Influence of lipophilicity and stereochemistry at the C7

Research Article

Influence of Lipophilicity and Stereochemistry at the C7 Position on the Cardioprotective and Antioxidant Effect of Ginkgolides During Rat Heart Ischemia and Reperfusion

DDR

DRUG DEVELOPMENT RESEARCH 64:157–171 (2005)

Ludovic Billottet,1 Sophie Martel,2 Marcel Culcasi,1,3 Katy Drieu,4 Pierre-Alain Carrupt,2 and Sylvia Pietri1* 1 Laboratoire "Structure et Re´activite´ des Espe`ces Paramagne´tiques" UMR 6517 du Centre National de la Recherche Scientifique, Universite´s d’Aix-Marseille I & III, Marseille, France 2 Laboratoire de Chimie The´rapeutique, Ecole de Pharmacie Gene`ve-Lausanne, Section des Sciences Pharmaceutiques, Universite´ de Gene`ve, Sciences II, Gene`ve, Switzerland 3 S.A.R.L. Oxylab, Martigues, France 4 Ipsen-Beaufour-Pharma, Paris, France

Strategy, Management and Health Policy Enabling Technology, Genomics, Proteomics

Preclinical Research

Preclinical Development Toxicology, Formulation Drug Delivery, Pharmacokinetics

Clinical Development Phases I-III Regulatory, Quality, Manufacturing

Postmarketing Phase IV

ABSTRACT The extent to which the cardioprotective effect of ginkgolides is related to their lipophilicity rather than to their anti-platelet activating factor (PAF) effect was addressed in isolated rat hearts submitted to ischemia and reperfusion. A new derivative of ginkgolide C (1), the 7-a-O-(4methylphenyl) ginkgolide C (4) was synthesized and compared to 7-O-(4-methylphenyl) ginkgolide C (2) that had the same absolute configuration at C7 as 1 for its lipophilicity, anti-PAF activity, and cardioprotective and antioxidant effects. Using reversed-phase high-performance liquid chromatography HPLC, 4 and 2 were found to be significantly more lipophilic (i.e., log kw of 3.4270.05 and 3.6470.07, respectively) than 1 (1.1570.03) and the strong PAF inhibitor ginkgolide B (GkB; 1.6570.03). The antiPAF activities (IC50 values in mM) were 8.2, 17.1, and 2.2 for 4, 1, and GkB, respectively, while 2 was inactive. In preischemic and/or reperfused hearts perfused with ginkgolides at 0.7 mM: (i) 2 and 4 were more efficient in improving postischemic hemodynamic and metabolic recovery than 1, (ii) a key-step in cardioprotection occurred during ischemia where 2 and 4 limited myocardial ATP depletion and contracture development, (iii) a strong anti-lipoperoxidant effect was observed with 2 and 4, but not 1. In vivo administration of 2 to rats (4 mg/kg/day for 20 days) was more effective than that of 1 regarding ischemic heart protection, suggesting a positive role for lipophilicity. It was concluded that a high lipophilicity is not an absolute prerequisite for a strong anti-PAF effect for ginkgolides, whereas it appears c 2005 Wiley-Liss, Inc. essential for cardioprotection. Drug Dev. Res. 64:157–171, 2005. Key words: ginkgolides; ischemia–reperfusion; lipophilicity; anti-PAF effect; stereochemistry

This work was supported by the CNRS (UMR 6517), SARL OXYLAB (Biochemical analysis) and IPSEN-Beaufour-PHARMA (Paris, France).

Saint-Je´roˆme (case 521), Avenue Escadrille Normandie-Niemen, F-13397 Marseille Cedex 20, France. E-mail: [email protected]

*Correspondence to Sylvia Pietri, PhD, Laboratoire "Structure et Re´activite´ des Espe`ces Paramagne´tiques" UMR 6517 du Centre National de la Recherche Scientifique, Universite´s d’Aix-Marseille I et III, Faculte´ des Sciences de

Received 1 July 2004; Accepted 11 October 2004

c 2005 Wiley-Liss, Inc.

Published online in Wiley InterScience (www.interscience. wiley.com) DOI: 10.1002/10424

158

BILLOTTET ET AL.

Introduction

The comprehensive study of the pharmacological properties of the standardized extract of Ginkgo biloba leaves EGb 761 has always represented a complex issue because several active chemical constituents contained in this extract can exert additive, antagonistic, and synergistic effects [De Feudis, 1998]. The main pharmacologically active constituents of EGb 761 have been characterized [Furukawa, 1932; Maruyama et al., 1967; Okabe et al., 1967] and, among them, the terpene trilactones ginkgolides, in particular ginkgolides A (GkA), B (GkB), and C (GkC; 1 see Fig. 1) are unique to the extract. Growing interest in ginkgolides has started with the discovery that these molecules, in particular GkB, possess activity as antagonists of platelet-activating factor (PAF), one of the most potent phospholipid mediators in mammals [Braquet et al., 1985; Braquet, 1987; Braquet et al., 1991]. GkA and GkB have been found to be practically completely bioavailable in rat [Biber and Koch, 1999] and human [Fourtillan et al., 1995] plasma after oral or intravenous administration of EGb 761. Many pharmacological activities have been described after in vitro and/or in

vivo treatment of GkA and/or GkB, including the regulation of glucocorticoid levels [Amri et al., 1996; Amri et al., 2003], the protective and antioxidant effect during myocardial ischemia and reperfusion [Pietri et al., 1997; Liebgott et al., 2000; Rioufol et al., 2003], the limitation of apoptosis and lipoperoxidation in rat hippocampal neurons after exposure to a peroxyl radicalgenerator, the 2,20 -azobis 2 amidinopropane [Rapin et al., 1998], the limitation of nitric oxide (NO) production in macrophages via the inducible NO-synthase expression [Cheung et al., 2001] and of NO-induced neurotoxicity [Zhao and Li, 2002], or the effect as potential priming agents of phagocyte function [Lenoir et al., 2002]. However, the extent to which these pharmacological properties of GkA and GkB are solely related to PAF inhibition remains a matter of debate. In in vitro and in vivo studies of rat cardiac ischemia/reperfusion, we have shown [Pietri et al., 1997; Liebgott et al., 2001] that the GkA and GkB concentrations necessary to achieve cardioprotection are two orders of magnitude lower than those required to inhibit PAF and antiarrhythmic effects [Koltai et al., 1989]. In our studies [Pietri et al., 1997; Liebgott et al., 2001], the weak PAF antagonist, GkA, demonstrated a better

Fig. 1. Structures of the ginkgolides (A) and synthesis of 7-a-O-(4-methylphenyl)ginkgolide C (B).

CARDIOPROTECTION BY EPIMER GINKGOLIDES

cardioprotective and antioxidant effects than GkB, suggesting that PAF inhibition is not an important mechanism involved in ginkgolide-induced improvement of postischemic recovery in isolated rat hearts. Particular structural features of ginkgolides may play a key role in their anti-PAF and antioxidant properties. Because it has been established that the presence of the tert-Bu substituent at the C8 position of ginkgolides (Fig. 1) is essential for PAF antagonism [Corey and Srinivas Rao, 1991], it may be anticipated that any structural modification around the C8 (e.g., change in local lipophilicity or steric crowding) may result in a modulation of anti-PAF activity of ginkgolides. A large series of C1 and/or C10 derivatives of GkB were tested as anti-PAF agents, and significant improvements in this property have been reported [Corey and Gavai, 1989; Chen et al., 1998; Hu et al., 1999, 2000; Stromgaard et al., 2002]. The importance of the nature of the substituents at the C7 position of ginkgolides on their pharmacological properties has been highlighted in myocardial ischemia/reperfusion studies [Pietri et al., 2001], showing that the weak PAF inhibitor 1, which should be less lipophilic than GkA and GkB owing to the increased number of hydroxyl substituents (Fig. 1), is also less cardioprotective. To clarify the role of lipophilicity (e.g., correlation with membrane affinity) on both PAF antagonism and cardioprotection in C7-substituted ginkgolides, 2, the 7-O-(4-methylphenyl) derivative of 1 was synthesized (Fig. 1) and its action during cardiac ischemia and reperfusion in isolated rat hearts was investigated [Pietri et al., 2001]. Although in vitro administration of 2 improved recovery of ischemic rat heart as compared to 1 and GkB, 2 was the first example of a C8-tert-Bu substituted ginkgolide with no measurable anti-PAF effect as measured in isolated rabbit platelets [Pietri et al., 2001]. These results [Pietri et al., 2001] show that even in a situation where lipophilicity is likely to increase (i.e., 2 vs. 1), two potentially membrane-related mechanisms (i.e., cardioprotection and PAF inhibition) can vary in opposite directions. The increased steric hindrance at the C7 in 2 with respect to 1, which could weaken the linkage of the C8-tert-Bu to PAF receptor, may be a plausible explanation for lowered PAF antagonism observed in 2. Because many drug– receptor interactions are strongly dependent on the stereochemistry of the active site of the drug, it was considered pertinent to assess the anti-PAF action of the epimer at C7 of 2. In the present study, we report on the two-step synthesis of 4, the 7-a-O-(4-methylphenyl) derivative of 1 (Fig. 1). Compound 4 was compared to GkC and 2 i) on PAF inhibition on rabbit isolated platelets, and ii)

159

on isolated ischemic-reperfused rat hearts for its hemodynamic, metabolic, and antioxidant effects. To further investigate the mechanisms implicated in the cardioprotective effects of ginkgolides and the bioavailability of these compounds in cardiac tissue, preischemic and postischemic perfusion were compared to in vivo administration in ischemia–reperfusion injury. Finally, the first reliable measurement of the lipophilic character of gingkolides (GkB, GkC, 2, and 4) was undertaken using reversed-phase high-performance liquid chromatography (HPLC). Methods

Chemistry Chemicals and analytical techniques Reagents and solvents used in syntheses were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). Preparative HPLC was performed on silica gel (with Merck (Darmstadt, Germany) ‘‘silica gel 60,’’ 0.015 mm–0.040 mm) using a Shimadzu (Duisburg, Germany) LC-8A solvent delivery unit equipped with a Novasep Prochrom (Pompey, France) LC-50 column. Melting points were measured on a Bu¨chi B450 apparatus and were uncorrected. Nuclear magnetic resonance (NMR) spectra were recorded in acetone-d6 on Bruker (Karlsruhe, Germany) DPX 300 (1H, 300.1 MHz; 13C, 75.5 MHz) or AC 100 (19F, 94.1 MHz) spectrometers. Chemical shifts d are given in parts per million (ppm) from (i) tetramethylsilane (TMS) as internal reference (1H and 13C), and (ii) trichlorofluoromethane as external reference (19F). In the 19F NMR spectrum of 3, upfield shift is quoted as negative. Ginkgolide C (BN 52022) was kindly provided by the Institut Henri Beaufour (Paris, France) and was used in water-soluble form (i.e., an equilibrium of open lactone forms at pH 8.75). Syntheses 7-O-(4-methylphenyl) ginkgolide C (2). The Oarylation of the 7-hydroxyl group of 1 was conducted with retention of configuration in three steps as previously described [Pietri et al., 2001], using in the key step a copper-catalyzed arylation with tris-(4methylphenyl) bismuth diacetate. Thus, 31.3 g of 1 yielded 1.0 g of purified 7-O-(4-methylphenyl) ginkgolide C (2) as a white powder (overall yield, 2.7%) which was purified by preparative HPLC eluting with toluene/acetone/acetic acid (70:30:0.5) and a 100 mL/min flow rate. 7-O-trifluoromethanesulfonyl ginkgolide C (3). A solution of 1 (4.4 g, 10 mmol) in dry pyridine (40 mL)

160

BILLOTTET ET AL.

was cooled to 201C under N2 bubbling, and trifluoromethanesulfonic anhydride (2.82 g, 10 mmol) was added dropwise. The reaction mixture was allowed to warm to 01C under stirring for 2 h and was then stirred at 201C for 16 h. Distillation under reduced pressure afforded a brown gum that was dissolved in methyl-t-butylether (100 mL), washed with 1N hydrochloric acid (2 50 mL) and with water (2 10 mL). The organic layer was stirred with charcoal, dried over MgSO4 and filtered. Heptane (100 mL) was added to the filtrate to give a white suspension. After filtration, the resulting solid was washed with heptane (2 50 mL) and dried at 501C for 48 h to yield 3 (5.21 g, 91%) as a white powder, m.p. 2321C (dec.); 1H NMR d (300.1 MHz, acetone-d6) 1.31 (3 H, d, JH16-H14 7.0 Hz, 16-Me), 1.34 (9 H, s, Cme3), 2.49 (1 H, d, JH8-H7 12.4 Hz, H8), 3.07 (1 H, q, JH14-H16 7.0 Hz, H14), 4.34 (1 H, dd, JH1-OH 3.8 and JH1-H2 7.6 Hz, H1), 4.74 (1 H, d, JOH-H1 3.8 Hz, 1-OH), 4.83 (1 H, d, JH2-H1 7.4 Hz, H2), 5.45 (2 H, m, H7 and H10), 5.58 (1 H, d, JH6-H7 4.3 Hz, H6), 5.60 (1 H, s, 3-OH), 6.34 (1 H, s, H12), 7.12 (1 H, d, JOH-H10 4.9 Hz, 10-OH); 13C NMR d (75.5 MHz, acetone-d6) 8.12, 28.99, 32.83, 42.67, 49.44, 64.34, 68.18, 70.09, 74.70, 75.51, 84.37, 86.09, 92.23, 99.40, 110.46, 119.08 (q, JC-F 319 Hz), 169.85, 172.93, 176.29; 19F NMR d (94.1 MHz, acetone-d6) 71.03 (s, 3 F); Calc. for C21H23O13SF3: C, 44.06; H, 4.05. Found: C, 44.34; H, 4.35%. 7-a-O-(4-methylphenyl) ginkgolide C (4). A suspension of sodium hydride (80%, 0.175 g, 5.83 mmol) in dry dimethylformamide (10 mL) was cooled to 01C under argon bubbling. A solution of 4hydroxytoluene (0.472 g, 4.36 mmol) in dry dimethylformamide (5 mL) was slowly added to the suspension. The reaction mixture was stirred at 201C for 3 h and cooled to 01C. A solution of 3 (2.0 g, 3.49 mmol) in dry dimethylformamide (10 mL) was slowly added to the reaction mixture, stirred at 201C for 4 h, and warmed at 501C for 30 min. The reaction mixture was cooled to 01C and hydrolysed with water (25 mL). The aqueous layer was extracted with ethyl acetate (2 50 mL) and the organic layer was washed with brine (75 mL) and dried over MgSO4. Distillation under reduced pressure produced a brown oil that was dissolved in 1:1 methyl-t-butylether:heptane (50 mL) to give a drab suspension that was filtered. The resulting solid was washed with heptane (2 15 mL) and purified by preparative HPLC eluting with heptane/ethyl acetate (75:25) with a 100 mL/min flow rate. Compound 4 (0.593 g, 32%) was obtained as a white powder, m.p. 3391C (dec.); 1H NMR d (300.1 MHz, acetone-d6) 1.31 (3 H, d, JH16-H14 7.2 Hz, 16-Me), 1.34 (9 H, s, CMe3), 2.20 (1 H, d, JH8-H7 3.2

Hz, H8), 2.32 (3 H, s, Ar-Me), 3.10 (1 H, q, JH14-H16 7.0 Hz, H14), 4.30 (2 H, m, H1 and 1-OH), 4.63 (1 H, d, JH2-H1 7.6 Hz, H2), 5.20 (1 H, d, JH7-H8 3.4 Hz, H7), 5.22 (1 H, s, H6), 5.46 (1 H, s, 3-OH), 5.51 (1 H, d, JH10-OH 5.5 Hz, H10), 6.31 (1 H, s, H12), 6.58 (1 H, d, JOH-H10 5.5 Hz, 10-OH), 7.05 (2 H, d, J 8.9 Hz, o-Ar), 7.21 (2 H, d, J 8.3 Hz, m-Ar); 13C NMR d (75.5 MHz, acetone-d6) 8.10, 20.44, 30.83, 33.84, 42.86, 53.45, 70.03, 70.06, 72.47, 75.27, 79.16, 82.87, 84.58, 91.64, 99.25, 112.10, 115.99, 131.13, 131.90, 154.96, 170.33, 173.54, 176.52; Calc. for C27H30O11: C, 61.13; H, 5.70. Found: C, 60.88; H, 5.71%. Lipophilicity Measurements Lipophilicity was expressed either as the logarithm of the capacity factor in pure water (log kw) or as the n-octanol/water partition coefficient (log Poct). The log kw values were determined using reversed-phase HPLC according to standard procedures [van de Waterbeemd et al., 1996; Lombardo et al., 2000]. HPLC analyses were performed at 221C and pH 5.1 using a Merck Hitachi liquid chromatograph (Merck KgaA, Darmstadt, Germany and Hitachi Instrument, Inc., San Jose, CA) equipped with a L-7200 auto sampler, a L-7614 degasser, and a L-7360 column oven. Detection was performed with a L-7400 UV detector operating at 220 nm. The chromatographic system was controlled by a D-7000 Interface module with the D7000 HPLC System Manager Program version 4.1 (Merck Hitachi). Retention time measurements were performed on a Discovery RP Amide C16 5 mm, 150 4.6-mm column (Supelco, Bellefonte, PA). In these experiments, uracil was used as a reference unretained compound. The mobile phases were mixtures of buffer at pH 5.1 (49% of citric acid 0.1 M and 51% of Na2HPO4 0.2 M) and methanol in varying proportions from 30 to 60% (vol/vol). The flow rates ranged from 0.7 mL/min to 1.5 mL/min according to the ginkgolides lipophilicity. The aqueous portion of the mobile phase was filtered through a 0.45-mm HA Millipore filter (Millipore, Milford, MA) and for each ginkgolide four percentages of methanol in the buffer were tested. Stock solutions (103 M) of ginkgolides were prepared in methanol and diluted to 104 M in mobile phase for injection (20 mL). All samples were injected at least three times for each mobile phase. In order to increase statistical precision, log kw values at 100% of buffer were calculated by linear extrapolation (isocratic log k versus percentages of MeOH in mobile phase) using all isocratic measurements (at least 12) for each compound. The log Poct values for tested ginkgolides were calculated from log kw values using the correlation


equation of Liu [2003] (Eq. 1). log Poct ¼ 1:13ð0:06Þ log kw þ 0:027 ð0:12Þ n ¼ 45; r2 ¼ 0:97; s ¼ 0:18; F ¼ 1; 445

ð1Þ

Experimental log kw and log Poct values as determined above were compared to the theoretical partition coefficients as calculated with the CLOGP fragmental system (CLOGP Medchem Software version 4.81, Medicinal Chemistry Project, Pomona College, Claremount, CA, 2002). BIOLOGY

Inhibition of Platelet Aggregation In Vitro The determination of anti-PAF activities of GkB, 1, 2, and 4 were determined on isolated platelets prepared from New Zealand rabbits weighing 2.5–3.5 kg (ESD, Chatillon, France) as has been previously described in detail [Pietri et al., 2001]. Six different concentrations were tested (in quadruplicate) for GkB, 1, 2, and 4 (0.1–100 mM). The concentration corresponding to 50% inhibition of platelet aggregation (IC50) was determined by plotting the percent inhibition against the concentration of ginkgolide. Perfusion Conditions and Hemodynamic Measurements For perfusion experiments, all reagents were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France) and were of the highest purity available. Male Wistar rats (340–370 g) purchased from CERJ (Le Genest St. Isle, France) were anesthetized with sodium pentobarbital (50 mg/kg, ip), and after thoracotomy, the heart was excised and immediately placed in cold buffer (41C) until contractions ceased. The aorta was then cannulated and retrograde Langendorff perfusion was initiated at 371C at a constant pressure of 95 cm H2O. The perfusion fluid was a modified Krebs-Henseleit bicarbonate buffer, filtered through a 0.22-mm Millipore filter, and containing 25 mM NaHCO3, 119 mM NaCl, 4.6 mM KCl, 1.2 mM KH2PO4, 1.2 mM MgSO4, 2.5 mM CaCl2, 0.5 mM EDTA, and 10 mM glucose (pH 7.35 adjusted by gassing with 95% O2, 5% CO2). A water-filled latex balloon was inserted into the left ventricle and inflated until left-ventricular end-diastolic pressure (LVEDP) was in the range of 8–12 mm Hg. The balloon was connected to a pressure transducer (Gould Statham P23) to monitor the contractile function. The following hemodynamic parameters were monitored every 5 min: left ventricular developed pressure (LVDP) measured as the difference between systolic and end-diastolic pressures, its first derivative with time (dP/dt), heart rate, and rate pressure product (RPP) calculated as the

161

product of LVDP and heart rate. Coronary flow was measured by collecting the coronary perfusate over 1 min. In order to determine a nontoxic concentration window for the in vitro perfusion of 4, four groups of six hearts were studied under normoxic conditions. After an initial 30-min stabilization period, a 45-min normoxic perfusion was extended with buffer containing different concentrations of 4, i.e., 0.20 mM; 0.70 mM, and 1.0 mM, or with buffer alone (untreated group). These test concentrations were chosen upon considering the high bioavailability of structurally related GkA, Gk and, to a lesser extent, 1, after oral ingestion of EGb 761 in humans and their respective percentages in the extract [DeFeudis, 1998]. Because none of the tested concentrations of 4 significantly altered hemodynamic parameters as compared to untreated hearts, the perfusion concentration selected for ginkgolides 1, 2, and 4 in ischemia– reperfusion experiments was 0.70 mM. For this purpose, five groups of hearts (n ¼ 12–16/group) were used. Hearts were equilibrated for 30 min, then subjected to 30 min of no-flow normothermic global ischemia and 60 min of reperfusion. Three groups of animals were used to evaluate the effects of in vitro application of 0.70 mM of 1, 2, and 4 during ischemia and reperfusion (groups 1, 2, and 4), each ginkgolide being added to the perfusion buffer during the last 20 min of control perfusion and the first 15 min of reperfusion. Two additional groups of animals were used to evaluate the effects of in vitro perfusion of 1 and 2 during the first 15 min of reperfusion only (groups 1REP and 2REP). The effect on cardiac ischemia–reperfusion of a 20-day per os administration of ginkgolides was assessed in two groups of 20 rats that were given 1 or 2 (4 mg in 0.2 mL of vehicle/kg/day; groups 1PO and 2PO, respectively) and compared to that obtained in a group of 20 rats having received vehicle alone (placebo group). In these latter experiments, vehicle was a mixture of corn oil and water (1:1 vol/vol). After sacrifice, hearts were perfused with normal buffer according to the experimental ischemia–perfusion protocol described above. The dose of 4mg/kg/day was selected because terpenes account for about 6% of the oral dose of 60 mg/kg/day of EGb 761 previously used in rats [Pietri et al., 1997], approximately corresponding to the therapeutic dose of 240 mg/kg/ day of EGb 761 in humans [DeFeudis, 1998]. At the end of reperfusion, hearts from each group were randomly allocated for ATP content and enzymatic activities determination, or for malondialdehyde (MDA) tissue assay (n ¼ 6/group for each measurement, see below).

162

BILLOTTET ET AL.

Biochemical Analyses in Freeze-Clamped Hearts At the end of reperfusion, six hearts randomly taken from the in vitro and per os groups described above were freeze-clamped with aluminum tongs precooled in liquid nitrogen and the frozen tissue was pulverized and used for enzymatic activities determination and myocardial ATP content. Creatine kinase (CK) and lactate dehydrogenase (LDH) activities were determined according to standard procedures (see [Pietri et al., 2001]); ATP content in heart tissue was determined by an HPLC procedure according to Takeo et al. [2000]. Protein determination was performed as previously described [Lowry et al., 1951]. The effect of ginkgolides 1, 2, or 4 on the enzymatic activities and/or the tissue ATP content at different stages of the ischemia–reperfusion protocol described above has been determined in 12 additional groups of six hearts from rats having received ginkgolides in vitro or per os at the concentrations or oral doses defined above. Freeze-clamping of hearts was performed either after the initial 30-min control period or after the 30-min global ischemic period. Cardiac Tissue MDA Assay At the end of reperfusion, 6 hearts from each group were allocated for MDA assay as an index of oxygen radical–induced membrane lipid peroxidation. For this purpose, the left ventricle was removed and homogenized in 9 vol of 1.15% KCl solution and centrifuged (1,000g for 10 min at 41C) in the presence of 10 mM of deferoxamine, 0.04% butylated hydroxytoluene, and 2% of ethanol to prevent auto-oxidation of the sample. The MDA content of the supernatant was measured by a modification [Ambrosio et al., 1991] of the thiobarbituric acid (TBA) method [Okhawa et al., 1979]. After addition of 0.6 mL of sodium dodecyl sulfate (8.1%) to 0.4 mL of heart homogenate supernatant, the mixture was treated with 3 mL of 20% acetic acid and 3 mL of 0.8% TBA, and incubated at 951C for 90 min. After cooling, the sample was extracted with 2 mL of butanol/pyridine (15:1) and the absorbance of the MDA–TBA complex was determined at 532 nm. The effect of ginkgolides 1, 2, or 4 on the cardiac MDA–TBA content at the end of the normoxic period was assessed in six additional groups of hearts (n ¼ 6/ group) of rats that had received ginkgolides in vitro or per os at the concentrations or oral doses defined above. MDA–TBA content was determined after the initial 30-min control period. The cardiac tissue MDA content is expressed as nanomoles of MDA–TBA per mg of protein [Ambrosio et al., 1991]. This study was conducted in conformity with the Guide for the Care and Use of Laboratory Animals,

U.S. Department of Health and Human Services, N.I.H. Publication No. 85-23, revised (1985). Statistical Analysis Hemodynamic and biochemical data (absolute values, or percent recovery of preischemic values) are presented as means7standard error of mean (s.e.m.). Statistical analysis was performed using a two-way analysis of variance (ANOVA) with repeated measures. If a difference was found, experimental groups were compared further using the Newman-Keuls test. A one-way ANOVA was then carried out to test for differences among the mean values of all groups at every time, and this was followed by an appropriate a posteriori multiple comparison test. Differences were considered significant when Po0.05. RESULTS

Chemistry Synthesis Synthesis of 4 was conducted in two steps (Fig. 1B) as previously described by Cazaux et al. [1995]. In a first step, the activation of the 7-hydroxyl group was performed by treating 1 with trifluoromethanesulfonic anhydride in pyridine and led to the 7-O-trifluoromethanesulfonyl derivative (3) in a good yield (91%). In a second step, the nucleophilic substitution on the C7 carbon was performed by reaction of 3 with in situ formed sodium paracresolate in dimethylformamide to yield 4 in a 32% yield. Log kw Measurement and Log Poct Determination The experimental (at 221C and pH 5.1) and calculated values of the lipophilicities of GkB 1, 2, and 4 are given in Table 1. Because of the statistical quality of the calibration regression line (r2 ¼ 0.97), the errors on estimated log Poct are larger than that on log kw. A pH value of 5.1 was chosen because GkB was found to exist only as a trilactonic form (non-ionized species) at this pH [Zekri et al., 1996; Lang and Wai, 1999]. According to their chemical structures, it is reasonable to assume for 1, 2, and 4 a similar behavior regarding the opening of lactone rings. Biology Inhibition of Platelet Aggregation In Vitro GkB (0.3–10 mM), 1 (1–100 mM), 2 (1–120 mM), and 4 (0.1–100 mM) were tested for their ability to inhibit rabbit platelet aggregation in vitro in the presence of 0.02 mM PAF. The IC50 concentrations required to inhibit platelet aggregation were 2.2 mM, 17.1 mM, and 8.2 mM for GkB, 1, and 4, respectively. In platelet samples treated with 120 mM of 2, the


percentage of PAF-induced platelet aggregation inhibition was 4.871.2%. With physiologically relevant concentrations of 2 (1–100 mM), PAF-induced platelet aggregation inhibition was not significant (0.1–1.5%).

TABLE 1. Lipophilicity of Ginkgolides B, C (1), and of 7-O-(4Methylphenyl) Ginkgolide C (2) and 7-a-O-(4-Methylphenyl) Ginkgolide C (4) log kaw

log Pboct

CLOGPc

1.1570.03 1.6570.03 3.6470.07 3.4270.05

1.5870.13 2.1570.15 4.4070.31 4.1570.28

2.48 1.45 0.60 0.60

Ginkgolide 1 Gk B 2 4 a

Measured retention factors extrapolated to 100% of water by linear regression (nZ12). b Estimated partition coefficients by Eq. 1. c Calculated by CLOG P Medchem Software version 4.81.

163

Evaluation of the Cardioprotective Effect of Ginkgolides In preliminary experiments, hearts submitted to a 45-min normoxic perfusion in the presence of increasing concentrations of 4 (0.20–1.0 mM) showed no significant alterations of cardiac function (data not shown). In a previous study [Pietri et al., 2001], 1 and 2 were tested up to 0.9 mM using the same normoxic perfusion protocol, and no cardiac dysfunction was evidenced; however, because 0.35 mM of 1 failed to protect ischemic–reperfused hearts, a concentration of 0.70 mM was selected for 1, 2, and 4 in the present studies. In perfusion experiments with 2 and 4, actual concentrations were assessed by HPLC with UV detection at 220 nm [Pietri et al., 2001]. Table 2 shows the hemodynamic parameters after 30 min of preischemic control period and after 60 min of reperfusion. None of the three ginkgolides 1, 2, and

TABLE 2. Effects of Ginkgolides C (1), 7-O-(4-Methylphenyl) Ginkgolide C (2) and 7-a-O-(4-Methylphenyl) Ginkgolide C (4) on Hemodynamic Parameters of Ischemic-Reperfused Isolated Rat Heartsa Group

LVEDP

HR

LVDP

RPP

dP/dt

Coronary flow

(mm Hg/s)

(mL/min/g wet wt)

3,981763 4,007753 3,806758 3,645785 3,902769 3,695773 3,741768 3,960751

15.470.5 17.170.6 15.570.5 16.670.4 14.970.7 16.170.6 17.070.4 15.270.6

At the end of the preischemic 30-min control period

Placebo 1 2 4 1REP 2REP 1PO 2PO

(mm Hg)

(beats/min)

(mm Hg)

10.471.2 9.771.0 9.671.4 11.570.9 12.071.1 9.671.0 10.071.4 9.871.0

238.875.5 234.778.0 242.378.0 258.075.4 238.876.1 236.177.1 244.477.5 233.976.9

121.573.7 130.274.2 118.873.8 116.672.7 115.272.6 116.975.6 117.874.3 124.274.1

(mm Hg beats/min) 29,1557817 29,83071005 28,5497970 29,9807496 27,7507998 27,2737873 28,5657558 28,9977963

After 60 min reperfusion (mmHg) Placebo 1 2 4 1REP 2REP 1PO 2PO a

73.774.0 53.473.5nyw 25.472.9n 37.272.1ny 68.573.6zy 63.972.3y 70.874.0zyw+ 50.672.6nyw

(% of PIV) 55.8710.6 72.476.5n 63.578.3 65.5711.9n 74.974.3n 77.577.3n 43.274.4z+ 81.173.3ny

(% of PIV) 26.173.4 35.874.0y 60.576.0n 44.074.3ny 30.974.3y 36.373.8ny 27.774.8z+ 62.877.1n

(% of PIV)

(% of PIV)

(% of PIV)

19.172.9 27.272.9ny 45.974.6n 33.773.6ny 23.572.1y 23.973.1y 17.972.7zy+ 53.276.9n

25.474.2 36.274.2ny 49.377.4n 32.674.7y 30.172.3y 34.172.7ny 23.672.2zy+ 50.373.5n

36.873.3 40.271.9 46.176.9n 46.874.9n 33.673.8y 42.471.9 42.474.9 48.172.8n

Perfusion protocol: see Materials and Methods. All chemical structures are indicated in Fig. 1. Ginkgolides 1, 2 and 4 (0.70 mM) were perfused during the last 20 min of control period and the first 15 min of reperfusion (groups 1 (n=14), 2 (n=16) and 4 (n=13)) or during the first 15 min of reperfusion only (groups 1REP (n=13) and 2REP (n=14)). Hearts from groups 1PO and 2PO (n=12), which were taken from animals having received 1 or 2 per os (4 mg in 0.2 mL of vehicle/kg/day during 20 days), respectively, were perfused with buffer throughout. One group of rats received 0.2 mL of vehicle orally during 20 days before sacrifice and hearts were perfused with buffer throughout (placebo group; n=20). Data are presented as means7s.e.m. LVEDP, left ventricular end-diastolic pressure; LVDP, left ventricular developed pressure; RPP, rate pressure product and dP/dt, first derivative of left ventricular pressure with time. PIV, preischemic value taken at 30 min control period. Statistics: One-way ANOVA (Po0.01) followed by Duncan test: nPo0.05 vs. placebo group; zPo0.05 vs. group 1; yPo0.05 vs. group 2; wPo0.05 vs. group 4, +Po0.05 vs. group 2PO.

164

BILLOTTET ET AL.

4 caused significant changes in overall hemodynamic parameters during the last 15 min of the normoxic preischemic period, as compared to the placebo group. The initiation of global ischemia was rapidly followed by a dramatic impairment of contractile activity of the hearts, with systolic oscillations stopping completely after 3–5 min of ischemia in all groups. Ventricular contracture on the balloon (which was held at a constant volume of 60–70 mL) assessed by diastolic pressure measurements (e.g., LVEDP), which ranged from 9.6 to 12.0 mm Hg during the control period, gradually increased in all groups after 10 min of ischemia. However, the time to onset of ischemic contracture was significantly delayed by at least 5 min by 2, given in vitro or in vivo and by 4 as compared to the placebo group, whereas 1 given in vitro was less effective (Fig. 2). Figure 2 shows that after 15 min of ischemia, the placebo and 1PO hearts reached a peak of contracture with LVEDP values corresponding to about 40% of preischemic LVDP. After peak of contracture, no significant relaxation of tension occurred in placebo and 1PO groups. This trend was markedly different in groups 2, 2PO, and 4, where

Fig. 2. Effect of treatment with ginkgolides on the increase of left ventricular end-diastolic pressure (LVEDP) of isolated rat hearts subjected to 30 min of no-flow ischemia. Drugs were added to the perfusion medium at 0.70 mM during the 20 last min of preischemic control period: GkC (1; &); 7-O-(4-methylphenyl) ginkgolide C (2;~); 7-a-O-(4-methylphenyl) ginkgolide C (4; !), or were given orally during 20 days before perfusion of the isolated heart with buffer: GkC (1PO;!); 7-O-(4-methylphenyl) ginkgolide C (2PO;K). Hearts from the placebo group (J) received buffer throughout. Symbols represent means7s.e.m. (n ¼ 12–20/group). Statistics (at 15 and 30 min of ischemia): *Po0.05 vs. placebo group, yPo0.05 vs. 2, +Po0.05 vs. 2PO; by one-way ANOVA followed by Duncan test.

the peak contracture only appeared after 20 min (Fig. 2), giving a significantly less severe contracture than placebo or 1 and 1PO-treated hearts after 15 min of ischemia. No significant fall-off of tension was seen after the peak contracture in groups 2, 2PO, and 4 (Fig. 2). Upon terminating ischemia, hearts treated by 2 reached LVEDP values (in mm Hg) of only 30.471.4 (group 2) and 33.471.2 (group 2PO), whereas values for groups 4, 1, and 1PO hearts raised to 39.671.8, 41.871.2, and 43.171.6, respectively, not significantly different from that seen in the placebo group (44.171.2). Initiation of reperfusion caused a marked elevation of contracture in all groups, LVEDP peaking at 99.973.8 mm Hg during the first 5 min of reperfusion for placebo hearts. In parallel, myocardial dysfunction and severe arrhythmias were observed in these hearts with a concomitant decrease in coronary flow (an index of coronary vascular resistance) to 35–44% of preischemic values. After 10 min, heart function recovered to relative stability in all groups, but hearts that received placebo still remained in intense contracture after 60 min of reflow. With respect to the placebo group, reperfusion-induced early elevation of LVEDP was significantly limited by preischemic infusion of 2, and to a lesser extent by 4 or 1 (data not shown). In contrast, in vitro treatments with 1 or 2 only at reperfusion (groups 1REP and 2REP) were not active in protecting hearts from ventricular contracture. As in the ischemic phase, the in vivo treatment in groups 1PO and 2PO showed a discrepancy between the two ginkgolides in reducing ventricular postischemic contracture, 2 being strongly protective while 1 was ineffective. It should be noted that the contracture limitation achieved when 2 was given in vivo was about 50% of the protection provided when 2 was infused in vitro at 0.70 mM. As expected regarding the behavior of contracture, coronary flow in the placebo group was poorly restored. In vitro application of 2 or 4, and in vivo treatment with 2, significantly enhanced the recovery of coronary flow, whereas postischemic in vitro treatments (groups 1REP and 2REP) and 1 given in vivo were ineffective in reversing postischemic vascular damage. Mechanical performance (i.e., HR, LVEDP, RPP) was markedly depressed at early reperfusion times in all groups and showed a progressive recovery over the entire period of reflow (see Fig. 3 for RPP). The same pattern of recovery was seen for dP/dt, an index of contractility. These hemodynamic indices were significantly improved by perfusion or oral administration of 2, by perfusion of 4 and, to a lesser extent, by perfusion with 1. Groups 1REP, 2REP, and 1PO hearts poorly recovered (Fig. 3).


165

Biochemical Analyses in Freeze-Clamped Hearts and Coronary Effluents After 30 min of normoxic control, no statistical difference was found between the hearts from the placebo and the treated groups regarding myocardial ATP content (Fig. 4) and CK and LDH activities

(Table 3). After 30 min of ischemia, a marked decrease of ATP content was observed in all groups, as compared to that of normoxic hearts, reaching 5% of the corresponding preischemic value (PIV) for the placebo group. Interestingly, ATP was significantly preserved during ischemia in both groups of hearts

Fig. 3. Effect of treatment with ginkgolides on the time course of recovery of rate pressure product of isolated rat hearts having undergone 30 min of no-flow ischemia. Drugs were added to the perfusion medium at 0.70 mM during the last 20 min of preischemic control and the 15 first min of reflow: GkC (1; &); 7-O-(4methylphenyl) ginkgolide C (2;~); 7-a-O-(4-methylphenyl) ginkgolide C (4;!), or at reflow only: GkC (1REP; &); 7-O-(4-methylphenyl) ginkgolide C (2REP; ~), or were given orally during 20 days before perfusion of the isolated heart with buffer: GkC (1PO;!); 7-O-(4methylphenyl) ginkgolide C (2PO; K). Hearts from the placebo group (J) received buffer throughout. Symbols represent means7s.e.m. (n ¼ 12–20/group). Statistics (over the last 30 min of reperfusion): *Po0.05 vs. placebo group, yPo0.05 vs. 2, +Po0.05 vs. 2PO; by twoway ANOVA followed by Newman-Keuls test.

Fig. 4. Effect of ginkgolides on the myocardial ATP content at different stages of ischemia–reperfusion. For drug administration, perfusion protocol, and experimental groups, see Fig. 3 legend. Symbols represent means7s.e.m. (n ¼ 12–20/group). Statistics: *Po0.05 vs. placebo group, yPo0.05 vs. 2, +Po0.05 vs. 2PO; by one-way ANOVA followed by Duncan test.

166

BILLOTTET ET AL.

TABLE 3. Effects of Ginkgolide C (1), 7-O-(4-methylphenyl) Ginkgolide C (2) and 7-a-O-(4-Methylphenyl) Ginkgolide C (4) on Myocardial Tissue Creatine Kinase and Lactate Dehydrogenase Activities of Normoxic and Ischemic-Reperfused Isolated Rat Heartsa Normoxic hearts Experimental group Placebo 1 2 4 1REP 2REP 1PO 2PO

Ischemic-reperfused hearts

CK (IU/mg protein)

LDH (IU/mg of protein)

CK (IU/mg of protein)

LDH (IU/mg of protein)

12.0270.21 12.0870.27 12.2670.35 11.7670.20 11.9970.31 11.9870.33 12.1170.32 11.9570.12

3.9970.15 4.0670.17 4.1270.21 4.0270.25 3.8970.23 4.0370.11 3.9370.19 4.0570.17

6.1270.19 6.6170.14 7.8970.11nzw 6.8270.18n 6.3770.10y 6.9370.11ny 6.2770.14y 8.5870.11nzw

2.9970.16 3.1170.21 3.8170.09nzw 3.1470.17 3.0870.11y 3.2470.17ny 2.9870.15y 3.9470.17nzw

a Data are presented as means7s.e.m. (n=6 in each group). CK, creatine kinase; LDH, lactate dehydrogenase; IU, international units. Statistics: one-way ANOVA (Po0.01) followed by Duncan test: nPo0.05 vs. placebo group; zPo0.05 vs. group 1; yPo0.05 vs. group 2; wPo0.05 vs. group 4. Groups are described in the footnote of Table 2. Drug treatments and the perfusion protocol are detailed in the Experimental section.

treated with 2 (i.e., 14% of PIV in group 2 and 16% of PIV in group 2PO) than in all other groups (Fig. 4). At the end of 60-min reperfusion, ATP content and enzymatic activities were still better preserved in all hearts treated with 2 and 4 (Fig. 4 and Table 3). For all the biochemical indices that were considered, 1 given in vitro or in vivo was not effective. Cardiac Tissue MDA Assay Cardiac tissue MDA-TBA content was assessed in hearts undergoing ischemia and reperfusion as an index of the antioxidant properties of ginkgolides. In normoxic hearts, the tissue MDA-TBA content from all groups ranged similarly at 0.95–1.11 nmol/mg of protein, showing that there was no preischemic treatment interference with control baseline MDA–TBA measurement (Fig. 5). After 60 min of reperfusion, increases in membrane lipid peroxidation resulted in accumulation of MDA–TBA in placebo reperfused hearts, reaching 1.70 nmol/mg of protein. This ischemia–reperfusioninduced elevation of tissue MDA-TBA was significantly reduced in hearts from groups 2 (30%; Po0.01), 2REP (15%; Po0.05) and 2PO (23%; Po0.05) and group 4 (16%; Po0.05) as compared to the placebo group. In groups 1 and 1PO, MDA–TBA accumulation was not reduced, as compared to either placebo hearts or hearts treated with 2 or 4. DISCUSSION

Chemistry As a consequence of increased steric hindrance in its environment, the C7-hydroxyl group is the least reactive of the secondary hydroxyl groups of 1 [Corey et al., 1992; Weinges and Schick, 1991; Weinges et al., 1993]. However, selective oxidation or acetylation of the C7-hydroxyl group of 1 without protection of the other

secondary hydroxyl groups have been reported [Corey et al., 1992; Jaracz et al., 2002]. It is expected that epimerization at the C7 position of 1 would occur under the conditions of the SN2 reaction provided the C7-OH group was activated by prior conversion to a better leaving group (Fig. 1). This was achieved by introducing a trifluoromethanesulfonyl group on the C7 carbon using a modification of a method previously described in the case of the conversion of 1 to ginkgolide B [Cazaux et al., 1995]. Treatment of 1 with trifluoromethanesulfonic anhydride in pyridine yielded 3 in a good yield and, as expected owing to steric hindrance of the C8-tert-Bu group, reaction of 3 with sodium paracresolate afforded 4, the epimer of 2, with a relatively poor yield. Steric hindrance of the neighboring tert-butyl group of the ginkgolide structure has been reported to cause poor yields in the synthesis of a variety of C7-substituted compounds [Vogensen et al., 2003]. 1H NMR analysis of the important changes in chemical shifts and coupling constants of the H7 in 4 (5.20 ppm; J7.8 ¼ 3.4 Hz) with respect to that in 2 (4.65 ppm; J7.8 ¼ 12 Hz, Pietri et al. [2001]) provided additional proof of the inverted configuration at C7 in 4. Lipophilicity of Ginkgolides Reversed-phase HPLC was chosen to measure lipophilicity of ginkgolides because of several advantages (rapidity, accuracy, economy of compounds and solvents, possibility to measure the lipophilicity of parent compounds even in a presence of small amounts of products resulting from the opening of ginkgolides lactone rings). This analytical technique has been used to quantify lipophilicity [van de Waterbeemd et al., 1996; Lombardo et al., 2000]. Table 1 confirms that introducing a potentially less hydrophilic moiety than the OH group at C7 increases the lipophilicity of


167

CLOGP prediction, were compared. Table 1 shows that the difference between estimated log Poct and CLOGP prediction is very large (3.55 to 4.06 depending on the tested molecule), suggesting that the CLOGP method, based on 2D chemical structure only, is not able to predict the lipophilicity for ginkgolides. Indeed, for these highly crowded structures, lipophilicity is largely modulated by 3D topology and, especially, by the diminution of the accessibility of polar moieties present in the ginkgolide core for hydrogen bonding with water. Of interest is the result that the difference in lipophilicity between Gk B and 1 is much lower when obtained experimentally (i.e., from log kw or log Poct) than that calculated from the CLOGP method by adding an aliphatic alcohol increment at C7 position (Table 1). Thus, it may be proposed that the potential for the hydrophilic character of the OH substituent at C7 of 1 is decreased by intramolecular hydrogen bonding with the cis lactone rings (Fig. 1). The data in the present study suggest that reversed-phase HPLC is a very reliable method to determine the lipophilicity of poorly water soluble ginkgolides. Biology

Fig. 5. Effect of ginkgolides on the myocardial malondialdehyde– thiobarbituric acid content at different stages of ischemia–reperfusion. For drug administration, perfusion protocol, and experimental groups, see Fig. 3 legend. Symbols represent means7s.e.m. (n ¼ 12–20/ group). Statistics: *Po0.05 vs. placebo group, yPo0.05 vs. 2, + Po0.05 vs. 2PO; by one-way ANOVA followed by Duncan test.

ginkgolides. Interestingly, the precision of log kw measurements allows a clear demonstration that 4 is less lipophilic than its epimer, 2. This difference in lipophilicity may be attributed to a hydrophobic collapse between tert-butyl and the p-tolyl substituents in 2 (see [Tsai et al., 1993]). To better understand the relationship between structure and lipophilicity for ginkgolide derivatives, two other widely used estimates for lipophilicity, i.e., n-octanol/water partition coefficients (log Poct) and

Because GkB is the most potent naturally occurring PAF antagonist, a number of derivatives have been prepared. Structural modifications at the secondary C1 and/or C10 sites have afforded GkB derivatives with significantly improved anti-PAF activities [Corey and Gavai, 1989; Chen et al., 1998; Hu et al., 1999, 2000; Stromgaard et al., 2002]. The relevance of this C1/C10 synthetic strategy to increase anti-PAF activity of ginkgolides has been reemphasized by the finding that 1, a naturally occurring GkB derivative substituted at C7, is a weak PAF antagonist. A possible explanation for this was that PAF antagonism is potentiated by structural factors strengthening the linkage of the tert-butyl group of the ginkgolide with the hydrophobic area of the PAF receptor membrane and that such positive interactions are impaired by the presence of a polar substituent, e.g., OH at the C7 position of 1 [Braquet, 1987; Braquet et al., 1991]. In apparent agreement, Vogensen et al. [2003] recently prepared the epimer of 1 and found that inversion of configuration at C7 caused a moderate increase in anti-PAF activity as compared to that of 1, concluding that stereochemistry at C7 plays a minor role in the anti-PAF properties of ginkgolides. Jaracz et al. [2002] and Vogensen et al. [2003] also reported that 7a-O-Ac-GkC and 7b-O-Ac-GkC, the two acetyl diastereoisomers of 1, show similar anti-PAF antagonism to 1. Together, these data [Braquet, 1987; Braquet et al., 1991; Jaracz et al., 2002; Vogensen et al., 2003]

168

BILLOTTET ET AL.

suggest that local electronic effects at C7 of ginkgolides determine their anti-PAF behavior. However, as a typical receptor membrane interaction (i.e., sensitive to lipophilicity), PAF antagonism by ginkgolides should also depend on steric factors. Pietri et al. [2001] found that 2, retaining the R stereochemistry at C7 of 1, had no measurable anti-PAF effect. Because the lipophilic character of 2 is 2.8 units of log Poct higher than that of 1 (Table 1), it is proposed that the lack of anti-PAF effect of 2 has a steric rather than electronic origin. The finding that 4 is active against PAF, although slightly less lipophilic than its nonactive epimer 2 (Table 1), supports a novel role for steric interactions on PAF antagonism in C7-substituted ginkgolides. Thus, lipophilicity cannot be considered as a unique parameter to control in the design of synthetic ginkgolides of improved anti-PAGF activities, as previously stated [Braquet et al., 1991]. The nature of the steric interactions increasing the anti-PAF effect in ginkgolides remains to be investigated, and it should be observed that a much better interaction with PAF membrane receptor was obtained in 4 versus 2 where the bulky Oaryl ether group at C7 is in cis position with respect to the C8-tert-Bu (Fig. 1). An effect on PAF superior to that shown here by 4 has been found for cis-C7-substituted halogen derivatives of 1 [Vogensen et al., 2003], suggesting that stereochemical, steric, and electronic effects synergize in anti-PAF antagonism of ginkgolides. An increasing body of evidence supports that ginkgolides may represent a promising alternative in the pharmacological treatment of cardiovascular disease [DeFeudis, 1998]. Initially, the potential of ginkgolides as cardioprotectants has been investigated via the recognized activity of GkB as an anti-PAF agent [Berti et al., 1989; Koltai et al., 1989; Lagente et al., 1989; Wainwright and Parratt, 1989]. The involvement of PAF in cardiovascular pathophysiological processes is well established [Montrucchio et al., 2000]. Considerable amounts of PAF are released from rabbit isolated heart during early postischemic reperfusion [Montrucchio et al., 1989; Berti et al., 1990; Katoh et al., 1993], which may induce myocardial dysfunction by triggering platelet/neutrophil cooperation [Alloati et al., 1992]. PAF is also produced in isolated guinea pig hearts after exposure to a free radical generator, dihydroxyfumaric acid [Alloati et al., 1994]. In these studies, the PAF-induced deleterious effects on myocardial function have been reversed by the use of antiPAF molecules, including GkB, applied at concentrations in the 3–30 mM range [Berti et al., 1989; Koltai et al., 1989; Montrucchio et al., 1989; Wainwright and Parratt, 1989; Alloati et al., 1992, 1994]. In the present study, the earlier [Pietri et al., 2001] concept of perfusing a ginkgolide to ischemic–

reperfused rat hearts at a low concentration that should not provide any significant anti-PAF effect has been reinvestigated using two functionally equivalent ginkgolides 2 and 4 that differ structurally at one stereocenter. The perfused concentration of ginkgolides was 0.70 mM, one order of magnitude lower than the IC50 value of 4, and negligible in terms of PAF interactions given that 2 showed no anti-PAF effect up to 120 mM. Under these conditions, however, 2 was significantly more effective than 4 in improving standard indices of the outcome of postischemic functional and metabolic recovery. These data reinforce our previous assumption [Pietri et al., 2001] that cardioprotection by ginkgolides has a component unrelated to PAF antagonism. To further investigate the mechanisms involved in the PAF-independent cardioprotective effects of ginkgolides, 2 and 1 (which is an even weaker anti-PAF agent than 4) were added at 0.70 mM in the perfusate of hearts at time of reperfusion only. Because the results for the respective groups 1REP and 2REP show that this administration mode was inefficient in improving postischemic recovery, it may be inferred that a significant part of the protective effect of ginkgolides is intimately linked to their presence into the myocardium during the ischemic phase. During ischemia, the groups pretreated in vitro with 2 or 4 displayed a significant delay of time to onset of contracture together with a reduction in peak contracture (Fig. 2) and a concomitant slower decrease of myocardial ATP levels (Fig. 4), as compared to the placebo and 1-treated groups. Ischemia-induced ATP depletion and the appearance of ischemic contracture are closely correlated, possibly because of the formation of rigor bridges [Ventura-Clapier and Vassort, 1981; Korestsune and Marban, 1990; Steenbergen et al., 1990; Ventura-Clapier and Veksler, 1994]. Other mechanisms have been highlighted, including accumulation of internal calcium and of potentially toxic metabolites [Hearse et al., 1977; Neely and Grotyohann, 1984; Harper et al., 1990; Ventura-Clapier et al., 1994]. Although the relationship between the metabolic events occurring during ischemia and the degree of recovery at reflow are still unclear [Neely and Grotyohann, 1984; Harper et al., 1990; Vanoverschelde et al., 1994], it has been proposed that reducing contracture during ischemia favors functional recovery [Hearse et al., 1977]. The observed improvement of functional and ATP recovery at reperfusion in groups 2 and 4 appears in line with this assumption, and it can be concluded that the major part of the protective effect of ginkgolides exerted during the ischemic phase likely occur via the preservation of ATP levels. In agreement, it has been demonstrated that ginkgolides


A and B protect mitochondrial respiratory activity and ATP decreases in endothelial cells or heart mitochondrial preparation subjected to hypoxic conditions [Janssens et al., 1995, 1999]. Interestingly, in this study [Janssens et al., 1999], lactate production by endothelial cells during hypoxia was inhibited in the presence of ginkgolides, showing that these terpenes do not act as activators of glycolysis but likely by preserving the coupling of mitochondria and the ATP production. Owing to the outstanding intrinsic PAF-unrelated cardioprotective activity of ginkgolides as demonstrated in the present and other [Pietri et al., 1997, 2001] studies, it was of interest to generate insights into their bioavailability. Earlier studies in rat [Biber and Koch, 1999] and human [Fourtillan et al., 1995] showed that GkA and GkB were fully absorbed without any metabolism, yielding high plasma concentrations. In the present study, the effects on isolated ischemic– reperfused hearts from rats that were given the weak PAF antagonists 1 or 2 orally were assessed on functional recovery, energy metabolism, and cellular integrity (groups 1PO and 2PO). These in vivo studies showed a trend similar to that of the in vitro studies, i.e., 2 induced a better global protective effect than the less lipophilic 1, suggesting that 2 should display a satisfactory bioavailability in rat heart tissue after oral ingestion. Regarding 1, oral administration was less effective than in vitro perfusion in protecting ischemic– reperfused hearts, suggesting that this rather hydrophilic ginkgolide is poorly available in heart tissue after oral treatment. To provide more conclusive information on the bioavailability of 2, 4, and GkC in the plasma and the myocardial tissue of rats given the ginkgolides orally, a GC/MS study is ongoing. Another mechanism by which ginkgolides protect the ischemic and reperfused heart may be related to their known antioxidant properties. Using spin-trapping with electron spin resonance detection, GkA and GkB, either given in vitro or perfused in vivo, potently inhibited the postischemic generation of oxygenderived free radical in rat coronary effluents [Pietri et al., 1997; Liebgott et al., 2001]. Because lipid peroxidation is one of the major biochemical consequences of oxygen free radical production in tissues, another rationale of the present study was to examine whether a relevant mechanism for the observed cardioprotection by low doses of ginkgolides could involve antiradicalar/antilipoperoxidant effects. In line with the hemodynamic, biochemical, and metabolic results, postischemic tissue MDA–TBA content was potently decreased by 2 (given in vitro and in vivo) and in a lesser extent by 4, as compared to 1-treated and placebo hearts. An increase of tissue MDA–TBA in reperfused dog or rat hearts has been noted [Ro-

169

maschin et al., 1990; Ambrosio et al., 1991; Pietri et al., 2003], which could be limited by free radical inhibitors. Conversely, Ambrosio et al. [1991] demonstrated that the major amount of MDA–TBA and of conjugated dienes found in isolated reperfused hearts arose from the accumulation of lipid peroxidation products at the level of mitochondria and lysosomal membranes. Therefore, a likely explanation for the observed preservation of ATP levels and decrease of MDA– TBA content by 2 and 4 could be due to the existence of an antioxidant mechanism at the site of mitochondria. Such a mechanism would be effective with only low concentrations of ginkgolides like those used in the present study, whereas efficient direct scavenging of free radicals requires higher doses [Pietri et al., 1997]. Considering that the lipophilicity of 1, 4, and 2 varies similarly as their cardioprotective and antioxidant effects, it may be postulated that ginkgolides can exert a direct interaction with mitochondrial membranes, preventing alterations in fluidity or the denaturation of the respiratory chain during ischemia, protecting the activity of essential enzymes when challenged by free radicals, in particular creatine kinase [Yuan et al., 1992]. As a strategy for a therapeutic limitation of myocardial damage, recent studies have focused on the understanding of mitochondrial dysfunction in ischemic conditions [Suleiman et al., 2001; Bouaziz et al., 2002]. In this context, the determination of the sites of action of ginkgolides within mitochondria is of particular interest. Finally, to unequivocally understand whether the cardioprotective effect could only be ascribed to the lipophilic character/tissue distribution or also to a different antioxidant activity, it would be of great interest to estimate the in vitro anti–free radical properties of the tested ginkgolides in a lipidic-like phase challenged with a lipophilic radical inducer and substrate. Acknowledgments

The authors acknowledge the help of Drs. J. Cazaux and D. Mancier (Expansia, Aramon, France) in synthesis and analytical procedures, and Dr J.A. Hoerter (INSERM, Chaˆtenay-Malabry, France) and Dr. J.-P. Finet (CNRS-UMR 6517 Marseille) for helpful discussions. References Alloatti G, Montrucchio G, Emanuelli G, Camussi G. 1992. Plateletactivating factor induced platelet/neutrophil co-operation during myocardial reperfusion. J Mol Cell Cardiol 24:163–171. Alloatti G, Montrucchio G, Camussi G. 1994. Role of plateletactivating factor (PAF) in oxygen-radical-induced cardiac dysfunction. J Pharmacol Exp Ther 269:766–771.

170

BILLOTTET ET AL.

Ambrosio G, Flaherty JT, Duilio C, Tritto I, Santoro G, Elia PP, Condorelli M, Chiariello M. 1991. Oxygen radicals generated at reflow induce peroxidation of membrane lipids in reperfused hearts. J Clin Invest 87:2056–2066. Amri H, Ogwuegbu SO, Boujrad N, Drieu K, Papadopoulos V. 1996. In vivo regulation of the peripheral-type benzodiazepine receptor and glucocorticoid synthesis by Ginkgo biloba extract (EGb 761) and isolated ginkgolides. Endocrinology 137:5707–5718. Amri H, Drieu K, Papadopoulos V. 2003. Transcriptional suppression of the adrenal cortical peripheral-type benzodiazepine receptor gene and inhibition of steroid synthesis by ginkgolide B. Biochem Pharmacol 65:717–729. Berti F, Magni F, Rossoni G, Puglisi L, Beretta A, Berti M, Tondo C, Villa L. 1989. The ginkgolide BN 52021 prevents myocardial contracture and arrhythmia in ischemic rabbit heart. In: Braquet P, editor. GinkgolidesFchemistry, biology, pharmacology and clinical perspectives. Vol 2. JR Prous Science Publisher, S.A. p 369–380. Berti F, Magni F, Rossoni G, De Angelis L, Galli G. 1990. Production and biologic interactions of prostacyclin and platelet activating factor in acute myocardial ischemia in the perfused rabbit heart. J Cardiovasc Pharmacol 16:727–735. Biber A, Koch E. 1999. Bioavailability of ginkgolides and bilobalide from extracts of Ginkgo biloba using GC/MS. Planta Med 65: 192–193. Bouaziz N, Redon M, Que´re´ L, Remacle J, Michiels C. 2002. Mitochondrial respiratory chain as a new target for anti-ischemic molecules. Eur J Pharmacol 441:35–45. Braquet P. 1987. The ginkgolides: potent platelet-activating factor antagonists isolated from Ginkgo biloba L.: chemistry, pharmacology and clinical applications. Drugs Future 12:643–699. Braquet P, Esanu A, Buisine E, Hosford D, Broquet C, Koltai M. 1991. Recent progress in ginkgolide research. Med Res Rev 11:295–355. Braquet PG, Spinnewyn B, Braquet M, Bourgain RH, Taylor JE, Etienne A, Drieu K. 1985. BN 52021 and related compounds: a new series of highly specific PAF-acether antagonists isolated from Ginkgo biloba L. Blood Vessels 16:558–572. Cazaux J-B, Dafniet M, Rebollo J, Teng BP. 1995. Two-step process for the preparation of ginkgolide B from ginkgolide C by intermediate sulfonate formation followed by borohydride reduction. Eur Pat Appl EP 680968 (Chem Abstr 1995. 124:176580). Chen J, Hu L, Jiang H, Gu J, Zhu W, Chen Z, Chen K, Ji R. 1998. A 3D-QSAR study on ginkgolides and their analogues with comparative molecular field analysis. Bioorg Med Chem 8:1291–1296. Cheung F, Siow YL, O K. 2001. Inhibition by ginkgolides and bilobalide of the production of nitric oxide in macrophages (THP1) but not in endothelial cells (HUVEC). Biochem Pharmacol 61:503–510. Corey EJ, Gavai AV. 1989. Simple analogs of ginkgolide B which are highly active antagonists of platelet activating factor. Tetrahedron Lett 30:6959–6962. Corey EJ, Srinivas Rao K. 1991. Enantioselective total synthesis of ginkgolide derivatives lacking the tert-butyl group, an essential structural subunit for antagonism of platelet activating factor. Tetrahedron Lett 32:4623–4626. Corey EJ, Rao KS, Ghosh AK. 1992. Intramolecular and intermolecular hydroxyl reactivity differences in ginkgolides A, B and C and their chemical applications. Tetrahedron Lett 33:6955–6958.

DeFeudis FV. 1998. Ginkgo biloba extract (EGb 761). From chemistry to the clinic. Wiesbaden: Ullstein Medical. Fourtillan JB, Brisson AM, Girault J, Ingrand I, Decourt J-P, Drieu K, Jouenne P, Biber A. 1995. Proprie´te´s pharmacocine´tiques du bilobalide et des ginkgolides A et B chez le sujet sain apre`s administrations intraveineuses et orales d’extrait de Ginkgo biloba (EGb 761). The´rapie 50:137–144. Furukawa S. 1932. Untersuchung u¨ber die bestandteile von ‘‘Ginkgo biloba L’’. Chem Zentralbl II:3901–3902. Harper IS, Merwe EVD, Owen P, Opie LH. 1990. Ultrastructural correlates of ischaemic contracture during global subtotal ischaemia in the rat heart. J Exp Path 71:257–268. Hearse DJ, Garlick PB, Humphrey SM. 1977. Ischemic contracture of the myocardium: mechanisms and prevention. Am J Cardiol 39:986–993. Hu L, Chen Z, Cheng X, Xie Y. 1999. Chemistry of ginkgolides: structure-activity relationship as PAF antagonists. Pure Applied Chem 71:1153–1156. Hu L, Chen Z, Xie Y, Jiang Y, Zhen H. 2000. Alkyl and alkoxycarbonyl derivatives of ginkgolide B: synthesis and biological evaluation of PAF inhibitory activity. Bioorg Med Chem 8:1515–1521. Janssens D, Michiels C, Delaive E, Eliaers F, Drieu K, Remacle J. 1995. Protection of hypoxia-induced ATP decrease in endothelial cells by Ginkgo biloba extract and bilobalide. Biochem Pharmacol 50:991–999. Janssens D, Remacle J, Drieu K, Michiels C. 1999. Protection of mitochondrial respiration activity by bilobalide. Biochem Pharmacol 58:109–119. Jaracz S, Stromgaard K, Nakanishi K. 2002. Ginkgolides: selective acetylations, translactonization and biological evaluation. J Org Chem 67:4623–4626. Katoh S, Toyama J, Kodama I, Koike A, Abe T. 1993. Role of platelet-activating factor in ischemia-reperfusion injury of isolated rabbit hearts: protective effect of a specific platelet activating factor antagonist, TCV-309. Cardiovasc Res 27:1430–1434. Koltai M, Tosaki A, Hosford D, Braquet P. 1989. Ginkgolide B protects isolated hearts against arrhythmias induced by ischemia but not reperfusion. Eur J Pharmacol 164:293–302. Korestsune Y, Marban E. 1990. Mechanism of ischemic contracture in ferret hearts: relative roles of Cai elevation and ATP depletion. Am J Physiol 258:H9–H16. Krieglstein J, Ausmeier F, El-Abhar H, Lippert K, Welsch M, Rupalla K, Henrich-Noack P. 1995. Neuroprotective effects of Ginkgo biloba constituents. Eur J Pharm Sci 3:39–48. Lagente V, Simonetti MPB, Fortes ZB, Braquet P. 1989. Protective effect of a specific platelet-activating factor antagonist, BN 52021, on bupivacaine-induced cardiovascular impairment in rats. Pharmacol Res 21:577–585. Lang Q, Wai CM. 1999. An extraction method for determination of ginkgolides and bilobalide in Ginkgo leaf extracts. Anal Chem 71:2929–2933. Lenoir M, Pedruzzi E, Rais S, Drieu K, Perianin A. 2002. Sensitization of human neutrophil defense activities through activation of platelet-activating factor receptors by ginkgolide B, a bioactive component of the Ginkgo biloba extract EGb 761. Biochem Pharmacol 63:1241–1249. Liebgott T, Miollan M, Berchadsky Y, Drieu K, Culcasi M, Pietri S. 2000. Complementary cardioprotective effects of flavonoid

CARDIOPROTECTION BY EPIMER GINKGOLIDES metabolites and terpenoid constituents of Ginkgo biloba extract (EGb 761) during ischemia and reperfusion. Basic Res Cardiol 95:368–377. Liu X. 2003. New approaches to measure lipophilicity: applications and interpretation. PhD Thesis, University of Lausanne, Switzerland. Lombardo F, Shalaeva MY, Tupper KA, Gao F, Abraham MH. 2000. ElogPoct: a tool for lipophilicity determination in drug discovery. J Med Chem 43:2922–2928. Lowry OH, Rosebrough NJ, Farr A, Randall RJ.1951. Protein measurement with the folinphenol reagent. J Biol Chem 193: 265–275. Maruyama M, Terahara A, Itagaki Y, Nakanishi K. 1967. The ginkgolides. I. Isolation and characterization of the various groups. Tetrahedron Lett 4:299–302. Montrucchio G, Alloatti G, Tetta C, De Luca R, Saunders RN, Emanueli G, Camussi G. 1989. Release of platelet-activating factor from ischemic-reperfused rabbit heart. Am J Physiol 256:H1236–H1242. Montrucchio G, Alloatti G, Camussi G. 2000. Role of plateletactivating factor in cardiovascular pathophysiology. Physiol Rev 80:1669–1699. Neely JR, Grotyohann LW. 1984. Role of glycolytic products in damage to ischemic myocardium. Dissociation of adenosine triphosphate levels and recovery of function of reperfused ischemic hearts. Circ Res 55:816–824. Okabe K, Yamada K, Yamamura S, Takada S. 1967. Ginkgolides. J Chem Soc C 2001–2006. Okhawa H, Ohishi N, Yagi K. 1979. Assay for lipid peroxides in animal tissues by thiobarbituric acid reaction. Anal Biochem 95:351–358. Pietri S, Maurelli E, Drieu K, Culcasi M. 1997. Cardioprotective and anti-oxidant effects of the terpenoid constituents of Ginkgo biloba extract (EGb 761). J Mol Cell Cardiol 29:733–742. Pietri S, Liebgott T, Finet JP, Culcasi M, Billottet L, BernardHenriet C. 2001. Synthesis and biological studies of a new ginkgolide C derivative: evidence that the cardioprotective effect of ginkgolides is unrelated to PAF inhibition. Drug Dev Res 54:191–201. Pietri S, Mercier A, Mathieu C, Caffaratti S, Culcasi M. 2003. Hemodynamic and metabolic effects of the ?-phosphorylated nitroxide 2-diethoxyphosphoryl-2,5,5-trimethylpyrrolidinoxyl during myocardial ischemia and reperfusion. Free Rad Biol Med 34:1167–1177.

171

Stromgaard K, Saito DR, Shindou H, Ishii S, Shimizu T, Nakanishi K. 2002. Ginkgolide derivatives for photolabeling studies: preparation and pharmacological evaluation. J Med Chem 45:4038–4046. Suleiman MS, Halestrap AP, Griffiths EJ. 2001. Mitochondria: a target for myocardial protection. Pharmacol Ther 89:29–46. Takeo S, Kajiwara H, Tanonaka K, Nasa Y, Nakajima Y, Ohya Y, Kaizuka Y. 2000. Cardioprotective effect of a novel cyclohexane dicarboximide derivative, ST-6, against ischemia/reperfusion injury. Drug Dev Res 51:20–28. Tsai R-S, Carrupt P-A, Testa B, El Tayar N, Grunewald GL, Casy AF. 1993. Influence of stereochemical factors on the partition coefficient of diastereomers in a biphasic octan-1-ol/water system. J Chem Res Synopses 8:298–299. van de Waterbeemd H, Kansy M, Wagner B, Fischer H. 1996. Lipophilicity measurement by reversed-phase high performance liquid chromatography (RP-HPLC). In: Pliska V, Testa B, van de Waterbeemd H, editors. Lipophilicity in drug action and toxicology. Weinheim: VCH Publishers. p 73–87. Vanoverschelde J-LJ, Janier MF, Bergman SR. 1994. The relative importance of myocardial energy metabolism compared with ischemic contracture in the determination of ischemic injury in perfused rabbit hearts. Circ Res 74:817–828. Ventura-Clapier R, Vassort G. 1981. Rigor tension during metabolic and ionic rises in resting tension in rat heart. J Mol Cell Cardiol 13:551–561. Ventura-Clapier R, Veksler V. 1994. Myocardial ischemic contracture. Metabolites affect rigor tension development and stiffness. Circ Res 74:920–929. Ventura-Clapier R, Veksler V, Hoerter JA. 1994. Myofibrillar creatine kinase and cardiac contraction. Mol Cell Biochem 133/ 134:125–144. Vogensen SB, Stromgaard K, Shindou H, Jaracz S, Suehiro M, Ishii S, Shimizu T, Nakanishi K. 2003. Preparation of 7-substituted ginkgolide derivatives: potent platelet activating factor (PAF) receptor antagonists. J Med Chem 46:601–608. Wainwright CL, Parratt JR. 1989. Effect of BN 52021 on platelet behavior and arrhythmias during myocardial ischemia and reperfusion. In: Braquet P, editor. GinkgolidesFchemistry, biology pharmacology and clinical perspectives. Vol 2. JR Prous Science Publisher, S.A. p 369–380. Weinges K, Schick H. 1991. Chemie der ginkgolide, IV. Herstellung von ginkgolid B aus ginkgolid C. Liebigs Ann Chem 81–83.

Rapin JR, Zaibi M, Drieu K. 1998. In vitro and in vivo effects of an extract of Ginkgo biloba (EGb 761), ginkgolide B, and bilobalide on apoptosis in primary cultures of rat hippocampal neurons. Drug Dev Res 45:23–29.

Weinges K, Ru¨mmler M, Schick H. 1993. Chemie der ginkgolide, VI. Herstellung von 1,10-dihydroxy und 1,7,10-trihydroxyginkgolid aus 1,3,7,10-tetrahydroxyginkgolid. Liebigs Ann Chem 1023– 1027.

Rioufol G, Pietri S, Culcasi M, Loufoua J, Staat P, Pop C, Drieu K, Ovize M. 2003. Ginkgo biloba extract EGb 761 attenuates myocardial stunning in the pig heart. Basic Res Cardiol 98:59–68.

Yuan G, Kaneko M, Masuda H, Hon RB, Kobayashi A, Yamazaki N. 1992. Decrease in heart mitochondrial creatine kinase activity due to oxygen free radicals. Biochim Biophys Acta 1140:78–84.

Romaschin AD, Wilson GJ, Thomas U, Feitler DA, Tumiati L, Mickle DAG. 1990. Subcellular distribution of peroxidized lipids in myocardial reperfusion injury. Am J Physiol 259:H116–H123.

Zekri O, Boudeville P, Genay P, Perly B, Braquet P, Jouenne P, Burgot JL. 1996. Ionization constants of ginkgolide B in aqueous solution. Anal Chem 68:2598–2604.

Steenbergen C, Murphy E, Watts JA, London RE. 1990. Correlation between cytosolic free calcium, contracture, ATP, and irreversible ischemic injury in perfused rat heart. Circ Res 66:135–146.

Zhao HW, Li X-Y. 2002. Ginkgolide A, B, and huperzine A inhibit nitric oxide-induced neurotoxicity. Int Immunopharmacol 2: 1551–1556.