Biological consequences of alteration of cellular poly(ADP-ribose) polymerase-1 expression in rodent cells

We have identified three novel structures for inhibitors of thepoly(ADP-ribose) polymerase (PARP), a nuclear enzyme activatedby strand breaks in DNA and implicated in DNA repair, apoptosis,organ dysfunction or necrosis. 2-[4-(5-Methyl-1H-imidazol-4-yl)-piperidin-1-yl]-4,5-dihydro-imidazo[4,5,1-i,j]quinolin-6-one(BYK49187), 2-(4-pyridin-2-yl-phenyl)-4,5-dihydro-imidazo[4,5,1-i,j]quinolin-6-one (BYK236864), 6-chloro-8-hydroxy-2,3-dimethyl-imidazo-[1,2]-pyridine (BYK20370), and 4-(1-methyl-1H-pyrrol-2-ylmethylene)-4H-isoquinolin-1,3-dione (BYK204165) inhibitedcell-free recombinant human PARP-1 with pIC50 values of 8.36,7.81, 6.40, and 7.35(pKi 7.97, 7.43, 5.90, and 7.05), and murinePARP-2 with pIC50 values of 7.50, 7.55, 5.71, and 5.38, respec-tively. BYK49187, BYK236864, and BYK20370 displayed no se-lectivity for PARP-1/2, whereas BYK204165 displayed 100-foldselectivity for PARP-1. The IC50 values for inhibition of poly(ADP-ribose) synthesis in human lung epithelial A549 and cervical car-cinoma C4I cells as well in rat cardiac myoblast H9c2 cells afterPARP activation by H2O2 were highly significantly correlated withthose at cell-free PARP-1 (r 0.89–0.96, P 0.001) but lesswith those at PARP-2 (r 0.78–0.84, P 0.01). The infarct sizecaused by coronary artery occlusion and reperfusion in the anes-thetized rat was reduced by 22% (P 0.05) by treatment withBYK49187 (3 mg/kg i.v. bolus and 3 mg/kg/h i.v. during 2-hreperfusion), whereas the weaker PARP inhibitors, BYK236864and BYK20370, were not cardioprotective. In conclusion, the imi-dazoquinolinone BYK49187 is a potent inhibitor of human PARP-1activity in cell-free and cellular assays in vitro and reduces myo-cardial infarct size in vivo. The isoquinolindione BYK204165 wasfound to be 100-fold more selective for PARP-1. Thus, both com-pounds might be novel and valuable tools for investigating PARP-1-mediated effects. Poly(ADP-ribose) polymerases (PARP) 1 and 2 are abun-dant nuclear enzymes in eukaryotic cells that have beenimplicated in the cellular response to DNA damage (Schre-iber et al., 2006). PARPs catalyze an energy-consuming re-action by transferring ADP-ribose moieties from the sub-strate NAD to nuclear acceptor proteins, including PARPitself, and to existing ADP-ribose adducts on protein, thusforming chains of poly(ADP-ribose) (PAR), to render dam-aged DNA accessible to the repair system and to maintaincell survival, genomic stability, and mammalian longevity(D’Amours et al., 1999). This beneficial, cytoprotective effectof PARP activity is apparent under conditions of low to mod-Article, publication date, and citation information can be found athttp://molpharm.aspetjournals.org.doi:10.1124/mol.108.048751. ABBREVIATIONS: PARP, poly(ADP-ribose) polymerase; PAR, poly(ADP-ribose); 3-AB, 3-aminobenzamide; NA, nicotinamide; 4-HQN, 4-hy-droxyquinazoline; ISQ, 1,5-dihydroxyisoquinoline; GPI-6150, 1,11b-dihydro-[2H]benzopyrano[4,3,2-de]isoquinolin-3-one; 5-AIQ, 5-aminoiso-quinolin-1(2H)-one; INO-1001, (6-fluoro-2,3,4,11b-tetrahydro-1H-fluoreno[1,9-cd]azepin-10-ylmethyl)-methyl-amine; PJ34, N-(6-oxo-5,6-dihy-dro-phenanthridin-2-yl)-N,N-dimethylacetamide; SPA, scintillation proximity assay; DMSO, dimethyl sulfoxide; PBS, phosphate-buffered saline;DMEM, Dulbecco’s modified Eagle’s medium; LAD, left anterior descending coronary artery; AAR, area at risk; PEG, polyethylene glycol;BYK49187, 2-[4-(5-methyl-1H-imidazol-4-yl)-piperidin-1-yl]-4,5-dihydro-imidazo[4,5,1-i,j]quinolin-6-one; BYK236864, 2-(4-pyridin-2-ylphenyl)-4,5-dihy-dro-imidazo[4,5,1-i,j]quinolin-6-one; BYK20370, 6-chloro-8-hydroxy-2,3-dimethyl-imidazo-[1,2]-pyridine; BYK204165, 4-(1-methyl-1H-pyrrol-2-ylmethylene)-4H-isoquinolin-1,3-dione; DPQ, 3,4-dihydro-5-[4-(piperidinyl1-yl)butoxy]isoquinolin-1(2H)-one; PND, 6-(5H)-phenantridinone;INH2BP, 5-iodo-6-amino-1,2-benzopyrone; 4-ANI, 4-amino-1,8-naphthalimide; AG14361, 1-(4-dimethyl-aminomethyl-phenyl)-8,9-dihydro-7H-2,7,9a-benzo[cd]azulen-6-one; FR261529, 2-(4-chlorophenyl)-5-quinoxalinecarboxamide; FR247304, 5-chloro-2-[3-(4-phenyl-3,6-dihydro-1(2H)-pyridinyl)propyl]-4(3H)-quinazolinone.0026-895X/08/7406-1587–1598$20.00MOLECULAR PHARMACOLOGYVol. 74, No. 6Copyright © 2008 The American Society for Pharmacology and Experimental Therapeutics48751/3411777Mol Pharmacol 74:1587–1598, 2008Printed in U.S.A. 1587atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom erate damage infliction. A more intense activation of PARP inresponse to abundant genotoxic stimuli activates an apopto-tic pathway to eliminate cells with insufficiently repairedDNA, mediated via release of apoptosis-inducing factor frommitochondria (Yu et al., 2002). Severe DNA damageor consequences of a variety of cardiovascular and inflamma-tory diseases, such as shock, ischemia, diabetes, and neuro-degenerative disorders, can cause excessive activation ofPARP, which depletes the intracellular pools of NAD andsubsequently ATP, ultimately leading to cellular dysfunctionand necrosis by rapid energy consumption (Pieper et al.,1999; Virág and Szabó, 2002; Amè et al., 2004). Conse-quently, depending on the circumstances, pharmacologicalinhibitors of PARP have the potential to either enhance thecytotoxicity of antitumor treatment, or to provide remarkableprotection from tissue damage in various forms of reperfu-sion organ injury, inflammation, and neurotoxicity in animalmodels (Virág and Szabó, 2002; Beneke et al., 2004; Jagtapand Szabó, 2005; de la Lastra et al., 2007). Although themajor isoform PARP-1, encoded by one of the seventeen cur-rently known members of the human PARP gene family, wasthought to be responsible for all the DNA damage-dependentPAR synthesis in mammalian cells, a second DNA damage-dependent isoform, PARP-2, was subsequently discoveredbased on the presence of residual DNA-dependent PARPactivity in cells from parp-1( / ) mice (Shieh et al., 1998).Distinct binding modes necessary for discrimination betweenligands and each isoenzyme have been discovered, enablingsynthesis of PARP-1 selective quinazolinones and PARP-2selective quinoxalines (Iwashita et al., 2004a,b; Ishida et al.,2006).PARP-1 activation contributes to the tissue injury causedby ischemia and reperfusion in various organs, including heart(Eliasson et al., 1997; Thiemermann et al., 1997; Liaudet et al.,2001). A reduction in infarct size and/or improved cardiac con-tractility after myocardial ischemia in rats has been demon-strated for PARP inhibitors of different chemical structure [e.g.,3-AB, NA, 4-HQN, ISQ, 5-AIQ, GPI-6150, PJ34, and INO-1001(Thiemermann et al., 1997; Zingarelli et al., 1997; Bowes et al.,1998; Docherty et al., 1999; McDonald et al., 2000; Pieper et al.,2000; Wayman et al., 2001; Faro et al., 2002)]. However, theirPARP inhibitory effect in vivo is not determined solely by theirpotency in vitro, but most notably governed by their ability tocross cell membranes and their low lipophilicity. Thus, al-though different new chemical structures of potent PARP in-hibitors have been discovered in the last decade (Southan andSzabó, 2003; Jagtap and Szabó, 2005), the need for developingselective inhibitors that are both potent and sufficiently water-soluble is still of pivotal importance (Woon and Threadgill,2005).In the present study, we describe the biochemical and phar-macological properties of two new imidazoquinolinone com-pounds, BYK49187 and BYK236864, the imidazopyridineBYK20370, and the isoquinolindione BYK204165, of whichonly the latter compound bears a benzamide structure mim-icking to some degree the nicotinamide moiety of the sub-strate NAD (Fig. 1). We characterized these compoundswith respect to inhibitory potency and selectivity on cell-freerecombinant human PARP-1 and murine PARP-2, includinganalysis of their kinetics and reversibility of PARP-1 inhibi-tion. We also tested the compounds in various cellular sys-tems (i.e., human lung epithelial A549, human cervical car-cinoma C4I, and rat cardiac myoblast H9c2 cells) comparedwith a series of standard PARP inhibitors of various chemicalclasses and potencies. Two of the new compounds were testedfor inhibition of PARP in parp-1( / ) and parp-1( / )mouse fibroblasts. In addition, we evaluated three of thecompounds for their ability to reduce myocardial reperfusioninjury, measured as infarct size in the anesthetized rat, areliable model in which treatment with PARP-inhibitors ofdifferent chemical structures have been shown to reduce theinfarct size and to improve cardiac contractility (Bowes et al.,1998; Docherty et al., 1999; Pieper et al., 2000; Wayman etal., 2001; Virág and Szabó, 2002; Szabó et al., 2004). Materials and MethodsPARP-1 Assay. The enzymatic reaction of the recombinant hu-man PARP-1 was performed by scintillation proximity assay (SPA)run in a 384-well format using microtiter F-plates (Greiner, Frick-enhausen, Germany). The assay was carried out in a total volume of50 l, comprising 100 mM Tris-HCl, pH 7.8, 10 mM MgCl2, 10 mMdithiothreitol, 1 M NAD , 0.067 Ci of [H]NAD (GE Healthcare,Chalfont St. Giles, Buckinghamshire, UK), 1 g of double-strandedoligonucleotide GGAATTCC (ARK Scientific, Darmstadt, Germany),100 ng of PAR antibody (from Dr. M. Frey, Steinbeis-Transfer Cen-tre, Mannheim, Germany, or Alexis Corporation, Läufelfingen, Swit-zerland), in the absence and presence of various concentrations ofPARP inhibitors (dissolved in DMSO). Final DMSO concentrationsin the assay did not exceed 0.3%. An equal amount of vehicle wasadded to the control samples. The enzymatic reaction was started byadding 75 ng of recombinant partially purified human PARP-1 (Dr.M. Frey). After vigorous shaking of the microtiter-plates for a fewminutes, protein A-coated SPA-polyvinyltoluene beads (Amersham)were added. The mixture was vortexed again and kept at roomtemperature for 18 to 20 h. Bead-bound radioactivity (counts perminute) was measured by liquid scintillation spectrometry.NAD concentrations were 1 M to calculate the pIC50 values forall compounds for half-maximal inhibition of enzyme activity, andvaried from 0.2 to 200 M for determination of Km value of thesubstrate and for analyzing the type of inhibition of the novel com-pounds by Lineweaver-Burk plots as well as for determination of pKivalues by Schild plots. In the latter case, three to four differentinhibitor concentrations spaced by a factor of 2.5 to 3 (0.4 to 0.5 logunits) were used for graphical calculation of pKi and slope of regres-sion. All experiments were performed in duplicate or more, and theaverage of the results was used for analysis. Fig. 1. Chemical structures of the imidazoquinolinones BYK49187 andBYK236864, the imidazopyridine BYK20370, and the isoquinolindioneBYK204165.1588 Eltze et al. atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom PARP-2 Assay. Likewise, the enzymatic reaction of the recombi-nant mouse PARP-2 was quantified by SPA run on a 96-well formatusing microtiter V-plates (Wallac Isoplate; PerkinElmer Life andAnalytical Sciences, Waltham, MA). At the time these experimentswere initiated, human PARP-2 was not available. When it becameavailable during the course of the experiments, we decided to con-tinue with the murine enzyme for consistency. The assay was carriedout in a total volume of 100 l, comprising 100 mM Tris-HCl, pH 7.8,10 mM MgCl2, 10 mM dithiothreitol, 1 M NAD , 0.067 Ci of[H]NAD , 100 ng of PAR antibody (Alexis Corporation), 100 ng ofrecombinant mouse PARP-2 (Alexis Corporation) initially dissolvedin 50 mM Tris-HCl, pH 7.5, 14 mM -mercaptoethanol, 0.5 mMphenylmethylsulfonyl fluoride, and 10% glycerol, protein A-coatedSPA-PNT antibody-binding beads (GE Healthcare), in the absence orpresence of various concentrations of PARP inhibitors (dissolved inDMSO). Final DMSO concentrations in the assay did not exceed0.3%. An equal amount of the vehicle was added to the controlsamples. The enzymatic reaction was started by adding 1 g ofdouble-stranded calf thymus DNA (DNase-digested; Sigma) and in-cubating at room temperature during shaking for 60 min and thenfor 16 h overnight. Radioactivity incorporated from [H]NAD intoPAR, and then being captured by PAR antibody and finally bound toSPA beads, was measured by liquid scintillation spectrometry.Immunofluorescence Analysis of PAR in Fibroblasts. Mouseembryonic fibroblasts (3T3) from parp-1( / ) and parp-1( / ) micewere cultured to confluence in Dulbecco’s modified Eagle’s medium(DMEM) containing 4.5 g/l glucose supplemented with 0.58 g/l L-glutamine, penicillin G (100 units/ml), streptomycin (100 g/ml), and10% heat-inactivated fetal calf serum at 37°C in a humidified 5%CO2-95% air incubator. Confluent cells were washed in PBS andtreated as described below. For immunofluorescence analysis ofPAR, cells were trypsinized, plated on sterile coverslips at a densityof 2 10 cells/cm in 12-well culture dishes, and allowed to adhereovernight (Wagner et al., 2007). After exposure of cultures to theinhibitor (0.3–3 M or 0.3–10 M, final DMSO concentration 0.3%)for 30 min, cells were washed with PBS and treated with H2O2 [5mM for parp-1( / ) fibroblasts, 50 mM for parp-1( / ) fibroblasts]for 5 min at 37°C to stimulate PAR formation. Cells were then fixed[methanol/acetic acid, 3:1 (v/v)] for 10 min at room temperature.After three washings with PBS, cells were incubated with monoclo-nal antibody 10H directed against PAR at a dilution of 1:250 inblocking solution (5% nonfat milk powder in PBS and 0.05% Tween20) for 1 h at 37°C in a humid chamber. After three washings withPBS, antibody-antigen complexes were detected with Alexa Fluor488-conjugated goat secondary antibody (Invitrogen, Karlsruhe, Ger-many) for 45 min at 37°C. The cells were washed three times andthen counterstained with 4 ,6 -diamidino-2-phenylindole. Cells wereexamined under a fluorescence microscope for detection of PAR.Cellular PARP Assay. Human cervical carcinoma C4I cells(American Type Culture Collection, Manassas, VA), human lungepithelial A549 cells, and rat H9c2 cardiomyocytes (American TypeCulture Collection) were grown to confluence in culture flasks con-taining minimum essential medium (RPMI-1640 medium; Sigma)and 10% fetal calf serum. After reaching confluence, cells weretrypsinized (0.05%). After centrifugation at 50g for 5 min, pelletedcells were resuspended in RPMI-1640 medium containing 2 mML-glutamine. Cells were incubated in 96-well plates for 3 days at37°C until confluence was reached, at which point the cell numberwas between 2 and 5 10 cells/plate. Cellular supernatant wasremoved from the wells by aspiration and wells were washed oncewith 100 l of DMEM. DMEM (60 l) was added to the adherent celllayer. Inhibitor dilution series were prepared in 100% DMSO anddiluted 100-fold in DMEM. Thirty microliters of inhibitor solutionwas added to 60 l of DMEM in each well, yielding a total of 90 l ofinhibitor/DMEM solution, with a final DMSO concentration of 0.3%.Cells were preincubated with inhibitors for 30 min; then, intracellu-lar PARP was activated by the addition of 10 l of H2O2 (10 mM;final concentration, 1 mM). Cells were incubated for 10 min at 37°C.The reaction was terminated by adding 100 l of fixation solution(70% methanol/30% acetone [v/v]; precooled to 20°C) for 10 min.The supernatant was aspirated, and plates were dried for 30 min.For rehydration of the cells, 100 l of phosphate-buffered saline(PBS) was added for 10 min at room temperature. PBS was removedby aspiration and 100 l of blocking solution (5% nonfat milk powderin PBS containing 0.05% Tween 20) was added to each well followedby 30-min incubation at room temperature. After removal of theblocking solution, mouse monoclonal PAR-antibody 10H (final con-centration, 20 g/ml; Steinbeis-Transfer Centre) was added in 100 lof blocking solution. Cells were incubated for 1 h at 37°C. Wells werewashed twice for 5 min with 100 l of Tween 20/PBS and thesecondary fluorescein isothiocyanate-conjugated goat anti-mouse an-tibody (50-fold dilution in blocking buffer; Sigma) was added. Cellswere incubated for 30 min at 37°C and then washed two times with100 l of Tween 20/PBS. Fluorescence was measured with the dryplates in a fluorescence counter (Wallac Victor; PerkinElmer Lifeand Analytical Sciences) at 485/536 nm.All compounds were subjected to a uniform solution and dilutionprocedure, with DMSO as solvent not exceeding final concentrationsof 0.3% in all noncellular and cellular assays. In accordance withprevious observations (Banasik et al., 2004), this DMSO concentra-tion did not interference with the cell-free and cellular assaysystems.Coronary Artery Ligation and Myocardial Infarct Size inthe Rat. The method of coronary artery occlusion and reperfusion inthe anesthetized rat was performed as described previously (Way-man et al., 2001). The care and the use of animals in this work werein accordance with UK Home Office guidelines on the Animals (Sci-entific Procedures) Act 1986 and the European Community guide-lines for the use of experimental animals. Wistar rats (male, 200–300g; Tuck, Rayleigh, Essex, UK) receiving standard diet and water adlibitum were anesthetized with thiopentone sodium (Intraval, 120mg/kg i.p.; Rhône-Merrieux, Essex, UK) and thereafter intubatedand ventilated with a Harvard ventilator. Body temperature wasmaintained at 38 1°C. The right carotid artery was cannulated andconnected to a pressure transducer (MLT 1050; AD Instruments Ltd,Hastings, UK) to monitor mean arterial blood pressure and heartrate. The right jugular vein was cannulated for administration ofdrugs and Evans Blue (at the end of the experiment). A lateralthoracotomy was performed, and the heart was suspended in atemporary pericardial cradle. A snare occluder was placed aroundthe left anterior descending coronary artery (LAD); after that, theanimals were allowed to stabilize for 30 min before LAD ligation. Thecoronary artery was occluded at time 0, and at 20 min into myocar-dial ischemia, a bolus injection of either vehicle or test compoundwas administered intravenously. After 25 min of acute myocardialischemia, the occluder was reopened to allow the reperfusion for 2 h,during which the vehicle or test compound was continuously infused.After that, the coronary artery was reoccluded, and Evans Blue [1 mlof 2% (w/v)] injected into the left ventricle, via the right jugular veincannula, to distinguish between still perfused and nonperfused [areaat risk (AAR)] sections of the heart. After death of the animals by anoverdose of anesthetic, the heart was excised and sectioned intoslices of 3 to 4 mm. The right ventricular wall was removed, and theAAR (pink) was separated from the nonischemic (blue) area fordetermination of AAR portion in percent of the left ventricular por-tion. The AAR was cut into small pieces and incubated with p-nitroblue tetrazolium (0.5 mg/ml) for 30 min at 37°C. In the presence ofintact dehydrogenase enzyme systems in viable myocardium, nitroblue tetrazolium forms a dark blue formazan, whereas areas ofnecrosis lack dehydrogenase activity and therefore fail to stain.Pieces were separated according to staining and weighed to deter-mine the infarct size as a percentage of the weight of the AAR.The following groups of animals (all n 10) were studied: 1)Sham-operated control group of rats subjected to the surgical proce-dure alone (without LAD occlusion) and treated with vehicle (20%polyethylene glycol-400 [PEG; Serva, Heidelberg, Germany], 15% 1Novel PARP Inhibitors1589 atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom N HCl, 65% distilled water); 2) vehicle control group of rats subjectedto myocardial ischemia for 25 min followed by reperfusion (2 h) andtreated with vehicle; 3) treatment groups of rats subjected to myo-cardial ischemia and reperfusion and treated with BYK49187,BYK236864, or BYK20370 at 1 or 3 mg/kg i.v. (each n 10).BYK204165 was not investigated in vivo because of its poorer watersolubility and its short half-time (t1/2) of 23 min measured at ratmicrosomes in vitro, compared with the other compounds with t1/2values 40 min (not shown).The test compounds were initially dissolved in PEG-400 (Serva)and then diluted to the necessary concentrations with 1 N HCl anddistilled water. The final concentrations of PEG and 1 N HCl were 20and 15% (v/v), respectively.Ex Vivo PARP-1 Assay. At the end of experiment, venous bloodsamples were taken (in EDTA-coated tubes) under anesthesia. Bloodplasma was generated via centrifugation (2250g, 10 min, 4°C) andstored at 80°C. The ex vivo PARP-1 assay was done in analogy asdescribed, but respective blood samples were added instead of testdrug solutions.Lactate Dehydrogenase Assay. Cellular toxicity was deter-mined by lactate dehydrogenase release measured by the CytoTox 96assay kit from Promega (Mannheim, Germany).Materials. BYK49187, BYK236864, BYK20370, and BYK204165were synthesized at NYCOMED GmbH (formerly ALTANA Chem-ical Research, Konstanz, Germany). ISQ, 4-ANI, NA, 3-AB, 3,4-dihydro-5-[4-(piperidinyl1-yl)butoxy]isoquinolin-1(2H)-one(DPQ), 4-HQN, 6-(5H)-phenantridinone (PND), 5-iodo-6-amino-1,2-benzopyrone (INH2BP), GPI-6150, and PJ34 were purchasedfrom Alexis Corporation.Human recombinant PARP-1 was supplied by Dr. M. Frey (Stein-beis-Transfer Centre). Mouse recombinant PARP-2 was obtainedfrom Alexis Corporation. [H]NAD was purchased from Amersham(now Perkin Elmer, UK). Monoclonal antibody against PAR wasfrom Alexis Corporation (10 g/ml) or from Dr. M. Frey (Steinbeis-Transfer Centre). Goat anti-mouse antibody (85 g/ml fluoresceinisothiocyanate) was from Sigma. All other chemicals were from com-mercial suppliers with highest grade of purity.Statistical Analysis. The pIC50 values of test compounds forhalf-maximal inhibition of cell-free PARP-1 and PARP-2 as well ofPARP in the cell lines were calculated from concentration-responsecurves by using Prism 5.0 (GraphPad Inc., San Diego, CA). In anal-ogy to antagonist-receptor interaction, Schild plots were constructedfrom data derived from enzyme kinetic experiments to estimate thepKi value of the inhibitor and the slope of regression as an importantparameter in that it defines whether or not the data fit the simplecompetitive model of substrate-inhibitor interaction. Calculation ofthe correlation coefficient r and the slope of regression line of datausing two sets of inhibitory potencies (pIC50 values) were performedto compare the results obtained from different experimental assays.All data are presented as means S.E.M. Infarct size in rats wasanalyzed by single-factorial analysis of variance, followed by a Dun-nett’s test for comparison of a treated group to the vehicle or shamgroup. P values 0.05 were considered statistically significant. ResultsStructure and Solubility of BYK49187, BYK236864,BYK20370, and BYK204165. Fig. 1 shows the imidazo-quinolinones BYK49187 and BYK236864, the imidazopyri-dine BYK20370, as well as the isoquinolindione BYK204165,of which only the latter compound bears the benzamide moi-ety, typically present in known PARP inhibitors such as3-AB, PND, GPI-6150, DPQ, or 4-ANI. With the exception of3-AB, NA, 4-ANI, 5-AIQ, and INH2BP, which are known tobe readily soluble in saline, all reference compounds, includ-ing PND, DPQ, GPI-6150, and ISQ, were relatively insolublein water, which was also true for BYK49187, BYK236864,BYK20370, and BYK204165, which had maximal attainableconcentrations in saline of 0.42, 0.02, 2.0, and 0.009 mM,respectively, mirrored by calculated logP values of 2.3, 3.8,2.5, and 2.3, respectively (Table 1). Therefore, regardless oftheir solubility, all test compounds were dissolved in DMSOand further diluted in 10% DMSO to the desired test drugconcentrations. Final DMSO concentrations in cell-free andcellular PARP-inhibition assays did not exceed 0.3%, a con-centration known to exert no inhibitory effect on PARP-1activity (Banasik et al., 2004).Identification of Compounds of the Imidazochinoli-none, Imidazopyridine, and Isoquinolindione Struc-ture as Potent Inhibitors of Human PARP-1. In ourcell-free recombinant human PARP-1 assay, enzymatic activ-ity was measured by quantification of pmol[H]ADP-ribosebound to antibody binding beads within 24 h by using alow substrate concentration of 1 M NAD . This rapid andreliable biochemical screen, using parts of the compoundlibrary at NYCOMED GmbH, identified two imidazochinoli-nones, BYK49187 and BYK236864, and the isoquinolindione TABLE 1Potencies of BYK49187, BYK236864, BYK20370, and BYK204165 in comparison with reference compounds to inhibit cell-free human PARP-1,mouse PARP-2, and H2O2-activated cellular PARP in human lung epithelial A549, human cervical carcinoma C4I, and rat cardiac myoblast H9c2cellsInhibitory potencies are expressed as pIC50 values with slope values (in parentheses) of the concentration-response curves. Given are means S.E.M. of n 3 to 5experiments for hPARP-1 and mPARP-2 and n 4 to 6 experiments for each cellular PARP assay and compound. Numbering of the reference compounds (in brackets) refersto Fig. 8. LogP values were calculated by Hansch method.CompoundhPARP-1mPARP-2A549 CellsC4I CellsH9c2 CellslogPBYK491878.36 0.11 (1.08)7.70 0.13 (0.93)7.80 0.087.02 0.117.65 0.032.3BYK2368647.81 0.09 (1.13)7.55 0.10 (0.89)6.41 0.03N.T.6.70 0.033.8BYK203706.40 0.13 (1.17)5.71 0.14 (1.05)6.51 0.146.05 0.096.02 0.042.5BYK2041657.35 0.10 (0.96)5.38 0.08 (1.08)6.64 0.155.75 0.076.91 0.052.3PJ34 17.72 0.08 (0.93)7.21 0.06 (1.19)N.T.N.T.N.T.1.64-ANI 27.66 0.09 (1.24)7.49 0.05 (1.03)6.88 0.066.24 0.126.94 0.081.4GPI-6150 37.08 0.14 (1.31)6.73 0.17 (1.03)6.81 0.116.25 0.116.86 0.122.6PND 47.07 0.13 (1.39)6.48 0.16 (0.91)6.77 0.136.37 0.156.62 0.122.2ISQ 56.82 0.11 (1.09)6.35 0.13 (0.87)6.51 0.095.47 0.096.51 0.112.4DPQ 66.43 0.14 (1.07)5.76 0.13 (0.75)6.60 0.175.47 0.126.64 0.032.55-AIQ 75.93 0.07 (1.02)5.74 0.08 (1.18)N.T.N.T.N.T.0.44-HQN 85.23 0.11 (1.16)4.59 0.09 (1.01)4.97 0.124.46 0.114.84 0.101.7INH2BP 95.07 0.17 (0.95)4.75 0.11 (0.98)4.80 0.083.87 0.094.62 0.041.93-AB 104.89 0.09 (1.06)4.38 0.06 (1.01)4.95 0.114.16 0.114.81 0.140.3NA 114.30 0.08 (1.13)3.68 0.10 (1.25)4.29 0.123.73 0.174.10 0.110.2N.T., not tested.1590 Eltze et al. atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom BYK204165, as potent inhibitors of human PARP-1, whereasthe imidazopyridine BYK20370 was less potent. Mean pIC50values were 8.36, 7.81, 7.35, and 6.40 for BYK49187,BYK236864, BYK204165, and BYK20370, respectively. Theslope values of the inhibition curves were near unity, point-ing to a homogeneous population of enzyme and an NADcompetitive behavior of all compounds (Table 1). None of thereference compounds (e.g., PJ34, 4-ANI, GPI-6150, and PNDwith pIC50 values of 7.72, 7.66, 7.08, and 7.07, respectively)reached the nearly nanomolar potency of BYK49187 for in-hibition of PARP-1 but were 5to 20-fold weaker. Among theother compounds investigated, 3-AB and NA proved to be theweakest inhibitors of PARP-1 (pIC50 of 4.89 and 4.30, respec-tively) (Table 1).The selectivity of the compounds was then assessed byinhibition of recombinant mouse PARP-2, using the same lowsubstrate concentration (1 M NAD ) as in the PARP-1assay. Potencies for inhibition of PARP-2 (pIC50 values) byBYK49187 (7.70), BYK236864 (7.55), BYK20370 (5.71), andall reference inhibitors investigated were less than 0.7 logunits (i.e., a factor of 5) lower than those for inhibition ofPARP-1, and by definition these compounds must be re-garded as unselective, although some compounds (e.g., PND,DPQ, and 4-HQN), together with BYK49187 and BYK20370,discriminated both isoforms by a factor of 3 but 5, therebydisplaying a small preference for PARP-1 (Table 1). By con-trast, the isoquinolindione BYK204165 discriminated be-tween PARP-1 and PARP-2 by a factor of 100 (pIC50 7.35versus 5.38).NAD Competitive Inhibition and Reversibility ofHuman PARP-1 by BYK49187, BYK236864, BYK20370,and BYK204165. The mechanism of inhibition of humanPARP-1 was investigated by measurement of enzyme veloc-ity at increasing NAD concentrations (0.2 up to 200 M) inthe absence and presence of increasing inhibitor concen-trations. In the absence of inhibitor, Km values were be-tween 5 and 9 M and Vmax values between 7 and 11 pmolof[H]ADP-ribose/mg of bead-bound enzyme. Addition ofBYK49187 (10–300 nM), BYK236864 (50–400 nM),BYK20370 (1–30 M), or BYK204165 (30 nM-3 M) re-sulted in rightward shifts of the apparent Km value ofNAD , whereas Vmax of the reaction essentially did notchange (Figs. 2–5). The Lineweaver-Burk plots shown inthese figures demonstrate that all compounds acted asNAD competitive inhibitors. By analyzing the data inrespective Schild plots for determination of inhibitor pKivalues from the intercepts with the abscissa, the followingvalues were obtained: BYK49187, pKi 7.97 (slope 0.64;significantly different from unity, P 0.01); BYK236864,pKi 7.43 (slope 1.06; not significantly different fromunity, P 0.05); BYK20370, pKi 5.90 (slope 0.94; notsignificantly different from unity, p 0.05); andBYK204165, pKi 7.05 (slope 0.98; not significantly dif-ferent from unity, P 0.05). In general, affinity constants(pKi values) were in good agreement with respective pIC50values, but on average 0.4 log units lower, which can beexplained by the fact that the latter were determinedat a lower NAD concentration (1 M) than its Km value( 5 M).Consistent with the Lineweaver-Burk plots, data fromdilution experiments confirmed that even total inhibitionof PARP-1 by BYK49187, BYK236864, BYK20370, andBYK204165 (at starting concentrations 100-fold of the re-spective IC50 values) was fully reversible after a 1000-folddilution in assay buffer with constant substrate concentra-tion (1 M NAD ), thereby reaching an enzyme activitycomparable with that in the absence of inhibitors (100%). Ineach case, a half-maximal reversal of PARP-1 inhibition ap-proximately occurred by a 100-fold dilution of each startingconcentration, enabling calculation of pIC50 values that didnot differ by more than 0.2 log units from those previouslydetermined by single concentrations (not shown).Inhibition of PAR Formation in parp-1( / ) andparp-1( / ) Mouse Fibroblasts. To confirm PARP-1 selec-tivity of BYK204165, we further investigated this compoundin comparison with the unselective BYK236864 in parp-1( / ) and parp-1( / ) fibroblasts. Nuclear PAR was visu-alized by immunofluorescence analysis using the PAR-speFig. 2. A, substrate dependence of human PARP-1 activity in the pres-ence of increasing BYK49187 concentrations (10, 30, 100, and 300 nM). B,Lineweaver-Burk plot, showing respective data from saturation experi-ments and revealing the competitive type of inhibition. C, pKi value forBYK49187 determined by Schild plot analysis (pKi 7.97; intercept withthe abscissa at a slope of 0.64). Each point represents the mean ofduplicate determinations.Novel PARP Inhibitors 1591 atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom cific monoclonal antibody 10H. The assay demonstrated thecharacteristic granular distribution pattern of PAR forma-tion in nuclei upon DNA-damaging treatment of the cellswith H2O2 (Fig. 6). The unselective inhibitor BYK236864completely abrogated immunostaining at 3 M and above inboth cell lines, whereas the PARP-1 selective inhibitorBYK204165 did not preclude residual PAR formation, evenat 10 M in parp-1( / ) or at 3 M in parp-1( / ) fibro-blasts. The latter finding is perfectly compatible with ongoingPARP-2 activity in both cell lines and clearly demonstratesthe high selectivity of BYK204165 for PARP-1.Inhibition of PARP in Various Intact Cells. The cellu-lar potency of PARP inhibition by the compounds and stan-dard inhibitors was tested in human lung epithelial A549,human cervical carcinoma C4I, and rat cardiomyocyte H9c2cells, in which activation of PARP was performed by additionof H2O2. None of the compounds (BYK49187, BYK236864,BYK20370, and BYK204165) showed cellular toxicity up to100 M, as evidenced by the lack of detectable lactate dehy-drogenase activity in cellular supernatants.PAR formation in A549, C4I, and H9c2 cells was inhibitedby BYK49187 with pIC50 values of 7.80, 7.02, and 7.65, Fig. 3. A, substrate dependence of human PARP-1 activity in the pres-ence of increasing BYK236864 concentrations (50, 125, and 400 nM). B,Lineweaver-Burk plot, showing respective data from saturation experi-ments and revealing the competitive type of inhibition. C, pKi value forBYK236864 determined by Schild plot analysis (pKi 7.43; intercept withthe abscissa at a slope of 1.06). Each point represents the mean ofduplicate determinations.Fig. 4. A, substrate dependence of human PARP-1 activity in the pres-ence of increasing BYK20370 concentrations (1, 3, 10, and 30 M). B,Lineweaver-Burk plot, showing respective data from saturation experi-ments and revealing the competitive type of inhibition. C, pKi value forBYK20370 determined by Schild plot analysis (pKi 5.90; intercept withthe abscissa at a slope of 0.94). Each point represents the mean ofduplicate determinations.1592 Eltze et al. atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom respectively (Table 1). However, the other imidazoquinolin-one, BYK236864, which has been shown to be only 3-foldweaker than BYK49187 at cell-free human PARP-1, was 10-to 25-fold weaker than BYK49187 in these cellular assays(pIC50 6.41 in A549 cells and 6.70 in H9c2 cells), possiblybecause of its lower membrane permeability (20-fold) comparedwith BYK49187. The cellular potencies of the less water-insol-uble imidazopyridine BYK20370 (pIC50 6.02–6.51) were com-parable with the value derived from the cell-free PARP-1 assay(pIC50 of 6.40), whereas those of the less water-soluble isoquino-lindione BYK204165 differed from the cell-free assay value bymore than 5-fold (Table 1). From the reference compounds in-vestigated in these cellular assays, once again 4-ANI was themost potent inhibitor but did not reach the potency ofBYK49187. The respective pIC50 values for 4-ANI in A549, C4I,and H9c2 cells were 6.88, 6.24, and 6.94, respectively, whereas3-AB together with 4-HQN, INH2BP, and NA was one of theweakest inhibitors (Table 1). Concentration-response curves ofBYK49187, BYK236864, BYK20370, and BYK204165 for inhi-bition of PARP in A549 and H9c2 cells are depicted in Fig. 7.Although the correlation of pIC50 values of the compoundsderived from PARP inhibition in A549 cells with those ob-tained in C4I cells was highly significant [r 0.95, P0.001; slope 0.97 0.07 (mean S.E.M.), n 12], gener-ally 3to 10-fold higher concentrations were necessary toachieve half-maximal inhibition of PARP in C4I cells, possi-bly reflecting a higher endogenous NAD concentration inthese cells (Fig. 8A). An even better correlation (r 0.97,P 0.001; slope 1.05 0.06, n 13) was obtained bycomparing the pIC50 values of the compounds in A549 cellswith respective values in rat cardiac myoblast H9c2 cells,strongly suggesting comparable penetration of the com-pounds into these cells, despite uncertainty about their in-tracellular NAD substrate concentrations (Fig. 8B).As a further analysis, we compared the pIC50 values ob-tained from cell-free human PARP-1 and mouse PARP-2assays with respective values of the compounds derived from Fig. 5. A, substrate dependence of human PARP-1 activity in the pres-ence of increasing BYK204165 concentrations (30 nM, 100 nM, 300 nM,1 M, and 3 M). B, Lineweaver-Burk plot, showing respective data fromsaturation experiments and revealing the competitive type of inhibition.C, pKi value for BYK204165 determined by Schild plot analysis (pKi 7.05;intercept with the abscissa at a slope of 0.98). Each point is the mean ofduplicate determinations.Fig. 6. Immunofluorescence analysis of H2O2 induced PAR formation incultured fibroblasts from parp-1( / ) (A) and parp-1( / ) (B) mice. Nospecific PAR staining is observed in the absence of H2O2 ( H2O2). PARformation induced by H2O2 ( H2O2) in parp-1( / ) fibroblasts (A) as aresult of activation of both PARP-1 and PARP-2 is characterized by largenumber of intense, granular signals in the cell nuclei and is totallyinhibited by BYK236864 (0.3–10 M), whereas residual PAR formation isdetectable in the presence of BYK204165 (0.3–10 M). Note that the“soft” staining visible in the panel H2O2 10 M BYK236864 is non-specific cytoplasmic background (see below). In parp-1( / ) fibroblasts(B), PAR formation (arrows) induced by H2O2 ( H2O2) as a result ofactivation of PARP-2 only is much weaker than in parp-1( / ) fibro-blasts, as expected, and is abrogated in the presence of BYK236864 (0.3–3M) but remains unaffected by BYK204165 (0.3–3 M). Note that thephotographic exposure time for the parp-1( / ) samples was muchlonger, in view of the reduced overall signal intensity. As a consequence,a nonspecific, “soft,” cytoplasmic background emerges in all parp-1( / )samples that is, however, easily distinguishable from the genuine, gran-ular, intranuclear PAR signals.Novel PARP Inhibitors 1593 atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom the three cell lines (Fig. 9). In general, the IC50 values forinhibition of PAR synthesis in A549, C4I, and H9c2 cells afterPARP activation by H2O2 were 3to 35-fold higher than atthe cell-free human PARP-1. Increasingly better correlationsand slopes closer to unity were obtained by comparingPARP-1 values with those at A549 cells (r 0.89, P 0.001;slope 0.78 0.08, n 13), at C4I cells (r 0.92, P0.001; slope 0.83 0.08, n 12) and finally at H9c2 cells(r 0.96, P 0.001; slope 0.87 0.06, n 13). Except forBYK236864, there was no great difference in loss of potencyin cellular PARP assays related to cell-free PARP-1 assaybetween known inhibitors with good water-solubility (e.g.,3-AB, 4-ANI, PJ34, INH2BP, and 5-AIQ) and those with poorwater-solubility (e.g., DPQ, PND, ISQ, and GPI-6150), sug-gesting that different penetration rates of the compoundsinto the cells were probably not responsible for this effect butshould be related to higher endogenous NAD concentrationswithin the cells. Less significant correlations were obtainedby comparing pIC50 values derived from mouse PARP-2 as-say with those at A549 cells (r 0.78, P 0.01; slope0.71 0.06, n 13), C4I cells (r 0.84, P 0.01; slope0.79 0.09, n 12) and H9c2 cells (r 0.83, P 0.01;slope 0.79 0.09, n 13), thereby excluding a majorparticipation of PARP-2 activation in these cells in responseto H2O2 treatment. It is noteworthy that the correlationscomparing pIC50 values of the inhibitors between the cell-free PARP-1, PARP-2, and cellular PARP assay(s) generatedslopes of regression lines that were consistently less thanunity, which may lead to an apparent underestimation of themore potent drugs in the cellular assays. However, thismight be readily explained by a lower drug diffusion gradientand, consequently, a longer time to reach equilibrium be-tween aqueous medium and intracellular space as a result oflower concentrations needed for those more potent drugs tocalculate their IC50 values.Effect of Test Drugs on Myocardial Infarct Size inthe Rat. In the rat model for regional myocardial infarction,mean values for the AAR were similar in all groups studiedand ranged from 42 to 52% irrespective of treatment withvehicle (PEG plus HCl) or test drugs (Fig. 10A). In animalstreated with vehicle, occlusion of the LAD (for 25 min) folFig. 7. Concentration-response curves for inhibition of H2O2-activatedPARP in human lung epithelial A549 cells (A) and in rat cardiac myoblastH9c2 cells (B) by BYK49187, BYK236864, BYK20370, and BYK204165.Given are means S.E.M. of four to six experiments.Fig. 8. Correlation of pIC50 values for BYK compounds and referencecompounds to inhibit PARP in human lung epithelial A549 versus humancervical carcinoma C4I cells (A) (r 0.95, P 0.001; slope 0.97 0.07,n 12) and versus rat cardiac myoblast H9c2 cells (B) (r 0.97, P0.001; slope 1.05 0.06, n 13). The numbering of the referencecompounds refers to their listing in Table 1.1594 Eltze et al. atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom lowed by reperfusion (for 2 h) resulted in an infarct size of57 3% of the AAR (mean S.E.M., n 11). Intravenousadministration of the lower dose of BYK49187 (1 mg/kg bolusfollowed by 1 mg/kg/h infusion) was nearly ineffective (6%reduction in infarct size; not significantly different from ve-hicle, P 0.05), whereas the higher dose (3 mg/kg followed by3 mg/kg/h) caused a significant reduction in infarct size of22% compared with vehicle (P 0.05; Fig. 10B). Sham oper-ation alone did not result in a significant degree of infarctionin any of the animal groups studied ( 2% of the AAR). Theother test drugs investigated in this model, which had anapproximately 5to 100-fold lower potency to inhibit PARP-1in cell-free and cellular test system(s) [i.e., BYK236864 (1mg/kg i.v. followed by 1 mg/kg/h i.v.) and BYK20370 (3 mg/kgi.v. followed by 3 mg/kg/h i.v.)] neither reduced nor increasedmyocardial infarct size (not shown). BYK204165, because ofits poorer water solubility, was not tested in vivo.To gain insight into the pharmacokinetic properties of thecompounds tested in vivo, blood samples from drug-treated Fig. 9. A, correlation of pIC50 values for BYK compounds and referencecompounds to inhibit cell-free human PARP-1 versus human lung epi-thelial A549 cells (top; r 0.89, P 0.001; slope 0.78 0.08, n 13),versus human cervical carcinoma C4I cells (middle; r 0.92, P 0.001;slope 0.83 0.08, n 12), and versus rat cardiac myoblast H9c2 cells(bottom; r 0.96, P 0.001; slope 0.87 0.06, n 13). B, correlationof pIC50 values for BYK compounds and reference compounds to inhibitcell-free murine PARP-2 versus human lung epithelial A549 cells (top; r0.78, P 0.01; slope 0.71 0.11, n 13), versus human cervicalcarcinoma C4I cells (middle; r 0.84, P 0.01; slope 0.79 0.11, n12), and versus rat cardiac myoblast H9c2 cells (bottom; r 0.83, P0.01; slope 0.79 0.11, n 13).Fig. 10. AAR (A) and infarct size (B) after regional myocardial ischemia(25 min) and reperfusion (2 h) in the anesthetized rat treated withBYK49187. The animals were subjected to the surgical procedure alone(Sham, n 10) or subjected to coronary artery occlusion and reperfusionand treated with either vehicle (n 11) or with BYK49187 at 1 mg/kg i.v.bolus followed by 1 mg/kg/h i.v. or 3 mg/kg i.v. bolus followed by 3 mg/kg/hi.v. (each n 10). Given are means S.E.M. , P 0.05 compared withvehicle. C, blood samples drawn at the end of infusion (2 h) of the testdrugs BYK20370, BYK236864, and BYK49187 (each at 3 mg/kg i.v.) wereanalyzed ex vivo for their ability to inhibit human PARP-1 (in the pres-ence of DNA). Resulting enzyme activities expressed as cpm per sampleare depicted as data points of 7 to 15 samples for each treatment withmean values thereof. , P 0.001 compared with sham. n.s., notsignificant.Novel PARP Inhibitors 1595 atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom rats were drawn at the end of each drug infusion and testedfor their ability to inhibit human PARP-1. Blood samples ofrats treated with 3 mg/kg i.v. BYK236864 or BYK49187significantly inhibited PARP-1 by 54 and 80%, respectively,compared with sham operation (both P 0.001), whereas nosignificant inhibition (13%, P 0.05) was observed with thesame dose of BYK20370 (Fig. 10C). DiscussionIn search for various chemical compounds as potentialinhibitors for PARP-1, we found three new chemical entitiesthat comprised imidazoquinolinone, imidazopyridine, orisoquinolindione structure. Both imidazoquinolinones (i.e.,BYK49187 and BYK236864) as well as the imidazopyridine,BYK20370, lack the classic benzamide structure as a mimicof the nicotinamide moiety of the substrate NAD , whereasthe isoquinolindione BYK204165 formally resembles substi-tuted naphthalimides containing the constrained arylamidemotif, which has become one of the consensus pharmacoph-ore for drug design of PARP-1 inhibitors and of which 4-ANIrepresents its best known member (Schlicker et al., 1999).Using recombinant human PARP-1 and murine PARP-2 astarget enzymes, we investigated all compounds under thesame experimental conditions. Because most of the reportedcompounds are NAD competitive, and the respective pIC50values are therefore directly dependent on the substrate con-centration used in the assay system, all compounds weretested at a NAD concentration (1 M) lower than Km (5–9M), to get reliable estimates of their inhibitory potency andto avoid errors inherent to the use of higher substrate con-centrations. Under these conditions, pIC50 values deter-mined from concentration-response curves approach the re-spective pKi values (Cheng and Prusoff, 1973). We alsocalculated IC50 in the presence of 50 M NAD for our leadcompounds. The IC50 values for BYK49187, BYK236864,BYK20370, and BYK204165 increased by factors of 19, 65,25, and 45, respectively, reflecting quite well the highersubstrate concentration in the cellular assays (see below).Of all inhibitors investigated in this study, BYK49187 andBYK236864 were the most potent drugs; IC50 values of bothPARP-1 and PARP-2 approached the nanomolar range, fol-lowed by 4-ANI and PJ34. A second group of compounds withinhibitory activity in the 100 nM range comprised GPI-6150,PND, and ISQ. Medium affinity in the micromolar range forboth isoenzymes were found for DPQ, BYK20370, and 5-AIQ,whereas INH2BP, 4-HQN, and 3-AB had IC50 values in the10 M range. It is noteworthy that BYK204165 was morepotent at PARP-1 (pIC50 7.35) than at PARP-2 (pIC50 5.38),thus being 100-fold selective for PARP-1, whereas all otherdrugs investigated did not reach a factor of 10 and in thisrespect must be classified unselective. The ability of PNDto weakly discriminate between PARP-1 and PARP-2 bya factor of 3 has been previously reported (Perkins et al.,2001) and was confirmed in the present study, whereas PJ34and 3-AB have been characterized as unselective inhibitors(Iwashita et al., 2004b). In accordance with previous data(Zhang et al., 2000), there was no selectivity for GPI-6150between PARP isoenzymes.In kinetic experiments with human PARP-1, BYK49187,BYK236864, and BYK204165 exhibited potent and competi-tive inhibition of enzyme activity, yielding pKi values of 7.97,7.43, and 7.05, respectively, whereas BYK20370 was found tobe less potent (pKi 5.90). Consistent with the kinetic analysisdemonstrating a competitive type of inhibition of PARP-1by BYK49187, BYK236864, BYK20370, and BYK204165, di-lution experiments revealed that even total inhibition ofPARP-1 by high concentrations of all compounds is fullyreversible after dilution in assay buffer. Relating to potencyon PARP-1, BYK49187 was identified as a PARP-1 inhibitorin the near nanomolar range and is comparable with recentlydescribed potent inhibitors, such as AG14361 (pKi 8.3; Cala-brese et al., 2004) and KU0058684 (pIC50 8.4; McCabe et al.,2005).Inhibition of cellular PAR synthesis in response to H2O2treatment by BYK49187, BYK20370, and BY204165 in hu-man A549 and C4I cells as well as in rat H9c2 cells afterPARP activation showed only 2to 5-fold lower pIC50 valuesthan at the isolated human PARP-1 enzyme. BYK236864,probably because of its poorer water solubility and higherlipophilicity, lost potency by a factor approximately 10 in allcellular assays. This is line with the observation that 3-ABand NA, having the lowest logP values, displayed the small-est loss in potency in cellular assays. By and large, thepotencies of inhibitors at the isolated human PARP-1 satis-factorily reflect potency at the cellular level and were corre-lated with high significance (P 0.001), particularly in H9c2cells. With regard to potency, there is a good compatibility ofdata between the cell-free and these three cellular assaysystems; however, a complete congruence cannot be expectedbecause the cellular uptake of different structural classes ofcompounds varies as a result of their different physicochem-ical properties (e.g., logP values). It is noteworthy that in-creasingly better correlations were obtained by comparingpotencies of the inhibitors from the cell-free human PARP-1assay with respective values derived from the three cell lines(i.e., A549 cells C4I cells H9c2 cells), which renders theH9c2 cell system an attractive and reliable model that isparticularly suitable for assessment of human PARP-1 inhi-bition at the cellular level. Less significant correlations wereobtained by comparing pIC50 values of the inhibitors at thethree cell lines with respective values from mouse PARP-2(P 0.01), thereby excluding a major contribution of PARP-2activation to the cellular response to H2O2 treatment.Two compounds, BYK204165 and DPQ, exhibited low po-tency of PARP inhibition in C4I cells (pIC50 of 5.75 and 5.47,respectively), which was apparently not compatible with in-hibition of PARP-1 (pIC50 of 7.35 and 6.43, respectively), butrather of PARP-2 (pIC50 of 5.38 and 5.76, respectively) (Table1). To address the possibility that PARP-2 rather thanPARP-1 is involved in this cell type, we compared pIC50values at mouse PARP-2 and human PARP-1 with those inC4I cells from eight compounds (i.e., BYK49187, BYK20370,BYK204165, PND, DPQ, 4-HQN, 3-AB, and NA) capable ofdiscriminating between both enzyme isoforms at least by afactor of 3. However, the correlations obtained under theseconditions did not fit better with either PARP-2 (r 0.92,P 0.01; slope 0.87 0.08, n 8) or PARP-1 (r 0.94,P 0.01; slope 0.81 0.10, n 8) (not shown). Thus, thisapproach did not reveal any preference for PARP-2 overPARP-1 activation in response to H2O2 stimulation in humancervical carcinoma C4I cells, either.In the rat model of regional myocardial ischemia andreperfusion used here, only treatment with BYK49187 at an1596 Eltze et al. atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom intravenous dose of 3 mg/kg bolus plus infusion of the samedose for 2 h caused a significant reduction of infarct sizeof 22%, whereas BYK49187 (at 1 mg/kg), BYK236864, orBYK20370 (both at 1 or 3 mg/kg) were not effective. Thisfinding is consistent with our data from ex vivo experiments,where blood samples taken after BYK49187 at 3 mg/kg i.v.produced a significant PARP-1 inhibition of 80%, revealingthat the blood levels of this compound in the rat were obvi-ously high enough to afford cardioprotection, whereas bloodsamples after 3 mg/kg BYK20370 failed to have a significanteffect in this respect. The ability of BYK236864 blood sam-ples to significantly inhibit PARP-1 ex vivo by more than50%, but its unexpected failure to act cardioprotective couldbe due to the model, in which a 25-min ischemia might havebeen too severe, whereas a shorter ischemic period mighthave revealed a significant reduction of infarct size by thecompound. It has also been argued that a much greatertherapeutic benefit in conditions associated with the conse-quences of ischemia-reperfusion could be attained with morepotent and water-soluble inhibitors of PARP. Available datafrom the literature, however, indicate no clear superiority ofPARP inhibitors with good water-solubility (e.g., 3-AB,5-AIQ, PJ34, INO-1001) over poorly water-soluble inhibitors(e.g., ISQ, GPI-6150), at least regarding reduction of infarctsize in the rat heart. In line with our observation, the max-imally obtainable reduction of myocardial infarct size in therat reported in the literature was 17 to 36% (Zingarelli et al.,1997; Pieper et al., 2000; Liaudet et al., 2001; Wayman et al.,2001; Faro et al., 2002). However, the cardiac infarct size inparp-1( / ) mice subjected to global myocardial ischemiaand reperfusion is maximally decreased by 35% relative tountreated wild-type mice, possibly because of a residual poly-(ADP-ribosyl)ation activity mediated by alternative iso-form(s) of PARP in this tissue (Pieper et al., 2000).From the high degree of homology of the PARP catalyticdomain between species, it has been suggested that PARPinhibitors might exhibit no difference in terms of potency inhuman, rat, and mouse tissues (de Murcia et al., 1994;Iwashita et al., 2004b,c; Kinoshita et al., 2004), and it wasspeculated that none of the PARP inhibitors existing at thattime would be able to discriminate between PARP-1 andPARP-2 (Oliver et al., 2004). However, Perkins et al. (2001)discovered compounds of the quinazolinone and phthalazi-none structure with modest selectivity for PARP-1 andPARP-2, respectively. Distinct binding modes necessary fordiscrimination between ligands and each isoenzyme havebeen identified, enabling the synthesis of quinazolinones(e.g., FR247304), with selectivity for PARP-1, and quinoxa-lines (e.g., FR261529), with selectivity for PARP-2 (Iwashitaet al., 2004a,b; Ishida et al., 2006), thus demonstrating thefeasibility of designing PARP-isoform selective ligands.In terms of selectivity for PARP-1, BYK204165 outper-forms that of recently reported quinazolinones, such asFR247304, with 10to 39-fold selectivity for PARP-1 overPARP-2 (Iwashita et al., 2004b; Ishida et al., 2006). Theenzymatic selectivity of BYK204165 for PARP-1 over PARP-2is maintained at the cellular level in parp-1( / ) and parp-1( / ) mouse fibroblasts. Our data clearly confirm the highselectivity of BYK204165 for inhibition of PARP-1, based onits failure to inhibit PARP-2 in both cell lines, whereas theunselective inhibitor BYK236864 completely abrogates PARformation by both PARP-1 and PARP-2. The method em-ployed here might provide a novel and convenient functionalapproach toward assessment of the contribution of PARP-1and PARP-2 to DNA damage-induced PAR formation in in-tact cells, because the enzymatic activity of the two isoformscan be assessed by use of a selective PARP-1 inhibitor.In conclusion, among the new compounds studied, theimidazoquinolinone BYK49187 emerged as a potent andreversible but unselective PARP-1/2 inhibitor in various invitro assays. This compound also reduced myocardial in-farct size in the rat, whereas the less potent PARP inhib-itors BYK236864 and BYK20370 did not. Furthermore, theisoquinolindione BYK204165 displayed 100-fold selectivityfor PARP-1 and therefore may have two important uses[i.e., 1) as a lead for further drug development and 2) as atool to dissect cellular functions of PARP-1 and PARP-2].Thus, BYK49187, because of its high potency, andBYK204165, because of its high selectivity, representnovel and valuable tools for investigating PARP-1-medi-ated effects. AcknowledgmentsWe thank Gilbert de Murcia (Illkirch-Graffenstaden, France) forparp-1( / ) and parp-1( / ) mouse embryonic fibroblasts. Wegratefully acknowledge I. Gruhler, K. Graf, E. Herrmann, and S.Haas for excellent technical assistance. ReferencesAmé JC, Spenlehauer C, and de Murcia G (2004) The PARP superfamily. Bioessays26:882–893.Banasik M, Stedeford T, Strosznajder RP, Persad AS, Tanaka S, and Ueda K (2004)The effects of organic solvents on poly(ADP-ribose) polymerase-1 activity: impli-cations for neurotoxicity. Acta Neurobiol Exp (Wars) 64:467–473.Beneke S, Diefenbach J, and Bürkle A (2004) Poly(ADP-ribosyl)ation inhibitors:promising drug candidates for a wide variety of pathophysiologic conditions. Int JCancer 111:813–818.Bowes J, Ruetten H, Martorana PA, Stockhausen H, and Thiemermann C (1998)Reduction of myocardial reperfusion injury by an inhibitor of poly (ADP-ribose)synthetase in the pig. Eur J Pharmacol 359:143–150.Calabrese CR, Almassy R, Barton S, Batey MA, Calvert AH, Canan-Koch S, DurkaczBW, Hostomsky Z, Kumpf RA, Kyle S, et al. (2004) Anticancer chemosensitizationand radiosensitization by the novel poly(ADP-ribose) polymerase-1 inhibitorAG14361. J Natl Cancer Inst 96:56–67.Cheng Y and Prusoff WH (1973) Relationship between the inhibition constant(K1)and the concentration of inhibitor which causes 50 per cent inhibition (I50) of anenzymatic reaction. Biochem Pharmacol 22:3099–3108.D’Amours D, Desnoyers S, D’Silva I, and Poirier GG (1999) Poly(ADP-ribosyl)ationreactions in the regulation of nuclear functions. Biochem J 342:249–268.de la Lastra CA, Villegas I, and Sánchez-Fidalgo S (2007) Poly(ADP-ribose) poly-merase inhibitors: new pharmacological functions and potential clinical implica-tions. Curr Pharm Des 13:933–962.de Murcia G, Schreiber V, Molinete M, Saulier B, Poch O, Masson M, Niedergang C,and Ménissier de Murcia J (1994) Structure and function of poly(ADP-ribose)polymerase. Mol Cell Biochem 138:15–24.Docherty JC, Kuzio B, Silvester JA, Bowes J, and Thiemermann C (1999) Aninhibitor of poly(ADP-ribose) synthetase activity reduces contractile dysfunctionand preserves high energy phosphate levels during reperfusion of the ischaemicrat heart. Br J Pharmacol 127:1518–1524.Eliasson MJ, Sampei K, Mandir AS, Hurn PD, Traystman RJ, Bao J, Pieper A, WangZQ, Dawson TM, Snyder SH, et al. (1997) Poly(ADP-ribose) polymerase genedisruption renders mice resistant to cerebral ischemia. Nat Med 3:1089–1095.Faro R, Toyoda Y, McCully JD, Jagtap P, Szabo E, Virag L, Bianchi C, Levitsky S,Szabo C, and Sellke FW (2002) Myocardial protection by PJ34, a novel potent poly(ADP-ribose) synthetase inhibitor. Ann Thorac Surg 73:575–581.Ishida J, Yamamoto H, Kido Y, Kamijo K, Murano K, Miyake H, Ohkubo M,Kinoshita T, Warizaya M, Iwashita A, et al. (2006) Discovery of potent andselective PARP-1 and PARP-2 inhibitors: SBDD analysis via a combination ofX-ray structural study and homology modeling. Bioorg Med Chem 14:1378–1390.Iwashita A, Mihara K, Yamazaki S, Matsuura S, Ishida J, Yamamoto H, Hattori K,Matsuoka N, and Mutoh S (2004a) A new poly(ADP-ribose) polymerase inhibitor,FR261529 [2-(4-chlorophenyl)-5-quinoxalinecarboxamide], ameliorates metham-phetamine-induced dopaminergic neurotoxicity in mice. J Pharmacol Exp Ther310:1114–1124.Iwashita A, Tojo N, Matsuura S, Yamazaki S, Kamijo K, Ishida J, Yamamoto H,Hattori K, Matsuoka N, and Mutoh S (2004b) A novel and potent poly(ADP-ribose)polymerase-1 inhibitor, FR247304 (5-chloro-2-[3-(4-phenyl-3,6-dihydro-1(2H)-pyridinyl)propyl]-4(3H)-quinazolinone), attenuates neuronal damage in in vitroand in vivo models of cerebral ischemia. J Pharmacol Exp Ther 310:425–436.Iwashita A, Yamazaki S, Mihara K, Hattori K, Yamamoto H, Ishida J, Matsuoka N,Novel PARP Inhibitors 1597 atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom and Mutoh S (2004c) Neuroprotective effects of a novel poly(ADP-ribose) polymer-ase-1 inhibitor, 2-[3-[4-(4-chlorophenyl)-1-piperazinyl] propyl]-4(3H)-quinazolin-one (FR255595), in an in vitro model of cell death and in mouse 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine model of Parkinson’s disease. J Pharmacol Exp Ther309:1067–1078.Jagtap P and Szabó C (2005) Poly(ADP-ribose) polymerase and the therapeuticeffects of its inhibitors. Nat Rev Drug Discov 4:421–440.Kinoshita T, Nakanishi I, Warizaya M, Iwashita A, Kido Y, Hattori K, and Fujii T(2004) Inhibitor-induced structural change of the active site of human poly(ADP-ribose) polymerase. FEBS Lett 556:43–46.Liaudet L, Szabó E, Timashpolsky L, Virág L, Cziráki A, and Szabó C (2001)Suppression of poly (ADP-ribose) polymerase activation by 3-aminobenzamide in arat model of myocardial infarction: long-term morphological and functional conse-quences. Br J Pharmacol 133:1424–1430.McCabe N, Lord CJ, Tutt AN, Martin NM, Smith GC, and Ashworth A (2005)BRCA2-deficient CAPAN-1 cells are extremely sensitive to the inhibition of Poly(ADP-Ribose) polymerase: an issue of potency. Cancer Biol Ther 4:934–936.McDonald MC, Mota-Filipe H, Wright JA, Abdelrahman M, Threadgill MD, Thomp-son AS, and Thiemermann C (2000) Effects of 5-aminoisoquinolinone, a water-soluble, potent inhibitor of the activity of poly (ADP-ribose) polymerase on theorgan injury and dysfunction caused by haemorrhagic shock. Br J Pharmacol130:843–850.Oliver AW, Amé JC, Roe SM, Good V, de Murcia G, and Pearl LH (2004) Crystalstructure of the catalytic fragment of murine poly(ADP-ribose) polymerase-2.Nucleic Acids Res 32:456–464.Perkins E, Sun D, Nguyen A, Tulac S, Francesco M, Tavana H, Nguyen H, Tugend-reich S, Barthmaier P, Couto J, et al. (2001) Novel inhibitors of poly(ADP-ribose)polymerase/PARP1 and PARP2 identified using a cell-based screen in yeast. Can-cer Res 61:4175–4183.Pieper AA, Verma A, Zhang J, and Snyder SH (1999) Poly (ADP-ribose) polymerase,nitric oxide and cell death. Trends Pharmacol Sci 20:171–181.Pieper AA, Walles T, Wei G, Clements EE, Verma A, Snyder SH, and Zweier JL(2000) Myocardial postischemic injury is reduced by poly(ADP-ribose) polymer-ase-1 gene disruption. Mol Med 6:271–282.Schlicker A, Peschke P, Bürkle A, Hahn EW, and Kim JH (1999) 4-Amino-1,8-naphthalimide: a novel inhibitor of poly(ADP-ribose) polymerase and radiationsensitizer. Int J Radiat Biol 75:91–100.Schreiber V, Dantzer F, Ame JC, and de Murcia G (2006) Poly(ADP-ribose): novelfunctions for an old molecule. Nat Rev Mol Cell Biol 7:517–528.Shieh WM, Amé JC, Wilson MV, Wang ZQ, Koh DW, Jacobson MK, and Jacobson EL(1998) Poly(ADP-ribose) polymerase null mouse cells synthesize ADP-ribose poly-mers. J Biol Chem 273:30069–30072.Southan GJ and Szabó C (2003) Poly(ADP-ribose) polymerase inhibitors. Curr MedChem 10:321–340.Szabó G, Liaudet L, Hagl S, and Szabó C (2004) Poly(ADP-ribose) polymeraseactivation in the reperfused myocardium. Cardiovasc Res 61:471–480.Thiemermann C, Bowes J, Myint FP, and Vane JR (1997) Inhibition of the activityof poly(ADP-ribose) synthetase reduces ischemia-reperfusion injury in the heartand skeletal muscle. Proc Natl Acad Sci U S A 94:679–683.Virág L and Szabó C (2002) The therapeutic potential of poly(ADP-ribose) polymer-ase inhibitors. Pharmacol Rev 54:375–429.Wagner S, Hussain MZ, Beckert S, Ghani QP, Weinreich J, Hunt TK, Becker HD,and Königsrainer A (2007) Lactate down-regulates cellular poly(ADP-ribose) for-mation in cultured human skin fibroblasts. Eur J Clin Invest 37:134–139.Wayman N, McDonald MC, Thompson AS, Threadgill MD, and Thiemermann C(2001) 5-aminoisoquinolinone, a potent inhibitor of poly (adenosine 5 -diphosphateribose) polymerase, reduces myocardial infarct size. Eur J Pharmacol 430:93–100.Woon EC and Threadgill MD (2005) Poly(ADP-ribose) polymerase inhibition—wherenow? Curr Med Chem 12:2373–2392.Yu SW, Wang H, Poitras MF, Coombs C, Bowers WJ, Federoff HJ, Poirier GG,Dawson TM, and Dawson VL (2002) Mediation of poly(ADP-ribose) polymerase-1-dependent cell death by apoptosis-inducing factor. Science 297:259–263.Zhang J, Lautar S, Huang S, Ramsey C, Cheung A, and Li JH (2000) GPI 6150preventsH2O2 cytotoxicity by inhibiting poly(ADP-ribose) polymerase. BiochemBiophys Res Commun 278:590–598.Zingarelli B, Cuzzocrea S, Zsengellér Z, Salzman AL, and Szabó C (1997) Protectionagainst myocardial ischemia and reperfusion injury by 3-aminobenzamide, aninhibitor of poly (ADP-ribose) synthetase. Cardiovasc Res 36:205–215. Address correspondence to: Dr. Thomas Klein, Boehringer-IngelheimPharma GmbH & Co. KG, 88397 Biberach an der Riss, Germany. E-mail:[email protected] Eltze et al. atKagerLibriAGCirclationDep-univKontanzBilioonNovemer5,2008moharm.aspeurnals.orgDownladedfom Overexpression of PARP-1 delays DNA repair but increases genomicstability in response to cytotoxic stimuli Tobias Eltze, Andrea Kunzmann, Yvonne Rudigier, Raphael Hahn, LauraRossi, A. Ivana Scovassi, Sascha Beneke, Marcus Müller, Cecilia Ström,Thomas Helleday, Alexander Bürkle*a Molecular Toxicology Group, University of Konstanz, D-78457 Konstanz, Germanyb Istituto di Genetica Molecolare CNR, I-27100 Pavia, Italyc German Cancer Research Center, Dept of Tumor Virology, D-69120 Heidelberg,Germanyd Department of Genetics, Microbiology and Toxicology, Arrhenius Laboratory,Stockholm University, S-106 91 Stockholm, Sweden # Equal contributors *Corresponding author. Tel.: +49 7531 884035Fax: +49 7531 884033E-mail address: [email protected]