Tuning drug binding

Inhibitors of poly(ADP-ribose) polymerase 1 (PARP-1) are used to treat ovarian and breast cancer (1). PARP-1 is activated on binding single-stranded DNA breaks (SSBs) and released from DNA by automodification (auto-PARylation) (2, 3). By preventing auto-PARylation, PARP inhibitors induce PARP-1 trapping on DNA (4). PARP inhibitors have similar in vitro potency in reducing PARP-1 activity, but their ability to induce cell killing differs substantially, most likely owing to differential potency in trapping PARP-1 on DNA (5, 6). In addition to catalytic inhibition, PARP-1 trapping was proposed to rely on reverse allosteric changes within PARP-1 from the catalytic domain to the DNA binding domain (5–7). On page 46 of this issue, Zandarashvili et al. (8) dissect the allosteric effects of different PARP inhibitors and show how these can be harnessed for targeted design of new protrapping or pro-release PARP inhibitors, which may have greater efficacy and versatile application potential.

A particularly enigmatic form of allosteric regulation lies at the heart of PARP-1 activation (2). A cascade of structural changes is triggered when a single PARP-1 molecule encounters an SSB: The flexibly connected PARP-1 domains no longer behave like beads on a string but instead engage each other in communicating the DNA damage signal from the amino-terminal zinc fingers toward the Trp-Gly-Arg (WGR) domain and the carboxyl-terminal domain, which consists of the helical domain (HD) and the catalytic adenosine diphosphate (ADP)–ribosyltransferase domain. Partial unfolding of the HD relieves its autoinhibitory effects and allows the NAD⁺ (oxidized form of nicotinamide adenine dinucleotide) cofactor to access the catalytic site. As a result, the catalytic domain is allosterically activated (2, 7, 9).

PARP inhibitors compete with NAD⁺ for the PARP-1 active site. Clinically relevant PARP inhibitors were proposed to induce PARP-1 trapping through “reverse allostery” (5), whereby allosteric changes spread in a reverse direction from the catalytic domain onto the HD, WGR domain, and DNA binding domain (2, 7). So far, reverse allostery has been demonstrated for the nonclinical PARP inhibitor benzamide adenine dinucleotide (BAD), a nonhydrolyzable NAD⁺ analog (7). Zandarashvili et al. used hydrogen-deuterium exchange mass spectrometry to probe the effect of clinically relevant PARP inhibitors on allosteric changes within different PARP-1 domains. Furthermore, to assess how allosteric changes interlink with PARP-1 DNA trapping, the authors determined PARP-1 SSB affinity and dissociation rates in the presence of PARP inhibitors in vitro.

In accordance with previous results (7), another NAD⁺ analog, EB-47, induced pronounced allosteric changes and strong PARP-1 DNA trapping in vitro, consistent with the reverse allostery model (5, 7). Zandarashvili et al. classified this NAD⁺ analog as a type I pro-retention inhibitor. Clinically relevant PARP inhibitors induced either no allosteric changes or opposite allosteric changes from those observed for the NAD⁺ analog, which calls into question reverse allostery as the mechanism of PARP-1 trapping by clinical PARP inhibitors. The PARP inhibitors talazoparib and olaparib caused weak or no allosteric changes, destabilization of the HD, and weak PARP-1 DNA trapping in vitro. These were classified as type II mild or no pro-retention inhibitors. Other PARP inhibitors—veliparib, niraparib, and rucaparib—induced pronounced allosteric changes that involved stabilization of the HD and increased PARP-1 release from SSBs. Thus, these were classified as type III pro-release inhibitors. This is concordant with weak cellular PARP-1 trapping reported for these inhibitors (5). PARP-1 trapping is favorable in cytotoxic anticancer treatment, where pro-retention PARP inhibitors induce replication stress, mitotic catastrophe, and cancer cell death (1). PARP inhibitors are also indicated for stroke and heart failure, where PARP-1 overactivation causes energy depletion and cell death (10). In such conditions, pro-release PARP inhibitors would be the drugs of choice because they do not induce cell death (see the figure).

Cellular PARP-1 trapping by clinical PARP inhibitors is typically assessed by measuring the shift in PARP-1 distribution in the nucleus from being soluble to chromatin-bound (4–6). Although the in vitro measurements of Zandarashvili et al. classify talazoparib as a mild pro-retention PARP inhibitor, cellular measurements reveal its strong pro-retention properties(6). Adding further to the complexity of biochemical and physiological system comparisons, ovarian cancer patient mutation Arg⁵⁹¹Cys in the PARP-1 WGR domain was found to reduce olaparib-mediated PARP-1 trapping in cells (11). This residue is critical for allosteric pro-retention effects of PARP inhibitors, suggesting that allosteric regulation could also play a role in olaparib-mediated PARP-1 trapping in cells.

Allosteric modulation.

Poly(ADP-ribose) polymerase 1 (PARP-1) is allosterically activated upon binding to single-stranded DNA breaks. Pro-retention and pro-release PARP inhibitors differentially harness PARP-1 allostery to induce PARP-1 trapping or release from DNA, which can be applied in the clinic depending on whether cytotoxicity or cytoprotection is desirable.

The apparent disparity between PARP-1 trapping measurements in vitro and in cells suggests that additional factors determine the efficiency of cellular PARP-1 trapping. Indeed, a myriad of chromatin factors, DNA damage response proteins, and RNAs are known to interact with PARP-1. PARP-1–interacting protein histone PARylation factor 1 (HPF1) recently emerged as a fascinating example of a PARP-1 modulator that directs catalytic specificity toward serine ADP-ribosylation by forming a joint active site with PARP-1 or PARP-2. HPF1 likely also modulates cellular PARP-1 trapping by PARP inhibitors (12). Moreover, the dynamic nature of PARP-1 enables its activation on binding not only SSBs but also unligated Okazaki fragments during DNA replication, illustrating its engagement in different pathways: DNA repair and replication fork stability (13). Therefore, its allosteric multidomain cascade provides a rapid switch for activation as well as pathway-specific regulation (2).

On the basis of the structural insights into allosteric changes induced by type I pro-retention inhibitors, Zandarashvili et al. converted the type III inhibitor veliparib into a new-generation type I veliparib variant, UKTT15, which induces type I proretention allosteric effects and shows increased cytotoxicity compared with veliparib. Among all clinically relevant PARP inhibitors, veliparib shows the highest selectivity for PARP-1 and PARP-2 inhibition (14). Enhancing PARP-1 trapping while preserving high selectivity is an example of intelligent drug design that can be extended to other available PARP inhibitors for customized application in different disease settings. However, this strategy may increase cytotoxicity to normal cells, which could increase side effects (1, 15). The potential of targeting allosteric changes for drug discovery is only just beginning to be harnessed.