CDK9 Is a Vulnerability in GATA-3 Driven and MCL-1 Independent T-Cell Lymphomas

Background: The transcription factor GATA-3 is a proto-oncogene in subsets of mature T cell lymphomas (TCL). CDK9 is a transcriptional regulator that, by phosphorylating paused RNA polymerase II (RNAPII), enables efficient transcriptional elongation. Therefore, we were motivated to examine the extent to which CDK9 is required for controlling transcription and expression of GATA-3 and its target genes.

Methods: A pharmacologic strategy, using the selective CDK9 inhibitor AZD4573, was utilized to identify CDK9 dependent transcripts. Traveling ratios, defined as the ratio of RNAPII bound at promoter to RNAPII bound within the gene body, were determined by RNAPII ChIP-seq in cell treated with AZD4573 (or vehicle control). CDK9 dependent genes were defined as those with increased traveling ratio and decreased transcripts. MCL-1 dependency was determined in these cell lines by BH3 profiling. The extent to which CDK9 inhibition transcriptionally reprograms TCL was further examined in complementary ex vivo and in vivo studies using primary TCL specimens and GEM and PDX models.

Results: By conditionally deleting SMARCB1 (SNF5), PTEN, and/or TP53, we generated complementary GEM models that recapitulate characteristics of the genetic landscape recurrently observed in GATA-3 driven TCL. In the GEM and PDX models utilized, CDK9 inhibition significantly impaired TCL progression and prolonged overall survival (p < 0.05). Therefore, we sought to evaluate the mechanism by which CDK9 inhibition impairs TCL growth and survival in MCL-1 independent models. We performed RNAPII ChIP-seq and RNA-seq in an MCL-1 independent cell line (H9) and a PTCL, NOS patient specimen treated with AZD4573 (or vehicle control) to identify CDK9 dependent transcripts. Gene-set enrichment analysis utilizing CDK9-regulated genes by RNA-seq identified representative pathways with established roles in TCL, including cell cycle/proliferation, and genes, including GATA-3 itself. Subsequent ChIP-seq and RNA-seq studies in independent specimens confirmed that GATA-3 transcription is regulated by and dependent upon CDK9 activity. The demonstration that CDK9 inhibition significantly reduced expression of a luciferase reporter drived by GATA-3 promoter also supports this view. We similarly examined the extent to which GATA-3 targets are CDK9 driven. We examined GATA-3 target genes as previously identified by ChIP-seq and RNA-seq, and integrated these prior datasets with the newly identified CDK9-dependent genes we identified. We observed that the majority (>75%) of GATA-3 target genes were also CDK9 dependent and thus their expression were impaired upon CDK9 inhibition in AZD4573-treated cells. A subset of GATA-3 target genes, including those that are therapeutically relevant (e.g. ITK), were selected for further validation in 3 cell lines, 4 PDX models, and primary SS specimens. As CDK9 is apparently a dominant regulator of the GATA-3 dependent transcriptional program, we sought to further investigate the mechanism by which CDK9 regulates the expression of GATA-3 target genes. Co-immunoprecipitation studies demonstrated that GATA-3 interacts with both CDK9 and cyclin T1. Consistent with these findings, we observed significant alignment of both GATA-3 and CDK9/Cyclin T1 binding peaks in TCL ChIP-seq datasets. We also utilized CRISPR/Cas9 technology to knock out (KO) GATA-3 in CTCL cell line and performed GATA3, CDK9, RNAPII, and pRNAPII ChIP-seq. In genome-wide, CDK9, RNAPII and CDK9 binding density were decreased upon GATA-3 KO cells, but restored when GATA-3 was re-expressed in GATA-3 KO cells.

Conclusions: Collectively then, our findings demonstrated that GATA-3 transcription is CDK9 dependent. Consequently, CDK9 inhibition impaired TCL growth in GATA-3 driven (and MCL-1 independent) cells. Furthermore, CDK9 and GATA-3 are binding partners, and CDK9 inhibition significantly impaired transcription of the GATA-3 dependent transcriptome. Therefore, CDK9 inhibition, by impairing expression of the GATA-3 dependent transcriptome, is an attractive therapeutic strategy for GATA-3 driven TCL, many of which are largely resistant to conventional chemotherapeutic approaches.