The Covalent CDK7 Inhibitor THZ1 Potently Induces Apoptosis in Multiple Myeloma Cells In Vitro and In Vivo

Cyclin-dependent kinases (CDK) catalyze the phosphorylation of cyclins, proteins intimately involved in the progression of cells through the cell cycle. However, in addition to their cell-cycle regulatory activities, CDKs exert several functions unrelated to cell-cycle control. For example, CDK9 acts as a transcriptional regulator as a participant in the positive-transcription elongation factor b (p-TEFb) complex, which phosphorylates the carboxy-terminal domain (CTD) of RNA Pol II (1). Indeed, CDK9 antagonists have been shown to downregulate the expression of short-lived proteins such as the antiapoptotic BCL-2 family member MCL-1 (2).

CDK7 represents a component of the TFIIH (general transcription factor IIH) multiprotein complex, which cooperates with p-TEFb to regulate RNA Pol II transcriptional activity (1). In addition to its transcriptional regulatory role, TFIIH has also been implicated in DNA repair (1). Moreover, CDK7 regulates the phosphorylation and activity of most CDKs through its participation in the cyclin-activating kinase (CAK) complex (3). The transcriptional addiction of many tumor types prompted the development of CDK7 inhibitors as anticancer agents. THZ1 is a covalent inhibitor of CDK7, which potently represses transcription and downregulates several short-lived proteins such as MYC (4, 5). In this regard, THZ1 has shown activity in multiple MYC-driven tumor types including neuroblastoma (4), breast cancer (6), and small cell lung cancer (7, 8). In contrast, THZ1 activity in T-cell acute lymphoblastic leukemia has been related to disruption of the RUNX1 transcription factor (9).

Multiple myeloma is an accumulative disorder of mature plasma cells that despite the introduction and approval of multiple novel agents (e.g., proteasome inhibitors, immunomodulatory agents, and antibodies; ref. 10) is in most cases incurable. Consequently, new and more effective approaches are urgently needed, particularly in the case of relapsed or refractory disease. Notably, several short-lived proteins, for example, MCL-1 and MYC have been implicated in myelomagenesis as well as resistance to established therapies (11, 12). The potential dependence of multiple myeloma cells on these proteins raised the possibility that a transcriptional CDK7 inhibitor like THZ1 might be particularly effective in this disease. Currently, the impact of CDK7 interruption has not yet been assessed in multiple myeloma models. Here we report that THZ1 potently inhibits multiple myeloma cell proliferation and survival in a MYC-, MCL-1-, and BCL-X_L-dependent manner, and potentiates the activity of proteasome inhibitors (carfilzomib, bortezomib) and BH3-mimetics (venetoclax) in both cell lines and primary patient samples. It also significantly improves survival in a multiple myeloma xenograft model with minimal toxicity. Together, these findings argue that CDK7 inhibitors like THZ1 warrant attention as therapeutic agents in multiple myeloma.

Human NCI-H929, U266, OPM2, and RPMI8226 cells were all from ATCC and maintained as described previously (13). Btz-resistant cells, U266/PS-R and 8226/V10R were established and maintained as described previously (14). Revlimid-resistant (R10R) RPMI8226 sublines were maintained as before (15). U266/MCL-1, U266/MYC, and 8226/BCL-X_L were established by stably transfecting full-length human MCL-1, MYC, and BCL-X_L cDNA separately as described previously (13). KMS28-BM and KMS28-PE were from Japanese Cancer Research Resources Bank (JCRB).

All experiments utilized logarithmically growing cells (3–5 × 10⁵ cells/mL). MycoAlert (Lonza) assays were performed, demonstrating that all cell lines were free of mycoplasma contamination.

THZ1 was purchased from Medchem Express. Bortezomib (Btz), Carfilzomib (Cfz), and Venetoclax (ABT-199) were purchased from ChemieTek. The caspase inhibitor Z-VAD-FMK was obtained from Enzo Life Sciences, Inc. All drugs were dissolved in DMSO, aliquoted, and stored at −80°C. In all experiments, final DMSO concentrations did not exceed 0.1%.

Construction of lenti-CRISPR/Cas9 vectors targeting CDK7 was performed following the protocol associated with the backbone vector (#45, Addgene; ref. 16). The following sequences were chosen from the published literature (6).

sgGFP (fwd: CACCGGGGCGAGGAGCTGTTCACCG;

rv: AAACCGGTGAACAGCTCCTCGCCCC),

sgCDK7-1 (fwd: CACCGGAAGCTGGACTTCCTTGGGG

rv: AAACCCCCAAGGAAGTCCAGCTTCC);

sgCDK7-2 (fwd: CACCGATCTCTGGCCTTGTAAACGG

rv: AAACCCGTTTACAAGGCCAGAGATC).

Virus infection were performed as described previously (13)

For clonogenic cell growth assays, cells in which lenti-CRISPR vectors were introduced, following puromycin selection for 5 days, were harvested and seeded into round-bottom 96 wells at 250 to 500 cells/well diluted by an equal volume of RPMI1640 with 10% FBS, incubated for varying intervals (e.g., 2, 4, 6, and 8 days) after seeding, and images captured using an Olympus IX71 Inverted System Microscope.

For 96-well plate assays, cells as described above were plated at a density of 2,000 cells per well, and CellTiter 96 Aqueous One Solution Cell Proliferation Assay (MTS) Kit (Promega) was used to monitor cell viability at 2, 4, 6, and 8 days according to the manufacturer's instructions.

For IC₅₀ determinations, multiple myeloma cells were seeded in 96-well plates at a density of 1.0 × 10⁴ cells per well. Each experiment was performed utilizing 3 independent wells. After 24 hours, cell viability was determined using a CellTiter Kit. Absorbance was measured at 490 nm using a microplate spectrophotometer (Thermo or Promega). To determine IC₅₀, 6 concentration points (corresponding to 2-fold increases) were chosen for each cell line. Experiments were performed 2 to 3 times independently.

All animal studies were IACUC approved and performed in accordance with AAALAC, USDA, and PHS guidelines. For the orthotopic murine model, NOD/SCIDγ mice were injected intravenously with either 5 × 10⁶ U266 stably transfected with constructs encoding luciferase (U266/Luc). For the flank murine model, NOD/SCIDγ mice were inoculated subcutaneously with 5 × 10⁶ PS-R/Luc cells. Control animals received equal volumes of vehicle (10% DMSO in D5W, 5% dextrose in water).

Tumors were measured in 2 dimensions using manual calipers. Tumor volume was calculated using the formula: V = 0.5 × length × width². Animal with tumor established (mean tumor volume of ∼200 mm³) were randomly divided into 2 groups, which were then treated with vehicle (10% DMSO in D5W), THZ1 (3 mg/mL, prepared in vehicle solutions) at the dose of 10 mg/kg via intraperitoneally twice daily (BID), 5 days a week.

Mice were monitored for tumor growth with an IVIS 200 imaging system (Xenogen Corporation), and tumor volume was monitored every 2 to 3 days. Measurement of animal body weight was performed every other day throughout the study to monitor toxicity. Upon harvesting, tumors were dissected into small pieces, and frozen in liquid nitrogen for preparation of lysates and immunoblotting.

Values represent the means ± SD for at least 3 independent experiments performed in triplicate. The significance of differences between experimental variables was determined using the Student t test or one-way ANOVA with Tukey–Kramer multiple comparisons test. The significance of P values are *<0.05, **<0.01, or ***<0.001, wherever indicated. Analysis of synergism was performed by median dose effect analysis using the software Calcusyn (Biosoft). Kaplan–Meier analysis of mouse survival performed with GraphPad Prism 6 software.

Cell-cycle analysis by propidium iodide (PI) staining was performed by flow cytometry (FCM) using the Modfit LT2.0 software (Verity Software House) as described previously (17).

See Supplementary Materials and Methods for transfection, qRT-PCR, immunoblot analysis, immunofluorescence, chromatin IP, isolation of primary multiple myeloma cells, analysis of cell death, and cell viability assay.

Exposure (24 hours) of multiple myeloma cell lines (OPM2, RPMI8226, H929, U266, PS-R, KMS28-BM, and KMS28-PE) to increasing concentrations of THZ1 variably reduced cell survival (Fig. 1A), but in all cases IC₅₀ values were in the low nmol/L range (e.g., <300 nmol/L; Supplementary Table S1). Low nmol/L THZ1 concentrations also killed bortezomib- or revlimid-resistant 8226 cells (Fig. 1B; ref. 15). Concordant findings were obtained when PARP and caspase-3 cleavage were monitored (Fig. 1C), including in bortezomib-resistant U266 cells (PS-R), previously shown to exhibit MCL-1 upregulation and BIM downregulation (18). THZ1 also induced expression of γH2A.X, an indicator of DNA double-strand breaks in multiple cell lines (Fig. 1C), consistent with the role of CDK7 in the DNA repair response (19).

Cell-cycle analysis revealed that submicromolar THZ1 concentrations (e.g., 50–150 nmol/L; 16 hours) moderately but significantly increased the G₂–M and diminished the S-phase fraction in H929 and U266 cells (Supplementary Fig. S1A–S1C). These effects were associated with diminished T161 CDK1 and T160 CDK2 phosphorylation (Supplementary Fig. S1D), consistent with the cyclin-activating kinase activity of CDK7 (20).

Studies in U266, their bortezomib-resistant counterparts PS-R (15) and OPM2 multiple myeloma cells demonstrated that low nmol/L THZ1 concentrations (50–400 nmol/L, 24 hours) potently diminished phosphorylation of RNA Pol II at both Ser2 and Ser5 sites, while also dephosphorylating CDK9 at the T186 activation site (Fig. 2A). In contrast, the BET inhibitor JQ1, known to disrupt the transcriptional regulatory apparatus (21), administered at a higher concentration (500 nmol/L), had little effect on these events. Time course studies in U266 cells showed discernible dephosphorylation of Pol II and CDK9 at 3 hours, which became more pronounced over the ensuing 24 hours (Fig. 2B). Finally, studies in U266 and PS-R cells demonstrated that these changes persisted despite THZ1 washout (Fig. 2C), consistent with the covalent nature of THZ1/CDK7 interactions (22).

Exposure of U266 cells to low nmol/L concentrations of THZ1 (50–400 nmol/L, 24 hours) led to downregulation of c-MYC, MCL-1, and BCL-X_L protein levels (Fig. 3A). JQ1 also downregulated c-MYC, consistent with previous reports (21), as well as BCL-X_L, but not MCL-1. Virtually identical findings were obtained in H929 cells (Supplementary Fig. S2A). Time course analysis of the latter showed discernible c-MYC, MCL-1, and BCL-X_L protein downregulation associated with PARP cleavage after 6 hours of THZ1 exposure, which was considerably more pronounced downregulation at 24 hours (Supplementary Fig. S2B). Notably, coadministration of the caspase inhibitor Z-VAD-fmk blocked THZ1-mediated PARP and caspase cleavage but not MCL-1, BCL-X_L, or c-MYC down-regulation (Supplementary Fig. S2C), arguing against the possibility that these events represented secondary caspase-dependent phenomena.

To elucidate mechanisms underlying THZ1 actions, RT-PCR analysis was performed. RT-PCR demonstrated that 200 nmol/L THZ1 significantly diminished c-MYC, MCL-1, and BCL-X_L mRNA levels in U266 cells as early as 1 to 3 hours after drug exposure (Fig. 3B). Comparable MCL-1, BCL-X_L, and c-MYC mRNA downregulation was observed in H929 cells (Supplementary Fig. S2D).

To assess the functional significance of transcriptional repression of these proteins in the antimyeloma activity of THZ1, lines were generated ectopically expressing c-MYC, MCL-1, or BCL-X_L. U226 cells ectopically expressing c-MYC (Fig. 3C, inset) were significantly less sensitive to 50 to 200 nmol/L THZ1-induced cell death than empty-vector controls, but equally sensitive to 500 nmol/L JQ1 (Fig. 3C). Analogously, ectopic expression of MCL-1 (Fig. 3D, inset) protected U266 cells from low concentrations of THZ1 but not JQ1 (Fig. 3D). The ability of MCL-1 to block THZ1-mediated caspase-3 cleavage is shown in Fig. 3E. Concordant results were obtained in cells ectopically expressing BCL-X_L (Fig. 3F). Of note, Although enforced expression of BCL-X_L clearly blocked PARP degradation in U266 cells exposed to THZ1, it did not prevent MCL-1 or c-MYC downregulation (Fig. 3G). Collectively, these findings argue that transcriptional downregulation of c-MYC, MCL-1, and BCL-X_L contribute functionally to THZ1-induced cell death in multiple myeloma cells.

It has been recently reported that THZ1 also inhibits CDK12 and CDK13 kinase activity (23), which control the expression of DNA damage response (DDR) genes. Therefore, the effects of THZ1 on DDR genes were examined in multiple myeloma lines. THZ1 dramatically reduced the expression of most DNA repair proteins (CtIP, FANCD2, RAD51, BRAC1, and ERCC1) examined (Supplementary Fig. S3A), and also downregulated their mRNA levels (Supplementary Fig. S3B), accompanied by increased expression of γH2A.X, indicative of double-stranded DNA breaks. These findings are concordant with the notion that THZ1 inhibits both CDK7 and CDK12 activities, consistent with what has been reported in the literature for other cell types (23). They are also consistent with the concept that downregulation of DNA repair genes may also contribute to THZ1 activity in multiple myeloma cells.

The effects of stromal cells on THZ1-mediated cell death was then examined. When luciferase-labeled U266 cells were exposed to sub-μmol/L concentrations of THZ1 in the presence of HS-5 stromal cells, a modest reduction in cell death was observed (Supplementary Fig. S4A). No reductions in bortezomib sensitivity were observed. Interestingly, HS-5 cell coculture failed to protect cells from THZ1 lethality in highly bortezomib-resistant PS-R cells (Supplementary Fig. S4B). Fluorescent microscopic images revealed the robust dose-dependent induction of cell death (7-AAD positivity) by THZ1 in the presence of HS-5 coculture, particularly in PS-R cells (Supplementary Fig. S4C).

The functional significance of CDK7 inhibition by THZ1 or CDK7 knockout on multiple myeloma cell growth and responses to targeted agents was then examined. To this end, 2 OPM-2 CRISPR-Cas CDK7 knockout cell lines (sg_CDK7-1 and -2) were generated which exhibited markedly reduced or virtually absent CDK7 expression (Fig. 4A, inset). Cell viability was monitored by the Titer-Glo assay, which reflects both cell death as well as loss of cell proliferation. The knockout lines displayed dramatically reduced viability and proliferation compared with control lines, which may reflect both THZ1's antiproliferative and proapoptotic actions (Fig. 4A). Light-microscopic images of cells plated at 250 cells/well and incubated for 6 days are shown in Fig. 4B. Similar growth inhibition was observed in H929 CDK7 knockdown cells (Supplementary Fig. S5A). The effects of CDK7 knockdown on the response of multiple myeloma cells to the clinically relevant agents bortezomib, carfilzomib, or ABT-199 were then investigated. Both of the knockdown lines were significantly more sensitive to these agents than controls, which demonstrated only a modest reduction in cell death (7-AAD uptake; Fig. 4C). Parallel studies were then performed using THZ1. Treatment of U266 cells with 50 or 100 nmol/L THZ1 significantly increased the lethal effects of all 3 agents compared with untreated controls (**P < 0.01 in each case; Fig. 4D). Equivalent results were obtained in bortezomib-resistant PS-R cells (Supplementary Fig. S5B). Isobologram analysis in U266 cells revealed CI ratios substantially less than 1.0 for each agent, indicating synergistic interactions (Fig. 4E). Synergistic interactions were also observed in H929 cells (Supplementary Fig. S5C). These findings indicate that pharmacologic disruption of CDK7 mimics the effects of genetic CDK7 interruption in increasing the susceptibility of multiple myeloma cells to PIs and BH3-mimetics. Given the pleiotropic actions of THZ1 (23), we examined the effects of CDK7 deficiency on c-MYC, MCL-1, and BCL-X_L protein expression. c-MYC, MCL-1, and BCL-X_L expression were clearly reduced in CDK7 knockout OPM2 cells (Supplementary Fig. S6A). To determine whether c-MYC, MCL-1, or BCL-X_L could be rescued by CDK7 in knockout cells, HA-CDK7 (24) was reintroduced in CDK7 knockdown cell expressing shRNA targeting 3-UTR. Notably, wild-type CDK7 effectively rescued c-MYC, MCL-1, and BCL-X_L expression in CDK7 knockdown cells (Supplementary Fig. S6B). Significantly, however, CDK7 did not reverse downregulation of these proteins following THZ1 treatment (Supplementary Fig. S6C). As THZ1 targets both CDK7 and CDK12/13 (23, 25), it is possible that both CDK7 and CDK12/13 inhibition may be involved in downregulation of MYC, MCL-1, and BCL-X_L by THZ1 in these cells, and thus not rescued by CDK7 alone. The effects of THZ1, alone or in combination, were then examined in primary multiple myeloma cells. As shown in the fluorescence photomicrographs in Fig. 5A, exposure (16 hours) of CD138⁺ cells from 2 patients with multiple myeloma to 300 nmol/L THZ1 sharply increased expression of activated caspase-3. In a series of primary patient with multiple myeloma samples (N = 17) exposed to 300 nmol/L THZ1; 16 hours, a very significant reduction in survival (Annexin V/7-AAD staining) was observed in both the CD138⁺ fraction as well as in the more primitive CD138⁻, CD19⁺, CD20⁺, CD27⁺ population (N = 11; P < 0.01 in each case; Fig. 5B, right; ref. 26). In marked contrast, an identical THZ1 exposure was minimally toxic to normal CD34⁺ cord blood cells (Fig. 5B, left). In addition, coadministration (16 hours) of THZ1 (25 nmol/L) and bortezomib (2 nmol/L) very significantly reduced survival compared with individual treatment in both of these multiple myeloma cell populations (Fig. 5C and D), but had little effect on the survival of normal cord blood CD34⁺ cells (Fig. 5I). Similar results were obtained when THZ1 was combined with carfilzomib (3 nmol/L; Fig. 5E, F, and J) or ABT-199 (200 nmol/L; Fig. 5G, H, and 5K).

It has been reported that CDK7 inhibition can disrupt super-enhancers (25), which are frequently associated with genes like MYC, MCL-1, BCL-X_L, and CCND2 (27) that figure prominently in multiple myeloma biology. Consequently, chromatin IP of histone H3K27 acetylation, a mark of an active enhancer, was therefore performed. Super-enhancer–associated genes such as c-MYC, MCL-1, and BCL-X_L displayed diminished H3K27 acetylation with THZ1 treatment (Supplementary Fig. S7), raising the possibility that MYC, MCL-1, and BCL-X_L expression may be regulated by histone H3K27 acetylation, which can in turn be disrupted by THZ1 in myeloma cells.

Finally, the in vivo activity of THZ1 was examined in systemic and flank multiple myeloma models. NSG mice were injected intravenously with 1 × 10⁶ luciferase-labeled U266 cells, and treated with 10 mg/kg THZ1 (i.p. twice daily, 5 days/week). As shown in Fig. 6A, THZ1 treatment sharply reduced the luciferase signal from days 19 to 61 compared with controls, without significantly reducing body weight (Fig. 6B). Kaplan–Meier analysis revealed that THZ1 also very significantly prolonged survival of mice compared with untreated controls (P < 0.0069; Fig. 6C). Parallel studies using a luciferase-labeled bortezomib-resistant PS-R cell flank model showed a sharp decrease in tumor size and weight following THZ1 treatment (Fig. 6D), which were both statistically significant (Fig. 6E and F). As in the systemic model, THZ1 administration did not lead to significant weight loss (Fig. 6G). Finally, as observed in vitro, Western blot analysis of excised tumors revealed downregulation of MYC, MCL-1, and BCL-X_L in tumors obtained from 2 representative THZ1-treated animals (Fig. 6H), analogous to results obtained in vitro. In addition, T161 CDK1 and T160 CDK2 phosphorylation were modestly decreased in tumors obtained from THZ1-treated animals compared with untreated controls (Fig. 6H), raising the possibility that THZ1 might exert its antitumor effects in vivo through transcriptional inhibition as well as antiproliferative actions.

Although the activity of CDKs may reflect inhibition of cell-cycle progression, other actions are likely to contribute to the antitumor activity of this class of agents. For example, CDK9 inhibitors disrupt the function of the pTEFb complex and by extension, RNA Pol II, leading to transcriptional repression of short-lived proteins necessary for tumor cell survival (28). Similar observations have been reported in the case of CDK7, which participates in the function of the TFIIH transcription factor (29), implicated in transcriptional regulation, among other functions (30). Consistent with previous findings in other tumor types, THZ1 blocked phosphorylation of the CTD of RNA Pol II at serine 2 and 5 sites (31, 32), and diminished transcription of several key antiapoptotic proteins in multiple myeloma cells. Given the pleiotropic actions of TFIIH (29), it might be anticipated that a CDK7 inhibitor such as THZ1 would act through multiple mechanisms to diminish multiple myeloma cell proliferation and survival. In this regard, the CAK-inhibitory activity of THZ1 (3) was associated with diminished phosphorylation/activation of CDK9, but also CDK1 and 2, accompanied by reductions in S-phase progression as well as G₂–M arrest. Moreover, TFIIH has also been implicated in the DNA damage response (29, 30), and CDKs are known to participate in DNA repair (1, 19), particularly CDK12/13, which potentially accounts for the increase in γH2A.X expression and the diminished expression of DDR genes (e.g., CtIP, FANCD2, RAD51, BRCA1, and ERCC1) in THZ1-treated cells. The latter phenomenon takes on added significance in light of recent evidence that multiple myeloma cells may be particularly susceptible to replicative stress and DNA damage (33). It is tempting to speculate that each of these actions, for example, downregulation of prosurvival proteins, inhibition of cell-cycle progression, and promotion of DNA damage cooperate to reduce multiple myeloma cell proliferation and promote cell death.

The c-MYC oncogene is involved in diverse oncogenic pathways, including those related to cell proliferation, survival, metabolism, and immune surveillance, among others (34). It plays a critical role in several tumor types, including neuroblastoma (35), certain lymphoid leukemia (36), and lymphomas, for example, double-hit DLBCL (37). De-regulation of c-MYC has also been implicated in myelomagenesis and c-MYC is considered a validated target in this disease (12). However, the lack of a druggable binding pocket has made direct pharmacologic disruption of c-MYC challenging, and it has been described, along with KRAS, as an undruggable target (38, 39). Nevertheless, there have been multiple attempts to target c-MYC indirectly, either by inhibiting upstream pathways or disrupting complexes in which it participates. Agents reported to interfere with c-MYC function or expression include inhibitors of β-catenin (40), aurora kinase inhibitors (41), and bromodomain extra-terminal domain (BET) inhibitors (42), which interfere with c-MYC transcription. More recently, CDK7 inhibitors, including THZ1, have been shown to be potent antagonists of c-MYC expression (4, 32). In accord with these findings, THZ1 robustly reduced c-MYC mRNA levels and protein expression in multiple myeloma cells. Notably, on a molar basis, THZ1 was considerably more potent than the BETi JQ1 in downregulating c-MYC. The observation that ectopic expression of c-MYC significantly diminished the antiproliferative and antisurvival actions of THZ1 argues that transcriptional downregulation of c-MYC plays an important role in the anti-multiple myeloma activity of this agent.

Exposure of multiple myeloma cells to THZ1 also downregulated transcription and expression of MCL-1 and BCL-X_L, both of which have been identified as determinants of both proteasome inhibitor and BH3-mimetic responsiveness (43, 44). Because of a short half-life (e.g., 2–4 hours; ref. 45), MCL-1 has been shown to be an important target of transcriptional repressive CDKIs such as alvocidib (46) and more recently, THZ1 (4, 7, 34). Although it is conceivable that downregulation of these proteins could reflect caspase-dependent cleavage occurring in apoptotic cells, the finding that downregulation persisted in cells exposed to a broad caspase inhibitor argues against this possibility. Notably, CRISPR-Cas knockdown of CDK7 sharply reduced multiple myeloma cell proliferation and survival, and significantly increased the sensitivity of multiple myeloma cells to several anti-multiple myeloma agents, including proteasome inhibitors and the BCL-2 antagonist venetoclax. Furthermore, genetic disruption of CDK7 was phenocopied by the pharmacologic agent THZ1 which potentiated the activity of each of these agents, supporting the notion that THZ1-mediated CDK7 inhibition was responsible for the observed synergism. In this context, we have recently reported that the CDK9 inhibitor alvocidib promoted cell death induction in high-risk multiple myeloma cells exhibiting low BCL-2 to MCL-1 ratios [e.g., t(4;14) multiple myeloma] and venetoclax resistance (14). In light of its ability to downregulate MCL-1 as well as BCL-X_L, it is likely that THZ1 would exert a similar activity, and efforts to confirm this possibility are currently underway. Finally, the ability of ectopic expression of c-MYC, MCL-1, and BCL-X_L to ameliorate the lethal effects of THZ1 alone or in combination with other agents argues that transcriptional repression of these proteins plays a key functional role in THZ1 anti-multiple myeloma actions.

It is important to note that THZ1 administered alone and particularly in combination with PIs or BH3-mimetics resulted in significant increases in cell death in primary, patient-derived multiple myeloma cells (CD138⁺) as well as in a population putatively enriched for multiple myeloma stem cell-like cells (CD138⁻, CD19⁺, CD20⁺, CD27⁺; ref. 47), but was relatively nontoxic to normal CD138⁻ and CD34⁺ cord blood cells. However, synergism between THZ1 and other agents (e.g., venetoclax, carfilzomib) was not as dramatic in primary myeloma cells as in their continuously cultured counterparts. This may reflect the greater susceptibility of cell lines to diverse agents compared with primary myeloma cells due to a various factors, for example, greater cytokinetic activity versus quiescent primary CD138⁺ cells. The basis for the observed selectivity of CDKIs toward multiple myeloma cells is not clear, but has been attributed to the intrinsic vulnerability of transformed cells to cell-cycle deregulation compared with their normal counterparts (48). The increased susceptibility of multiple myeloma cells to replicative stress (49) and the requirement for active transcription to circumvent this phenomenon could also contribute. Alternatively, the presence of super-enhancers at MYC and other key genes, for example, MCL-1 and BCL-X_L associated with multiple myeloma led us to test the hypothesis that super-enhancers may be particularly sensitive to CDK7 inhibitors like THZ1 in multiple myeloma cells. This appeared to be the case. However, to determine whether this phenomenon contributes to the selectivity of THZ1 toward transformed cells will require additional studies, including ChIP-seq analysis and comparisons in normal hematopoietic cells. In addition to these possibilities, several alternative mechanisms have been implicated in interactions between CDKIs and other agents in multiple myeloma cells, including upregulation of proapoptotic proteins (50, 51), induction of ER stress (52), and in the case of BH3-mimetics, disruption of cytoprotective autophagy (53). Whether any of these mechanisms contributes to the preferential targeting of multiple myelom cells by THZ1 alone or in combination remains to be determined.

Consistent with its relative sparing of normal hematopoietic cells, THZ1 administration displayed little in vivo toxicity, but significantly increased animal survival in a multiple myeloma xenograft model, while exerting minimal toxicity. CDK and checkpoint (e.g., CHK1) inhibitors often have only modest single-agent activity, but are most effective when combined with other agents. In addition, the xenograft studies demonstrated recapitulation of several of the in vitro pharmacodynamic observations. Importantly, THZ1 reduced multiple myeloma cell expression of several proteins, for example, c-MYC, MCL-1 shown to be downregulated in in vitro studies. Deregulation of both c-MYC (12) and MCL-1 (11) have been implicated in myeloma pathogenesis, raising the possibility that CDK7 inhibitors such as THZ1 may be particularly appropriate in this disease. We also found that p-CDK1 and p-CDK2 levels were modestly reduced in THZ1-treated tumors compared with untreated controls (Fig. 6H), at least at the 40-day interval. However, it is possible that different results might be obtained at alternative time intervals. Given these results, we cannot conclusively determine whether the in vivo growth suppressive effects of THZ1 stem predominantly from transcriptional repression, cell-cycle effects, or a combination of the two. Whether a CDK7 inhibitor like THZ1 will prove superior to other CDK inhibitors, for example the pan CDKi alvocidib, remains to be determined, although the ability of THZ1 to disrupt the THFII complex (31) may have therapeutic implications. The recent entry of CDK7 inhibitors (e.g., SY-1365; NCT03134638) into the clinical arena (7), along with their activity against multiple myeloma cell lines and primary multiple myeloma cells, in vivo efficacy, and synergistic interactions with PIs and BH3-mimetics argue that these agents warrant attention in multiple myeloma. Successor studies examining in vivo interactions between THZ1 and other agents, ideally in multiple myeloma PDX models as they become available, would be a logical plan. Efforts to test this notion are currently being explored.

No potential conflicts of interest were disclosed.

Conception and design: Y. Zhang, L. Zhou, D. Bandyopadhyay, S. Grant

Development of methodology: Y. Zhang, D. Bandyopadhyay, K. Sharma, A.J. Allen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Sharma, M. Kmieciak

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zhang

Writing, review, and/or revision of the manuscript: Y. Zhang, D. Bandyopadhyay, S. Grant

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zhang

Study supervision: S. Grant

This work was supported by awards CA205607 and UH2TR001373 (to S. Grant) from NCI, and R6508-18 from Leukemia and Lymphoma Society (LLS). Services and products in support of the research project were generated by the Virginia Commonwealth University Cancer Mouse Models Core Laboratory, supported, in part, with funding from NIH-NCI Cancer Center Support Grant P30 CA016059.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.