Targeting CDK12 for Cancer Therapy: Function, Mechanism, and Drug Discovery

Cyclin-dependent kinase 12 (CDK12) is one of the 20 members of cyclin-dependent kinases (CDK). Each of CDKs binds with cyclin play important roles in the control of cell cycle and cell division and regulates transcription in response to various cellular processes. CDKs can be divided into two categories: one is cell cycle–related CDKs, including CDK1, CDK2, CDK3, CDK4, CDK5, CDK6, CDK14, CDK15, CDK16, CDK17, and CDK18; and the other is transcription-related CDKs (Fig. 1A). There are extensive studies regarding CDK4, CDK6, CDK7, and CDK9 (1, 2). However, functions of several other CDKs remain largely unknown, such as CDK15, CDK16, CDK17, CDK18, CDK19, and CDK20. The comparisons of key features (subcellular location, molecular weight, cyclin partner, chromosomal location, molecular function, and key biological process) of representative CDKs are summarized in Table 1. Among the CDK family members, CDK12 and CDK13 possess the highest sequence homology with largest molecular weight. CDK13 has some similar biological functions with CDK12 because of the similar sequences; however, studies on CDK13 are not abundant yet. Among all CDKs, only CDK12 locates on the chromosome 17q12, which always contains oncogenic features and genetic alterations in various tumors (3). In this review, we mainly focus on CDK12.

Initial studies of CDK12 used the Ctk1 metazoan counterpart in Drosophila (4). CDK12 is also named as Cdc2-related kinase (CrkRS) and cell division cycle 2-related protein kinase 7 (CRK7; refs. 5, 6). CDK12 encompass two proline-rich motifs that are critical for protein–protein interactions, a carboxy-terminal kinase domain (KD) and an arginine/serine (RS) domain (Fig. 1A), which is commonly found in splicing factors of the RS-rich family (5). CDK12 mainly localizes in the nucleus and cytoplasm, which is important for embryonic development and the maintenance of genomic stability. Knockout of CDK12 (CDK12^−/−) leads to embryonic lethality (7). Other CDKs knockout mouse models phenotypes were summarized in Table 1. And CDK12 knockdown in cortical neurons decreases the length of averaged axonal by regulating CDK5 expression (8).

The cyclin partners of CDK12 initially identified were cyclin L1 and L2 (9). Cyclin K was later proved to be the credible CDK12-interacting cyclin, which is critical for CDK12 kinase activity (10–12). CDK12/Cyclin K complex plays important roles in gene expression regulation by phosphorylating RNA polymerase II (RNA Pol II; ref. 12). CDK12 can also regulate RNA splicing through interaction with RNA processing factors (13). Furthermore, CDK12 couples mRNA 3′ end processing via phosphorylation of RNA Pol II (14). Besides, CDK12 takes a significant role in regulating intronic polyadenylation and translation (15–17). CDK12 alterations including mutations, amplifications, fusions, and deletions were found in different human cancers (18–21). CDK12 is crucial in tumor progression by regulating c-MYC expression, WNT/β-catenin signaling, ErbB–PI3K–AKT signaling, MAPK signaling as well as noncanonical NF-κB pathway, and DNA damage response (DDR) signaling (22–26). It is recognized that CDK12 can act as both a biomarker and a therapeutic target in different cancer types (2, 25, 27–29). Inhibitors of CDK12 were developed and a number of studies have proved that a CDK12 inhibitor is highly effective for targeted therapy and combination therapy with anticancer drugs as well as de novo therapeutic resistance in several cancers (30–34). Herein, we summarize the literatures on CDK12 from the function, mechanism, and inhibitors, highlighting that it is an effective and promising therapeutic target in different types of human cancers.

Phosphorylation of cyclin E1 at Ser366 mediated by CDK12/cyclin K restrains the interaction of CDK2 with cyclin E1 in early G₁ phase, which suggests a novel role of CDK12 in regulating cross-talk between replication of DNA and transcription (Supplementary Fig. S1A; ref. 35). As a kinase, CDK12 phosphorylates the RAN Pol II of carboxy-terminal domain (CTD) at Ser2 and Ser5 in vitro and phosphorylates at Ser7 preferentially when it is prephosphorylated (14, 36). In addition, other non-CTD substrates and CDK12-binding partners have characterized by phosphor–proteomics and affinity–purification mass spectrometry. These non-CTD substrates are categorized as multiple protein kinases and RNA-splicing factors, with examples such as CDK5, mRNA 5′ cap-binding repressor, 4E-BP1, noncanonical NF-κB, p21-activated kinase 2 (PAK2; refs. 8, 17, 24, 26). However, the exact consequences of the phosphorylation and binding of these non-CTD CDK12 partners remain incomplete and are not well studied.

The RS motifs of CDK12 may play a role in pre-mRNA processing, CDK12 colocalizes with SC35 (a component of spliceosome, SRSF2 or SFRS2) to nuclear speckles and spliceosome, and thus is believed to be important in RNA splicing (5). CDK12-related proteins may be involved in RNA processing factors, RNA splicing machinery, as well as ALE (alternative last exon) splicing of genes containing large number exons and long transcripts (10, 13). CDK12 couples transcription and mRNA 3′ end processing by phosphorylating Ser2-RNA Pol II and interacting with the process of polyadenylation and termination machinery, recruiting the breakage and polyadenylation factor CstF77, leading to efficient 3′ end formation (Supplementary Fig. S1B; refs. 37, 38). CDK12 is required for MYC first rank pre-mRNA processing, with reduced levels of polyadenylated MYC RNA by gene depletion, indicating that CDK12 is critical for regulating intronic polyadenylation (Supplementary Fig. S1C; ref. 38). Despite the limited evidence shows that CDK12 is involved in splicing, the precise mechanisms are not fully understood. It remains for further study on mechanism to elucidate how gene selectivity is accomplished and what other partners involved in this process.

CDK12 depletion does not change transcription globally, but only alters a subset of genes associated with DDR. The most altered genes are those with large number of exons, which includes BRCA1, ATR, FANCI, and FANCD2 (11, 14). Loss of CDK12 leads to reduced transcription of BRCA1 and increased sensitivity to DNA-damaging agents. More importantly, mutations of CDK12 kinase domain lead to impairment of DNA double-strand breaks (DSB) through homologous recombination (HR; ref. 11). CDK12 is required for the expression of oxidative stress response-related genes and function of stress-activated Nrf2 transcription factor (39). In summary, CDK12 regulates the particular subclass of genes that participate in cellular responses to DNA damage, stress, and heat shock (11, 39, 40). However, molecular mechanisms for this specificity remain not entirely understood.

CDK12 has the highest alteration frequency among all CDKs with large percentage of mutations, amplifications, fusions, and multiple alterations (Fig. 1B). Hundreds of mutations spread over the whole protein sequence in different domains (Fig. 1B). Genomic alterations of CDK12 have identified in different human cancer types, including esophageal cancer, bladder cancer, colon cancer, prostate cancer, breast cancer, lung cancer, ovarian cancer, and other cancers (Fig. 1B). The high mutation frequency tissues belong to ovary, adrenal gland, stomach, breast, and urinary tract among different human tissues (Fig. 1B), but there is no more study of CDK12 mutations in other tissues besides ovary, prostate, breast, and lung.

CDK12 mutation frequency ranks the top 3 in patients with advanced non–small cell lung cancer (NSCLC) with bone metastasis. A nonfunctional novel CDK12 site mutation (G879V) in NSCLC can cause genomic instability and enhance sensitivity to chemotherapy (19, 41, 42). It was reported that a 61% mutation rate of CDK12 in 107 Chinese patients with HER2-positive breast cancer (43). CDK12 locates proximal to HER2 at approximately 200 kb; they are frequently coamplified in breast tumors, which accounts for around 90% of HER2-positive breast cancer (13, 25, 44, 45). The homozygous point mutations in the kinase domain of CDK12 are the main mutations in high-grade serous ovarian carcinoma (HGSOC), which abolish the catalytic activity of CDK12, cause multiple DNA repair pathways defects, leading to decreased HR and genomic instability (46, 47).

The well-studied mutation of CDK12 belongs to tandem duplications (TD), mainly in ovarian and prostate cancer (20). TD is a structural rearrangement that generates physically contiguous, head-to-tail duplications of a segment of DNA. The phenotypes of TD was uncovered in tumors show no obvious single coding or noncoding location, producing hundreds of small copy number gains, suggesting a systemic etiology rooted in a defect in DNA repair or replication, specifically, interfering with heterogeneous combinations of tumor suppressors and chromatin topological associated domains, while simultaneously replicating oncogenes and super enhancers (48–50). The number of focal tandem duplications (FTD) is significantly increased in CDK12 loss of function cases compared with CDK12 wild-type cases in ovarian, breast, gastric/esophageal, and endometrial cancer (20). The rate of CDK12 loss is around 3% to 7% in patients with metastatic prostate cancer, with a genomic instability signature, and patients with CDK12 mutation exhibited fast metastasis from original cancer and progression of castration-resistant disease compared with other genomic subtypes. It also showed fast progression on first-line ARPI (androgen receptor pathway inhibitor) treatment of metastatic castration-resistant disease (51–56). There are evidences suggesting that inactivating CDK12 alterations in prostate cancer correlates with poor outcomes to hormonal, taxane, and PARP inhibitors, while promoting sensitivity to immunotherapy (20, 57, 58). It is reported that CDK12 mutation was correlated with poor disease-free survival in Chinese patients with prostate cancer (59). The retrospective analysis of CDK12 polymorphisms in patients with late-stage epithelial ovarian cancer revealed that the G/G genotype of CDK12 polymorphism (rs1054488) predicted worse overall survival and progression-free survival than the genotype A/A-A/G (60). Several clinical trials evaluating CDK12 mutational status as a biomarker in various cancer types have been conducted (Supplementary Table S1; ref. 2). All these findings indicate that CDK12 alterations may be a potential biomarker for response to immune checkpoint blockade in cancer.

Cell-cycle processes consistent with DDR mechanisms ensure genomic stability. Cell division or exogenous DNA-damaging agents induce DSBs of DNA that can cause genomic instability. Aborted DSBs repair may cause chromosome breakage or rearrangement. These two kinds of chromosomal damages may be repaired by HR and NHEJ (nonhomologous end joining) typically (61–65). CDK12 protein absence in breast cancer cells may reduce the expression of many DDR-related proteins including ATR, γH2AX, PARP1, DNA-PK, and Ku70/Ku80. Lacking of CDK12 will result in defects in gene length-dependent elongation, inducing PCPA (premature cleavage and polyadenylation) and missing of long genes expression (>45 kb). CDK12 phosphorylates pre-mRNA processing factors directly, including those genes regulating PCPA. DDR genes may susceptible to CDK12 inhibition (15, 16, 25). CDK12 can work with mTORC1 and control a specialized translation network that is critical to the stability of mitotic chromosomes (17). Cell-cycle checkpoints are used in situations that are detrimental to complete cell division, such as when DNA damage is present (66, 67). As a major effector of mammalian cell progression, many DNA damage checkpoint signaling pathways focus on the inactivation of the CDK/cyclin complex (68, 69). It was reported that CDK12 kinase activity is essential for core DNA replication to gene transcription and G₁–S phase progression (14). Knocking down of CDK12 in human gastric cancer cells induces G₂ phase arrest, demonstrating that CDK12 is involved in checkpoint signaling, which needs more investigation in the detailed molecular basis (14).

MYC is an overall transcription factor and core driving force in tumor progression. Inhibition of CDK12 triggers extensive downregulation of c-MYC and its related genes (70–72). Furthermore, MYC downregulation causes replicative stress via accelerating replication rate. The overlapped functions between MYC and CDK12 suggest that targeting CDK12 may be an effective strategy against MYC-dependent human cancers (73–75). It is especially important because MYC has been proved difficult to inhibit directly. NF-κB transcription factors are pivotal for multitudinous cellular processes, including cell proliferation, differentiation, survival, apoptosis, angiogenesis, inflammation, immunity, tumorigenesis, as well as regulating innate and adaptive (76–78). The noncanonical NF-κB (n-NF-κB) pathway is a promising therapeutic strategy for a number of diseases and cancers (79, 80). Canonical NF-κB complexes are retained by IκBα in the cytoplasm; however, n-NF-κB complexes are reserved by p100 (encoded by NFKB2), and the accumulation of NIK (NF-κB–inducing kinase, encoded by MAP3K14). CDK12 was found to promote the ligand-induced increase in MAP3K14 and NFKB2 mRNAs abundances (Fig. 1C; ref. 24). This molecular basis suggests that targeting CDK12 to regulate NF-κB may be useful for inhibiting NF-κB–dependent cancers.

Cell signal transduction is a fundamental and pivotal process in tumor development and progression; dysregulation of various cell signaling pathways promotes cancer cell proliferation, differentiation, motility, metabolic disturbance, and immune response regulation, leading to aberrant tumor growth, invasion, angiogenesis, and metastasis (81). ErbB–PI3K–AKT, WNT/β-catenin, and MAPK signaling are the canonical and critical signal transduction pathways in carcinogenesis. Inhibition of these pathways is one of the main strategies for cancer targeted therapy (82–84). CDK12 can activate ErbB–PI3K–AKT or WNT/β-catenin signaling by phosphorylating RNA POL II to promote tumor initiation, invasion, and metastasis, as well as maintain cancer stem cells self-renewal in breast cancer (Fig. 1C; refs. 22, 23). These findings suggest that CDK12 is a novel therapeutic strategy for anticancer therapy, especially for HER2-positive breast cancer. Although the CTD substrates of CDK12 have identified a lot, the non-CTD substrates of CDK12 remain deserted. Our group identified that PAK2 can act as a non-CTD substrate of CDK12. CDK12 phosphorylates PAK2 at threonine 134/169 activating MAPK signaling pathway to promote tumor development in human gastric cancer (Fig. 1C; ref. 26). It reminds us CDK12 can act as a novel therapeutic target for human gastric cancer besides VEGFR-2, HER2, and PD-1 (85–88). We believe that other non-CTD substrates and CDK12-related proteins will be discovered and their roles in tumorigenesis will be verified.

Given the critical roles of CDK12 in diseases and cancers, we expect that the inhibitor of CDK12 can be used for cancer prevention and therapy. The overview of CDK inhibitors is shown in Supplementary Table S2. The specific CDK12 inhibitor is CDK12-IN-3, and the others are nonspecific CDK12 inhibitors. Most of them inhibit tumor growth by suppressing DDR, phospho-RNA POLII and have synergistic effect with PARPi.

It is reported that dinaciclib (also known as SCH 727965), an inhibitor of CDK1, 2, 5, 9, additionally has activity against CDK12 ablating restored HR and reverses PARP inhibitors (PARPi) resistance, supporting the widespread use of combined CDK12 and PARPi in triple-negative breast cancer (TNBC; refs. 89–92). A study verified that inhibiting CDK12 by using THZ1 and THZ531 (covalent small molecule) can suppress expression of DDR genes and is synergistic with PARPi in Ewing sarcoma (33). THZ531 induced significant DDR in hepatocellular carcinoma (HCC) cell lines, and combination of THZ531 with sorafenib showed strong synergy in HCC, suggesting a potential combination therapy strategy for HCC (32). Another study identified that inhibiting CDK7, CDK12, and CDK13 by THZ1 significantly downregulates expression of MYC, leading to dramatic inhibition of tumor growth and MYC expression, providing an effective way for MYC-dependent cancer treatment (75). CDK12 inhibition by THZ1 also resulted in global transcriptional changes and accumulation of NIK via reducing RNA Pol II phosphorylation, which can be used as a therapeutic strategy for autoimmunity and cancers (24). E9 (structurally like dinaciclib) is an irreversible inhibitor of CDK12, leading to a phosphorylated and total RNA poly II decrease in together with MYC and Mcl-1 downregulated, while PARP cleavage increasing (93). Highly selective CDK12 inhibitor CDK12-IN-3 shows potent inhibition of phosphorylation of Ser2 RNA Pol II on the CTD repeat domain in OV90 cells (30). SR-4835, a highly selective dual inhibitor of CDK12 and CDK13, was reported to inhibit TNBC cells by triggering the cleavage of intron polyadenylation sites; the expression of core DNA damage reactive proteins is inhibited, thus promoting the synergistic effect with DNA damage chemotherapy and PARPi (94). One study reported that silencing BRCA1 or CDK12 sensitizes tumor cells to CHK1 inhibitors regardless of p53 status, suggesting that inhibition of CHK1 is a strategy against BRCA1- or CDK12-deficient tumors (34). Our group discovered that procaterol, a clinical used β₂ receptor agonist, can act as CDK12 inhibitor (26, 95, 96). Procaterol inhibits human gastric cancer cell proliferation and tumor growth by inhibiting CDK12 kinase activity, which may be translated into clinic after conducting clinical trial (26). In summary, CDK12 inhibitors can be used for targeted therapy, can overcome drug resistance, and show synergistic effect with other anticancer drugs in different human cancers. But there is no CDK12 inhibitor now used in clinic for CDK12-targeted therapy.

The new era of immunotherapy has changed the practice of clinical oncology. There is an urgent need to develop new strategies to modalize the clinical outcomes of immunotherapy and to extend its benefits beyond the PD-1/PD-L1 signaling pathway to a broader population of patients with cancer (97). Several nonclassical molecular immune targets have been shown to act as feedback resistance circuits to shut down the classical immune checkpoint inhibitor–mediated antitumor immune response, including CD40, CD47, CD134, T-cell inducible costimulator, Toll-like receptors, and CDK12 (27, 98–108). Novel combinatorial approaches to improve the effect of cancer immunotherapy are needed based on the classical immunotherapies and strategies. It was reported that deletions of CDK12 bialleles showed genomic instability and increased neoantigen load, followed by enhanced tumor T-cell infiltration, and 50% of patients with mCRPC responded positively to PD-1 blocking (reduced PSA levels; refs. 27, 109). This report suggests that CDK12 loss in mCRPC may act as a hopeful prognostic biomarker for the potential benefits of immune checkpoint immunotherapy, and a new combination method applying CDK12 inhibitors as potential sensitizing agents to heighten the response to immune checkpoint antibody therapy may be useful in prostate tumors. We expect that the combination of CDK12 inhibitors with immune therapy has a wider application for the foreseeable future. In addition, it was reported that a novel compound (DDD853651/GSK3186899) is efficacious in a Visceral leishmaniasis (VL) mouse model by inhibiting CDC2-related kinase 12 (CRK12), the Leishmania homolog of CDK12, indicating that targeting CDK12 may have broad indications in human diseases beyond cancer (110).

CDK12 is emerging as a critical signaling molecular for cancer development. As a transcription associated CDK, CDK12 binds with cyclin K phosphorylate RNA Pol II at CTD promoting transcription elongation. CDK12 can also interact with RNA processing factors to regulate RNA splicing. Furthermore, CDK12 mediated phosphorylation of RNA Pol II couples mRNA 3′ end processing to participate in transcription termination. Besides, CDK12 plays critical role in regulating intronic polyadenylation, epigenetics, as well as translation. CDK12 alterations were found in different human cancers. Loss of function of CDK12 mutations or deletion can cause genomic instability, which may act as a biomarker in ovarian cancer and prostate cancer. The critical role of CDK12 in tumorigenesis and progress may mainly via activating c-MYC/β-catenin signaling, ErbB–PI3K–AKT, or WNT signaling cascades, MAPK signaling, noncanonical NF-κB pathway, and DDR signaling, and a number of studies have proved that CDK12 inhibitor is highly effective in cancer-targeted therapy and combination therapy or together with immune therapy (Fig. 1D).

Although CDK12 has received sustained attention for nearly 15 years, there are still many key questions waiting for investigation. For example, how CDK12 interacts with others to regulate complex molecular processes? What intrinsic redundancies are in place? How CDK12 works with other CDK members to affect cancer hallmarks? What are the consequences of CDK12 loss or accumulation in tumorigenesis of different cancer types and the posttranslational modification including phosphorylation, acetylation, methylation, ubiquitination, and degradation? Furthermore, the diseases beyond cancer that may be affected by CDK12 and benefited from CDK12 inhibitors also remain to be studied. In addition, much more work on developing effective specific CDK12 inhibitors is important, as the current inhibitors of CDK12 have not been used clinically yet. To uncover the answers of these questions, researchers may need more work, such as using CDK12 conditional knockout mouse to verify the role of CDK12 in tumorigenesis precisely for different cancer types. Also, we need to generate the phospho-CDK12 antibody to elucidate the consequence of phosphorylated CDK12 in diseases and cancers. As for CDK12 inhibitor, researchers can also find some natural compounds from herbs or fruits, which may inhibit CDK12 and used for chemoprevention or therapy of CDK12-related cancers.

Understanding the function, mechanism, and inhibition of CDK12 is an exciting area of oncology. We are waiting for the entry of CDK12 inhibitors into clinical trials, as well as looking forward to the identification of an effective combination therapy of CDK12 inhibitors with other anticancer agents or immune checkpoint inhibitors with elucidative meticulous mechanisms.

No disclosures were reported.

This work was supported by the China-US (Henan) Hormel Cancer Institute.