Cyclin-dependent Kinase 9 as a Potential Target for Anti-TNF-resistant Inflammatory Bowel Disease

Abstract

Background & aims: Resistance to single cytokine blockade, namely anti-tumor necrosis factor (TNF) therapy, is a growing concern for patients with inflammatory bowel disease (IBD). The transcription factor T-bet is a critical regulator of intestinal homeostasis, is genetically linked to mucosal inflammation and controls the expression of multiples genes such as the pro-inflammatory cytokines interferon (IFN)-γ and TNF. Inhibiting T-bet may therefore offer a more attractive prospect for treating IBD but remains challenging to target therapeutically. In this study, we evaluate the effect of targeting the transactivation function of T-bet using inhibitors of P-TEFb (CDK9-cyclin T), a transcriptional elongation factor downstream of T-bet.

Methods: Using an adaptive immune-mediated colitis model, human colonic lymphocytes from patients with IBD and multiple large clinical datasets, we investigate the effect of cyclin-dependent kinase 9 (CDK9) inhibitors on cytokine production and gene expression in colonic CD4⁺ T cells and link these genetic modules to clinical response in patients with IBD.

Results: Systemic CDK9 inhibition led to histological improvement of immune-mediated colitis and was associated with targeted suppression of colonic CD4⁺ T cell-derived IFN-γ and IL-17A. In colonic lymphocytes from patients with IBD, CDK9 inhibition potently repressed genes responsible for pro-inflammatory signalling, and in particular genes regulated by T-bet. Remarkably, CDK9 inhibition targeted genes that were highly expressed in anti-TNF resistant IBD and that predicted non-response to anti-TNF therapy.

Conclusion: Collectively, our findings reveal CDK9 as a potential target for anti-TNF-resistant IBD, which has the potential for rapid translation to the clinic.

Keywords: CDK9; Crohn’s Disease; IBD; Inflammation; Ulcerative Colitis.

Graphical abstract

Figure 1

CDK9 inhibition suppresses IFN-γ and TNF-α production in murine and human CD4⁺T cells.A, Proportions of murine CD4⁺ Th1 cells positive for intracellular cytokines following CDK9 inhibition with flavopiridol (FP) relative to untreated (DMSO control). Mean ± standard error of the mean (SEM); n = 5. Paired Student t test. B, Cytokine concentration in the supernatants of TCT colitis explants following 27-hour treatment with FP. Mean ± SEM; IFN-γ assays, n = 4; TNF-α assays, n = 6. Mann-Whitney U test. C, Immunoblots for RNA pol II S2P and total RNA pol II in PB CD4⁺ T cells purified from a patient with UC and a patient with CD and re-stimulated for 3 hours in the presence of DMSO, FP, NVP-2, or AT7519. D, RNA pol II S2P and total RNA pol II levels, normalized to β-actin. Mean ± SEM; n = 2. E, Proportions of human patient with IBD PB CD4⁺ T cells positive for intracellular cytokines following 3-hour treatment with AT7519 or NVP-2 relative to untreated (DMSO control). Mean ± SEM; n = 4. F, IFN-γ and TNF-α concentration in supernatant from patients with IBD PB CD4⁺ T cells restimulated for 16 hours in the presence of AT7519 or NVP-2. Mean ± SEM; n = 2.

Figure 2

CDK9 inhibition leads to histological improvement of TCT colitis.A, Percentage change from initial bodyweight (IBW) of mice following intraperitoneal injection of 0.5 ×10⁶ WT CD4⁺ CD25^- CD62L⁺ CD44^low T cells. Mean ± standard error of the mean; TCT colitis (n = 28), PBS (n = 5), 1% DMSO (n = 11), 3 mg/kg flavopiridol (FP, n = 5), 1 mg/kg NVP-2 (n = 10). Two-way analysis of variance. B, Normalized colon weights (colon weight [g] / bodyweight [g]) of control mice that received PBS instead of T cells (n = 7) compared with TCT mice following treatment with DMSO (n = 16), FP (n = 5), or NVP-2 (n = 10). Mann-Whitney U test. C, Histological colitis scores of TCT mice following treatment with DMSO (n = 16), FP (n = 5), or NVP-2 (n = 10). One-tailed Mann-Whitney U test. D, Representative colon micrographs for control mice (PBS) compared with DMSO-, FP-, or NVP-2-treated TCT mice (hematoxylin and eosin stained). Scale bars 250 μm. E, Proportions of CD4⁺ T cells positive for intracellular cytokines from colons, mesenteric lymph nodes (mLNs), or spleens of mice treated with 1 mg/kg NVP-2 (n = 10) or 1% DMSO (n = 8). Mann-Whitney U test.

Figure 3

Short treatments of CDK9 inhibition do not cause cell cycle arrest.A, Proportion of CD4⁺ LPMCs in G1, S, and G2 phase following activation with anti-CD3 and anti-CD28 for 48 hours and 3-hour, 6-hour, or 24-hour treatment with 1 or 4 mM AT7519. Activated control represents CD4⁺ LPMCs cultured with anti-CD3 and anti-CD28 alone for 48 hours. Resting control represents CD4⁺ LPMCs cultured without anti-CD3 and anti-CD28. Mean ± standard error of the mean; n = 6. Unpaired 2-tailed Student t test estimating significance of differences in % cells in G1. B, Proportion of live colonic LPMCs following 3 hours, 6 hours, and 24 hours of treatment with AT7519. Mean ± standard error of the mean, n = 6. Unpaired 2-tailed Student t test.

Figure 4

CDK9 inhibition suppresses transcription of Th1 and Th17 cytokines.A, Proportions of CD4⁺ LPMCs positive for intracellular cytokines following 3-hour incubation with 10 μM flavopiridol (n = 14), 4 μM AT7519 (n = 21), or 1 μM NVP-2 (n = 10) relative to untreated (DMSO control). Median ± interquartile range. B, Dose-response curves for intracellular cytokines in CD4⁺ LPMCs following 3-hour treatment with flavopiridol (n = 5), AT7519 (n = 6), or NVP-2 (n = 5); Mean ± standard error of the mean. C, Proportions of CD4⁺ LPMCs positive for intracellular cytokines following 3-hour incubation with 10 μM flavopiridol (n = 14), 4 μM AT7519 (n = 21), or 1 μM NVP-2 (n = 10) relative to untreated (DMSO) control. Results stratified by patients on anti-TNF therapy with active disease (anti-TNF resistant; flavopiridol, n = 5; AT7519, n = 5; and NVP-2, n = 2), those with quiescent disease (anti-TNF responsive; AT7519, n = 2; NVP-2, n = 2), and those not on anti-TNF therapy (flavopiridol, n = 8; AT7519, n = 14; and NVP-2, n = 6) at the time of endoscopy. Mean ± standard error of the mean. Two-way analysis of variance. D, RT-qPCR measuring TNF, IFNG, IL17A, and IL22 transcripts following 3-hour treatment with 4 μM AT7519 or 1 μM NVP-2. Mean ± standard error of the mean; TNF, IFNG, and IL17A, n = 7; IL22 n=4. Mann-Whitney U test.

Figure 5

Restimulated colonic CD4⁺T cells are highly responsive to CDK9 inhibition.A, Proportions of purified colonic CD4⁺ T cells positive for intracellular cytokines following primary stimulation (PS) or restimulation (RS). Median ± interquartile range; PS – IFN-γ & TNF-α (n = 10), IL-17A (n = 7), IL-22 (n = 6). RS - IFN-γ, TNF-α, IL-17A, IL-22 (n = 10). B, Proportions of purified colonic CD4⁺ T cells positive for intracellular cytokines following treatment with CDK9 inhibitors during PS or RS relative to untreated (DMSO control). 2μM AT7519 (AT) (n = 4), 4 μM AT7519 (AT) (n = 10), 1 μM NVP-2 (n = 10), 2 μM NVP-2 (n = 4). Two-way analysis of variance with the Šidák multiple comparisons test. C, RT-qPCR measuring IFNG and TNF expression in colonic CD4⁺ T cells following 3-hour treatment with AT7519 or NVP-2 during PS or RS. Mean ± standard error of the mean; n = 3. Unpaired 1-tailed Student t test. D, Heat map depicting changes in transcript levels for the top 500 genes ranked by inter-sample variability in NVP-2 treated and untreated restimulated colonic CD4⁺ T cells purified from 1 patient with UC and 2 patients with CD. E, Differential expression of genes specific to naïve (n = 19) or memory (n = 27) phenotypes in colonic CD4⁺ T cells following 72-hour activation with anti-CD3 and anti-CD28. Mean ± standard error of the mean. Mann-Whitney U test. F, Volcano plots of differential gene expression in cells treated with NVP-2 during PS or RS compared with cells left untreated during PS or RS. Blue dots represent genes upregulated (log₂ fold-change ≥2; FDR <0.01) following PS or RS of untreated cells. G, Correlation between the degree of gene activation during PS and RS of untreated cells and the degree of gene repression by NVP-2 during PS and RS. Pearson correlation with t test. H, Degree of transcriptional repression caused by treatment of colonic CD4⁺ T cells with NVP-2 during PS compared with treatment during RS. Pearson correlation and t test (left) and violin plot and Mann-Whitney U test (right).

Figure 6

Genes bound by P-TEFb and T-bet are more susceptible to CDK9 inhibition.A, GSEA of genes bound by P-TEFb in Th1 cells compared with changes in gene expression caused by NVP-2 treatment during primary stimulation (PS) or restimulation (RS) of colonic CD4⁺ T cells. B, Correlation between the change in P-TEFb occupancy upon cell stimulation and change in gene expression following NVP-2 treatment during PS and RS. n = 1727. Spearman correlation. C, Change in expression of P-TEFb bound genes following NVP-2 treatment during PS compared with RS. Median ± interquartile range, n = 936. Mann-Whitney U test. D, GSEA of genes associated with T-bet-bound enhancers compared with changes in gene expression caused by NVP-2 treatment during PS or RS of colonic CD4⁺ T cells.

Figure 7

CDK9i-repressed transcripts are enriched in anti-TNF resistant IBD.A, Ingenuity pathway analysis showing enrichment of genes repressed by NVP-2 (defined as log₂ FC <−2 and FDR <0.01 in NVP-2 vs untreated restimulation [RS] cells) in inflammatory signaling and cancer pathways. Bars indicate −log₁₀ (P-value), and dots indicate z-score. B, GSVA of NVP-2-repressed genes and genes insensitive to NVP-2 (defined as log₂ FC = 0 in NVP-2/RS) in colonic CD and UC compared with healthy controls (HCs). Results for 3 independent IBD datasets (Mucosal Gene Expression Defects in IBD [GSE16879], PURSUIT-SC [GSE92415], and PROTECT [GSE109142]). Unpaired Student t test. C, GSVA of NVP-2-repressed and insensitive genes in anti-TNF responders (R) vs non-responders (NR), and remission (Rem) or active disease (Act) at week 4. Results for 3 independent IBD datasets outlined above. Unpaired Student t test. D, Receiver operator characteristic analysis of NVP-2-repressed genes and genes insensitive to NVP-2 in UC and colonic CD, distinguishing infliximab responders and non-responders in the Mucosal Gene Expression Defects in IBD data set and golimumab responders and non-responders in the PURSUIT-SC data set. E, Correlation of enrichment score of NVP-2-repressed genes with clinical parameters including fecal calprotectin (fCal), PUCAI, Mayo score, and histology score. Spearman correlation and t test.