Enhancer Remodeling during Adaptive Bypass to MEK Inhibition Is Attenuated by Pharmacologic Targeting of the P-TEFb Complex

Abstract

Targeting the dysregulated BRAF-MEK-ERK pathway in cancer has increasingly emerged in clinical trial design. Despite clinical responses in specific cancers using inhibitors targeting BRAF and MEK, resistance develops often involving nongenomic adaptive bypass mechanisms. Inhibition of MEK1/2 by trametinib in patients with triple-negative breast cancer (TNBC) induced dramatic transcriptional responses, including upregulation of receptor tyrosine kinases (RTK) comparing tumor samples before and after one week of treatment. In preclinical models, MEK inhibition induced genome-wide enhancer formation involving the seeding of BRD4, MED1, H3K27 acetylation, and p300 that drives transcriptional adaptation. Inhibition of the P-TEFb-associated proteins BRD4 and CBP/p300 arrested enhancer seeding and RTK upregulation. BRD4 bromodomain inhibitors overcame trametinib resistance, producing sustained growth inhibition in cells, xenografts, and syngeneic mouse TNBC models. Pharmacologic targeting of P-TEFb members in conjunction with MEK inhibition by trametinib is an effective strategy to durably inhibit epigenomic remodeling required for adaptive resistance.Significance: Widespread transcriptional adaptation to pharmacologic MEK inhibition was observed in TNBC patient tumors. In preclinical models, MEK inhibition induces dramatic genome-wide modulation of chromatin, in the form of de novo enhancer formation and enhancer remodeling. Pharmacologic targeting of P-TEFb complex members at enhancers is an effective strategy to durably inhibit such adaptation. Cancer Discov; 7(3); 302-21. ©2017 AACR.This article is highlighted in the In This Issue feature, p. 235.

Figure 1

Response to MEK inhibition in TNBC patient tumors from window-of-opportunity clinical trial. (A) Transcriptional response to trametinib in TNBC patient tumors in pre-treatment needle core biopsies (NCBs) and in corresponding surgical resections following 7 day trametinib treatment. Total number of expressed genes are indicated in black; percentage of transcripts induced (red) or suppressed (green) > 2 fold after trametinib treatment are indicated. (B) Tyrosine kinome transcriptional response (> 1.5 fold) to trametinib treatment in BL;BL patient tumors (blue) or CL;CL patient tumor (red). (C) Differential Expression-Seq2 (DESeq2) analysis comparing pre-trametinib and post-trametinib BL;BL tumors. Shown are differentially expressed kinases using 0.05 FDR for significance. Pt. 4 was excluded from the DESeq2 analysis because of high immune kinase expression but presented in the heat map for comparison. CL;CL tumor 6 is presented in the heat map for comparison to the BL;BL tumor response. (D) Adaptive response RTK protein upregulation in BL;BL patient tumors. (E) Scatterplot of RSEM transcript abundance values vs. MIB/MS TK MIB binding as a ratio of trametinib-treated surgical resection:pre-treatment NCB in BL;BL Pt. 5 (blue) and CL;CL Pt. 6 (red). Non-TKs are indicated with black circles. Arrows highlight decreased MEK1/2 MIB binding following trametinib.

Figure 2

Trametinib elicits a subtype-specific transcriptional response in TNBC cells. (A) Myc protein loss and RTK upregulation in SUM-159PT CL cells and HCC1806 BL cells after 48 h 10 nM trametinib treatment. (B) RNAseq results (mean RSEM of biological duplicates) in the indicated BL or CL human cell lines following 24 h 500 nM trametinib. (C) DESeq2 differential expression analysis of the kinome response to 24 h 500 nM trametinib comparing basal-like cell lines (HCC1806, SUM-149PT EpCAM+/CD49f+, MDA-MB-468) or claudin-low cell lines (SUM-159PT, Hs 578T, WHIM12). Log2 fold trametinib changes for kinases significant with a threshold of 0.05 FDR (Supplementary Table S1) are displayed in heat map following hierarchical clustering using (1 − Spearman Correlation)/2 as the distance metric and row scaling. (D) BL;BL patient tumor trametinib-upregulated kinases (Fig. 1C) overlapping with basal-like cell line trametinib-upregulated kinases (Supplementary Table S2). Hypergeometric test p value for overlap = 0.007 (patient FDR = 0.05, cell line FDR = 0.01). (E) Mean tyrosine kinome response to 24 h 500 nM trametinib of biological duplicates. TK transcripts upregulated > 1.5 fold are displayed and Log₂ magnitude of response plotted for BL (HCC1806, SUM-149PT EpCAM+; blue) and CL (SUM-159PT, Hs 578T, WHIM12; red) cells. TK transcripts upregulated > 1.3 fold are displayed for MDA-MB-468 cells due to low relative transcriptome-wide responsiveness to trametinib (B). (F) Sorting SUM-229PE parental cells into distinct BL EpCAM+/CD49f+ and CL EpCAM−/CD49f− populations using flow cytometry. (G) Top: Commonly (grey) and uniquely (blue, red) induced (> 2 fold) transcripts following 24 h 30 nM trametinib treatment of EpCAM+ or EpCAM− subpopulations. Bottom: mRNAseq in EpCAM+ or EpCAM− subpopulations showing response to 24 h 30 nM trametinib treatment. (H) Kinases induced > 1.5 fold or suppressed > 1.5 fold after 24 h 30 nM trametinib treatment that are either unique to EpCAM+ cells (blue) or EpCAM− cells (red), or both subpopulations (grey). (I) Upregulation of BL adaptive response RTK FGFR2 in EpCAM+ cells and CL adaptive response RTK PDGFRB in EpCAM− cells following 24 h 30 nM trametinib.

Figure 3

Remodeling of epigenomic landscape induced by MEK inhibition. (A) SUM-159PT ChIPseq density tracks at the DDR1 adaptive response RTK locus in the presence or absence of 24 h 100 nM trametinib. (B) Response of BRD4, H3K27ac, MED1, and p300 ChIPseq density to 24 h 100 nM trametinib alone, or the combination of 300 nM JQ1 at the highest 50 ranking BRD4 peaks defined by trametinib induction magnitude. Quantification of enhancers and super-enhancers by BRD4 density following 24 h 100 nM trametinib, 300 nM JQ1 or the combination in SUM-159PT (C) or HCC1806 (D) cells. (E) Enhancer quantification by BRD4 density following 24 h 30 nM trametinib in SUM-229PE EpCAM+/CD49f+ (dotted lines) or EpCAM−/CD49f− (solid lines) cells. (F) Left: Fold change of genome-wide BRD4 stitched peak ChIPseq density vs. transcriptional fold change of genes whose TSS resides +/− 200 kb from the BRD4 peak density in SUM-159PT cells with 24 h 100 nM trametinib. Right: Zoom of plot on left with warmer colors representing higher density of points showing enrichment in upper right quadrant. Empirical p value (< 10⁻⁵) from randomization test indicates that in each of 10,000 cycles of data randomization the number of points under the null hypothesis of no enrichment did not exceed the number of observed counts for this quadrant.

Figure 4

Proteasome or BET bromodomain inhibition attenuates trametinib-induced enhancers at kinase loci. (A) Time course of BRD4 density induced by 100 nM trametinib treatment at the DDR1 enhancer. (B) Classical enhancer (left) or super-enhancer (right) quantification by BRD4 density over 100 nM trametinib time course. (C) Time course of Myc protein levels following trametinib treatment showing anticorrelation of DDR1 protein induction and BRD4 density (A) or enhancer induction (B). (D) Western blot showing Myc stabilization and loss of PDGFRB, DDR1, and KDR upregulation with 8 h co-treatment of 100 nM trametinib and 30 nM bortezomib. (E) Loss of trametinib-induced DDR1 (left) and KDR (right) BRD4 enhancer density upon co-treatment with 30 nM bortezomib. (F) Upregulation of adaptive response RTK KDR upon doxycycline induction of Myc shRNA in stable SUM-159PT cells. (G) BRD4 density change at the highest ranking 50 trametinib-induced regions in response to 48 h 100 nM trametinib or 1 μg/ml doxycycline induction of Myc shRNA. (H) BRD4 density induction following 48 h 100 nM trametinib or 1 μg/ml doxycycline induction of Myc shRNA at PIK3R1, WNT5A or KDR1 adaptive response loci. (I) BRD4 ChIPseq density tracks depicting enhancer formation following 24 h 100 nM trametinib and enhancer blockade following co-treatment with 300 nM JQ1 at the DDR1, PIK3R1, and KDR SUM-159PT adaptive response genes. (J) Top: CRISPR/Cas9 deletion of the SUM-159PT DDR1 trametinib-induced enhancer. Bottom: Attenuation of DDR1 protein induction following 24 h 100 nM trametinib in stable SUM-159PT cell lines either heterozygous or homozygous for the enhancer deletion.

Figure 5

BET bromodomain inhibition enhances growth suppression elicited by MEK inhibition. (A) SUM-159PT four day growth curve with 30 nM trametinib, 300 nM JQ1, or the combination. (B) MDA-MB-231 (CL) four day growth curve with 30 nM trametinib, 500 nM I-BET151, or the combination. (C) Cell counting assay showing growth suppression in SUM-159PT cells in the presence or absence of 72 h 1 nM trametinib and the indicated siRNAs, normalized to non-targeting control siRNA. In the DMSO condition, all super-enhancer associated siRNAs yielded significantly different (p < 0.05) growth suppression relative to control siRNA except for PIK3R1. P values are indicated for siRNAs that showed significantly different growth suppression between DMSO and trametinib conditions. (D) Westerns with indicated antibodies in SUM-159PT cells treated 24 h with 100 nM trametinib, 500 nM I-BET151, or the combination. (E) Westerns showing loss of adaptive response RTKs in SUM-159PT cells (left) or SUM-229PE parental cells (right) in cells after 48 h 10 nM trametinib and BRD4 siRNA. (F) Top: SUM-159PT RNAseq showing percentage of genes induced (red) or suppressed (green) > 2 fold by 100 nM trametinib. Bottom: Percentage of trametinib-induced genes further induced (red) or suppressed (green) > 2 fold by co-treatment with 300 nM JQ1. (G) Long tail plot of trametinib mRNA induction or JQ1 mRNA suppression (100 nM trametinib : DMSO, or 100 nM trametinib + 300 nM JQ1 : 100 nM trametinib) for enhancer-associated genes with > 2 fold trametinib-induced expression change. (H) Four-week crystal violet assays in SUM-159PT parental cells (top) or SUM-159R (bottom) cells in the presence or absence of 30 nM trametinib or 300 nM JQ1.

Figure 6

MEK inhibition and BET bromodomain inhibition synergy in vivo. (A) Tumor volume in SUM-159PT xenografts: vehicle, 2 mg/kg daily trametinib, 30 mg/kg daily I-BET151, or combination treatment. Percent change in tumor volume from T11 (B) or 2225 (C) orthotopic serial transplant (OST) models following 2 week treatment of 1.0 mg/kg (chow) trametinib, 30 mg/kg (3X weekly, IP) I-BET151, or the combination. Error bars show +/− SEM. (D) Trametinib-induced mRNA upregulation and I-BET151-mediated suppression of DDR1 and PDGFRB as assayed from total RNA (left) or riboTRAP RNA (right) isolated from SUM-159PT xenografts (n=3). Error bars show SD from mean. (E) Top: SUM-159PT xenograft mean (n=3) transcriptome showing percentage of genes induced (red) or suppressed (green) > 1.5 fold by trametinib treatment. Bottom: Percentage of trametinib-induced genes further induced (red) or suppressed (green) > 1.5 fold by co-treatment with 300 nM JQ1. (F) mRNA fold change of SUM-159PT xenograft tyrosine kinases induced > 1.5 fold by trametinib treatment and corresponding JQ1-mediated suppression. Data are mean +/− SD, n=3 tumors.

Figure 7

Attenuation of adaptive response to MEK inhibition by P-TEFb complex perturbation. (A) PDGFRB and DDR1 western blot of SUM-159PT cells treated with the indicated P-TEFb complex siRNAs for 48 h followed by 24 h 100 nM trametinib. (B) Western with the indicated antibodies of SUM-159PT cells treated with CDK9 siRNA for 48 h followed by 24 h 100 nM trametinib. (C) Attenuation of 24 h 100 nM trametinib-induced PDGFRB and DDR1 upregulation by co-treatment with 100 nM HY-16462. (D) BRD4 (top) and p300 (bottom) ChIPseq density tracks at the DDR1 enhancer upon 24 h 100 nM trametinib alone, or in combination with either 300 nM JIB-04 or 1 μM SGC-CBP30. (E) Dose-dependent blockade of PDGFRB, KDR, and DDR1 upregulation to 24 h 100 nM trametinib by SGC-CBP30 co-treatment in SUM-159PT cells. (F) Loss of PDGFRB and DDR1 upregulation to 100 nM 24 h trametinib by co-treatment with 300 nM JIB-04 in SUM-159PT cells. (G) Enhancement of SUM-159PT (left) or MDA-MB-231 (right) growth suppression with 10 nM trametinib treatment with 5 μM SGC-CBP30. (H) Enhancement of SUM-159PT growth suppression with 5 nM trametinib and 500 nM JIB-04. (I) Left: Percentage of SUM-159PT transcripts further upregulated > 2 fold (red) or downregulated > 2 fold (green) with either 300 nM JQ1, 500 nM I-BET151, or 3 μM SGC-CBP30 in combination with 24 h 100 nM trametinib. Right: Trametinib-induced genes commonly suppressed by JQ1 and SGC-CBP30. (J) Model of dynamic enhancer formation in the adaptive response to MEK inhibition (left) and targeting strategies for different P-TEFb complex members to attenuate the response (right).