CDK9 inhibitors in acute myeloid leukemia

Together with regulatory subunits (cyclins), CDKs form functional complexes responsible for the control of cell proliferation, differentiation, apoptosis, and DNA repair [17]. Whereas many CDKs (eg, CDK1, CDK2, CDK3, CDK4, and CDK6) control cell-cycle progression, ensuring timely and accurate cell replication, others (ie, CDK8 and CDK9) function as gene transcription controllers [18]. CDK9 plays a critical role in controlling global (non-ribosomal) transcription, notably including expression of genes that are regulated by super enhancers, large clusters of DNA regulatory elements (“enhancers”) that drive transcription of genes involved in cell identity [19]. Such genes include MYC, a downstream proto-oncogene involved in cell growth and cell-cycle progression, and MCL-1, an apoptosis regulator. CDK9 also appears to be involved in several physiologic processes in the cell outside of transcription, including differentiation, apoptosis, and signal transduction [15].

CDK9 was first designated PITALRE based on a characteristic amino acid motif (Pro-Ile-Thr-Ala-Leu-Arg-Glu), and its function was first elucidated in studies of human immunodeficiency virus [20, 21]. CDK9 exists in two isoforms, the originally identified major 42 kDa protein (CDK9₄₂) and a minor 55 kDa (CDK9₅₅) protein that is translated from an in-frame mRNA that arises from an upstream transcriptional start site [22, 23].

Both CDK9 isoforms generate a heterodimer with regulatory cyclins T1, T2a, or T2b to form the main component of the positive transcription elongation factor b (P-TEFb) complex that stimulates transcription elongation by phosphorylating the carboxy-terminal domain (CTD) of the largest subunit of RNA polymerase II (RNA Pol II); the CTD contains tandem repeats of a 7 amino-acid sequence that is phosphorylated by CDK7 at Ser5 (YSPTSer5PS) and CDK9 at Ser2 (YSer2PTSPS). Ser5 phosphorylation results in activation of RNA Pol II such that transcription is initiated and Ser2 phosphorylation allows productive transcriptional elongation (Fig. 1 [15]). Therefore CDK9 inhibition prevents productive transcription and is associated with a global reduction in mRNA, including genes, such as MYC and MCL-1, which regulate proliferation and survival of cancer cells [15, 24‐26]. Cyclin K can also interact with CDK9 isoforms in vitro and in vivo, and the CDK9–cyclin K complex can activate transcription when tethered to RNA, but not to DNA, in vitro [24]. Further investigations have shown that CDK9 is involved in co-transcriptional histone modification, mRNA processing, mRNA export, and DNA repair [16, 27, 28].

Although their phosphorylation patterns may be identical, the CDK9 isoforms display differences in subcellular localization and expression patterns, regulation, and tissue distribution [23, 28]. CDK₄₂ has been reported to localize to the nucleoplasm whereas CDK9₅₅ localizes to the nucleolus [23, 26]. Also, CDK₅₅, but not CDK9₄₂, specifically associates with Ku70 and appears to play a role in DNA repair and cell viability through a distinct function [28]. In addition, there are numerous reports of differential expression patterns, including in hematologic cells [25]. Interestingly, CDK9₅₅ is preferentially expressed following induced differentiation of human primary monocytes into macrophages [23], whereas stimulation of human macrophages with lipopolysaccharide, or infection with the human immunodeficiency virus type 1 (HIV-1) increases CDK9₄₂ expression [29]. Also activation of primary lymphocytes results in increased CDK9₄₂ and decreased CDK9₅₅ expression [23]. Taken together, these findings suggest that the function of the two CDK9 isoforms is likely to be at least partially distinct, although further studies are required to produce definitive evidence [26] and importantly in the cancer context, including in AML.

The CDK9-related pathway has emerged as a prioritized target for cancer therapy across a range of tumor types [30]. Multiple studies have shown that a dysregulated CDK9 signaling system may have important implications in the development and/or maintenance of a malignant cell phenotype [30]. Dysregulation of the CDK9 pathway has been observed in a variety of human tumors, which may induce increased expression and/or hyperactivity of cellular oncogenic factors. Studies on cancers, such as lymphoma [31, 32], prostate cancer [33], neuroblastoma [34], and other malignancies [35], demonstrate that CDK9-related pathways are dysregulated, suggesting that CDK9 overexpression promotes the cell proliferation and the synthesis of antiapoptotic factors like MCL-1, BCL,-2 and XIAP [36], which are determinants for cancer-cell survival [37]. The levels of gene products with short half-lives, such as MYC and MCL-1, are reduced most rapidly on exposure to CDK9 inhibitors, thereby leading to reduced cell proliferation and survival.

Regarding CDK9 expression in AML, the HemaExplorer [38] curated database of processed mRNA gene expression profiles provides accessible data on CDK9 expression in hematopoietic cells at different maturation stages. Expression data from distinct subtypes of human AML, defined by karyotypes, are included in the database allowing researchers to directly compare gene expression of leukemic cells with those of their closest normal counterparts. Such data point to a numeric increase in CDK9 mRNA expression in AML samples relative to common myeloid progenitor cells and also illustrates some variation in expression between AML subtypes.

However, it is important to note that overall CDK9 pathway activity is a critical aspect of dysregulation rather than CDK9 expression only. Critically, CDK9-mediated transcription of MCL-1 and MYC plays an important role in growth and survival of cancer cells, and dysregulation of this component of the CDK9 pathway is prominent in a number of hematologic malignancies [17]. The MCL-1 and MYC aspects of the CDK9 pathway have been associated with the pathogenesis of AML. For example, various translocation products of the MLL gene found in leukemias such as AML associate with P-TEFb and constitutively activate transcription [17, 39]. High MCL-1 expression is linked to AML development in murine models [40], and MCL-1 plays a key role in survival and expansion of murine and human AML cells [41]. MCL-1 is also upregulated in about half of cases of R/R AML and is associated with a poor prognosis [42]. Levels of the P-TEFb inhibitor hexamethylene bisacetamide-inducible protein 1 (HEXIM1) are upregulated during differentiation of murine leukemia cells [43], and HEXIM1 has been shown to be involved in the tumorigenesis of AML cell lines bearing the NPMc + mutation, the cytoplasmic-mislocated mutant form of NPM seen in approximately 35% of patients with AML [44]. In addition, HEXIM1 mRNA overexpression is almost mutually exclusive with MYC overexpression in primary AML samples, suggesting that HEXIM1 plays a key role in the growth inhibition and apoptosis of AML cells [45].

Such disease linkage evidence encouraged studies evaluating the potential of CDK9 inhibitors as anticancer therapeutics, initially in pre-clinical models. CDK9 inhibition has been reported to lead to apoptosis in a variety of leukemia and solid tumor cell lines. Notably, the most sensitive cancer cells lines included hematologic tumor cells, especially AML [46], thereby providing functional evidence for the dependency of AML on the CDK9 pathway.

Given CDK9 is a kinase it is considered as a relatively tractable target for drug discovery and provides a route for the indirect targeting of MCL-1 and MYC that may be considered as currently more challenging targets in drug discovery [32].

CDK9 inhibitors have been investigated as therapeutics for a variety of hematologic cancers and solid tumors. Table 1 [46‐59] provides CDK inhibition profiles for CDK9 inhibitors that have reached the clinic or have been evaluated in preclinical studies in AML and other hematologic cancers (Additional file 1: Table S1 summarizes the clinical status of other CDK9 inhibitors across a broader range of tumor types). Current CDK9 inhibitors are competitive inhibitors of the ATP-binding site, which is highly conserved across the CDK family; consequently CDK9 inhibitors lack specificity and generally also inhibit other CDKs to varying extents [60, 61]. Although some progress has been made against other CDKs in discovery of allosteric inhibitors with greater selectivity potential by targeting residues outside the kinase domain (CDK12/13 [62, 63]), no such inhibitors have been described for CDK9. Although they display activity against a variety of CDKs and enzymes, CDK9 inhibitors are referred to as such because they typically exhibit increased half maximal inhibitory concentration (IC₅₀) values for CDK9 compared with other CDKs/enzymes. As described in the sections that follow, CDK9 inhibitors in general exhibit a variety of effects in AML cells and in vivo models, including reduced phosphorylation of RNA Pol II; reduced levels of proteins such as MYC, MCL-1, XIAP, and cyclin D1; induction of apoptosis; and inhibition of tumor growth and prolonged survival in animal models. There is increasing interest in identifying predictive biomarkers of response to conventional and investigational-targeted therapies in AML, including CDK9 inhibitors. For example, measuring the function of B-cell lymphoma 2 (BCL-2) family proteins using BCL-2 homology domain 3 (BH3) profiling has been shown to provide useful information in discriminating AML treatment response with traditional cytarabine-based therapy and investigational AML regimens [64‐68]. The underlying principle of BH3 profiling is that mitochondrial depolarization following exposure to BH3 domain peptides serves as a functional biomarker to predict cell sensitivity to individual antiapoptotic proteins [69]. For example, sensitivity of cells to the NOXA-BH3 peptide provides a direct functional measurement of MCL-1 dependency, whereas sensitivity of cells to BAD-BH3 provides a measurement of BCL-2 dependency.