Introduction
The human genome is constantly encountering a diverse array of endogenous and exogenous DNA damaging agents. All DNA damage can be categorized into two groups: (1) spontaneous DNA damage generated during normal metabolism and (2) induced DNA damage deriving from environmental factors. Spontaneous DNA damage includes dNTP misincorporation, base deamination, DNA depurination, base modifications, and DNA strand breaks derived from ROS (reactive oxygen species) [1]. Induced DNA damage results from multiple physical and chemical factors, such as ionizing radiation (IR), ultraviolet (UV), environmental toxins and chemotherapy reagents, etc. [1]. Both spontaneous and induced damage are each estimated to occur 105 times per day in a cell [2].
DNA damage can cause genome instability which can result in cell death or tumorigenesis. To maintain genome integrity, cells have developed multiple strategies to efficiently repair the different kinds of DNA damage. These mechanisms include mismatch repair (MMR), base excision repair (BER), nucleotide excision repair (NER), single-strand break repair (SSBR), double-strand break repair (DSBR), etc. [3]. Homologous recombination (HR) and nonhomologous end joining (NHEJ) are two major DSBR pathways. Compelling data from previous studies indicate that mutation of genes responding to DNA damage greatly promotes tumorigenesis. In addition, inhibiting proteins involved in the DNA damage response can efficiently lead to tumor cell death. Therefore, studies of the DNA damage response can elucidate the fundamental mechanisms triggering tumorigenesis and provide novel strategies for tumor therapy.
PARP1 (poly(ADP-ribose) polymerase 1), one of the key players in SSBR, plays important roles in multiple DNA damage response pathways. The therapeutic applications of PARP inhibitors in potentiating the killing action of IR have been well-documented and are attracting increasing interest in tumor therapy. Moreover, compelling data indicate that PARP inhibitors result in synthetic lethality in HR-deficient tumor cells. As an important novel class of anticancer drugs, there are now more than 40 clinical trials ongoing or in development with PARP inhibitors in the treatment of cancer [4]. However, the underlying mechanism of PARP inhibitors in tumor therapy remains controversial. Within this review, we briefly summarize the recent research progress on PARP1 in the DNA damage response and PARP inhibitors in tumor cytotoxicity. Hopefully the information in this review article summarizing the recent research advances will lead to a better understand of the mechanisms of PARP1 in genome stability maintenance and provide valuable clues in the selection of patient populations that will respond to PARP inhibitors.
Introduction of PARP1 protein and poly(ADP-ribosyl)ation
PARP1 is a member of PARPs (poly(ADP-ribose) polymerases) protein family. In human cells, there are hitherto 17 members reported in the family [5]. Among them, PARP1 and PARP2, PARP3, PARP4, TNKS (also known as tankyrase 1) and TNKS2 have poly(ADP-ribosyl)ation activity, which means that they can transfer ADP-ribose molecules one by one from nicotinamide adenine dinucleotide (NAD + ) to target proteins to form poly(ADP-ribosyl)ated (PARylated) proteins [4].
Poly(ADP-ribosyl)ation (PARylation) of proteins, like phosphorylation and ubiquitination, is a common form of post-translation modification (PTM). It can participate in multiple metabolic processes, such as recognition and repair of DNA damage, chromatin remodeling, transcription, programmed cell death and mitosis. The dendroid PAR chains with strong negative charges can modify the three-dimensional structures of proteins, thus affecting their functions and the interactions with other proteins and DNA.
PARP1 is the first reported PAR polymerase. It includes a nuclear localization signal (NLS) and is an abundant nuclear protein. PARP1 mainly contains 3 functional domains (Fig. 1): the N-terminal DNA binding domain (DBD), the central automodification domain (AMD) and the C-terminal catalytic domain (CD). The DBD includes 3 zinc finger motifs. The first two (Zn I and Zn II) participate in the recognition of DSB/SSB and mediate the binding of PARP1 to DNA. The newly identified third zinc finger motif (Zn III) mediates the regulation of the DBD on the catalytic activity and is not believed to be involved in DNA binding [6,7]. The AMD contains specific glutamate and lysine residues serving as acceptors for ADP-ribose moieties and also a BRCT domain that can interact with many DNA damage response proteins. The CD includes a PARP signature motif and a WGR motif and catalyzes the formation of PAR. The PARP signature motif forms the active site and binds NAD + . The function of the WGR motif is unknown [8-10].
PARP1 can be activated by several forms of DNA damage and subsequently catalyze the PARylation of multiple downstream proteins, including itself, histones, topoisomerase 1 (TOP1), DNA-dependent protein kinase (DNA-PK) and others, ensuing protein interactions in vivo [11]. In general, PARP1 accounts for 75%-90% of cellular PAR formation upon genotoxic stress [12]. In addition, the PAR chains on PARP1 can be hydrolyzed by poly(ADP-ribose) glycohydrolase (PARG) [4] and make PARP1 recycled. PARP1 can be degraded through two different pathways: by ubiquitination and hydrolysis via the proteasome or by the caspase pathway. Under some conditions, such as heat-shock, PARP1 would be degraded by the former pathway [13]. During cell apoptosis and sometimes under resting conditions, PARP1 is hydrolyzed by caspase enzymes into two distinct fragments: a DNA binding fragment (24 kDa) and an apoptosis-promoting fragment (89 kDa) [14-16].
The functions of PARP1 in DNA damage response
As a sensor of DNA damage, PARP1 binds to SSBs and DSBs and gets activated [1]. It then catalyzes the addition of PAR chains on proteins to recruit DNA damage response factors to chromatin at breaks. PARP-1 has been implicated in BER and SSBR. Recent studies indicate that PARP1 also functions in DSBR.
The role of PARP1 in SSBR
The role of PARP1 in SSBR is quite clear at present. It is believed that SSBs resulting from different factors such as IR, reactive oxygen species (ROS) and apurinic/apyrimidinic sites (AP sites) [17] can be recognized by PARP1 through its first two zinc finger motifs in the DBD. Once recruited to SSBs, PARP1 will be activated within a few minutes [18] and immediately transfer PARs to itself and other targeting proteins. The interaction between negative charges of PAR on histones and DNA relaxes the structure of chromosomes, then facilitates other proteins to associate and assemble a DNA repair complex consisting of X-ray repair cross-complementing 1 (XRCC1)/DNA Ligase III, polynucleotide kinase 3′-phosphatase (PNKP) and others at SSBs [18]. Subsequently, PARylated PARP1 dissociates from DNA due to repulsion of the negative charges of PAR with those of DNA and the PAR on PARP1 is degraded by PARG. The entire cycle described takes just a few minutes [19]. Therefore, the function of PARP1 in SSBR is only as a sensor to recognize DNA damage after which other repair proteins are recruited to finish SSBR (Fig. 2, left panel).
The role of PARP1 in HR
HR is essential for the accurate repair of DSBs by utilizing sister chromatids as template. It involves multiple steps and is strictly regulated by DNA damage checkpoints. In HR, DSBs can be recognized by the Mre11-Rad50-Nbs1 (MRN) complex, which utilizes its nuclease activity to process DSBs and thus promotes the activation of ataxia-telangiectasia-mutated (ATM). ATM can phosphorylate histone variant H2AX and produce γH2AX (phosphorylated-H2AX) which is believed to be the initial signal for subsequent accumulation of DNA repair proteins to DSBs. MDC1 (mediator of DNA damage checkpoint protein 1) binds ATM to further propagate γH2AX spreading [1].
The overall role of PARP1 in HR is complex. It has been reported that inhibition or deletion of PARP1 has been shown to increase spontaneous sister chromatid exchange (SCE) levels [21] and HR [22]. However, the rate of gene conversion induced by the rare-cutting restriction endonuclease I-SceI remains unchanged [23-25]. PARP1 was largely not found co-localized with Rad51 foci (a marker of HR) following treatment with hydroxyurea (HU), a reagent that stalls DNA replication, from which Schultz et al. (2003) inferred that PARP1 could not participate in carrying out HR. The hyper-recombination phenotype in PARP1-deficient cells was speculated to relate to a general defect in BER, rather than PARP1 being involved in catalyzing HR per se [24]. However, some recent data indicate that PARP1 in fact participates in HR [26]. A defective ATM-kinase activity and reduced γH2AX foci formation in response to γ-irradiation were observed in PARP1-deficient cells, indicating a role of PARP1 in ATM activation and HR [4,27]. Hochegger et al. (2006) indicated that PARP1 competed with Ku (a DSB binding factor that mediates NHEJ) binding to DNA ends to promote HR [28]. Through laser microirradiation, Haince et al. (2007) reported that PARP1 could be recruited to DSBs[18]. In addition, PARP1 is thought to mediate efficient recruitment of Nijmegen breakage syndrome 1 (Nbs1) and Mre11 to DSBs in a γH2AX- and MDC1-independent manner. By detecting the localization of PARs instead of PARP1, Bryant et al. (2009) found that both replication protein A (RPA) foci and PARs induced by HU co-localized and hence reasoned that the stalling of replication forks might activate PARP1 [20]. They also found that knockdown or inactivation of PARP1 significantly impaired the recruitment of Mre11 to stalled replication forks, foci formation of RPA and Rad51, HR and replication restart. Therefore the authors suggested that PARP1 can detect disrupted replication forks and attract MRN complex for end processing that is required for subsequent recombination repair and restart of replication forks (Fig. 2, right panel). Meanwhile the two-ended DSBs resulted from I-SceI cutting could not activate PARP1, ensuing no change in gene conversion rate upon inhibition or loss of PARP1 [24]. Therefore, PARP1 seems to play a dual role in HR: first by promoting BER/SSBR to reduce DSB formation and HR and second by sensing stalled or collapsed replication forks and recruiting MRN complex to initiate the HR process.
The role of PARP1 in NHEJ
Unlike HR that is only active in S and G2 phases when sister chromatids become available, the NHEJ pathway can function throughout the cell cycle and is especially active in G1 phase [29]. Therefore, NHEJ is the major repair pathway for DSBs in mammalian cells. However, NHEJ is more error-prone than HR.
NHEJ includes D-NHEJ (also called C-NHEJ, representing classical NHEJ) and PARP1-dependent B-NHEJ (backup NHEJ, also called Alt-NHEJ) [26,30]. D-NHEJ is the major NHEJ pathway in vivo. The initiation of D-NHEJ requires the DNA-PK complex (Ku70/Ku80 and DNA-PKcs). The XRCC4/DNA Ligase IV complex is responsible for the final ligation step in the D-NHEJ pathway (Fig. 3). When the D-NHEJ pathway is inactivated, DSBs can be repaired by the B-NHEJ pathway which often leads to a deletion with microhomology at the repair junction. In addition, error-prone B-NHEJ surfaces as a main repair mechanism resulting in chromosome translocations associated with both spontaneous and therapy-related cancer [31]. Although the components and regulation of B-NHEJ are still largely unknown, PARP1 may be indispensible for it. Chemical inhibition and PARP1-depletion demonstrated that the B-NHEJ in vivo is completely dependent upon functional PARP1 [32]. Additionally, the requirement for PARP1 depends on the absence of Ku but not on DNA-PKcs. It has been documented that B-NHEJ activity is normally suppressed by D-NHEJ [33]. Compelling evidence suggests that this suppression is mainly attributed to the presence of Ku which inhibits PARP1 recruitment and subsequent PAR synthesis and single-stranded DNA production in response to DSBs [32,34]. When B-NHEJ is triggered, PARP1 binds to DSBs and is activated, facilitating the subsequent recruitment of MRN to DSBs. MRN then carries out limited resection and the XRCC1/DNA Ligase III complex finishes the final ligation step (Fig. 3).
More recently, it was proposed that PARP1 could inhibit the functions of D-NHEJ by some unknown mechanism [35]. The authors observed that suppressing D-NHEJ rescued cell viability rather than increased the toxicity of PARP1 inhibitors on BRCA2-deficient tumor cells. Considering that Ku can be PARylated by PARP1 and PARylation of Ku reduces its affinity to DSBs [36], PARP1likely functions as a negative regulator of D-NHEJ in HR-deficient cells [35] (Fig. 3).
PARP inhibitors and cancer therapy
The functions of PARP inhibitors in cancer therapy
As described above, PARP1 plays key roles in multiple DNA damage response pathways and thus maintains genome integrity. Notably, most of PARP1’s functions are dependent on its PARylation ability, which renders the latter a potential target for tumor therapy. The development of PARP inhibitors originated from the observation that nicotinamide, a product of PARP catalytic activity, is a weak PARP inhibitor. So far, multiple PARP inhibitors have been identified and exhibit promising anti-cancer potential. In general, clinical development of PARP inhibitors follows two distinct strategies based on the molecular status of a cancer cell. First, PARP1 inhibitors may find potential use in chemosensitization in combination therapies. Second, PARP inhibitors may be useful as a single agent in killing tumors in HR-deficient patients based on synthetic lethality, which means that blocking the functions of two gene products simultaneously can cause cell death, while blocking either one is nonlethal [37].
PARP inhibitors
Given its prominent antineoplastic potential, the development of specific, effective and safe PARP inhibitors has become a hot topic in the PARP field. At present, several PARP inhibitors have been used either for biochemical assays, such as 3′-AB (3′-aminobenzamide) and DIQ (1,5-dihydroxyisoquinolinediol) [18,27,30], or for clinical trials, such as ABT-888 (Veliparib), BSI-201(Iniparib), CEP9722, MK4827, Olaparib and PF01367338 (AG014699) [4]. In addition, various phase I and II clinical trials are currently underway to evaluate the use of PARP inhibitors in treatment of a variety of tumors including breast cancers, ovarian cancers, prostate cancer, colorectal cancer, gastric cancer, cancer of the small intestine, hepatocellular carcinomas, sarcomas, lymphomas, melanomas, pancreatic neoplasms and glioblastomas [12]. A phase II study of BSI-201 (Sanofi-Aventis) in combination with gemcitabine/carboplatin in patients with metastatic triple-negative breast cancer has shown a significant clinical benefit [38]. At present a subsequent phase III trial of BSI-201 with gemcitabine/carboplatin for breast cancer and squamous cell lung cancer therapy is being conducted.
PARP inhibitors in combined therapies
Many cancer therapies utilize DNA-damaging agents to kill tumor cells, which often triggers DNA repair and renders the cancer cells resistant to the therapies. Considering that PARP1 plays essential roles in multiple DNA repair pathways, it is expected that blocking PARP will sensitize tumors to chemotherapy or radiotherapy, thereby improving the therapeutic index of such approaches.
Several preclinical and clinical studies using PARP inhibitors in conjugation with cytotoxic agents including the monofunctional alkylating agent temozolomide (TMZ), the DNA-crosslinking agent cisplatin, and ionizing radiation (IR) have been conducted. PARP inhibitors can potentiate the antitumor effects of TMZ in multiple tumor types and currently they are being tested in combination therapy with TMZ for the treatment of metastatic malignant melanoma, glioblastoma multiforme, childhood neuroblastoma, advanced solid tumors, refractory solid tumors and lymphomas. PARP inhibitors also increase the cytotoxicity of platinum complexes in cisplatin-resistant ovarian tumor cells and BRCA2-deficient mammary tumor cells. Moreover, PARP inhibitors are currently evaluated as radiosensitizers in phase I and II clinical trials of treatment of head and neck cancers as well as CNS neoplasms [12].
PARP inhibitors in synthetic lethal therapies
BRCA2, which is required for HR repair of DSBs, is one of the major genes identified to be associated with susceptibility to breast and ovarian cancers. Two pioneering studies have independently demonstrated that PARP inhibitors selectively kill BRCA2-deficient tumor cells [39,40]. Since cancer cells are believed to become dependent on DNA repair pathways other than the one that leads to its initial mutability [37], and given the fact that PARP is essential for multiple DNA repair pathways, it is not surprising that inhibition of PARP in these cells can result in increased genomic instability and ultimately cell death.
The original model postulated that the antitumor effects of PARP inhibitors in BRCA2-deficient cells were due to defects in both BER/SSR and HR. However, recent data indicated that downregulation of XRCC1 (a key player in BER/SSBR) in BRCA2-deficient cells led to no synthetic lethality, while disabling NHEJ diminished the genomic instability and lethality of PARP inhibition in BRCA2-deficient cells [35]. PARP1 is proposed to inhibit the functions of DNA-PK and D-NHEJ in BRCA2-deficient cells. The treatment with PARP inhibitors will reverse the suppression and induce aberrant activation of D-NHEJ in BRCA2-deficient cells, and this activation is responsible for the ensuing genomic instability and eventual lethality. However, the PARP1 inhibition mechanisms on D-NHEJ are still elusive.
Recent studies indicate that PARP inhibitors could also cause synthetic lethality in cancers defective in other HR components. Additionally, genes involved in nucleotide excision repair (DDB1 and XAB2), XRCC1 and tumor suppressor gene PTEN have been found to be synthetically lethal with a PARP inhibitor through high-throughput RNA interference screenings [37,41,42]. These results will extend the utility of PARP inhibitors to a larger group of patients beyond those with BRCA mutations.
However, despite the significant antitumor potential of PARP inhibitors, it is worth noting that long-term PARP inhibition may cause deleterious effects such as secondary malignancies, especially in combined therapies with DNA-damaging agents. Studies using PARP1 mutant mice have implicated PARP1 in the suppression of tumorigenesis [43-45]. Although the incidence of spontaneous tumors in both PARP-1- / - and PARP-1 + / + groups is similar, spontaneous tumors appear earlier in PARP-1 - / - mice compared to the wild type group [44]. Additionally, malignant tumors including uterine tumors, lung adenocarcinomas and hepatocellular carcinomas, develop at a significantly higher frequency in PARP-1 - / - mice. Inactivation of PARP1 has also been shown to result in a high frequency of T cell lymphomas in severe combined immunodeficiency mice [45]. Moreover, PARP1 mutant mice are susceptible to chemical carcinogen-induced tumorigenesis [46]. Therefore, whether long-term PARP1 inhibition will have any deleterious effects such as secondary malignancies requires careful investigation, particularly when inhibitors are administered with DNA-damaging agents [10,47]. Therefore, safety profile of PARP inhibitors in cancer therapies must be addressed in the near future.
Tumor resistance of PARP inhibitors
Acquired resistance of PARP inhibitors
Although PARP inhibitors have unprecedented therapeutic potential for cancer patients, accumulating evidence demonstrates that resistance to these drugs develops in tumors in both preclinical and clinical settings [47].
Due to their carrying of the protein-truncating c.6174delT BRCA2 frameshift mutation, human CAPAN1 pancreatic cancer cells are defective in HR and are extremely sensitive to treatment with potent PARP inhibitors [48]. However, the Ashworth group derived PARP-inhibitor-resistant (PIR) clones from the KU0058948 (a PARP inhibitor)-exposed CAPAN1 cells. They found that new BRCA2 isoforms were expressed in the resistant lines as a result of intragenic deletion of the c.6174delT mutation and restoration of the open reading frame. These PIR clones could form DNA damage-induced RAD51 foci and were competent for HR. Hence, PIR can arise in BRCA2-deficient cells due to the reactivation of the gene by secondary mutations [10,49].
Considering that PARP inhibitor-treated BRCA2-deficient cells are defective in both HR and BER/SSBR, it is possible that other acquired mutations mediated by alternative error-prone repair pathways or translesion DNA synthesis may also contribute to the PIR.
Primary resistance of PARP inhibitors
It is known that not all BRCA1 or BRCA2 mutation carriers respond to the inhibition of PARP [50]. Although the exact cause for PIR remains elusive, the genomic instability in these cells might lead to additional mutations occurring during cancer progression that would give rise to PARP inhibitor resistance.
Some BRCA1-deficient cancers were found with null or low PARP1 expression [51]. Additionally, PARP is found to be hyperactivated in replicating BRCA2-defective cells, which correlated well with increased sensitivity to PARP inhibitors [52]. In contrast, PARP inhibitor–resistant BRCA2 mutant cells have normal levels of PARP activity. Therefore, low levels of PARP1 expression or PARP activity may render cancer cells refractory to treatment with PARP inhibitors.
Meanwhile, it was found that a subset of BRCA1-deficient breast cancers have lost 53BP1 protein expression (an important factor in DNA repair and checkpoint control) [53]. Data from recent studies suggest that loss of 53BP1 is also able to reverse the HR defect in BRCA1-deficient cells and render them resistant to PARP inhibitors [54,55].
Moreover, PARP1 has broad roles in chromatin remodeling, epigenetic regulation and metabolism, etc. It is possible that these functions might also contribute to primary and acquired PIR [47].
So far, synthetic lethal therapies using PARP inhibitors prove beneficial only in a small subset of the patient population. From a therapeutic prospective, identification and incorporation of biomarkers to predict responders to PARP inhibitors will be beneficial. Although relatively little biomarker information is currently available for patient stratification in PARP inhibitor therapies, assays to measure HR ability and PARP activity will be vital to discriminate patient populations responding or resistant to PARP inhibitors [56].
Given the unprecedented clinical potential of PARP inhibitors, the further research on PARP1 will have great significance, not only in elucidation of the complex mechanisms of PARP1 with respect to DNA repair, but also in selection of patient populations that will respond to treatment.