SAM-DNMT3A, a strategy for induction of genome-wide DNA methylation, identifies DNA methylation as a vulnerability in ER-positive breast cancers

Abstract

DNA methylation is an epigenetic mark that plays a critical role in regulating gene expression. DNA methyltransferase (DNMT) inhibitors, inhibit global DNA methylation and have been a key tool in studies of DNA methylation. A major bottleneck is the lack of tools to induce global DNA methylation. Here, we engineered a CRISPR based approach, that we initially designed, to enable site-specific DNA methylation. Using the synergistic activation mediator (SAM) system, we unexpectedly find that regardless of the targeted sequence any sgRNA induces global genome-wide DNA methylation. We term this method SAM-DNMT3A and show that induction of global DNA methylation is a unique vulnerability in ER-positive breast cancer suggesting a therapeutic approach. Our findings highlight the need of caution when using CRISPR based approaches for inducing DNA methylation and demonstrate a method for global induction of DNA methylation.

Fig. 1. SAM-DNMT3A induces high levels of DNA methylation at desired sites.

A Vectors used for development of SAM-DNMT3A. The sgRNA vector contains puromycin resistance and a PP7-DNMT3A fusion protein. The DNMT expression vector contains a blasticidine resistance gene and a dCas9 enzyme fused to a DNMT protein. B The SAM-DNMT system. At each target site three DNMT enzymes are recruited. C sgRNAs used for induction of DNA methylation at the BRCA1 promoter. D DNA methylation measured using HRM assay at BRCA1 promoter 3 days following transfection of HEK293T cells with the SAM-DNMT3A system. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = 3.69e-03 (sgGFP), 2.94e-07 (sgBRCA1_1), 5.06e-07 (sgBRCA1_2)). E DNA methylation measured using HRM assay at BRCA1 promoter 3 days post-transfection of HEK293T cells with the SAM-DNMT3B system. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = 4.68e-02 (sgGFP), 1.32e-07 (sgBRCA1_1), 2.07e-07 (sgBRCA1_2)). F DNA methylation measured using HRM assay at BRCA1 promoter 3 days post-transfection of HEK293T cells with the SAM-DNMT1 system. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = 7.35e-02 (sgGFP), 2.35e-06 (sgBRCA1_1), 7.24e-07 (sgBRCA1_2)). G DNA methylation measured using HRM assay at BRCA1 promoter 3 days post-transfection of HEK293T cells with the SunTag-DNMT3A system. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = 3.16e-03 (sgGFP), 1.31e-08 (sgBRCA1_1), 2.88e-08 (sgBRCA1_2)). NS, not significant, ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001.

Fig. 2. SAM-DNMT3A pooled screens do not identify any hits associated with proliferation.

A Strategy for pooled DNA methylation screen used to identify DNA methylation sites that affect cell proliferation. Correlation (Pearson) of sgRNA abundance between cells with active and inactive DNMT3A. p-value calculated using a two-tailed Pearson correlation analysis. for (B) MCF7 (P = 4.34e-56). C BRE80-T5 (P = 5.21e-32). D T47D (P = 1.04e-115). Correlation (Pearson) of gene scores (average of all sgRNAs targeting a methylation region) between cells expressing active or inactive DNMT3A p-value calculated using a two-tailed Pearson correlation analysis for (E) MCF7 (P = 7.84e-10). F BRE80-T5 (P = 8.76e-14). G T47D (P = 7.16e-45).

Fig. 3. SAM-DNMT3A induces global non-specific DNA methylation.

A Illustration of experiment aimed to identify global methylation changes induced by SAM-DNMT3A. B For each methylation site the fold change in methylation was calculated by comparing methylation levels in cells with SAM-DNMT3A and SAM-DNMT3A-inactive. The average fold change across all DNA methylation sites is plotted for each sgRNA. C Manhattan plot showing all methylation sites in BRE80-T5 cells with no sgRNAs or with sgAAVS1_118. D Manhattan plot showing all methylation sites in T47D cells with no sgRNAs or with sgAAVS1_118. E MDS dimensionality plot using all replicates of EPIC 1.0 array for BT549 cells transduced with mock (control), SAM-DNMT3A-inactive or SAM-DNMT3A-active with sgAAVS1_118. F Manhattan plot showing all methylation sites in BT549 cells with no sgRNAs or with sgAAVS1_118. Each point is the average of three independent replicates. G Manhattan plot showing all methylation sites in MCF7 cells with no sgRNAs or with sgAAVS1_118. Each point is the average of three independent replicates. H HRM assay using LINE-1 probes in MCF7 cells expressing SAM-DMT3A-active or SAM-DNMT3A-inactive with sgAAVS1_118 in cells treated for 48 h with 0.2 µM of the DNMT inhibitor decitabine. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = SAM-DNMT3A-active 8.65e-07 (0.025 µM), 4.87e-07 (0.05 µM), 1.01e-05 (0.1 µM), 1.31e-03 (0.2 µM), SAM-DNMT3A-inactive 1.54e-04 (0.025 µM),6.07e-05 (0.05 µM), 1.33e-03 (0.1 µM), 1.00 (0.2 µM)). NS, not significant, ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001.

Fig. 4. SAM-DNMT3A scans the genome in the presence of an sgRNA.

Dynamics of HaloTag-SAM-DNMT3A measured in HeLa cells transfected with SAM-DNMT3A in the presence or absence an AAVS1 sgRNA. A Quantification of dCas9-DNMT3A movement in the presence or absence of a sgRNA. Data was collected from 534 (SAM-DNMT3A + sgRNA, n = 5 cells) and 366 (SAM-DNMT3A No sgRNA, n = 3 cells) trajectories. B Representative images (from n = 5 cells for sgRNA and n = 3 cells for No sgRNA control experiments) of trajectory movement of dCas9-DNMT3A in the presence or absence of an sgRNA. Colour bar represents degree of movement.

Fig. 5. Induction of DNA methylation is a vulnerability in ER positive breast cancers.

A DNA methylation measured in ER-positive and ER-negative breast cells using HRM assay at LINE-1 sequences following transduction of SAM-DNMT3A with sgAAVS1_118. Data are mean ± SD, n = 3 biological replicates. p-value was calculated using unpaired two-tailed T.test (P = 0.015). B Images of Crystal violet staining of ER positive and ER negative breast cancer cell lines containing active or inactive DNMT3A 7 days post-transduction with no sgRNA or AAVS1 targeting sgRNAs. C Proliferation changes of all AAVS1 sgRNAs in ER-positive and ER-negative breast cancers. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = 2.00e-04, 1.00e-04 (BRE80-T5), 0.11, 0.76 (BT549), 0.32, 0.40 (CAL120), 1.77e-08, 1.25e-08 (MCF7), 1.7e-09, 1.9e-09 (MDAMB415), 1.19e-05, 1.41e-05 (T47D), 8.45e-07, 3.88e-07 (EFM19)). D Proliferation changes (from (C)) in ER-positive and ER-negative breast cancer cell lines following transduction with SAM-DNMT3A. Each dot is an average of three biological replicates. p-value was calculated using unpaired two-tailed T-test (P = 5.05e-20). E Proliferation of MCF7 cells with SAM-DNMT3A or SAM-DNMT3A-inactive and an AAVS1 targeting sgRNA (sgAAVS1_118) treated with the indicated concentrations of the DNMT inhibitor decitabine. Data are mean ± SD, n = 3 biological replicates. p-value was calculated using one-way ANOVA. Data are mean ± SD, n = 3 biological replicates. (P = SAM-DNMT3A 4.5e-03 (0.2 µM), 9e-04 (0.5 µM), 4e-03 (1 µM), 0.31 (2 µM), 0.47 (5 µM), SAM-DNMT3A-inactive 4e-15 (0.2 µM), 2e-15 (0.5 µM), 1e-15 (1 µM), 1e-15 (2 µM), 1e-15 (5 µM)).