LAUSR

Dear Editor, Signaling downstream of the pathogenetic mutations in JAK2, MPL, or calreticulin, along with co-occurring mutations in chromatin modifiers or transcription factors, results in dysregulated transcriptome and proteome that is responsible for the transformation of myeloproliferative neoplasms (MPN) to sAML, as well and for conferring therapy refractoriness [1, 2]. Although routinely used and effective in symptomatic and advanced MPN, JAK inhibitor (JAKi), e.g., ruxolitinib, and/or treatments with standard AML chemotherapy are ineffective against postmyeloproliferative neoplasm (MPN) sAML [3]. Binding and activity of transcription factors (TFs), including lineage-specific master regulators and signaling TFs such as STAT3/5, RELA, and MYC, at enhancers and promoters of their targets involves recruitment of transcriptional co-factors and epigenetic regulators, including HAT (histone acetyltransferase), bromodomain extra-terminal (BET) protein (BETP) BRD4 and pTEFb (positive transcript elongation factor b) [4]. Collectively, activities of these TFs and co-factors induce promoter-proximal pause-release of the poised RNA pol II (RNAP2) to stimulate productive mRNA transcript elongation, leading to the dysregulated transcriptome that confers the aggressive phenotype and therapy refractoriness in post-MPN sAML cells [2, 5]. The non-cell-cycle regulatory CDK9 is the catalytic subunit of the positive transcript elongation factor b (pTEFb) [5, 6]. In association with its regulatory subunit cyclin T1, CDK9 mediates phosphorylation of serine 2 in the tandem heptad repeats of the C-terminal domain of RNA pol II (RNAP2), which induces pauserelease of RNAP2. CDK9 also phosphorylates and inactivates the negative transcription regulators NELF (negative elongation factor) and DSIF (DRB sensitivity-inducing factor), thereby further promoting RNAP2-mediated transcription [5, 6]. CDK9 activity is required to maintain constant production of mRNAs of short-lived proteins, e.g., MCL1 and c-Myc, promoting growth and survival of AML cells [6, 7]. Dysregulated c-Myc in AML cells, either due to amplification or protein stabilization, interacts with and recruits pTEFb to its own enhancers and promoter and those of its target genes to mediate RNAP2 pause-release [6, 7]. Of the two protein isoforms of CDK9, the smaller 42 kDa isoform is more abundant in AML cells [6]. Selective CDK9 inhibitors that exhibit high affinity interaction with the ATP binding site of CDK9 inhibit phosphorylation of serine 2 of RNAP2, and CDK9-mediated phosphorylation of the negative regulators of RNAP2-mediated transcription, i.e., NELF and DSIF, thus inhibiting RNAP2-mediated transcription of oncogenes, e.g., c-Myc and MCL-1 [5–7]. In the present studies, we determined effects of two chemically distinct CDK9 inhibitors, BAY-1143572 and NVP2, on the post-MPN sAML cell lines SET-2 and HEL92.1.7 (HEL) and on primary patient-derived (PD) post-MPN sAML cells [8, 9]. Genetic alterations in the cell lines and PD post-MPN sAML cells are presented in Fig. S1A. As shown, NVP2 was more potent than BAY-1143572 to dose-dependently induce apoptosis in SET-2 compared to HEL cells (Fig. 1A). NVP2 and BAY-1143572 also exerted similar levels of lethal activity against the previously reported, in vitro isolated and characterized, ruxolitinib-persister (tolerant)/resistant (P/R) post-MPN sAML SET-2-RuxP and HELRuxP cells, as compared to the parental SET-2 and HEL92.1.7 cells (Fig. 1B and S1B) [10]. SET-2-RuxP and HEL-RuxP cells had been shown to exhibit non-genetic resistance to ruxolitinib, with LD50 values over 2000 nM in RuxP cells compared to 560 nM and 1350 nM for parental SET-2 and HEL92.1.7 cells (Fig. S1C), and cross-resistance to other JAKi [10]. Treatment with NVP2 and BAY-1143572 also dose-dependently induced in vitro loss of viability in multiple samples of PD, CD34+ sAML cells, harvested from patients who had been previously treated with ruxolitinib (Fig. 1C). In contrast, exposure to similar doses and exposure interval to BAY-1143572 or NVP2 did not induce loss of viability in CD34+ normal progenitor cells (Fig. S1D). We also determined effects of the CDK9is on the chromatin and transcriptome of post-MPN sAML cells. Following treatment of SET-2 cells with BAY-1143572 or NVP2, assessment of the accessible chromatin by ATAC-Seq demonstrated large numbers of lost and gained peaks (Fig. 1D) [11]. Rank-sorted TF motifs lost from the chromatin following treatment with BAY-1143572 or NVP2 included those of ERG, PU.1, RUNX1, STAT3/5, and c-Myc (Fig. S2A). Additionally, treatment with BAY-1143572 or NVP2 significantly reduced ATAC-Seq peaks over the superenhancers of MYC, BCL2, and CDK6 (Fig. 1E, S2B, C). A previously reported RNA-Seq analysis had shown that NVP2 treatment repressed vastly greater numbers of mRNA expressions in leukemia cells, with log2 fold-reduction in the mRNAs of MYC, PIM1, MYB, LMO2, NFkB2, MCL1, BIRC3, BCL2L1, BCL2, and CDK6 (Fig. S2D, E) [9]. Utilizing qPCR analysis, we confirmed that NVP2 treatment repressed these same mRNAs, and of SPI1, RELB, PRDM1, and cFLIP in SET-2 and HEL cells (Fig. 1F and S3A). It is noteworthy, that treatment with NVP2, without affecting protein levels for the 42 kDa protein isoform of CDK9, also attenuated protein levels of cyclin T1, p-Rbp1 subunit of RNAP2, c-Myc, XIAP, and MCL1 in SET-2 and HEL92.1.7, as well as in SET-2-RuxP cells (Fig. 1G and S3B). Similar effects on the protein expressions above were also observed following treatment with BAY1143572 in SET-2 and HEL92.1.7 cells (Fig. S3C). These findings underscore that, following treatment with CDK9i, the repression of pro-growth and anti-apoptotic proteins observed could contribute to lowering of the apoptotic threshold and loss of viability of post-MPN sAML cells. We next determined in vivo activity of BAY-1143572 against HEL92.1.7 cells transduced with, and expressing, luciferase, following their infusion via tail vein and engraftment in the immune-depleted NSG mice. After engraftment of HEL92.1.7 cells, mice were treated by oral gavage