New CDK8 inhibitors as potential anti-leukemic agents – Design, synthesis and biological evaluation

Cyclin-dependent kinase 8 (CDK8) plays a vital role in regulating cell transcription either through its association with the mediator complex or by the phosphorylation of transcription factors. CDK8-mediated activation of oncogenes has proved to be important in a variety of cancer types including hematological malignancies. We have designed and synthesized a series of new synthetic steroids. The compounds were evaluated as CDK8 inhibitors in vitro. The three most potent compounds exhibit Kd-values towards CDK8 in the low nanomolar range (3.5–18 nM). Furthermore, the compounds displayed selectivity for CDK8 in a panel of 465 different kinases. The cell studies indicated a selectivity to kill AML-cancer cell lines compared to normal cell lines.


Introduction
Cyclin-dependent kinases (CDKs) are a family of serine/threonine protein kinases that control critical regulatory events during cell cycle and transcription. 1 Given these fundamental roles, it is unsurprising that deregulation of CDKs is a common feature of many cancers. In particular, Cyclin-dependent kinase 8 (CDK8), a ubiquitously expressed, primarily transcriptional member of the CDK family, has come under focus owing to investigations of its central roles in transcription and oncogenesis. [2][3][4] Several genetic and biochemical studies have established CDK8 as a key oncogenic driver in a variety of different cancer forms like colon cancer 5 , prostate cancer 6 and melanoma 7 . It is also reported that a striking correlation exists between the expression of CDK8 and the duration of relapse-free survival in patients with breastand ovarian cancer 8 . For example, breast cancer patients with belowmedian expression of CDK8 survive without the disease approximately seven years longer than patients with an above-median expression of CDK8. 9 From a biochemical perspective, CDK8 plays an important role in regulating transcription either through its association with the mediator complex or by phosphorylating transcription factors. Specifically, CDK8-mediated activation of oncogenic Wnt-β-catenin signaling, transcription of estrogen-inducible genes and suppression of super enhancer-associated genes contribute to oncogenesis in colorectal-, mammary-and hematological malignancies, respectively. Inhibitors of CDK8 have also been shown to be active in AML cells that have high activation of the signal transducer and activator of transcription 5 (STAT5). 10 STAT5 mediates several malignant properties in AML cells. Inhibition of STAT5 phosphorylation, through inhibition of for instance Flt3 or PIM-kinases, has shown therapeutic value. 11,12 Another aspect with CDK8 in cancer is that it is involved in tumor surveillance of NKcells. Inhibition of CDK8 is known to increase NK cell cytotoxicity towards cancer cells. 13,14 The use of CDK8 inhibitors can therefore have a double action: First, by acting directly on the cancer cells, and secondly indirectly by activation the NK cells to more efficiently lyse the cancer cells. Hence, an interest to develop compounds inhibiting CDK8 has increased. As of today, based on high throughput screening (HTS) and computational methods towards different kinase panels, a few compounds have appeared as either CDK8 ligands or inhibitors. [15][16][17][18] (Fig. 1).
In 2012, the senexin compounds, exemplified by Senexin A (1) Fig. 1, were revealed as a class of selective CDK8 inhibitors. This group of compounds was discovered using HTS with more than 100,000 compounds for the downstream inhibition of p21-activation transcription. An optimization of the compounds led to Senexin A that was characterized as a kinase inhibitor with K d = 0.83 µM and K d = 0.31 µM for CDK8 and its structural homolog CDK19, respectively. The kinase activity of CDK8 was inhibited with an IC 50 = 0.28 µM. Cortistatin A (2) is a naturally occurring steroidal alkaloid isolated from the marine sponge Corticum simplex. 19 It was reported in 2009 that 2 is a high-affinity binder, with K d = 17 nM, towards CDK8. 20 In addition to its strong binding affinity, the compound showed a high level of selectivity for CDK8. Cortistatin A was tested at 10 µM in a high-throughput binding assay against a panel 359 non-mutant kinases, and the calculated selectivity score at 35% control activity S(35) 21 , proved to be 0,045. The authors suggested in a binding model that hydrogen bonding between the nitrogen on the isoquinoline side chain to Ala100 in CDK8 was important. 20 More recently, structural and mechanistic studies have revealed that compound 2 is a highly potent, ATP-competitive, CDK8 inhibitor, and utilizes its isoquinoline moiety to bind to the kinase hinge segment through the nitrogen of the isoquinoline and the NeH of Ala100. The steroidal-like core makes extensive contacts with the ATP-binding cavity. 22 Furthermore, the study investigated the potential of cortistatin A (2) as a treatment for acute myeloid leukemia (AML) and a 71% tumor growth inhibition on day 33, in mice containing SET-2 AML xenograft tumors, was reported. Importantly, cortistatin A had an acceptable pharmacokinetic profile in mice for once-daily intraperitoneal dosing and the reduction in the progression of the disease was reported to be dose-dependent.
The heterocyclic compound CCT251921 (3) was originally identified as a potent inhibitor of the Wnt signaling pathway. 23 A follow-up study using a chemical proteomics-based approach identified CDK8/ CDK19 as its primary targets. 24 The compound was claimed to be selective for CDK8/CDK19 in a panel of 279 kinases at 1.0 µM; however CDK8 itself was not included in the panel as a positive control. Preclinical testing of these compounds showed that the compound inhibited Wnt-dependent activity in a number of colorectal cell lines, as well as the phosphorylation of the transcription factor Signal transducer and activator of transcription 1 (STAT1) at Ser727. The binding mode of CCT251921 (3) to CDK8 revealed that its pyridine nitrogen interacted with the kinase hinge segment by a similar interaction to the NeH of Ala100 as that of the isoquinoline nitrogen of cortistatin A (2). From this interaction, the indazole substituent of 3 is orientated towards the ATP-binding cavity in similarity to the steroidal core of cortistatin A. Additionally, the chlorine-atom, attached to the pyridine ring, in a meta-relationship to both the nitrogen atom and the indazole, further increased the affinity of compound 3 by an interaction with Phe97. In connection with our interest and earlier efforts in using steroids as lead compounds for the development of new anti-cancer agents, [25][26][27][28][29][30] we present our experimental approach by taking advantage of this information in the design of easily accessible steroid-based inhibitors of CDK8.

Chemistry -Design and synthesis
In March 2018, Hatcher and coworkers demonstrated that several steroidal cortistatin A (2) analogs were effective inhibitors of CDK8. 31 In this work, we aim to advance their approach further by combining information of the binding modes of the pyridine core of CCT251921 (3) and the isoquinoline of cortistatin A (2). Compared to the work published by Hatcher and coworkers, we have included several more isoquinolinyl-, as well as pyridinyl and indolyl substituents attached to the unsaturated carbon 17 at the steroidal core. Encouraged by the work of Baran and coworkers and to simplify the SAR-studies, 32 we choose not to reduce the double bond on the D-ring of the analogs. Furthermore, the orientation of dimethylamino group of Cortistatin A is maintained in our analogs at the carbon number of the steroid nucleus. By this approach, several analogs have been made (see Scheme 1).
The synthesis of compounds 8-17 commenced with a Mitsunobutype inversion of the chiral center at C-3 of epiandrosterone (4) to the corresponding androsterone-3-azide 5 using triphenylphosphine, diisopropyl azodicarboxylate (DIAD) and diphenylphosphorylazide (DPPA) in THF, from which androsterone-3-azide 5 was isolated in 73% yield. Then, in a one-pot sequence, a modified Staudinger reaction using trimethylphosphine in THF was employed to afford the intermediate steroidal amine, which was further converted to the dimethyl amine 6, by an Eschweiler-Clarke type reaction [33][34][35] The enol-triflate 7 was made from the ketone 6 in a reaction using N-phenyl-bis(trifluoromethanesulfonimide) in the presence of NaHMDS at −78°C. The key intermediate 7 was further converted to the desired compounds 8-17 using the Suzuki-Miyaura cross-coupling reaction with different boronic acids as coupling partners. Tetrakis(triphenylphosphine)palladium and caesium carbonate in THF/H 2 O (1:1) enabled the preparation of compounds 8-17 in 31-42% overall yield (see Fig 2).

CDK8 profiling
Compounds 8-17 were tested for their ability to inhibit probe binding to protein kinase CDK8 in vitro at 50 nM concentration. In this assay, the ability of a test compound to compete with an immobilized, active site directed ligand is quantitatively measured and reported as percent of DMSO control (POC), with lower numbers indicating higher binding affinity. The results are displayed in Table 1. Except for 16, all compounds displayed inhibitory activity towards CDK8 at 50 nM concentration. For the five most promising compounds, their K d -values were determined. The most potent compound, the 5-isoquinoline analog 12, inhibited CDK8 with a K d = 3.5 nM, which is 3-4 times lower than the reported K d -value of our lead compound cortistatin A (2) (17 nM) in a similar assay. 20 Compound 12 has a 5-isoquinoline substituent in the 17-position of the steroid core skeleton. For compound 9, with a 5-chloropyridin-3-yl substituent as in CCT251921 (3), the K dvalue was determined to 13 nM in the same range as for cortistatin A (2).
To evaluate the kinase selectivity, we performed a KinomeScan™ binding analysis against a near comprehensive panel of 468 kinases at a concentration of 50 nM (Fig. 3). This profiling revealed selectivity of compounds 12 and 9 for CDK8, both with selectivity scores, S(10) at 0.01. 21 Furthermore, compound 14 displayed a K d = 18 nM and a S(10)

In vitro studies
In order to assess whether the analogs could be lead compounds in acute myeloid leukemia (AML) therapy, we tested them against three cell lines derived from AML patients. These were cells with different mutation status and cytogenetics. The MOLM-13 and MV4-11 cells are from the monocytic lineage, and display heterozygotic and homozygotic internal tandem duplication (ITD) of Flt3, respectively, which both are associated with high risk. The OCI-AML3 cell line has wildtype Flt3, associated with low risk, but a complex karyotype, which turns this cell line into a high-risk model. In line with this, we have found that the OCI-AML3 are more resistant than MOLM-13 and MV4-11 cells to other kinase inhibitors, which did not correspond to expression of drug transporters. 36,37 The CDK8 inhibitors tested had EC 50 values between 2.7 µM and 10 µM for the AML cells ( Table 1). The analog 13 was the most potent, with values below 4.1 µM for all AML cell lines (Table 1). Interestingly, the analogs did not show significantly different activity towards the three AML cell lines. This suggests that their cell death targets are different from other cytostatics like daunorubicin and kinase inhibitors, which for instance have low activity towards the cytogenetically complex OCI-AML3 cells compared to their activity towards MOLM-13 and MV4-11. The analogs were next tested against the two non-malignant cell lines NRK (rat epithelial kidney cells) and H9c2 (cardio myoblasts) to find if the compounds also could affect normal tissue. In comparison to the malignant cell lines, these cells tolerated two to four times higher concentration of most of the analogs ( Table 2).

Conclusion
Herein, ten new synthetic compounds have been reported and evaluated for their inhibitory properties towards CDK8. The initial screening of the compounds proved inhibitory effects toward CDK8 at a concentration of 50 nM. For the five most potent compounds, the individual K d -values toward the same enzyme were determined, which further proved the ability of the compounds to bind to the kinase in vitro. The K d -values of the three most potent compounds are in line or even lower compared to the previous reported CDK8 inhibitors. 9,10,20,24 Additionally, when tested towards a panel of 468 different kinases, the compounds within this study demonstrated a high degree of selective inhibition of CDK8. Compound 14 is a similar analog to what has already been reported by Hatcher and coworkers. 31 By further structure activity relationship studies, we have improved both the affinity and selectivity towards CDK8 for this class of compounds.
Interestingly, the cellular studies identified compound 13 to be the most cytotoxic compound. This observation is in line with the observation done by Corey and coworkers, where several isoquinoline steroidal analogs where tested for their antiangiogenic activity. 38 However, low selectivity of this compound towards the AML cell lines, compared to the non-malignant cell lines, were observed (Table 2). For the most potent CDK8 inhibitor, compound 12, the EC 50 values were slightly higher, i.e. it was less cytotoxic towards the AML cell lines (Fig. 3). However, compound 12 proved more selective in killing AML cell lines, compared to compound 13. The EC 50 values are three to four times lower for compound 12 towards the AML cell lines compared to the non-malignant cell lines.

General methods -Chemistry
All reagents and solvents were used as purchased without further purification unless stated otherwise. Melting points are uncorrected. Analytical TLC was performed using silica gel 60 F254 aluminum plates (Merck). Flash column chromatography was performed on silica gel 60

Protein kinase assay
For the CDK8 and CDK11 profiling we used a kinase selectivity and profiling assay (DiscoveRx). 39 The assay uses kinase-tagged T7 phage strains which were grown in parallel in 24-well blocks in an E. coli host derived from the BL21 strain. E. coli were grown to log-phase and infected with T7 phage from a frozen stock (multiplicity of infection = 0.4) and incubated with shaking at 32°C until lysis (90-150 min). The lysates were centrifuged (6000g) and filtered (0.2 μm) to remove cell debris. Streptavidin-coated magnetic beads were treated with biotinylated small molecule ligands for 30 min at room temperature to generate affinity resins for kinase assays. The liganded beads were blocked with excess biotin and washed with blocking buffer (SeaBlock (Pierce), 1% BSA, 0.05% Tween 20, 1 mM DTT) to remove unbound ligand and to reduce non-specific phage binding. Binding reactions were assembled by combining kinases, liganded affinity beads, and test compounds in 1x binding buffer (20% SeaBlock, 0.17x PBS, 0.05% Tween 20, 6 mM DTT). Test compounds were prepared as 40x stocks in 100% DMSO and directly diluted into the assay. All reactions were performed in polypropylene 384-well plates in a final volume of 20 µL. The assay plates were incubated at room temperature with shaking for 1 h and the affinity beads were washed with wash buffer (1x PBS, 0.05% Tween 20). The beads were then re-suspended in elution buffer (1x PBS, 0.05% Tween 20, 0.5 μM non-biotinylated affinity ligand) and incubated at room temperature with shaking for 30 min. The kinase concentration in the eluates was measured by qPCR. The test compounds were screened at 50 nM, and results for primary screen binding interactions are reported as POC (percent of control). The negative control consists of adding an equal DMSO volume without a test compound and the positive control consist of a control compound. From this POC is calculated: ((test compound signal − positive control signal)/(negative control signal − positive control signal)) × 100%. Negative control = DMSO (100% Ctrl); Positive control = control compound (0% Ctrl), where lower numbers indicate stronger hits in the matrix.
For the binding constant (K d ) determination an 11-point 3-fold serial dilution of each test compound was prepared in 100% DMSO at 100x final test concentration and subsequently diluted to 1x in the assay (final DMSO concentration = 1%). Most K d values were determined using a compound top concentration = 30,000 nM. The K d values were calculated with a standard dose-response curve using the Hill equation and are expressed as an average of two determinations.
All analogs were dissolved in DMSO (D5879, Sigma-Aldrich) to a stock concentration of 25 mM before testing, and stored at −80°C. For cytotoxic testing, the three AML cell lines were seeded at 10,000 cells/ well, while NRK cells were seeded as 2000 cells/well and H9c2 as 2500 cells/well in 96-well Microplates (#167008, Thermo Scientific™ Nunc™ MicroWell™) with 0.1 mL medium/well. The cells were exposed to various concentrations of CDK8 inhibitors for 48 h before the cell proliferation reagent WST-1 (11644807001, Roche Diagnostics, Sigma-Aldrich) was used to assess the viability, in accordance with the manufacturer's instructions. The cells were next fixed in 2% buffered formaldehyde (pH 7.4) with the DNA-specific dye Hoechst 33,342 (Polysciences Inc.) and assessed for cell death by UV-microscopy based on nuclear morphology. Viability assessment with these two methods is shown to correlate well with cells exposed to kinase inhibitors 37 .
EC 50 values were determined based on data from WST-1 assay in combination with microscopic evaluation of cell death, using fourparameter regression analysis function in the SigmaPlot software (Systat Software inc. San Jose, CA): where Y is the response (WST-1 signal or percent apoptosis), min and max are minimum and maximum response, X is concentration of analog, EC50 equals the point of inflection, i.e. the point that gives half of maximum response, and h is the Hill's slope of the curve.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.