Deregulations in the Cyclin-Dependent Kinase-9-Related Pathway in Cancer: Implications for Drug Discovery and Development

The CDK9-related pathway is an important regulator of mammalian cell biology and is also involved in the replication cycle of several viruses, including the human immunodeficiency virus type 1. CDK9 is present in two isoforms termed CDK9-42 and CDK9-55 that bind noncovalently type T cyclins and cyclin K. This association forms a heterodimer, where CDK9 carries the enzymatic site and the cyclin partner functions as a regulatory subunit. This heterodimer is the main component of the positive transcription elongation factor b, which stabilizes RNA elongation via phosphorylation of the RNA pol II carboxyl terminal domain. Abnormal activities in the CDK9-related pathway were observed in human malignancies and cardiac hypertrophies. Thus, the elucidation of the CDK9 pathway deregulations may provide useful insights into the pathogenesis and progression of human malignancies, cardiac hypertrophy, AIDS and other viral-related maladies. These studies may lead to the improvement of kinase inhibitors for the treatment of the previously mentioned pathological conditions. This review describes the CDK9-related pathway deregulations in malignancies and the development of kinase inhibitors in cancer therapy, which can be classified into three categories: antagonists that block the ATP binding site of the catalytic domain, allosteric inhibitors, and small molecules that disrupt protein-protein interactions.

This paper focuses on the relevance of the deregulations in CDK9-related pathway in cancer and on the standpoint of the discovery and development of CDK inhibitors for cancer therapy.

Biological Functions of the CDK9-Related Pathway
The CDK9-related pathway is one of the primary factors in the regulation of mammalian cell biology [9,13,25,27,28]. Studies in the human and murine systems showed that CDK9 is present in all kinds of tissues and is highly expressed in terminally differentiated cells [2,3,9]. As anticipated,  there are two CDK9 isoforms in mammalian cells: CDK9-42 and CDK9-55 [4][5][6]. The CDK9-55 isoform has 117 additional amino acid residues in front of the amino terminus of CDK9-42 [4]. The genetic organization of the human CDK9 promoter comprises two transcription starts [4,5]. The human CDK9 promoter that encodes for CDK9-42 mRNA does not have a functional TATA box and contains a GC-rich sequence, and the region −352 to −1 includes the required transcriptional elements to sustain full promoter activity. These factors result in constitutive high levels of human CDK9-42 promoter activity, which is similar to a housekeeping gene promoter [7]. In contrast, the human CDK9-55 promoter contains a TATA box, which is situated approximately 500 bp upstream of the human CDK9-42 transcription start [4,7]. The CDK9-42 isoform is present both in the cell cytoplasm and nucleus [12], whereas CDK9-55 is predominantly localized in the cell nucleus [4]. Differential patterns of tissue distribution in the human and mouse model were observed between the two CDK9 isoforms [4,5]. For instance, CDK9-42 expression has a bias for testis and spleen, whereas CDK9-55 is predominantly found in liver tissues, brain, and lung [4,5]. Similar findings in differential patterns of expression were observed in human and mouse cell culture systems. On one hand, CDK9-42 is primarily expressed in human cervical carcinoma HeLa cell line [76], human primary undifferentiated monocytes [8], and mouse NIH/3T3 fibroblasts [4]. On the other hand, CDK9-55 expression becomes predominant after induced differentiation of human primary monocytes along the macrophage lineage [8]. A considerable decline in CDK9-55 expression was reported following the activation of human lymphocytes, in which CDK9-42 becomes the predominant form [8]. Another study showed that CDK9-55 is primarily expressed in rat hepatocytes, when these cells are inside the liver, but CDK9-42 surmounts CDK9-55 expression once the rat hepatocytes are established in primary cell cultures [8]. Thus, the two CDK9 isoforms are differentially expressed depending on the cell signaling system and/or cell type. Lastly, several studies showed that CDK9-55 is associated with the regulation of the cell differentiation program of various tissues, such as the hematopoietic compartment [4,5,8], the muscle [68,77], and adipogenesis [78].
Both CDK9 isoforms can combine either with type T cyclins or cyclin K to generate a heterodimer (Tables 1 and 2), which is the principal constituent of the positive transcription elongation factor b (P-TEFb) ( Figure 4) [5,8,25,[79][80][81][82]. The P-TEFb complex phosphorylates the carboxyl-terminus domain (CTD) of the RNA pol II, which, in turn, stabilizes the elongation of the RNA transcript [25,[80][81][82]. The kinase activity of P-TEFb is specifically inhibited by 5,6-dichloro-1--D-ribofuranosylbenzimidazole (DRB) [5,8,25]. The nonphosphorylated RNA pol II can start RNA transcription in the initial stages of the transcription complex assembly but is not able to support the full elongation of the nascent RNA transcript ( Figure 4) [25,80,83]. This is due to the association between RNA pol II and the negative transcription elongation factor (N-TEF) ( Figure 4) [80,83]. The presence of the P-TEFb accomplishes a dual function: the ejection of N-TEF from the transcription complex and the subsequent phosphorylation of the RNA pol II CTD, which becomes effective in maintaining the elongation of the nascent RNA transcript ( Figure 4) [25,80,83]. The human RNA pol II CTD contains 52 repeats of the heptapeptide Y 1 S 2 P 3 T 4 S 5 P 6 S 7 [84][85][86]. The nonphosphorylated RNA pol II is firmly linked with the various factors of the preinitiation complex, which include the TATA-binding protein (TBP) and the mediator complex [84][85][86]. A study conducted in the in vitro system showed that the CDK7-mediated phosphorylation of Ser5 of the heptapeptide reduces the binding affinity between the RNA pol II CTD and the other components of the preinitiation complex [84]. Then, the capping enzyme binds to the phosphorylated Ser5 CTD, and the RNA capping structure is associated with the nascent RNA transcript, which is approximately 25-30 nucleotides long [84]. However, Ser5 phosphorylation of RNA pol II CTD is not per se sufficient to stabilize the elongation of the RNA transcript, which also needs    the CDK9-mediated phosphorylation of Ser2 of the CTD heptapeptide [84]. Interestingly, an in vivo study showed that CDK7 and CDK9 were able to substitute each other in the phosphorylation of Ser5 and Ser2 of the RNA pol II CTD [87], which is in contrast with the observation of the in vitro model [84,87]. This study was conducted on human glioblastoma and human prostate cancer cell lines and utilized siRNA molecules for the silencing of CDK9-and/or CDK7-related pathways [87]. The biological efficacy of the silencing of the two pathways was monitored by Western blot assay with antibodies that were specific either for phosphorylated Ser2 or phosphorylated Ser5 of the RNA pol II CTD [87]. The best inhibition of Ser2 and Ser5 phosphorylation required the simultaneous silencing of CDK9 and CDK7 [87]. Indeed, this finding may have significant implications both for the design of kinase inhibitors in cancer therapy and the development of a platform for microarray analysis of the two signaling systems in malignant cells [87]. The phosphorylation of the RNA pol II CTD is crucial for the CDK9-related pathway regulation of gene expression in mammalian cell biology [13,25,27]. Interestingly, the CDK9-related pathway also interacts with several other cellular and viral factors (Table 2) [3, 10, 12, 13, 24-28, 56-67, 88-107]. These interactions may consist of phosphorylation and/or physical association between the CDK9related pathway and various cellular or viral components.
The CDK9-mediated posttranslational modification of cellular factors constitutes an additional prospect for the genetic regulation of mammalian cell biology [13,25,27]. The same applies for the phosphorylation and/or association between the CDK9-related pathway and viral factors, which, once they are modified, take part in the gene regulation of the viral agent within the host cell [13,25,27].
In addition, the CDK9-related pathway arbitrates the regulation of the cotranscriptional mRNA processing, chromatin modification, activation of quiescent B and/or T lymphocytes, cell differentiation, and cell survival (  a main player in the regulation of the differentiation program in the hematopoietic compartment [4,5,8], muscle tissues [77,107], central nervous system [67], and adipogenesis [78]. There is a great deal of interest in the involvement of the CDK9-related pathway and cell survival and/or protection from apoptosis. Previous studies demonstrated that the overexpression of a dominant negative CDK9 (CDK9-dn) increased susceptibility to apoptosis in human monocytes [12], human U-937 promonocytic cells, and human Jurkat T cell line [100]. Indeed, the CDK9-related signaling system is an important pathway in mammalian cell biology and covers a wide range of functions, which comprise cell survival and protection from apoptotic injuries. The latter might have important implication in the context of a deregulated CDK9 pathway in the establishment and maintenance of a transformed cell phenotype.
Increased expression levels of CDK9 and cyclin T1 were observed in B and T cell precursor-derived lymphomas, anaplastic large T cell lymphoma, and follicular lymphomas, while Hodgkin and Reed-Sternberg cells of classical Hodgkin's lymphoma were characterized by a strong nuclear staining for both proteins [66]. In addition, abnormal mRNA levels of CDK9 and cyclin T1 were found in Burkitt's lymphoma, diffuse large B cell lymphoma with germinal center phenotype, classical Hodgkin's lymphoma-derived cell lines, and follicular lymphoma [66].
These findings, taken together, indicate that CDK9, along with other CDKs, might be a suitable therapeutic target for cancer therapy.

The Standpoint of Protein Kinase Development for Cancer Therapy
The pharmacological inhibition of CDKs for the treatment of malignancies has indeed attracted a great deal of interest over the last years [13,[26][27][28]. The majority of protein kinase inhibitors are nucleoside analogs and aim at neutralizing the adenosine triphosphate (ATP) binding site situated in the CDK enzymatic moiety [130][131][132][133]. Many CDK inhibitors have been utilized in phase I and phase II clinical trials for the treatment of an ample variety of cancers (Table 3). In this respect, seliciclib and alvocidib belong to the first generation of CDK inhibitors and have been utilized in phase I and phase II clinical trials for the treatment of many types of tumors [13,[26][27][28]. However, these clinical trials reported a modest clinical efficacy in oncological patients treated either with seliciclib or alvocidib [13,[26][27][28].

Future Directions for Drug Discovery and Development in Cancer Therapy
As already mentioned, most of protein kinase inhibitors are nucleoside analogs and were designed to block the ATP binding site of the catalytic domain of various CDKs, such as alvocidib (or flavopiridol) and seliciclib [13,[26][27][28]. However, this therapeutic approach is associated with toxicity [13]. In addition, alvocidib and seliciclib exhibited a modest pharmacological activity in clinical trials for the treatment of patients with different kinds of tumors [13]. Several reasons may account for the toxicity of the two compounds [13].
One of these reasons might be related to the targeting of the ATP binding site of the CDK enzymatic domain, which has a conserved structure among protein kinases [152]. Thus, nucleoside analog CDK inhibitors become responsible for off-target effects, which may impair to some extent the normal functions of other types of protein kinases and, as a result, cause a considerable harm in normal cells and/or tissues [152]. In order to circumvent this issue, the field of drug design is developing the so-called allosteric inhibitors, which either modify the protein kinase conformation to inhibit the function of the CDK ATP binding site [153] or compete directly with the binding of the regulatory subunit of the kinase to the protein substrate [153][154][155][156][157]. Some examples of allosteric kinase inhibitors that impair ATP binding to the catalytic site comprise Gleevec (imatinib mesylate), BIRB796 (Doramapimod), BAY43-9006 (Sorafenib or Nexavar), and AAL-993 [153]. Gleevec was designed to inhibit many types of tyrosine kinases, such as Bcr-Abl, stem cell factor receptor (c-Kit), and platelet-derived growth factor receptor (PDGF-R) [153]. BIRB796 targets serine/threonine kinase p38 MAPK [153]. BAY43-9006 is a Raf inhibitor, whereas AAL-993 is an antiangiogenic factor that suppresses the kinase activity of VEGF-1R, VEGF-R2, and VEGF-R3 [153]. Other socalled ATP noncompetitive allosteric inhibitors that interfere with the interaction between the regulatory subunit of the kinase and the protein substrate include PD09859, CMPD1, Coumarin derivatives G8935 and G0328, API-2, amino-functionalized quinoxaline 5, Pyrazinone derivative 14f, Akt-I-1, Thiadiazolines, Chloromethyl thienylketone 17, and BMS-345541 [153]. Specifically, PD09859 is an MEK inhibitor, whereas CMPD1 and Coumarin derivatives G8935 and G0328 target the extracellular signal regulated kinase (ERK) mitogen-activated protein kinases (MAPKs) [153]. The AKT pathway is inhibited by API-2, amino-functionalized quinoxaline 5, Pyrazinone derivative 14f, and Akt-I-1. Thiadiazolines and Chloromethyl thienylketone 17 impair the functions of Glycogen synthase kinase 3- [153]. Lastly, BMS-345541 competes with the binding between the protein substrate and threonine/serine I B kinases [153]. Interestingly, other classes of protein kinase inhibitors are currently under development for the targeting of wide interfaces between two proteins [158][159][160][161][162][163][164][165]. On one hand, this approach holds great therapeutic potential, but, on the other hand, the formulation of molecules that have the ability to disrupt the interaction between two proteins poses a daunting challenge. In fact, the contact surfaces between two interacting proteins may be in the range of 1,500 to 3,000 A 2 [166,167], whereas the interactions between a protein and a small molecule typically involve contact surfaces that may vary from 300 to 1,000 A 2 [168,169]. Furthermore, contact interfaces between two interacting proteins are usually flat and do not have pockets and/or grooves, which may be quite 8 ISRN Oncology normally found on surfaces of proteins that bind to small molecules [170]. However, several mutational studies showed that a small number of amino acid residues involved in holding together protein-protein interface account for most of the free energy for the binding [171][172][173][174][175]. These so-called hotspots are usually found at the center of the contact interface and take less than half of the total contact interface implicated in protein-protein interactions. For example, the binding interface between CDK9 and cyclin T1 is flat and hydrophobic ( Figure 2) [176]. The hydrophobic residues that constitute the hydrophobic pocket in the binding region of CDK9 and cyclin T1 comprise Leucine 64 (CDK9), Phenylalanine 59 (CDK9), Isoleucine 67 (CDK9), Isoleucine 84 (CDK9), Isoleucine 99 (CDK9), and Phenylalanine 146 (cyclin T1). Phenylalanine 146 residue of cyclin T1 is also present among the other cyclin partners of CDK9, such as cyclin T2a, cyclin T2b, and cyclin K (Figure 1) [176]. Remarkably, studies conducted on Iron chelators ICL670 and ICL311 reported kinase activity inhibition following the separation of CDK9 from cyclin T1, which, in turn, resulted either in suppression of HIV-1 replication [164,165] or repression of tumor growth in vitro and in animal models [177]. The exact mechanism of Iron chelator-mediated dissociation of the CDK9 and cyclin T1 complex is still under investigation.
As indicated by the studies on the binding interface between CDK9 and cyclin T1, almost identical hotspot regions might be in common among other protein-protein interacting surfaces, such as CDK7 and cyclin H, for example, [178,179]. Other reports described the presence of promiscuous contact surfaces in a number of protein-protein interactions [180]. Therefore, the design of small molecules for the inhibition of protein-protein interactions is always susceptible to undesired off-target effects. Nevertheless, encouraging achievements were reported in designing small molecules that exhibited specific targeting for protein-protein interactions [181][182][183][184]. A number of inhibitors for protein-protein interfaces were developed for interleukin-2 (IL-2), Bcl-X , human protein double minute 2 (HDM2), human papilloma virus (HPV) transcription factor E2, membrane-anchored bacterial ZipA protein, and binding of tumor necrosis factor (TNF) to its receptors (TNFR1 and TNFR2) [158]. Other studies are aiming at producing new classes of macromolecules termed "bis-peptides", which have molecular weights ranging from 750 to 2000 Da [185,186]. Such bis-peptides have the ability to create large, extended preorganized surfaces, which may provide outstanding drug characteristics for the disruption of protein-protein interfaces [185,186]. This approach allowed for the synthesis of a functionalized bis-peptide that inhibited the ubiquitination of wild-type p53 in human liver cancer cell lines [187]. The bis-peptide reproduced the activation domain of the tumor suppressor gene p53 and was utilized to neutralize HDM2 in human liver cancer cell lines [187]. HDM2 is a human E3 ubiquitin ligase and a suppressor of p53 expression through ubiquitination. Therefore, the bispeptide-mediated disruption of HDM2/p53 binding caused the inhibition of wild-type p53 ubiquitination [187,188]. The accumulation of wild-type p53 may subsequently trigger apoptosis in malignant cells [187,188].

Conclusion
The CDK9-related pathway has emerged as a target of extreme for cancer therapy. Deregulations in this pathway were observed in a variety of human tumors, which may induce increased expression and/or hyperactivity of cellular oncogenic factors. In fact, the use of kinase inhibitor in clinical trials for the treatment of patients with chronic lymphocytic leukemia showed that the inhibition of CDK9and CDK7-related pathways was responsible for the decrease of the antiapoptotic factor Mcl-1 [125][126][127]. On these grounds, there is a keen interest in producing new compounds with enhanced specificity for CDK9-and/or CDK7-related pathways.
The field of drug design is currently striving to improve the therapeutic index of kinase inhibitors for cancer therapy, in order to minimize the toxicity associated with kinase inhibitors that block the ATP binding site of the catalytic domain of the enzyme. In this respect, drug designers are pursuing two main strategies: the engineering of allosteric kinase inhibitors and of inhibitors that disrupt proteinprotein interactions.