MYC dysregulation confers a poor prognosis to diffuse large B-cell lymphoma (DLBCL), and effective therapeutic strategies are lacking in relapsed/refractory DLBCL, Burkitt lymphoma and intermediate forms.1, 2 As a master transcriptional regulator, MYC recruits transcription complexes containing RNA polymerase II (Pol II) to facilitate effective transcriptional elongation of MYC gene targets.3 Pol II is fully activated by phosphorylation of a critical serine residue at position 2 within heptapeptide repeats in the carboxy-terminal domain (CTD), a function performed by the positive transcription elongation factor b (P-TEFb; comprising CDK9 and cyclin T1).4 It has been shown that MYC binds and recruits P-TEFb to its targets as a means to activate Pol II.3, 5, 6 More recently, CDK9-mediated transcriptional elongation was reported as essential for tumor maintenance in a genetically defined MYC-driven model of hepatocellular carcinoma.7 Thus, CDK9 dependence may represent a druggable vulnerability in lymphomas with dysregulated MYC expression.

Dinaciclib (Merck, Boston, MA, USA) is a novel CDK inhibitor that has reached phase 1b/2 of clinical trials for a range of solid-organ malignancies, as well as for myeloma and chronic lymphocytic leukemia.8 We hypothesized that CDK9 inhibition by dinaciclib would represent a rational pharmacologic approach to target the transcription of critical MYC-regulated oncogenic effector proteins. Here we describe durable in vivo responses to dinaciclib in aggressive MYC-driven lymphoma, mediated by downregulation of Pol II-mediated Mcl-1 transcription.

Dinaciclib has 50% kinase inhibitory concentrations of 1, 1, 3 and 4 nM for CDK2, CDK5, CDK1 and CDK9, respectively.8 Dinaciclib potently killed Eμ-Myc and human IG-cMYC-translocated cell lines independent of p53 function, but not untransformed murine fibroblast cells, at low nanomolar concentrations approximating those observed for kinase inhibition (Figures 1a and b, Supplementary Figure S1).

Figure 1
figure 1

Dinaciclib potently induces apoptosis of murine Eμ-Myc and human IG-cMYC-translocated lymphomas with rapid and selective suppression of Mcl-1 transcription and protein levels. (a) Wild-type p53 (#4242) and p53-null (#3391) Eμ-Myc lymphomas were cultured in vitro with dimethylsulfoxide (DMSO) vehicle control or dinaciclib for 24 h and then analyzed using flow cytometric analysis for annexin-V/propidium iodide (PI) positivity. (b) Human IG-cMYC-translocated BL-41 and Ramos cell lines were cultured in vitro with DMSO or dinaciclib for 48 h before the analysis of annexin-V/PI positivity using flow cytometry. (c) Mcl-1 and Bcl-2 mRNA expression in lymphoma #4242 following 3-h in vitro treatment with DMSO or 20 nM dinaciclib. Transcript levels are represented as fold change compared with DMSO. NS, not significant; *P<0.0001. (d) Chromatin immunoprecipitation-PCR of Eμ-Myc lymphoma #4242 cells showing binding of phospho-RNA Pol II CTD serine 2 (pRpb1 Ser2) at the Mcl-1 locus. Error bars denote the s.e.m. from three independent primer sets across the Mcl-1 locus. (e) Eμ-Myc lymphoma #4242 was cultured in vitro for 3-h untreated or in the presence of DMSO or 20 nM dinaciclib before the preparation of lysates and immunoblotting for phospho-RNA Pol II CTD (pRpb1Ser2, pRpb1Ser5 and pRpb1Ser2/5), total Mcl-1, Bcl-2, Bcl-xL, c-Myc and HSP90 loading control. (f) Human IG-cMYC-translocated BL-41 and Ramos cell lines were cultured in vitro for 3 h in the presence of DMSO or 20 nM dinaciclib before the preparation of lysates and immunoblotting for total Mcl-1, Bcl-2, Bcl-xL, c-Myc, Tubulin and HSP90 loading controls. (g) Eμ-Myc lymphoma #4242 was transduced with murine stem cell virus expressing empty vector control or Mcl-1 and then cultured in vitro with dinaciclib for 24 h before flow cytometric analysis for annexin-V/PI positivity. **P<0.01 comparing treatments at 16 nm concentration. All graphs represent the mean±s.e.m (error bars) for three or more independent experiments.

As Bcl-2 and Mcl-1 have been implicated as important apoptotic regulators in Eμ-Myc lymphomas,9, 10 we assessed the effects of dinaciclib on these proteins. We hypothesized that CDK9 inhibition with dinaciclib would target Mcl-1 transcription, as has been observed with other CDK inhibitors in myeloma and mantle cell lymphoma.11, 12 Eμ-Myc and human IG-cMYC-translocated cell lines were treated with dinaciclib or dimethylsulfoxide control and interrogated using the quantitative PCR analysis for the effect on Mcl-1 and Bcl-2 mRNA. Dinaciclib treatment was associated with a significant reduction in Mcl-1 mRNA, with no significant effect on Bcl-2 transcript levels (Figure 1c, Supplementary Figure S2). Chromatin immunoprecipitation-PCR was used to show the binding of phosphorylated Pol II, subunit B1 carboxy-terminal domain (CTD) serine 2 (pRpb1 Ser2) as a marker of CDK9 activity at the Mcl-1 locus in a representative Eμ-Myc lymphoma cell line (Figure 1d). These findings support the hypothesis that dinaciclib transcriptionally downregulates Mcl-1.

We next examined Mcl-1 expression in Eμ-Myc and human IG-cMYC-translocated lymphoma cell lysates following the treatment with dinaciclib or vehicle. On-target CDK9 inhibition by dinaciclib was confirmed through inhibition of pRpb1 Ser2 at concentrations corresponding to apoptosis induction in Eμ-Myc cells (Figure 1e). Dinaciclib treatment also rapidly suppressed Mcl-1 protein expression, with no discernible reduction in Bcl-2 or Bcl-xL protein observed in murine (Figure 1e) or human (Figure 1f) cells. To determine the functional importance of Mcl-1 in regulating dinaciclib-mediated apoptosis, a representative Eμ-Myc lymphoma was stably transduced to express Mcl-1 off a retroviral promoter. As shown in Figure 1g, exogenously expressed Mcl-1 significantly protected Eμ-Myc cells from dinaciclib-induced apoptosis.

The in vivo efficacy of dinaciclib was then assessed by transplanting the same Eμ-Myc lymphomas into cohorts of syngeneic C57Bl/6 recipients. Compared with the vehicle control, dinaciclib treatment was well tolerated and associated with a highly significant survival advantage of tumor-bearing mice, including those bearing a p53-null lymphoma and a lymphoma with a spontaneous p53 mutation encoding a dominant-negative p53 protein (Figures 2a–c, Supplementary Figure S3). In contrast, dinaciclib-mediated therapeutic efficacy was severely attenuated in isogeneic p53-competent Eμ-Myc lymphoma overexpressing Mcl-1 (Figure 2d). In separate experiments, mice bearing transplanted Eμ-Myc cells were left untreated for 12 days to establish bulky nodal disease, at which time they received a single dose of dinaciclib or vehicle 1 or 4 h before being killed and before the lymph nodes were harvested. Consistent with the in vitro data, lymph node protein lysates showed reductions of pRpb1 and total Mcl-1 protein (Figure 2e), concomitant with the induction of apoptosis (Supplementary Figure S4). Finally, dinaciclib treatment of immunocompromised mice xenografted with the human IG-cMYC-translocated lymphoma was associated with reduced disease progression and significantly prolonged overall survival (Figures 2f and g).

Figure 2
figure 2

Dinaciclib therapy prolongs the survival of mice bearing Eμ-Myc and human IG-cMYC-translocated lymphomas. (a–d) Kaplan–Meier survival curves representing cohorts of C57Bl/6 mice transplanted with representative Eμ-Myc lymphomas 3 days before the therapy commencement with 20% hydroxypropyl-beta-cyclodextran (HPBCD) vehicle or 30 mg/kg dinaciclib by intraperitoneal injection twice weekly. Gray shading denotes the period of therapy. dn, dominant negative; P<0.0001 for each experiment. The median survival for vehicle- and dinaciclib-treated mice were 12 days and not reached (#4242), 16 and 48 days (#3391), 18 and 66 days (#106) and 13 and 18 days (#4242tMcl-1), respectively. (e) Lymph nodes were harvested from cohorts of C57Bl/6 mice 1 or 4 h following a single dose of dinaciclib or 20% HPBCD, 12 days following transplantation with Eμ-Myc lymphoma #4242. Protein lysates were then prepared and separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis before immunoblotting for the indicated targets. Each lane represents protein lysate from the lymph nodes of an individual mouse. (f) Bioluminescence imaging of NOD-scid IL2Rγnull mice transplanted with human IG-cMYC-translocated BL-41.luc2 lymphoma 3 days before the commencement of the therapy with 20% HPBCD vehicle or 30 mg/kg dinaciclib by intraperitoneal injection twice weekly. Mice were imaged at 7 and 14 days post transplantation. (g) Overall survival of the mice from the experiment is described in f. Gray shading denotes the period of therapy. The median survival for vehicle and dinaciclib-treated mice were 19 and 26 days, respectively (P<0.001).

In conclusion, our findings indicate that CDK9 inhibition by dinaciclib is highly effective in aggressive MYC-driven lymphomas, including ‘poor-risk’ p53-deficient clones, via selective inhibition of critical MYC targets including Mcl-1 (which is currently undruggable with existing BH3 mimetics).13, 14 Our data suggest a linear and druggable dependency between MYC and Mcl-1 that is contingent on CDK9 signaling. These findings are of particular interest in the context of a recent publication by Kelly et al.,15 further highlighting the dependency of MYC-driven B-cell lymphoma to Mcl-1. Rapid clinical translation of CDK9 inhibitors to MYC-dysregulated lymphoid malignancy should now be considered.