2. Inhibition of DNA Methylation
DNA methyltransferases (DNMTs) are critical mediators of epigenetic gene regulation, responsible for genome-wide de novo and maintenance methylation. Aberrations in methylation have been widely implicated in cancer development, progression, and response to treatment
[1][2], and consequently, DNMT inhibitors (DNMTi) including decitabine and 5-azacytidine have been developed and are now FDA-approved for the treatment of myelodysplastic syndrome
[3][4]. These agents are cytosine analogs that become incorporated into replicating DNA, where they are targeted for methylation by DNMTs. Due to their altered structure, they cannot be released by DNMT by β-elimination, leading to the covalent entrapment of DNMT into the DNA
[5][6].
There is evidence of a biological interplay between DNMT1, the enzyme responsible for maintenance methylation, and PARP1, that provides a rationale for combination DNMTi-PARPi therapy. DNMT1 and PARP1 are members of a multiprotein complex that localizes to sites of oxidative DNA damage
[7], where the presence of PARylated PARP1 inhibits methylation activity by DNMT1
[8], possibly to maintain an open chromatin structure to permit repair. It has been demonstrated that combining low-dose DNMTi treatment with the potent PARP-trapping PARPi talazoparib enhances PARP1-DNA binding, synergistically enhancing cytotoxicity across a number of
BRCA-wildtype cancer types with minimal toxicity in in vivo models
[9][10][11] or human subjects
[12][13]. Similar synergism has also been observed when talazoparib is combined with the second-generation DNMTi guadecitabine
[14].
In addition to a direct reduction in the free enzyme pool, DNMT entrapment also induces ubiquitin-E3 ligase-mediated proteasomal degradation of free DNMT1
[15][16]. Accordingly, low doses of DNMTi are sufficient to alter methylation patterns across the genome, leading to widespread alterations in multiple molecular pathways including the DNA damage response (DDR) and apoptosis
[17]. Among the myriad of pathways contributing to the Hanahan and Weinberg ‘hallmarks of cancer’
[18] that are altered by DNMTi, a recent examination of the DNA repair reactome in non-small cell lung
[10], breast, and ovarian cancers
[11] demonstrated a significant reduction in DSB repair, particularly involving the FA pathway. Of note was the downregulation of FANCD2, which is mono-ubiquitinated by other FA pathway members in response to DNA damage, leading to colocalization with BRCA1 and BRCA2 during homologous recombination repair of DSBs, and resulting in it being ascribed a role as a BRCAness gene
[19]. Furthermore, FANCD2 monoubiquitination is required for interactions with FANCD2/FANCI-associated nuclease 1 (FAN1), which mediates the canonical FA roles of interstrand crosslink repair and the resolution of stalled replication forks, potentially including those induced by trapped PARP1 and/or DNMT1
[20]. In keeping with the loss of these repair roles, DNMTi-induced FANCD2 downregulation was associated with a BRCAness phenotype, including increased replication fork stalling, DSB accumulation as measured by γH2AX foci accumulation, and a reduction in RAD51-mediated DSB repair capacity. Accordingly, in several human cancer cell lines and murine xenograft models, combining a low dose DNMTi with the PARPi talazoparib produces a significant and synergistic increase in tumor cell cytotoxicity
[10][11]. These results have led to a dose-finding Phase 1 trial in untreated or relapsed/refractory AML using DNMTi decitabine and PARPi talazoparib
[21], and a Phase 1 trial in BRCA-proficient breast cancer treated using oral decitabine and talazoparib (
Table 1).
Table 1. Clinical trials evaluating epigenetic therapy in combination with PARPi.
DNMTi-induced reversal of cancer-associated methylation abnormalities can reactivate abnormally methylated tumor suppressor gene promoters. One emerging example is Schlafen 11 (SLFN11), which irreversibly inhibits replication in cells undergoing replication stress such as DNA-damaging chemotherapy
[23]. High levels of SLFN11 destabilizes the interaction between single-stranded DNA and replication protein A (RPA) at the sites of DNA damage, inhibiting downstream DSB repair and producing cell cycle checkpoint activation
[24]. The suppression of SLFN11 expression is observed in ~50% of cancer cell lines and is correlated with a resistance to DNA-damaging agents including PARP inhibitors
[25]. SLFN11 suppression appears to be primarily epigenetic in origin, linked to promoter methylation, histone deacetylation, and PRC-mediated histone methylation
[23]. Decitabine can reverse
SLFN11 promoter methylation, leading to the re-expression and re-sensitization to DNA-damaging agents, and similar results have also been observed following EZH2
[26] or HDAC inhibitors
[27] (see below), providing a further rationale for future studies combining epigenetic agents with PARP inhibitors.
3. Maintenance of Chromatin Repressive States
Histone deacetylases (HDACs) remove acetyl groups from ϵ-N-acetyl lysine residues on histones, leading to chromatin condensation and transcriptional repression. Abnormal acetylation resulting from HDAC overexpression can downregulate the expression of various tumor suppressive mechanisms, including cyclin-dependent kinases, differentiation factors, and proapoptotic signals, leading to the uncontrolled proliferation, de-differentiation, and survival that is characteristic of oncogenesis and metastasis
[28]. The eighteen identified members of the HDAC family have been classified into four groups (class I, IIa/b, V, and III/sirtuins) based on homology to yeast HDACs. Compounds with anti-HDAC activity are numerous, and can be divided into pan-HDAC inhibitors (HDACi), which exhibit activity against all non-sirtuin HDACs, or selective HDACi, which target specific HDACs
[29]. FDA approval has been granted for the treatment of various hematological malignancies for three pan-HDACi (vorinostat, belinostat, and panobinostat) and one HDAC1/2-selective HDACi (romidepsin)
[30].
Acetylation exerts effects over the chromatin structure that impacts the recognition and repair of DNA damage
[31], and accordingly, HDACi have been reported to alter DSB repair capacity
[32]. While deacetylation activity of HDAC1/2 has been shown to both directly and indirectly decrease c-NHEJ activity
[33][34], the role of HDACs in HR, and hence the therapeutic potential of HDACi for the induction of BRCAness, is less clearly defined. Of note, HR proteins including BRCA1, BRCA2, and RAD51 have been reported to be suppressed by HDACi in a variety of cancers
[35][36], sensitizing to PARPi
[37][38][39][40][41][42][43]. Based on these results, phase I trials combining olaparib with vorinostat are underway in advanced lymphoma and breast cancer (
Table 1).
Notably, the inhibition of the deacetylation activity following HDACi exposure leads to PARP1 hyperacetylation and enrichment in chromatin that resembles PARPi-induced PARP trapping. When combined with PARPi, HDACi treatment further increases PARP trapping, synergistically sensitizing to the PARP-trapping PARPi talazoparib
[44]. Synergism has also been observed when HDACi are combined with DNMTi, specifically by enhancing the re-expression of genes silenced by abnormal promoter methylation
[12][45]. Valdez et al. have reported synergistic inhibition of AML and lymphoma cell proliferation by the triple combination of PARPi niraparib, DNMTi decitabine, and HDACi romidepsin or pabinostat, associated with the activation of ATM-mediated DDR, increased ROS production, and the induction of apoptosis
[46]. These effects were hypothesized to be the sequelae of DSB accumulation induced by triple combination through three mechanisms: significantly enhanced PARP trapping; acetylation and inhibition of DNA repair proteins including Ku70/80 and PARP1; and the downregulation of the nucleosome-remodeling deacetylase complex, a transcriptional repressor with chromatin remodeling activity that is functionally linked to efficient DNA repair
[47]. While further preclinical study is required, these results provide a rationale for the future development of combination therapy using PARPi, HDACi, and DNMTi.
4. Polycomb Repressive Complex 2
An enhancer of the zeste homolog 2 (EZH2) is the histone methyltransferase subunit of polycomb repressive complex 2 (PRC2), which methylates histone H3 on lysine 27 (H3K27me3) to mark chromatin as transcriptionally silent. PRC2 plays an important oncogenic role through the modulation of the DDR
[48]. EZH2 overexpression, which is common in many cancers
[49], induces the downregulation of RAD51 homolog expression
[50], cytoplasmic BRCA1 retention
[51], and impaired HR that is associated with increased genomic instability. PRC2 appears to play a role in the DSB repair pathway choice, being recruited to DSBs in a Ku-dependent mechanism to promote efficient NHEJ
[52], and accordingly, EZH2 depletion favors HR, impairs NHEJ, and sensitizes to irradiation damage
[53]. Recent evidence indicates that this DSB repair pathway switch can be therapeutically targeted by PARPi in a subset of HR-proficient tumors overexpressing the oncogene coactivator-associated arginine methyltransferase 1 (CARM1). The overexpression of CARM1 promotes the EZH2 silencing of
MAD2L2, a member of the shieldin complex that limits DNA end resection to favor NHEJ. Accordingly, in CARM1-high cells, EZH2 inhibition upregulates
MAD2L2, increasing error-prone NHEJ activity and associated chromosomal abnormalities, and producing mitotic catastrophe in combination with PARPi treatment
[54]. An ongoing phase II clinical trial, evaluating multiple targeted therapies in a biomarker-guided precision therapy approach, includes the novel agents SHR2554 (EZH2 inhibitor) and SHR3162 (PARPi) (
Table 1).
5. BET Proteins
The conserved bromodomain and extraterminal (BET) family of proteins are characterized by two tandem bromodomains that bind to activated lysine residues on target proteins
[55][56]. BET members preferentially interact with hyperacetylated histones, leading to an accumulation at the transcriptionally active regulatory elements
[57]. BET family member BRD4 acts as a transcriptional cofactor, influencing the expression of a wide range of genes involved in cell fate determination. In cancer, BRD4 has been implicated in the activation of a multitude of oncogenes, co-occupying a set of promoter super-enhancers associated with prominent oncogenic drivers such as c-MYC
[58][59]. High affinity small molecules targeting the BET bromodomains demonstrate preclinical efficacy in a wide range of cancers associated with transcriptional suppression of key proto-oncogenes including
c-MYC,
N-MYC,
FOSL1, and
BLC2 (reviewed in
[57].
The first report of potential synergism between BET inhibition and PARPi was based on a drug combination screen testing PARPi olaparib in
BRCA-wildtype triple negative breast (TNBC), ovarian, and prostate cancer in combination with 20 well-characterized epigenetic modulators across seven classes, demonstrating synergism for all tested BET inhibitors (BETi)
[60]. BETi treatment significantly enhanced PARPi-induced DSB accumulation independent of PARP-trapping, associated with the repression of
BRCA1 and
RAD51 transcription, which is suggestive of induced BRCAness. Notably, BETi treatment could disrupt the enrichment of BRD2/3/4 at the
BRCA1 and
RAD51 promoter regions, in addition to the putative super-enhancer region downstream of the
BRCA1 promoter that exerted a stronger transcriptional enhancing activity than the promoter region alone
[60]. Validation of these results, with similar BETi-induced repression of
BRCA1 and
RAD51, induction of an HR defect, and sensitivity to PARPi, has since been reported in TNBC
[61].
A subsequent study used publicly available transcriptional profiling data to demonstrate that BRD4 inhibition modulates a previously validated HR defect gene signature
[62], though finding minimal impact on
BRCA1 or
RAD51 expression in cell lines of multiple cancer types. Instead, a consistent and marked downregulation of CtIP was observed, in keeping with ChIP-seq analysis that indicates that both the
CtIP promoter and an associated enhancer region are directly targeted by BRD4. BETi induced PARPi sensitivity in 40 of 55 cancer cell lines and five in vivo models spanning breast, ovarian, and pancreatic cancer, as well as resensitizing PARPi-resistant cells
[63]. Despite the mechanistic discrepancies between the studies, these results indicate the therapeutic potential of the PARPi-BETi combination that warrants further investigation.