Chemogenomics CRISPR-Cas9 Screen to Study Cell Cycle Progression.

(A): Schematic of whole-genome CRISPR-Cas9 screen.

(B): Volcano plots and 1nM camptothecin whole-genome screen results. The Orobas calculated “Differential Gene Effect” was plotted against the -log10(p-value) for this effect for each gene targeted in the screen, as calculated by the Orobas pipeline. Red dotted line indicates the established cut-off. Highlighted dots are genes with known roles in response to each treatment, with blue or yellow dots indicate genes that when inactivated resulted in sensitivity or resistance, respectively, to camptothecin.

(C): Representative STRING analysis networks for protein complexes with known roles in pathways that we identified as sensitive in our 1nM camptothecin chemogenomics screen. Blue dots in the STRING network indicate genes that when inactivated resulted in sensitivity to camptothecin.

(D): Same as in (B) but for 9.2nM colchicine whole-genome screen results.

(E): Same as in (C) but for 9.2nM colchicine whole-genome screen results.

(F): Same as in (B) but for 0.7µM palbociclib whole-genome screen results.

(G): Same as in (C) but for 0.7µM palbociclib whole-genome screen results.

Analysis of Camptothecin and Colchicine Chemogenomics Screen Reveals Novel Players in Cell Cycle Regulation.

(A): Dot plot comparison of the effect of gene mutation across three different screen conditions. Circle color indicates the strength of the positive or negative differential gene effect, circle size indicates its -log10(p-value) from our Orobas analysis.

(B): Volcano plot of novel genes identified in the 1nM camptothecin chemogenomics screen. Volcano plots are plotted as in (Fig. 1B), with highlighted dots representing novel genes identified in the camptothecin screen. (C): Dot plot of Metascape analysis of significant genes that sensitized or de-sensitized cells to 1nM camptothecin. The -log10(p-value) was plotted against each term. The Metascape determined enrichment for observed over expected genes associated with a given term is indicated by color of circle. The percentage of the genes above our established cut-off associated with a given term is indicated by the size of the circle.

(D): STRING analysis of genes identified from our analysis of the 1nM camptothecin chemogenomics screen. (E), (F) and (G) Same as in (B), (C) and (D) except for 9.2nM colchicine chemogenomics screen.

Mutation of Mitochondria Genes Attenuates the Sensitivity to Palbociclib.

(A): Dot plot of Metascape analysis of significant genes in the 0.7µM palbociclib chemogenomics screen. The -log10(p-value) was plotted against each term. The Metascape determined enrichment for observed over expected genes associated with a given term is indicated by color of circle. The percentage of the genes above our established cut-off are associated with a given term is indicated by the size of the circle.

(B): Volcano plot of genes identified from our analysis of the 0.7µM palbociclib chemogenomics screen. Volcano plots are plotted as in Fig. 1D, with highlighted dots representing novel genes identified in the palbociclib screen.

(C): STRING networks of novel protein complexes identified in palbociclib screen. Dots in the STRING network indicate genes that when inactivated resulted in sensitivity (blue, left) of resistance (yellow, right) to 0.7µM palbociclib.

(D): Dose-response curve of palbociclib-induced growth inhibition with oxidative phosphorylation inhibitors. HAP1 cells were exposed to palbociclib with or without increasing concentrations of the oxidative phosphorylation inhibitors rotenone, TTFA and oligomycin. Cell proliferation in each treatment was determined after 48 hours by PrestoBlue assay. All data were normalized to the initial dose of each oxidative phosphorylation inhibitor in indicated concentration of palbociclib. Error bars represent the mean of three replicates ±StdDev.

Loss of Polycomb Repressive Complex Components Display Specific Resistance to Palbociclib.

(A): Volcano plots in Fig. 3B except with members of PR-DUB, Polycomb Repressive Complex 1 and 2 highlighted.

(B): STRING analysis network of Polycomb Repressive Complex components. Yellow dots in STRING network indicate that inactivation of these genes conferred resistance to 0.7µM palbociclib in our chemogenomics screen.

(C): Dot plot of comparison of the effect of PRC2 complex member gene mutation across three different screen conditions, as in Fig. 2B.

(D): Dose-response curve of a nine-day quantitative Crystal Violet assay demonstrating rescue of palbociclib-induced growth inhibition with GSK126. All data were normalized to untreated cells and represents the mean of three technical replicates ±StdDev.

(E): Results of competitive growth assay for each indicated time point, normalized to the initial GFP+/GFP-ratio of the pool. The performance of each sgRNA in 1.5µM palbociclib vs Mock is shown, after normalizing to the control sgRNA. Data represents the mean ±SEM of the GFP+/GFP-ratios of three independent sgRNAs. (F): Dose-response curve of nine-day quantitative Crystal Violet assay results demonstrating rescue of palbociclib-induced growth in MTF2Δ and JARID2Δ cells. Data represent mean Crystal Violet staining of cell culture wells of three independently isolated monoclonal knockout cell lines ±StdDev.

(G): BrdU incorporation assay for Wild-type, SUZ12Δ, MTF2Δ and JARID2Δ cell lines were incubated in 1.5µM palbociclib for 24 hours. Left – Representative BrdU incorporation vs propidium iodide FACS analysis. Right – Quantification of BrdU incorporation assay. Data represents the mean S-phase cells for three independently isolated monoclonal cell lines ±StDev. *: p-value<0.05, n.s.: not significant, as determined by two-tailed unpaired Student’s t-test.

Polycomb 2.1 and PRC2.2 are Differentially Recruited to Promoters with CpG Island in HAP1.

(A): Left - Western blot of Wild-type, SUZ12Δ, MTF2Δ and JARID2Δ HAP1 cell lines, probed with the indicated antibodies. Right –Quantification of H3K27me3 signal, normalized to H3 intensity. Each bar represents mean normalized H3K27me3 signal intensity across three independently isolated monoclonal cell lines ±StDev. *: p-value<0.05, n.s.: not significant, as determined by two-tailed unpaired Student’s t-test.

(B): Left - Venn diagram of the overlap in promoters with decreased H3K27me3 read counts in CUT&RUN experiment between MTF2Δ or JARID2Δ cell lines compared to Wild-type cells. Promoters with differential enrichment of H3K27me3 total counts for a promoter was considered significant if the between monoclonal knockout and Wild-type log2 fold-change was ±1 and the adjusted p-value <0.1. Right - Venn diagram of the overlap in differentially increased transcript levels in RNA-Seq experiment between MTF2Δ or JARID2Δ cell lines compared to Wild-type cells. A transcript was considered as significantly differentially expressed if the log2 fold-change was ±1 and the adjusted p-value <0.05.

(C): Dot-plot of selected Metascape terms from analysis of promoters or mRNAs annotated as protein coding genes and displaying either significantly increased or decreased levels of H3K27me3 or transcripts. Color of the circle indicates the -log10(p-value) of the term and the size of circle indicates the percentage of the genes from the input list were represented in that term.

(D): Bedgraphs of promoters with significantly decreased H3K27me3 and increased transcripts that was either specifically dependent on MTF2 (left), JARID2 (center) or co-dependent on the presence either MTF2 or JARID2 (right). Tracks represent combined BED files from two clonal biological replicates.

(E): Heat map of H3K27me3 signal over a 10kb window for 1,877 peaks overlapping with CpG islands with the highest average signal intensity were identified in wild-type cells and plotted for the same loci in MTF2Δ and JARID2Δ cells. Plots are of one replicate of the two biological replicates used in the CUT&RUN experiment. CGI: CpG Island.

(F): Left – Heat map of H3K27me3 signal in promoters containing at least one CpG island for the top 2,000 promoters with the highest average H3K27me3 signal were identified in wild-type cells and plotted for the same loci in MTF2Δ and JARID2Δ cells in 5kb window around transcription start site (TSS). Left boarder of plot is −4kb from TSS, right boarder is +1kb from TSS. Plots are of one replicate of the two biological replicates used in the CUT&RUN experiment. Right - CUT&RUN signal averaged for all 25,124 CpG island-containing promoters for wild-type, MTF2Δ, and JARID2Δ cells in 5kb window around transcription start site (TSS). (G): Bar plot of log10(p-value) of Reactome (teal bars) and MSigDB (red bars) terms associated with protein coding genes that contain at least one CpG island.

CCND1 and CCND2 Expression is Increased in MTF2Δ Mutants.

(A): Volcano plot of DESeq2 calculated changes in log2 fold-change in H3K27me3 signal in promoters versus the log10(p-value) in enrichment in MTF2Δ cells determined by CUT&RUN.

(B): Same as in (A) but for transcript abundance determined by RNA-seq.

(C): Scatterplot of log2 fold-changes calculated by DESeq2 for MTF2Δ/Wild-type ratio of mRNA expression in the RNA-seq (x-axis) versus H3K27me3 signal in promoters from our CUT&RUN (y-axis) experiments. Genes whose log2 fold change in the RNA-seq experiment had an adjusted p-value of <0.05 and an adjusted p-value <0.1 in the CUT&RUN experiment where plotted.

(D): Bedgraph of H3K27me3 and transcript signal as well as CpG island location within the CCND1 promoter region (chr11:69,634,655-69,644,656; CpG island coordinates - chr11:69,636,368-69,643,828) and CCND2 promoter region (chr12:4,266,723-4,280,248; CpG island coordinates - chr12:4,269,237-4,275,239).

(E): Top – Western blots of Cas9-expressing pools of HAP1 cells transduced three independent sgRNAs targeting the indicated genes. Membranes were probed with the indicated antibodies Bottom – Quantification of CCND1, CCND2 and CCND3 signal, normalized to ß-actin intensity in HAP1 pooled knockouts. Mean signal intensity for three independent sgRNAs was plotted ±StDev. *: p-value<0.05, **: p-value<0.005, ***: p-value<0.0005, n.s.: not significant, Two tailed unpaired Student’s t-test.

(F): qRT-PCR relative quantification of CCND1, CCND2 and CCND3 mRNA levels in monoclonal wild-type, SUZ12Δ, MTF2Δ and JARID2Δ cells. Each bar is the mean for three biological replicates, each in performed in technical triplicate, ±StDev. *: p-value<0.05, **: p-value<0.005, ***: p-value<0.0005, n.s.: not significant, two tailed unpaired Student’s t-test.

(G): Same as in (E), but in wild-type, SUZ12Δ, MTF2Δ and JARID2Δ monoclonal cell lines. The same protein extracts used in Fig. 5A were used here, but membranes probed with indicated antibodies.

(H): Left – Representative Western blot of total RB1 and P-S807/8111-RB1 with increasing [palbociclib] in WT, MTF2Δ and JARID2Δ cells. Membranes were probed with secondary antibodies against mouse (RB1) or rabbit (P-S807/8111-RB1) conjugated to different fluorophores to enable simultaneous detection of phosphorylated and total forms of RB1. Right – Quantification of the ratio of P-S807/8111-RB1 to total RB1 signal determined by Western blot, plotted against [palbociclib]. Each point represents the P-RB1/RB1 ratio for two independently isolated monoclonal cell lines, error bars ±range. The experiment was repeated three times with similar results.

Dosing to Determine Inhibitor Concentration for Chemogenomics Screen.

(S1A): Drug dosing experiments were performed to determine screening concentrations. Cells were counted during passage in increasing doses of camptothecin (left), palbociclib (center) and colchicine (right).

(S1B): Representative images of flow cytometry traces from untreated cells or cells treated with 0.7µM palbociclib, 9.2nM colchicine or 1nM camptothecin treated cells for three days, then stained propidium iodide. Plots represent the number of stained cells with a given propidium iodide intensity.

(S1C): Venn diagrams showing overlap for significant genes that sensitized (left) or de-sensitized cells (right) to each condition tested. Genes that were determined as significant in all three screens were omitted in further analyses (see Table S2 for a list of the omitted genes).

Assays to Determine Resistance of PRC2 Component Mutants to CDK4/6 Inhibitors.

(S2A): Schematic of internally controlled competitive growth assay used to validate chemogenomics results or monoclonal cell line proliferation when treated with palbociclib. In experiments where we generated polyclonal pooled knockouts, GFP+ cells expressing Cas9 were mixed with GFP- cells without Cas9 (as in Fig 4E). For competitive growth experiments with monoclonal knockout cell lines, GFP+, Cas9 expressing cells were mixed with GFP- monoclonal knockout lines (as in Fig. S2C).

(S2B): Western blots demonstrating the efficacy of indicated sgRNA used in our competitive growth assay and use of polyclonal pooled knockouts. Cells were transduced with the indicated sgRNA, Cas9 expression was then induced for three days with doxycycline to generate polyclonal pooled knockouts. Whole-cell lysates from these pooled knockouts were then probed with antibodies against the indicated proteins.

(S2C): Competitive growth assay for or monoclonal knockout cell lines. Wild-type, MTF2Δ and JARID2Δ cell lines (GFP-) were mixed with wild-type HAP1 cells expressing Cas9 and GFP (GFP+) and treated with either DMSO (mock) or 1.5µM palbociclib (left), 3.5µM ribociclib (center) or 0.4µM abemaciclib (right) for twelve days. Cells were split every three days and the GFP-/GFP+ ratio was assessed every six days by flow cytometry. (S2D): Western blot of protein extracts from cells treated with DMSO (mock) or 1.5µM palbociclib for 48 hours, probed with indicated antibody. PARP cleavage and BIM over expression controls are included in the last lane from protein extracts from RPE1 cells over-expressing a doxycycline-inducible HA-tagged BIM to induce apoptosis.

Analysis of Changes in H3K27me3 Distribution in CUT&RUN and Differentially Expressed Genes in RNA-Seq Experiments.

(S3A): Top - PCA plot of H3K27me3 peaks called by macs2 from CUT&RUN experiment done in biological duplicate. Bottom – PCA plot of RNA-seq reads for experiment done in biological triplicate.

(S3B): Venn diagrams of the Gencode Annotations of promoters that had significantly up regulated (top row) and down regulated H3K27me3 (bottom row) from our CUT&RUN experiment for MTF2Δ (left) and JARID2Δ cells (right). Significant promoters were determined as having a log2 fold change ±1 and an adjusted p-value of <0.1.

(S3C): Same as in (S3B) only for our RNA-Seq experiments and significant promoters were determined as having a log2 fold change ±1 and an adjusted p-value < 0.05.

(S3D): Average H3K27me3 distribution over a 10kb window for 1,877 peaks overlapping with CpG islands with the highest average signal intensity were identified in Wild-type cells and plotted for the same loci in MTF2Δ and JARID2Δ cells.

(S3E): Bar plot of -log10(p-value) for the enrichment of a given transcription factors from ENCODE and ChEA databases binding to the list of promoters with overlapping CpG islands and H3K27me3 peaks.

Analysis of Differential H3K27me3 Distribution and Transcript Expression of D-type Cyclins in CUT&RUN and RNA-Seq Data Sets.

(S4A): Volcano plot of DESeq2 calculated changes in log2 fold-change in H3K27me3 signal in promoters versus the log10(p-value) in enrichment in JARID2Δ cells determined by CUT&RUN. CCND1 and CCND2 location within the dataset are indicated by yellow dots.

(S4B): Same as in (A) but for transcript abundance determined by RNA-seq.

(S4C): Scatter plot of log2 fold-change in transcript abundance vs H3K27me3 promoter signal for genes with an adjusted p-value <0.1 in our CUT&RUN and adjusted p-value <0.05 in our RNA-Seq from our JARID2Δ cell lines.

(S4D): Bedgraph of H3K27me3, transcript coverage and CpG island location within the CCND3 promoter region (chr6:41,940,179-4,195,017; CpG island coordinates: chr6:41,941,008-41,941,973).

(S4E): Quantification of protein signal from Western blot in Fig. 6G for CCND1 (left), CCND2 (center), and CCND3 (right) normalized to Actin. Each bar is the mean for three biological replicates, error bars ±StDev. *: p-value<0.05, **: p-value<0.005, ***: p-value<0.0005, n.s.: not significant, two tailed unpaired Student’s t-test.

(S4F): Dot plot of log2 fold-change for indicated mRNAs in MTF2Δ and JARID2Δ cells as determined by DESeq2. Established cut of ±1 log2 fold-change in transcript abundance indicated by dashed grey line. (S4G): Western blot for a panel of G1 regulators from whole-cell lysates of wild-type, SUZ12Δ, MTF2Δ and JARID2Δ cell lines from three independently isolated monoclonal knockouts. Membranes were probed with the indicated antibodies.