Advancements in Chemical Biology Targeting the Kinases and Phosphatases of RNA Polymerase II-mediated Transcription

Abstract

Phosphorylation of RNA polymerase II (RNAP II) coordinates the temporal progression of eukaryotic transcription. The development and application of chemical genetic methods have enhanced our ability to investigate the intricate and intertwined pathways regulated by the kinases and phosphatases targeting RNAP II to ensure transcription accuracy and efficiency. Although identifying small molecules that modulate these enzymes has been challenging due to their highly conserved structures, powerful new chemical biology strategies such as targeted covalent inhibitors and small molecule degraders have significantly improved chemical probe specificity. The recent success in discovering phosphatase holoenzyme activators and inhibitors, which demonstrates the feasibility of selective targeting of individual phosphatase complexes, opens up new avenues into the study of transcription. Herein, we summarize how chemical biology is used to delineate kinases’ identities involved in RNAP II regulation and new concepts in inhibitor/activator design implemented for kinases/phosphatases involved in modulating RNAP II-mediated transcription.

Graphical Abstract

A new era of studying eukaryotic transcription began with the successful purification of RNAP II fifty years ago [1]. Since then, RNAP II regulation has become the center of extensive investigation interrogated by multi-disciplinary approaches [2]. Genetic analysis using functional mutants is robust and highly specific in revealing fundamental mechanisms. However, results can be convoluted and difficult to interpret due to inadvertent catalytic inactivation, structural disruption, compensation mechanisms, or even gain-of-function effects. In this aspect, small molecules impeding the transcription process exhibit obvious advantages and complement genetic methods [3]. Indeed, small molecules can instantly impact the target upon entering the cells, which elicits direct physiological effects and provides high temporal resolution for biological observations. This acute effect alleviates the survival pressure commonly associated with genetic manipulations, which might activate a compensatory pathway or secondary mechanism. The time and dosage-dependent action and the reversibility of many chemical probes reveal the cascade of regulatory pathways and differentiate direct biological effects from secondary compensatory mechanisms. However, off-target effects continue to be a major concern regarding the use of chemical tools but can be circumvented by strategically designing chemical compounds with high specificity.

Phosphorylation/dephosphorylation is a well-established mechanism for transcriptional regulation. Classic examples include the phosphorylation of CREB through the cAMP/PKA pathway in response to hormonal stimulations [4] and regulation of the NFAT pathway by the Ca²⁺-dependent protein phosphatase calcineurin in response to calcium stimulation [5]. In addition to the gene-specific regulation through individual transcriptional factors, phosphorylation can also regulate the general transcription with RNAP II as the primary target. Phosphorylation sites are highly enriched in the C-terminal domain of the largest subunit of RNAP II (CTD) [6]. This domain undergoes extensive post-translational modification temporally throughout transcription, which recruits various regulatory proteins at specific stages of the eukaryotic transcriptional cycle and coordinates the whole process of transcription [7]. Thus, kinases and phosphatases work together to ensure effective and accurate transmission of information during transcription (Figure 1). The temporal control gained by the application of natural products (e.g., 5,6-Dichloro-1-β-ribo-furanosylbenzimidazole and flavopiridol) inhibiting CTD kinases gave the first glimpse of transcription events such as promoter-proximal pausing and termination [8,9] (Figure 1). Here, we review emerging chemical biology strategies regarding the elucidation of sophisticated transcription mechanisms mediated by RNAP II phosphorylation. Specifically, we discuss the recent advances in the chemical regulators of the RNAP II regulatory kinases/phosphatases and their pharmacological potential.

Figure 1. The phosphorylation regulation of the CTD.

While the CTD sequence seems very simple with the consensus heptad (Y₁S₂P₃T₄S₅P₆S₇) repeated 17 to 52 times depending on species, this domain undergoes extensive spatiotemporal phosphorylation throughout transcription by various kinases and phosphatases. These enzymes are highly conserved through eukaryotes. Abbreviation used in the figure: Saccharomyces cerevisiae, S.c.; Saccharomyces pombe, S.p.; mitotic catastrophe suppressor 6, MCS6; Cyclin-dependent Kinase 7, CDK7; Positive transcription elongation factor b, P-TEFb; Cyclin-dependent Kinase 9, CDK9; Pombe cyclin C homolog 1, PCH1; general transcription factor II E, TFIIE; DRB sensitivity-inducing factor, DSIF; negative elongation factor, NELF; Serine/threonine-protein phosphatase 2A, PP2A; Serine/threonine-protein phosphatase 1, PP1; capping enzyme, CE; RNA polymerase II-associated protein 2, RPAP2; Cyclin-dependent kinase 12/13, CDK12/13.

Elucidating the function of kinases that target RNAP II with analog-sensitive allele

Historically, investigators have faced the challenge of differentiating the direct effects of enzymatic activity from secondary effects when studying phosphorylation/dephosphorylation of RNAP II. A chemical genetic method called analog-sensitive (AS) kinase design enables specific substrate identification for kinases (the rationale of the design is summarized in Figure 2s and its legend). For brevity, this review focuses on the most recent and impactful discoveries using the AS method. Comprehensive and in-depth analysis on the functions of CTD kinases during transcription can be found in the recent review by Parua and Fisher (2020) [10].

Figure 2. Scheme of the analog-sensitive kinase design method.

The application of this method to study kinases has elucidated the transcriptional events control by RNAP II phosphorylation. A “gate-keeper” residue is identified and mutated in the target kinase’s active site to increase the pocket volume without compromising kinase activity, as in Figure 2a. This mutation accommodates the binding of a bulky ATP-analog, which cannot inhibit any wild type kinases due to steric hindrance (Figure 2c). In the presence of the bulkier inhibitors, the engineered kinase is specifically inhibited while no other kinase pathway is affected. (a) Chemical structure of bulky ATP-analogs which selectively inhibit Analog-sensitive CDKs. (b) Sequence alignment shows highly conserved ‘gate-keeper’ phenylalanine residue among CDKs, mutated into glycine or alanine to create AS mutants. (c) Structural modeling of CDK12 WT (light blue ribbon) & AS mutant (wheat ribbon) illustrates how steric hindrance between the ‘gate-keeper’ residue and bulky ATP-analog (yellow stick) becomes a determinant for selective inhibition. Structural modeling was based on WT CDK12 structure (PDB code: 4NST) and 1-NM-PP1 (Bulky ATP-analog) binding mode with analog-sensitive src kinase (PDB code: 4LGH)

The use of AS method has been incredibly effective in revealing the transcriptional roles of CTD kinases during different stages of transcription (Figure 1). For example, CDK7 and CDK9 are the essential kinases in RNAP II-mediated transcription, whose functions are highly conserved throughout the kingdom of eukaryotes (Figure 1). The CDK7 kinase module, as part of the TFIIH complex, orchestrates the initiation of transcription and phosphorylates Ser5 and Ser7 in the CTD heptad repeats [11]. CDK7 also activates the kinase activity of P-TEFb in metazoans, promoting productive elongation as a checkpoint by phosphorylating both Ser2 of CTD and SPT5 subunit of DSIF [12,13] (Figure 1). Analysis of the AS kinase allele by precision nuclear run-on sequencing captures the effect of CDK9 inhibition — a dramatic reduction in elongation speed [14]. P-TEFb depends on the activation by CDK7 to overcome pausing barriers to elongation [15]. Once activated, CDK9 provides feedback to CDK7-mediated initiation by increasing the initiation frequency while reducing the time the RNAP II stuck at the pausing site [16]. In addition to its role in pausing release and elongation, CDK9 seems to function as a communication mediator between RNAP II and chromatin through its association with histone modification enzymes such as mono-ubiquitination enzymes H2Bub1 [17] and BRD4 [18].

The AS kinase strategies are also successful in identifying the functions of CTD kinases that were not well-defined previously. Although early studies in S.cerevisiae suggest overlapping functions of CDK8 and CDK7 [19], CDK8 exhibits unique activity in promoting the genes involved with glycolysis in human cells in AS alleles [20]. CDK12/13, which phosphorylate Ser2 of the CTD as late transcriptional events, reduces the overall phosphorylation in RNAP II, leading to decreases in elongation rates and processivity [21] (Figure 1). The loss of one of the pair seems to have no global inhibitory effects on transcription [22]. Instead, it triggers the transcriptional pathway associated with DNA damage response in human cells, with CDK12 inhibition showing a much more significant effect [23]. The identification of HRR25 as the Thr4 kinase in S. cerevisiae reveals that its inhibition results in defective transcription termination of a subset of noncoding small nucleolar RNA genes [24]. A pTyr1-specific antibody’s development and commercial availability bring attention back again to Tyr1 phosphorylation of the CTD heptad repeats [25] (Figure 1). Tyr1 phosphorylation alters the kinase specificity of P-TEFb from Ser5 to promote Ser2 phosphorylation [26]. However, unlike the other CDKs, the identification of AS alleles for Tyrosine kinase ABL proves to be challenging. The only reported ATP-analog ABL mutant is BCR-ABL (T315A) by the small molecule 4-amino-1-tert-butyl-3-(1-naphthyl)pyrazolo[3,4-D]pyrimidine [27]. Yet, the mutant selectivity is not biochemically characterized [27]. Overall, chemical genetic methods like the analog-sensitive technology reveal the biological functions of kinases phosphorylating RNAP II (Figure 1).

Targeted covalent inhibition towards RNAP II kinases

Kinase activities regulating RNAP II have been implicated in various disease states [28–30]. For example, several natural product inhibitors towards CDK7 or CDK9 have strong cytotoxic effects on various cancer cell lines because specific oncogene expression highly depends on hi-jacking transcription machinery to drive malignant growth [8,9]. Combined inhibition of CDK12/13 and Poly(ADP-ribose) polymerase (PARP) results in synthetic lethality to diseased cells due to their involvement in regulating the expression of genes essential to DNA repair pathways [31]. These considerations warrant the development of compounds both as potential probes to study kinases’ mechanisms in RNAP II-mediated transcription — and ultimately leads for therapeutic development. Early inhibitor design attempts were inspired by these natural product scaffolds that block RNAP II kinase activities [32]. However, due to the highly conserved active sites and sequence similarity among the 20 kinases within the CDK family, isozyme-selective inhibitors have been elusive (Figure 3a-b). Although many reported chemical compounds exhibit CDK7 or CDK9 inhibition, very few could proceed to clinical testing due to broad inhibition profiles.

Figure 3. Strategies for selective CDK targeted covalent inhibition and degradation.

(a) Sequence alignment of the active sites of CDK kinases. Highly conserved residues close to the ATP-binding site are highlighted in yellow. The identical residues are annotated with “..” and highly similar residues with “.”. (b) Structural locations of highly conserved residues close to the active site of CDK7. The binding pocket is shown with a gray surface around the CDK7 crystal structure (PDB code: 1UA2), while conserved residues are highlighted as yellow sticks. (c) Chemical structures of CDK covalent inhibitor THZ1 and THZ531. THZ531 shares the same ATP-analog moiety and cysteine-reactive moiety with THZ1 but with different linker orientations between the two moieties. (d) Superposition of the complex structures of THZ1 with CDK7 (PDB code: 6XD3) and THZ531 with CDK12 (PDB code: 5ACB). THZ1 is shown as a white stick and CDK7 as a light blue ribbon, whereas THZ531 a yellow stick and CDK12 as a light pink ribbon. ATP-analog moiety of covalent inhibitors is occupying the ATP-binding site of CDKs, while each covalent inhibitor is forming covalent bonds with cysteine at different positions. (e) Chemical structures of selective CDK7 covalent inhibitors, YKL-1–116, YKL-5–124, and SY-1365. (f) Chemical structure of THAL-SNS-032, a PROTAC compound designed for selective degradation of CDK9.

The recent development of targeted covalent inhibitors revitalizes interest in discovering RNAP II kinase inhibitors [33]. The targeted covalent inhibitor design achieves high selectivity by forming a covalent bond with non-conserved cysteine residues. Therefore, it poses an additional filter for inhibition by the location of the cysteine unique to the target. The nucleophiles in these targeted covalent inhibitors are usually moderate to weak reactors to avoid interaction with all cysteines. The compounds are designed to first bind to the targets non-covalently, followed by the slow formation of a covalent bond in situ. The covalent bond formation can be reversible or non-reversible. The reversible covalent inhibitors provide an additional advantage by allowing the release of inhibitory compounds in non-productive complex formations [33]. Since the targeted cysteines are intentionally selected based on a lack of conservation, only the target containing the cysteine of interest engaged by the inhibitor achieves long-term inhibition. These covalent inhibitors show potential for long-lasting pharmacological effects due to their extended residence time and high selectivity compared to conventional reversible inhibitors.

The design of targeted covalent inhibitor for CDKs was first implemented in the inhibitor discovery for CDK7, with THZ1 forming a covalent bond with a remote cysteine outside the conserved kinase domain [34] (Figure 3c-d). Recently, the complex structure of THZ1 with human CDK7 kinase heterotrimer (containing CDK7, cyclin H, and MAT1) was determined by electron microscopy at the resolution of 3.30 Å, providing a clear visualization of the covalent bond formation between THZ1 and Cys312 of CDK7 (Figure 3c-d) [35]. The prolonged effect resulting from covalent inhibition reveals that some cancer cell lines are susceptible to the compounds with vulnerability conferred by some super-enhancers [34]. However, the effort to push THZ1 towards clinical usage is hampered by its remaining cross-inhibition towards CDK12 and CDK13 and poor bioactivity.

New chemical scaffolds were explored to distinguish CDK7 from CDK12/13. For example, YKL-5–124 exhibits high potency in k_inact/K_I towards CDK7 but no detectable activity for CDK12/13 [36] (Figure 3e). This selective inhibition on CDK7 causes strong cell-cycle arrest at G1/S, consistent with the function of CDK7 in activating other cyclin-dependent kinases. Another selective covalent inhibitor of CDK7 with a similar scaffold, YKL-1–116 (Figure 3e), induces apoptosis in p53 activated cells, rationalizing the combination usage of CDK7 inhibitors with p53 pathway chemical activators to activate the p53-controlled transcriptional gene to promote cell death [37]. Using a selective covalent inhibitor for CDK7, SY-351, additional substrates of CDK7 were identified, implicating its role as a “master regulator” in activating other CTD kinases (CDK9, CDK12, and CDK13), as well as spliceosome components [38]. Recently, a selective covalent CDK7 inhibitor, SY-1365, proceeded into clinical trials for ovarian and breast cancers (NCT03134638) [39] (Figure 3e). The cancer cells with low BCL-XL expression seem to be sensitized with SY-1365. The compound shows the most optimal growth inhibition when combined with BCL2 pathway inhibitors [39].

Optimization of the THZ1 scaffolds has led to the design of covalent inhibitors specific to CDK12/13, accomplished by targeting cysteine residues present only in CDK12/13. For example, THZ531 targets a cysteine unique to CDK12/13 (Cys1039 in CDK12 and 1017 for CDK13) and achieves high selectivity (Figure 3c-d) [40]. This compound emphasizes CDK12/13 in late transcription events, mostly affecting elongation and hyperphosphorylation in RNAP II but showed no overlapping activity with P-TEFb during pausing release. CDK12/13 inhibition causes a substantial reduction in the expression of a subset of genes mediating DNA damage response involving homologous recombination repair pathway and essential super-enhancer–associated transcription factor genes [23,41]. The mechanism rationalizes the combination of CDK12 and PARP inhibitors, which have shown cytotoxic effects in triple-negative breast cancer [31].

The final fate of the CDKs inhibited by targeted covalent inhibitors is degradation. In recent years, proteolysis-targeting chimeras (PROTACs) and similar chemical approaches have led to the revival of small molecule degraders, which combine the advantages of temporal control by chemical compounds with the genetic effects of a knockout [42]. Rapid and efficient degradation of a target protein is achieved by “hooking” the target protein with a specific ligand and shuttling it to the ubiquitin degradation system using an E3 ligase ligand. The effort for designing PROTAC compounds targeting phosphorylation regulators of RNAP II is still in an early stage. Happily, the inherent issue of specificity in many CDK inhibitors seems to be alleviated by the PROTAC approach. The first PROTAC designed to target CDK9 utilizes the compound THAL-SNS-032, which consists of a pan-CDK inhibitor SNS-032 ligand linked to a thalidomide derivative binding the E3 ubiquitin ligase Cereblon (Figure 3f) [43]. The degradation of CDK9 was achieved rapidly with no detectable loss of other proteins, a stark difference from the parent compound SNS-032, which cross-inhibits multiple CDKs. The selectivity might arise from the prolonged pharmaco-dynamic effect of PROTAC compounds in cells. Their outstanding performance in selectivity justifies a continued effort to design and optimize PROTAC inhibitors for studying the phosphorylation states of RNAP II.

Compounds modulating the activity of RNAP II phosphatases

The identification of phosphatases involved in RNAP II regulation has been elusive due to the lack of powerful chemical biology approaches like analog-sensitive alleles to identify RNAP II kinases. Genetic screening identified FCP1 [44] and SSU72 [45] as the “cleanup crews” to remove Ser2 and Ser5 phosphorylation at the end of each transcription cycle (Figure 1). However, recent reports on the phosphatases’ biological functions that dephosphorylate RNAP II highlight their regulatory roles during individual eukaryotic transcription with the recent implication of PP1 and PP2A in transcription (Figure 1) [46]. Both PP1 and PP2A achieve their substrate specificity via various regulatory subunits, identifying transcription-specific regulatory subunit as the bottleneck for phosphatase studies. Advancement in gene editing and proteomics are beginning to reveal these long-awaited key molecules are.

The CDK9-PP1 kinase-phosphatase pair regulates the elongation and termination transition of RNAP II by controlling the phosphorylation state of the SPT5 [13]. When associated with its regulatory domain, PNUTS, PP1 controls the elongation speed in preparation for polyadenylation and termination by decelerating the moving transcription machinery through dephosphorylation of SPT5 [47–49]. In turn, the CDK9 subunit of P-TEFb phosphorylates the PP1 isoform (DIS2 in S. pombe) to inhibit its phosphatase activity [13]. PP1-PNUTS has also been implicated in other steps of transcription —namely initiation [50], splicing [51], and termination [52].

The role of PP2A in general transcription has been suspected for decades. Still, the elusive regulatory subunit that recruits PP2A to RNAP II-mediated transcription remained a missing piece of the puzzle until recently. The integrator complex has been identified as the regulatory subunit for the transcription-specific PP2A holoenzyme using CRISPR screening. Specifically, INTS6 and INTS8, two subunits of the Integrator complex, are crucial for PP2A recruitment to the transcription machinery [53,54]. The timely structural determination of the integrator core subunits complexed with the PP2A catalytic and scaffold subunits provided the direct visualization of this unique PP2A holoenzyme (Figure 4a) [55]. Once recruited to RNAP II, PP2A seems to counter-act the kinase activity of the CDK9 subunit of P-TEFb, opposing its activity in pausing release and thus preventing productive elongation [53]. This molecular mechanism suggests that combinatorial therapy of kinase inhibitors with compounds regulating PP1 or PP2A activity could have profound clinical outcomes. Indeed, the activation of PP2A has been found to combat cancer cells with CDK9 inhibitors synergistically [53].

Figure 4. Chemical modulators of phosphatases involved in transcription pathway.

(a) The EM structure of PP2A-integrator complex (PDB code: 7CUN). PP2A A and C subunits are part of the large integrator complex. Unlike canonical PP2A holoenzymes, both INTS6 and INTS8 are closely associated with PP2A subunits for complex formation. (b) Chemical structures of small molecules that inhibit assembly of phosphatase complexes. (c) Scheme of biased stabilization of B56α-PP2A holoenzyme trimer with a small molecule activator DT-061. (d) DT-061 in complex with B56α-PP2A holoenzyme trimer (PDB code: 6NTS). DT-061 is stabilizing the interface between A subunit (pale green), B56alpha subunit (light blue), and C subunit (orange).

PP1 and PP2A are the primary phosphatases in cells, responsible for the majority of dephosphorylation events. Therefore, widespread activation or inhibition of their catalytic domains can lead to catastrophic consequences. Selective inhibition of metal-dependent phosphatases like PP1 and PP2A has proven hard to achieve by conventional methods. However, chemical modulators for specific inhibition exploit the necessity of PP1 and PP2A to form phosphate complexes with scaffold/regulatory subunits. Indeed, natural products like cyclosporin A and FK-506 prevent the assembly of phosphatase complexes by blocking the substrate recruitment sites to selectively inhibit calcineurin (also called PP2B and PP3) [56] (Figure 4b). Similarly, since PP1 utilizes different docking sites to recruit substrate, selective inhibition towards a specific substrate can be achieved by blocking the substrate-binding sites [57]. A selective inhibitor, Raphin1, is reported to selectively inhibit the R15B-PP1 complex without interfering with the highly similar R15A-PP1 complex when using a holoenzyme phosphatase complex instead of the catalytic domain itself during compound screening [58] (Figure 4b).

To suppress the P-TEFb kinase activity, activation of PP2A phosphatases is needed to inhibit the overactive CDK9 in tumors [59]. Although such compounds have been challenging to obtain by rational design, recent developments in the field of PP2A have provided the first example that specific inhibition or activation of phosphatase complexes that regulate RNAP II is feasible. Specifically, the chemical compound DT-061 identified from chemical screening can activate the B56α-PP2A holoenzyme trimer (Figure 4c) [60]. The structure of PP2A with DT-061 revealed that the compound functions as a “molecular glue” at the trimer interface to stabilize the holoenzyme formation (Figure 4d). Intriguingly, this compound only stabilizes the heterotrimer formation involving the B56α regulatory subunit but not any other PP2A holoenzymes. Consequently, DT-061 promotes the PP2A-mediated dephosphorylation of oncogenic c-Myc and silences the Myc-transformation pathway. Since PP2A inhibition ablated mitogen-activated protein kinase (MEK) inhibitor response, the combination of DT-061 with MEK inhibitors suppressed both p-AKT and MYC well as tumor regression in two KRAS-driven lung cancer mouse models [61]. Another example of PP2A inhibition is a well-tolerated inhibitor for PP2A and Serine/threonine-protein phosphatase 5 catalytic domains with promising pharmacological properties called LB-100 [62]. LB-100 potentiates potency against solid and hematologic cancers when used as part of a cocktail with standard cytotoxic drugs or radiation [62]. Currently, the specific PP2A holoenzyme targeted by LB-100 is not yet defined, as it is often the bottleneck for phosphatase studies.

The development of these new compounds shakes the long-held belief that chemical compounds targeting PP1 and PP2A are too toxic to allow their development as pharmacological treatments. Although still in its infancy, the new strategies for holoenzyme-specific compounds, such as PP2A activator (DT-061) and R15B-PP1 inhibitor (Raplin1), demonstrate that selective chemical modulation of an individual phosphatase complex is attainable.

Perspective

The successful implementation of recent chemical genetic methods to study enzymes that regulate RNAP II dramatically advances our understanding of eukaryotic transcription. Analog-sensitive kinase alleles and selective inhibitors have successfully illuminated the transcriptional function of enzymes that modify RNAP II. Although specific engagement of kinases and phosphatases has been challenging with traditional approaches, targeted covalent inhibition and early efforts in PROTAC inhibitors show significant improvement in selectivity. New and exciting concepts in inhibitor and activator design are gaining momentum and proving feasibility for their clinical application as solo or combination therapy. With the advent of these new technologies, our arsenal of chemical biology tools continues to expand, allowing us to understand the sophisticated mechanisms of transcription further and eventually combat diseases.

Abbreviation

RNAP II

RNA polymerase II

CTD

C-terminal domain the largest subunit of RNA polymerase II subunit RPB1

AS

Analog-sensitive

PARP

Poly(ADP-ribose) polymerase

PROTAC

proteolysis-targeting chimera

MEK

mitogen-activated protein kinase kinase