Cyclin-dependent kinase 8 mediates chemotherapy-induced tumor-promoting paracrine activities

Abstract

Conventional chemotherapy not only kills tumor cells but also changes gene expression in treatment-damaged tissues, inducing production of multiple tumor-supporting secreted factors. This secretory phenotype was found here to be mediated in part by a damage-inducible cell-cycle inhibitor p21 (CDKN1A). We developed small-molecule compounds that inhibit damage-induced transcription downstream of p21. These compounds were identified as selective inhibitors of a transcription-regulating kinase CDK8 and its isoform CDK19. Remarkably, p21 was found to bind to CDK8 and stimulate its kinase activity. p21 and CDK8 also cooperate in the formation of internucleolar bodies, where both proteins accumulate. A CDK8 inhibitor suppresses damage-induced tumor-promoting paracrine activities of tumor cells and normal fibroblasts and reverses the increase in tumor engraftment and serum mitogenic activity in mice pretreated with a chemotherapeutic drug. The inhibitor also increases the efficacy of chemotherapy against xenografts formed by tumor cell/fibroblast mixtures. Microarray data analysis revealed striking correlations between CDK8 expression and poor survival in breast and ovarian cancers. CDK8 inhibition offers a promising approach to increasing the efficacy of cancer chemotherapy.

Fig. 1.

Effects of Senexin A on the induction of transcription by IPTG-inducible p21. (A) Structures of some SNX2-class compounds. SNX2 and SNX14 were isolated through HTS; Senexin A and SNX2-1-108 were generated through chemical optimization. (B) Effects of Senexin A on CMV-GFP expression in HT1080-p21-9 cells, untreated or treated for 48 h with 50 μM IPTG (quadruplicate assays). y axis: GFP fluorescence normalized by Hoechst 33342 DNA staining (a measure of relative cell number). (C) Same as in A, except that GFP fluorescence was normalized by sulphorhodamine B staining (a measure of protein amount). (D) Effects of Senexin A on GFP expression from NF-κB–dependent consensus promoter in HT1080 p21-9 cells, untreated or treated for 72 h with 50 μM IPTG. y axis: Mean GFP fluorescence per live cell (measured by flow cytometry). (E) Immunoblotting analysis of p21 protein in HT1080 p21-9 cells, untreated or treated with 50 μM IPTG and 5 μM Senexin A, singly and in combinations. (F) Effects of Senexin A on the inhibition and induction of gene expression in HT1080-p21-9 cells treated with 50 μM IPTG alone (x axis) or with 50 μM IPTG and 5 μM Senexin A (y axis) for 48 h (microarray data). Fold changes in gene expression are plotted as log2; genes showing IPTG-induced fold changes with log2 < 0.5 are excluded. See Fig. S2 for quantitative PCR confirmation of gene expression changes. (G) Box-whisker plots of fold changes in the expression of all of the genes in the indicated GO categories in cells treated with Senexin A, IPTG, or IPTG plus Senexin A.

Fig. 2.

Identification of CDK8 as a mediator of p21-induced transcription and target of SNX2-class compounds. (A) Effects of 10 μM SNX14 on the activity of 442 kinases, measured by ATP binding competition assay. The kinases are displayed in the form of an evolutionary dendrogram; red circles indicate the inhibited kinases. (B) Same analysis as in A conducted with 2 μM SNX2-1-108. (C) Effect of Senexin A on CDK8/cyclin C kinase activity in a cell-free assay. (D) Luciferase expression from β-catenin–dependent promoter (TOPflash) or its β-catenin–independent version (FOPflash) in HCT116 cells, after 48 h treatment with Senexin A (quadruplicate assays). (E) Quantitative PCR analysis of EGR1 mRNA expression in HT1080 p21-9 cells upon serum starvation for 48 h, followed by readdition of serum for the indicated periods of time, in the presence or absence of 5 μM Senexin A (triplicate assays). Primer sequences are listed in Table S3. (F) Mean fluorescence intensity of CMV-GFP expression in live (propidium iodide-negative) cells, infected with the indicated dilutions of pLKO.1 lentiviral vector (control) or with pLKO.1-expressing shRNAs targeting CDK8 or CDK19, singly or in combination, with or without 3-d treatment with 50 μM IPTG. ***Decrease relative to cells infected with control lentivirus, with P = 0 (two-tailed t test). (G) CMV-GFP expression in cells, untransduced or transduced with a control lentivirus or with lentiviruses expressing two different shRNAs against CDK8, with or without 48-h treatment with 50 μM IPTG or 5 μM Senexin A, alone or in combination. Bars represent fold changes in mean fluorescence intensity of the GFP-expressing live cells in test samples relative to the same untreated cells, in biological triplicates. (H) Coimmunoprecipitation analysis of CDK8, Med12, cyclin C, and p21 in HT1080 p21-9 cells, untreated, or treated for 48 h with 10 μM Senexin A and 50 μM IPTG, singly or in combination. (I) Dose-dependent effects of recombinant p21 on CDK2/cyclin E and CDK8/cyclin C kinase activities in cell-free assays (a representative from three experiments).

Fig. 3.

Effects of p21 and CDK8 inhibitor on paracrine tumor-promoting activities. (A) Survival of luciferase-labeled C8 cells in low-serum media without coculture or in coculture with wild-type, p53-null, or p21-null HCT116 cells, untreated or pretreated for 72 h with 150 nM doxorubicin (quadruplicate assays). (B) Same assays as in A conducted with wild-type and p21-null HCT116 cells in the presence or in the absence of 5 μM Senexin A. (C) Same assays as in A and B conducted with wild-type HCT116 cells, untreated or pretreated for 72 h with 150 nM doxorubicin in the absence or presence of Senexin A (eight replicate assays). (D) Effects of Senexin A on p21 expression in HCT116 cells, untreated or treated for 72 h with 150 nM doxorubicin. (E) Effects of Senexin A on p21 expression in WI38 fibroblasts, untreated or exposed 3 d earlier to 10 Gy ionizing radiation. (F) Same assays as in A–C conducted with human WI38 fibroblasts, untreated or exposed 3 d earlier to 10 Gy ionizing radiation, in the absence or presence of Senexin A (triplicate assays). (G) A549 cell growth in conditioned media from MEF that were either untreated or treated for 24 h with 200 nM doxorubicin, singly or in combination with 5 μM Senexin A (triplicate assays). MEF-conditioned media were collected 48 h after removing the drugs; A549 cells were counted after 48 h.

Fig. 4.

Effects of CDK8 inhibitor and clinical correlations of CDK8 expression in vivo. (A) Effect of pretreatment with doxorubicin, with or without Senexin A, on A549 xenograft tumor engraftment in SCID mice. Mice were either untreated (n = 6) or treated with a single i.p. injection of 4 mg/kg doxorubicin, followed by five daily i.p. injections of either carrier (n = 6) or 20 mg/kg Senexin A (n = 8). A total of 2 × 10⁶ A549 cells were injected s.c. 5 d after doxorubicin injection. The time when tumors became detectable by palpation was recorded. (B) Effects of sera from mice that were untreated (n = 9) or pretreated i.p. with doxorubicin, with (n = 8) or without (n = 5) 5-d treatment with Senexin A, on the growth of A549 cells in culture. Serum was isolated 5 d after the initiation of therapy. Cell number was assessed 48 h after the addition of mouse sera (each sample assayed in triplicate). (C) Survival of xenograft tumors formed in SCID mice by A549 mixed with MEF (1:1). Once tumors became palpable, mice were treated with a single i.p. injection of 4 mg/kg doxorubicin, followed by five daily i.p. injections of either carrier or 20 mg/kg Senexin A. The time until tumors became undetectable by palpation was recorded. (D) Correlations of CDK8, CDK19, and CCNC expression with relapse-free patient survival in microarray data from 2,897 breast cancers, determined using an online survival analysis tool. Kaplan–Meier correlations with relapse-free survival are plotted for high (above-median) and low (below-median) expression of each gene.