Poly(ADP-ribose) polymerase inhibitor therapeutic effect: are we just scratching the surface?

Abstract

Poly(ADP-ribose) polymerase (PARP) research has come a long way since the discovery of the enzyme 50 years ago. Since the development of first-generation PARP inhibitors (PARPi), numerous clinical trials have been performed to validate their safety and efficacy, bringing us to the stage at which a PARPi is now a valuable treatment option for patients with ovarian cancer. Nevertheless, the exact molecular mechanism of the PARPi anti-tumor effect is under debate and PARPi are not specific for a single enzyme. Moreover, the anti-inflammatory activity of PARPi in preclinical experiments has not been explored much so far. Thus, further basic and preclinical research is needed to advance the use of PARPi in the treatment of tumors and potentially other inflammation-associated diseases.

Keywords:

• ADP-ribosyltransferases diphtheria toxin-like
• poly(ADP-ribose) polymerase inhibitors
• poly-ADP-ribose
• synthetic lethality
• tumor

1. Anti-tumor effect of poly(ADP-ribose) polymerase inhibitors: current dogma

The discovery of the founding member of the poly(ADP-ribose) polymerase (PARP) family in mammalian cells and of the abundant, yet simple polymer structure made out of ADP-ribose repeats (poly(ADP-ribose), PAR) in the early 1960s [1] had a large impact on research and medicine that could hardly be foreseen. PARPs catalyze the post-translational modification of proteins by transferring negatively charged ADP-ribose moieties from nicotinamide-adenine-dinucleotide (NAD⁺) to amino acid residues, or to an existing ADP-ribose group, forming PAR chains. Such protein modification, also called PARylation, alters protein–protein and protein–DNA interactions. In humans, the PARP family consists of 18 enzymes with crucial roles in multiple cellular processes [2]. Since recent studies have shown that some PARPs are in fact mono-ADP-ribosyltransferases, catalyzing MARylation, a more appropriate nomenclature for these enzymes based on their structural homology to bacterial enzymes has been proposed [3]. Members of the group of membrane-bound/extracellular enzymes are referred to as ADP-ribosyltransferases cholera toxin-like, whereas intracellular enzymes are called ADP-ribosyltransferases diphtheria toxin-like (ARTDs).

ARTD1/PARP1, the best-studied intracellular enzyme, has proven to be an excellent therapeutic target for treating cancer owing to its pivotal role in the cellular response to genotoxic stress. Since the discovery of the low-potency PARP inhibitor (PARPi) 3-aminobenzamide, the development of PARPi has greatly progressed [4]. Yet, most of the PARPi in development continue to mimic the nicotinamide moiety of NAD⁺, blocking the binding of NAD⁺ to the enzyme. Thus, even the newest generation of PARPi (e.g., olaparib, veliparib, rucaparib and MRL-45696) not only inhibit ARTD1/PARP1, but also other ARTDs/PARPs and potentially also other NAD⁺-consuming proteins, such as sirtuins [5]. Some PARPi have been reported to not only inhibit PAR formation, but also to trap enzymes such as ARTD1/PARP1 to the DNA. A large number of PARPi have been characterized in preclinical and clinical studies [6]. PARPi are interesting drugs because they are easily delivered, potent inhibitors of PAR formation and are associated with only few mild-to-moderate side effects like fatigue, vomiting and mild nausea, with rare incidences of more serious symptoms such as temporary cognitive deficits, myelosuppression or anemia.

Two conceptually unrelated anticancer strategies for PARPi have been suggested: either as single agent or in combination with either chemotherapy (e.g., with DNA-methylating agents or topoisomerase I inhibitors) or radiotherapy [4]. The rationale for using PARPi in combination therapies is to prevent the action of ARTD1/PARP1 in the genotoxic stress response induced during chemotherapy or radiation. In monotherapy, PARPi have been proposed to induce synthetic lethality through inhibiting base excision repair (BER) and/or trapping of ARTD1/PARP1 at the DNA, both causing an increase in DNA SSBs, which are converted during replication to irreparable toxic DSBs in homologous recombination (HR) repair-defective cancer cells (i.e., BRCA-deficient cells), a feature often referred to as ‘BRCAness’ [7]. Because sustained treatment with PARPi may potentially provoke genomic instability, some concerns have been raised about the clinical use of PARPi. However, recent data show that prolonged PARPi treatment does not induce genomic instability, at least in mice [8]. The combination of therapeutic efficacy with minimal toxicity has recently led to the approval of the PARPi olaparib by the US FDA and the European Commission for the treatment of advanced ovarian cancer in patients with BRCA mutations. The approval of olaparib together with its accompanying diagnostic test for treating BRCA-deficient ovarian cancers is a milestone in personalized cancer therapy. However, after almost a decade of avid investigation and high expectations, enthusiasm has been tempered by disappointing outcomes of other clinical trials, in which ‘BRCAness’ could not predict favorable responses [9]. The acquisition of secondary resistance in initially responsive patients [10], and the lack of standardized biomarkers to identify tumor sensitivity are serious challenges to the further clinical advance of PARPi. Moreover, there are a number of recent reports challenging the model of how PARPi induces synthetic lethality [11]. An extensive body of work has now revealed the potential of PARPi in cancer therapy beyond the presence of BRCA mutations, indicating that there may be additional determinants of PARPi sensitivity. An emerging feeling is that the BRCA-centered view might have missed the contribution of DNA repair-independent mechanisms to the outcome of PARPi therapy, and that PARPi-induced synthetic lethality might thus also involve HR-unrelated proteins.

2. Non-HR targets of PARPi anti-tumor activity

It is becoming increasingly apparent that most PARPi affect cellular processes that influence aspects of tumor development, progression and treatment response, including several hallmarks of cancer, that have not yet been considered as primary targets of PARPi anti-tumor action. For example, recent findings indicate that ARTD1/PARP1 positively regulates the activity of key components of hypoxic adaptation, angiogenesis and of the inflammatory response, namely hypoxia-inducible factor 1α (HIF-1α) and NF-κB [12]. As a consequence, PARPi have been shown to reduce HIF-1α- and/or NF-κB-driven tumor proliferation, angiogenesis and metastatic ability in different preclinical models. ARTD1/PARP1 also controls the expression of molecules involved in the epithelial–mesenchymal transition, such as vimentin and Snail1. Similarly, PARPi hamper invasion, intravasation and metastasis of prostate cancer cells that overexpress fusion proteins of ETS transcription factors, by disrupting the ETS-mediated transcription of metastasis-related genes, such as EZH2.

Furthermore, a striking relationship between ARTD1/PARP1 and HER2 inhibition has recently been observed in breast cancer cells overexpressing HER2, in which PARPi enhanced the anti-tumor effects of the anti-HER2 antibody trastuzumab [13]. Combinatorial strategies with PARPi and other targeted agents also include inhibitors of heat-shock protein 90 or histone deacetylases, which are expected to reduce the expression of proteins by mechanisms involving proteasomal degradation and transcriptional regulation. For example, co-inhibition of HDAC sensitizes triple-negative breast cancer cells to PARPi and synergistically reduces the viability of prostate cancer cells. However, whether and to what extent PARPi also affect the epigenome and transgeneration inheritance remains to be investigated.

Together, these findings unveil an as yet poorly considered scenario for PARPi that entails the possibility to interfere with tumor-promoting inflammation, cancer metabolism and senescence, angiogenesis, and metastasis.

3. Non-tumor-related inflammation-associated diseases: a new therapeutic application for PARPi?

The generation of reactive oxygen and nitrogen species by non-tumor conditions, such as myocardial infarction, heart transplantation and autoimmune β-cell destruction associated with diabetes mellitus, could lead to ARTD1/PARP1 overactivation and subsequent NAD⁺ depletion, driving cells to necrosis through energy deprivation, thus leading to an inflammatory condition in these and other diseases. The relationship between PARP inhibition and non-tumor-associated inflammation has been widely investigated and has provided strong evidence for a protective effect of PARPi towards a number of inflammatory conditions [12]. The beneficial therapeutic effect of PARPi has been described for disease models such as heart failure, cardiomyopathies, circulatory shock, cardiovascular diseases, atherosclerosis and post-ischemic brain damage [12]. A favorable effect of PARPi has also been described for models of diabetes and diabetic complications, and can be attributed to ARTD1/PARP1 as a causative factor, as ARTD1/PARP1-deficient mice are protected from streptozotocin-induced diabetes. A protective effect of PARPi has also been observed in models of asthma, airway inflammation and other lung diseases. Only a few studies have been conducted in humans: an in vivo study that tested the anti-inflammatory effects of flavonoids that act as PARPi revealed that they attenuate cytokine release in blood from patients with chronic obstructive pulmonary disease or type 2 diabetes. The mechanisms underlying this PARPi effect are not solely attributable to the above-described paradigm ‘PARP hyperactivation – NAD⁺ consumption – necrosis’, but also imply a role of PARPi in regulating inflammatory factors such as NF-κB and the pro-inflammatory and anti-apoptotic genes it drives.

4. Pan-PARPi versus single enzyme-specific PARPi

None of the currently available PARPi specifically inhibits only a single ARTD/PARP member, although some progress towards this goal has been made. The currently available PARPi are also being tested for tumors regulated by other ARTD/PARP family members than ARTD1/PARP1 (e.g., ARTD8/PARP14 in multiple myeloma or prostate cancer). We have recently shown that PARPi treatment prevents and cures Helicobacter-induced, inflammation-induced gastric pre-neoplasia by impairing Helicobacter-specific T-cell priming and Th1 polarization in the gut-draining mesenteric lymph nodes [14]. Interestingly, the onset and severity of pre-neoplasia formation was not any different in ARTD1- or ARTD2-deficient mice, suggesting that other ARTDs/PARPs are involved in the inflammation-induced neoplasia.

The extreme versatility of the ARTD/PARP family implicates the existence of functions independent of the genotoxic response, which may provide novel therapeutic opportunities, although they remain immature for clinical exploitation at present. Therefore, blocking ARTDs/PARPs more specifically may offer a unique opportunity to tackle tumor- and non-tumor-related clinical pathologies. The crystal structure of the catalytic domain of human ARTD2/PARP2 has defined its interaction with the PARPi ABT-888 and has opened new perspectives into the development of selective inhibitors. Similarly, the recent determination of the crystal structure of ARTD3/PARP3 has revealed that the loops surrounding its active site are different from the ones in ARTD1/PARP1, thus providing the basis for developing more specific inhibitors against this enzyme. New inhibitors are also being developed against more recently described ARTD/PARP family members, such as tankyrases, which has provided new knowledge and additional facets on PARylation-related tumor biology [15]. However, studies to draw a solid comparison of the consequences of pan-PARP versus selective PARP targeting do currently not exist. Under certain medical conditions, a global inhibition of ARTDs/PARPs may even be more beneficial, as such an inhibitor prohibits functional compensation between different ARTDs/PARPs. On the other hand, a selective inhibition may be more beneficial under conditions in which only a particular ARTD/PARP form contributes to a disease.

5. Expert opinion

PARPi are an exceptional example of a synthetic lethal therapeutic strategy successfully developed from bench to bedside. In addition to the antitumor action of PARPi, the evidence summarized above supports a high potential for PARPi to be therapeutically beneficial in diseases mediated by the reactive oxygen species/NF-κB pathway, such as various inflammatory diseases. While the clinical benefits of the existing PARPi are currently being further explored, additional new areas of research are opening up at the preclinical front, which should eventually lead to the discovery of new and more effective PARPi, but also to additional therapeutic applications such as its use in inflammation-associated diseases. Finally, while mostly focusing on the ARTDs/PARPs that catalyze PARylation, one should also more strongly consider approaches aimed at inhibiting MARylation and regulating the reversal of protein ADP-ribosylation (e.g., through activation/inactivation of poly-ADP-ribosyl glycohydrolase).

Declaration of interest

Research in the MOH laboratory on PARPi is funded in whole or in part with funds from the Swiss National Science Foundation, Zürcher Krebsliga and the University of Zurich Research Priority Program (URPP) ‘Translational Cancer Research’. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.