Targeting the transcription cycle and RNA processing in cancer treatment

Abstract

The transcriptional program and RNA splicing machinery are highly and frequently dysregulated in human cancers due to genomic and epigenomic alterations during tumorigenesis. This leads to cancer-specific dependencies on components of the transcriptional program and RNA splicing machinery, providing alternative and targetable “Achilles’ heels” for cancer treatment in the clinic. To target these vulnerabilities in cancer cells, potent and specific transcriptional CDK inhibitors and chemical compounds that impair splicing have been developed and evaluated in preclinical cancer models. Several novel combination approaches with immune or targeted therapies have also been proposed for cancer treatment. More recently, inhibitors targeting transcriptional CDKs, splicing, or PRMT5 have shown promising therapeutic potential in preclinical studies, and many of them have rapidly advanced into early clinical trials for treatment of human cancer.

Introduction

Cancer cells exhibit accumulated genomic alterations, such as mutations, copy number changes and gene fusions, leading ultimately to uncontrolled cell growth and aberrant gene expression [1,2]. Components of the transcription program, including transcription factors, cofactors, chromatin regulators and cis-regulatory elements, are frequently altered to support the elevated level of gene expression required by the higher demands of cancer cell metabolism, proliferation, and viability compared to normal cells [3–7]. Therefore, the transcriptional machinery is always dysregulated during tumorigenesis. Increasing evidence indicates that cancer cells are highly dependent on, or “addicted to,” certain components of the transcriptional program [3–7]. Given that cancer cells are remarkably susceptible to abnormal transcription (termed “transcription addiction”) [3], targeting the transcription cycle and RNA processing have emerged as new strategies in cancer therapy [3–7]. Many tissue-specific or master transcription factors, which play central roles in regulating gene expression, are frequently and recurrently altered in human cancers; however, with the exception of nuclear receptors, targeting of transcription factors has remained a clinical challenge [3–7]. For example, MYC amplification and hyperactivation have been observed in more than 28% of tumor specimens across the TCGA sample cohort; however, most strategies directly targeting MYC have failed in preclinical development over the last two decades. Alternatively, targeting druggable proteins involved in the transcription cycle or RNA processing has been reported as a potential therapeutic strategy in cancer treatment [3–7].

Targeting transcriptional cyclin-dependent kinases (CDKs)

The family of cyclin-dependent kinases (CDKs) is composed of serine-threonine kinases that play critical roles in controlling the cell cycle and regulating transcription [8,9]. CDKs are classified as cell cycle-related CDKs or transcriptional CDKs that involved in regulation of basal transcription. The activity of a CDK depends on its regulatory partner cyclin. Upon the binding of cyclin, a cyclin-CDK complex transmits active signaling to multiple downstream proteins through phosphorylation [8,9]. Over the past decades, the biology of CDKs and their roles in diseases, including cancer, have been extensively studied [8,9]. Cell cycle-related CDKs directly regulate the progress of cell cycle. Transcriptional CDKs regulate phosphorylation of the C-terminal domain (CTD) of RNA polymerase II (Pol II), therefore controlling gene transcriptions. Given that cancer cells are always undergoing uncontrolled proliferation and dysregulated transcription, it is unsurprising that CDKs have been considered attractive targets for cancer drug development [3–5,10,11]. Development of selective CDK inhibitors to target certain CDKs is key for the application of CDKis in the clinic. During recent years, an increasing number of selective inhibitors targeting cell cycle-regulated CDKs, for example CDK4/6, have been approved for the clinical care of patients with ER⁺ and HER2⁻ advanced metastatic breast cancer [10,11]. More recently, selective inhibitors targeting transcriptional CDKs have also emerged as a new class of anti-cancer agents [3–5]. For example, CDK7, CDK8/19, CDK9, CDK11 and CDK12/13 inhibitors have shown promising therapeutic potential in preclinical cancer models as well as in early-stage clinical trials [12–17] (Table 1). However, given that more than ten transcriptional CDKs have been identified in the human genome [8,9], identification and prioritization of appropriate candidates for cancer treatment remain challenging. Recently, by analyzing recurrent copy number alterations, mutations, and fusions of 21 CDK genes across more than 10,000 human tumor specimens, we found that CDKs are pervasively subject to recurrent genomic alterations across cancers [18]. Somatic copy number alteration is dominant among the genomic changes identified in CDK genes, and copy number status is positively correlated with mRNA expression levels. Notably, the copy number alterations in cell cycle-related CDKs are distinct from those in transcriptional CDKs. Cell cycle-related CDKs tend to be recurrently amplified, while transcriptional CDKs show recurrent copy number loss across cancers (Figure 1). Although transcription addiction in cancer is well recognized, the role of copy number losses in transcriptional CDKs in tumors is still largely unknown. More intriguingly, homogenous deletion is very rare whereas hemizygous loss is the dominant form of copy number deletions in transcriptional CDKs, suggesting their important roles in maintaining cancer cell growth and survival [18]. Hemizygous loss of an essential gene in cancer could create vulnerabilities which can be targeted for cancer treatment [19–24]. Some of these CYCLOPS genes (copy number alterations yielding cancer liabilities owing to partial loss [20]) have been identified and evaluated in preclinical models [20–23]. Excitingly, the very first successful example of this concept is the approval of lenalidomide (Revlimid), which targets CK1α (located in 5q32) to treat myelodysplastic syndrome with loss of chromosome 5q [24]. Given that transcriptional CDKs are essential for cell growth and survival, targeting transcriptional CDKs would be a promising therapeutic strategy for tumor cells with hemizygous loss of transcriptional CDKs [18]. For example, recurrent and focal CDK7 losses were found in multiple cancer types, and 27.3% (2,991/10,950) of TCGA tumors showed hemizygous CDK7 copy number loss [18]. The high frequency of CDK7 hemizygous loss in adult solid tumors suggests that CDK7 inhibition may trigger treatment-induced “haploinsufficiency”. Thus, low-dose CDK7 inhibitor treatment may be sufficient to kill tumor cells which are subject to transcription addiction and show hemizygous loss of CDK7. In this setting, the treatment may thus be tolerable to normal cells harboring two copies of the CDK7 gene. Taken together, targeting transcriptional CDKs with recurrent genomic losses could be a novel cancer therapeutic strategy, and such genomic alterations may also serve as biomarkers to stratify patient populations that will benefit from the treatment [18]. Additionally, inhibition of transcriptional CDKs may preferentially suppress the expression of the genes regulated by super enhancers [3–5], such as MYC, MCL1 and RUNX1 in cancer cells, providing additional therapeutic windows in patients with tumors that depend on hyperactivation of these oncogenes.

Table 1.

List of small molecule compounds targeting transcriptional cycle or splicing process.

Compound	Target	CID	PMID	Clinical Identifier
BS-181	CDK7	49867929	19638587
LDC4297	CDK7	78161839	25047832
THZ1	CDK7	73602827	25043025
THZ2	CDK7	78357763	26406377
CT7001	CDK7	91844733	29545334	NCT03363893
YKL-5–124	CDK7	121443990	30905681
SY-1365	CDK7	118426108	31064851	NCT03134638
LY3405105	CDK7			NCT03770494
SY-5609	CDK7			NCT04247126
LY2857785	CDK9	78357764	24688048
BAY-1251152	CDK9	74767009		NCT02635672, NCT02745743
Atuveciclib	CDK9	71618220	28961375	NCT01938638, NCT02345382
MC180295	CDK9	137333456	30454645
NVP-2	CDK9	66937006	29251720
AZD4573	CDK9	124155204	31699827	NCT04630756
CCT251545	CDK8/19	77050682	26502155
SEL-120	CDK8/19	73776232		NCT04021368
AS2863619	CDK8/19	139600293	31653719
BI-1347	CDK8/19	132180482	32024684
OTS964*	CDK11	89675898	31511426
THZ531	CDK12/13	118025540	27571479
SR-4835	CDK12/13	139600338	31668947
FR901463	SF3B1	70678766	9031664, 9031665
FR901464	SF3B1	10553647	9031664, 9031665
FR901465	SF3B1	10673616	9031664, 9031665
Herboxidiene	SF3B1	6438496	12523818
Pladienolides B	SF3B1	16202130	15152802, 15152803, 15152804
Pladienolides D	SF3B1	52944386	15152802, 15152803, 15152804
spliceostatin A	SF3B1	10673568	17643111
E7107	SF3B1	16202132	17643112	NCT00459823, NCT00499499
Meayamycin	SF3B1	16111971	17279752
Sudemycins	SF3B1		21344922
H3B-8800	SF3B1	92135969	29457796	NCT02841540
TG003	CLK1, CLK2, CLK4	1893668	15010457
SRPIN340	SRPK1, SRPK2	2797577	16840555
cpd-1/2/3	SRPK1, SRPK2, CLK1, CLK2		25581376
Indisulam	RBM39	216468	10508428	NCT00060567, NCT00165594, NCT01692197, NCT00059735, NCT00080197, NCT00165854, NCT00165880, NCT00003976, NCT00003981, NCT00014625, NCT00165867,
C220	PRMT5		32669286
PRT811	PRMT5			NCT04089449
PRT543	PRMT5			NCT03886831
GSK3368715	PRMT1	90462880	31257072	NCT03666988
JNJ-64619178	PRMT5	126637809		NCT03573310
GSK3326595	PRMT5	90241742	29946150	NCT02783300, NCT03614728, NCT04676516
PF-06939999	PRMT5			NCT03854227

CID = PubChem Compound CID; https://pubchem.ncbi.nlm.nih.gov/

PMID = PubMed Reference PMID; https://pubmed.ncbi.nlm.nih.gov/

Clinical Identifier = ClinicalTrials.gov Identifier; https://www.clinicaltrials.gov/ct2/home

^*OTS964 was originally identified as a TOPK (T–lymphokine-activated killer cell–originated protein kinase) inhibitor. Recent study indicated that OTS964 is potent CDK11 inhibitor.

Figure 1. Genomic alterations of CDKs/cyclins are dominantly driven by somatic copy number aberrations in cancer.

Somatic copy number aberrations of the genes encoding CDKs/cyclins were characterized across the TCGA sample cohort (33 cancer types and more than 10,000 tumor specimens) by SNP arrays. Summary of the overall G-scores, which consider both the amplitudes of the aberrations and the frequencies of their occurrence across samples, of the function-defined CDKs/cyclins in cell cycle (left panel) and transcription (right panel) pathways. The size of the bubble: overall G-score; red: gain; blue: loss.

Targeting splicing dysregulations

To maintain uncontrolled cell growth and aberrant gene expression, the RNA processing machinery is also frequently altered during tumorigenesis [6,7]. RNA splicing is one of the essential steps for gene expression, and is necessary for producing diverse transcripts required by multiple cellular functions including proliferation, survival, and differentiation [6,7]. Dysregulation and mutations of core components of the spliceosome complex and splicing regulatory proteins, as well as mutations in regulatory sequences of the pre-mRNA of cancer-associated genes, have been identified in human cancers [1,2]. Consistent with this, transcriptome-wide aberrant splicing in tumor cells has been confirmed in large-scale pan-cancer analyses. For example, significant amounts of mis-spliced transcripts were found in tumor cells compared to their normal counterparts, generating cancer-specific gene isoforms that support cancer growth and causing the cells to develop splicing dependency [25–27]. Finally, in addition to splicing-disrupting mutations, splice-site-creating mutations that generate novel splicing sites have also been identified across cancers [28]. Taken together, dysregulated splicing emerges as a targetable vulnerability for cancers, especially for cancers bearing spliceosome mutations [29–33]. In this regard, small molecules targeting abnormal splicing have been identified during the past decades (Table 1). For example, inhibitors targeting the core spliceosome, such as E7107 (analogue of pladienolide B), have been developed and tested in clinical trials [34]. However, unexpected toxicity was observed in early-stage clinical studies of splicing inhibitors, leading to the early termination of these trials. Excitingly, novel derivatives of pladienolide B compounds (e.g., H3B-8800) with lower cytotoxicity have been developed and are under clinical evaluation [32]. In addition to directly targeting the core spliceosome, recent studies reported that inhibition of PRMT5 has anti-tumor potency in vitro and in pre-clinical models in multiple cancer types [35–40]. Several PRMT5 inhibitors are currently under evaluation in clinical trials for both hematologic cancers and solid tumors (Table 1). PRMT5 is an arginine methyltransferase which methylates Sm proteins of U2 snRNP, which are crucial in maintaining splicing fidelity in cells. Knocking down PRMT5 leads to abnormal splicing, such as exon skipping and intron retention on genes related to cell survival. Notably, tumors with splicing defects (e.g., splicing factor mutations) are more sensitive to PRMT5 inhibition [35,41]. Moreover, studies showed that MYC-driven cancers are also preferentially vulnerable to spliceosome inhibition, including treatment with PRMT5 inhibitors, because many splicing regulators and snRNA assembly factors are direct transcription targets of MYC [35,42,43]. Upregulation of these factors contributes to oncogenic transformation. For example, MYC-upregulated SRSF1 promotes mammary epithelial cell transformation together with MYC [43,44]. hnRNP1 and hnRNP2 induced by MYC lead to isoform switching of target gene; they regulate alternative splicing of pyruvate kinase and increase expression of its cancer-associated PKM2 isoform [42]. In addition, MYC also directly upregulates PRMT5 transcription; the increased expression of PRMT5 may be in response to an increased demand for accurate gene splicing in cells [35,45]. Depletion of PRMT5 results in mis-splicing of genes required for the growth and survival of MYC-driven lymphomas [35]. Furthermore, by leveraging the enhanced sensitivity of MYC-driven tumors to splicing perturbations, researchers identified synthetic lethal genes for MYC such as BUD31, which is a component of the core spliceosome [46]. Depletion of BUD31 expression in MYC-driven breast cancer cells resulted in reduced cell viability and increased apoptosis. Finally, RNA-based therapy has also been explored as an approach to treat dysregulated splicing in cancer. Antisense oligonucleotides (ASOs) modulating splicing (termed splice-switching oligonucleotides, SSOs) have been effective in correcting abnormal splicing events in vivo, and were approved for the treatment of Duchenne muscular dystrophy and spinal muscular atrophy in the clinic [47,48]. Sequence-specific SSO binds to corresponding pre-mRNA at spice sites or splicing regulatory motifs, preventing the recruitment of splicing machinery. The binding of SSOs to splicing enhancers or silencers promotes exon exclusion or inclusion, respectively. Despite the success of SSOs in treating non-oncology diseases, their application in cancer treatment remains challenging. The global dysregulation of splicing in cancer makes it uncertain whether modulating only a single splicing event in cancer would have an anti-tumor effect. However, recent results have demonstrated that correcting mis-spliced BRD9 by SSO in uveal melanoma inhibited cancer cell growth in vivo and in vitro, showing the potential of SSOs in cancer therapy [49]. Taken together, targeting dysregulated splicing in cancer has been proposed as a promising approach; however, more efforts are still needed for further understanding of the underlying mechanism in the context of tumorigenesis, as well as development of more specific and tolerable splicing inhibitors.

Combination with immune and targeted therapies

Although single-agent treatment of drugs targeting dysregulated transcription and RNA processing have shown promising results in cancers, recent studies revealed the possibility of advances from a combination of CDK inhibitors or splicing inhibitors with immune and targeted therapies. For example, the combination of CDK7 inhibitor (CDK7i) with p53 activators has been reported in cancer cell in vitro [50], and the combination of CDK7i or CDK12 inhibitor (CDK12i) with targeted drugs, such as PARP inhibitors (PARPis) [18,51,52], have been evaluated in animal models and early stage of clinical trials. The PARP (Poly(ADP-ribose) polymerase) family is a group of enzymes that post-translationally modify its target proteins by conjugating polymeric chains of ADP-ribose. PARPs have been found to be involved in multiple cellular processes including DNA repair. PARPs, especially PARP1, are important in recruiting homologous recombination (HR) proteins such as BRCA1 to DNA breaks. Cells with PARP inhibition show hyper-recombinogenic phenotypes with increasing HR frequency. Multiple PARPis have been successfully applied in the treatment of HR-deficient tumors; however, their effects were limited in cancers with primary or acquired HR-proficiency. Therefore, strategies suppressing expression of HR genes (chemically induced BRCAness) have been combined with PARPis or other DNA damage agents as a novel strategy to treat HR-proficient tumors [53]. Both CDK7is and CDK12is preferentially repress expression of genes in the DNA damage repair pathways, thereby impairing the activity of HR [18,51,54]. Loss of CDK7 has been reported to be correlated with elevated sensitivities to PARPis in ovarian and breast cancer cells [18]. Decreased CDK7 activity by CDK7i treatment represses the expression of BRCA1, BRCA2 and RAD51, subsequently sensitizing HR-proficient tumors to PARP inhibition. Low-dose CDK7i treatment is sufficient to act synergistically with PARPis in preclinical models of ovarian and breast cancers. Additionally, CDK7i also represses the expression of mutant BRCA1 with restored function after secondary mutations, or blocks the expression of HR genes in the context of other resistance mechanisms. CDK12 has been reported to regulate HR gene expression by suppressing intronic polyadenylation, subsequently leading to HR-deficiency [54]. By such depletion of HR gene expression, inhibition of CDK12 sensitizes or re-sensitizes cancer cells with primary or acquired PARPi resistance to PARPi therapy; this has been demonstrated in BRCA-mutant and wildtype triple negative breast cancers [51]. Thus, combination of transcriptional CDK inhibitors and PARPis may not only benefit cancer patients with primary HR-proficient tumors, but may also overcome PARPi resistance in patients with acquired HR-proficiency.

In human cancer, altered splicing has been connected with the formation of neoantigens, which are proteins that are absent in normal tissues but specifically present on the surface of cancer cells. Multiple studies have demonstrated that peptides derived from cancer-specific splicing events serve as neoepitopes by binding to histocompatibility complex class I (MHC I) molecules [26,28,55]. These studies also revealed that peptides derived from cancer-specific neojunctions significantly increase the target space for immunotherapy [26], and splice-site-creating mutations in cancer generate more than two times the number of neoepitopes per mutation compared with non-synonymous mutations. Moreover, a higher neoantigen burden is associated with elevated immune response, and splice-site-creating mutations increase the overall immunogenicity of cancers, suggesting the potential for application of immunotherapy in these cancers. Recently, PRMT5i has been reported to enhance the tumor immune response [56]. Inhibition of PRMT5 increased expression levels of interferon and chemokines, as well as abundance of MHC I molecules through regulation of target genes, therefore modulating antigen presentation of melanoma cells. Combination of PRMT5i with immune checkpoint therapy successfully inhibited tumor growth in a melanoma mouse model, showing enhanced efficacy compared to monotherapy alone. This evidence strongly suggests the potential of combining PRMT5is and immunotherapy for further clinical development.

Conclusion

There is an urgent unmet medical need to develop novel and alternative therapeutic strategies for patients with cancer. Given that the transcriptional program and RNA splicing machinery are highly dysregulated in human cancers, resulting in cancer dependencies on specific components of the transcription and splicing machinery, targeting transcriptional CDKs and splicing are emerging as new strategies for cancer treatment. Potent and selective transcriptional CDK inhibitors and chemical compounds that impair splicing have been developed and evaluated in preclinical models of cancer. Many of them have been advanced into early stage of clinical trials as a single therapeutic agent or in combination with immune and targeted therapeutic drugs.