Quantification of histone H3 Lys27 trimethylation (H3K27me3) by high-throughput microscopy enables cellular large-scale screening for small-molecule EZH2 inhibitors

Abstract

EZH2 inhibition can decrease global histone H3 lysine 27 trimethylation (H3K27me3) and thereby reactivates silenced tumor suppressor genes. Inhibition of EZH2 is regarded as an option for therapeutic cancer intervention. To identify novel small-molecule (SMOL) inhibitors of EZH2 in drug discovery, trustworthy cellular assays amenable for phenotypic high-throughput screening (HTS) are crucial. We describe a reliable approach that quantifies changes in global levels of histone modification marks using high-content analysis (HCA). The approach was validated in different cell lines by using small interfering RNA and SMOL inhibitors. By automation and miniaturization from a 384-well to 1536-well plate, we demonstrated its utility in conducting phenotypic HTS campaigns and assessing structure-activity relationships (SAR). This assay enables screening of SMOL EZH2 inhibitors and can advance the mechanistic understanding of H3K27me3 suppression, which is crucial with regard to epigenetic therapy. We observed that a decrease in global H3K27me3, induced by EZH2 inhibition, comprises two distinct mechanisms: (1) inhibition of de novo DNA methylation and (II) inhibition of dynamic, replication-independent H3K27me3 turnover. This report describes an HCA assay for primary HTS to identify, profile, and optimize cellular active SMOL inhibitors targeting histone methyltransferases, which could benefit epigenetic drug discovery.

Keywords: EZH2; KMT6; chromatin modulators; high-content analysis; histone methyltransferase.

Figure 1.

H3K27me3 was directly suppressed in a time-dependent manner after small interfering RNA (siRNA)–mediated knockdown of EZH2. (A, B) HeLa S3 cells, treated with siRNA, were immunostained with specific antibodies against EZH2 and H3K27me3 followed by image acquisition and image analysis. (A) The first panel on the left displays EZH2 (green) merged with a staining of the nuclei (blue), the second panel displays EZH2 alone (monochrome), the third panel displays H3K27me3 (red) merged with a staining of the nuclei (blue), and the fourth panel displays H3K27me3 alone (monochrome) at indicated days after siRNA knockdown. Scale bar = 10 µM. (B) Quantification of the relative nuclear EZH2 protein and H3K27me3 together with the EZH2 messenger RNA (mRNA) level. Responses are plotted as percentage of the lipid control. Mean values represent the average of approximately 1500 nuclei analyzed. Error bars show the standard deviation from three replicates. ****p < 0.0001. (C) Schematic image analysis for phenotypically quantifying cellular EZH2 and H3K27me3 using high-content analysis. Using the DNA stain, immunostained nuclei were segmented (DNA, nuclei 1–5) and analyzed for size and shape. Based on the segmentation, binary object masks were generated and subsequently superimposed over the EZH2-and H3K27me3-specific staining of the image set to accurately quantify the antibody-specific signal intensity of every segmented nucleus. Fields appearing in the figure are smaller than a complete field of view.

Figure 2.

Small-molecule EZH2 inhibition induced a progressive H3K27me3 suppression over 3 days accompanied by a genome-wide modification switch from H3K27me3 to H3K27ac in MDA-MB-231 cells. Cells, treated with the EZH2 tool inhibitor for a maximum of 3 days at 37 °C in 5% CO₂, were immunostained followed by image acquisition and image analysis. The graphs show the relative H3K27 modification levels at (A) after various treatment times using 3 µM tool inhibitor and (B) after 3 days of treatment at tool inhibitor concentrations indicated. Responses are plotted as percentage of the DMSO control. Mean values represent the average of 2000 nuclei analyzed. Error bars show the standard deviation from three replicates. By multiple comparisons, every mean was compared with the mean representing zero hours of EZH2 inhibition (0 h). **p ≤ 0.01. ***p ≤ 0.001. ****p < 0.0001. (C) The upper panel displays a staining of H3K27me3 (red) merged with a staining of the nuclei (blue) with increasing EZH2 tool inhibitor concentrations, as indicated. The lower panel displays a staining of H3K27ac (green) merged with a staining of the nuclei (blue) at similar EZH2 tool inhibitor concentrations. Scale bar = 5 µM.

Figure 3.

Different EZH2 inhibitors demonstrated distinct cellular inhibitory activity on H3K27me3. MDA-MB-231 cells, treated with varying concentrations of the different compounds over 3 days at 37 °C in 5% CO₂, were immunostained followed by image acquisition and image analysis. The graphs show the concentration response of the inhibitors in the relative H3K27me3 level (A) and in the relative number of analyzed nuclei (B). Responses are plotted as percentage of the DMSO control. Mean values represent the average of approximately 1500 nuclei analyzed. Error bars show the standard deviation from three replicates. (C) Chemical structures of used EZH2 inhibitors.

Figure 4.

A decrease in inhibitor potency occurred as a result of miniaturization and automation from 384-well plate format to 1536-well plate format with the same order in potency. (A, B) MDA-MB-231 cells were treated with varying concentrations of the EZH2 tool inhibitor and three structurally related EZH2 inhibitors over 3 days at 37 °C in 5% CO₂ using the 384-well plate format (A) or 1536-well plate format (B). The graphs show the concentration response of the different EZH2 inhibitors in the relative H3K27me3 level at inhibitor concentrations indicated. Responses are plotted as percentage of the DMSO control. Error bars show the standard deviation from three replicates. (C) Overview of the EZH2 inhibitor IC₅₀ values for the 384-well plate format and 1536-well plate format. All IC₅₀ values are the average of at least three determinations.

Figure 5.

Small-molecule EZH2 inhibition reveals a dynamic, cell cycle–independent H3K27me3 fraction. (A–C) MDA-MB-231 cells were treated with 3 µM tool inhibitor and (B, C) a fluorescently labeled thymidine analogue EdU (10 µM) for 6 h at 37 °C in 5% CO₂. (A) The graph shows the relative H3K27me3 level in different cell cycle phases before and after 6 h of EZH2 inhibition. All nuclei showed a comparable and significant degree of H3K27me3-demethylation, regardless of which cell cycle phase they were in. No statistically significant difference was determined between those three groups after 6 h of EZH2 inhibition, tested by multiple comparisons. (B) The graph show the percentage of cells with (blue) and without (gray) incorporated EdU after 6 h of EZH2 inhibition, separating the cells into two populations: one population that has synthesized newly DNA versus one population that has not during the 6 h of EZH2 inhibition. (C) The graph shows the relative H3K27me3 level of cells with and without incorporated EdU after 6 h of EZH2 inhibition. Cells showing incorporated EdU (blue, containing newly synthesized DNA) showed an H3K27me3 reduction of 31%. Notably, cells showing no Edu incorporation (gray, containing no newly synthesized DNA) showed an H3K27me3 reduction of 40%. Mean values represent the average of 500 to 1500 nuclei analyzed. Responses are plotted as percentage of the DMSO control. Error bars show the standard deviation from three replicates. **p ≤ 0.01. ***p ≤ 0.001. ****p < 0.0001.