DLBCL accounts for >10,000 deaths in the U.S. annually. Despite efficacy of upfront chemo-immunotherapy, up to 50% of patients die within 5 years. While inhibitors of B-cell receptor signaling transformed the therapeutic paradigm in indolent lymphomas, they show poor activity in DLBCL, which thus represents an unmet clinical need.

MYC oncogene, a transcription factor which regulates cell proliferation and growth, is a key driver in lymphomagenesis. Myc overexpression is associated with chemoresistance in NHL, while "double-hit" lymphomas characterized by MYC and BCL2 rearrangement are most aggressive. Myc proteins have extraordinarily short half-life. To regulate transcription, Myc recruits pTEFb, an elongation factor whose catalytic core comprises CDK9/cyclin T complexes. Hence, Myc expression and function may be susceptible to CDK9 inhibition. Non-specific inhibitors of multiple CDKs have shown promise in B-cell malignancies, where their pro-apoptotic effect has been attributed to a reduction in transcription and downmodulation of short lived pro-survival proteins, e.g. MCL1. However, such inhibitors lack a therapeutic window or a refined mechanism of action. Here we selectively targeted CDK9 in a pre-clinical study in DLBCL and demonstrate that this therapeutic approach disrupts Myc oncogenic function.

Experiments were conducted in activated B-cell(ABC)-type (OCI-LY3, U2932, NuDUL-1) and germinal center(GC)-type (OCI-LY18/19, SUDHL4/6/10/16, VAL) DLBCL cell lines and in primary lymph node-derived B-cells from patients with refractory DLBCL. Tumor xenografts were established in NSG mice. We used AZ5576, a novel selective CDK9 inhibitor (Astra Zeneca), pan-CDK inhibitors (flavopiridol, dinaciclib) and genetic knockdown of CDK9.

Treatment with AZ5576 promoted rapid apoptosis induction (within 2 hours) in DLBCL cell lines and primary tumor cells, as evident by PARP cleavage. Quantification of Annexin V+ cells after 24 hour drug exposure rendered an IC50 of 300-500 nM in DLBCL cell lines independent of cell of origin and 100 nM in primary DLBCL cells. Loss of RNA polymerase phosphorylation at Ser2 residue, a CDK9-dependent site, preceded apoptosis. Contrary to pan-CDK inhibitors, AZ-5576 did not inhibit CDKs 2, 4/6, and 7 (as measured by pRb and pRNApol [Ser5]), confirming its selective effect.

CDK9 inhibition resulted in rapid (within 1 hour) and dose-dependent downregulation of Myc and MCL1 mRNA and protein in DLBCL cell lines and primary cells. Cell cycle profiling demonstrated that treatment with AZ-5576 led to S phase reduction, consistent with loss of Myc, while pan-CDK inhibitors downmodulated S/G2/M, reflecting inhibition of multiple CDKs. Cyclohexamide chase experiments demonstrated enhanced degradation of Myc in DLBCL cells treated with AZ-5576. Myc protein degradation is tightly regulated: while phosphorylation at Ser62 enhances Myc stability, MycT58 is primed for degradation. CDK9 inhibition decreased pMycS62 prior to apoptosis induction in DLBCL cell lines. Raji cells, which carry a stabilizing MycT58A mutation, were less sensitive to AZ-5576. This suggests that CDK9 may govern Myc protein turnover, thus regulating its expression through multiple mechanisms.

Overexpression of Myc was associated with enhanced sensitivity to CDK9 inhibition. Treatment with 300 nM AZ-5576 for 24 h induced apoptosis of 35-70% of Myc-high cells (e.g., U-2932, VAL), compared with 10-20% of Myc-low cells (e.g., OCI-LY3, SU-DHL10). CDK9 inhibition did not disrupt Myc-CDK9 complex formation. Despite this, it abrogated Myc transcriptional activity in a functional reporter assay. Meanwhile, sensitivity to AZ-5576 did not correlate with expression of MCL1, BCL2 or BCLX.

Our findings were confirmed in a mouse xenograft model. VAL cells were inoculated subcutaneously and treatment with AZ-5576 (30 mg/kg orally twice weekly) or vehicle control began when tumors reached 10 mm in size. Treatment led to reduced tumor progression (Figure), prolonged survival, decreased cell proliferation (Ki-67), enhanced apoptosis (Annexin V), and reduction in Myc total and p-S62 levels and MCL1.

Thus, our data suggests that targeting CDK9 is a promising therapeutic strategy poised to disrupt Myc oncogenic activity in DLBCL and provides rationale for clinical development of AZ-5576.


Chen:Seattle Genetics: Consultancy. Drew:Astra Zeneca: Employment. Danilov:ImmunoGen: Consultancy; Pharmacyclics: Consultancy; Astra Zeneca: Research Funding; Prime Oncology: Honoraria; Dava Oncology: Honoraria; GIlead Sciences: Research Funding; Takeda: Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.