1 Introduction

Cyclin-dependent kinase 7 (CDK7), along with cyclin H and MAT1, comprises the CDK-activating kinase (CAK), which provides the T-loop phosphorylation required for activation of CDKs 1,2, 4 and 6, which drive cell cycle progression (Table 1, Fig. 1a) [1,2,3,4]. CAK also has a role in the regulation of transcription, as a component of the general transcription factor TFIIH. At active gene promoters, CDK7 phosphorylates the C-terminal domain (CTD) of RNA polymerase II (Pol II), at serine 5 (Ser5), to facilitate transcription initiation (Table 1, Fig. 1b) [5,6,7]. CDK7 also phosphorylates CDK9, which in turn phosphorylates the Pol II CTD at Ser2, to drive transcription elongation [8]. The activities of a variety of transcription factors, including p53 [9, 10], retinoic acid receptor [11,12,13], oestrogen receptor [14, 15] and androgen receptor [16, 17], are also regulated by CDK7-mediated phosphorylation (Table 1).

Table 1 CDK7 substrates
Fig. 1
figure 1

Overview of the regulation of CAK and the role of CDK7 in regulating the cell cycle (a) and transcription (b). CAK = CDK activating kinase, CDK = cyclin-dependent kinase, CK2 = protein kinase CK2, G1 = gap phase 1, G2 = gap phase 2, M = mitosis, P = phosphate, PKCι = protein kinase C iota, Pol II = RNA polymerase II, S = synthesis, TFIIH = transcription factor II H

Because of its dual role in regulating the cell cycle and transcription, CDK7 has been studied as an anticancer drug target, and a number of selective inhibitors of CDK7 have been developed and investigated as cancer therapies. Preclinical studies have revealed that cancer cells can be preferentially targeted by transcriptional inhibition, at least in part because they are more reliant than normal cells on high levels of super-enhancer (SE)-driven transcription [18, 19] mediated by specific oncogenic drivers, such as RUNX1 in acute lymphoblastic lymphoma (ALL) [20] and N-MYC in neuroblastoma [21]. To date, four selective CDK7 inhibitors, ICEC0942 [22], SY-1365 [23], SY-5609 [24, 25] and LY340515 [26], have progressed to Phase I/II clinical trial for the treatment of advanced solid malignancies.

In this review we outline the role of CDK7 in both normal and tumour cells and the rationale for inhibiting CDK7 in cancer. We also discuss the development of selective CDK7 inhibitors, their mechanism of action in cancer and their potential for use in combination therapies.

2 CDK7 function

2.1 CAK structure and regulation

CDK7 is a 346 amino acid kinase, having a predicted molecular mass of 39 kDa, with an N-terminal cyclin H-binding region and a C-terminal MAT1 binding region [27]. A single crystal structure has been reported for CDK7 bound to ATP, in the inactive conformation, the structure being similar to that of the inactive conformation of ATP-bound CDK2 [28]. Cyclin H binding is obligatory for CDK7 kinase activity, whilst the addition of MAT1 stabilises the trimeric CAK complex and anchors it to TFIIH [27]. In addition, cyclin H and MAT1 binding have been shown to regulate CDK7 substrate specificity, with the trimeric CDK7-cyclin H-MAT1 complex having greater kinase activity for Pol II, in comparison to CDK7-cyclin H, which preferentially phosphorylates CDK2 [27,28,29].

The T-loop of CDK7 can be phosphorylated at two positions, threonine 170 (Thr170) and Ser164, enhancing both its kinase activity and ability to bind cyclin H [6]. Furthermore, T-loop phosphorylation of CDK7 seems to direct substrate specificity, with Thr170 phosphorylation stimulating activity towards Pol II over CDK2 [29]. In vitro, CDK1 and CDK2 can phosphorylate CDK7 and as substrates of CDK7 themselves; this hints at the possibility of a reinforcement activation loop between these CDKs [30]. In addition, protein kinase C iota (PKCι), acting downstream of PI3K signalling, can phosphorylate CDK7 at Thr170 (Fig. 1a) [31,32,33,34,35].

Regulation of CAK activity may also be mediated through phosphorylation of cyclin H. CK2 can activate CAK in vitro, via phosphorylation of cyclin H at Thr315 (Fig. 1a) [36], whereas CDK8 has been shown to negatively regulate transcription initiation, via phosphorylation of cyclin H at Ser5 and Ser304 (Fig. 1a) [37]. Furthermore, CDK7 complexed with cyclin H and/or the trimeric CAK can phosphorylate cyclin H in vitro. This autophosphorylation reduces activity of CDK7-cyclin H but has no apparent effect on CDK7-cyclin H-MAT1 activity. This suggests that MAT1 binding aids maintenance of the transcriptional activity of CAK by preventing regulation by cyclin H phosphorylation [28].

An additional means of CDK7 regulation has been observed in mouse neural progenitor cells, where the microRNA (miRNA) miR-210 regulates cell cycle progression by modulating expression levels of CDK7 [38]. This raises the possibility that there may be additional miRNAs that regulate CAK expression and activity in other cellular contexts. There is clearly more to be discovered with regard to the regulation of CDK7 and CAK activity and the identification of players acting upstream of CDK7 could potentially provide additional means by which to manipulate CDK7 activity.

2.2 CDK7 in the cell cycle

CDK7 controls the cell cycle by phosphorylating the cell cycle CDKs 1, 2, 4 and 6 in their T-loops, to promote their activities (Fig. 1a) [1]. Both CDK1 and CDK2 are activated by CDK7-mediated T-loop phosphorylation, at Thr161 and Thr160, respectively (Table 1) [2, 20,21,22, 39]. Inhibiting CDK7 during G1 prevents CDK2 activation and delays S phase, whilst inhibition of CDK7 during S/G2 prevents CDK1 activation and mitotic entry [2, 22]. Whilst CDK7 can phosphorylate CDK2 prior to its binding to cyclin, and is not strictly required for the formation of CDK2-cyclin complexes, CDK7 phosphorylates CDK1 in concert with cyclin B binding and is required for the stabilisation of CDK1-cyclin B complexes [2, 40].

Full commitment to the cell cycle is controlled at the restriction point, through phosphorylation of retinoblastoma (RB) by CDK4/6-cyclin D, in response to mitogens (Fig. 1a). CDK7 phosphorylates both CDK4 and CDK6 in their T-loops, at Thr172 and Thr177 (Table 1), respectively, and CDK7 inhibition prevents their RB kinase activity, halting G1 progression [3, 4]. Although expression levels of the CAK components remain constant throughout the cell cycle, T-loop phosphorylation of CDK7 increases when cells are released from serum starvation [3]. Therefore, a mitogen-induced cascade of CDK T-loop phosphorylation regulates progression through G1 [3].

Unlike cyclin-bound CDK2, which remains phosphorylated for up to 12 hours after CDK7 inhibition, CDK4 and CDK6 activity is rapidly lost following CDK7 inhibition [3]. This difference is likely due to structural differences between the complexes; the T-loop of CDK2 is protected from dephosphorylation by cyclin binding, whereas the T-loops of cyclin D-bound CDK4/6 remain exposed to phosphatases [3]. As a result, CDK7 activity is required to maintain CDK4/6 activity during G1 whilst being required only for initial activation of CDK1 and CDK2 during S/G2 [3].

2.3 CDK7 in transcription

CDK7 regulates gene expression, as a component of the general transcription factor complex, TFIIH (Fig. 1b). TFIIH is composed of two distinct sub-complexes: the core complex, which contains two DNA helicases, xeroderma pigmentosum type B (XPB) and xeroderma pigmentosum type D (XPD), along with five other structural and regulatory proteins, and the CAK complex. CAK is recruited to the core TFIIH complex via a reversible interaction between the ARCH domain of XPD and the latch domain of MAT1 [41, 42]. TFIIH is recruited by TFIIE to active gene promoters, where it joins the other assembled general transcription factors (TFs), and Pol II, in the preinitiation complex (PIC) [27]. The composition of TFIIH and the structure of the PIC have recently been reviewed by Rimel and Taatjes [27].

After DNA is unwound at the transcription start site (TSS) by XPB [43], Pol II must be released from the PIC to initiate transcription, in a CDK7-regulated process termed promoter escape [5]. The CTD of mammalian RPB1, the largest subunit of Pol II, contains 52 repeats of a heptad sequence, conforming to the consensus Y1-S2-P3-T4-S5-P6-S7, the residues of which can be sequentially phosphorylated to regulate Pol II activity throughout the transcription cycle [44]. Whilst unphosphorylated, Pol II remains anchored to the PIC, via an interaction with the mediator complex (another PIC component) [5]. CDK7 phosphorylates Ser5 and Ser7 of the Pol II CTD at gene promoters [6, 7]; Ser5 phosphorylation facilitates the release of Pol II from mediator, allowing Pol II to escape the PIC and initiate transcription (Table 1, Fig. 1b) [5, 45]. The precise function of CDK7-directed Ser7 phosphorylation is as yet unclear, but evidence suggests that Ser7 phosphorylation may promote the transcription and post-transcriptional processing of small nuclear RNA transcripts, by facilitating an interaction between the integrator complex and Pol II [46].

After promoter escape, Pol II generally generates a transcript of around 20–80 bases, before halting progress, in a process known as promoter-proximal pausing, which likely functions as a checkpoint to ensure the establishment of a range of co-transcriptional processes [6, 47]. CDK7 is required for the recruitment of two complexes, the DRB sensitivity inducing factor (DSIF) and the negative elongation factor (NELF), both of which are required to establish the promoter-proximal pause [6, 8, 48,49,50]. For the release of paused Pol II and commencement of the productive elongation phase of transcription, the activity of CDK9, as a component of the positive transcription elongation factor (P-TEFb), is required [8]. Like the cell cycle CDKs, for full functionality, CDK9 must undergo T-loop phosphorylation by CDK7 (Table 1, Fig. 1b) [8]. Therefore, CDK7 plays a role in both establishing the promoter-proximal pause and in release from the pause, and inhibition of CDK7 has been shown to increase the amount of Pol II paused at promoter-proximal regions [6, 51]. Active CDK9 phosphorylates the Pol II CTD, on Ser2, promoting transcriptional elongation [52]; therefore, there is an indirect requirement for CDK7 activity after Pol II pause release.

CDK7 also regulates further transcriptional processes; for example, CTD phosphorylation by CDK7 allows the co-transcriptional interaction of Pol II with enzymes that add the 5′-monomethyl-guanosine cap to nascent RNA transcripts [50]. Additionally, CDK7 is necessary for appropriate transcription termination, with read-through transcription observed upon CDK7 inhibition [6]. CDK12 and CDK13 are also involved in regulating transcription by phosphorylating the Pol II CTD during elongation [53]. In vitro, CDK12 can phosphorylate Ser2, Ser5 and Ser7 [54], whereas CDK13 can phosphorylate Ser2 and Ser5 [55]. Like the previously discussed CDKs, T-loop phosphorylation is necessary for CDK12/13 activation and is likely mediated by CDK7 [54, 56]; thus, it is probable that additional transcriptional substrates of CDK7, and further roles in transcriptional regulation, remain to be identified.

Genetic targeting of Mat1 or Cdk7 in mice is early embryonic lethal and cells cultured from embryos of these animals fail to enter S phase [57, 58]. The activities of Cdks 2, 4 and 6 are reduced in mouse embryonic fibroblasts (MEFs) with Cdk7 knockout, indicating that Cdk7 has an essential role in cell proliferation [58]. Cdk7 targeting in adult animals results in phenotypically normal low-proliferating tissues, such as the liver, kidney or cerebellum. However, in rapidly dividing epithelial tissues, Cdk7 expression is retained due to tissue renewal sustained by stem cells with incomplete Cdk7 knockout. This eventually leads to stem cell exhaustion and premature ageing [58]. Interestingly, MEFs lacking Cdk7 expression have unaltered Pol II CTD Ser5 phosphorylation and a largely unchanged gene expression program, indicating that Cdk7 is dispensable for de novo transcription [58]. This raises the possibility that another Pol II CTD kinase can compensate for a lack of Cdk7.

2.4 CDK7 as a regulator of transcription factor activity

Alongside its critical role in directing transcription by Pol II, CDK7 phosphorylates a number of TFs, functioning to either promote their activities and/or regulate their degradation (Table 1). The activity of retinoic acid receptor ⍺ (RAR⍺) is promoted by XPD-dependent phosphorylation of Ser77 by CDK7 [11, 13]. Likewise, the activity of RARγ is also modulated by phosphorylation by TFIIH-incorporated CDK7 [12]. CDK7, as part of TFIIH, mediates ligand-dependent phosphorylation of oestrogen receptor ⍺ (ER⍺) at Ser118 [14, 15], regulating the activity and turnover of the TF [59, 60]. Phosphorylation by CDK7, at Ser515 in the transcription activation function of androgen receptor (AR), has also been reported [16, 17]. Additionally, CDK7 can phosphorylate p53 in a MAT1-dependent fashion, at both the C-terminus (between residues 311 and 393) [10] and the N-terminus, at Ser33 [9], the former of which has been shown to stimulate p53 binding to DNA. Evidence that CDK7 phosphorylates Ets1 [61], peroxisome proliferator-activated receptors (PPARs) [62] and E2F1 has also been demonstrated, the latter functioning to trigger E2F1 degradation [63] (Table 1). Recently, the stabilisation of the transcriptional regulators YAP/TAZ was shown to be mediated by CDK7, with phosphorylation of YAP at Ser128 and TAZ at Ser90, preventing their ubiquitination and degradation [64]. At present we have an incomplete understanding of the role CDK7 plays in regulating the activities of sequence-specific transcriptional regulators. Further knowledge in this area may be helpful in informing the use of CDK7 inhibitors in specific cellular contexts.

2.5 CDK7 in DNA repair

TFIIH plays a key role in the nucleotide excision repair (NER) pathway [27], which repairs single-stranded DNA damage, particularly that caused by ultraviolet light. TFIIH is recruited to damaged DNA, where the NER protein, xeroderma pigmentosum group A (XPA), catalyses the release of CAK from the core TFIIH complex, allowing NER to proceed [65]. After DNA repair, CAK reassociates with TFIIH, and the complex resumes its role in transcription [65]. Inhibition of CDK7 kinase activity improves NER efficiency, suggesting that CDK7 negatively regulates NER, directly or indirectly, via phosphorylation of an as yet unidentified substrate(s) [66].

3 CDK7 in cancer

3.1 CDK7 expression in tumours

Two decades ago, immunohistochemical analyses on a range of tumour types indicated that CDK7 expression is elevated in tumour cells compared with their normal counterparts [67]. Since then, numerous studies have provided support for this finding [68,50, 51]; however, it was recently shown that its anti-transcriptional and antitumour activities are reliant on inhibition of CDK12 and CDK13, in addition to CDK7 [130]. Consequently, YKL-5-124 was developed, with a strategy that combined the covalent warhead of THZ1 with the pyrrolidinopyrazole core of the PAK4 inhibitor, PF-3758309 [130]. Like THZ1, YKL-5-124 covalently links to Cys312 of CDK7 but does not affect the activities of CDK12 and 13 (Fig. 3b, Table 3) [130]. An analogue of THZ1, with altered regiochemistry of the acrylamide and increased in vivo stability has also been developed and was designated THZ2 (Fig. 3b, Table 3) [

6 Conclusions

CDK7 has a dual role in driving the cell cycle and transcription, is upregulated in a variety of cancers and has emerged as a promising cancer therapeutic target. At least ten selective inhibitors of CDK7, with activity against a wide range of cancer types, have been developed, their antitumour action likely mediated both through cell cycle arrest and inhibition of oncogenic transcriptional programs. In the preclinical setting, these inhibitors have demonstrated potential to overcome treatment-resistant cancer, both as monotherapies, and in combination with other cancer drugs. To date, four CDK7 inhibitors have progressed to Phase I/II clinical trial for the treatment of advanced solid malignancies. Whilst ABC-transporters can mediate resistance to some CDK7 inhibitors, additional factors that influence tumour response to CDK7 inhibition are yet to be identified. Further efforts to elucidate mechanisms of response, and to define patient selection strategies, will help to facilitate the clinical utility of CDK7 inhibitors.