Targeting CDK9 Reactivates Epigenetically Silenced Genes in Cancer

Hanghang Zhang; Somnath Pandey; Meghan Travers; Hongxing Sun; George Morton; Jozef Madzo; Woonbok Chung; Jittasak Khowsathit; Oscar Perez-Leal; Carlos A Barrero; Carmen Merali; Yasuyuki Okamoto; Takahiro Sato; Joshua Pan; Judit Garriga; Natarajan V Bhanu; Johayra Simithy; Bela Patel; Jian Huang; Noël J-M Raynal; Benjamin A Garcia; Marlene A Jacobson; Cigall Kadoch; Salim Merali; Yi Zhang; Wayne Childers; Magid Abou-Gharbia; John Karanicolas; Stephen B Baylin; Cynthia A Zahnow; Jaroslav Jelinek; Xavier Graña; Jean-Pierre J Issa

doi:10.1016/j.cell.2018.09.051

. Author manuscript; available in PMC: 2019 Nov 15.

Published in final edited form as: Cell. 2018 Oct 25;175(5):1244–1258.e26. doi: 10.1016/j.cell.2018.09.051

Targeting CDK9 Reactivates Epigenetically Silenced Genes in Cancer

Hanghang Zhang ¹, Somnath Pandey ¹, Meghan Travers ², Hongxing Sun ¹, George Morton ³, Jozef Madzo ¹, Woonbok Chung ¹, Jittasak Khowsathit ^4,⁵, Oscar Perez-Leal ⁶, Carlos A Barrero ⁶, Carmen Merali ⁶, Yasuyuki Okamoto ¹, Takahiro Sato ¹, Joshua Pan ⁷, Judit Garriga ¹, Natarajan V Bhanu ⁸, Johayra Simithy ⁸, Bela Patel ¹, Jian Huang ¹, Noël J-M Raynal ⁹, Benjamin A Garcia ⁸, Marlene A Jacobson ³, Cigall Kadoch ⁷, Salim Merali ⁶, Yi Zhang ¹, Wayne Childers ³, Magid Abou-Gharbia ³, John Karanicolas ⁴, Stephen B Baylin ², Cynthia A Zahnow ², Jaroslav Jelinek ¹, Xavier Graña ¹, Jean-Pierre J Issa ^1,^10,^*

¹Fels Institute for Cancer Research, Lewis Katz School of Medicine at Temple University, Philadelphia, PA 19140, USA.

²The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD 21231, USA.

³Moulder Center for Drug Discovery Research, Temple University School of Pharmacy, Philadelphia, PA 19140, USA.

⁴Molecular Therapeutics Program, Fox Chase Cancer Center, Philadelphia, PA 19111, USA.

⁵Department of Molecular Biosciences, University of Kansas, Lawrence, KS 66045, USA

⁶Department of Pharmaceutical Sciences, Temple University School of Pharmacy, Philadelphia, PA 19140, USA.

⁷Department of Pediatric Oncology, Dana-Farber Cancer Institute and Harvard Medical School, Boston, MA 02215, USA.

⁸Epigenetics Institute, Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA

⁹Département de pharmacologie et physiologie, Université de Montréal, Montréal H3C 3J7, Canada.

¹⁰Lead Contact

AUTHOR CONTRIBUTIONS:

H.Z. and J.-P. I. designed experiments, analyzed data and wrote the manuscript. H.Z. conducted most experiments. H.Z. and S.P. performed ChIP-qPCR experiments. S.P. performed Co-IP studies. S.P. and O. P-L. designed BRG1 mutants and performed confocal microscopy experiments. B.P. and J.J. performed and analyzed DREAM and RRBS. W.C. performed ATAC-seq. J.G. and X.G. provided CDK9-CyclinT1 constructs in this study. J.P. and C.K. provided a lentiviral BRG1 construct. H.Z., M. T., H. S., Y. Z. and C.A.Z. assisted with mouse studies. G. M., W. C. and M. A-G. designed and synthesized different analogs. H.Z., T.S., Y.O. and J. M. performed bioinformatics analysis. J. K. and J. K. performed molecular docking analysis. C.A.B., C.M., S.M. and X.G. performed and analyzed in vitro phosphoproteomic data. N.V.B., J.S. and B.A.G. performed and analyzed histone mass spectrometry. All authors provided critical analysis of the data and editing of the manuscript.

Correspondence: jpissa@temple.edu

PMCID: PMC6247954 NIHMSID: NIHMS1510720 PMID: 30454645

SUMMARY:

Cyclin-Dependent Kinase 9 (CDK9) promotes transcriptional elongation through RNAPII pause release. We now report that CDK9 is also essential for maintaining gene silencing at heterochromatic loci. Through a live cell drug screen with genetic confirmation, we discovered that CDK9 inhibition reactivates epigenetically silenced genes in cancer, leading to restored tumor suppressor gene expression, cell differentiation, and activation of endogenous retrovirus genes. CDK9 inhibition dephosphorylates the SWI/SNF protein BRG1, which contributes to gene reactivation. By optimization through gene expression, we developed a highly selective CDK9 inhibitor (MC180295, IC50=5nM) that has broad anti-cancer activity in-vitro and is effective in in-vivo cancer models. Additionally, CDK9 inhibition sensitizes to the immune checkpoint inhibitor α-PD-1 in vivo, making it an excellent target for epigenetic therapy of cancer.

In brief

Inhibition of a kinase typically associated with heterochromatin formation leads to reactivation of tumor suppressor genes and increase sensitivity to immunotherapy in cancer models

INTRODUCTION

Epigenetic reprogramming in cancer leads to heritable changes in gene expression, such as silencing of tumor suppressor genes (TSG) (Jones et al., 2016). These altered epigenetic marks come about as a result of aging and acquisition of genetic and epigenetic changes in readers/writers/editors of the epigenome (Issa, 2014; Jones et al., 2016). DNA methylation in the promoter of TSGs results in gene silencing through recruitment of methyl-binding domain (MBD) proteins such as MeCP2 (Jones et al., 2016; Taby and Issa, 2010). In turn, MBDs recruit repressor complexes resulting in heterochromatin formation (Nan et al., 1998). Reversing epigenetic modifications by drugging the epigenome led to the field of epigenetic therapy. However, clinically, epigenetic treatment options remain limited.

Cyclin-dependent kinases (CDKs) come in two partially overlapping classes – regulators of the cell cycle (e.g. CDK1, 2, 4, 6, 7) and regulators of transcription (e.g. CDK7, 8, 9, 10–13). CDK9, the catalytic subunit of P-TEFb, is a transcriptional activator recruited to promote RNAPII promoter-proximal pause release by phosphorylating negative elongation factors (DSIF and NELF) (Adelman and Lis, 2012; Garriga and Graña, 2004). CDK9-mediated phosphorylation of the C-terminal domain (CTD) of RNAPII on Serine-2 allows recruitment of RNA processing factors, which work on the nascent RNA as it emerges from RNAPII. P-TEFb promotes transcriptional elongation of several signal-responsive genes that regulate proliferation, development, stress and/or damage responses (Adelman and Lis, 2012; Garriga and Graña, 2004), such as MYC (Rahl et al., 2010), NFkB (Barboric et al., 2001) and MCL1 (MacCallum et al., 2005).

We have been using the YB5 cell-based system derived from the human colon cancer cell line, SW48, as a platform for epigenetic drug screens (Raynal et al., 2012). YB5 contains a single insertion of the cytomegalovirus (CMV) promoter driving the green fluorescent protein (GFP) gene. GFP is silenced in YB5 by epigenetic mechanisms including DNA hypermethylation leading to closed chromatin formation, and its expression can be reactivated by treatment with DNA methylation inhibitors and/or HDAC inhibitors. In a screen for GFP reactivation in YB5, we identified a novel drug class that reactivates silenced genes by targeting CDK9 without affecting DNA methylation. We show that CDK9 is required for gene silencing in cancer cells, in part by phosphorylation of BRG1.

RESULTS:

Identification of CDK9 as a novel epigenetic target

Phenotypic screens provide an unbiased approach for identifying targets and drugs. We screened YB5 against the NDL-3040 natural compound library (Figure S1A). This assay had a z-factor of 0.6, indicating adequacy for high-throughput screening. At a stringent criterion for GFP induction [(mean of all compounds) + 3 standard deviations], 33 compounds (1.1%) were positive. A further selection of hits with ≥ 25% relative activity compared to the positive control yielded 18 compounds, 15 of which were successfully validated by 24hr dose curves, fluorescence microscopy and qPCR. Five of these hits had similar structures including an aminothiazole core (Figure S1A). Structure and data on HH0, a representative aminothiazole compound are shown in Figures 1A, S1B, and S1C. We started chemical optimization by screening a library of 93 aminothiazole analogs at multiple doses (ranging from 2.5 μM to 50 μM) and identified HH1 as the most potent (Figure SA). HH1 was active at 5 μM in YB5 (Figure 2A) and was also validated in HCT116-GFP, a colon cancer cell line where GFP is inserted downstream of the hypermethylated promoter SFRP1 (Cui et al., 2014), and in MCF7-GFP, a breast cancer line derived by introducing GFP under the control of a methylated CMV promoter (Figure SD). Thus, HH1 can reactivate silenced GFP in three distinct live cell assays.

Figure 1: — CDK9 inhibition reactivates epigenetically silenced genes. See also Figure S1 and Table S3.

(A) Expression of GFP (measured by FACS) after 24hr treatment of YB5 with HH0 (structure at the top). Depsipeptide (Depsi), an HDACi, was used as a control. Insert: representative fluorescent microscopy of YB5 treated with DMSO (left) and HH0 (right).

(B) GFP expression 24hr after treatment with CDK inhibitors measured by FACS. Corresponding structures are shown on top.

(C)Reactivation of GFP, SYNE1 and MGMT 24hr after treatment with CDK9 inhibitors as detected by qPCR.

(D) Reactivation of GFP, SYNE1 and MGMT 72hr after dominant-negative CDK9 (dnCDK9) overexpression (TET-off) detected by qPCR. HH1 (25μM, 24hr) was used as a positive control. The Western Blot (right) shows that dnCDK9 is overexpressed when cells are grown in the absence of tetracycline, together with decreased phosphorylation of RNA Pol ll at Ser2 (pSer2).

(E) Overexpression of P-TEFb (CDK9 and Cyclin T1) for 72hr abolished the effect of CDK9 inhibitors (24hr) on the activation of GFP, SYNE1 and MGMT as detected by qPCR. Depsipeptide was used as a negative control (uninhibited by CDK9 overexpression). The Western Blot (right) shows the overexpression of CDK9 and Cyclin T1 72hr after viral transduction. Single transduction of CDK9 or Cyclin T1 was used as a size control.

Figure 2: — Structure activity optimization identifies MC180295, a novel potent and selective CDK9 inhibitor. See also Figure S2 and Table S4.

(A) GFP expression measured by flow cytometry four days after single dose treatment with aminothiazole analogs in YB5. Corresponding structures are shown on top. Data are shown as mean+SD, N=3. *p<0.05, ***p < 0.001.

(B) Kinase phylogenetic tree showing the distribution of human kinases inhibited by 1μM MC180295.

(C) IC50 of MC180295 against 10 CDK/Cyclins.

(D) Quantification of Western Blots 2hr after MC180295 treatment against pSer2 (CDK9 target), phosphor-Rb at T870/811, phosphor-Rb at T826, p130 (CDK4/6 targets), phosphor-CDK Substrate Motif (K/H)pSP and phosphor-PRC1 (CDK1/2 targets). At 500nM (top), there is specific inhibition of CDK9 only while at 5μM, CDK9, 4 and 6 are inhibited. Data on other doses are in Figure S2D.

(E) In our model (*green*), the aminothiazole core of MC180295 engaged the CDK9 hinge region (left panel) with interactions that mimic that of dasatinib (shown here bound to cSrc, PDB ID 3G5D, *pink*). MC180295 engaged the conserved Lys48-Glu66 hydrogen bond (*green*) (middle panel); the multi-CDK inhibitor flavopiridol also made a similar interaction (PDB ID 3BLR, *pink*). The norbornyl group from MC180295 requires that the C-terminus of the hinge region adopts a slightly lower conformation; this conformation is shared amongst the many crystal structures of CDK9 (right panel) (*yellow:* structures of CDK9 bound to ATP and to another Type I inhibitor (PDB IDs 3BLQ and 4BCJ), and our model of MC180295), but this loop conformation is rarely observed in structures of other CDK kinases (*blue:* representative structures of CDK1/2/5/6/7, each bound to ATP or a Type I inhibitor (PDB IDs 5LQF/1HCK/1UNH/2EUF/1UA2)).

Chemically, aminothiazole compounds do not resemble known epigenetic drugs. To determine their relevant target(s), we first studied DNA methylation and found no demethylation at either CMV or LINE-1 (measured by bisulfite pyrosequencing - data not shown) or globally (measured by Reduced Representation Bisulfite Sequencing (RRBS)), indicating that they are not DNMT inhibitors (Figure S1E). We next used biochemical assays to analyze a panel of HDACs (class I, IIb and IV) and found no inhibitory activity (Figure S1F). We also screened HH1 against a panel of 30 histone methylases and demethylases and found no significant inhibition of enzymatic activity (Figure S1G and Table S3). We then measured global histone acetylation and methylation after 48hr treatment with HH1 and found no significant changes, except for a modest increase in H3K79 methylation, a mark of transcriptional elongation (Figure S1H). Thus, HH1 had no substantial activity against the main known regulators of epigenetic silencing.

We next used connectivity mapping (Lamb et al., 2006) which identifies drugs with similar transcriptional profiles. Using RNA-seq after HH1 treatment for 24hr, the closest drugs to HH1 were CDK inhibitors. To validate these unexpected targets, we tested 7 different CDK inhibitors and they all induced GFP in YB5 with a range of 5–15% after 24hr (Figure 1B). The fact that these drugs with diverse chemical structures (Figure 1B) are all active suggests that CDKs are indeed the relevant drug targets. We also tested the expression of SYNE1 and MGMT, two known genes hypermethylated in YB5 and found that HH1 and other CDK inhibitors (flavopiridol and iCDK9 (Lu et al., 2015)) led to gene reactivation after 24hr treatment (Figure 1C).

The tested drugs target many CDKs but the ones most effective at gene reactivation had the lowest IC50s against CDK9 (Figure S1I). To genetically validate CDK9 as a regulator of gene silencing, we used an inducible dominant negative CDK9 (dnCDK9). We found striking re-expression of GFP and of endogenously silenced genes upon induction of dnCDK9 (Figure 1D and Figure S1J). This effect could also be seen in HCT116-GFP cells (Figure S1K). By contrast, dnCDK1 or dnCDK2 showed no effects (Figure S1L). Conversely, activation of GFP and endogenously silenced genes by HH1 and other CDK inhibitors was completely prevented by overexpression of P-TEFb (CDK9 + Cyclin T1) (Figure 1E and Figure S1M). Collectively, these data strongly suggest that CDK9 is the target of the newly identified aminothiazole compounds and is therefore required to maintain transcriptional repression at epigenetically silenced loci.

Development of novel CDK9 inhibitors

We next tested newly synthesized HH1 analogs for gene expression-based Structure Activity Relationship (SAR) discovery. A time course qPCR showed that a one-dose 4-day treatment with 25μM HH1 induced the highest levels of GFP and hypermethylated genes in YB5 (Figure S3A). We tested 77 novel aminothiazole analogs using this approach and identified MC180295 as the most potent (active at 50nM and leading to ~60% GFP+ cells at 500nM) (Figure 2A), together with three analogs with similar activities (Figure 2A and S2A). We tested the selectivity of MC180295 against a panel of 250 kinases at 1μM and found it to be highly selective against CDKs within the human kinome, though glycogen synthase kinase 3 (GSK-3α and GSK-3β) was also inhibited (Figure 2B and Table S4). Two specific GSK-3 inhibitors (CHR99021 and LiCl) showed no GFP reactivation in YB5, suggesting that GSK-3 is not relevant to gene activation (Figure S2C). We then generated a dose-response curve for MC180295 against 10 different CDKs. The drug was most active against CDK9 (IC50=5nM) and was at least 22-fold more selective for CDK9 over other CDKs (Figure 2C and S2B). To confirm on-target selective CDK9 inhibition, we measured phosphorylation levels of Ser2 (pSer2) on the heptad repeats of the CTD of RNAPII, which is phosphorylated by P-TEFb, at different doses after 2hrs MC180295 treatment. Ser2 was dephosphorylated at nanomolar doses, while substrates of CDK4/6 (Calbó et al., 2002; Harbour et al., 1999) were only affected at substantially higher doses, and CDK1/2 substrates were not affected at all (Figure 2D and S2D). Similar results were seen after 8hrs of MC180295 treatment (Figure S2E). Thus, optimization of HH1 based on gene expression alone yielded a highly specific CDK9 inhibitor.

To understand MC180295 specificity, we built a model of the CDK9-MC180295 complex. From its planar aminothiazole core with alternating hydrogen bond donors and acceptors, we surmised that MC180295 would be an ATP-competitive inhibitor with a pattern of interactions matching other kinase inhibitors (Wang et al., 2014). Based on this, we elected to model the structure of MC180295 bound to CDK9 by analogy to other ligands in complex with CDKs. We first compiled a set of all 389 structures of CDKs with bound ligands available in the Protein Data Bank (PDB). As described in the STAR methods section, we structurally aligned lowenergy conformations of MC180295 to each of these ligands, and then replaced the kinase structure with that of CDK9. We refined these models and found that a single dominant cluster emerged among the top-scoring complexes (i.e. an identical binding mode in 14 of the top 30 models). We took this as the predicted pose for the MC180295-CDK9 complex and examined this model further.

In our model, the aminothiazole core makes canonical hydrogen bonding interactions to the CDK9 hinge region. Though it was not included among our templates, the interactions that emerge mimic closely those made by dasatinib, a broad-spectrum inhibitor that is also built on an aminothiazole core (Figure 2E). Like dasatinib, MC180295 is modeled in a “DFG-in” conformation (making it a Type I inhibitor), and its binding mode strongly resembles that of the CDK4/6 inhibitor palbociclib (Lu and Schulze-Gahmen, 2006). Whereas palbociclib hydrogen bonds directly to the backbone of the DFG motif, however, MC180295 uses a nitro group to engage the Lys-Glu salt bridge that is invariant in essentially all kinases; a similar interaction to this has also been observed in the structure of CDK9 bound to the multi-CDK inhibitor flavopiridol (alvocidib) (Baumli et al., 2008) (Figure 2E). The strong similarity in this part of the binding site among CDKs, coupled with the similarity of these interactions to those of other multi-CDK inhibitors, implied that the basis for CDK9-selectivity did not derive from this part of the compound.

On the opposite site of the MC180295 model, a norbornyl group fits on top of the C-terminus of the hinge region and occupies a shallow hydrophobic cleft. This region of the binding site is also occupied by many other inhibitors of CDKs; however, each of these uses individual (flat) ring structures instead of the bulkier norbornane. Careful comparison amongst all CDK crystal structures available in the PDB revealed a subtle difference: the C-terminal part of the CDK9 hinge region can adopt a slightly lower conformation that distinguishes it from the other CDKs where this loop has a higher conformation that sterically prevents the norbornyl group from being accommodated (Figure 2E). Thus, this model explains the observed SAR for analogs of MC180295 and this compound’s preference for CDK9 over other CDKs.

CDK9 inhibition leads epigenetic derepression genome-wide

To characterize CDK9 targets dynamically, we performed a time course qPCR on GFP and SYNE1. They could both be induced as early as 8hrs and their expression levels peaked four days after a single exposure (Figure S3A). To confirm that the gene induction is associated with on-target CDK9 inhibition, we measured the expression of MYC, a known P-TEFb target and found it to be suppressed at all these time points (Figure S3B). The timing of gene expression activation is likely related to the fact that chromatin remodeling at silenced loci requires time and cell division. However, it could also be related to selection of drug resistant survivor cells at 4 days. To exclude this, we measured drug sensitivity of survivor cells by culturing them drug-free for two weeks and re-exposing them to drugs. Rather than resistance, we observed that “survivors” of CDK9 inhibition were even more sensitive to re-treatment, indicating delayed sensitization, an effect typical of epigenetic drugs (Figure S3C).

We next performed time-course RNA-seq using HH1 (Figure S3D). Short term CDK9 inhibition (2hrs, 4hrs) led mainly to gene downregulation (e.g. 1242 genes down vs. 404 genes up at 2hr). Gene Ontology analysis showed that genes downregulated after 2hrs were enriched for regulation of transcription (Figure S3E). These started to recover by 4hrs and showed delayed upregulation as previously shown for P-TEFb targets (Garriga and Graña, 2014; Lu et al., 2015) (Figure S3F). In contrast to this early response, we observed massive gene upregulation (2981 up vs. 278 down) 4 days after first drug exposure. Even after excluding the P-TEFb targets identified earlier (downregulated at 2hrs/4hrs), we still observed upregulation of 2597 genes. This large effect (12.3% of the transcriptome) was consistent with the GFP data. Over the entire transcriptome, the effect of HH1 on gene upregulation was predominant at 4 days (mean induction of 1.4 fold), and the effect was most profound for silenced genes (mean induction of >4 fold) (Figure 3A and S3G). To better characterize this, we focused on the 1806 genes that showed low expression at baseline (RPKM < 0.31) and that were induced by HH1 (Figure 3B).

Most of these silenced/upregulated genes were highly hypermethylated in their promoters (Figure 3C). While expression peaked at four-days on average, they generally showed progressive induction, starting to be detectable at 4 to 8 hours after treatment (Figure 3D). Gene Ontology analysis showed that they were enriched for cell adhesion, a signature also seen for upregulated genes after treatment with the DNMT inhibitor decitabine (DAC) (Figure S3H). To compare gene reactivation by CDK9 vs. DNMT inhibition, we selected 10 hypermethylated genes reactivated by HH1 and validated their expression dynamics using MC180295, as compared to DAC. Eight out of 10 hypermethylated genes could be reactivated by MC180295 as early as 8 hours after drug treatment and their expression levels peaked after 4-day CDK9 inhibition. This reversal of gene silencing happened faster than could be achieved with DAC; for most of these, reactivation by DAC took at least 24hrs and also peaked at 4 days (Figure S3I).

The data above indicate that HH1 has a bimodal effect (one gene subset downregulated early and another upregulated late). To verify that this is due to CDK9 inhibition, we examined RNAseq after dnCDK9. Principal component analysis of the entire transcriptome showed that the baseline conditions (cells treated with DMSO at different time points and TET-on dnCDK9) clustered together and that there was a time-dependent progressive gene induction after HH1 treatment (Figure 3E). Strikingly, dnCDK9 overexpression clustered closest to four-day HH1 treatment, and there was a strong correlation between HH1 and dnCDK9 effects by RNA-seq (Figure 3F). Genes upregulated after either HH1 four-day treatment or dnCDK9 overexpression had very low baseline expression, consistent with the hypothesis that CDK9 is essential to maintain epigenetic silencing (Figure 3G).

Interestingly, the transcriptional profiles of DNMT inhibition (by DAC) clustered closely to 4-day HH1 (Figure 3E), and upstream analysis of genes induced by HH1 showed a strong enrichment for genes induced by DAC (Figure 3H). Because HH1 did not induce demethylation, we tested for synergy between HH1 and DAC. The two drugs were highly synergistic for GFP and SYNE1 reactivation (Figure S3A). The synergistic effects were validated using either siDNMT1 or siCDK9 (Figure S3J). We next performed RNA-seq using DAC in combination with HH1. After excluding the early response genes (downregulated at 2hr/4hr), we found that, compared to DAC alone (1238 up vs.7 down) or HH1 alone (1806 up vs. 13 down), silenced genes (RPKM < 0.31) were significantly upregulated by the combination (3948 up vs. 3 down) (Figure 3A). Thus, long-term CDK9 inhibition shows similar transcriptional profiles to DNMT inhibition and exhibits synergy with DNMT inhibition.

BRG1 is a direct phosphorylation-substrate of CDK9

In a search for the mechanism of CDK9 mediated repression, we examined Ingenuity Pathway Analysis (IPA) of the RNA-seq data on genes upregulated by HH1. SMARCA4, a member of the SWI/SNF family of genes was one of the top activated regulators (Figure 3H). SMARCA4, also known as BRG1, is an ATP-dependent helicase that is a central component of the SWI/SNF family and can use ATP hydrolysis to regulate chromatin structure (St Pierre and Kadoch, 2017). To test if CDK9 inhibition affects chromatin structure, we used ATAC-seq and examined cells before and after HH1 treatment (at 4-days). ATAC-seq data (done in triplicate) showed that both before and after HH1, there was a strong correlation between chromatin relaxation and gene expression (Figure S4A). When normalized using highly expressed genes that do not change after drug treatment, we found that HH1 results in striking global and diffuse relaxation of chromatin, suggesting that CDK9 is required for maintenance of heterochromatin compaction (Figure 4A). Examples of ATAC-seq peaks appearing after HH1 treatment are shown in Figure 4B. Consistent with this, we found higher occupancy of H3K4me2 at the promoter regions of hypermethylated CDK9 targeted genes after HH1 treatment (Figure S4B and S4C).

Figure 4: — CDK9 regulates BRG1 to de-repress silenced genes. See also Figure S4.

(A) Enrichment of ATAC-seq signal around transcription start sites (TSS). Data show merged triplicates. On the left is aggregated enrichment of all genes around TSS for cells treated with DMSO (blue) or HH1 (green). On the right is aggregated enrichment around TSS of genes that are induced by HH1 treatment (green) compared to DMSO (blue).

(B) Representative traces of ATAC-seq after HH1 treatment. SPOCK2 and CYP1B1 are significantly (FC>2) upregulated by HH1 and are methylated in their promoter regions. Arrows indicate peaks gained in promoter region. “Normalized HH1” refers to ATAC-seq reads after normalization based on invariantly expressed genes.

(C) (C–F) CDK9 immunoprecipitates with BRG1. In each panel, the immunoprecipitation antibody is shown on top while the Western Blot antibody is shown on the right. In (C), BRG1 immunoprecipitates with transfected FLAG-tagged CDK9 in HEK293T. In (D), (top) CDK9 immunoprecipitates with transfected FLAG-tagged BRG1 and (bottom) BRG1 immunoprecipitates with transfected GFP-tagged CDK9 in SW48.

(E) shows endogenous Co-IPs in HEK293T while (F) shows endogenous Co-IPs in SW48. IgG was used as a negative control for panels E and F. The IgG heavy chain (Hc) was also shown in panel F.

(G) Isotope kinase activity assay using recombinant active full-length CDK9 and BRG1 with or without CDK9 inhibitors (flavopiridol (FVP) and MC180295 (295)) in the presence of ³²γ-ATP. RNA-Pol-ll, CTD, BRG1 and CDK9 itself were all phosphorylated by CDK9 and unphosphorylated after CDK9 inhibition. Coomassie blue staining is shown on the bottom of the graph to verify equal loading.

(H) BRG1 overexpression overcomes gene silencing and synergizes with CDK9 inhibition. YB5 cells were transfected with different V5-tagged BRG1 constructs (WT (wild-type), 5STOA (five serine residues substituted by alanine residues) and NO5S (five serine residues were deleted)) for 48hr prior to drug treatment for 48hr. The number of GFP+ cells was measured using confocal microscopy. Data are shown as mean+SD, N=3. *p<0.05, **p < 0.01, ***p < 0.001.

Based on the above, we hypothesized that CDK9 regulates BRG1 by direct phosphorylation. Publicly available proteomic data showed possible binding of CDK9 to BRG1 (Rouillard et al., 2016). We used Co-immunoprecipitation (Co-IP) to confirm this. We first performed an IP using a FLAG-tagged CDK9 construct in HEK293T and successfully pulled down BRG1 (Figure 4C). Next, we overexpressed FLAG-tagged BRG1 in SW48, the parental cell line of YB5, and also successfully pulled down CDK9 (Figure 4D). We then overexpressed GFP-tagged CDK9 and successfully pulled down BRG1 (Figure 4D). To confirm these interactions, we performed reciprocal endogenous co-IPs in both HEK293T and YB5 (Figure 4E and 4F). IP of BRG1 successfully pulled down CDK9 along with other SWI/SNF complex members (Figure S4D). IP of CDK9 pulled down BRG1 along with its P-TEFb complex partner, CyclinT1 (Figure S4E).

We next performed an in vitro kinase assay using purified CDK9 and BRG1 proteins and found that CDK9 directly phosphorylates BRG1 in vitro (Figure 4G). To identify potential phosphorylation sites in BRG1 regulated by CDK9 inhibition, we performed LC-MS/MS using purified recombinant active CDK9 and BRG1 proteins in the presence or absence of drug treatment. In an experiment run in quadruplicate, we identified a series of five serine residues at the C-terminal domain of BRG1 as phosphorylation targets of CDK9 (Figure S4F). These residues are highly conserved across different species (Figure S4F) and are just downstream of the bromodomain of BRG1. Publicly available data showed that these serine residues could also be found phosphorylated in vivo in cancer (Hornbeck et al., 2015).

BRG1 is essential for gene reactivation by CDK9 inhibition

To confirm that BRG1 phosphorylation is relevant to gene activation by CDK9 inhibitors, we overexpressed BRG1 in YB5. BRG1 overexpression significantly enhanced the effects of CDK9 inhibitors on GFP reactivation (Figure 4H), a finding that was confirmed using different BRG1 expression vectors (Figure S4G, S4H, S4I and S4J). Using sensitive confocal microscopy, we found that BRG1 overexpression alone could activate GFP, albeit at low levels (Figure 4H). We generated a BRG1 expression vector where all 5 serine sites identified as phosphorylated by CDK9 were mutated to alanine to generate a de-phosphomimetic construct mirroring CDK9 inhibition. This de-phosphomimetic BRG1 alone could also trigger GFP induction (Figure 4H) and showed an even greater synergy when combined with MC180295 (Figure 4H). We then deleted these five serines and found that the truncated BRG1 behaved similarly to the dephosphomimetic construct (Figure 4H, S4I and S4J). To test whether BRG1 is required for gene activation, we inhibited BRG1 expression by siRNA prior to treatment with CDK9 inhibitors and found significantly reduced GFP induction (Figure S4K). We also used the BRG1 inhibitor PFI-3 (Vangamudi et al., 2015) and found that it inhibited GFP activation by CDK9 inhibitors in a dose-dependent manner (Figure S4L).

If the SWI/SNF complex is involved in gene reactivation, one would expect to see effects on chromatin. Indeed, as described earlier, CDK9 inhibition resulted in broad opening of chromatin as measured by ATAC-seq. Moreover, when we merged RNA-seq data with H3K9me2 ChIP-seq data in YB5, we found that the HH1-upregulated genes were highly enriched for H3K9me2 at baseline (Figure S4M). We validated this enrichment at the CMV-GFP locus and at the promoters of MGMT and SYNE1 by ChIP-qPCR (Figure S4N). Also, we showed earlier that CDK9 inhibition resulted in site-specific increases in H3K4me2, consistent with chromatin relaxation. Collectively, the data suggest a model whereby CDK9-mediated phosphorylation of BRG1 prevents it from being recruited to heterochromatin loci, while CDK9 inhibition allows BRG1 to remodel chromatin and alter gene expression.

CDK9 inhibition has anti-tumoral effects in vitro and in vivo.

Anti-tumoral effects of CDK9 inhibition have been attributed to MCL1 and/or MYC suppression, but these studies were based on drugs that inhibit multiple CDKs. For example, the “prototypical” CDK9 inhibitor flavopiridol also inhibits CDK1 and CDK4 (IC50 20nM for CDK9, 30nM for CDK1, 100nM for CDK4, etc.) (Asghar et al., 2015). We therefore tested whether more specific CDK9 inhibitors have anti-tumoral effects. We tested proliferation four days after one-time HH1 or MC180295 exposure. Compared to a normal lung fibroblast (IMR90), both HH1 and MC180295 were more effective in reducing proliferation of cancer cells (Figure 5A). Cell cycle analysis showed no cell cycle arrest after HH1 or MC180295 (Figure S5A), but an increase in the sub-G1 subpopulation (Figure S5B). We next tested colony formation in soft agar and found that a single dose pre-exposure of HH1 and MC180295 for four days can blunt colony formation by 30–80% in YB5 and HCT116 (Figure 5B and 5C). We also tested effects on the differentiation marker CD11b using the HL60 cell line and found significant induction (Figure 5D and S5C). Lastly, we tested in vivo effects of MC180295 and the structurally related CDK9 inhibitor, SNS032. We subcutaneously injected SW48 into NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice and treated them every other day (qod) for 10 weeks at 20mg/kg. We found that tumors grew slower after drug treatment and MC180295 improved mouse survival (Figure 5E) without causing overt toxicity as measured by body weight (Figure S5D). Similar data were obtained in an experiment where luciferase labeled SW48 were injected IP into NSG mice followed by MC180295 treatment (Figure S5F, S5G and S5H). Analysis of tumor tissues showed that GFP and a hypermethylated CDK9 target gene (GLD) were reactivated in-vivo after 2 weeks of drug exposure (Figure S5E). Lastly, we transplanted an ovarian cancer mouse cell line, ID8, into syngeneic immunocompetent mice and tested the anti-tumor efficacy of SNS-032. 10mg/kg SNS-032 given qod reduced tumor burden and extended survival (Figure 5F). In this cell line, we also saw reactivation of ovarian cancer-specific hypermethylated TSGs upon CDK9 inhibition (Figure S5I).

Figure 5: — *In vitro* and *in vivo* efficacy of CDK9 inhibition. See also Figure S5.

(A) Proliferation responses of a normal lung fibroblast (IMR90) and cancer cell lines treated with a single-dose of 5μM HH1 or 0.1μM MC180295 and counted four days after treatment by trypan blue exclusion.

(B) (C) Soft agar assays of SW48 (B) and HCT116 (C) cells following a single dose of HH1 or MC180295 for four days.

(D) HL60 differentiation measured by expression of CD11b (flow cytometry) after a single dose exposure to HH1 or MC180295 for four days. 1μM ATRA and high concentrations of DMSO (1.25%) were used as positive controls.

(E) Anti-tumoral effect of MC180295 *in vivo*. (Left) NSG mice were inoculated (s.c.) with 2×10⁶ SW48 cells. Eleven days later, when tumors were palpable, 20mg/kg MC180295 or vehicle was administered (i.p.) qod. Tumor sizes were measured using a caliper. Data are shown as mean+SEM (Student's t-test). (Right) Survival of the mice in days. Significance was calculated using a log-rank (Mantel-Cox) test.

(F) Efficacy of CDK9 inhibition in a syngeneic cancer model. Measurement of ascites fluid in the VEGF-DEF ID8 ovarian cancer mouse model is an indicator of tumor burden. *In vivo* treatment with the CDK9 inhibitor SNS-032 every 3 days demonstrated a decrease in tumor burden at weeks 4 and 5 (left). Addition of α-PD-1 led to a further decrease in tumor burden at week 5. Data are shown as mean+SEM (Mann Whitney test). Survival of NSG mice in (G) was calculated using a log-rank (Mantel-Cox) test. MC180295 significantly extended survival of the mice in this model (right). In all panels, *p<0.05, **p < 0.01, ***p < 0.001. Bar graphs represent mean+SD of at least biological triplicates.

CDK9 inhibition activates endogenous retroviruses and an Interferon response

DNMT inhibitors can trigger the IFNγ pathway within tumor cells, in part by activation of Endogenous Retroviruses (ERVs), leading to epigenetic immunosensitization (Chiappinelli et al., 2015; Roulois et al., 2015). RNA-seq after HH1 identified DNMTi and IFNγ signatures (Figure 3H). We therefore analyzed ATAC-seq and RNA-seq data after HH1 treatment. There was a significant increase in the number of ATAC-seq reads coming from repetitive elements after drug treatment (Figure 6A and S6A). RNA-seq data showed that multiple repetitive elements were induced (Figure 6A), including ERVs. We validated four of the ERVs by qPCR in SW48 (Figure 6B) and HCT116 (Figure S6B). HLA-A, HLA-B and HLA-C were all significantly upregulated upon CDK9 inhibition. PD-L1, an IFNγ inducer, was downregulated early after CDK9 inhibition but rebounded to baseline levels by 4 days (Figure S6C). Next, as previously shown for DNMTi (Chiappinelli et al., 2015), we identified 326 immune-related genes that could be activated by HH1 (Figure S6D). This CDK9 immune signature (CIM) was queried in the TCGA database (Network, 2015) and identified a subset of melanoma patients carrying high expression of CIM and significantly better outcomes (Figure 6C). A similar pattern was found in colon cancer (Network, 2012) (Figure S6E). We then queried RNA-seq data from 19 melanoma patients treated with anti-CTLA4 (Snyder et al., 2014) and found that the subset of cases with long-term benefit tended to have a higher expression level of CIM signature genes (Figure S6F). These data supported the hypothesis that CDK9 inhibition could sensitize to immune checkpoint inhibitors. We directly tested this in the ID8 mouse model and found that CDK9 inhibition could sensitize to anti-PD1 in vivo (Figure 5F). CDK9 inhibition increased the numbers of CD45+ immune cells and the percentages of CD3+ T cells and activated dendritic cells in the tumor environment, while combining CDK9 inhibition with checkpoint blockage further boosted the immune responses in vivo (Figure 6D and S6G). Finally, we injected human peripheral blood mononuclear cells directly into NSG mice followed by either 10 mg/kg or 20 mg/kg MC180295 treatment for 14 days. We found that MC180295 did not kill human T lymphocytes and did not affect the ratio of CD4 and CD8 T cells in vivo (Figure 6E and S6H). Thus, in addition to having single-agent anti-tumoral activity, CDK9 inhibition is a promising strategy for epigenetic immunosensitization.

Figure 6: — CDK9 inhibition triggers upregulation of endogenous retroviruses (ERV) and synergizes with immune checkpoint inhibitors *in vivo*. See also Figure S6.

(A) ATAC-seq reads mapping to repetitive elements enriched or depleted after 4-day HH1 treatment are shown on the left. There is a high preponderance of enrichment, consistent with broad chromatin decompaction. Differentially expressed repetitive elements as measured by RNA-seq after 4-day HH1 are shown on the right. The majority of these are activated repeats, consistent with the broad epigenetic effects of CDK9 inhibition.

(B) ERV activation measured by qPCR four-days after single dose CDK9 inhibitor treatment in YB5 cells (n=3). DAC was used as a positive control. Data are shown as mean+SD. *p<0.05, **p < 0.01, ***p < 0.001.

(C) CDK9 IMmune Signature (CIM) gene expression panel clustered TCGA melanoma patients into high and low immune signatures. CIM-high patients have a longer survival than CIMlow patients.

(D) *In vivo* treatment of mice with SNS-032 resulted in increased populations of immune cells (CD45+) and T cells (CD3+) in the tumor microenvironment. *p<0.05, **p < 0.01, ***p < 0.00).

(E) MC180295 did not inhibit human T-cell growth *in vivo*. 20 million human PBMCs from a healthy donor were injected i.p. on day 0 into NSG mice. 20mg/kg MC180295 was injected (i.p.) qod for 12 days and whole blood was collected on day 14. Flow cytometry was performed using anti-CD45, anti-CD4 and anti-CD8 antibodies.

DISCUSSION:

Epigenetic mediators of gene silencing are validated cancer targets (Jones et al., 2016). Here, we report on a novel epigenetic silencing target - CDK9 - and a new inhibitor that shows nanomolar potency and has 20-fold selectivity for CDK9 compared to other CDKs. Reactivation of silenced genes requires cell division to help reset chromatin (Cui et al., 2014); thus, optimization using gene expression also selects against drugs that inhibit the cell cycle. This may explain why the drug discovered shows high selectivity for CDK9 compared to cell cycle regulating CDKs. Our model of the CDK9-MC180295 complex suggests that this inhibitor achieves selectivity through the norbornyl group, by taking advantage of a subtle structural variation in the active site. This general principle has emerged as a paradigm to explain the selectivity of other selective Type I inhibitors as well (Wang et al., 2014).

This is the first time to our knowledge that CDK9 is linked to gene silencing in mammals. The data pointing to CDK9’s role in epigenetic silencing are based on the effects of pharmacologic (exquisitely specific drugs) and genetic manipulation (dominant negative) of CDK9, as well as overexpression of CDK9 which blocks the effects of the drugs on gene silencing. In terms of timing and targets, this gene induction pattern is broadly similar to what is seen with DNMTi, and the synergy with DNMTi suggests potential pathways for clinical development. CDK9 inhibition reactivates genes by remodeling chromatin but without affecting DNA methylation, as previously seen with HDAC inhibition (Raynal et al., 2012).

CDK9 mediated phosphorylation of SWI/SNF complex components, including BRG1, has previously been reported in HIV-1 infected T-cells and phosphorylation of the SWI/SNF component Baf53 can lead to its release from DNA (Van Duyne et al., 2011). We hypothesize that CDK9 mediated phosphorylation of BRG1 can lead to its release from chromatin. The specific phosphorylation sites we identified are near the bromodomain of BRG1 and mutations or deletion of these sites enhances the effects of BRG1 on gene activation, suggesting that phosphorylation of this domain keeps the BAF complex away from heterochromatin. We propose that CDK9 inhibition triggers a cascade of events that start with allowing BRG1 to remodel chromatin, followed by nucleosomal sliding, access to activating histone modifiers and ultimately access to transcription factors that result in gene activation. Given that BRG1 overexpression alone only modestly activates gene expression, it is likely that CDK9 inhibition also has other effects on the gene silencing machinery. It is interesting to consider why CDK9 evolved to simultaneously maintain high-level gene expression (at super-enhancer driven loci) and gene silencing (at heterochromatin loci). Rapidly cycling cells have the potential for transcription “errors” due to the need to remodel chromatin for DNA replication, and it is likely that proteins such as CDK9 evolved to ensure fidelity of expression (both high and low) at newly replicated loci. In this context, it will be interesting to determine whether other CDKs are also invovled in gene silencing.

CDK targeting has been an active area of research in oncology. CDK4/6 inhibitors were developed to treat Cyclin D-dependent cancers and were approved by the FDA for treating ER-positive and HER2-negative breast cancer (Sherr et al., 2016). A recent report on immune effects of CDK4/6 inhibition (Goel et al., 2017) may explain their activity as combination therapeutics, and it is broadly similar to what we observed here with CDK9 inhibition. In fact, the CDK4/6 inhibitor used in these studies also inhibits CDK9 (Sherr et al., 2016), suggesting the possibility that some of the effects seen were also due to CDK9 inhibition. CDK7/9 targeting has previously been proposed as a strategy to suppress the expression of super-enhancer driven oncogenes (e.g. MYC) (Wang et al., 2015). Moreover, inhibition of P-TEFb has been shown to downregulate the expression of EMT transcription factors (Slug, FOXC2, ZEB2, and Twist1), thus delaying tumor progression (Ji et al., 2014). Our new data show that CDK9 targeting reactivates tumor-suppressor genes and induces a cellular immune response that may sensitize to checkpoint inhibitors. Thus, CDK9, which is overexpressed in many cancers (based on cBioPortal) (Cerami et al., 2012; Gao et al., 2013), has multiple properties that make it an excellent target for drug development in cancer.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Jean-Pierre Issa (jpissa@temple.edu ).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell lines and cell culture

YB5 cell line was derived from SW48 (female) colon cancer cell line in our lab (Si et al., 2010). SW48/YB5 cells were maintained in L-15 (Corning) supplemented with 10% FBS (Atlanta Biologicals) and 1% penicillin/streptomycin (Corning) at 1% CO₂ in 37°C. MCF7-GFP cell line was derived from MCF7 (female) breast cancer cell line in our lab. HCT116-GFP (male) colon cancer cell line (Cui et al., 2014) was provided by Dr. Stephen Baylin (Johns Hopkins University). Prostate cancer cell lines DU145 (male) and LnCaP (male), a normal fibroblast cell line IMR90 (female), HEK293T (female) and leukemia cell lines KG-1 (male) and HL-60 (female) were obtained from ATCC. HEK293T, HCT116/HCT116-GFP, MCF7/MCF7-GFP, IMR90, LnCaP and DU145 cells were cultured in DMEM (Corning), McCoy’s 5A (Corning), DMEM (Corning), MEM (Corning), RPMI-1640 (Corning) and MEM (Corning) respectively, with 10% FBS and 1% penicillin/streptomycin at 37°C in 5% CO ₂. KG-1 and HL-60 were cultured in IMDM supplemented with 20% FBS at 37°C in 5% CO ₂. Mouse ovarian cancer ID8 (female) cells were provided by Dr. Cynthia Zahnow (Johns Hopkins University) and grown in RPMI 1640, 10% FBS and gentamicin sulfate (5μg/μL) (Corning) at 5% CO₂ in 37°C.

NSG mice treated with MC180295

SW48-luc cell line was generated by transfecting pFUGW-FerH-ffluc2-eGFP into SW48 cells. GFP positive cells were sorted out one week after transfection and expanded for the in vivo experiments. Both female and male NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice were used for the experiment. Age and gender matched mice were randomly assigned to each group.

Experimental protocols were approved by the Temple University’s Committee on Use and Care of Animals. 8–10-week-old NSG mice were inoculated (i.p.) with 1×10⁵ SW48-luc cells. One week later, at which time substantial tumor burden was evident by bioluminescence imaging, mice were randomized and 5–20mg/kg MC180295 or drug solvent was administered (i.p.) every other day. 200uL of diluted Pierce™ D-Luciferin, Monosodium Salt (Thermo Fisher Scientific) (working concentration: 15mg/mL) was administered (i.p.) into each mouse and was imaged using IVIS imaging system 5 minutes after the administration. Three vehicle treated mice and five drug treated mice were used in this study. Natural death or accumulation of ascites fluid was used as the endpoint for this model.

For the s.c. SW48 model, 2×10⁶ cells were injected into the flank of randomly assigned gendermatched 8–10-week-old NSG mice. 20mg/kg MC180295 was injected (i.p.) every other day when the tumors were palpable 9 days after engraftment. Seven mice were used in each group for this experiment and they were sacrificed before tumors volumes exceeded 400 mm².

MC180295 was dissolved in NMP (Fisher Scientific), Captisol (20% w/v) (CyDex), PEG-400 (Millipore Sigma) and normal saline (PBS) (Corning) in a ratio of 1:4:4:11. NMP first, followed by Captisol and PEG-400. PBS was added last.

Humanized PBMC NSG mice treated with MC180295

Human PBMCs were isolated from blood samples collected from healthy adult donors (two donors total) under an institutional review board (IRB) approved protocol (Temple University). 20 million human PBMCs were injected (i.p.) on day 0 into 8–10-week-old female randomly assigned NSG mice. 20mg/kg MC180295 were injected (i.p.) every other day for 12 days and whole blood was collected on day 14 followed by flow cytometry analysis.

Pharmacodynamics study in NSG mice treated with MC180295

Six gender-matched 8–10-week-old NSG mice were randomly assigned to two groups (Vehicle vs. Drug treatment). 2×10⁶ cells were injected into the flank of NSG mice, 20mg/kg MC180295 was injected (i.p.) every other day when the tumors were palpable 9 days after engraftment for two weeks. Tumors were then collected, homogenized (FastPrep-24 homogenizer) and total mRNA was extracted followed by qPCR.

Mouse experiments with in vivo treatment of SNS-032

2.5×10⁵ ID8-VEGF-Defensin cells were injected i.p. into 7–8-week-old randomly assigned female C57BL/6NHsd (C57Bl/6) mice. All animal experiments were approved by the Johns Hopkins Animal Care and Use Committee. All animal care and protocols followed were in accordance with guidelines of the institutional Animal Care and Use Committee (IACUC). Three days after injection, 1 mg/kg SNS-032 (Selleckchem), 10 mg/kg SNS-032, or 5% DMSO in PBS (vehicle control) was administered i.p. every 3 days for the duration of the experiment.α-PD-1 (200ug/mouse) or IgG control (Leinco Technologies) were given on days 17, 20, 24, and 27 after injection.α-PD-1 (1mg/mL in saline) was kindly provided by Dr. Michael Lim of the SKCCC, Johns Hopkins University. Mouse IgG isotype control was diluted in PBS.

METHOD DETAILS

NDL-3040 library screening and drug treatments

NDL-3040 compound library consists of 3040 chemically diverse compounds that are seminatural, derived from natural compounds, or synthetic compounds that are natural-compoundlike.The small molecules were arrayed in 96-well plates as 10mM stocks in 100% DMSO and were purchased from TimTec Inc. An aminothiazole analog library (93 small molecules) was also purchased from the same vendor and was in a 96-well plate format as 10mM stocks in 100% DMSO. The 3040 compounds were screened at 25μM for 24hr. All plates were kept at -80°C before use.YB5 cells growing in log phase (70–80% confluency) in 96-well plates were used. Each experimental 96-well plate contained 80 different compounds. A negative control (DMSO) and a positive control (5μM TSA) were placed at the edges as shown in Figure S1A. Compounds were dispensed using an INTEGRA VIAFLO96 96-well pipette. After a 24hr drug treatment, cells were trypsinized for 10 mins and re-suspended in L-15 medium containing propidium iodide (PI) to stain for dead cells. A total of 10000 cells per well were analyzed using Millipore Guava flow cytometer (EMD, Millipore). GFP positive percentage was calculated by excluding all the PI positive cells. After finishing the screening, the average Z-factor $(Z-factor = 1 - \frac{3 (σ_{p} + σ_{n})}{| μ_{p} - μ_{n} |})$ was calculated to test the robustness of the assay (the means (μ) and standard deviations (σ) of both the positive control (TSA) and negative controls (DMSO) (μ_p, σ_p and μ_n σ_n)). For a single-dose, four-day treatment schedule, different drugs were added 24 hours after cell seeding, drug-free medium was replenished three days later, and downstream experiments were performed on the following day. For daily treatment schedules, drug-free media were changed every day before supplementing new drugs. All drugs were originally in 100% DMSO stocks. The final concentration of DMSO in drug-treated cultures was 0.5%. Eighteen compounds identified as positive hits by the screening were purchased in powder form from TimTec Inc for validation. Several multi-CDK inhibitors (alsterpaullone (Sigma-Aldrich), GW8510 (Sigma-Aldrich), roscovitine (Millipore Sigma), RGB286147 (Millipore Sigma), dinaciclib (Selleckchem), SNS-032 (Selleckchem), iCDK9 (Chemscene, LLC) and flavopiridol (Santa Cruz Biotechnology) were also purchased and used in this project. Other chemicals used include PFI-3 (Selleckchem), CHIR99021 (Cayman), trichostatin A (TSA) (Sigma-Aldrich), SAHA (Sigma-Aldrich), depsipeptide (Sigma-Aldrich), valproic acid (Sigma-Aldrich), decitabine (DAC, Sigma-Aldrich) and tretinoin (ATRA, Selleckchem). All the compounds above were dissolved in 100% DMSO at 10mM stock concentration except for ATRA (dissolved in ethanol) and DAC (dissolved in water).

DNA extraction and DNA methylation analysis

DNA extraction, bisulfite conversion, and pyrosequencing were described and all primers were listed previously (Si et al., 2010). In brief, cell pellets were lysed in 600 microliters of the cell lysis solution (2% sodium dodecyl sulphate, 25 mM EDTA, pH 8.0). Two microliters of proteinase K (20 mg/ml) were added and the lysate were incubated at 55°C for 1 hour. One microliter of RNase A (100 mg/ml) was added afterwards and the lysate was incubated at 37°C for 15 minutes. Proteins were precipitated by adding 200 microliters of 10 M ammonium acetate. The tubes were cooled on ice and centrifuged at 12,000 rpm for 15 minutes. Supernatant containing DNA was poured in a tube containing 600 microliters of isopropanol. DNA was precipitated by inversion of the tube and brief vortexing. Precipitated DNA was pelleted by centrifugation, washed with 70% ethanol, and dissolved in 100 microliters of TE (TRIS 10 mM, EDTA 1 mM, pH 8.0). After DNA extraction, one microgram of genomic DNA was treated with bisulfite using the Epitect kit (QIAGEN). Bisulfite-treated DNA (40–80 ng) was amplified with gene-specific primers in a 2-step polymerase chain reaction (PCR). The second step of PCR was used to label single DNA strand with biotin using a universal primer tag. We measured levels of DNA methylation as the percentage of bisulfite-resistant cytosines at CpG sites by pyrosequencing with the PSQ HS 96 Pyrosequencing System (QIAGEN) and Pyro Gold CDT Reagents (QIAGEN). We found high concordance in methylation between adjacent CpG sites and therefore used mean values from all pyrosequenced CpG sites as a measure of methylation of a given gene.

siRNA Knockdown

ON-TARGETplus Non-targeting siRNA (siN) (D-001810–10), SMARTpool siSMARCA4 (L010431–00-0005) and CDK9 (L-003243–00-0010) were ordered from GE Dharmacon and diluted in water. Transfection was performed using Lipofectamine® RNAiMAX Reagent (ThermoFisher Scientific) according to the manufacturer’s instructions at a 20nM final working concentration.

Plasmids transfection and viral transduction

Cells were transfected with Cdc2-DN-HA, CDK2-DN-HA, Rc-dnCDK9, pCMV5 FLAG-BRG1, GFP-CDK9 and FLAG-CDK9 plasmids for 72hr to overexpress dnCDK1, dnCDK2, dnCK9, BRG1, and CDK9 using Lipofectamine 2000 (ThermoFisher Scientific) according to the manufacturer’s instructions. V5-tagged wild-type BRG1, 5STOA-BRG1 and NO5S-BRG1 constructs were transfected using Lipofectamine 2000 (ThermoFisher Scientific) according to the manufacturer’s instructions for 48hr prior to drug treatment for another 48hr.YB5 and HCT116-GFP cells were infected with Ad-T-dnCDK9 plus Adeno-X™ Tet-Off™ adenoviruses (dnCDK9) in the presence or absence of doxycycline (tet) as previously described (Garriga et al., 2010) for 72hr before processing for analysis. Wild type Ad-CyclinT1 and Ad-CDK9 were also transduced for 72hr to overexpress CyclinT1 and CDK9 (Garriga et al., 2003). Cdc2-DN-HA (#1889), CDK2-DN-HA (#1885), and pCMV5 BRGI-Flag (#19143) plasmids were purchased from Addgene. GFP-CDK9 and FLAG-CDK9 plasmid were a generous gift from Bassel E. Sawaya, Temple University. V5-BRG1 construct was a generous gift from Cigall Kadoch, DanaFarber Cancer Institute.

Biochemistry assays

HDAC inhibitory activity assays were performed using FLUOR DE LYS® HDAC fluorometric activity assay kit (ENZO) following the manufacturer’s instructions. GloMax®-Multi Detection System (Promega) was used to read the fluorescence signals. Histone methyltransferase and demethylase enzymatic assays were performed by BPS Bioscience at 10μM in duplicates (Table S3). Briefly speaking, all assays were performed in a 50μl mixture containing proper methyltransferase assay buffer, S-adenoslymethionine (BPS number 52120), enzyme (HMTs or HDMs), and the test compound (either HH0 or HH1 at 10μM) at room temperature in a substrate pre-coated plate. After enzymatic reactions for 60–960 minutes, reaction mixtures were discarded, wells were washed three times with Tris Buffered-saline Tween (TBST) buffer, and slowly shaken with Blocking Buffer (BPS catalog number 52100) for 10 minutes. Then, wells were emptied and 100 μl of diluted primary antibody was added. The plate was then slowly shaken for 60 minutes at room temperature, washed three times using TBST, and shaken with Blocking Buffer for 10 minutes at room temperature. After discarding the Blocking Buffer, 100 μl of diluted secondary antibody was added and the plate was shaken for 30 minutes at room temperature. Then, the plate was washed three times using TBST, and shaken with Blocking Buffer for 10 minutes at room temperature. Blocking Buffer was discarded and 100 μl of freshly prepared mixture of the HRP chemiluminescent substrates was added to each empty well. Immediately, sample luminescence was measured using a BioTek SynergyTM 2 microplate reader. The percent activity in the presence of each compound was calculated according to the following equation: % activity = (C-C0)/(Ce-C0), where C= the luminescence or A-screen intensity in the presence of the compound. The intensity (Ce) in each data set was defined as 100% activity in the absence of the compound. In the absence of enzyme, the intensity (C0) in each data set was defined as 0% activity. Kinase enzymatic assays were performed by Nanosyn using microfluidic technology. Briefly speaking, the in vitro kinase assay was performed using the microfluidic mobility shift platform of PerkinElmer. The target kinase was incubated with fluorescently labeled substrate and MC180295 in a standardized reaction mixture in a 384 well plate. Upon termination of the enzymatic reaction, samples were loaded onto microfluidic chips. Samples then migrated through channels and product and substrates were separated based on the difference in their charge. Enzyme activity was determined by comparison of the fluorescence in the product and substrate peaks. 250 kinome screening was done in duplicates using MC180295 at 1μM (Table S4). IC50 curves against 10 CDKs were created for MC180295. The human kinome tree was annotated using the online Kinome Render software (Chartier et al., 2013). Isotope kinase assay was performed as previously described (Garriga et al., 1998) using recombinant active full-length CDK9/CyclinT1 (Thermo Fisher Scientific) and BRG1 (Abcam) in the presence or absence of CDK9 inhibitors. Briefly speaking, 1.2μg of purified CDK9/Cyclin T1 (Thermo Fisher Scientific) was pre-incubated in the presence or absence of either 1μM of MC180295 or 1μM of FVP for 15min. Then, 0.8 μg of purified BRG1 (Abcam) was added and incubated at 30 ºC for 1hr in KAS buffer (with 5μCi/μl of ³²γ-ATP). The reaction was stopped by heating the samples at 65 ºC for 5min.

Synthesis of MC180295

(4-amino-2-(((2S)-bicyclo[2.2.1]heptan-2-yl)amino)thiazol-5-yl)(2-nitrophenyl)methanone

graphic file with name nihms-1510720-f0007.jpg

Step 1

(4-amino-2-(methylthio)thiazol-5-yl)(2-nitrophenyl)methanone

graphic file with name nihms-1510720-f0008.jpg

2-Bromo-2’-nitroacetophenone (1.005 mmol; 245 mg) and triethylamine (1.296 mmol; 180 μl) were added sequentially to a solution of cyanimidodithiocarbonic acid S-methyl ester Spotassium salt (0.902 mmol; 154 mg) in anhydrous dimethylformamide (4.0 ml). This mixture was stirred at 80^oC for 3 hours. It was cooled to room temperature and concentrated down. The residue was partitioned between ethyl acetate and water. The insoluble solids suspended between the organic and aqueous layers were filtered off and washed with ethyl acetate to afford the titled compound as a yellow solid. ¹H NMR (400 MHz, DMSO) δ 8.13 (dd, J = 8.08 Hz, J = 0.92 Hz, 1H), 7.97 (bs, 2H), 7.84 (td, J = 7.48 Hz, J = 1.08 Hz, 1H), 7.74 (td J = 8.08 Hz, J = 1.44 Hz, 1H), 7.68 (dd, J = 7.48 Hz, J = 1.36 Hz, 1H), 2.62 (s, 3H); ESIMS: m/z 296.0 [(M+H)⁺].

Step 2

(4-amino-2-(((2S)-bicyclo[2.2.1]heptan-2-yl)amino)thiazol-5-yl)(2-nitrophenyl)methanone

graphic file with name nihms-1510720-f0009.jpg

A solution of (4-amino-2-(methylthio)thiazol-5-yl)(2-nitrophenyl)methanone (0.1693 mmol; 50 mg) and exo-2-aminonorbornane (3.386 mmol; 401 μl) in ethanol (2 ml) was stirred at 100^oC in a glass pressure vessel overnight. This solution was cooled to room temperature and concentrated down. The crude product was purified by flash column chromatography on silica gel using a gradient solvent system of 0 to 100% of ethyl acetate in hexanes to afford the titled compound as an orange glassy solid. ¹H NMR (400 MHz, CDCl₃) δ 8.08 (d, J = 7.84 Hz, 1H), 7.67 (t, J = 7.52 Hz, 1H), 7.56 (t, J = 7.68 Hz, 2H), 5.65 (bd, J = 6.12 Hz, 1H), 3.17 (bs, 1H),2.31 (bs, 2H), 1.82 (m, 1H), 1.49 (m, 3H), 1.32 (m, 1H), 1.25 (m, 1H), 1.11 (m, 2H); ESIMS: m/z 359.0 [(M+H)⁺].

Flow cytometry

For drug screening and dose response validations, GFP positive cells were detected by Millipore Guava flow cytometer (EMD, Millipore). Cell cycle analysis was performed using BD FACSCalibur™ by propidium iodide staining four days after drug treatment. Sub-G1 population percentage was also included to measure apoptotic cell proportion. Data were analyzed using FlowJo software version 10.2. For the cell differentiation analysis, cells were washed and stained with propidium iodide (PI), CD11b (BD Biosciences) and the isotype control IgG (BD Biosciences). Flow cytometry analysis was performed on a Millipore Guava flow cytometer (EMD, Millipore). For the ID8 in vivo experiments, ascites was drained from 5–10 mice per group and incubated in ACK buffer (Thermo Fisher) to lyse red blood cells for 10 minutes, then washed. Ascites from each mouse was individually lysed and prepared for flow cytometry. Mononuclear cells collected were cultured for 4 hours in RPMI with 5% Fetal Bovine Serum and in the presence of Cell Stimulation Cocktail (plus protein transport inhibitors; eBioscience). Cells were then washed and stained for cell surface markers including Live/Dead (eBioscience), CD45 (BD Biosciences), CD3 (BD Biosciences), MHC II (Pacific Blue™ anti-mouse I-A/I-E Antibody, Isotype Control: Pacific Blue™ Rat IgG2b, κ Isotype Ctrl Antibody), CD80 (BD Biosciences), CD86 (BD Biosciences) and CD11c (BD Biosciences). Flow cytometry acquisition was performed on an LSRII cytometer (BD Biosciences) and data were analyzed using FlowJo software version 10.2. For the PBMC in vivo experiments, whole blood was collected from healthy donors and processed using Ficoll-Paque PLUS (GE Healthcare) following the manufacture’s protocol. Fourteen days after injection, blood was collected from tail vein, lysed using ACK lysing buffer and prepared for flow cytometry. Cells were stained for cell surface markers including CD45 (Biolegend), CD4 (Biolegend) and CD8 (Biolegend). Flow cytometry acquisition was performed on an LSRII cytometer (BD Biosciences) and data were analyzed using FlowJo software version 9.9.4.

Nuclear Extract Preparation

Cells were homogenized in Buffer A (25 mM HEPES (pH 7.6), 25 mM KCl, 0.05 mM EDTA, 10% glycerol, 5mM MgCl₂, 0.1% NP-40, supplemented with fresh 1mM DTT, protease inhibitors [Roche], and 1mM PMSF) on ice. Nuclei were sedimented by centrifugation (1,200 rpm), resuspended in Buffer C (10 mM HEPES (pH 7.6), 3 mM MgCl₂, 100 mM KCl, 0.1 mM EDTA, 10% glycerol, 1 mM DTT and protease inhibitors), and lysed by the addition of ammonium sulfate to a final concentration of 300 mg/mL. Soluble nuclear proteins were separated by ultracentrifugation (100,000 × g) and precipitated with 0.3 mg/ml ammonium sulfate for 20 min on ice. Protein precipitate was isolated by ultracentrifugation (100,000 × g) and resuspended in IP buffer (300 mM NaCl, 25 mM HEPES [pH 8.0], 0.1% Tween-20, 10% Glycerol, 1 mM DTT, 1 mM PMSF with protease inhibitors).

Co-immunoprecipitation

Endogenous Co-IP experiments were performed in HEK293T and SW48 cells using BRG1 (Santa Cruz, #sc-17796), CDK9 (Santa Cruz, #sc-484), BAF155 (Cell Signaling, #11956), CyclinT1 (Santa Cruz, #sc-10750), BAF60a (Santa Cruz, #sc-135843) antibodies.Exogenous co-IP studies were performed in HEK293T and SW48 cells using BRG1 (Santa Cruz, #sc17796), CDK9 (Santa Cruz, #sc-484), FLAG-M2 (Sigma, #F-1804) and GFP (NOVUS, #NB-100) antibodies. HEK293T cells were transiently transfected with either an empty vector (EV) or a FLAG-tagged CDK9. SW48 cells were transiently transfected with an empty vector (EV), a FLAG-tagged BRG1 or a GFP-tagged CDK9.

Nuclear extracts were resuspended in IP buffer and placed in protein low-bind tubes (Eppendorf). We quantified protein by standard BCA protocol (Pierce-23225, Thermo Scientific). Cell extracts were incubated with the corresponding IP antibodies. Separate lysate tube was prepared from samples for incubation with species- matched normal mouse IgG. Samples were incubated overnight with gentle rotation at 4°C. Pr otein G Dynabeads (Life Technologies) were incubated with the antigen-antibody complex for 2.5 hours the following day. Beads were washed four times with lysis buffer with gentle agitation for 5 minutes per wash. 2x Laemmli sample buffer (Bio-Rad) was used for elution of complex from beads followed by Western blotting along with the nuclear extract as inputs.

Generation of 5STOA and NO5S BRG1 constructs

The recombinant wild type BRG1 construct was generated using a plasmid PLEX 307 containing the open reading frame of the human SMARCA4/BRG1 gene (V5-BRG1). A mutant lacking the amino acids 1570–1631 (Called NO5S) and a mutant with substitutions of serine by alanine in positions 1570, 1575, 1586, 1627, 1631 (Called 5STOA) were used in the paper. The amino acid positions are based on the Uniprot (www.uniprot.org) canonical notation sequence # P51532–1.

For generating the mutants, the PLEX307 plasmid containing the SMARCA4 gene (V5-BRG1) was co-digested with FspAI and Eco32I (Thermofisher) and the mutations were inserted by ligating synthetic constructs (gblocks by IDT DNA Technologies) containing the desire mutations using the Gibson assembly method (Gibson et al., 2009). For this, three gblocks were designed as follows. A common gblock for the two mutants with the sequences below:

5’GGCACGAGGAGGAGTTTGATCTGTTCATGCGCATGGACCTGGACCGCAGGCGCGAGGA GGCCCGCAACCCCAAGCGGAAGCCGCGCCTCATGGAGGAGGACGAGCTCCCCTCGTGGA TCATCAAGGACGACGCGGAGGTGGAGCGGCTGACCTGTGAGGAGGAGGAGGAGAAGATG TTCGGCCGTGGCTCCCGCCACCGCAAGGAGGTGGACTACAGCGACTCACTGACGGAGAA GCAGTGGCTCAAGACCCTGAAGGCCATCGAGGAGGGCACGCTGGAGGAGATCGAAGAGG AGGTCCGGCAGAAGAAATCATCACGGAAGCGCAAGCGAGACAGCGACGCCGGCTCCTCC ACCCCGACCACCAGCACCCGCAGCCGCGACAAGGACGACGAGAGCAAGAAGCAGAAGAA GCGCGGGCGGCCGCCTGCCGAGAAACTCTCCCCTAACCCACCCAACCTCACCAAGAAGA TGAAGAAGATTGTGGATGCCGTGATCAAG-3’

A gblock for the mutant NO5S (D1570–1631):

5’ATGAAGAAGATTGTGGATGCCGTGATCAAGTACAAGGACAGCAGCAGTGGACGTCAGCT CAGCGAGGTCTTCATCCAGCTGCCCTCGCGAAAGGAGCTGCCCGAGTACTACGAGCTCAT CCGCAAGCCCGTGGACTTCAAGAAGATAAAGGAGCGCATTCGCAACCACAAGTACCGCAG CCTCAACGACCTAGAGAAGGACGTCATGCTCCTGTGCCAGAACGCACAGACCTTCAACCT GGAGGGCTCCCTGATCTATGAAGACTCCATCGTCTTGCAGTCGGTCTTCACCAGCGTGCG GCAGAAAATCGAGAAGGAGGATGACGAGGAGGAACAAGAGGAGGACCGCTCAGGAAGTG GCAGCGAAGAAGACAAGGGTGGGCGCGCCGACCCAGCTTTCTTGTACAAAGTGGTTGATA TCGGTAAGCCTATCCCTAACCCTCTCCTC-3’

A gblock for the substitution of serine by alanine in positions 1570, 1575, 1586, 1627, 1631:

5’ATGAAGAAGATTGTGGATGCCGTGATCAAGTACAAGGACAGCAGCAGTGGACGTCAGCT CAGCGAGGTCTTCATCCAGCTGCCCTCGCGAAAGGAGCTGCCCGAGTACTACGAGCTCAT CCGCAAGCCCGTGGACTTCAAGAAGATAAAGGAGCGCATTCGCAACCACAAGTACCGCAG CCTCAACGACCTAGAGAAGGACGTCATGCTCCTGTGCCAGAACGCACAGACCTTCAACCT GGAGGGCTCCCTGATCTATGAAGACTCCATCGTCTTGCAGTCGGTCTTCACCAGCGTGCG GCAGAAAATCGAGAAGGAGGATGACGCAGAAGGCGAGGAGGCAGAGGAGGAGGAAGAG GGCGAGGAGGAAGGCGCAGAATCCGAATCTCGGTCCGTCAAAGTGAAGATCAAGCTTGG CCGGAAGGAGAAGGCACAGGACCGGCTGAAGGGCGGCCGGCGGCGGCCGAGCCGAGG GTCCCGAGCCAAGCCGGTCGTGGCAGACGACGACGCAGAGGAGGAACAAGAGGAGGAC CGCTCAGGAAGTGGCAGCGAAGAAGACAAGGGTGGGCGCGCCGACCCAGCTTTCTTGTA CAAAGTGGTTGATATCGGTAAGCCTATCCCTAACCCTCTCCTC-3’

For the Gibson assembly of each mutant, 50 ng of digested vector were ligated with 37.5 ng of the common gblock and 37.5ng of the gblock of each mutant by using the Nebuilder HIFI DNA assembly master mix (New England Biolabs) following the manufacturer’s instructions. After 1 hour of incubation at 50 deg celcius, ~3 μl of the reaction mixture was used to transform STBL3 competent cells (Thermo Fisher) and selected in carbenicillin-agar plates. Overexpression of these plasmids was verified by western blot using an anti-V5 antibody (Thermo Fisher).

Cell microscopy and High Content Imaging Analysis

YB5 cells were seeded in 96 well plates for high content imaging (Cell Carrier Ultra by Perkin Elmer) and each well was individually transfected with 100 ng of the plasmids using Lipofectamine 2000 following the manufacturer’s instructions. 48 hours after transfection, YB5 cells were treated with either DMSO, 500nM MC180295 or 500nM iCDK9 for an additional 48 hours before being processed for cell microscopy. For the microscopy analysis, Hoechst 33342 was added into every well at a final concentration of 1 mg/ml, followed by incubation at 37 degrees Celsius for 20 minutes. Microscopic image acquisition was then performed using a confocal Operetta CLS-Live High Content Imaging System (Perkin Elmer – model # HH16000000) equipped with temperature and CO₂ environment control for live cells. The images were acquired with a 20X water immersion objective by using the following combination filter settings for fluorescence imaging: Hoechst Nuclear staining was detected with a 406/450 nanometer wavelength excitation/emission filter setting. GFP expressing cells were detected with a 488/513 nanometer wavelength excitation/emission filter setting. Image capture and analysis were performed with Harmony 4.6 software (Perkin Elmer). For this analysis, the instrument was setup to capture sixteen high resolution images per well on the same location in every well across all the experimental plates. Cell numbers and GFP quantifications across all experimental conditions were detected using Harmony 4.6 image analysis module.

qPCR

Total RNA was extracted using TRIzol reagent (Invitrogen) following the manufacturer’s protocol and RNA concentrations were measured using Nanodrop. cDNAs were synthesized using High Capacity cDNA Reverse Transcription Kit (ThermoFisher) and qPCR was performed using either ready-made TaqMan® assays or SYBR-green using custom-designed primers. All the data were analyzed using Applied Biosystems software (StepOne™ Software V2.3). For all experiments, relative expression levels of the target genes were determined by calculating the 2^−ΔCt values. All experiments were performed at least in triplicates. Either GAPDH or 18S rRNA was used as an internal normalization control for protein coding genes. RPLPO was used as an internal normalization control for ERVs. All SYBR-green primers were described previously and are listed in Table S1 along with TaqMan® probes used.

Anti-proliferation assay

Cells were seeded in 96-well plates at 40% confluency. Fresh medium was changed the next day and drugs were added directly. After mixing thoroughly, plates were cultured in a 37°C incubator for two more days. Drug-free fresh medium was changed the fourth day. The cells were collected on day 5 by trypsin, suspended in medium, mixed with trypan blue (Thermo Fisher Scientific) (1:1 ratio), and counted using LUNA II automated cell counter. Each sample was counted at least three times and the average numbers were used for the analysis. Each treatment condition was performed in triplicates.

Soft Agar Colony Formation Assay

Cells for colony-formation assays were pretreated with different concentrations of HH1 and MC180295 and drugs were kept in the medium for two more days and drug-free medium was changed the day before seeding (four days total). 1000 cells were then seeded in 35mm×10mm tissue culture dishes and cultured in a 37°C incuba tor for two weeks before staining using 0.005% crystal violet (dissolved in autoclaved water with 10% EtOH). Difco™ Agar Noble (BD Biosciences) was used to make soft agar. 2x medium supplemented with 20% FBS and 2% penicillin/streptomycin was used to culture colonies. Bottom layer was made of 0.6% agar and top layer was made of 0.3% agar. Feeder layer with 0.3% agar was added every week. All visible colonies were counted manually.

Histone extraction

Histones were extracted and prepared from isolated nuclei as described previously (Sidoli et al., 2016). Briefly, nuclei were incubated with 0.2 M H₂SO₄ for 2 hours and precipitated with 33% trichloroacetic acid (TCA) overnight to extract histones from the chromatin. Purified histones were dissolved in 30 μL of 50 mM NH₄HCO₃, pH 8.0, and a mixture of propionic anhydride with acetonitrile (ratio of 1:3 (v/v)) was added to the histone sample in the ratio of 1:4 (v/v) for 20 minutes at room temperature. This reaction was performed twice. Histones were then digested with trypsin (enzyme:sample ratio 1:20, 6 hours, room temperature) in 50 mM NH₄HCO₃. After digestion, derivatization was repeated to propionylate peptide N-termini. Samples were desalted prior LC-MS analysis using C18 Stage-tips.

Mass Spectrometry Analysis

Samples were then separated using a 75 μm ID × 17 cm Reprosil-Pur C18-AQ (3 μm; Dr. Maisch GmbH, Germany) nano-column mounted on an EASY-nLC nanoHPLC (Thermo Scientific, San Jose, Ca, USA). The HPLC gradient was as follows: 2% to 28% solvent B (A = 0.1% formic acid; B = 95% MeCN, 0.1% formic acid) over 45 minutes, from 28% to 80% solvent B in 5 minutes, 80% B for 10 minutes at a flow-rate of 300 nL/min. nLC was coupled online to an LTQ-Orbitrap Elite mass spectrometer (Thermo Scientific) and data were acquired using targeted scans and data-dependent acquisition (DDA). MS acquisition was divided into three segments, each beginning with a full MS scan: (i) MS/MS of the top seven most abundant ions (14 min), (ii) targeted CID fragmentation of common isobaric species (H3 peptide aa 9–17 with 1 acetyl, H3 peptide aa 18–26 with 1 acetyl and histone H4 peptide aa 4–17 with 1/2/3 acetyl groups) followed by CID fragmentation of the top five most abundant ions (27 min), (iii) CID fragmentation of the top ten most abundant ions (19 min). MS/MS was acquired using collision induced dissociation (CID) with normalized collision energy of 35 and collected in centroid mode. Data were searched using EpiProfile (Yuan et al., 2015). The peptide relative ratio was calculated using the total area under the extracted ion chromatograms of all peptides with the same amino acid sequence (including all of its modified forms) as 100%. For isobaric peptides, the relative ratio of two isobaric forms was estimated by averaging the ratio for each fragment ion with different mass between the two species. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD007925 and 10.6019/PXD007925.

In Vitro Phosphoproteomics

An in vitro kinase assay was performed using purified CDK9/Cyclin T1 (4.8μg) (Thermo Fisher Scientific)and BRG1 (3.2μg) (Abcam), in the presence or absence of 1μM of MC180295 using KAS buffer (with “cold” ATP) and incubate at 30 ºC for 1hr in quadruplicate. Protein reaction was stopped by heating at 65 ºC for 5 minutes and samples were desalted using 3kDa cutoff filter (nanosep 3K Omega PALL®). Samples were processed for mass spectrometer analysis using in-StageTip method for digestion and peptide purification before performing LC-MS/MS proteomics analysis.

The label-free proteomics analysis was performed using the nanoelectrospray ionization (ESI) tandem MS with a LTQ Orbitrap Elite mass spectrometer (Thermo Scientific). The complete system was fully controlled by Xcalibur software (Version 3.0.63). Mass spectra processing was performed using Proteome Discoverer 2.2.0.388. The generated de-isotoped peak list was submitted to an in-house Mascot Server 2.2.07 for searching against the Homo sapiens SwissProt database (TaxID=9606, released 2017–05-10. 42,153 sequences). Mascot search parameters were set as follows: species, Homo sapiens; enzyme: trypsin; maximal two missed cleavage; dynamic modifications: phospho (S, T) and phospho (Y); mass tolerance: 20 ppm for precursor peptide ions and 0.4 dalton tolerance for MS/MS fragment ions. Phosphopeptides matches were filtered using an ion score cutoff of 20. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD009573.

ChIP-qPCR

Chromatin immunoprecipitation (ChIP) was performed as described previously (Raynal et al., 2012) using antibodies for anti-histone H3K4me2 (Abam) and anti-histone H3K9 dimethylation (Abcam). In brief, YB5 cells were fixed with 1% formaldehyde, sonicated in SDS lysis buffer (1% SDS, 10uM EDTA, 50uM Tris pH 8.1) and immunoprecipitated using different antibodies. DNA was extracted using a QIAquick PCR Purification Kit (QIAGEN). Signal was quantified by qPCR and normalized using % of input. All ChIP primers are listed in Table S2.

QUANTIFICATION AND STATISTICAL ANALYSIS

RNA-seq

RNA from experiments in biological triplicates was isolated using RNeasy Mini Plus Kit (QIAGEN) following the manufacturer’s instructions. Strand-specific RNA libraries were generated from 1 μg of RNA using TruSeq stranded total RNA with Ribo-Zero Gold (Illumina). Sequencing was performed using single end reads (50 bp, average 50 million reads per sample) on the HiSeq2500 platform (Illumina) at Fox Chase Cancer Center Genomic Facility. Sequenced reads were aligned to the hg19 genome assembly using TopHat2 (Kim et al., 2013). The expression level and fold change of each treatment group was evaluated using EdgeR (Robinson et al., 2010). Genes that had 0 reads across all samples were excluded. In order to get rid of batch effects, samples were normalized using RUVr method from the RUVseq package (Risso et al., 2014). For the repetitive element analysis, RNAseq reads were mapped against known repetitive element sequences and data were normalized with the total number of mappable reads. Significantly regulated genes were defined as log2FC >1 or < -1, and FDR < 0.5. All the RNA-seq data were normalized to time-matched DMSO samples for each time point, while TET-off dnCDK9 was normalized to TET-on dnCDK9.

RRBS

Triplicate samples of YB5 cells treated with 10 μM HH1 and DMSO controls were analyzed for DNA methylation changes by reduced representation bisulfite sequencing (RRBS) (Gu et al., 2011). We followed the NEB protocol for methylated adaptors. Briefly, 1 microgram of genomic DNA was spiked 100 picograms of lambda phage DNA as the unmethylated standard and digested with MspI. Ends of restriction fragments were filled in, 3’-dA tailed and methylated adaptors (NEB E7535) were ligated to the ends of restriction fragments. Bisulfite treatment using the Epitect kit (Qiagen) followed. Bisulfite-converted libraries were amplified using EpiMark Taq DNA polymerase (NEB) and primers with barcode indices. The libraries were pooled and sequenced at Fox Chase Cancer Center Genomics Facility on Illumina HiSeq2500 instrument using single end reads of 50 bases. We used Bismark v0.18.1 (Krueger and Andrews, 2011) to align the sequences to hg19 human genome assembly. We used methylKit v1.3.3 (Akalin et al., 2012) to analyze differential methylation. Logistic regression was used to calculate P-values. P-values were adjusted to Q-values using SLIM method (Wang et al., 2011).

Digital restriction enzyme analysis of methylation (DREAM)

DREAM is a method established in our lab for DNA methylation analysis at tens of thousands of CpG sites across the genome (Jelinek et al., 2012). Sequential digests of genomic DNA with restriction endonucleases SmaI and XmaI creates specific signatures at unmethylated and methylated CpG sites. The signatures are resolved by high throughput sequencing.Briefly, two samples of 2 μg of genomic DNA from ID8 ovarian cancer cell line were digested with 20 units of SmaI (8 h at 25°C, NEB) and 20 units of XmaI (~16 h at 37°C, NEB), resulting in distinct DNA methylation signatures at CCCGGG sites. 3’ ends of the DNA fragments were repaired using Klenow fragment (3’→5’ exo-) DNA polymerase and dCTP, dGTP, and dATP nucleotides. Illumina sequencing adapters were ligated to the DNA fragments and the libraries were sequenced by paired-end 40 nt sequencing on Illumina HiSeq2500.The sequencing reads were mapped to the mm9 genome and methylation values were calculated as the ratio of the number of the reads with the methylated XmaI signature over the total number of tags mapped to a given SmaI/XmaI site. The coverage threshold was set to greater than 10 reads per sample.

ATAC-seq

ATAC-seq was performed as described (Buenrostro et al., 2013) using YB5 cells on three DMSO-treated (control) and three HH1-treated (4-day) samples. ATAC-seq reads were aligned to the human genome (build hg19) with the bowtie2 software suite. After alignment, we filtered out the reads mapped to ENCODE blacklisted regions, mitochondrial DNA, and duplicated reads using SAMtools. We used MACS2 to call peak regions. The list of 188 genes with consistently high expression in both control and drug treated samples were used to normalize ATAC-seq reads counts between conditions. We used ngs.plot.r to plot aggregate enrichment around TSS.

Data and Software Availability

The accession number of RNA-seq reported in this paper is GEO: GSE104837.

The accession number of ATAC-seq reported in this paper is GEO: GSE113608.

The accession number of RRBS reported in this paper is GEO: GSE104998.

The accession number of DREAM reported in this paper is GEO: GSE104997.

The accession numbers of LC-MS/MS data of histone post-translational modifications reported in this paper are PRIDE: PXD007925 and 10.6019/PXD007925.

The accession number of LC-MS/MS data of phosphoproteomics is PRIDE: PXD009573.

Comparative modeling of MC180295 in complex with CDK9

The SMILES string NC1=C(C(C2=C([N+]([O-])=O)C=CC=C2)=O)SC(N[C@H]3CC4CCC3C4)=N1 (corresponding to compound MC180295) was used to generate 100 low-energy conformers using the program OMEGA (Hawkins and Nicholls, 2012; Hawkins et al., 2010), via the command line:

omega2 -in input_file.smi -out output_file.sdf.gz -prefix ligand_name -warts -maxconfs 100

At the time of our study, the Protein Data Bank (PDB) contained 389 structures of CDK kinases in complex with ligands bound at the ATP site. The protein component from each structure was aligned to a single reference structure: for this we selected the crystal structure of human CDK9/cyclinT1 in complex with ATP (PDB ID: 3BLQ) (Baumli et al., 2008). The transformation applied to the protein was also applied to the ligand from each complex, yielding starting models of CDK9 in complex with a diverse variety of template ligand poses.

Each of the 100 low-energy conformers of MC180295 was sequentially aligned to each of the 389 ligand templates (i.e. a total of 38,900 overlays) using the ROCS software (Hawkins et al., 2007), via the command line:

rocs -dbase 180295_conformers.sdf.gz -query /extracted_ligand_library.pdb -prefix structure name -cutoff -1.0 -oformat sdf scdbase true -maxhits 100 -maxconfs 100 -outputquery false -qconflabel title

By concatentating the protein structure from human CDK9 bound to ATP with the MC180295 pose from aligning to these 389 ligand templates, this approach provided a set of complete, but unrefined, comparative modeling templates. We carried out a full-atom gradient-based energy minimization for each complex using the Rosetta macromolecular modeling suite (Leaver-Fay et al., 2011), then sorted the resulting models on the basis of protein-ligand interaction energy. Four of the top-scoring ten models adopted a nearly identical pose, whereas the other six had a broad variety of other poses. Based on consistency with the available structure-activity data, we confirmed this dominant cluster as the most likely pose (as described in the Main Text).

TCGA analysis

RNA-seq data for melanoma and colon cancer patients were downloaded from the TCGA data portal. Unsupervised hierarchical clustering was performed using ArrayTrack. Cox regression analyses and Kaplan-Meier curves were generated for CIM-high and CIM-low patients.

RNA-seq data from anti-CTLA4 treated melanoma patients

Patients were described previously (Chiappinelli et al., 2015). RNA-seq data were downloaded from dbGaP (phs001038). Unsupervised hierarchical clustering was performed using ArrayTrack.

All statistical analysis were calculated using the GraphPad Prism software. Student’s t-test was used to calculate significance between two groups and One-way ANOVA test was used to calculate differences for more than two groups.

For Figure 1, in all panels, *p<0.05, **p < 0.01, ***p < 0.001. One-way ANOVA, Dunnett's multiple comparison test was used in panels A, B and C while Student's t-test was used in panels D and E. Bar graphs represent mean+SD of at least biological triplicates.

For Figure 5, in all panels, *p<0.05, **p < 0.01, ***p < 0.001. Student's t-test was used in panel A while One-way ANOVA, Dunnett's multiple comparison test was used in panels B, C and D. Bar graphs represent mean+SD of at least biological triplicates.

One-way ANOVA, Dunnett's multiple comparison test was also used in Figures 2A and 6B.

Student’s t-test was used in Figure 4H and Mann Whitney test was used in Figure 6D.

All other statistical details of experiments are included in the Figure legends.

Supplementary Material

Figure S1: An unbiased screen identified CDK9 as the primary target to reactivate silenced genes in cancer. Related to Figure 1.

Figure S1A: Drug screening workflow using YB5 as a phenotypic based screening system. Drug development funnel shows the criteria for the selection of top hits.

Figure S1B: FACS analysis of YB5 treated with DMSO (negative control), 5μM TSA (positive control) and 10μM HH0 for 24 hours. On the FACS scatter plot, the x-axis and the y-axis represent GFP and propidium iodide (PI) fluorescence, respectively.

Figure S1C: GFP mRNA expression detected by qPCR after a 24-hour treatment using 10μM HH0 (N=3). DAC (100nM, daily treatment for 72hr) and Depsi (20nM, 24hr treatment) were used as positive controls. Data are shown as mean+SD. ***p < 0.001 (Student's t-test).

Figure S1D: GFP re-expression using different doses of HH1 and analogs in HCT116 (24hr) and MCF7 cells (four days after single dose treatment) (N=3) measured by flow cytometry. Data are shown as mean+SD. *p<0.05, ***p < 0.001 (One-way ANOVA, Dunnett's multiple comparison test).

Figure S1E: (Left) DNA methylation analysis of CMV promoter after drug treatment (24hr) analyzed by bisulfite pyrosequencing. DAC (24hr) was used as a positive control (N=3). Data are shown as mean+SD, ***p < 0.001 (Student's t-test). (Right) 10 μM HH1 (four days after single dose treatment) did not change DNA methylation compared to DMSO control, as measured by RRBS (Reduced Representation Bisulfite Sequencing) at 218,879 CpG sites with the minimum coverage of 10 reads. Red line shows linear regression. R^2 = 0.98, p <2 .2e-16.

Figure S1F: HDAC inhibitory activity assays were analyzed in vitro at 10μM in triplicates. Three aminothiazole compounds (HH0, HH1 and HH2) have no HDAC inhibitory activity. Four known HDACis (TSA, SAHA, depsipeptide (Depsi) and valproic acid (VPA)) were used as positive controls. N=3. Data are shown as mean+SD.***p < 0.001 (Student's t-test).

Figure S1G: Histone methyltransferase (HMT) and demethylase (HDM) inhibitory activities were assessed using either HH0 or HH1 at 10μM. No significant enzymatic inhibition was found for either HH0 or HH1.

Figure S1H: Global histone acetylation and methylation analysis after 48hr treatment with different CDK9 inhibitors showed a modest H3K79me2 increase detected by LC-MS. Depsipeptide was used as a positive control here. SNS-032 and GW8510 are two known CDK inhibitors. Fold change was calculated over the DMSO baseline (average value of duplicates).

Figure S1I: IC50 of three potent CDK inhibitors against different CDKs.

Figure S1J: Two endogenously hypermethylated genes (PYGM and RRAD) were reactivated upon dominant negative CDK9 (dnCDK9) overexpression (72hr) (TET-off) (N=3). Data are shown as mean+SD. HH1 (25μM for 48hr) was used as a positive control. ***p < 0.001 (Student's t-test).

Figure S1K: GFP reactivation upon dominant negative CDK9 (dnCDK9) overexpression (72hr) (TET-off) in HCT116-GFP cells (n=3). Cre virus was used as a negative control. N=3. Data are shown as mean+SD. **p < 0.01 (Student's t-test).

Figure S1L: GFP and two endogenously hypermethylated genes (MGMT and SYNE1) were reactivated upon CMV-dnCDK9 construct overexpression (72hr) (N=3). CMV-dnCDK1 and CMV-dnCDK2 constructs (72hr) did not trigger gene reactivation in YB5. The Western Blot shows the overexpression of dnCDK1, dnCDK2 and dnCDK9 after transfection. Data are shown as mean+SD. ***p < 0.001 (Student's t-test).

Figure S1M: CDK9 inhibition mediated GFP induction was abolished when overexpressing CDK9 and Cyclin T1 (72hr overexpression prior to drug treatment for 24hr). GFP fluorescence was detected by FACS (N=3). Data are shown as mean+SD. Depsipeptide was used as a negative control (uninhibited by CDK9 overexpression).***p < 0.001 (Student's t-test).

NIHMS1510720-supplement-1.pdf^{(501.2KB, pdf)}

Figure S2: GFP based structure activity relationship identified MC180295 as a potent and selective CDK9 inhibitor. Related to Figure 2.

Figure S2A: Reaction schemes for the lead compounds.

Figure S2B: IC50 curves of MC180295 against different CDKs show a high selectivity for CDK9.

Figure S2C: Two GSK-3 inhibitors (CHIR99021 and LiCl) were tested at multiple doses after single exposure for four days in YB5 with no GFP reactivation (N=3). Data are shown as mean+SD. Despipeptide was used as a positive control. ***p < 0.001 (Student's t-test).

Figure S2D: Western Blot after 2hr MC180295 treatment at different doses against pSer2 (a CDK9 target), phosphor-Rb at T870/811, phosphor-Rb at T826, p130 (all CDK4/6 targets), phosphor-CDK Substrate Motif [(K/H)pSP and phosphor-PRC1 (CDK1/2 targets).

Figure S2E: Western Blot after 8hr MC180295 treatment at different doses against pSer2 (a CDK9 target), phosphor-Rb at T870/811, phosphor-Rb at T826, p130 (all CDK4/6 targets), phosphor-CDK Substrate Motif [(K/H)pSP and phosphor-PRC1 (CDK1/2 targets).

NIHMS1510720-supplement-2.pdf^{(288.3KB, pdf)}

Figure S3: CDK9 inhibition reactivates epigenetically silenced genes and synergizes with DAC. Related to Figure 3.

Figure S3A: Time course of GFP and SYNE1 expression after HH1 treatment (25μM) for up to 22hr (left), 24hr, 48hr, 72hr, 96hr daily treatment, and four-day (single exposure) treatment (middle) (P values correspond to paired t-tests comparing the DMSO condition to subsequent time points.) and in combination with DAC (100nM DAC daily treatment for 48hr followed by four days after first HH1 exposure at 10μM) (right) (P values correspond to paired t-tests comparing the 1μM condition to subsequent doses.) in YB5. GFP and SYNE1 can be reactivated after 810hr HH1 treatment. DAC synergizes with HH1 in terms of GFP and SYNE1 reactivation significantly. N=3. Data are shown as mean+SD. *p<0.05, **p < 0.01, ***p < 0.001.

Figure S3B: Time course of MYC suppression after 25μM HH1 treatment for up to 16hr (top). MYC was sustainably suppressed four days after single exposure using multiple CDK9 inhibitors (bottom) (N=3). Data are shown as mean+SD. ***p < 0.001 (One-way ANOVA, Dunnett's multiple comparison test).

Figure S3C: Anti-proliferation assay measured after single exposure using multiple CDK9 inhibitors for 4 days, culturing the cells drug-free for two weeks and re-exposing them to drugs. P values correspond to paired t-tests comparing the drug condition to the corresponding DMSO control. N=3. Data are shown as mean+SD. *p<0.05, **p < 0.01.

Figure S3D: The number of genes upregulated and downregulated by 10μM HH1 treatment at each time point measured by RNA-seq (N=3, FC>2 or <0.5, FDR<0.1). dnCDK9 (72hr), DAC (daily treatment for 48hr, 100nM) and combinatorial treatment (100nM DAC daily treatment for 48hr followed by four days after first HH1 exposure at 10μM) were also included.

Figure S3E: Gene Ontology analysis of genes that were significantly downregulated (FC<0.5, FDR<0.1) after two-hour HH1 treatment at 10μM shows enrichment for transcription.

Figure S3F: Dynamics of gene expression for genes that were significantly downregulated (FC<0.5, FDR<0.1) after 2hr HH1 treatment at 10μM.The yellow dotted lines represent two-fold change.

Figure S3G: Dynamics of all genes (left), expressed genes (RPKM >= 0.31) and silenced genes (RPKM < 0.31).The yellow dotted lines represent two-fold change.

Figure S3H: Gene Ontology analysis of upregulated genes after HH1 treatment (four days after single exposure) at 10μM shows enrichment for cell adhesion.

Figure S3I: Comparison of gene reactivation at DNA methylated/silenced loci by the CDK9 inhibitor MC180295 and the DNMT inhibitor DAC. 10 genes (GLDC, c9orf172, MUC20, ANPEP, SUSD4, COLEC11, OLFML2A, GFP, MGMT and SYNE1) were selected based on promoter hypermethylation and gene silencing. YB5 cells were treated with a single dose of MC180295 (500nM) or daily doses (100nM) of DAC for 3 days. Cells were harvested on day 4. Gene expression was measured in triplicate by qPCR and shown is the average of all 10 genes. MC180295 achieved even faster gene reactivation than DAC. P values correspond to paired ttests comparing the 2hr time point to subsequent time points.

Figure S3J: Synergistic effect of HH1 with siDNMT1 (left) (either Non-targeting siRNA (siN) or siDNMT1 was transfected on day 0 and drugs were added on day 3. Drug-free media were changed on day 6 and FACS analysis was performed on day 7) (N=3) and DAC with siCDK9 (right) (either Non-targeting siRNA (siN) or siCDK9 was transfected on day 0. 50nM DAC was added on day 1 and FACS analysis was performed on day 4) in terms of GFP induction measured by FACS (N=6). Data are shown as mean+SD. **p < 0.01, ***p < 0.001 (paired t-test).

NIHMS1510720-supplement-3.pdf^{(5.9MB, pdf)}

Figure S4: CDK9 regulates the activity of BRG1 through phosphorylation. Related to Figure 4.

Figure S4A: ATAC-seq after 4-day either DMSO or 10μM HH1 treatment in YB5. Aggregate enrichment of reads around all TSS plotted by gene expression in DMSO (blue) and HH1 (green). High: 1st quartile based on gene expression. Medium: 2nd and 3rd quartile based on gene expression. Low: 4th quartile based on gene expression. None: not expressed genes.

Figure S4B: An increase of H3K4me2 occupancy at CMV/GFP loci after HH1 treatment (four days after single exposure) at 10μM measured by ChIP-qPCR (N=6) in YB5. *p<0.05, **p < 0.01 (Student's t-test).

Figure S4C: An increase of H3K4me2 occupancy at promoter regions of MGMT and SYNE1 after HH1 treatment (four days after single exposure) at 10μM measured by ChIP-qPCR in YB5 (N=3). *p<0.05 (Student's t-test).

Figure S4D: BRG1 co-immunoprecipitates with BAF155 and BAF60a in HEK293T cells. Immunoprecipitation was performed using an antibody against BRG1 and the respective coprecipitations were assessed using Western Blot.

Figure S4E: CDK9 co-immunoprecipitates with Cyclin T1 in HEK293T cells. Immunoprecipitation was performed using an antibody against CDK9, and the respective coprecipitation Cyclin T1 was assessed using Western Blot.

Figure S4F: Phosphorylated peptides that are in control (BRG1+CDK9/Cyclin T1) but absent after drug treatment (1μM MC180295 for 1hr) detected by LC-MS/MS. Phosphorylated sites are highlighted in red (top). These five serine residues on BRG1 are highly conserved across different species (bottom).

Figure S4G: YB5 cells were transfected with a FLAG-tagged BRG1 construct for 48hr prior to drug treatment for 48hr. GFP positive percentages were measured by FACS (N=3). Data are shown as mean+SD.*p<0.05, **p < 0.01 (paired t-test).

Figure S4H: Western Blot checking overexpression of WT (wild-type), 5STOA (five serine residues are substituted by alanine residues) and NO5S (five serine residues are deleted) BRG1 constructs in YB5 cells after 48hr transfection. An empty vector (EV) was used as a negative control.

Figure S4I: YB5 cells were transfected with different V5-tagged BRG1 constructs (WT, 5STOA and NO5S) for 48hr prior to drug treatment for another 48hr. GFP positive percentages were measured by FACS (n=3). Student’s t-tests were used to compare the number of GFP positive cells in the different conditions. Data are shown as mean+SD. N=3. *p<0.05, **p < 0.01, ***p < 0.001.

Figure S4J: YB5 cells were transfected with different V5-tagged BRG1 constructs (WT, 5STOA and NO5S) for 48hr prior to drug treatment for 48hr. GFP positive cells were captured using confocal microscopy (N=3). Representative images are shown in the figure.

Figure S4K: Disruption of BRG1/SMARCA4 using siSMARCA4 reduces gene activation by CDK9 inhibitors in YB5. To achieve a higher knocking-down efficiency, transfection was done every other day for a total of three times. Drugs were added to the medium 48hrs after the third transfection. Data are shown as mean+SD, n=3. **p < 0.01, ***p < 0.001 (paired t-test).

Figure S4L: PFI-3, a BRG1 inhibitor, was used to block BRG1 enzymatic activity (72hr daily pretreatment followed by 24hr co-treatment) (right). Inhibition of BRG1 diminished the effect mediated by CDK9 inhibitors (HH1 or iCDk9) on GFP induction in YB5. *p<0.05, **p < 0.01, ***p < 0.001 (One-way ANOVA, Dunnett's multiple comparison test).

Figure S4M: ChIP-seq for H3K9me2. Average read count per million mapped reads of genes upregulated by each condition in YB5 cells plotted around gene bodies. UNC0638, a G9a inhibitor was used as a positive control.

Figure S4N: High H3K9me2 (a repressive histone mark) occupancy at baseline CMV/GFP region, promoter regions of MGMT and SYNE1 measured by ChIP-qPCR in YB5. LINE-1, a repetitive repressive element was used as a positive control and GAPDH was used as a negative control (N=3). Data are shown as mean+SD. ***p < 0.001 (One-way ANOVA, Dunnett's multiple comparison test).

NIHMS1510720-supplement-4.pdf^{(374.5KB, pdf)}

Figure S5: Anti-tumoral efficacy of CDK9 inhibitors in vitro and in vivo. Related to Figure 5.

Figure S5A: Cell cycle analysis after drug treatment (four days after single exposure) using different CDK9 inhibitors (HH1, MC180295, flavopiridol (FVP) and iCDK9) at multiple doses in YB5 (N=3) shows no cell cycle arrest. Data are shown as mean+SEM.

Figure S5B: Cell apoptosis measured by sub-G1 sub-population after CDK9 inhibitor treatment (four days after single exposure) in YB5 (N=3). Data are shown as mean+SEM.

Figure S5C: Histograms of HL60 cell differentiation measured as CD11b expression under different conditions by flow cytometry. On each graph, the green peak represents the IgG control and the red peak represents FITC-CD11b fluorescence intensity.

Figure S5D: In the s.c. SW48 mouse model, 20mg/kg every other day treatment did not affect body weights in NSG mice. Data are shown as mean+SEM for each group.

Figure S5E: Two-week every other day MC180295 treatment reactivated GFP and GLDC in a dose dependent manner in vivo (N=3). Data are shown as mean+SEM. *p<0.05 (Student's t-test).

Figure S5F: NSG mice were inoculated (i.p.) with 5×105 SW48-luc cells. Four days later, at which time substantial tumor burden was evident by bioluminescence imaging, MC180295 or vehicle was administered (i.p.) every other day. Three vehicle controls and five drug-treated mice were included. Images were taken until week 8 post drug treatment.

Figure S5G: Survival of the mice in days. Significance was calculated using a log-rank (MantelCox) test. MC180295 can significantly extend survival of the mice in the i.p. SW48 mouse model.

Figure S5H: In the i.p. SW48 mouse model, every other day MC180295 treatment did not affect body weights in NSG mice. Data are shown as mean+SEM for each group.

Figure S5I: Reactivation of three tumor suppressor genes hypermethylated in ovarian cancer (selected by merging promoter hypermethylated genes identified by a DREAM assay with known ovarian cancer tumor suppressor genes from the TSGene database) after 24hr treatment with CDK9 inhibitors in mouse ID8 ovarian cancer cells detected by qPCR (N=3). Data are shown as mean+SD. *p<0.05, **p < 0.01, ***p < 0.001 (One-way ANOVA, Dunnett's multiple comparison test).

NIHMS1510720-supplement-5.pdf^{(394.1KB, pdf)}

Figure S6: Immunosensitization mediated by CDK9 inhibition. Related to Figure 6.

Figure S6A: Total number of reads mapped to repetitive sequences (left). Scatter plot of difference in abundance of reads between control and treatment. Each dot represents one repetitive element (right). Single exposure of HH1 for four days increased number of reads mapped to repetitive sequences in YB5.

Figure S6B: Single ERV activation measured by qPCR four-days after single dose CDK9 inhibitor treatment in HCT116 cells (N=3). Data are shown as mean+SD. *p<0.05, **p < 0.01, ***p < 0.001 (One-way ANOVA, Dunnett's multiple comparison test).

Figure S6C: Dynamics of gene expression (PD-L1, HLA-A, HLA-B and HLA-C) at different time points measured by qPCR in YB5 (N=3). Data are shown as mean+SD. *p<0.05, **p < 0.01 (One-way ANOVA, Dunnett's multiple comparison test).

Figure S6D: Generation of CDK9 IMmune signature (CIM) gene panel. RNA-seq data were analyzed by GSEA (Gene Set Enrichment Analysis) and 328 immune related genes were identified and upregulated after HH1 treatment (four days after single exposure).

Figure S6E: (Top) CIM (CDK9 IMmune signature) gene expression panel (328 genes) clusters TCGA colon cancer patients into high and low immune signatures. (Bottom) CIM-high patients tend to have a better survival than CIM-low patients.

Figure S6F: Anti-CTLA4 treated melanoma patients with long-term benefit tend to have higher expression levels of CIM signature related genes.

Figure S6G: In vivo treatment of mice with SNS-032 resulted in increased populations of activated dendritic cells (CD80/CD86+ and CD11c/MHCII+) in the tumor microenvironment, particularly with the addition of α-PD-1. Mean+SEM are shown. *p<0.05, **p < 0.01 (Mann Whitney test). Mononuclear cells isolated from ascites fluid were washed and stained for cell surface markers and analyzed via flow cytometry.

Figure S6H: MC180295 did not kill human immune cells in vivo. 20 million human PBMCs from a healthy donor (donor #2) were injected (i.p.) on day 0 into NSG mice. 10mg/kg MC180295 were injected (i.p.) every other day for 12 days and whole blood was collected on day 14. Flow cytometry was then performed using anti-CD45, anti-CD4 and anti-CD8 antibodies. Upper panels: representative pictures by flow cytometry showing cell gating.

NIHMS1510720-supplement-6.pdf^{(306.6KB, pdf)}

Table S1: SYBR green primers and Taqman probes used for qPCR. Related to the STAR Methods.

Table S2: Primers for ChIP-qPCR. Related to the STAR Methods.

Table S3: Histone methyltransferase (HMT) and demethylase (HDM) inhibitory activities were assessed using either HH0 or HH1 at 10μM in duplicate. Related to Figure 1.

NIHMS1510720-supplement-7.pdf^{(111.3KB, pdf)}

Table S4: Quantitative distribution of MC180295 inhibitory effect against a panel of 250 kinases at 1μM in duplicate experiments. Related to Figure 2.

NIHMS1510720-supplement-8.xlsx^{(21.4KB, xlsx)}

Highlights:

CDK9 inhibition reactivates epigenetically silenced genes in multiple cancer cells.
CDK9 binds to and phosphorylates BRG1, which contributes to gene silencing.
We developed a novel potent and selective CDK9 inhibitor, MC180295.
MC180285 shows promising anti-tumoral and immunosensitization activity.

ACKNOWLEDGMENTS:

This work was supported by NIH grants CA158112 and CA100632 to JPJI, GM117437 and CA216134–01 to XG, GM110174 to BAG, GM123336 to JK and DK105267 to OP. Research funding was also provided by the Van Andel Research Institute through the Van Andel Research Institute – Stand Up To Cancer Epigenetics Dream Team. Stand Up To Cancer is a program of the Entertainment Industry Foundation, administered by AACR. The work used the Extreme Science and Engineering Discovery Environment (XSEDE) allocation MCB130049, which is supported by National Science Foundation grant number ACI-1548562. JPI is an American Cancer Society Clinical Research professor supported by a generous gift from the F. M. Kirby Foundation. We thank Dr. Roland Dunbrack for providing a superposed set of all CDK kinase structures from the Protein Data Bank. We are grateful to OpenEye Scientific Software (Santa Fe, NM) for providing an academic license for the use of ROCS and OMEGA. We thank Dr. Bassel E. Sawaya for providing CDK9 constructs for the co-IP experiments.

Footnotes

DECLARATION OF INTERESTS:

H.Z., G. M., W. C. and M. A-G. and J.-P. I. are co-inventors of the pending patent: Novel Bridged Bicycloalkyl-Substituted Aminothiazoles and their Methods of Use (Application Number: PCT/US18/14465). All the newly synthesized aminothiazole analogs disclosed in the manuscript are covered by the patent. C.K. is a Scientific Founder, Board of Directors member, Scientific Advisory Board member, Shareholder, and Consultant for Foghorn Therapeutics, Inc. (Cambridge, MA). Other co-authors declare no competing interests.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES:

Adelman K, and Lis JT (2012). Promoter-proximal pausing of RNA polymerase II: emerging roles in metazoans. Nat Rev Genet 13, 720–731. [DOI] [PMC free article] [PubMed] [Google Scholar]
Akalin A, Kormaksson M, Li S, Garrett-Bakelman FE, Figueroa ME, Melnick A, and Mason CE (2012). methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles. Genome Biol 13, R87. [DOI] [PMC free article] [PubMed] [Google Scholar]
Asghar U, Witkiewicz AK, Turner NC, and Knudsen ES (2015). The history and future of targeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug Discov 14, 130–146. [DOI] [PMC free article] [PubMed] [Google Scholar]
Barboric M, Nissen RM, Kanazawa S, Jabrane-Ferrat N, and Peterlin BM (2001). NF-kappaB binds P-TEFb to stimulate transcriptional elongation by RNA polymerase II. Mol Cell 8, 327–337. [DOI] [PubMed] [Google Scholar]
Baumli S, Lolli G, Lowe ED, Troiani S, Rusconi L, Bullock AN, Debreczeni JE, Knapp S, and Johnson LN (2008). The structure of P-TEFb (CDK9/cyclin T1), its complex with flavopiridol and regulation by phosphorylation. EMBO J 27, 1907–1918. [DOI] [PMC free article] [PubMed] [Google Scholar]
Buenrostro JD, Giresi PG, Zaba LC, Chang HY, and Greenleaf WJ (2013). Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat Methods 10, 1213–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
Calbó J, Parreño M, Sotillo E, Yong T, Mazo A, Garriga J, and Grana X (2002). G1 cyclin/cyclindependent kinase-coordinated phosphorylation of endogenous pocket proteins differentially regulates their interactions with E2F4 and E2F1 and gene expression. J Biol Chem 277, 50263–50274. [DOI] [PubMed] [Google Scholar]
Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, et al. (2012). The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov 2, 401–404. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chartier M, Chénard T, Barker J, and Najmanovich R (2013). Kinome Render: a stand-alone and web-accessible tool to annotate the human protein kinome tree. PeerJ 1, e126. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chiappinelli KB, Strissel PL, Desrichard A, Li H, Henke C, Akman B, Hein A, Rote NS, Cope LM, Snyder A, et al. (2015). Inhibiting DNA Methylation Causes an Interferon Response in Cancer via dsRNA Including Endogenous Retroviruses. Cell 162, 974–986. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cui Y, Hausheer F, Beaty R, Zahnow C, Issa JP, Bunz F, and Baylin SB (2014). A recombinant reporter system for monitoring reactivation of an endogenously DNA hypermethylated gene. Cancer Res 74, 3834–3843. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, et al. (2013). Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 6, pl1. [DOI] [PMC free article] [PubMed] [Google Scholar]
Garriga J, Bhattacharya S, Calbó J, Marshall RM, Truongcao M, Haines DS, and Graña X (2003). CDK9 is constitutively expressed throughout the cell cycle, and its steady-state expression is independent of SKP2. Mol Cell Biol 23, 5165–5173. [DOI] [PMC free article] [PubMed] [Google Scholar]
Garriga J, and Graña X (2004). Cellular control of gene expression by T-type cyclin/CDK9 complexes. Gene 337, 15–23. [DOI] [PubMed] [Google Scholar]
Garriga J, and Graña X (2014). CDK9 inhibition strategy defines distinct sets of target genes. BMC Res Notes 7, 301. [DOI] [PMC free article] [PubMed] [Google Scholar]
Garriga J, Peng J, Parreño M, Price DH, Henderson EE, and Graña X (1998). Upregulation of cyclin T1/CDK9 complexes during T cell activation. Oncogene 17, 3093–3102. [DOI] [PubMed] [Google Scholar]
Garriga J, Xie H, Obradovic Z, and Graña X (2010). Selective control of gene expression by CDK9 in human cells. J Cell Physiol 222, 200–208. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gibson DG, Young L, Chuang RY, Venter JC, Hutchison CA, and Smith HO (2009). Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat Methods 6, 343–345. [DOI] [PubMed] [Google Scholar]
Goel S, DeCristo MJ, Watt AC, BrinJones H, Sceneay J, Li BB, Khan N, Ubellacker JM, Xie S, Metzger-Filho O, et al. (2017). CDK4/6 inhibition triggers anti-tumour immunity. Nature 548, 471–475. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gu H, Smith ZD, Bock C, Boyle P, Gnirke A, and Meissner A (2011). Preparation of reduced representation bisulfite sequencing libraries for genome-scale DNA methylation profiling. Nat Protoc 6, 468–481. [DOI] [PubMed] [Google Scholar]
Harbour JW, Luo RX, Dei Santi A, Postigo AA, and Dean DC (1999). Cdk phosphorylation triggers sequential intramolecular interactions that progressively block Rb functions as cells move through G1. Cell 98, 859–869. [DOI] [PubMed] [Google Scholar]
Hawkins PC, and Nicholls A (2012). Conformer generation with OMEGA: learning from the data set and the analysis of failures. J Chem Inf Model 52, 2919–2936. [DOI] [PubMed] [Google Scholar]
Hawkins PC, Skillman AG, and Nicholls A (2007). Comparison of shape-matching and docking as virtual screening tools. J Med Chem 50, 74–82. [DOI] [PubMed] [Google Scholar]
Hawkins PC, Skillman AG, Warren GL, Ellingson BA, and Stahl MT (2010). Conformer generation with OMEGA: algorithm and validation using high quality structures from the Protein Databank and Cambridge Structural Database. J Chem Inf Model 50, 572–584. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hornbeck PV, Zhang B, Murray B, Kornhauser JM, Latham V, and Skrzypek E (2015). PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res 43, D512–520. [DOI] [PMC free article] [PubMed] [Google Scholar]
Issa JP (2014). Aging and epigenetic drift: a vicious cycle. J Clin Invest 124, 24–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jelinek J, Liang S, Lu Y, He R, Ramagli LS, Shpall EJ, Estecio MR, and Issa JP (2012). Conserved DNA methylation patterns in healthy blood cells and extensive changes in leukemia measured by a new quantitative technique. Epigenetics 7, 1368–1378. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ji X, Lu H, Zhou Q, and Luo K (2014). LARP7 suppresses P-TEFb activity to inhibit breast cancer progression and metastasis. Elife 3, e02907. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jones PA, Issa JP, and Baylin S (2016). Targeting the cancer epigenome for therapy. Nat Rev Genet 17, 630–641. [DOI] [PubMed] [Google Scholar]
Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, and Salzberg SL (2013). TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol 14, R36. [DOI] [PMC free article] [PubMed] [Google Scholar]
Krueger F, and Andrews SR (2011). Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics 27, 1571–1572. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, Lerner J, Brunet JP, Subramanian A, Ross KN, et al. (2006). The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science 313, 1929–1935. [DOI] [PubMed] [Google Scholar]
Langmead B, and Salzberg SL (2012). Fast gapped-read alignment with Bowtie 2. Nat Methods 9, 357359. [DOI] [PMC free article] [PubMed] [Google Scholar]
Leaver-Fay A, Tyka M, Lewis SM, Lange OF, Thompson J, Jacak R, Kaufman K, Renfrew PD, Smith CA, Sheffler W, et al. (2011). ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol 487, 545–574. [DOI] [PMC free article] [PubMed] [Google Scholar]
Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, and Subgroup GPDP (2009). The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lu H, and Schulze-Gahmen U (2006). Toward understanding the structural basis of cyclin-dependent kinase 6 specific inhibition. J Med Chem 49, 3826–3831. [DOI] [PubMed] [Google Scholar]
Lu H, Xue Y, Yu GK, Arias C, Lin J, Fong S, Faure M, Weisburd B, Ji X, Mercier A, et al. (2015). Compensatory induction of MYC expression by sustained CDK9 inhibition via a BRD4-dependent mechanism. Elife 4, e06535. [DOI] [PMC free article] [PubMed] [Google Scholar]
MacCallum DE, Melville J, Frame S, Watt K, Anderson S, Gianella-Borradori A, Lane DP, and Green SR (2005). Seliciclib (CYC202, R-Roscovitine) induces cell death in multiple myeloma cells by inhibition of RNA polymerase II-dependent transcription and down-regulation of Mcl-1. Cancer Res 65, 5399–5407. [DOI] [PubMed] [Google Scholar]
Nan X, Ng HH, Johnson CA, Laherty CD, Turner BM, Eisenman RN, and Bird A (1998). Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature 393, 386–389. [DOI] [PubMed] [Google Scholar]
Network CGA (2012). Comprehensive molecular characterization of human colon and rectal cancer. Nature 487, 330–337. [DOI] [PMC free article] [PubMed] [Google Scholar]
Network CGA (2015). Genomic Classification of Cutaneous Melanoma. Cell 161, 1681–1696. [DOI] [PMC free article] [PubMed] [Google Scholar]
Quinlan AR, and Hall IM (2010). BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rahl PB, Lin CY, Seila AC, Flynn RA, McCuine S, Burge CB, Sharp PA, and Young RA (2010). c-Myc regulates transcriptional pause release. Cell 141, 432–445. [DOI] [PMC free article] [PubMed] [Google Scholar]
Raynal NJ, Si J, Taby RF, Gharibyan V, Ahmed S, Jelinek J, Estécio MR, and Issa JP (2012). DNA methylation does not stably lock gene expression but instead serves as a molecular mark for gene silencing memory. Cancer Res 72, 1170–1181. [DOI] [PMC free article] [PubMed] [Google Scholar]
Risso D, Ngai J, Speed TP, and Dudoit S (2014). Normalization of RNA-seq data using factor analysis of control genes or samples. Nat Biotechnol 32, 896–902. [DOI] [PMC free article] [PubMed] [Google Scholar]
Robinson MD, McCarthy DJ, and Smyth GK (2010). edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rouillard AD, Gundersen GW, Fernandez NF, Wang Z, Monteiro CD, McDermott MG, and Ma'ayan A (2016). The harmonizome: a collection of processed datasets gathered to serve and mine knowledge about genes and proteins. Database (Oxford) 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Roulois D, Loo Yau H, Singhania R, Wang Y, Danesh A, Shen SY, Han H, Liang G, Jones PA, Pugh TJ, et al. (2015). DNA-Demethylating Agents Target Colorectal Cancer Cells by Inducing Viral Mimicry by Endogenous Transcripts. Cell 162, 961–973. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shen L, Shao N, Liu X, and Nestler E (2014). ngs.plot: Quick mining and visualization of nextgeneration sequencing data by integrating genomic databases. BMC Genomics 15, 284. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sherr CJ, Beach D, and Shapiro GI (2016). Targeting CDK4 and CDK6: From Discovery to Therapy. Cancer Discov 6, 353–367. [DOI] [PMC free article] [PubMed] [Google Scholar]
Si J, Boumber YA, Shu J, Qin T, Ahmed S, He R, Jelinek J, and Issa JP (2010). Chromatin remodeling is required for gene reactivation after decitabine-mediated DNA hypomethylation. Cancer Res 70, 6968–6977. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sidoli S, Bhanu NV, Karch KR, Wang X, and Garcia BA (2016). Complete Workflow for Analysis of Histone Post-translational Modifications Using Bottom-up Mass Spectrometry: From Histone Extraction to Data Analysis. J Vis Exp. [DOI] [PMC free article] [PubMed] [Google Scholar]
Snyder A, Makarov V, Merghoub T, Yuan J, Zaretsky JM, Desrichard A, Walsh LA, Postow MA, Wong P, Ho TS, et al. (2014). Genetic basis for clinical response to CTLA-4 blockade in melanoma. N Engl J Med 371, 2189–2199. [DOI] [PMC free article] [PubMed] [Google Scholar]
St Pierre R, and Kadoch C (2017). Mammalian SWI/SNF complexes in cancer: emerging therapeutic opportunities. Curr Opin Genet Dev 42, 56–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
Taby R, and Issa JP (2010). Cancer epigenetics. CA Cancer J Clin 60, 376–392. [DOI] [PubMed] [Google Scholar]
van den Heuvel S, and Harlow E (1993). Distinct roles for cyclin-dependent kinases in cell cycle control. Science 262, 2050–2054. [DOI] [PubMed] [Google Scholar]
Van Duyne R, Guendel I, Narayanan A, Gregg E, Shafagati N, Tyagi M, Easley R, Klase Z, Nekhai S, Kehn-Hall K, et al. (2011). Varying modulation of HIV-1 LTR activity by Baf complexes. J Mol Biol 411, 581–596. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vangamudi B, Paul TA, Shah PK, Kost-Alimova M, Nottebaum L, Shi X, Zhan Y, Leo E, Mahadeshwar HS, Protopopov A, et al. (2015). The SMARCA2/4 ATPase Domain Surpasses the Bromodomain as a Drug Target in SWI/SNF-Mutant Cancers: Insights from cDNA Rescue and PFI-3 Inhibitor Studies. Cancer Res 75, 3865–3878. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vizcaíno JA, Csordas A, del-Toro N, Dianes JA, Griss J, Lavidas I, Mayer G, Perez-Riverol Y, Reisinger F, Ternent T, et al. (2016). 2016 update of the PRIDE database and its related tools. Nucleic Acids Res 44, D447–456. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wang HQ, Tuominen LK, and Tsai CJ (2011). SLIM: a sliding linear model for estimating the proportion of true null hypotheses in datasets with dependence structures. Bioinformatics 27, 225–231. [DOI] [PubMed] [Google Scholar]
Wang Q, Zorn JA, and Kuriyan J (2014). A structural atlas of kinases inhibited by clinically approved drugs. Methods Enzymol 548, 23–67. [DOI] [PubMed] [Google Scholar]
Wang Y, Zhang T, Kwiatkowski N, Abraham BJ, Lee TI, Xie S, Yuzugullu H, Von T, Li H, Lin Z, et al. (2015). CDK7-dependent transcriptional addiction in triple-negative breast cancer. Cell 163, 174186. [DOI] [PMC free article] [PubMed] [Google Scholar]
Xi Q, He W, Zhang XH, Le HV, and Massagué J (2008). Genome-wide impact of the BRG1 SWI/SNF chromatin remodeler on the transforming growth factor beta transcriptional program. J Biol Chem 283, 1146–1155. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yuan ZF, Lin S, Molden RC, Cao XJ, Bhanu NV, Wang X, Sidoli S, Liu S, and Garcia BA (2015). EpiProfile Quantifies Histone Peptides With Modifications by Extracting Retention Time and Intensity in High-resolution Mass Spectra. Mol Cell Proteomics 14, 1696–1707. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, et al. (2008). Model-based analysis of ChIP-Seq (MACS). Genome Biol 9, R137. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1: An unbiased screen identified CDK9 as the primary target to reactivate silenced genes in cancer. Related to Figure 1.

Figure S1A: Drug screening workflow using YB5 as a phenotypic based screening system. Drug development funnel shows the criteria for the selection of top hits.

Figure S1I: IC50 of three potent CDK inhibitors against different CDKs.

NIHMS1510720-supplement-1.pdf^{(501.2KB, pdf)}

Figure S2: GFP based structure activity relationship identified MC180295 as a potent and selective CDK9 inhibitor. Related to Figure 2.

Figure S2A: Reaction schemes for the lead compounds.

Figure S2B: IC50 curves of MC180295 against different CDKs show a high selectivity for CDK9.

NIHMS1510720-supplement-2.pdf^{(288.3KB, pdf)}

Figure S3: CDK9 inhibition reactivates epigenetically silenced genes and synergizes with DAC. Related to Figure 3.

Figure S3E: Gene Ontology analysis of genes that were significantly downregulated (FC<0.5, FDR<0.1) after two-hour HH1 treatment at 10μM shows enrichment for transcription.

Figure S3F: Dynamics of gene expression for genes that were significantly downregulated (FC<0.5, FDR<0.1) after 2hr HH1 treatment at 10μM.The yellow dotted lines represent two-fold change.

Figure S3G: Dynamics of all genes (left), expressed genes (RPKM >= 0.31) and silenced genes (RPKM < 0.31).The yellow dotted lines represent two-fold change.

Figure S3H: Gene Ontology analysis of upregulated genes after HH1 treatment (four days after single exposure) at 10μM shows enrichment for cell adhesion.

NIHMS1510720-supplement-3.pdf^{(5.9MB, pdf)}

Figure S4: CDK9 regulates the activity of BRG1 through phosphorylation. Related to Figure 4.

Figure S4B: An increase of H3K4me2 occupancy at CMV/GFP loci after HH1 treatment (four days after single exposure) at 10μM measured by ChIP-qPCR (N=6) in YB5. *p<0.05, **p < 0.01 (Student's t-test).

NIHMS1510720-supplement-4.pdf^{(374.5KB, pdf)}

Figure S5: Anti-tumoral efficacy of CDK9 inhibitors in vitro and in vivo. Related to Figure 5.

Figure S5B: Cell apoptosis measured by sub-G1 sub-population after CDK9 inhibitor treatment (four days after single exposure) in YB5 (N=3). Data are shown as mean+SEM.

Figure S5D: In the s.c. SW48 mouse model, 20mg/kg every other day treatment did not affect body weights in NSG mice. Data are shown as mean+SEM for each group.

Figure S5E: Two-week every other day MC180295 treatment reactivated GFP and GLDC in a dose dependent manner in vivo (N=3). Data are shown as mean+SEM. *p<0.05 (Student's t-test).

Figure S5G: Survival of the mice in days. Significance was calculated using a log-rank (MantelCox) test. MC180295 can significantly extend survival of the mice in the i.p. SW48 mouse model.

Figure S5H: In the i.p. SW48 mouse model, every other day MC180295 treatment did not affect body weights in NSG mice. Data are shown as mean+SEM for each group.

NIHMS1510720-supplement-5.pdf^{(394.1KB, pdf)}

Figure S6: Immunosensitization mediated by CDK9 inhibition. Related to Figure 6.

Figure S6F: Anti-CTLA4 treated melanoma patients with long-term benefit tend to have higher expression levels of CIM signature related genes.

NIHMS1510720-supplement-6.pdf^{(306.6KB, pdf)}

Table S1: SYBR green primers and Taqman probes used for qPCR. Related to the STAR Methods.

Table S2: Primers for ChIP-qPCR. Related to the STAR Methods.

Table S3: Histone methyltransferase (HMT) and demethylase (HDM) inhibitory activities were assessed using either HH0 or HH1 at 10μM in duplicate. Related to Figure 1.

NIHMS1510720-supplement-7.pdf^{(111.3KB, pdf)}

Table S4: Quantitative distribution of MC180295 inhibitory effect against a panel of 250 kinases at 1μM in duplicate experiments. Related to Figure 2.

NIHMS1510720-supplement-8.xlsx^{(21.4KB, xlsx)}

[R1] Adelman K, and Lis JT (2012). Promoter-proximal pausing of RNA polymerase II: emerging roles in metazoans. Nat Rev Genet 13, 720–731. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] Akalin A, Kormaksson M, Li S, Garrett-Bakelman FE, Figueroa ME, Melnick A, and Mason CE (2012). methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles. Genome Biol 13, R87. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] Asghar U, Witkiewicz AK, Turner NC, and Knudsen ES (2015). The history and future of targeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug Discov 14, 130–146. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] Barboric M, Nissen RM, Kanazawa S, Jabrane-Ferrat N, and Peterlin BM (2001). NF-kappaB binds P-TEFb to stimulate transcriptional elongation by RNA polymerase II. Mol Cell 8, 327–337. [DOI] [PubMed] [Google Scholar]

[R5] Baumli S, Lolli G, Lowe ED, Troiani S, Rusconi L, Bullock AN, Debreczeni JE, Knapp S, and Johnson LN (2008). The structure of P-TEFb (CDK9/cyclin T1), its complex with flavopiridol and regulation by phosphorylation. EMBO J 27, 1907–1918. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] Buenrostro JD, Giresi PG, Zaba LC, Chang HY, and Greenleaf WJ (2013). Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat Methods 10, 1213–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] Calbó J, Parreño M, Sotillo E, Yong T, Mazo A, Garriga J, and Grana X (2002). G1 cyclin/cyclindependent kinase-coordinated phosphorylation of endogenous pocket proteins differentially regulates their interactions with E2F4 and E2F1 and gene expression. J Biol Chem 277, 50263–50274. [DOI] [PubMed] [Google Scholar]

[R8] Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, et al. (2012). The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov 2, 401–404. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] Chartier M, Chénard T, Barker J, and Najmanovich R (2013). Kinome Render: a stand-alone and web-accessible tool to annotate the human protein kinome tree. PeerJ 1, e126. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] Chiappinelli KB, Strissel PL, Desrichard A, Li H, Henke C, Akman B, Hein A, Rote NS, Cope LM, Snyder A, et al. (2015). Inhibiting DNA Methylation Causes an Interferon Response in Cancer via dsRNA Including Endogenous Retroviruses. Cell 162, 974–986. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] Cui Y, Hausheer F, Beaty R, Zahnow C, Issa JP, Bunz F, and Baylin SB (2014). A recombinant reporter system for monitoring reactivation of an endogenously DNA hypermethylated gene. Cancer Res 74, 3834–3843. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, et al. (2013). Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 6, pl1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] Garriga J, Bhattacharya S, Calbó J, Marshall RM, Truongcao M, Haines DS, and Graña X (2003). CDK9 is constitutively expressed throughout the cell cycle, and its steady-state expression is independent of SKP2. Mol Cell Biol 23, 5165–5173. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] Garriga J, and Graña X (2004). Cellular control of gene expression by T-type cyclin/CDK9 complexes. Gene 337, 15–23. [DOI] [PubMed] [Google Scholar]

[R15] Garriga J, and Graña X (2014). CDK9 inhibition strategy defines distinct sets of target genes. BMC Res Notes 7, 301. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] Garriga J, Peng J, Parreño M, Price DH, Henderson EE, and Graña X (1998). Upregulation of cyclin T1/CDK9 complexes during T cell activation. Oncogene 17, 3093–3102. [DOI] [PubMed] [Google Scholar]

[R17] Garriga J, Xie H, Obradovic Z, and Graña X (2010). Selective control of gene expression by CDK9 in human cells. J Cell Physiol 222, 200–208. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] Gibson DG, Young L, Chuang RY, Venter JC, Hutchison CA, and Smith HO (2009). Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat Methods 6, 343–345. [DOI] [PubMed] [Google Scholar]

[R19] Goel S, DeCristo MJ, Watt AC, BrinJones H, Sceneay J, Li BB, Khan N, Ubellacker JM, Xie S, Metzger-Filho O, et al. (2017). CDK4/6 inhibition triggers anti-tumour immunity. Nature 548, 471–475. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] Gu H, Smith ZD, Bock C, Boyle P, Gnirke A, and Meissner A (2011). Preparation of reduced representation bisulfite sequencing libraries for genome-scale DNA methylation profiling. Nat Protoc 6, 468–481. [DOI] [PubMed] [Google Scholar]

[R21] Harbour JW, Luo RX, Dei Santi A, Postigo AA, and Dean DC (1999). Cdk phosphorylation triggers sequential intramolecular interactions that progressively block Rb functions as cells move through G1. Cell 98, 859–869. [DOI] [PubMed] [Google Scholar]

[R22] Hawkins PC, and Nicholls A (2012). Conformer generation with OMEGA: learning from the data set and the analysis of failures. J Chem Inf Model 52, 2919–2936. [DOI] [PubMed] [Google Scholar]

[R23] Hawkins PC, Skillman AG, and Nicholls A (2007). Comparison of shape-matching and docking as virtual screening tools. J Med Chem 50, 74–82. [DOI] [PubMed] [Google Scholar]

[R24] Hawkins PC, Skillman AG, Warren GL, Ellingson BA, and Stahl MT (2010). Conformer generation with OMEGA: algorithm and validation using high quality structures from the Protein Databank and Cambridge Structural Database. J Chem Inf Model 50, 572–584. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] Hornbeck PV, Zhang B, Murray B, Kornhauser JM, Latham V, and Skrzypek E (2015). PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res 43, D512–520. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] Issa JP (2014). Aging and epigenetic drift: a vicious cycle. J Clin Invest 124, 24–29. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] Jelinek J, Liang S, Lu Y, He R, Ramagli LS, Shpall EJ, Estecio MR, and Issa JP (2012). Conserved DNA methylation patterns in healthy blood cells and extensive changes in leukemia measured by a new quantitative technique. Epigenetics 7, 1368–1378. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] Ji X, Lu H, Zhou Q, and Luo K (2014). LARP7 suppresses P-TEFb activity to inhibit breast cancer progression and metastasis. Elife 3, e02907. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] Jones PA, Issa JP, and Baylin S (2016). Targeting the cancer epigenome for therapy. Nat Rev Genet 17, 630–641. [DOI] [PubMed] [Google Scholar]

[R30] Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, and Salzberg SL (2013). TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol 14, R36. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] Krueger F, and Andrews SR (2011). Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics 27, 1571–1572. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, Lerner J, Brunet JP, Subramanian A, Ross KN, et al. (2006). The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science 313, 1929–1935. [DOI] [PubMed] [Google Scholar]

[R33] Langmead B, and Salzberg SL (2012). Fast gapped-read alignment with Bowtie 2. Nat Methods 9, 357359. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] Leaver-Fay A, Tyka M, Lewis SM, Lange OF, Thompson J, Jacak R, Kaufman K, Renfrew PD, Smith CA, Sheffler W, et al. (2011). ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol 487, 545–574. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, and Subgroup GPDP (2009). The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R36] Lu H, and Schulze-Gahmen U (2006). Toward understanding the structural basis of cyclin-dependent kinase 6 specific inhibition. J Med Chem 49, 3826–3831. [DOI] [PubMed] [Google Scholar]

[R37] Lu H, Xue Y, Yu GK, Arias C, Lin J, Fong S, Faure M, Weisburd B, Ji X, Mercier A, et al. (2015). Compensatory induction of MYC expression by sustained CDK9 inhibition via a BRD4-dependent mechanism. Elife 4, e06535. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] MacCallum DE, Melville J, Frame S, Watt K, Anderson S, Gianella-Borradori A, Lane DP, and Green SR (2005). Seliciclib (CYC202, R-Roscovitine) induces cell death in multiple myeloma cells by inhibition of RNA polymerase II-dependent transcription and down-regulation of Mcl-1. Cancer Res 65, 5399–5407. [DOI] [PubMed] [Google Scholar]

[R39] Nan X, Ng HH, Johnson CA, Laherty CD, Turner BM, Eisenman RN, and Bird A (1998). Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature 393, 386–389. [DOI] [PubMed] [Google Scholar]

[R40] Network CGA (2012). Comprehensive molecular characterization of human colon and rectal cancer. Nature 487, 330–337. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] Network CGA (2015). Genomic Classification of Cutaneous Melanoma. Cell 161, 1681–1696. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] Quinlan AR, and Hall IM (2010). BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] Rahl PB, Lin CY, Seila AC, Flynn RA, McCuine S, Burge CB, Sharp PA, and Young RA (2010). c-Myc regulates transcriptional pause release. Cell 141, 432–445. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] Raynal NJ, Si J, Taby RF, Gharibyan V, Ahmed S, Jelinek J, Estécio MR, and Issa JP (2012). DNA methylation does not stably lock gene expression but instead serves as a molecular mark for gene silencing memory. Cancer Res 72, 1170–1181. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] Risso D, Ngai J, Speed TP, and Dudoit S (2014). Normalization of RNA-seq data using factor analysis of control genes or samples. Nat Biotechnol 32, 896–902. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] Robinson MD, McCarthy DJ, and Smyth GK (2010). edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] Rouillard AD, Gundersen GW, Fernandez NF, Wang Z, Monteiro CD, McDermott MG, and Ma'ayan A (2016). The harmonizome: a collection of processed datasets gathered to serve and mine knowledge about genes and proteins. Database (Oxford) 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] Roulois D, Loo Yau H, Singhania R, Wang Y, Danesh A, Shen SY, Han H, Liang G, Jones PA, Pugh TJ, et al. (2015). DNA-Demethylating Agents Target Colorectal Cancer Cells by Inducing Viral Mimicry by Endogenous Transcripts. Cell 162, 961–973. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] Shen L, Shao N, Liu X, and Nestler E (2014). ngs.plot: Quick mining and visualization of nextgeneration sequencing data by integrating genomic databases. BMC Genomics 15, 284. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] Sherr CJ, Beach D, and Shapiro GI (2016). Targeting CDK4 and CDK6: From Discovery to Therapy. Cancer Discov 6, 353–367. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] Si J, Boumber YA, Shu J, Qin T, Ahmed S, He R, Jelinek J, and Issa JP (2010). Chromatin remodeling is required for gene reactivation after decitabine-mediated DNA hypomethylation. Cancer Res 70, 6968–6977. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] Sidoli S, Bhanu NV, Karch KR, Wang X, and Garcia BA (2016). Complete Workflow for Analysis of Histone Post-translational Modifications Using Bottom-up Mass Spectrometry: From Histone Extraction to Data Analysis. J Vis Exp. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R53] Snyder A, Makarov V, Merghoub T, Yuan J, Zaretsky JM, Desrichard A, Walsh LA, Postow MA, Wong P, Ho TS, et al. (2014). Genetic basis for clinical response to CTLA-4 blockade in melanoma. N Engl J Med 371, 2189–2199. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] St Pierre R, and Kadoch C (2017). Mammalian SWI/SNF complexes in cancer: emerging therapeutic opportunities. Curr Opin Genet Dev 42, 56–67. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R55] Taby R, and Issa JP (2010). Cancer epigenetics. CA Cancer J Clin 60, 376–392. [DOI] [PubMed] [Google Scholar]

[R56] van den Heuvel S, and Harlow E (1993). Distinct roles for cyclin-dependent kinases in cell cycle control. Science 262, 2050–2054. [DOI] [PubMed] [Google Scholar]

[R57] Van Duyne R, Guendel I, Narayanan A, Gregg E, Shafagati N, Tyagi M, Easley R, Klase Z, Nekhai S, Kehn-Hall K, et al. (2011). Varying modulation of HIV-1 LTR activity by Baf complexes. J Mol Biol 411, 581–596. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] Vangamudi B, Paul TA, Shah PK, Kost-Alimova M, Nottebaum L, Shi X, Zhan Y, Leo E, Mahadeshwar HS, Protopopov A, et al. (2015). The SMARCA2/4 ATPase Domain Surpasses the Bromodomain as a Drug Target in SWI/SNF-Mutant Cancers: Insights from cDNA Rescue and PFI-3 Inhibitor Studies. Cancer Res 75, 3865–3878. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R59] Vizcaíno JA, Csordas A, del-Toro N, Dianes JA, Griss J, Lavidas I, Mayer G, Perez-Riverol Y, Reisinger F, Ternent T, et al. (2016). 2016 update of the PRIDE database and its related tools. Nucleic Acids Res 44, D447–456. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R60] Wang HQ, Tuominen LK, and Tsai CJ (2011). SLIM: a sliding linear model for estimating the proportion of true null hypotheses in datasets with dependence structures. Bioinformatics 27, 225–231. [DOI] [PubMed] [Google Scholar]

[R61] Wang Q, Zorn JA, and Kuriyan J (2014). A structural atlas of kinases inhibited by clinically approved drugs. Methods Enzymol 548, 23–67. [DOI] [PubMed] [Google Scholar]

[R62] Wang Y, Zhang T, Kwiatkowski N, Abraham BJ, Lee TI, Xie S, Yuzugullu H, Von T, Li H, Lin Z, et al. (2015). CDK7-dependent transcriptional addiction in triple-negative breast cancer. Cell 163, 174186. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R63] Xi Q, He W, Zhang XH, Le HV, and Massagué J (2008). Genome-wide impact of the BRG1 SWI/SNF chromatin remodeler on the transforming growth factor beta transcriptional program. J Biol Chem 283, 1146–1155. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] Yuan ZF, Lin S, Molden RC, Cao XJ, Bhanu NV, Wang X, Sidoli S, Liu S, and Garcia BA (2015). EpiProfile Quantifies Histone Peptides With Modifications by Extracting Retention Time and Intensity in High-resolution Mass Spectra. Mol Cell Proteomics 14, 1696–1707. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, et al. (2008). Model-based analysis of ChIP-Seq (MACS). Genome Biol 9, R137. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Targeting CDK9 Reactivates Epigenetically Silenced Genes in Cancer

Hanghang Zhang

Somnath Pandey

Meghan Travers

Hongxing Sun

George Morton

Jozef Madzo

Woonbok Chung

Jittasak Khowsathit

Oscar Perez-Leal

Carlos A Barrero

Carmen Merali

Yasuyuki Okamoto

Takahiro Sato

Joshua Pan

Judit Garriga

Natarajan V Bhanu

Johayra Simithy

Bela Patel

Jian Huang

Noël J-M Raynal

Benjamin A Garcia

Marlene A Jacobson

Cigall Kadoch

Salim Merali

Yi Zhang

Wayne Childers

Magid Abou-Gharbia

John Karanicolas

Stephen B Baylin

Cynthia A Zahnow

Jaroslav Jelinek

Xavier Graña

Jean-Pierre J Issa

SUMMARY:

In brief

INTRODUCTION

RESULTS:

Identification of CDK9 as a novel epigenetic target

Figure 1:

Figure 2:

Development of novel CDK9 inhibitors

CDK9 inhibition leads epigenetic derepression genome-wide

Figure 3:

BRG1 is a direct phosphorylation-substrate of CDK9

Figure 4:

BRG1 is essential for gene reactivation by CDK9 inhibition

CDK9 inhibition has anti-tumoral effects in vitro and in vivo.

Figure 5:

CDK9 inhibition activates endogenous retroviruses and an Interferon response

Figure 6:

DISCUSSION:

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell lines and cell culture

NSG mice treated with MC180295

Humanized PBMC NSG mice treated with MC180295

Pharmacodynamics study in NSG mice treated with MC180295

Mouse experiments with in vivo treatment of SNS-032

METHOD DETAILS

NDL-3040 library screening and drug treatments

DNA extraction and DNA methylation analysis

siRNA Knockdown

Plasmids transfection and viral transduction

Biochemistry assays

Synthesis of MC180295

Step 1

Step 2

Flow cytometry

Nuclear Extract Preparation

Co-immunoprecipitation

Generation of 5STOA and NO5S BRG1 constructs

Cell microscopy and High Content Imaging Analysis

qPCR

Anti-proliferation assay

Soft Agar Colony Formation Assay

Histone extraction

Mass Spectrometry Analysis