Improving the Therapeutic Ratio of Radiotherapy by Targeting the DNA Damage Response

Fig. 1.1

Responses of mammalian cells to DNA DSBs induced by gamma irradiation following induction of DNA DSBs by ionising radiation, a DNA damage response consisting of cell cycle checkpoint activation and DNA repair is generated. If successful, this will result in DNA repair and cell survival

Detection of Radiation-Induced DNA Damage and Initiation of DDR

Efficient DDR relies upon rapid detection of DNA damage and subsequent escalation of appropriate DDR pathways. The MRN complex consisting of MRE-11, NBS-1 and Rad50 proteins represents the major DNA DSB detector within mammalian cells. Ku70/Ku80 proteins , which are key effectors of the non-homologous end joining (NHEJ) pathway, also bind directly to DNA DSBs facilitating early repair of most DNA DSBs. Following detection of DNA damage, this signal must be amplified and coordinated in order to facilitate a cellular environment conducive to DNA repair. This is achieved by the actions of three apical DDR proteins: ataxia telangiectasia mutated (ATM) , ataxia telangiectasia and Rad 3 related (ATR) and DNA-dependent protein kinase (DNAPK) which are phosphatidylinositol 3-kinase-related kinases (PIKKs) . The apical PIKKs phosphorylate a repertoire of DNA repair and checkpoint control proteins ensuring timely activation of cell cycle checkpoints and initiation of DNA repair mechanisms appropriate to DNA lesion stimuli and allow modification of heterochromatin and other more general intracellular environmental features in order to promote cellular survival. Apical PIKKs are appealing targets for radiosensitisation strategies and their functions and are described below.

Ataxia Telangiectasia Mutated (ATM)

ATM is a highly prolific kinase which phosphorylates many substrates in response to DNA DSBs and has a dual role in both cell cycle control and repair of a subset of DNA DSBs. For a detailed review, see Shiloh et al. [1]. Mutations in ATM are responsible for the radiosensitivity syndrome ‘ataxia telangiectasia’, first described in 1975 [2]. Cells derived from patients with ataxia telangiectasia show deficient G1/S, S and G2/M checkpoints and deficient DNA DSB repair. ATM exists as an inactive dimer or multimer until DNA damage occurs, upon which autophosphorylation at serine 1981 occurs, allowing the dissociation of ATM dimers into active monomers. The exact mechanism of ATM activation is debated in current literature, and activation may occur via direct interaction with DNA DSBs, in response to conformational changes in heterochromatin structure or via the MRN complex [3, 4].

The proportion of DNA DSBs that cannot be repaired in ATM-mutant cells is estimated at around 10–20 % of the total DSB burden. ATM has a role in promoting DSB repair executed by NHEJ in G1 phase and by both NHEJ and HR in G2, and the proportion of ATM-dependent DSBs is similar in both phases of the cell cycle. Goodarzi et al. investigated the role of ATM in chromatin modification and demonstrated that ATM has a role in repair of heterochromatic DSBs [5]. This model proposes that in G1 phase, around 75 % of DNA DSBs occur in euchromatin regions and that ATM is not required for the repair of these lesions. However, in heterochromatic regions, nucleosome flexibility is constrained by factors such as KAP-1 , which severely limits DSB repair. In this model, DSBs in heterochromatin are responsible for the slow phase of DSB repair , since the cell needs to execute additional steps to rejoin DSBs occurring in this relatively inaccessible chromatin context. ATM is able to phosphorylate KAP-1 , thereby generating sufficient elasticity in DNA tertiary structure to allow repair. It has previously been suggested that ATM’s primary role is to deal with complex DNA DSB lesions, since Artemis and ATM defects create epistatic DNA repair defects and Artemis has a vital role in end resection for facilitation of NHEJ [6]. However, the proportion of ATM-dependent DNA DSBs appears not to increase following irradiation with high LET radiation types which cause more complex DSBs, which implies that ATM-dependent repair is not necessarily associated with complex DNA DSBs .

Nevertheless, ATM is also known to have roles in specialised DSB repair mechanisms that are not related to heterochromatin such as VDJ class switching and meiotic recombination. Alvarez-Quilon et al. demonstrated that ATM is necessary for the repair of DNA DSBs with blocked ends and that this requirement is independent of chromatin status [7]. The authors speculated that ATM could promote nucleolytic activity to eliminate blockage at DNA ends via the MRN complex, CtIP or Artemis or it could restrict excessive nucleolytic degradation of DNA ends by inhibiting these same nucleases or by phosphorylation of H2AX. These two models are not necessarily conflicting, since ATM may have roles in both complex DNA lesion repair and modification of chromatin .

Ataxia Telangiectasia and Rad 3 Related (ATR)

ATR has a critical role in the DDR by protecting cells from replication stress. Replication stress can be defined as the slowing or stalling of replication forks during duplication of DNA and is characterised by the presence of single-stranded DNA (ssDNA) within the nucleus. Cancers in general are known to exhibit high levels of replication stress, which is thought to be induced primarily by oncogene activation, leading to upregulation and increased dependence upon the ATR-Chk1 pathway [8]. Furthermore, the DNA damage induced by ionising radiation (both SSBs and DSBs) is a significant source of replicative stress in the irradiated cell. The role of ATR in the DDR is reviewed in Marechal et al. [9]. ATR has an essential role in the survival of proliferating cells, and its deletion leads to embryonic lethality in mice and lethality in human cells [10]. ATM and ATR share many phosphorylation substrates; however, they have distinct roles in DDR and cannot be viewed as redundant in function. ATR is activated by RPA-coated ssDNA; hence, any situation leading to the formation of ssDNA will result in the activation of ATR. ATR phosphorylates Chk1 which leads to G2/M checkpoint activation, allowing time for damage repair. However, both ATR and Chk1 have additional important functions in maintaining the integrity of replication forks. Replication fork collapse is characterised by the dissociation of replisome contents and may result in generation of a DSB. This process is still poorly understood and may be the result of replisome dissociation/migration, nuclease digestion of a reversed fork or replication runoff [11]. ATR is activated by ssDNA generated at stalled replication forks and acts to stabilise the fork and initiate cell cycle checkpoint activation and inhibition of DNA replication origin firing on a global scale throughout the cell nucleus. ATR activation inhibits origin firing via the phosphorylation of the lysine methyltransferase MLL, which alters chromatin structure around replication origins [12]. In this manner, the stalled fork can then be restarted when the replication stress stimulus has been resolved.

DNA-Dependent Protein Kinase (DNA-PK)

DNA-PK has a critical role in DDR via its function in NHEJ, as discussed below. It phosphorylates a smaller number of substrates in comparison to ATR and ATM. However, DNA-PK is able to phosphorylate some substrates of ATM in ATM-defective cells, allowing a degree of functional redundancy. In particular, DNA-PK is able to phosphorylate histone H2AX in the absence of ATM [13].

Activation of the apical DDR PIKKs results in cell cycle checkpoint initiation and attempted DNA repair. These processes will be considered separately as follows .

Cell Cycle Checkpoint Control

Mammalian cells have three main cell cycle checkpoints that are activated following DNA damage: G1, intra-S and G2/M. These are shown in Fig. 1.2. The checkpoints regulate progression through the cell cycle, preventing a cell from progressing into the next phase of the cell cycle prior to satisfying the requirements of the previous phase. Progression through the cell cycle is controlled by cyclin-dependent kinases (CDKs) and cyclins, the names alluding to their cyclical accumulation and destruction through the cell cycle. These proteins form cyclin-CDK complexes whose activity ultimately regulates the machinery responsible for cycle progression. For a review of the cellular machinery controlling cell cycle checkpoints, see Lukas et al. [14].

Fig. 1.2

Cell cycle control in response to DNA damage . Simplified diagram of cell cycle control following activation of the upstream PIKKs ATR and ATM. ATM is activated by DNA DSBs and influences all three major checkpoints, whereas ATR is activated by RPA-coated ssDNA and has its major roles in the intra-S checkpoint and maintenance of the G2/M checkpoint

The G1 checkpoint is usually very robust in eukaryotic cells; however, in malignant cells, the G1 checkpoint is frequently absent due to mutations affecting the p53 pathway. For example, glioblastoma and other cancer cells frequently fail to initiate a G1 checkpoint response to irradiation. Normal G1 checkpoint function requires functioning p53, which is phosphorylated in response to DNA damage by both ATM and Chk2 proteins. This leads to a reduction in the binding of MDM2 to p53, and subsequent p53 activation, resulting in its nuclear accumulation and stabilisation . Increased levels of p53 protein stimulate increased transcription of p21, which binds and inhibits CDK2-cyclin E activity, preventing the cell from entering S phase. The G1/S checkpoint is highly sensitive, but limited by the time required for p21 upregulation [15]. Alternative activation of the G1/S checkpoint is mediated via phosphorylation of Cdc25A, again by ATM and Chk2, which then targets Cdc25A for proteasomal degradation. Cdc25A removes inhibitory phosphate groups on CDK2, allowing progression into S phase [16].

The intra-S checkpoint is activated in response to replication stress or other difficulties encountered by the cell during S phase. It operates to slow DNA replication rather than stop it entirely and is p53 independent. The components of the S phase checkpoint suppress origin firing and slow replication fork progression to reduce the rate of DNA replication. Abnormalities in S phase checkpoints result in the radioresistant DNA synthesis (RDS) phenotype, i.e. cells are unable to stop or delay the synthesis of DNA following induction of DNA damage by radiation.

Cancer cells frequently demonstrate an increased dependency upon G2/M checkpoint activation to allow repair of DNA damage prior to entering mitosis, since the G1/S phase checkpoint is often dysfunctional in malignant cells due to deficiencies in the p53 pathway. Progression through the G2/M checkpoint with unrepaired DNA damage can result in cell death, and therefore it is essential that control of the G2/M checkpoint is maintained. Activation of the G2/M checkpoint occurs via ATM and ATR which phosphorylate Chk1 and Chk2, leading to phosphorylation of Cdc25 phosphatases. The G2/M checkpoint has a defined threshold of sensitivity, with activation and maintenance of G2/M arrest appearing to require 10–15 DSBs [17]. The G2/M cell cycle checkpoint arrests heavily damaged cells in G2 to provide time for repair of DSBs, and it is proposed that this may be important for slow phase repair in G2 via homologous recombination. However, the G2/M checkpoint is inherently insensitive and allows cells to enter mitosis carrying a measurable number of unrepaired DSB [18].

DNA Repair Processes

Exposure to a 2Gy dose of radiation will produce on average around 2000 SSBs and 80 DSBs. DNA DSBs are much more difficult for cells to repair and have long been considered the lesions responsible for lethality following irradiation. Figure 1.3 illustrates DNA DSB repair kinetics in mammalian cells following gamma radiation adapted from Goodarzi et al. [19]. There is an initial fast phase of repair lasting 1–3 h which represents DNA DSBs that can be efficiently repaired by the cell. In addition to the fast phase of repair, there is a longer ‘tail’ which is termed the slow phase of DNA DSB repair and can extend past 24 h. Both slow phase and fast phase repair occur simultaneously. If left unrepaired, even a single DNA DSB can result in loss of genetic information and cell death [20] so it is unsurprising that mammalian cells have developed complex and highly efficient systems for their repair. DNA DSBs are repaired predominantly by two pathways, homologous recombination (HR) and non-homologous end joining (NHEJ), although back up pathways such as microhomology-mediated end joining (MMEJ) also exist. For a review of DNA DSB repair, see Shibata and Jeggo [21].

Fig. 1.3

Illustrative schematic of kinetics of DNA DSB repair following irradiation in mammalian cells. The majority of DSBs are repaired a short time after irradiation in the ‘fast’ phase of DNA DSB repair via NHEJ. However, a subset of DNA DSBs requires much more time for repair, due to complexity and/or chromatin context, and is represented by a ‘slow’ phase tail on the above illustration. Slow phase repair is achieved via NHEJ in G1 phase and HR repair in G2 phase. Adapted from [19]

Non-homologous End Joining (NHEJ)

The bulk of DNA DSB repair in mammalian cells is undertaken by NHEJ, exclusively so in G1 cell cycle phase where cells have a diploid DNA content. NHEJ is involved in both fast phase repair and slow phase repair in G1 cells and in the fast phase of repair in G2 cells [6]. NHEJ involves the processing of broken DNA termini to form compatible ends which can then be ligated back together. NHEJ is a relatively simple, rapid and efficient method of DNA DSB repair but is error prone and associated with loss of genetic information. The mechanisms of NHEJ can be simplified into three steps: (for a comprehensive review, see Weterings et al. [22]) (1) capture of both ends of the broken DNA molecule, (2) bridging of the two broken DNA ends and (3) religation of the broken DNA molecule. NHEJ is thought to make the first attempt at rejoining the majority of DNA DSBs , even in G2 phase where HR is competent, due partly to the cellular abundance of Ku70 and Ku80 and their high affinity for DNA termini [23, 24]. NHEJ and its major protein components are summarised in simplified form in Fig. 1.4.

Fig. 1.4

Schematic diagram of non-homologous end joining (NHEJ) repair. NHEJ is initiated by the binding of Ku70/Ku80, followed by the recruitment of DNA-PKcs and its subsequent autophosphorylation . End processing is achieved via Artemis, and additional factors before the broken DNA ends are ligated

An alternative mechanism of NHEJ is thought to occur via microhomology-mediated end joining (MMEJ) [25, 26]. For a detailed review, see McVey et al. [27]. MMEJ has a requirement for limited MRN-dependent end resection and relies upon homologous matching of 5–25 base pairs on both strands in order to correctly align the DNA DSB ends. Any overhanging or mismatched bases are removed and missing bases inserted. The process is particularly error prone, since it does not identify sequences lost around the DSB. MMEJ appears to act as a reserve DSB repair pathway but can also repair DSBs generated at collapse of replication forks. The process is dependent upon ATM, PARP-1, MRE-11, CtIP and DNA ligase IV but operates independently of Ku or DNA-PKcs [27]. The extent to which MMEJ contributes to DSB repair in normal cells is unknown, but it has been shown to assume importance in cancer cells bearing defects in other DSB repair pathways [28].

Homologous Recombination (HR)

The homologous recombination (HR) pathway represents a more complex and sophisticated mechanism of DNA DSB repair. Although NHEJ repairs the majority of DNA DSBs , HR contributes to the repair of DSBs in specific circumstances, such as the one-ended DSB created by the collapse of DNA replication forks and a subset of DNA DSBs in G2 that are repaired with slow kinetics [23, 29, 30]. HR is conventionally considered to be limited to S and G2 phases of the cell cycle, since it relies upon homologous DNA sequences (in the form of the duplicated DNA strand of a sister chromatid) to effect repair. Because of this, however, it is highly accurate. For a more detailed review of the process, see Filippo et al. [31], Li et al. [32] and Krejci et al. [33]. In brief, HR is initiated by resection of the 5′ DNA end of the DSB in order to create 3′ SS DNA which can then invade a partner chromosome. End processing creates 3′ ends following resection of nucleotides from the 5′ break ends. Extension of resection is tightly regulated by the repositioning of 53BP1 via a BRCA 1-dependent process (9 Jeggo 2014 review). Resected 3′ ends are then quickly bound by replication protein A (RPA) , which protects ssDNA and removes DNA secondary structure in order to facilitate formation of a ‘presynaptic filament ’ consisting of Rad51-coated ssDNA [34, 35]. Rad51 is a recombinase, i.e. an enzyme which facilitates genetic recombination and forms a helical filament on ssDNA which holds it in an extended conformation to aid the search for homology. BRCA 2 has an essential role in the loading of Rad51 onto ssDNA.

Once assembled, the presynaptic filament captures a duplex DNA molecule and begins its search for the homologous sequence. Rad51 facilitates the physical connection between the invading DNA strand and the DNA duplex structure leading to the formation of heteroduplex DNA (‘D loop’) with a Holliday junction (HJ) , as described in Fig. 1.5. Synthesis of DNA and repair of the DSB lesion then occurs using the undamaged DNA strand of the heteroduplex DNA molecule as a template. Following successful repair, resolution of the heteroduplex DNA molecule occurs, generating crossover or non-crossover products .

Fig. 1.5

Schematic diagrams of homologous recombination (HR) repair. HR repair is initiated by end resection and coating of ssDNA by RPA and subsequently Rad51. The search for a homologous sequence on the sister chromatid is initiated by strand invasion and subsequent Holliday junction formation. Synthesis of new complementary DNA sequence and Holliday junction resolution results in successful DNA DSB repair

Poly (ADP-Ribose) Polymerases (PARPs)

Whilst the main features of the DDR to DNA DSB have been explored, it should not be forgotten that responses to single-strand DNA breaks also influence the eventual outcome of radiation-induced DNA damage . Gamma or X-radiation induces around 25-fold more SSBs than DSBs, but these are usually repaired promptly. If SSBs are not resolved efficiently, however, they can have significant effects on cell survival via the generation of DSBs. The PARP family of proteins is known to facilitate base excision repair (BER) which is one of the main cellular single-strand break repair pathways.

PARPs form a large protein family with diverse cellular functions including DNA repair, mitotic segregation, telomere homeostasis and cell death. PARPs are characterised by their catalytic function, which is poly(ADP-ribosylation). There are 18 reported family members; however, not all have definite poly(ADP-ribose) catalytic function, and only PARPs 1–3 have well-characterised roles in DNA repair. For an in-depth review of PARP function, see D’Amours and Burkle [36, 37]. PARP-1 is the most abundant and best understood family member, so the term ‘PARP’ will be used to refer to the actions of PARP-1 for the rest of this chapter.

Activated PARP modifies its substrates via covalent, sequential addition of ADP-ribose molecules that form branching poly(ADP-ribose) (PAR) polymers on its targets. The substrate from which PAR is formed is nicotinamide adenine dinucleotide (NAD+). Poly(ADP-ribosylation) is a commonly occurring post-translational modification in the cell. It creates negative charge on target proteins altering their three-dimensional structure and regulating interactions with other proteins and with DNA [38].

PARP is an efficient sensor of DNA damage and its rapid binding to damaged DNA results in its activation (Fig. 1.6). PARP can bind to a variety of DNA damage structures including SSBs and DSBs [39–42] and plays a major role in PAR synthesis following DNA damage: approximately 90 % of PAR production is attributable to PARP-1 in this context [43]. DNA-bound PARP undergoes automodification via the addition of long, negatively charged PAR polymers [36]. This autoPARylation promotes dissociation of PARP from the DNA molecule, allowing access of other DNA repair components to the damaged DNA [44–46] and facilitating their recruitment to the damaged sites. The list of substrates of PARP is extensive, and their DDR function can be modified both by PARylation and by direct interaction with PARP.

Fig. 1.6

The role of PARP in SSB repair. PARP detects SSBs and facilitates efficient repair via interactions with a variety of base excision repair (BER) factors. Automodification of PARP facilitates its dissociation from the damaged site

Although the precise role of PARP in DNA repair is still being elucidated, an important contribution to the repair of SSB lesions is well documented. Rather than being essential for SSB repair, however, PARP appears to increase the efficiency and kinetics of this process [47–49]. Activation of PARP promotes recruitment of the scaffold protein XRCC1 to damages sites [50]; PARP modifies and interacts directly with XRCC1 during this process. Lesions then undergo end processing before being repaired by either short patch or long patch mechanisms. PARP is known to interact with and modulate many SSB repair proteins, including DNA Lig III, DNA Pol Beta and others, whilst playing a clear role in base excision repair (BER) does not appear to be an absolute requirement for the function of this pathway [49]. The radiosensitising effects of PARP inhibition will be discussed below.

DNA Damage Response as a Therapeutic Target

From the discussion above, it can be predicted that targeting of the tumour cell DDR will lead to radiosensitisation via two distinct mechanisms. Inhibition of cellular checkpoint activation will promote transit of malignant cells into mitosis before DNA damage can be completed, thus increasing the probability of cell death, whilst inhibition of DNA repair will increase the incidence and persistence of unrepaired DNA breaks, thus enhancing the lethal effects of irradiation. Some of the key DDR effectors (e.g. ATM) are involved in both of these processes.

Exploitation of DDR as a therapeutic target often raises understandable concerns regarding toxicity to normal tissues. The concept of ‘tumour specificity ’ is vitally important in cancer therapy and particularly so when considering strategies that increase the biological effects of ionising radiation . If DDR inhibition were to sensitise normal tissues to the same degree as tumour cells, then no therapeutic gain would be made, since any increased tumour effect would be accompanied by an unacceptable increase in normal tissue toxicity.

Supporting the prospect of tumour-specific radiosensitisation , important differences between the DDR of tumours and normal tissues have been well documented. At the most fundamental level, the DDR presents a barrier to carcinogenesis during the early stages of tumour development [51]. Cellular populations in the process of carcinogenesis face selective pressures that promote survival of cells bearing mutations associated with altered DDR that increase their ability to tolerate oncogenic proliferative stress . At the population level, dysfunctional DDR can be advantageous, endowing a minority of tumour cells the capacity to generate and tolerate genomic instability and heterogeneity, leading to adaptability and a survival advantage in the hostile tumour microenvironment. Consistent with this, there is evidence to suggest that tumours may be profoundly deficient in some aspects of DDR, rendering them overly dependent on other DDR pathways to carry out necessary DNA repair. Examples of this behaviour are seen in the widespread loss of G1/S checkpoint integrity in solid tumours due to p53 mutation and resulting dependence upon G2/M checkpoint integrity. A further example is seen in the context of ‘synthetic lethality ’ in HR-deficient tumours, which are sensitive to therapies such as PARP inhibitors that create DNA lesions requiring HR for repair. Given that genomic instability is now considered a ‘hallmark’ of cancer, it is likely that DDR abnormalities are common in cancer cells [52]. Indeed the main reason why radiotherapy is a successful cancer treatment is because tumour cells are less able than the surrounding normal tissues to deal with the DNA damage caused by ionising radiation. The intact DDR of normal tissues ensures that a therapeutic ratio exists between tumour and normal tissue, allowing radiation to eradicate tumour cells whilst normal tissues are able to survive or tolerate the resulting DNA damage. Therefore, pharmacological inhibition of DDR exploits an inherent vulnerability of many cancer cells and represents a valid and promising therapeutic strategy.

Recently, a variety of small molecule inhibitors have become commercially available that possess the ability to specifically and potently inhibit individual DDR proteins. Although many of these are not yet sufficiently advanced to be anything more than laboratory tools, others such as the PARP inhibitor class have been licensed as single agents and are entering phase I and II clinical trials in combination with radiotherapy. A discussion on the current landscape of DDR inhibition in the context of radiation therapy now follows.

PARP Inhibition

PARP inhibitors represent the most developed class of DDR modifiers, largely due to early successful trials as monotherapy in the ‘synthetic lethality ’ setting [53]. There are now several PARP inhibitors entering clinical trials as radiosensitisers including AZD2281 (olaparib) , AG014699 (rucaparib) and ABT888 (veliparib) . Extensive preclinical investigation into their role as radiosensitising agents has been carried out and is summarised below.

In vitro work has demonstrated that PARP inhibitors (PARPi) provide modest radiosensitisation . Sensitiser enhancement ratios (SER) , which are a measure of the fold increase in radiation dose necessary to produce a given level of survival observed in the absence of the sensitising drug, have been reported in the range of 1.1–1.7, depending on the PARP inhibitor and cell line tested.

Brock et al. [54] demonstrated this effect in fibroblast and murine sarcoma cell lines, with SERs (at 10 % survival) of 1.4–1.6 using the PARP inhibitor INO-1001. Interestingly they also showed an enhanced sensitisation effect when INO-1001 was combined with fractionated radiotherapy, suggesting that PARPi was able to block interfraction repair of sublethal damage. This effect was also reported in a study of glioblastoma cell lines [55].

Other authors have confirmed the radiosensitising effects of PARPi in vitro in a variety of different tumour cell lines; these are summarised in Table 1.1 and include head and neck squamous cancer ; prostate cancer ; glioblastoma ; pancreatic , colon and cervix cancer ; and lung carcinoma cell lines.

Table 1.1

Summary of in vitro studies of radiosensitising effects of PARP inhibitors

Author	Parp inhibitor and radiation dose	Cell line	Assays	Outcome
Brock et al. [54]	INO-1001 10 μM, IR 0–8 Gy	CHO rodent fibroblast, c37 human fibroblast, SaNH murine sarcoma cell lines	Clonogenic survival and apoptosis	Decreased clonogenic survival in PARPi plus IR, effect enhanced by fractionation No increase in apoptosis
Albert et al. [56]	ABT888 (veliparib) 5 μM, IR 0–6 Gy	H460 lung carcinoma cell lines	Clonogenic survival, apoptosis, endothelial damage assay	Decreased clonogenic survival in PARPi plus IR vs. IR alone Increased apoptosis Inhibition of endothelial tubule formation
Dungey et al. [55]	AZD2881 (olaparib) 1 μM , IR 0–5 Gy	T98G and U87MG glioblastoma cell lines	Clonogenic survival, gamma H2AX foci	Decreased clonogenic survival in PARPi plus IR vs. IR alone, decreased DNA repair, DNA replication-dependent effect of PARPi, fractionation-sensitive effect
Loser et al. [57]	AZD2881 (olaparib) 500 nmol/l plus IR 0–8 Gy	Human and murine primary cells defective in Artemis, ATM, DNA ligase IV	Clonogenic survival, alkaline comet assay, gamma H2AX foci	PARPi radiosensitisation enhanced in ATM, Artemis and DNA ligase IV-deficient cells. Clonogenic survival decreased in rapidly dividing and DNA repair-deficient cells
Calabrese et al. [58]	AG14361 0.4 μM plus IR 8 Gy	LoVo and SW620 human colonic carcinoma cell lines	Clonogenic survival	PARPi plus IR decreased survival by inhibiting recovery from potentially lethal damage
Russo et al. [59]	E7016 3–5 μM plus IR 0–8 Gy	U251 glioblastoma, MiaPaCa pancreatic, DU145 prostatic carcinoma cell lines	Clonogenic survival, gamma H2AX foci, mitotic catastrophe, apoptosis	PARPi plus IR increased clonogenic cell kill and mitotic catastrophe, however no increase in apoptosis
Liu et al. [60]	ABT 888 (veliparib) 2.5 μM plus IR 5 Gy	H1299 lung cancer cells, DU145 and 22RV1 prostate carcinoma cell lines	Clonogenic survival, repair foci assay	PARPi plus IR reduced clonogenic survival, with effect seen in acute hypoxic cells and oxic cells

PARP inhibitors have been shown to decrease clonogenic survival and increase apoptosis and mitotic catastrophe in irradiated cells in vitro. The pro-apoptotic effects of PARPi vary between studies and are likely to be cell line dependent. Noel et al. demonstrated lack of radiosensitisation of asynchronously dividing human cell lines treated with PARPi, whilst HeLa cells synchronised in S phase were significantly sensitised to radiation by the addition of PARPi, suggesting that sensitisation was dependent upon DNA replication [61]. This was confirmed by Dungey et al. [55] who showed that radiosensitisation was enhanced by synchronisation in S phase and abrogated by aphidicolin (which creates an early S phase block). PARPi delayed repair of DNA damage and was associated with a replication-dependent increase in DNA DSBs as measured by gamma H2AX and Rad51 foci. Again radiosensitisation was increased with a fractionated schedule, indicating impaired repair of sublethal damage in PARPi-treated cultures. The authors proposed a mechanism whereby PARPi reduced the rate of SSB repair which, in replicating cells, increased the burden of DSBs due to generation of collapsed replication forks during S phase (see Fig. 1.7). They also proposed that the DNA lesions produced by collapsed replication forks in the presence of PARPi might be more complex and hence more difficult to repair. Persistent binding of chemically inhibited PARP to DNA (via steric hindrance) would prevent efficient recruitment of DNA repair proteins to the lesion, providing a potential explanation for this theory [62]. The observation that DNA replication is required in order for PARP inhibition to radiosensitise cells indicates that direct effects on DSB repair are unlikely.

Fig. 1.7

Mechanism of radiosensitisation by PARP-1 inhibition. PARP inhibition does not affect binding of PARP-1 to DNA SSBs but prevents their efficient repair by inhibiting recruitment of key BER effectors and by blocking access of repair elements to damaged sites. This results in delayed SSB repair and increases the likelihood of replication fork collapse by which mechanism SSBs are converted into cytotoxic DSBs during S phase

Loser et al. investigated the radiosensitising effects of PARPi on cells that were deficient in various DDR pathways, an effect which has been termed ‘synthetic sickness’. Pre-existing DDR pathway abnormalities were found to enhance the radiosensitising effects of PARPi when compared with effects in DDR competent cell lines. Whilst the underlying mechanism varied according to the specific DDR pathway abnormality, the addition of PARPi appeared to render DDR-deficient cells more vulnerable to radiation-induced DNA lesions that would otherwise have been repaired by alternative pathways [57].

Important work by Liu and colleagues [60] examined the effects of acute hypoxia on radiosensitisation by PARPi . Firstly, the clinical PARPi ABT-888 was shown to inhibit intracellular PARP activity in prostate and non-small cell lung carcinoma cell lines under conditions of hypoxia. Secondly, tumour cells under conditions of acute hypoxia were radiosensitised to the same degree as oxic cells. The authors concluded that ABT-888 remained an effective radiosensitiser under conditions of acute hypoxia, which is an important consideration in translating PARPi into clinical practice because most tumours are hypoxic to some degree [63, 64]. Chronic hypoxia induces downregulation of HR, which may allow targeting of chronically hypoxic cancer cells with a PARPi synthetic lethal strategy. Chan et al. have shown that PARPi-treated tumour xenografts with hypoxic subregions exhibited increased gamma H2AX signalling and reduced survival in an ex vivo clonogenic assay. However, the specific radiosensitising effects of PARPi in the context of chronic hypoxia were not investigated [65]. Nevertheless, the ability of PARPi to selectively target chronically hypoxic cancer cells is of significant clinical interest.

The radiosensitising effects of PARPi have been replicated by several authors in in vivo models. The results of these studies are summarised in Table 1.2. As an example, a recent paper by Tuli et al. demonstrated tumour growth inhibition and prolonged survival in an in vivo orthotopic model of pancreatic carcinoma [69].

Table 1.2

Summary of in vivo studies of radiosensitising effects of PARP inhibitors

Author	PARP inhibitor and radiation dose	Cell line	Assay	Outcome
Khan et al. [66]	GPI-15427 10, 30, 100, 300 mg/kg po, IR 2 Gy for 2 days	JHU012 and JHU012 head and neck cancer xenografts	Tumour growth delay apoptosis	PARPi plus IR inhibited tumour regrowth vs. IR Increased apoptosis
Clarke et al. [67]	ABT 888 7.5 mg/kg po bd, Temozolomide 33 mg/kg/day, IR 20 Gy over 11 days	Glioblastoma intracranial xenografts (MGMT hypermethylated)	Animal survival, body weight	PARPi-TMZ-IR prolonged survival vs. IR alone, minimal weight loss
Donawho et al. [68]	ABT 888 25 mg/kg/day via osmotic pumps, IR 20 Gy over 10 days	HCT116 xenograft human colorectal carcinoma	Animal survival	PARPi plus IR increased mean survival time vs. IR alone
Albert et al. [56]	ABT 888 25 mg/kg ip for 5 days, IR 10 Gy over 5 days	H460 xenograft, human lung carcinoma	Tumour growth delay, Ki67 staining, apoptosis, blood vessel density	PARPi plus IR delayed tumour regrowth vs. IR alone Decreased tumour vasculature Decreased proliferation Increased apoptosis
Calabrese et al. [58]	AG143615 or 15 mg/kg/day ip, IR 10 Gy over 5 days	SW620 human colon carcinoma	Tumour growth delay	PARPi plus IR delayed tumour regrowth vs. IR alone
Russo et al. [59]	E7016 30 mg/kg po, IR 4 Gy single fraction	U251 glioblastoma xenograft	Tumour growth delay	PARPi-TMZ-IR delayed tumour regrowth vs. IR alone
Tuli et al. [69]	ABT 888 25 mg/kg, IR 5 Gy single fraction	Pancreatic carcinoma	Tumour growth delay and survival	PARPi plus IR delayed tumour regrowth and prolonged survival

Reviewing these data, there is an indication that the radiosensitising effects of PARPi are enhanced in in vivo models, with several studies showing radiosensitising effects that exceed those predicted by in vitro data. This is unlikely to be explained by radiotherapy fractionation effects alone, since several of the studies used large single fraction radiotherapy doses similar to those used in vitro. The enhanced effects observed in vivo may be at least partly explained by effects of PARPi on the tumour vasculature, which may in turn be attributed to the structural similarities of many PARPi to nicotinamide , which is a potent vasodilator. Vasodilatory effects of PARPi on tumour blood vessels might alleviate tumour hypoxia whilst simultaneously increasing drug delivery and enhancing radiosensitisation [58, 70]. As yet, the clinical relevance and therapeutic potential of these effects remain unproven.

The normal tissue toxicity implications of a PARPi radiosensitisation strategy have not been extensively investigated, partly because few animal models yield clinically meaningful radiation toxicity data. However, several mechanistic arguments predict at least a degree of tumour specificity, as described below. Likely toxicities will of course depend upon the tumour site irradiated. As single agents, PARP inhibitors have been shown to have highly favourable toxicity profiles [53], so toxicities outwith the irradiated field would be unexpected, unless concomitant chemotherapy was also incorporated into the treatment regimen.

Only gold members can continue reading. Log In or Register to continue