Mediator Kinase Inhibitor Selectivity and Activity in Colorectal Cancer

Abstract

The Mediator complex is a regulator of gene expression, influencing chromatin structure and RNA polymerase II-mediated transcription. Its activity is controlled by a protein kinase module, which includes cyclin-dependent kinases 8 and 19, that phosphorylates RNA polymerase II and transcription factors to regulate gene expression. Using orthogonal approaches combining chemical and genetic tools, we demonstrated the selectivity of our small-molecule inhibitors derived from 3,4,5-trisubstituted pyridine and 3-methyl-1H-pyrazolo[3,4-b]pyridine chemical series in human colorectal cell culture and tumor xenograft models. The lack of activity of our inhibitors in CDK8/19 double knockout models, with respect to molecular, proliferative, and antitumor end points, revealed their specificity and dependence on these kinases. Using our chemical probes and knockout models, we explored Mediator kinase function in human colorectal cancer cells. Phospho-proteome profiling revealed substrates enriched with transcription and chromatin regulators, while promoter reporter experiments identified transcription factor binding sites, including TCF/LEF and AP1, regulated by Mediator kinases. Additionally, altered phosphorylation of several Mediator subunits suggests a mechanism for the rapid regulation of the Mediator complex. Overall, our results demonstrate that CDK8 and CDK19 play pivotal roles in regulating gene expression associated with oncogene activation and signaling pathways. Further studies are warranted to elucidate their broader cellular roles and regulatory mechanisms. The selective inhibitors validated in this study will provide valuable tools for such mechanistic investigations into Mediator kinase functions and their potential therapeutic exploitation.

1

Inhibition of CDK8/19 following treatment of COLO205 human colorectal cancer cells with CCT251545 decreases phosphorylation of transcriptional regulators. (A) Chemical structures of active and less active compounds used in this study. IC₅₀ values for in vitro CDK8 binding were determined using a reporter displacement assay for CCT251545 (5 nM), CCT251921 (4.9 nM), CCT251099 (103 nM), MSC2530818 (4.4 nM) and MSC2530819 (590 nM). (B) IC₅₀ values were calculated for 32 human CDK/cyclin pairs using a radiometric protein kinase assay (Supporting Information and Data). The dotted line indicates a 1 μM IC₅₀ threshold. (C) Linear regression plots show the response of COLO205 (n = 3 independent repeats) expressing a TCF/LEF WNT-reporter construct (F1756) following 6 h exposure to compound. Data are fitted using a log­(inhibitor) versus response model with a four-parameter variable slope; symbols and error bars are omitted for simplicity. (D) Plot of enriched phospho-peptides. Label-free intensities were measured in two independent biological experiments each with two technical repeats. Proteins with known links to transcription regulation are highlightedred and blue text indicates phospho-peptides increased and decreased >1.5-fold respectively in both biological repeats after 6 h treatment with 350 nM CCT251545. (E) Hierarchical clustering of label-free intensities of Mediator subunit peptides. Red indicates increased and blue decreased intensities relative to the median. DMSO (vehicle) and CCT251545-treated samples are shown (A/B = biological repeats; 1/2 = technical repeats; * = phosphorylated residue). Predicted phospho-sites are highlighted in red, blue text indicates sites altered by CCT251545 treatment. (F) Summary plot of log₂ fold changes of CCT251545 treatment/DMSO control versus label-free phospho-peptide intensities (red dotted line indicates 1.5-fold decrease).

2

Treatment of COLO205 human colorectal cancer cells with CCT251545 alters transcription factor activity and gene expression. (A) Impact of 350 nM CCT251545 treatment for 6 h on promoter reporter activity in COLO205 cells (n = 4; mean ± s.e.m). The black line represents no change relative to vehicle control, while the red horizontal line highlights reporter activities repressed by >1.5-fold following treatment. The blue column indicates the basal transcriptional reporter, and red columns denote significance (p _adj < 0.001; ANOVA with Sidak’s correction). (B) GSEA of selected gene sets from microarray gene expression profiles of COLO205 cells treated with 350 nM CCT251545 for 2 h (p _adj < 0.05; Supporting Information). Symbol size is proportional to the number of altered genes in each gene set, color indicates the degree of significance and the gene ratio (y-axis) = number of altered genes/total genes in the gene set.

3

Mediator kinase knockout alters STAT1^SER727phosphorylation and promoter activity in SW620 human colon cancer cells. (A) Capillary immunoassay of Mediator kinase module expression and pSTAT1^SER727 biomarker in multiple independent knockout clones. GAPDH was included as a loading control. (B) CDK-substrate motif immunoprecipitation from parent SW620 cells treated for 6 h with 350 nM CCT251545, the less active control (CCT251099), or untreated knockout clones (top gel panel). Controls included IgG alone for background, DMEM for untreated cells, and DMSO as a vehicle control. MED14 was detected as a doublet pair of bands by immunoblotting of cell lysates (middle gel panel). The upper band (indicated by an arrow) is reduced by active compound treatment or CDK8 knockout. The lower panel includes vinculin as a loading control. (C) Promoter reporter results from CDK8 knockout clone 1 relative to parent SW620 cells. Data represent mean ± s.e.m (n = 3). The black horizontal line indicates unchanged reporter activity, while the horizontal red line highlights reporters with altered activity ±1.5-fold. The blue column represents basal transcriptional reporter activity, and the red column indicates significance (p _adj < 0.001; ANOVA with Sidak’s correction). (D) Summary of promoter reporter assays showing significance in one of the assayed conditions (>1.5 = fold, p _adj < 0.001; Figure A and Supporting Figure S5) from COLO205 and SW620 cells treated with CCT251545 (relative to vehicle control) or SW620 knockout clones (relative to parent SW620).

4

CDK8 activity is required for inhibition of colony growth of SW620 human colorectal cancer cells by Mediator kinase chemical probes. (A) Colony assays of parent SW620 cells or three individual clones per knockout. The cells were treated with 350 nM CCT251545 or CCT251099, a structurally related less active control compound, for 10 days (n = 2 independent repeats). (B) Immunoblot of CDK8 expression for CDK8 knockout clone CDK8-KO-cl1 transfected with an empty vector, wild-type CDK8 (CDK8^WT), or the kinase-inactive CDK8^D173A (CDK8^KD) mutant. GAPDH was included as a loading control. (C) Transfected cells from (B) were treated for 10 days with 350 nM of either CCT251545, CCT251921 (from the 3,4,5-trisubstituted pyridine series), or MSC2530818 (from the 3-methyl-1H-pyrazolo­[3,4-b]­pyridine series). Structurally related but less active controls, CCT251099 and MSC2530819, were included for comparison (n = 2).

5

Growth of untreated SW620 human colorectal cancer tumor xenografts and pharmacokinetics of CCT251921. (A) Growth of untreated parent SW620 or CRISPR/Cas9 knockout xenografts. No significant difference in growth rates between parent and knockouts were observed. (B and C) SW620 parental or knockout clones were grown as subcutaneous solid tumor xenografts in NCRI nude mice and treated for 3 days with CCT251921 (42 mg/kg p.o. qd). Plasma and tumors were collected 6 h after the final dose. (B) Phospho-STAT1^SER727 levels relative to total STAT1 protein were quantified in tumors using an electrochemiluminescent ELISA (mean ± s.e.m). An ordinary one-way ANOVA analysis with Dunnett’s correction showed significant differences for basal levels in the parent verses the CDK8 and CDK8/19 clones (all p < 0.0001) and less significance for CDK19cl (p < 0.05) and CDK19-KO-cl2 (p = ns). Treatment of the parent cells and CDK19 clones significantly reduced phospho-STAT1^SER727 levels (p < 0.0001), in contrast the CDK8 and CDK8/19 knockouts showed no significant change, except for CDK8-KO-cl2 (p < 0.05). (C) Plasma concentrations of CCT251921, an ordinary one-way ANOVA found no significant difference between plasma levels in parent and individual knockout clones.

6

Loss of both CDK8 and 19 abrogates the antitumor activity of Mediator kinase chemical probe CCT251921 in SW620 human solid tumor xenografts. Parental SW620 cells or knockout clones were grown as tumor xenografts in NCRI nude mice and treated for 14–20 days with CCT251921 (42 mg/kg p.o qd) for 5/7 days. Three separate clones were run for each knockdown condition (n = 7 control or treated mice per group for the parent and each individual clone). Tumors and plasma were collected 6 h after the final dose. For all graphs an ordinary one-way ANOVA with Dunnett’s correction was run to determine significance. (A) Summary graph of CCT251921 antitumor activity. Each point represents data from a knockout clone. The parent and each of the three CDK8/19 knockout clones showed a significant difference in antitumor response (p < 0.005). (B) Plasma total concentrations of CCT251921 (mean ± s.e.m). With the exception of CDK8-KO-cl2 (p < 0.05), there were no significant differences in CCT251921 plasma levels between the parent and knockout clones. (C) Quantification of tumor phospho-STAT1^SER727 relative to total STAT1 protein following vehicle or CCT251921 treatment (mean ± s.e.m). All CDK8 and CDK8/19 clones had significantly reduced phospho-STAT1^SER727 (p < 0.0001). Treatment with CCT251921 significantly reduced phospho-STAT1^SER727 in the parent, CDK19 (all p < 0.0001) and CDK8 clones (CDK8-KO-cl1 p < 0.001 and CDK8-KO-cl2 p < 0.05), but not the CDK8/19 clones.