Inhibition of CDK12 elevates cancer cell dependence on P-TEFb by stimulation of RNA polymerase II pause release

Abstract

P-TEFb and CDK12 facilitate transcriptional elongation by RNA polymerase II. Given the prominence of both kinases in cancer, gaining a better understanding of their interplay could inform the design of novel anti-cancer strategies. While down-regulation of DNA repair genes in CDK12-targeted cancer cells is being explored therapeutically, little is known about mechanisms and significance of transcriptional induction upon inhibition of CDK12. We show that selective targeting of CDK12 in colon cancer-derived cells activates P-TEFb via its release from the inhibitory 7SK snRNP. In turn, P-TEFb stimulates Pol II pause release at thousands of genes, most of which become newly dependent on P-TEFb. Amongst the induced genes are those stimulated by hallmark pathways in cancer, including p53 and NF-κB. Consequently, CDK12-inhibited cancer cells exhibit hypersensitivity to inhibitors of P-TEFb. While blocking P-TEFb triggers their apoptosis in a p53-dependent manner, it impedes cell proliferation irrespective of p53 by preventing induction of genes downstream of the DNA damage-induced NF-κB signaling. In summary, stimulation of Pol II pause release at the signal-responsive genes underlies the functional dependence of CDK12-inhibited cancer cells on P-TEFb. Our study establishes the mechanistic underpinning for combinatorial targeting of CDK12 with either P-TEFb or the induced oncogenic pathways in cancer.

Graphical Abstract

Figure 1.

Inhibition of CDK12 triggers the release of P-TEFb from the inhibitory 7SK snRNP complex. (A) Schematic representation of controlling the kinase activity of P-TEFb with 7SK snRNP. Upon the release of inactive P-TEFb (CDK9 in red) from 7SK snRNP, the interaction of P-TEFb with HEXIM1 does not take place, resulting in the activation of P-TEFb (CDK9 in green). For simplicity, a large-scale conformational switching of 7SK upon the release of P-TEFb from the core of 7SK snRNP is not depicted. (B) CoIP of HEXIM1 with CDK9 from whole cell extracts of HCT116 cells. Cells were treated with DMSO (lane 1), THZ532 (400 nM) and two different doses of THZ531 for the indicated duration prior to preparation of whole cell extracts and detection of HEXIM1 and CDK9 in HEXIM1 immunoprecipitations (αHEXIM1 IP) and whole cell extracts (Input) by Western blotting. (C) Specificity of the targeting of CDK12 with THZ531 in target engagement assay. HCT116 cells were treated with DMSO and two different doses of THZ531 for the indicated duration (in hours) prior to preparation of whole cell extracts and detection of CDK12 and CDK7 in Biotin-THZ1 (1 μM) affinity purifications (Biotin AP) and whole cell extracts (Input) by Western blotting. (D) CoIP of HEXIM1 with CDK9 from whole cell extracts of the indicated HCT116 cell lines. Cells were treated with DMSO (lanes 1 and 5) and 3-MB-PP1 for the indicated duration (in hours) prior to preparation of whole cell extracts and detection of HEXIM1 and CDK9 in HEXIM1 immunoprecipitations (αHEXIM1 IP) and whole cell extracts (Input) by Western blotting. (E) HCT116 cells were treated with DMSO (lanes 1 and 6), THZ531 (400 nM) and triptolide (1 μM) as indicated for the indicated duration (in hours) prior to preparation of whole cell extracts and detection of the indicated proteins by Western blotting. (F) CoIP of HEXIM1 with CDK9 from whole cell extracts of HCT116 cells. Cells were treated with DMSO (-), THZ531 (400 nM) and KU-60019 (5 μM) alone and in combination as indicated for the indicated duration prior to preparation of whole cell extracts and detection of HEXIM1 and CDK9 in HEXIM1 immunoprecipitations (αHEXIM1 IP) and the indicated proteins in whole cell extracts (Input) by Western blotting.

Figure 2.

Transcriptional induction in CDK12-inhibited HCT116 cells occurs through P-TEFb-stimulated Pol II pause release. (A) MA plot representation of differentially transcribed genes (n = 16 560; P-adj ≤ 0.05; FC ≥ 1.25 and ≤ 0.8) from PRO-seq experiments (n = 2) of HCT116 CDK12^as/as cells treated with 3-MB-PP1 (5 μM) and NVP-2 (10 nM) as indicated for 1 h or 4.5 h compared to DMSO. Induced (UP), repressed (DOWN) and non-significantly regulated genes are depicted. FC, fold-change. (B) Change in the density of Pol II at promoter-proximal and body regions of genes with CDK12 inhibition-induced gain of Pol II at 1 h and 4.5 h from PRO-seq experiments (n = 2) of (A). (C) Change in pausing index of genes with CDK12 inhibition-induced gain of Pol II at 1 h and 4.5 h from PRO-seq experiments (n = 2) of (A). Median pausing index for each group is indicated. (D) Average density of engaged Pol II at genes with CDK12 inhibition-induced gain of Pol II at 4.5 h from PRO-seq experiments (n = 2) of (A). The promoter-proximal region is measured as a linear scale from –1000 to +1000 nucleotide from the TSS, gene body scaled into 50 bins per gene, and end of the gene as a linear scale from –1000 to +2000 nucleotide from the CPS. The insets show magnified promoter-proximal, early gene body and CPS gene regions. TSS, transcription start site. CPS, cleavage and polyadenylation site. nt, nucleotide.

Figure 3.

Inhibition of CDK12 induces gene expression downstream of key cancer pathways through P-TEFb. (A) MA plot representation of differentially expressed genes (n = 12 495; P-adj ≤ 0.001; Log₂ FC ≥ 1) from nuclear RNA-seq experiments (n = 3) of HCT116 CDK12^as/as cells that were synchronized by serum deprivation for 72 h prior to their culture in the serum‐containing medium for 4.5 h in the presence of DMSO or 3-MB-PP1 (5 μM). Induced (UP), repressed (DOWN) and non-significantly regulated genes are depicted. BBC3, FOS and CDKN1A model genes are highlighted. FC, fold-change. (B) GSEA of gene expression changes for protein-coding genes of (A) and in A-375 cells treated with DMSO and THZ531 (500 nM) for 6 h. Top gene sets from the hallmark gene set collection that are associated with induced genes are shown (FDR q-val ≤ 0.001). (C) Enrichment plots of the top NF-κB and p53 pathway gene sets of (B) from the GSEA of transcription changes of protein-coding genes obtained from PRO-seq experiments (n = 2) of HCT116 CDK12^as/as cells treated with DMSO, 3-MB-PP1 (5 μM) and NVP-2 (10 nM) as indicated for 1 h. NES, normalized enrichment score; positive values indicate enrichment among induced genes, negative values enrichment among down-regulated genes. FDR, false discovery rate. (D-F) HCT116 cell lines were treated with DMSO (-), THZ531 (400 nM), 3-MB-PP1 (5 μM), NVP-2 (20 nM) and KU-60019 (5 μM) alone and in combination as indicated for 3 h (D, F) or 12 h (E) prior to quantifying pre-mRNA levels of FOS, CDKN1A and BBC3 with RT-qPCR. Results normalized to the levels of GAPDH mRNA and DMSO-treated cells are presented as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, n.s., non-significant, determined by Student's t test. (G) HCT116 cell lines were treated with DMSO (–), THZ531 (400 nM) and NVP-2 (20 nM) alone and in combination as indicated for 16 h prior to preparation of whole cell extracts and detection of the indicated proteins by Western blotting.

Figure 4.

Co-targeting of CDK12 and P-TEFb decreases viability of cancer cells. (A, B) 6 × 6 cytotoxicity matrices with combinatorial titrations of THZ531 with NVP-2 at indicated doses to test for the synthetic lethality of compounds in HCT116 cell lines, depicting cytotoxicity (top) and synergy (bottom) of the combinations. Cytotoxicity values obtained at 48 hr of the treatments using CellTox Green assay were normalized to the DMSO control. Results represent the average of independent experiments (n = 3). Combinations with the highest Bliss synergy scores are highlighted (gold). (C–E) Cytotoxicity of HCT116 cell lines treated with DMSO (–), THZ531 (200 nM), 3-MB-PP1 (5 μM), NVP-2 (10 nM) and KU-60019 (5 μM) alone and in combination as indicated for 48 h measured using CellTox Green Cytotoxicity Assay. Results are presented as fluorescence values relative to the values of DMSO-treated cells and plotted as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; n.s., non-significant, determined by Student's t test. (F) Cell growth of HCT116 cell lines treated with DMSO, THZ531 (50 nM) and NVP-2 (1.25 nM) alone and in combination as indicated for seven days measured using live cell imaging. Results are presented as % confluency and plotted as the mean ± s.e.m. (n = 3). Arrows indicate the replenishment of medium and drugs at day 4. ****P < 0.0001, determined by two-way ANOVA using THZ531 and THZ531 + NVP-2 data sets.

Figure 5.

Co-targeting of CDK12 and P-TEFb stimulates p53-dependent apoptosis of HCT116 cells. (A) Quantification of cells (%) in the indicated cell-cycle phases based on flow cytometry profiles of propidium iodide-stained HCT116 cells that were synchronized by serum deprivation for 24 h prior to their culture in the serum‐containing medium for 24 h in the presence of DMSO (–), THZ531 (50 nM) and NVP-2 (50 nM) as indicated. (B) Apoptosis of HCT116 cell lines treated with DMSO, THZ531 (100 nM) and NVP-2 (5 nM) alone and in combination as indicated. Results obtained at the time points indicated below the graphs using RealTime-Glo Annexin V Apoptosis and Necrosis Assay are presented as luminescence values relative to the values of DMSO-treated cells at 2 h and plotted as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, determined by Student's t test using THZ531 and THZ531 + NVP-2 data sets. (C, D) Apoptosis of HCT116 cell line spheroid cultures treated with DMSO, THZ531 (100 nM), 3-MB-PP1 (5 μM) and NVP-2 (10 nM) alone and in combination as indicated. Spheroids were formed for 48 h prior to the treatments. Results obtained at the time points indicated below the graphs using RealTime-Glo Annexin V Apoptosis and Necrosis Assay are presented as luminescence values relative to the values of DMSO-treated cells at 2 h and plotted as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, determined by Student's t test using THZ531 and THZ531 + NVP-2 (C), and 3-MB-PP1 and 3-MB-PP1 + NVP-2 (D) data sets.

Figure 6.

Co-targeting of CDK12 and the NF-κB pathway decreases proliferation of HCT116 cells. (A) Cartoon depicting canonical (left) and DNA damage-induced (right) NF-κB signaling pathway. In the canonical pathway, trimerization of TNFRSF1A by TNF-α leads to activation of TAK1 (step 1), which activates IKKβ of the IKK complex (step 2). In turn, IKKβ induces degradation of IκBα (step 3), allowing NF-κB to translocate to the nucleus and stimulate gene transcription (step 4). In the DNA damage-induced pathway, induction of DSBs stimulates ATM and nuclear import of a fraction of NEMO (step 1), after which both proteins are exported from the nucleus (step 2) to stimulate TAK1-mediated activation of IKKβ within the indicated signalosomes (step 3). Steps 3 and 4 of the canonical pathway ensue. Pharmacological targeting of IKKβ by MLN120B and ATM by KU-60019 is indicated. TNF-α, tumor necrosis factor α; TNFRSF1A, TNF receptor superfamily member 1A; TAK1, TGF-β-activated kinase 1; TABs, TAK1 binding proteins; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; IKKα, inhibitor of κB (IκB) kinase α; IKKβ, inhibitor of IκB kinase β; NEMO/IKKγ, NF-κB essential modulator; p65/p50, heterodimer of NF-κB; ATM, ataxia-telangiectasia mutated; TRAF6, tumor necrosis factor receptor–associated factor 6. (B) Cytotoxicity of HCT116 cell lines treated with DMSO, TNF-α (20 ng/ml) and NVP-2 (10 nM) alone and in combination as indicated for 48 h measured using CellTox Green Cytotoxicity Assay. Results are presented as fluorescence values relative to the values of DMSO-treated cells and plotted as the mean ± s.e.m. (n = 3). **P < 0.01; ***P < 0.001, determined by Student's t test. (C) Apoptosis of HCT116 cell line spheroid cultures treated with DMSO, TNF-α (20 ng/ml) and NVP-2 (10 nM) alone and in combination as indicated. Spheroids were formed for 48 h prior to the treatments. Results obtained at the time points indicated below the graphs using RealTime-Glo Annexin V Apoptosis and Necrosis Assay are presented as luminescence values relative to the values of DMSO-treated cells at 2 h and plotted as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, determined by Student's t test using TNF-α and TNF-α + NVP-2 data sets. (D) Viability of HCT116 cell lines treated with DMSO, TNF-α (5 ng/ml) and NVP-2 (2.5 nM) alone and in combination as indicated for seven days measured using CellTiter-Glo 2.0 Cell Viability Assay. Results are presented as luminescence values relative to the values of DMSO-treated cells and plotted as the mean ± s.e.m. (n = 3). ***P < 0.001, determined by Student's t test. (E, F) HCT116 cell lines were treated with DMSO (–), THZ531 (400 nM), MLN120B (20 μM) and KU-60019 (5 μM) alone and in combination as indicated for 3 h prior to quantifying pre-mRNA levels of FOS and CDKN1A with RT-qPCR. Results normalized to the levels of GAPDH mRNA and DMSO-treated cells are presented as the mean ± s.e.m. (n = 3). **P < 0.01; ***P < 0.001, determined by Student's t test. (G, H) HCT116 cells were treated with DMSO (–), THZ531 (400 nM), NVP-2 (20 nM), MLN120B (20 μM) and KU-60019 (5 μM) alone and in combination as indicated for 16 h prior to preparation of whole cell extracts and detection of the indicated proteins by Western blotting. (I,J) Growth of HCT116 cell lines treated with DMSO, THZ531 (50 nM), MLN120B (2.5 μM) and KU-60019 (50 nM) alone and in combination as indicated for seven days measured using live cell imaging. Results are presented as % confluency and plotted as the mean ± s.e.m. (n = 3). Arrows indicate the replenishment of medium and drugs at day 4. ****P < 0.0001, determined by two-way ANOVA using THZ531 and THZ531 + MLN120B (I) and THZ531 and THZ531 + KU-60019 (J) data sets.

Figure 7.

Inhibition of CDK12 or P-TEFb switches the fate of CDK7-inhibited HCT116 cells from cell-cycle arrest to apoptosis. (A, B) Cytotoxicity of the indicated HCT116 cell lines treated with DMSO, YKL-5–124 (100 nM), THZ531 (200 nM) and NVP-2 (10 nM) alone and in combination as indicated for 48 h measured using CellTox Green Cytotoxicity Assay. Results are presented as fluorescence values relative to the values of DMSO-treated cells and plotted as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001; n.s., non-significant, determined by Student's t test. (C, D) Viability of the indicated HCT116 cell lines treated with DMSO, YKL-5–124 (5 nM), THZ531 (50 nM) and NVP-2 (2.5 nM) alone and in combination as indicated for seven days measured using CellTiter-Glo 2.0 Cell Viability Assay. Results are presented as luminescence values relative to the values of DMSO-treated cells and plotted as the mean ± s.e.m. (n = 3). **P < 0.01, determined by Student's t test. (E, F) Apoptosis of the indicated HCT116 cell line spheroid cultures treated with DMSO, YKL-5–124 (100 nM), THZ531 (150 nM) and NVP-2 (10 nM) alone and in combination as indicated. Spheroids were formed for 48 h prior to the treatments. Results obtained at the time points indicated below the graphs using RealTime-Glo Annexin V Apoptosis and Necrosis Assay are presented as luminescence values relative to the values of DMSO-treated cells at 2 h and plotted as the mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, determined by Student's t test using YKL-5–124 and YKL-5–124 + THZ531 (C), and YKL-5–124 and YKL-5–124 + NVP-2 (D) data sets.