SAH inhibited EZH2 with an IC₅₀ of 2.9±0.2 μM at SAM K_m . As expected, in the presence of high concentrations of SAM, SAH was significantly less potent with an IC₅₀ of 104±8 μM. We also explored the use of an H3-based peptide in which the target lysine is mutated to methionine (K27M) as a peptide substrate-competitive inhibitor. We have used a similar approach recently to inhibit the activity of LSD1 and JMJD2C demethylases.³⁸ The average IC₅₀ value for the K27M H3-peptide was determined to be 10.7±1.1 nM, and as expected, the IC₅₀ did not shift when measured in the presence of SAM at 20× K_m (Fig. 6D). Collectively, these data demonstrated that our optimized EZH2 assay was suitable for the detection and characterization of inhibitors with various mechanisms of action, and we next sought to explore the performance of our assay in an HTS setting.

Fig. 6.

Mechanism of action and IC₅₀ determinations using the AlphaLISA assay for EZH2. Enzyme reactions were carried out in an optimized EZH2 assay buffer using 3 nM PRC2 Complex, 500 nM SAM, and 100 nM H3K27me0 peptide and incubating for 3 h at RT. Inhibition is plotted as percent of control (POC), based on observed S/B obtained for negative (no enzyme) controls and positive (vehicle only) controls. (A) Representative IC₅₀ determination for poly-l-Lys solution. (B) IC₅₀ determination for sinefungin. (C) IC₅₀ for SAH was determined in the presence of SAM at K_m (500 nM) and at 20× K_m (10 μM) to investigate the mechanism of action (SAM competitive vs. noncompetitive). (D) IC₅₀ for H3K27M inhibitory peptide was determined in the presence of SAM at K_m and at 20× K_m . In (B–D), points represent mean values of triplicates, and error bars represent standard error of the mean.

EZH2 AlphaLISA Performance in a High-Throughput Screen

Given the high structural similarity in the purine and ribose portions of adenosine triphosphate (ATP) and SAM, we focused on compounds that were originally synthesized to target the ATP pocket of various kinases for initial screening. The hypothesis was that this subset of compounds would improve the probability of identifying inhibitors of EZH2, despite the differences in the H-bond networks formed with the respective enzyme classes. In addition, an internal computational program was run to generate create a diverse subset of 6,000 compounds based on Daylight fingerprint similarity.³⁹ This collection of compounds contained molecules originally designed and synthesized for binding to the ATP pocket of kinases. In this screen, compounds were screened at 10 μM using the optimized EZH2 AlphaLISA assay. We obtained an average-plate S/B of 17, and average Z′ was 0.7 (n=18 plates). Using an arbitrary primary hit cutoff of 50% inhibition, 280 primary hits (a hit rate of 4.6%) were chosen for further characterization. Since the AlphaLISA technology is known to be sensitive to several buffer additives, heavy metal ions, compounds that are fluorescent, colored, or interact with the ¹O₂ transfer,^40,41 it was necessary to triage the primary hits in an AlphaLISA artifact detection assay, in which all compounds were added after completion of the enzyme reaction and immediately before the addition of detection reagents. Thus, all primary hits were counterscreened to identify compounds that were potential false-positive hits due to interference with the detection technology rather than inhibition of EZH2. From the hits identified in this subset of 6,000 compounds, ∼60% of the primary hits reduced the AlphaLISA signal in the artifact assay by more than 30% and were therefore flagged as artifacts. These AlphaLISA false positives were also confirmed in an orthogonal mass spectrometry assay to not inhibit EZH2 (data not shown).

The EZH2 AlphaLISA potency of all primary hits was determined in parallel with the artifact assay. Using a top concentration of 50 μM, 95 compounds produced measurable IC₅₀ values. Only those compounds that had IC₅₀ values <5 μM were considered as on-target hits and were progressed further. This corresponded to 83 validated hits and a validated hit rate of 1.4% from the original screen of 6,000 compounds, demonstrating the potential of this approach to identify inhibitors of methyltransferases. The EZH2 AlphaLISA IC₅₀ values for the 83 remaining compounds were reproducible and correlated well with a Pearson r of 0.81 and a linear relationship of R ²=0.7 (Fig. 7).

Fig. 7.

Correlation plots to determine EZH2 AlphaLISA assay quality. Assay performance was assessed by correlating replicate Log₁₀ IC₅₀ values (M) obtained for a set of 83 compounds, which were confirmed as hits with IC₅₀ values <5 μM and reduced the AlphaLISA signal in the artifact assay by <30% POC at 10 μM. All initial hits that did not meet these criteria were excluded from further analysis. Linear regression (R ²) and correlation (Pearson r) of two sets of IC₅₀ values were performed to assess the reproducibility of the assay for determining compound potency. R ²=0.81 and Pearson r=0.7 for two independent sets of IC₅₀ determinations.

Overall, the EZH2 AlphaLISA assay was straightforward to implement in a high-throughput setting. However, it is unclear whether an assay that uses a peptide substrate would be predictive of EZH2-catalyzed turnover of more complex and biologically relevant substrates in cells. In the case of the PRC2 complex, larger and more complex substrates may interact with much higher affinity. Binding of these substrates to remote sites in the PRC2, as shown in this study with the addition of an H3K27me3 peptide to EED, may result in the allosteric modulation of EZH2 activity. In doing so, compounds that may act in this manner may provide ways of specifically targeting EZH2. Furthermore, there may be additional methylation sites available in more biologically relevant complex substrates, and depending on the nature of the interaction of the PRC2 with different methylation sites, it is important to keep in mind that the potency of compounds identified using simpler biochemical assays may not necessarily correlate with cellular activity. It is interesting to note that recent work by Diaz et al. demonstrated that the same tractable chemical series could be identified with either peptide or nucleosome as the substrate for EZH2.⁴²

In conclusion, we developed an AlphaLISA assay for EZH2 that allowed methylation of K27 of Histone H3 to be measured reliably in a high-throughput setting. The assay was nonradiometric and sensitive, requiring only 3 nM of PRC2 Complex (<1 nM EZH2) to achieve a robust signal (Z′>0.7) upon methylation of an H3 peptide substrate. This assay may serve as a valuable tool for the rapid screening of large compound libraries to aid in the identification of inhibitors of EZH2 activity and ultimately for the further development of innovative anticancer therapeutics that are capable of targeting abnormal changes in the epigenome.

Footnotes

Acknowledgments

AlphaLISA is a registered trademark of PerkinElmer Corp. The authors thank Dan La and Yax Sun for helpful discussions.

Disclosure Statement

No competing financial interests exist.

Abbreviations

References

Daub

, Specht

, Ullrich

. Strategies to overcome resistance to targeted protein kinase inhibitors. Nat Rev Drug Discov, 2004; 3:1001–1010.

Zhang

, Kim

. Small-molecule fluorescent chemosensors for Hg2+ ion. Anal Sci, 2009; 25:1271–1281.

Comley

. Epigentics: an emerging target class for drug screening. Drug Discov World, 2011; 12,Spring 2011:40–55.

Loyola

, Almouzni

. Marking histone H3 variants: how, when and why? Trends Biochem Sci, 2007; 32:425–433.

Arrowsmith

, Bountra

, Fish

, Lee

, Schapira

. Epigenetic protein families: a new frontier for drug discovery. Nat Rev Drug Discov, 2012; 11:384–400.

Cao

, Zhang

. SUZ12 is required for both the histone methyltransferase activity and the silencing function of the EED-EZH2 complex. Mol Cell, 2004; 15:57–67.

Cao

, Zhang

. The functions of E(Z)/EZH2-mediated methylation of lysine 27 in histone H3. Curr Opin Genet Dev, 2004; 14:155–164.

Czermin

, Melfi

, McCabe

, Seitz

, Imhof

, Pirrotta

. Drosophila enhancer of Zeste/ESC complexes have a histone H3 methyltransferase activity that marks chromosomal polycomb sites. Cell, 2002; 111:185–196.

Kirmizis

, Bartley

, Kuzmichev

et al. Silencing of human polycomb target genes is associated with methylation of histone H3 Lys 27. Genes Dev, 2004; 18:1592–1605.

10.

Kuzmichev

, Nishioka

, Erdjument-Bromage

, Tempst

, Reinberg

. Histone methyltransferase activity associated with a human multiprotein complex containing the enhancer of Zeste protein. Genes Dev, 2002; 16:2893–2905.

11.

Muller

, Hart

, Francis

et al. Histone methyltransferase activity of a Drosophila polycomb group repressor complex. Cell, 2002; 111:197–208.

12.

Fussbroich

, Wagener

, Macher-Goeppinger

et al. EZH2 depletion blocks the proliferation of colon cancer cells. PLoS One, 2011; 6:e21651.

13.

Kleer

, Cao

, Varambally

et al. EZH2 is a marker of aggressive breast cancer and promotes neoplastic transformation of breast epithelial cells. Proc Natl Acad Sci USA, 2003; 100:11606–11611.

14.

Simon

, Kingston

. Mechanisms of polycomb gene silencing: knowns and unknowns. Nat Rev Mol Cell Biol, 2009; 10:697–708.

15.

Simon

, Lange

. Roles of the EZH2 histone methyltransferase in cancer epigenetics. Mutat Res, 2008; 647:21–29.

16.

Varambally

, Dhanasekaran

, Zhou

et al. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature, 2002; 419:624–629.

17.

Morin

, Johnson

, Severson

et al. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat Genet, 2010; 42:181–185.

18.

Sneeringer

, Scott

, Kuntz

et al. Coordinated activities of wild-type plus mutant EZH2 drive tumor-associated hypertrimethylation of lysine 27 on histone H3 (H3K27) in human B-cell lymphomas. Proc Natl Acad Sci USA, 2010; 107:20980–20985.

19.

Margueron

, Justin

, Ohno

et al. Role of the polycomb protein EED in the propagation of repressive histone marks. Nature, 2009; 461:762–767.

20.

Martinez-Garcia

, Licht

. Deregulation of H3K27 methylation in cancer. Nat Genet, 2010; 42:100–101.

21.

Tsang

, Cheng

. Epigenetic regulation of signaling pathways in cancer: role of the histone methyltransferase EZH2. J Gastroenterol Hepatol, 2011; 26:19–27.

22.

, Bian

, Yang

et al.

Binding of different histone marks differentially regulates the activity and specificity of polycomb repressive complex 2 (PRC2)

Proc Natl Acad Sci USA, 2010; 107:19266–19271.

23.

Dorgan

, Wooderchak

, Wynn

et al. An enzyme-coupled continuous spectrophotometric assay for S-adenosylmethionine-dependent methyltransferases. Anal Biochem, 2006; 350:249–255.

24.

Hendricks

, Ross

, Pichersky

, Noel

, Zhou

. An enzyme-coupled colorimetric assay for S-adenosylmethionine-dependent methyltransferases. Anal Biochem, 2004; 326:100–105.

25.

Gauthier

, Caron

, Pedro

et al. Development of homogeneous nonradioactive methyltransferase and demethylase assays targeting histone H3 lysine 4. J Biomol Screen, 2012; 17:49–58.

26.

Quinn

, Allali-Hassani

, Vedadi

, Simeonov

. A chemiluminescence-based method for identification of histone lysine methyltransferase inhibitors. Mol Biosyst, 2010; 6:782–788.

27.

Kruidenier

, Chung

, Cheng

et al. A selective jumonji H3K27 demethylase inhibitor modulates the proinflammatory macrophage response. Nature, 2012; 488:404–408.

28.

Wigle

, Herold

, Senisterra

et al. Screening for inhibitors of low-affinity epigenetic peptide-protein interactions: an AlphaScreen-based assay for antagonists of methyl-lysine binding proteins. J Biomol Screen, 2010; 15:62–71.

29.

Collazo

, Couture

, Bulfer

, Trievel

. A coupled fluorescent assay for histone methyltransferases. Anal Biochem, 2005; 342:86–92.

30.

Wang

, Leffler

, Thompson

, Hrycyna

. A general fluorescence-based coupled assay for S-adenosylmethionine-dependent methyltransferases. Biochem Biophys Res Commun, 2005; 331:351–356.

31.

Han

, Xing

, Hu

, Zhang

, Liu

, Chai

. Structural basis of EZH2 recognition by EED. Structure, 2007; 15:1306–1315.

32.

Smith

, Gaitatzes

, Saxena

, Neer

. The WD repeat: a common architecture for diverse functions. Trends Biochem Sci, 1999; 24:181–185.

33.

Choi

, Seeliger

, Panjarian

et al. N-myristoylated c-Abl tyrosine kinase localizes to the endoplasmic reticulum upon binding to an allosteric inhibitor. J Biol Chem, 2009; 284:29005–29014.

34.

Habig

, Blechschmidt

, Dressler

et al. Efficient elimination of nonstoichiometric enzyme inhibitors from HTS hit lists. J Biomol Screen, 2009; 14:679–689.

35.

, Huang

, Jiang

, Seefeld

, McQueney

, Macarron

. The effect of triton concentration on the activity of undecaprenyl pyrophosphate synthase inhibitors. J Biomol Screen, 2003; 8:712–715.

36.

PerkinElmer Reagents Technical Support Wiki: The hook effect. https://perkinelmerreagents.onconfluence.com/display/ts/The+hook+effect. 2012 May 23.

37.

Richon

, Johnston

, Sneeringer

et al. Chemogenetic analysis of human protein methyltransferases. Chem Biol Drug Des, 2011; 78:199–210.

38.

, Fisch

, Long

et al. High-throughput TR-FRET assays for identifying inhibitors of LSD1 and JMJD2C histone lysine demethylases. J Biomol Screen, 2012; 17:27–38.

39.

Daylight Chemical Information Systems, Inc. www.daylight.com. 2012 May 23.

40.

PerkinElmer Reagents Technical Support Wiki: Buffer selection for Alpha assays. https://perkinelmerreagents.onconfluence.com/display/ts/Buffer+selection+for+Alpha+assays. 2012 May 23.

41.

Baell

, Holloway

. New substructure filters for removal of pan assay interference compounds (PAINS) from screening libraries and for their exclusion in bioassays. J Med Chem, 2010; 53:2719–2740.

42.

Diaz

, Machutta

, Chen

et al. Development and validation of reagents and assays for EZH2 peptide and nucleosome high-throughput screens. J Biomol Screen, 2012; 17:1279–1292.