Molecular Pathways: Overcoming Radiation Resistance by Targeting DNA Damage Response Pathways

M.A. Morgan reports receiving a commercial research grant from AstraZeneca. No potential conflicts of interest were disclosed by the other author.

The following editor(s) reported relevant financial relationships: P.S. Steeg reports receiving a commercial research grant from Sanofi.

The members of the planning committee have no real or apparent conflict of interest to disclose.

Upon completion of this activity, the participant should have a better understanding of cellular responses to radiation-induced DNA damage as well as the therapeutic strategies for sensitizing tumor cells to radiotherapy through targeting DNA damage response/repair pathways.

This activity does not receive commercial support.

Many factors influence tumor cell sensitivity or resistance to ionizing radiation, including the tumor microenvironment (1, 2), membrane signaling (3), and the immune system (4). However, as the principal target of radiation, DNA damage is the most critical factor governing radiation-induced cell death (5). DNA is subject to an array of different types of damage following exposure to ionizing radiation, including base and sugar damage, cross-links, as well as both single- and double-strand breaks (SSB and DSB, respectively). In particular, DNA DSBs are largely responsible for the cellular lethality of radiation. Hence, the ability of cells to recognize and respond to DSBs is fundamental in determining the sensitivity (or resistance) of cells to radiation. In this review, we focus on radiation-induced DNA damage and repair as well as the cell-cycle checkpoint pathways activated to mitigate this damage, with an emphasis on therapeutic targeting of these pathways to improve radiotherapy outcomes.

DNA DSBs can be simple or complex depending on several factors, including the physical characteristics of the break, the chromatin architecture surrounding the break, and the kinetics with which the break is repaired (6–8). Simple DSBs are often classified as two-ended DSBs directly induced by radiation, occurring in euchromatic regions that are repaired with fast kinetics (Fig. 1A). On the other hand, complex DSBs are characterized as two-ended DSBs occurring in close proximity to other types of damage (cross-links, SSBs, etc.) or within heterochromatic regions. In addition, replication-associated DSBs, which occur as a result of a SSB colliding with an active DNA replication fork, referred to herein as one-ended DSBs, also represent a complex type of DSB. All of these complex types of DSBs are repaired with relatively slow kinetics. Given the involvement of persistent SSBs in the generation of one-ended DSBs, SSBs are also relevant to radiation-induced DSBs. SSBs may arise via direct radiation-induced DNA damage or as intermediaries formed during repair of other single-strand lesions [e.g., base excision repair (BER) of oxidized bases].

Ataxia telangiectasia mutated (ATM), in cooperation with the trimeric protein complex composed of MRE11-RAD50-NBS1 (MRN), is the earliest responder to DNA DSBs. Activation and localization of these proteins at sites of DSBs in “radiation-induced foci” initiate phosphorylation of the histone variant, γH2AX and the recruitment of mediator of DNA damage checkpoint protein 1 (MDC1) as well as additional ATM and MRN molecules, necessary for the orchestration of subsequent DNA DSB repair and cell-cycle checkpoints. In addition to ATM, ataxia telangiectasia mutated and RAD3-related (ATR) plays a role in initiating the response to DNA DSBs in S and G₂ phases of the cell cycle, especially in the context of replication-associated one-ended DSBs (9). In contrast with DSBs, direct SSBs are detected by PARP1, which catalyzes the formation of poly-ADP-ribose (PAR) chains on itself and other acceptor proteins to facilitate the recruitment of DNA repair factors such as X-ray cross-complementing protein 1 (XRCC1) and DNA polymerase β (POLβ).

DNA DSB repair is composed of two major and mechanistically distinct processes: nonhomologous end-joining (NHEJ) and homologous recombination (HR; Fig. 1B). NHEJ and HR contribute to the fast and slow components of DSB repair with half-lives in the range of 5 to 30 minutes and 2 to 5 hours, respectively (10, 11). Although highly efficient, NHEJ is error prone, given its ability to catalyze simple rejoining reactions between DNA ends with no sequence homology. NHEJ repairs the majority of direct two-ended radiation-induced DSBs and is the predominant DSB repair mechanism in G₁ phase, although activity is present in all phases of the cell cycle. Following recognition of DNA DSBs, NHEJ begins with the stabilization of free DNA ends mediated by binding of the KU70 and KU80 heterodimers and subsequent recruitment of DNA-PKcs. 53BP1 and KU70/80 prevent DNA resection, an initial step of other DSB repair processes. NHEJ proceeds in a DNA-PK–dependent manner with recruitment of other core NHEJ proteins, including XRCC4 (X-ray cross-complement protein 4), LIG4 (ligase 4), XLF (XRCC4-like factor), and the recently identified PAXX (paralog of XRCC4 and XLF; ref. 12), which together mediate alignment and ligation of DNA ends. NHEJ likely represents the a priori DNA DSB repair mechanism as other DSB repair pathways, discussed below, only occur following initial attempts to repair by NHEJ (13).

Alternative end-joining (alt-EJ; also referred to as alt-NHEJ or B-NHEJ) mechanisms are also being elucidated and represent pathways that may compensate in the absence of core NHEJ proteins or when NHEJ or HR initiates, but fails to complete repair (reviewed in ref. 14). Alt-EJ is generally distinguished from classical NHEJ based on its lack of requirement of core NHEJ proteins such as DNA-PKcs and KU70/80, its slower kinetics, and its dependence on PARP1, XRCC1, and LIG1 or LIG3. In some cases, alt-EJ is associated with microhomologous DNA sequences (e.g., microhomology-mediated end-joining), which likely represent a sub-pathway of alt-EJ. The factors involved in alt-EJ (and its probable subpathways) are still being elucidated, as it is not yet clear whether other factors found at alt-EJ junctions such as KU70/80, MRN, and CtIP [(carboxy-terminal binding protein) interacting protein] are direct contributors to alt-EJ or are simply remnants from failed repair attempts by NHEJ or HR.

HR repair is a relatively slow, yet highly accurate repair process owing to its requirement for extensive end resection and homologous DNA sequences. Although the contribution of HR to the repair of simple radiation-induced DSBs is likely minor, HR is crucial for the repair of complex DNA DSBs (Fig. 1B). This is because, as lesions become increasingly complex, homologous templates are necessary for faithful repair. Given the requirement for homologous DNA sequences, HR can only operate during the S and G₂ phases of the cell cycle when sister chromatids are present. The core components of HR execute the major steps of the pathway beginning with resection of 5′ DNA ends, which involves the MRN complex and CtIP, followed by replication protein A (RPA) binding to single-stranded 3′ DNA ends. Initiated by BRCA2, RAD51 displaces RPA and mediates strand invasion to a homologous DNA strand forming the characteristic Holliday junction structure, followed by the final steps of HR, which include synthesis at the single-strand DNA regions, branch migration, and junction resolution.

As alluded to above, unrepaired SSBs contribute to the formation of complex DSBs. The SSB repair pathway is a common pathway shared for repair of direct SSBs as well as the SSB intermediates created during BER by excision of a damaged base and its corresponding phosphodiester bond. Following SSB detection (described above), PARP1 facilitates SSB repair by recruitment of the XRCC1 scaffolding protein to the SSB (Fig. 1B). XRCC1 promotes recruitment and stabilization of SSB repair enzymes, including those that mediate end-processing (e.g., APE1), gap synthesis (e.g., POLβ), and DNA ligation (e.g., LIG1 or LIG3; reviewed in ref. 15).

In response to DNA damage, cell-cycle checkpoints are activated to block cell-cycle progression and prevent propagation of cells with damaged DNA. The major DNA damage inducible checkpoints occur in G₁, S, and G₂ and are initiated by the same machinery responsible for DNA DSB repair recognition, ATM and ATR. The G₁ checkpoint is mediated mainly by ATM, which results in phosphorylation and activation of P53 transcriptional activity, leading to increased transcription of the cyclin-dependent kinase inhibitor P21, which induces G₁ arrest by inhibiting G₁ and early S-phase cyclin–CDK complexes (Fig. 1C). Cells in S-phase at the time of radiation exhibit a slowing of DNA synthesis mediated by two distinct pathways, ATM/NBS1/SMC1 and ATM/CHK2/CDC25A/CDK2 (16). Although it is not clearly understood how the former pathway regulates DNA synthesis, there is evidence that recruitment of the MRN complex to replication sites by RPA is involved (17). In contrast, the latter pathway negatively regulates DNA replication by preventing loading of the replication factor, CDC45 onto replication origins (18). Activation of the G₂ checkpoint in response to DNA damage prevents entry of cells into mitosis and is initiated by ATM-mediated activation of CHK1 and CHK2. Although ATM is the initial activator of this pathway, delayed ATR activation contributes to a sustained checkpoint response and is especially important in the context of replication-associated DNA DSBs (9). Activated CHK1 and CHK2 phosphorylate CDC25C and trigger its cytoplasmic sequestration and inactivation. In the absence of CDC25C phosphatase activity, CDK1 remains bound by inhibitory phosphorylations (catalyzed by the WEE1 kinase) resulting in G₂ arrest.

Given the lethality of unrepaired DNA DSBs, the development of agents designed to target proteins involved in the DNA damage response, and thus prevent repair or cell-cycle checkpoints in response to DNA damage, is an intense area of investigation. Although most of these agents (Table 1) are effective radiosensitizers, tumor cell selectivity remains an outstanding issue in their development. As monotherapy, the main developmental thrust for these agents is based on the principles of synthetic lethality, such that agents targeting a given DNA damage response pathway are strategically being applied to tumors with aberrations in other DNA damage response pathways. This strategy confers tumor cell selectivity because normal tissues have intact DNA damage response pathways and thus are spared from synthetic lethality. For radiotherapy combination studies, however, these synthetic lethal interactions are generally not being utilized, despite their potential to maximize the efficacy of these agents with radiation. Given that modern radiotherapy planning and delivery facilitate the induction of tumor-selective DNA damage, radiation should potentiate synthetic lethal interactions between DNA damage response pathways and therefore may improve the efficacy of some less robust synthetic lethal interactions. In this section, we review synthetic lethal interactions between agents and genes involved in radiation-induced DNA damage responses. We also begin to address whether extending these synthetic lethal mechanisms to radiation combination studies can improve the therapeutic window for these agents.

As monotherapy, the prototypical example of synthetic lethality between an agent and a gene involved in the DNA damage response is PARP inhibition in BRCA1/2-mutant cancers. Since this discovery, numerous other synthetic lethal interactions between DNA damage response pathways have been elucidated. For example, ATM-defective cancers are particularly susceptible to DNA-PK or ATR inhibition (19, 20). Likewise, ERCC1-defective tumors are sensitive to ATR inhibition (21).

Although extensive evidence exists to support the ability of DNA damage response–targeted agents to radiosensitize (22–26), considerably less data are available to support their tumor cell selectivity. Perhaps, the best established tumor cell–selective mechanism is illustrated by the synthetic lethality occurring between CHK1 or WEE1 inhibitors and mutant TP53. In combination with radiation, CHK1 or WEE1 inhibitors are synthetically lethal in TP53-mutant cancers (26–28). This is due in part to abrogation of the G₂ checkpoint by CHK1 or WEE1 inhibitors, which is particularly detrimental in TP53-mutant cancers that lack a G₁ checkpoint. In addition, it is plausible that the ability of CHK1/WEE1 inhibitors to inhibit HR (29–31) may also play a role in their selectivity, as TP53-mutant cancer cells are more likely than normal cells to rely on HR for DSB repair due to their inability to arrest in G₁ where NHEJ is the dominant DSB repair mechanism. Similarly, TP53-mutant gliomas are preferentially radiosensitized by ATM inhibition, although the mechanism of this selectivity is not fully understood (32).

There are numerous ongoing clinical trials combining PARP1/2 inhibitors with radiation or chemoradiotherapy (Table 1). A suggested model for radiosensitization by PARP inhibition involves inhibition of BER/SSB repair, leading to persistent radiation-induced SSBs that are converted to cytotoxic one ended DSBs in replicating cells (33, 34). Given that one-ended DSBs require HR for repair and that PARP inhibition is highly effective in HR defective (i.e., BRCA mutant) cancers, it is plausible to speculate that BRCA1/2-mutant cancers should be radiosensitized by PARP inhibition. This, however, is likely an oversimplification of the potential mechanisms of interaction between radiation, BRCA mutation and PARP inhibition, underscoring the requirement for thorough investigation in this area. In addition to inhibition of BER/SSB repair, inhibition of alt-EJ is also involved in PARP inhibitor-mediated radiosensitization. This is supported by the findings that radiosensitization is replication independent in NHEJ-deficient cells (33) and that POLβ null cells, which are defective in BER/SSB repair, are maximally radiosensitized by PARP inhibition (35). Taken together, these studies indicate that radiosensitization by PARP inhibitors involves multiple factors, including replication, BRCA1/2 mutation, as well as NHEJ and BER/SSB repair status that require careful evaluation in order to maximize the therapeutic index of PARP inhibitors with radiation.

Given the observed synthetic lethal interactions between the BER/SSB and HR pathways, illustrated by the efficacy of PARP inhibitors in BRCA1/2-mutant tumors, the inverse has been investigated to determine whether HR inhibitors are effective in BER-defective cancers. Indeed, in BER-defective tumors, ATM inhibition or RAD51 deficiency preferentially induces radiosensitization in β-deficient relative to POLβ-proficient tumor cells (36). These studies underscore the importance of the development of selective HR inhibitors such as those targeting RAD51 (37, 38) and their testing as radiosensitizers, especially in cancers with a high frequency of POLβ mutations (reviewed in ref. 39).

Efforts are under way to extend these synthetic lethal interactions beyond genes directly involved in the DNA damage response to common oncogenes such as KRAS and MYC that, when deregulated, lead to replication stress, genomic instability, endogenous DNA damage, and ultimately an increased reliance on DNA damage response pathways such as those mediated by ATR/CHK1 (40–43). Underscoring the potential efficacy of targeting ATR/CHK1 in KRAS/MYC-driven cancers, initial studies have suggested that CHK1 inhibition (alone or in combination with a PARP inhibitor) confers greater radiosensitization in isogenic KRAS-mutant versus KRAS wild-type cancer cells (reviewed in ref. 44). Strategies for directly targeting oncogene-induced replication stress through the use of small molecules that disrupt RPA-mediated protein–protein or protein–single-stranded DNA interactions are also under way (45–47). Taken together, these types of studies will determine whether oncogene-induced replication stress is a targetable phenomenon and may provide a rationale for extending the use of other agents targeting the DNA damage response to cancers with oncogene-induced replication stress. Future studies are necessary to determine whether radiotherapy enhances these interactions.

In order to pair a targeted agent with a tumor cell defect in a specific DNA damage response pathway, reproducible assays that capture the activity of these pathways in human tissues are necessary. Although numerous mutations in genes involved in individual repair pathways have been found to influence sensitivity to DNA damage response–targeted agents, a complete understanding of the genes, their specific mutations, and their mutational landscape that leads to overall functional defects in DNA damage response pathways is still lacking (48). Therefore, functional assays provide valuable insights into the global activity of a given DNA damage response pathway. For example, HR status has been determined in breast tumor biopsies by assessing RAD51 focus formation following ex vivo irradiation (49). The utility of this functional HR assay is illustrated by its ability to identify HR-defective cancers beyond those with characteristic BRCA1/2 mutations (50). Furthermore, functional assays have been used to assess BER in tumor biopsies and may prove useful for predicting sensitivity to DNA damage response–targeted agents (51). Although functional assays for NHEJ are more difficult because core NHEJ proteins do not form readily visible foci at DSBs due to their low quantities, some success in this area was demonstrated by detection of phopsphorylated-KU70 foci (52).

In the absence of obvious defects in DNA damage response pathways, an additional strategy for achieving synthetic lethality is to use pairs of agents strategically selected to target epistatic DNA damage response pathways. In an effort to extend the synthetic lethality of PARP inhibitors beyond BRCA1/2-mutant tumors, agents that inhibit HR are an intense area of investigation. To date, many agents that indirectly inhibit HR have been identified and are being combined with PARP inhibitors, such as inhibitors of EGFR (53), PI3K (54), HSP90 (55), CHK1 (56), and WEE1 (30). Some of these agent pairs, such as those targeting WEE1-PARP or HSP90-PARP, appear to be especially effective when given in combination with radiotherapy (reviewed in ref. 44), supporting the notion that radiation-induced DNA damage can potentiate these interactions. Finally, based on the consistent observation that PARP inhibitor treatment in combination with radiotherapy leads to a more robust G₂ checkpoint (likely due to increased DNA damage), agents that inhibit the G₂ checkpoint, such as CHK1 and WEE1 inhibitors, lead to improved radiosensitization, especially in TP53-mutant cancers that lack a G₁ checkpoint (30, 56, 57). Given that pairs of agents do not afford the same tumor cell selectivity that an agent paired with a tumor-specific mutation does, the tumor cell selectivity of agent pairs needs to be carefully investigated.

In recent years, the number of known synthetic lethal interactions between DNA damage response pathways has expanded considerably, and in some instances these interactions are being exploited in preclinical radiation studies. Despite these advancements, however, none of the current clinical radiation studies (Table 1) are incorporating this information. To improve the therapeutic window for these DNA damage response–targeted agents in combination with radiotherapy, mechanisms of synthetic lethality in the context of radiosensitization need to be further investigated. In addition, both functional and genomic assays that permit identification of DNA damage response pathway defects in tumor cells need to continue to be developed. The large number of DNA damage–targeted agents in development and the frequency of tumor defects in DNA damage response pathways offer great potential to improve the efficacy of radiation therapy.

Conception and design: M.A. Morgan, T.S. Lawrence

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.A. Morgan, T.S. Lawrence

Writing, review, and/or revision of the manuscript: M.A. Morgan, T.S. Lawrence

The authors thank Dr. Thomas Wilson for helpful discussions and Dr. Qiang Zhang for article review and editing.

M.A. Morgan was supported by the NIH under award number R01CA163895. T.S. Lawrence was supported by the NIH under award numbers R01CA138723, P3CA130810, and P50CA130810 and an A. Alfred Taubman Scholarship.