Synthesis and Evaluation of a Mitochondria-Targeting Poly(ADP-ribose) Polymerase-1 Inhibitor

Abstract

The poly(ADP-ribose) polymerase (PARP) family of enzymes plays a crucial role in cellular and molecular processes including DNA damage detection and repair, and transcription; indeed, PARP inhibitors are under clinical evaluation as chemotherapeutic adjuncts given their capacity to impede genomic DNA repair in tumor cells. Conversely, overactivation of PARP can lead to NAD⁺ depletion, mitochondrial energy failure, and cell death. Since PARP activation facilitates genomic but impedes mitochondrial DNA repair, non-selective PARP inhibitors are likely to have opposing effects in these cellular compartments. Herein, we describe the synthesis and evaluation of the mitochondria-targeting PARP inhibitor, XJB-veliparib. Attachment of the hemigramicidin S pentapeptide isostere for mitochondrial targeting using a flexible linker at the primary amide site of veliparib did not disrupt PARP affinity or inhibition. XJB-veliparib was effective at low nanomolar concentrations (10–100 nM) and more potent than veliparib in protection from oxygen glucose deprivation (OGD) in primary cortical neurons. Both XJB-veliparib and veliparib (10 nM) preserved mitochondrial NAD⁺ after OGD; however, only XJB-veliparib prevented release of NAD⁺ into cytosol. XJB-veliparib (10 nM) appeared to inhibit poly(ADP-ribose) polymer formation in mitochondria and preserve mitochondrial cytoarchitecture after OGD in primary cortical neurons. After 10 nM exposure, XJB-veliparib was detected by LC-MS in mitochondria-but not nuclear-enriched fractions in neurons and was observed in mitoplasts stripped of the outer mitochondrial membrane obtained from HT22 cells. XJB-veliparib was also effective at preventing glutamate-induced HT22 cell death at micromolar concentrations. Importantly, in HT22 cells exposed to H₂O₂ to produce DNA damage, XJB-veliparib (10 µM) had no effect on nuclear DNA repair, in contrast to veliparib (10 µM) where DNA repair was retarded. XJB-veliparib and analogous mitochondria-targeting PARP inhibitors warrant further evaluation in vitro and in vivo, particularly in conditions where PARP overactivation leads to mitochondrial energy failure and maintenance of genomic DNA integrity is desirable, e.g. ischemia, oxidative stress, and radiation exposure.

Grahical Abstract

Proposed mechanisms of action for mitochondria-targeting PARP inhibitors. Abbreviations: AIF, apoptosis-inducing factor; ARH3, ADP-ribosylhydrolase 3; EXOG, exo/endonuclease G; mtDNA, mitochondrial DNA; PARG, poly(ADP-ribose) glycohydrolase; PARP, poly(ADP-ribose) polymerase; Polβ, DNA polymerase β; Polγ, DNA polymerase γ; XRRC1, X-ray repair cross-complementing protein 1. 127×79mm (300 × 300 DPI)

INTRODUCTION

Poly(ADP-ribosyl)ation (PARylation) plays a crucial role in cellular and molecular processes including DNA damage detection and repair, transcription, and the maintenance of genomic integrity.¹ The currently identified 17 members of the poly(ADP-ribose) polymerase (PARP) family induce the cleavage of NAD⁺ into nicotinamide and ADP-ribose moieties and mediate their polymerization on target proteins, with links to cellular redox homeostasis, inflammatory, and metabolic networks.^{2, 3} PARP1 binds to nuclear DNA single-strand break (SSBs) sites and recruits repair proteins to the DNA, subsequently dissociating itself from the SSB. The most potent small-molecule inhibitors, rather than just off-setting PARP1’s enzymatic activity, trap it at the site of DNA damage and stabilize PARP–DNA complexes, ultimately causing DNA double-strand breaks that require more complex repair mechanisms.⁴

DNA damage induced by irradiation or oxidative stress leads to overactivation of PARP1 and induces depletion of cellular NAD⁺ and ATP levels, leading to cell dysfunction and necrotic cell death. The primary location of NAD⁺ is in mitochondria, where it is utilized for oxidative phosphorylation. Furthermore, mitochondrial DNA (mtDNA) is constantly being exposed to damaging species such as reactive oxygen and nitrogen species and is efficiently repaired through at least a subset of the mechanisms involved in nuclear DNA repair, including regulation by PARP1.⁵ However, opposing roles for PARP1 have been observed in genomic versus mtDNA repair with PARP1 facilitating DNA repair in the nucleus and impeding DNA repair in mitochondria.⁶ Overactivation of PARP1 and NAD⁺ depletion has been linked to the pathogenesis of central nervous system (CNS) disorders, including ischemia, traumatic brain injury (TBI), neuroinflammation, and neurodegenerative diseases such as Alzheimer’s and Parkinson’s diseases, which have a pronounced mitochondrial component.^{1, 7} Preserving NAD⁺ levels may reverse mitochondrial dysfunction and prevent energy failure.⁸ If suitable PARP inhibitors could be targeted exclusively to the mitochondria, they might reduce NAD⁺ depletion-associated energy failure while allowing repair of damaged DNA in the nucleus and facilitating mtDNA repair.⁶

We have previously reported evidence for the localization of PARP in mitochondria and mitochondria-specific PARylation by immunocytochemistry, electron microscopy, and by western blot analysis of purified mitochondrial fractions.^{9, 10} Furthermore, we demonstrated that the inhibition of mtPARP preserved mitochondrial transmembrane potential and reduced neuronal cell death triggered by oxidative stress or excitotoxicity.⁹ However, to the best of our knowledge, the hypothesis that mitochondria-targeted PARP-inhibitors would be therapeutically effective in neurodegeneration and other CNS and non-CNS conditions associated with energy failure, and a desire to maintain genomic integrity, has not yet been addressed.

While there are broader therapeutic implications for synthetic PARP modulators, research and development of poly(ADP-ribose) polymerase-1 (PARP1) inhibitors in cancer research has intensified since the regulatory approval of olaparib and rucaparib for patients with BRCA1/2 mutant ovarian cancer, and niraparib for recurrent gynecologic cancers (Figure 1).^11–13 Veliparib (1, 2-[(S)-2-methylpyrrolidin-2-yl]-1H-benzimidazole-4-carboxamide, ABT-888) is a promising PARP1 inhibitor that has entered clinical phase I/II/III trials for several forms of cancer, including breast cancer and solid tumor neoplasm, either as a single agent or as a combination with chemotherapeutics. A phase III trial is currently ongoing to assess the efficacy, safety and tolerability of veliparib in patients with previously untreated ovarian cancer.¹⁴ The (S)- and (R)-enantiomers of veliparib have been shown to have identical biological activity,¹⁵ and relative to other FDA approved PARP1 inhibitors veliparib seems to be the least potent in terms of trapping PARP at the site of DNA damage.¹⁶

Figure 1.

PARP inhibitors in clinical use or in current trials.

The XJB pentapeptide isostere is a mitochondrial targeting unit^17–20 derived from the cyclopeptide antibiotic gramicidin S (GS). The XJB moiety has been used successfully to deliver various payloads to mitochondria, including nitroxides,¹⁹ a nitric oxide synthase (NOS) antagonist (AMT),²¹ and a derivative of the natural anticancer product β-lapachone.²² In line with these previous studies, we used XJB to deliver the payload veliparib to mitochondria. Here we show that XJB-veliparib is enriched in mitochondria, and that the compound potently inhibits PARP-mediated cell death in multiple models of cytotoxicity. XJB-veliparib is particularly effective in preventing cell death from hypoxia-ischemia in neurons, warranting further evaluation and development of this class of mitochondria-targeting PARP inhibitors for neurological diseases.

RESULTS AND DISCUSSION

Mitochondria-targeting Veliparib

The chemical synthesis of the mitochondria-targeting XJB-veliparib hybrid molecule is shown in Scheme 1 and fully described in the Methods. Since (S)- and (R)-enantiomers of veliparib have shown identical biological activity,¹⁵ we utilized racemic veliparib for the preparation of the XJB-linked hybrid molecule. Furthermore, we hypothesized that the attachment of a flexible linker at the primary amide would be unlikely to disrupt veliparib’s high PARP affinity since, while energetically not as favored, the amide could adopt a cis-configuration and engage Gly429 in PARP with two hydrogen bonds, or, in the trans-amide configuration, a hydrophopic alkyl linker chain could reinforce the single remaining amide carbonyl hydrogen bond to Gly429.²³ We selected the XJB pentapeptide isostere as the mitochondrial targeting sequence (Figure 2).^17–20 The alkene peptide isostere segment in XJB is a surrogate of the leucyl-D-phenylalanine dipeptide in the bacterial membrane-associating antibiotic gramicidin S (GS), and its side chain-protected ornithylvalylproline tripeptide subunit is taken directly from GS. The D-Phe-Pro sequence is based on the reverse turn inducing sequence of GS and folds into a type II′ β-turn structure that buries several polar amide groups inside the molecule and thus may facilitate membrane transport.^24–27 This moiety has previously been used in combination with a nitroxide payload to generate XJB-5–131, a reactive oxygen species (ROS) scavenger that validated the targeting design and was found to be ca. 600-fold enriched in mitochondria over the cytosol.¹⁹ XJB-5–131 has shown in vivo efficacy in rodent models of Huntington’s disease (HD),^28–30 TBI,³¹ ischemia-reperfusion injury,^{32, 33} and hemorrhagic shock.³⁴ In the radiation protector XJB-AMT, a nitric oxide synthase (NOS) antagonist (AMT) was conjugated to the targeting sequence, with the goal to counteract the activation of mitochondrial NOS by ionizing radiation, which can lead to inhibition of the respiratory chain, a burst of superoxide and peroxynitrite, and cellular damage.²¹ XJB-Lapachone introduced a derivative of the natural anticancer compound β-lapachone into mitochondria and triggered extensive cellular vacuolization and autophagy, as well as stimulating ROS generation in mitochondria.²²

Scheme 1.

Synthesis of XJB-Veliparib (refer to Methods for chemical transformations).

Figure 2.

Structures of mitochondrially targeted 4-amino-TEMPO (XJB-5–131), lapachone (XJB-lapachone) and veliparib (XJB-veliparib). The therapeutically active payload is in shown blue, the linker region is in red, and the XJB mitochondrial targeting moiety is in green.

XJB-veliparib and Veliparib PARP1 Enzyme Inhibition ex vivo

To determine whether the linkage to the mitochondria-targeting moiety on 15 affected PARP1 inhibition, we tested the capacity of veliparib and XJB-veliparib to inhibit active PARP1 enzyme (10 U) ex vivo. PARP1 inhibition was similar between untargeted veliparib and the XJB-veliparib conjugate 15 (Figure 3; experiments performed in triplicate), and is consistent with the reported K_i of veliparib of 5.2 nmol/L.³⁵

Figure 3.

Capacity of XJB-veliparib (15) to inhibit PARP1 enzyme. Varying concentrations of XJB-veliparib or veliparib (0–100 nM) were applied to active PARP1 enzyme (10 U) ex vivo (performed in triplicate).

XJB-veliparib and Veliparib Cytotoxicity in Primary Cortical Neurons

To investigate the biological properties and potential cytotoxicity of XJB-veliparib, we exposed rat primary cortical neurons to varying concentrations of XJB-veliparib and veliparib. Cytotoxicity was assessed by lactate dehydrogenase (LDH) release from dying neurons at 24 h. Both XJB-veliparib and veliparib showed a concentration-dependent cytotoxicity profile; however, XJB-veliparib was significantly less toxic compared with unconjugated veliparib (Figure 4; n = 6/group; mean ± standard deviation [SD]; *P < 0.05). Neurotoxicity defined as >10% cell death was seen with veliparib at 1 µM concentration, vs. a 10 µM concentration required for XJB-veliparib. Significant cytotoxicity has previously been reported when leukemia cells are exposed to micromolar concentrations of veliparib.³⁶ Cytotoxicity produced by PARP inhibitors in clinical use, including veliparib, have been associated with inability to efficiently repair DNA damage and genomic instability.³⁷ However, recent evidence further points to trapping of PARP1 enzyme itself in double-strand DNA (dsDNA) breaks by PARP1 inhibitors.¹⁶ Trapped PARP-dsDNA complexes retain catalytic activity and enhance genotoxicity and lethality of chemotherapeutic agents. Relevant here, veliparib concentrations of > 100 µM are tumoricidal in most cancer cell lines.³⁸ Importantly, mitochondria-targeting PARP inhibitors would avoid formation of trapped PARP-dsDNA complexes, and accordingly may have lower toxicity and a therapeutic advantage where prevention of cell death is desired.

Figure 4.

Cytotoxicity studies. Rat primary cortical neurons (DIV 10) were exposed to varying concentrations of XJB-veliparib or veliparib (0–100 µM) for 24 h. Cytotoxicity was determined by LDH release measured at 24 h (n = 6/group; mean ± SD).

Effect of XJB-veliparib and Veliparib after Oxygen-glucose Deprivation in Primary Cortical Neurons

To determine whether mitochondria-targeting XJB-veliparib can promote neuronal survival in ischemia-like conditions in vitro, we subjected primary cortical neurons at 12 DIV to oxygen-glucose deprivation (OGD).³³ Cultured neurons were exposed to a hypoxic and glucose-depleted environment for 2 h, followed by normal culture conditions for 24 h to mimic ischemia/reperfusion injury. As shown in Figure 5a, treatment with XJB-veliparib significantly attenuated OGD-induced cell death at low nanomolar concentrations. Specifically, 10 nM XJB-veliparib reduced cell death by 65% (LDH release 12.1 ± 3.1% vs. 34.2 ± 4.0%, 10 nM XJB-veliparib vs. vehicle; n = 30/group; mean ± SD; *P < 0.05).

Figure 5.

Ischemia-reperfusion injury (2 h of OGD) in rat primary cortical neurons treated with XJB-veliparib or veliparib.

(a) Cell death was determined by LDH release at 24 h. Drug concentrations of 1 nM −1 µM were protective against OGD, and XJB-veliparib was more potent than veliparib at concentrations of 10–100 nM (*P < 0.05 vs. veliparib; n = 12–30/group; mean ± SD).

(b) Mitochondrial and cytosolic NAD⁺ concentration in primary cortical neurons 24 h after OGD. XJB-veliparib or veliparib (each 10 nM) administered before OGD preserved mitochondrial NAD⁺ concentration (*P < 0.05 vs. veliparib; **P < 0.05 vs. vehicle; ***P < 0.05 vs. control; n = 6/group; mean ± SD).

(c) STED images showing mitochondrial morphology and PAR polymers 24 h after OGD. Mitochondria were labelled with anti-TOMM20 antibody (green), PAR polymers were labelled with anti-PAR antibody (red), and nuclei were labelled with DRAQ5 (blue). XJB-veliparib (10 nM) was administered before OGD and immunohistochemistry was performed 24 h after OGD. PAR co-localization in mitochondria after OGD is shown in higher magnification.

(d) Mitochondrial enrichment of XJB-veliparib. Rat primary cortical neurons (DIV 10) were exposed to 10 nM XJB-veliparib for 24 h. Cells were then harvested and mitochondrial and nuclear-enriched subfractions were obtained. XJB-veliparib was detected in mitochondria-but not nuclear-enriched fractions (3 independent experiments). Western blot showing mitochondrial and nuclear enrichment using antibodies against COX Vα and histone H3, respectively.

Treatment with concentrations >100 nM of either veliparib or XJB-veliparib conjugate provided no additional protection. Both XJB-veliparib and non-targeting veliparib appear more protective against OGD compared with other published PARP1 inhibitors, where protection is observed in micromolar ranges.^{39, 40} In vivo, PARP1 inhibition is highly effective at reducing neuronal death caused by ischemia-reperfusion injury.⁴¹

To determine whether XJB-veliparib and veliparib related neuroprotection was associated with preservation of intracellular NAD⁺, NAD⁺ concentrations in mitochondria-and cytosol-enriched fractions were measured 24 h after OGD in primary cortical neurons. Ten nM XJB-veliparib or non-targeting veliparib preserved mitochondrial NAD⁺ stores after OGD (Fig. 5b; n = 6/group; mean ± SD; *P < 0.05 vs. vehicle). However, XJB-veliparib, but not veliparib, also prevented efflux of NAD⁺ from the mitochondria to the cytosol after OGD (**P < 0.05 vs. vehicle and veliparib). These results are consistent with preservation of mitochondrial NAD⁺ as one of the mechanisms behind PARP-mediated neuroprotection after OGD.

To evaluate the impact of XJB-veliparib on mitochondrial structure after OGD, we used stimulation emission depletion (STED) microscopy and immunohistochemistry. Primary rat cortical neurons were treated with 10 nM XJB-veliparib or vehicle and then exposed to OGD for 2 h. Vehicle treated neurons showed increased PARP activation at 1 h, assessed by immunofluorescence staining with anti-PAR antibody, and swollen, circular mitochondria (labelled with translocase of outer mitochondrial membrane (OMM) 20 [TOMM20]), consistent with fission vs. coiling and compaction compared with control (no ischemia) neurons (Figure 5c; performed in triplicate). In contrast, PAR immunoreactivity was reduced in neurons treated with XJB-veliparib (vs. vehicle) and mitochondrial architecture appeared partially preserved. The relative increase in PAR immunoreactivity observed in mitochondria vs. nuclei in neurons after OGD may be explainable by more effective PAR metabolism by poly(ADP-ribose) glycohydrolase in cell nuclei vs. ADP-ribosylhydrolase 3 in mitochondria.⁴²

To verify mitochondrial enrichment at a therapeutic dose, rat primary cortical neurons were exposed to 10 nM XJB-veliparib for 24 h. Mitochondria- and nuclear-enriched subcellular fractions were obtained as described and verified by western blot using antibodies against the cytochrome oxidase subunit Vα (COX Vα) and histone H3, respectively (Figure 5d).⁹ Liquid chromatography-mass spectrometry (LC-MS) analysis of XJB-veliparib showed a concentration of 40.5 ± 15.8 pmol/10 µg protein in mitochondria-enriched fractions (Figure 5d; mean ± SD; n = 3 independent experiments). XJB-veliparib was undetectable in nuclear-enriched fractions.

Detection of XJB-veliparib in Mitoplasts

To verify that XJB-veliparib detected in mitochondria-enriched fractions shown in Figure 5d was not simply associated with the OMM, we isolated mitoplasts stripped of the OMM obtained from HT22 cells exposed to 10 nM XJB-veliparib for 45 min. Mitochondria were isolated using differential centrifugation following a standardized protocol (ThermoFisher #89874). The OMM was stripped from the inner mitochondrial membrane (IMM)/matrix using digitonin (0.15 – 1.5 mg/mL). The purity of the mitoplast containing fractions was verified by the absence of translocase of the outer mitochondrial membrane 40 (TOM40) and presence of manganese superoxide dismutase (MnSOD) observed on western blot. LC-MS analysis of XJB-veliparib showed a concentration of 62.6 ± 16.3 pM in mitoplast fractions (Figure 6; mean ± SD; n = 8 samples from 3 independent experiments). For comparison, the concentration of XJB-veliparib in whole mitochondria was 85.1 ± 5.3 pM in HT22 cells after a 45 min exposure to 10 nM XJB-veliparib (n = 3).

Figure 6.

XJB-veliparib concentrates in mitoplasts obtained from HT22 cells exposed to 10 nM XJB-veliparib for 45 min. The OMM was digested from IMM and matrix containing mitoplasts using digitonin (0.15 – 1.5 mg/mL). The purity of the mitoplast containing fractions was verified by the absence of TOM40 and presence of MnSOD observed on western blot (mean ± SD; n = 8 samples from 3 independent experiments). For comparison, the concentration of XJB-veliparib in whole mitochondria from HT22 cells exposed to 10 nM XJB-veliparib for 45 min is shown (n = 3). Mp, mitoplast; OM, outer mitochondrial membrane.

Effect of XJB-veliparib and Veliparib after Excitotoxic Injury in Primary Cortical Neurons and Hippocampal Neuronal Cells

Previous studies have shown an important role for nuclear PARP1 activation in inhibiting excitotoxic neuronal death in vitro.⁴³ In addition, PARP1 inhibition in vivo is effective at reducing N-methyl-D-aspartate (NMDA) but not non-NMDA excitotoxicity.⁴⁴ To determine whether XJB-Veliparib was also effective in attenuating excitotoxic cell death in vitro, primary cortical neurons were exposed to 10 µM L-glutamate and 10 µM glycine with varying concentrations of XJB-Veliparib or naked veliparib for 24 h as previously described.⁹ PARP inhibition reduced cell death after glutamate/glycine exposure in a concentration-dependent manner (Fig. 7a; n = 6/group; mean ± SD; *P < 0.05 vs. vehicle). Differing from in vitro ischemia/reperfusion, veliparib was more potent than XJB-veliparib in reducing LDH release. The underlying basis for the differential effectiveness of targeting versus non-targeting PARP inhibitors in models of OGD versus excitotoxicity is beyond the scope of the present study. However, one possible explanation is that since glutamate stimulates DNA repair via activation of transcription factors including cyclic AMP response element binding protein and upregulation of apurinic endonuclease 1,⁴⁵ any impact of non-targeting PARP inhibitors on nuclear DNA repair and formation of trapped PARP-dsDNA complexes may be mitigated.

Figure 7.

Excitiotoxicity in rat primary cortical neurons or HT22 cells treated with XJB-veliparib or veliparib.

(a) Glutamate-glycine excitotoxicity in primary cortical neurons. Neurons were exposed to 10 µM L-glutamate and 10 µM glycine, with 1–100 µM of XJB-veliparib or veliparib for 24 h (*P < 0.05 vs. veliparib, n = 6/group; mean ± SD).

(b) Glutamate excitotoxicity in an immortalized hippocampal neuronal HT22 cell line. XJB-veliparib and veliparib were both effective at inhibiting excitotoxic cell death, with XJB-veliparib slightly more effective at higher doses (*P < 0.05 vs. naked veliparib; n = 5/group; mean ± SD).

In addition to rat primary cortical neuron cultures, we also determined the effect of veliparib and XJB-veliparib in a stable mouse hippocampal neuronal cell line HT22, in which ferroptotic cell death is induced by high concentrations of glutamate.⁴⁶ HT22 cells grown to confluence were exposed to 5 mM glutamate and various concentrations of veliparib or XJB-veliparib with LDH release measured at 24 h. Both veliparib and XJB-veliparib similarly inhibit glutamate-induced ferroptosis in HT22 cells, with XJB-veliparib slightly more effective than veliparib at higher (10 µM) concentrations (Fig. 7b; n = 5/group; mean ± SD; *P < 0.05 vs. vehicle).

Effect of XJB-veliparib and Veliparib on DNA Repair in Hippocampal Neuronal Cells Exposed to H₂O₂

To determine whether XJB-veliparib and/or veliparib impacted DNA repair, we evaluated DNA damage serially in HT22 cells exposed to 200 µM H₂O₂ using the comet tail intensity assay as previously described.⁴⁷ Exposure to H₂O₂ produced DNA damage with evidence of DNA repair over 30 min (Fig. 8). Veliparib, but not XJB-veliparib (both 10 µM), impeded DNA repair (Fig. 8; *P < 0.05 vs. veliparib; **P < 0.05 vs. vehicle; n = 52–60 cells/group; mean ± SEM). These data suggest that mitochondria-targeting PARP inhibitors at a dose that protects HT22 cells from glutamate toxicity (Fig. 7b) do not impact repair of damaged DNA.

Figure 8.

DNA damage and repair in HT22 cells exposed to H₂O₂ treated with XJB-veliparib (10 µM) or veliparib (10 µM). HT22 cells were exposed to 200 µM H₂O₂ for 30 min. Tail moments were determined for each condition using the comet tail intensity assay and expressed as percent DNA damage remaining. Exposure to H₂O₂ produced DNA damage with evidence of DNA repair over 30 min. Veliparib, but not XJB-veliparib, impeded DNA repair (*P < 0.05 vs. veliparib; **P < 0.05 vs. vehicle; n = 52–60 cells/group; mean ± SEM).

Role for Mitochondria-targeting PARP Inhibitors

An undisputed role for mitochondrial PARP, and in particular mitochondrial PARP1, in health and disease is controversial;⁴⁸ however, at least a partial role for mitochondrial PARP in mtDNA repair and cellular metabolism is becoming more apparent.⁵ Several early studies have shown the presence of ADP-ribosylation in mitochondria from testes,⁴⁹ liver,^{50, 51} and brain.⁵² More recently, inhibition of mitochondrial PARP1 using genetic approaches was shown to increase the efficiency of mtDNA base excision repair via interaction with exo/endonuclease G and DNA polymerase γ;⁶ as opposed to inhibition of nuclear PARP1 which decreases the efficiency of genomic DNA repair.⁵³ Here we show that mitochondria-targeting XJB-veliparib has no effect on nuclear DNA repair after H₂O₂ exposure in neuronal cells, in contrast to non-targeting veliparib (Fig. 8).

PARP1 inhibition has been shown to increase mitochondrial metabolism via silent mating type information regulation 2 homolog 1 (SIRT1),⁵⁴ and directly preserve mitochondrial function and prevent cell death via maintenance of NAD⁺ stores.⁵⁵ Here we show that mitochondria-targeting XJB-veliparib is very effective at preserving mitochondrial NAD⁺ levels and preventing efflux of NAD⁺ into cytosol in neurons after OGD (Fig. 5b). We have previously reported PARylation of the electron transport chain proteins F₁F₀ ATPase, cytochrome c oxidase, and cytochrome c reductase, and preservation of respiration in isolated brain mitochondria under conditions of nitrosative stress with PARP inhibition or addition of poly(ADP-ribose) glycohydrolase.¹⁰ In addition, we have shown that mitochondrial PARP inhibition using PARP1 gene deletion or relatively selective PARP1 inhibitors protects primary cortical neurons and fibroblasts from oxidative stress and excitotoxicity.⁹

Powerful neuroprotection using non-mitochondria targetinggene deletion has been demonstrated in pre-clinical models of stroke,⁴¹ TBI,^{56, 57} and neurodegenerative disease.^{42, 58} In these and other CNS diseases, an important role for mitochondrial energy failure is implicated. XJB-veliparib (and similar compounds) that concentrate in mitochondria would be expected to directly inhibit mitochondrial PARP, reduce NAD⁺ depletion, prevent energy failure, and reduce cell death while having minimal impact on genomic DNA repair and integrity (please refer to Abstract Graphic). Furthermore, XJB-veliparib (and similar compounds) may also enhance mtDNA repair⁶ when used to treat these diseases. Mitochondria-targeting PARP inhibitors may also reduce PARylation of apoptosis-inducing factor (AIF), potentially mitigating AIF release from mitochondria, translocation of AIF into nuclei, and parthanatos-related cell death.⁵⁹

Conclusions

Mitochondria-targeting PARP inhibitors may have unique and important advantages over PARP inhibitors that are non-selective to cellular compartments, including directly preserving mitochondrial function and viability and preventing mitochondrial energy failure. Moreover, mitochondria-targeting PARP inhibitors could have clinical efficacy in diseases where the pathophysiology includes mitochondrial dysfunction and/or energy failure, but efficient nuclear DNA repair is desired—such as ischemia-reperfusion injury, radiation toxicity, and other diseases involving oxidative stress. As PARP inhibitors are being reconsidered for many clinical indications,⁶⁰ further development and in vitro and in vivo evaluation of mitochondria-targeting PARP inhibitors appears warranted.

METHODS

Synthesis of XJB-veliparib

For the synthesis of mitochondrially targeted veliparib, XJB-Veliparib, N-Cbz-L-proline methyl ester (7) was treated with NaHMDS and MeI, followed by saponification, to give the acid 9 in 98% yield over 2 steps (Scheme 1). One-pot EDCl coupling to methyl 2,3-diaminobenzoate (10) and acid-catalyzed cyclization provided the benzimidazole 11 in good overall yield. Hydrolysis of the methyl ester followed by acylation with the N-Boc-1,3-diaminopropane spacer group afforded 12. Removal of the Cbz-group via hydrogenation followed by Boc-deprotection provided 13, which was coupled to the mitochondrial targeting sequence 14⁶¹ to afford the desired XJB-Veliparib conjugate 15.

General Experimental Conditions. All moisture- and air-sensitive reactions were performed in oven dried glassware under a positive pressure of argon. All reagents and solvents were used as received unless otherwise specified. THF and Et₂O were distilled over sodium/benzophenone ketyl; CH₂Cl₂ was distilled over CaH₂, MeCN and DMF were dried over molecular sieves. Reactions were monitored by TLC analysis (pre-coated silica gel 60 F₂₅₄ plates, 250 µm layer thickness) and visualization was accomplished with a 254/280 nm UV light and/or by staining with KMnO₄ solution (1.5 g KMnO₄ and 1.5 g K₂CO₃ in 100 mL of a 0.1% NaOH solution), a ninhydrin solution (2 g ninhydrin in 100 mL EtOH), a PMA solution (5 g phosphomolybdic acid in 100 mL EtOH), or a p-anisaldehyde solution (2.5 mL p-anisaldehyde, 2 mL AcOH and 3.5 mL conc. H₂SO₄ in 100 mL EtOH). Flash chromatography was performed on silica gel (40–63 µm). Melting points were determined on a Mel-Temp II capillary melting point apparatus fitted with a Fluke 51 II digital thermometer. Infrared spectra were recorded on an ATR spectrometer. NMR spectra were recorded on 300, 400, 500 or 700 MHz instruments. Chemical shifts were reported in parts per million (ppm) and referenced to residual solvent. ¹H NMR spectra are tabulated as follows: chemical shift, multiplicity (br = broad, s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet), coupling constant(s), number of protons. ¹³C NMR spectra were obtained using a proton-decoupled pulse sequence and are tabulated by observed peak. LC-MS analyses were performed on a Shimadzu UFLC instrument equipped with an Applied Biosystem MDS SCIEX API 2000 mass spectrometer (ESI), under the following conditions: column: Varian Polaris C18-A (100 × 4.6 mm, 5µm) equilibrated at 40 °C; buffer A: 0.1% aqueous AcOH, buffer B: 0.1% AcOH in MeCN; 30 min gradient: 5% buffer B in buffer A for 1 min, then 5 to 95% buffer B in buffer A over 13 min, then 95% buffer B in buffer A for 4 min, then 95–5% buffer B in buffer A over 7 min, then 5% buffer B in buffer A for 5 min; flow rate: 0.2 mL/min; detection: TIC and/or UV λ = 254/280 nm. Intracellular distribution and HRMS data were obtained on a Thermo Scientific Exactive Orbitrap LC-MS (ESI positive ion mode) coupled to a Thermo Scientific Accela HPLC system using a 3.5 µM Water XTerra C18 column (2.1 × 50 mm; 10 min gradient elution with MeCN/H₂O/MeOH containing 0.1% formic acid at a flow rate of 500 µL/min from 3:92:5 at 0–0.5 min to 93:2:5 at 4.0 min, back to 3:92:5 from 6.0 to 7.5 min).

1-Benzyl 2-methyl 2-methylpyrrolidine-1,2-dicarboxylate (8).

To a −78 °C solution of N-Cbz-L-proline 7 (0.43 mL, 1.9 mmol) and iodomethane (0.24 mL, 3.8 mmol) in THF (3.5 mL) was added dropwise NaHMDS (1 M in THF, 3.8 mL, 3.8 mmol). The resulting mixture was warmed to −20 °C, stirred for 3 h, quenched with H₂O, acidified with 2N HCl and extracted with EtOAc (3 x). The combined organic layers were washed with brine (1x), dried over MgSO₄, filtered and concentrated to dryness. The residue was purified by chromatography on SiO₂ (30% EtOAc/hexanes) to give 8 as a pale yellow oil (0.52 g, 99%). Spectral data of the mixture of rotamers are in accordance with literature values.^{54 1}H NMR (300 MHz, CDCl₃) δ 1.54 (s, 1.5 H), 1.61 (s, 1.5 H), 1.86–1.99 (m, 3 H), 2.13–2.24 (m, 1 H), 3.46 (s, 1.5 H), 3.56–3.69 (m, 2 H), 3.71 (s, 1.5 H), 4.99–5.23 (m, 2 H), 7.24–7.40 (m, 5 H).

1-((Benzyloxy)carbonyl)-2-methylpyrrolidine-2-carboxylic acid (9).

Methyl ester 8 (1.53 g, 5.52 mmol) was dissolved in THF (10.5 mL) and H₂O (5.4 mL) and treated with a solution of LiOH (265 mg, 11.0 mmol) in H₂O (5.4 mL). MeOH (1.5 mL) was then added and the resulting homogenous solution heated to 60 °C overnight. The organic solvents were removed and the aqueous layer acidified to pH 2 using 2 N HCl and extracted with EtOAc (3x). The combined organic layers were washed with water (1x), dried (MgSO₄) filtered and concentrated to dryness to give 9 as a colorless solid (1.42 g, 98%). Spectral data of the mixture of rotamers are in accordance with literature values.^{54 1}H NMR (500 MHz, DMSO-d₆) δ 1.44 (s, 1.5 H), 1.45 (s, 1.5 H), 1.80–1.96 (m, 3 H), 2.02–2.19 (m, 1 H), 3.43–3.53 (m, 2 H), 4.94–5.11 (m, 2 H), 7.24–7.41 (m, 5 H), 12.51 (br s, 1 H).

Methyl 2-(1-((benzyloxy)carbonyl)-2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxylate (11).

To a solution of 9 (200 mg, 0.760 mmol) and methyl diaminobenzoate 10 (189 mg, 1.14 mmol) in CH₂Cl₂ (15 mL) was added DIPEA (0.13 mL, 0.76 mmol) followed by addition of EDCI (218 mg, 1.14 mmol), HOAt (155 mg, 1.14 mmol) and DMAP (9.3 mg, 0.076 mmol), and the resulting solution was stirred at room temperature overnight. The reaction mixture was quenched with sat. aqueous NH₄Cl solution and extracted with CH₂Cl₂ (2x). The combined organic layers were washed with brine (1x), dried (MgSO₄), filtered and concentrated to give the product as a brown oil. The crude residue was redissolved in AcOH (5 mL) and heated at reflux for 2 h. The solvent was evaporated, the crude residue poured onto sat. aqueous NaHCO₃ and extracted with EtOAc (3x). The combined organic layers were washed with brine (1x), dried (MgSO₄) and concentrated. The crude residue was purified via chromatography on SiO₂ (30–90% EtOAc/hexanes) to give 11 as a pale yellow oil (198 mg, 65% over 2 steps): HRMS (ESI⁺) m/z calcd for C₂₂H₂₄N₃O₄ [M+H] 394.1761, found 394.1758.

2-(1-((Benzyloxy)carbonyl)-2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxylic acid.

Methyl ester 11 (280 mg, 0.712 mmol) was dissolved in THF/H₂O (14:1, 21 mL) and cooled to −5 °C. A 40% n-Bu₄NOH (aqueous) solutionq (4.6 mL, 7.1 mmol) was added slowly and the reaction mixture was stirred for 30 min at −5 °C and at room temperature over night. The solution was acidified with aqueous AcOH and extracted with EtOAc (3x). The combined organic layers were washed with brine, dried (MgSO₄), filtered and concentrated under reduced pressure. The crude residue was purified by chromatography on SiO₂ (0–10% MeOH/CH₂Cl₂) to give 2-(1-((benzyloxy)carbonyl)-2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxylic acid as a pale-yellow foam (215 mg, 80%): ¹H NMR (400 MHz, DMSO-d₆) δ 7.86 (br s, 1 H), 7.78 (d, J = 7.5 Hz, 1 H), 7.41 – 7.36 (m, 4 H), 7.33 – 7.30 (m, 1 H), 7.30 – 7.25 (m, 1H), 7.06 (t, J = 7.4 Hz, 0.5 H), 6.91 (t, J = 7.0 Hz, 1H), 6.70 (br s, 1 H), 5.15 – 5.09 (m, 0.5 H), 5.07 – 5.04 (m, 1H), 3.87 – 3.54 (m, 3 H), 2.76 (br s, 1 H), 1.92 (s, 3 H), 1.87 (br s, 2 H), 1.86 (br s, 1 H); ¹³C-NMR (700 MHz, DMSO-d₆) δ 167.5, 160.5, 159.8, 154.4, 154.1, 137.4, 136.8, 128.9, 128.2, 127.9, 127.7, 127.1, 124.4, 124.3, 121.7, 66.3, 66.2, 62.9, 55.4, 48.8, 48.3, 42.7, 40.5, 24.4, 24.2, 22.9, 22.4; IR (ATR, neat) 1682.8, 1407.6, 1351.5, 1255.8, 746.4 cm⁻¹; Mp 102.3 – 104.6 °C; HRMS (ESI⁺) m/z calcd for C₂₁H₂₂N₃O₄ [M+H] 380.1605, found 380.1610.

Benzyl 2-(4-((3-((TERT-butoxycarbonyl)amino)propyl)carbamoyl)-1H-benzo[d]imidazol-2-yl)-2-methylpyrrolidine-1-carboxylate (12).

To a solution of N-Boc-1,3-propanediamine (0.28 mL, 1.6 mmol) and 2-(1-((benzyloxy)carbonyl)-2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxylic acid (342 mg, 1.05 mmol) in CH₂Cl₂ (12 mL) at 0 °C was added DIPEA (0.46 mL, 2.6 mmol), followed by the dropwise addition of T₃P (50 wt% in EtOAc, 0.92 mL, 1.6 mmol). The resulting mixture was stirred at room temperature overnight, washed with 5% aqueous Na₂CO₃ and brine, dried (MgSO₄), filtered and concentrated. The crude residue was purified via chromatography on SiO₂ (100% CH₂Cl₂ to 10% MeOH/CH₂Cl₂) to give 12 as a pale-yellow foam (503 mg, 89%): ¹H NMR (600 MHz, DMSO-d₆) δ 12.76 (br s, 0.5 H), 12.69 (br s, 0.5 H), 9.93 (br s, 1H), 7.86 – 7.80 (m, 1 H), 7.64 (d, J = 7.7 Hz, 0.5 H), 7.60 (d, J = 7.7 Hz, 0.5 H), 7.39 – 7.34 (m, 2 H), 7.33 – 7.27 (m, 2 H), 6.94 – 6.81 (m, 2 H), 6.73 – 6.65 (m, 1 H), 5.09 – 4.97 (m, 1 H), 4.94 – 4.79 (m, 1 H), 3.86 – 3.77 (m, 1 H), 3.70 – 3.60 (m, 1 H), 3.47 – 3.37 (m, 2 H), 3.11 – 3.00 (m, 2 H), 2.28 – 2.10 (m, 2 H), 2.02 – 1.94 (m, 2 H), 1.91 (br s, 1.5 H), 1.89 (br s, 1.5 H), 1.73 – 1.65 (m, 3 H), 1.38 (s, 9 H); ¹³C-NMR (600 MHz, DMSO-d₆) δ 165.1, 160.4, 160.2, 156.1, 154.1, 153.7, 140.7, 140.6, 137.5, 136.7, 135.3, 135.2, 128.8, 128.2, 127.8, 127.7, 127.0, 122.5, 122.4, 122.3, 122.2, 122.1, 115.2, 115.1, 77.9, 66.2, 62.6, 62.2, 55.4, 49.0, 48.2, 43.4, 42.1, 38.2, 36.9, 30.3, 30.3, 28.7, 24.4, 23.3, 23.0, 22.5 (mixture of rotamers); Mp 112.7 – 116.2 °C; HRMS (ESI⁺) m/z calcd for C₂₉H₃₈N₅O₅ [M+H] 536.2867, found 536.2867.

tert-Butyl (3-(2-(2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxamido)propyl)carbamate.

A mixture of 12 (500 mg, 0.933 mmol) and 10% Pd/C (99 18 mg, 0.093 mmol) in MeOH (5 mL) in a high-pressure hydrogenator was purged with argon and flushed with hydrogen (3x). Thepressure was set to 6 bar and the mixture was stirred at this pressure at room temperature overnight. The solution was filtered through a plug of Celite® and washed with CH₂Cl₂. The filtrate was concentrated and the crude residue was purified via chromatography on SiO₂ (100% CH₂Cl₂ to 10% MeOH/CH₂Cl₂) to give tert-butyl (3-(2-(2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxamido)propyl)carbamate as a white foam (375 mg, 85%): ¹H NMR (600 MHz, DMSO-d₆) δ 12.50 (br s, 2 H), 9.94 (br s, 1 H), 7.78 (d, J = 7.5 Hz, 1 H), 7.62 (d, J = 7.5 Hz, 1 H), 7.25 (t, J = 7.7 Hz, 1 H), 6.92 – 6.87 (m, 1 H), 3.45 – 3.38 (m, 2 H), 3.12 – 3.04 (m, 3H), 2.93 – 2.86 (m, 1 H), 2.45 – 2.38 (m, 1 H), 1.89 – 1.80 (m, 2 H), 1.72 – 1.64 (m, 3 H), 1.59 (br s, 3 H), 1.38 (s, 9 H); ¹³C-NMR (600 MHz, DMSO-d₆) δ 165.3, 156.2, 122.1, 121.8, 79.7, 79.4, 79.2, 77.9, 62.6, 60.2, 55.4, 49.1, 46.4, 38.1, 36.8, 30.4, 28.7, 27.6, 25.7, 21.3, 14.6; IR (ATR, neat) 3269, 2973, 1694, 1645, 1612, 1523, 1406, 1365, 1245, 1166, 1046, 988, 758 cm⁻¹; Mp 79.8 – 83.2 °C; HRMS (ESI⁺) m/z calcd for C₂₁H₃₂N₅O₃ [M+H] 402.2500, found 402.2499.

N-(3-Aminopropyl)-2-(2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxamide hydrochloride (13).

To a solution of tert-butyl (3-(2-(2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazole-4-carboxamido)propyl)carbamate (32.8 mg, 0.0817 mmol) in CH₂Cl₂ (1 mL) was added 4 M HCl in dioxane (0.2 mL, 0.8 mmol). The resulting mixture was stirred at room temperature for 2 h. The product was filtered off, washed with hexanes, and the resulting white solid (27.1 mg, 98%) was used in the next step without further purification: HRMS (ESI⁺) m/z calcd for C₁₆H₂₄N₅O [M+H] 302.1975, found 302.1974.

tert-Butyl ((4S,7S,E)-7-benzyl-2-methyl-8-((2S)-2-(((2S)-3-methyl-1-(((8S)-15-(2-(2-methylpyrrolidin-2-yl)-1H-benzo[d]imidazol-4-yl)-3,9,15-trioxo-1-phenyl-2-oxa-4,10,14-triazapentadecan-8-yl)amino)-1-oxobutan-2-yl)carbamoyl)pyrrolidin-1-yl)-8-oxooct-5-en-4-yl)carbamate (“XJB-Veliparib”, 15).

To a solution of XJB-acid 14 (36.9 mg, 0.0458 mmol) and 13 (23.2 mg, 0.0687 mmol) in DMF (0.92 mL, 0.05M) at 0 °C was added DIPEA (40 µL, 0.23 mmol), followed by the dropwise addition of T₃P (50% in DMF, 35 µL, 0.060 mmol). The reaction mixture was stirred at 0 °C for 1 h, quenched with aqueous NH₄Cl, and extracted with CH₂Cl₂ (3x). The combined organic layers were washed with 5% aqueous LiCl solution (1x), dried (MgSO₄), filtered and concentrated to give the crude product as a pale yellow oil that was purified by chromatography on SiO₂ (100% CH₂Cl₂ to 15% MeOH/CH₂Cl₂) to afford 15 as a colorless solid (18.5 mg, 37%): ¹H NMR (400 MHz, DMSO-d₆, 100 °C) δ 9.32 (br s, 1 H), 7.88 (d, J = 7.44 Hz, 1 H), 7.72 (d, J = 7.96 Hz, 1 H), 7.69 – 7.59 (m, 1 H), 7.47 – 7.38 (m, 1 H), 7.37 – 7.31 (m, 5 H), 7.30 – 7.26 ( m, 1 H), 7.25 – 7.12 (m, 5 H), 6.79 (br s, 1 H), 6.12 (d, J = 7.72 Hz, 1 H), 5.52 – 5.42 (m, 2 H), 5.02 (s, 2 H), 4.44 – 4.37 (m, 1 H), 4.30 – 4.23 (m, 1 H), 4.18 – 4.09 (m, 1 H), 3.93 – 3.84 (m, 1 H), 3.53 – 3.35 (m, 6 H), 3.29 – 3.21 (m, 3 H), 3.10 – 3.03 (m, 4 H), 2.72 – 2.00 (m, 3 H), 1.98 – 1.88 (m, 2 H), 1.84 (br s, 3 H), 1.80 – 1.71 (m, 5 H), 1.66 – 1.57 (m, 1 H), 1.55 – 1.43 (m, 3 H), 1.38 (s, 9 H), 1.30 – 1.25 (m, 4 H), 0.90 – 0.78 (m, 12 H); ¹³C NMR (700 MHz, DMSO-d₆) δ 172.7, 172.5, 172.2, 171.9, 171.8, 171.7, 171.4, 171.3, 171.2, 164.8, 156.6, 155.3, 155.2, 139.7, 139.5, 137.7, 135.0, 134.6, 129.6, 128.8, 128.4, 128.3, 128.2, 127.8, 126.4, 126.2, 122.8, 78.0, 77.8, 65.6, 64.9, 64.6, 59.7, 59.6, 59.3, 58.2, 53.1, 53.0, 50.7, 50.5, 49.3, 48.8, 47.2, 47.1, 45.5, 38.9, 38.2, 36.5, 36.4, 35.2, 32.4, 30.9, 30.7, 30.2, 29.8, 28.7, 26.5, 25.0, 24.7, 24.4, 24.3, 23.8, 23.0, 22.8; Mp 147.2 – 152.6 °C; IR (ATR, CH₂Cl₂) 3291.7, 2955.8, 1647.1, 1528.8, 1439.1, 1365.7, 1248.0, 1167.2, 1028.3, 758.2, 698.1 cm⁻¹; HRMS (ESI) m/z calcd for C₆₀H₈₅N₁₀O₉ 1089.6496, found 1089.6495.

PARP1 Activity

The capacity to inhibit PARP1 was determined using a commercial assay (Trevigen, Gaithersburg, MD) as per manufacturer’s direction. Various concentrations of XJB-veliparib or veliparib were added to histone-coated wells containing active PARP1 enzyme and NAD⁺ in surplus.

Cell Cultures

Primary cortical neuron-enriched cultures were prepared from 16–17 day old Sprague-Dawley rat embryos as described.⁹ Dissociated cell suspensions were filtered through a 70 µm nylon cell strainer and seeded in 96-well plates (5×10⁴ cells/well) or on poly-D-lysine coated glass coverslips, and maintained in Neurobasal medium with B27 supplements (Life Technologies, Carlsbad, CA). Experiments were performed 12 days in vitro (DIV).

HT22 cells were cultured at 37°C in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen Inc., Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific, San Jose, CA) and 1% penicillin-streptomycin (ATCC, Manassas, VA) in an atmosphere containing 5% CO₂. Cells were cultured for 24 to 48 h before use.

Separation of Subcellular Proteins

Separation of cytosol-, nuclear-, and mitochondria-enriched protein fractions was performed using differential centrifugations as previously described.⁶² Samples were homogenized in lysis buffer then centrifuged at 1025 g for 15 min at 4°C to separate nuclei. Pellets containing nuclei were resuspended in buffer and centrifuged at 16,000 g at 4°C for 20 min with these supernatants containing nuclear-enriched fractions. Supernatants after separation of nuclei containing mitochondria and cytosolic proteins were then centrifuged at 20,000 g for 50 min at 4°C to separate cytosol-enriched protein fractions. The pellets were lysed, sonicated until frothy, and re-suspended in lysis buffer then centrifuged at 16,000 g for 25 min at 4°C. These mitochondria-enriched pellets were then washed in 5 µM CaCl and centrifuged at 1,600 g for 10 min. Western blots were performed using antibodies against cytochrome c oxidase (BD Biosciences, San Jose, CA) and histone H3 (Abcam, Cambridge, MA) to verify purity of subcellular fractions.

Isolation of Mitoplasts

Mitochondria from HT22 cells (25 × 10⁶ cells/mL) were isolated using differential centrifugation following a standardized protocol using a commercial isolation kit (#89874; ThermoFisher Scientific, San Jose, CA). A low spin speed (3,000 g) was used to collect only heavy mitochondria for OMM removal. Mitochondria were resuspended in 325 µL mitochondrial isolation reagent C, then 100 µL of suspension was centrifuged at 3,000 g. The pellet was then digested with 0.15 – 1.5 mg/mL freshly dissolved digitonin in 125 µL volume. The samples were agitated at 750 rpm for 15 min at 4ºC, then digestion was stopped by adding an equal volume of reagent C. Suspensions were then centrifuged at 10,000 g for 10 min at 4ºC with the pellet containing mitoplasts (IMM and matrix) and the supernatant containing OMM and protein from the intermembrane space. The purity of the mitoplast containing fractions was verified by the absence of TOM40 and presence of MnSOD on western blot.

Quantification of XJB-veliparib

The mitochondrial fraction (50 µL) or nuclear fraction (50 µL), treated with XJB-veliparib was added to a 5:1 ratio of CH₂Cl₂:MeOH (950 µL) and vortexed (30 sec). Water (150 µL) was added and the solution was vortexed (15 sec) and set aside to equilibrate at room temperature (30 min). The resulting suspension was placed in an Eppendorf Centrifuge 5702 (4400 rpm, 20 °C) for 12 min. The organic layer was extracted and filtered through a 0.45 µm filter for analysis.

XJB-veliparib was quantified on a Thermo Scientific Exactive Orbitrap LC-MS (ESI positive ion mode) coupled to a Thermo Scientific Accela HPLC system using a 3.5 µM Water XTerra C18 column (2.1 × 50 mm; 20 min gradient elution with MeCN/H₂O containing 0.1% formic acid at a flow rate of 500 µL/min from 5:95 at 0–1.0 min to 95:5 at 12.0 min, back to 5:95 from 16.0 to 16.1 min). Calibration curves for XJB-veliparib were run in duplicate from 102 nM to 5.7 nM. Samples (10 µL) were injected in triplicate and Thermo Xcalibur software was used to determine the concentration of XJB-veliparib in mitochondrial and nuclear fractions (n = 3). The concentration was reported as pM concentration of XJB-veliparib per 10 µg of protein with corresponding standard deviation values.

Oxygen-glucose Deprivation

To model ischemia-reperfusion in vitro, culture medium was replaced with a pre-equilibrated low glucose (0.5 mM) medium. Neurons were transferred into a sealed hypoxic chamber (Coy Laboratory Products Inc., Grass Lake, MI) set to an atmosphere of 95% N₂ with 5% CO₂ at 37^∘C for 2 h. After OGD neurons were removed from the chamber and returned to the incubator.

Excitotoxicity

To model excitotoxicity in primary cortical neurons, cells were exposed to 10 µM L-glutamate with 10 µM glycine. Neuronal cells from HT22 cell line were exposed to 5 mM L-glutamate.

Assessment of Cell Death

Cell death was quantified by measuring lactate dehydrogenase (LDH) released into supernatant using a colorometric assay.⁹ LDH values were normalized to 100% cell death caused by 0.5% Triton X-100 exposure. Data are reported as the percentage of dead cells relative to total cells and presented as mean ± standard deviation (SD).

Measurement of NAD⁺

NAD⁺ levels were quantified in subcellular fractions by the enzymatic cycling method using alcohol dehydrogenase as described.⁹ A standard curve was generated using known concentrations of NAD⁺ and levels were calculated.

STED Imaging

Neurons grown on poly-D-lysine coated glass coverslips were fixed in 2% paraformaldehyde and permeabilized with TritonX-100. Coverslips were then incubated in a 1:200 dilution of mouse monoclonal antibody against PAR (SA216, ENzo Life Sciences, Inc., Farmingdale, NY) and an antibody against TOMM20 (Abcam, Cambridge, MA) followed by incubation in the appropriate secondary antibodies. STED imaging was performed using a Leica TCS SP8 super resolution STED microscope with a pulsed white light laser and AOBS detection system was used (Leica Microsystems, Wetzlar, Germany). Images were collected using the 775 nm STED laser line with 30% 3D STED using the Leica STED WHITE oil objective lens (HC PL APO 100x/1.40 OIL) with a 200 Hz scan speed and 2× line averaging. Pixel size was set to 45 nm/pixel, step size was set to 160 µm, and pinhole was set at 132.8 µm (0.875 AU). TOMM20 was visualized using Alexa Fluor 555, exciting at 553 nm and detecting between 558–599 nm and temporally gate between 0.83–4.33 nm. PAR was visualized with Alexa Fluor 594, exciting at 598 nm and detecting between 603–666 nm and temporally gated between 0.3–6.0 nsec. DRAQ5 (ThermoFisher Scientific, Waltham, MA) used to label cell nuclei was excited at 662 and detected between 667 to 780 nm and temporally gated between 0.3–6.0 nsec. Channels were collected between stacks, sequentially.

COMET Tail Assay

HT22 cells were incubated for 60 min at 37°C in growth media (DMEM) with vehicle (DMSO, 0.1%), 10 µM XJB-veliparib, or 10 µM veliparib. Cells were then washed with PBS and exposed to 200 µM H₂O₂ in the presence of vehicle, XJB-veliparib, or veliparib for 30 min on ice. Cells were rewashed to remove H₂O₂, and media containing vehicle, XJB-veliparib, or veliparib was replaced for specified times before collection via rubber policeman. Cells were washed in 4°C PBS, counted, resuspended in DMEM (10% DMSO, 40% FBS), and frozen (−80°C) for batch analysis. The comet tail intensity assay was performed as previously described.⁴⁷