In recent years, the prevalence of diabetes has increased alarmingly worldwide, giving it the dimension of an epidemic. It is estimated that 140 million people worldwide have diabetes, and this figure will increase to 300 million by the year 2025 (King et al., 1998). Type 2 diabetes accounts for ∼90% of all cases. Current therapeutic strategies for type 2 diabetes include five classes of oral agents—sulfonylureas, biguanides, meglitinides, thiazolidinediones, and α-glucosidase inhibitors (Loh and Leow, 2002), all of which have their limitations in terms of efficacy and side effects (Riddle, 2003; Tadayyon and Smith, 2003). Over the past few years, multiple approaches have been adopted to develop new therapies for the treatment of type 2 diabetes. Recent animal and human studies have demonstrated the promise of dipeptidyl peptidase-IV (DPP-IV) inhibitors for improving glucose tolerance (Hoffmann et al., 2001; Ahren et al., 2002; Hinke et al., 2002; Reimer et al., 2002; Sudre et al., 2002). DPP-IV is a membrane-bound and circulating serine protease that can rapidly cleave and inactivate two incretins, namely glucagon-like peptide-1 and gastric inhibitory peptide/glucose-dependent insulinotropic polypeptide (Mentlein, 1999). These gastrointestinal peptides could potentiate glucose-induced insulin secretion by sensitizing pancreatic β-cells to stimulation by glucose. Inactivation of these peptides can be prevented by DPP-IV inhibitors, leading to potentiation of the biological activity of the incretins (Drucker, 2003).

(S)-3-Fluoro-pyrrolidine derivatives are known to be potent and selective inhibitors of DPP-IV (Augustyns et al., 1997). Optimization of derivatives resulting from a series of fluorinated pyrrolidines with 4-substituted cyclohexylglycine subunits resulted in the selection of (1S)-1-(trans-4-{[(4-trifluoromethoxyphenyl)sulfonyl]amino}cyclohexyl)-2-[(3S)-3-fluoropyrrolidin-1-yl]-2-oxoethanaminium chloride (MRL-A) and (1S)-1-(trans-4-{[(2,4-difluorophenyl)sulfonyl]amino}cyclohexyl)-2-[(3S)-3-fluoropyrrolidin-1-yl]-2-oxoethanaminium chloride (MRL-B) (Fig. 1) as two potent and selective DPP-IV inhibitors with good pharmacokinetic properties and significant activity in an oral glucose tolerance test in lean mice (Caldwell et al., 2004)—favorable characteristics for development consideration at Merck Research Laboratories (MRL). Based on the strategy established at MRL to minimize reactive metabolite formation by structural modification (Evans et al., 2004), evaluation of the metabolic activation potential of these two compounds became relevant to determine whether further lead optimization was needed. Our studies herein described, designed to evaluate the potential for these two compounds to undergo metabolic activation and to understand the mechanism of metabolic activation, indicated that the fluoropyrrolidine ring is the labile site for metabolic activation.

Materials and Methods

Chemicals and Reagents. MRL-A, MRL-B, and a closely related structural analog used as an internal standard were obtained from the Medicinal Chemistry Department, Merck Research Laboratories, Rahway, NJ. [Benzenesulfonamide-2-³H]MRL-A ([³H]MRL-A) and [2,4-difluorophenyl-5-³H]MRL-B ([³H]MRL-B) were synthesized by the Labeled Compound Synthesis Group, Merck Research Laboratory, Rahway, NJ. The radioactive purity (99.4%) was determined by high-performance liquid chromatography (HPLC) with a radioactivity detector. N-acetylcysteine (NAC), reduced glutathione (GSH), semicarbazide, and NADPH were purchased from Sigma-Aldrich (St. Louis, MO). A bicinconinic acid protein assay kit was purchased from Pierce Chemical Co. (Rockford, IL). ScintiSafe Gel and all solvents (HPLC grade) were purchased from Fisher Scientific Co. (Pittsburgh, PA). Liver microsomes prepared from male Sprague-Dawley rats (∼250-350 g, ∼8 weeks old, n ≥ 10 per group) treated with saline, corn oil, phenobarbital (i.p. injections of 80 mg/kg in saline for 4 days), clofibric acid (i.p. injections of 200 mg/kg in corn oil for 4 days), isoniazid (i.p. injections of 200 mg/kg in saline for 4 days), pregnenolone-16α-carbonitrile (PCN) (i.p. injections of 50 mg/kg in corn oil for 4 days), dexamethasone (i.p. injections of 50 mg/kg in corn oil for 4 days), or 3-methylcholanthrene (3-MC) (i.p. injections of 27 mg/kg in corn oil for 4 days) were purchased from XenoTech LLC (Lenexa, KS).

Irreversible Binding to Rat Liver Microsomal Protein. A pool of 50 male Sprague-Dawley rat (∼225-250g, ∼8 weeks old) liver microsomes were prepared following procedures described in the literature (Raucy and Lasker, 1991). [³H]MRL-A (10 μM, 0.14 μCi/nmol) or [³H]MRL-B (10 μM, 0.14 μCi/nmol) were incubated with rat liver microsomes at 37°C for 30 and 60 min. The microsomal incubation mixture, in a final volume of 0.2 ml, contained 0.1 M phosphate buffer (pH 7.4), 2 mM MgCl₂, 1 mM NADPH, and 1 mg/ml rat liver microsomal protein in the presence and absence of 5 mM GSH or NAC. The reaction was initiated by adding NADPH to the mixture. Control incubations without NADPH were conducted under the same conditions. Determinations of irreversible binding levels were conducted following procedures previously described (Evans et al., 2004). Briefly, incubations were terminated with 800 μl of acetone. The precipitated protein was transferred onto a Whatman GF/B filter mat and washed with 80% methanol (1 liter per 48 wells; 20 ml per filter) to remove any radioactive material not covalently bound by using a Brandel cell/membrane harvester (Brandel Inc., Gaithersburg, MA). Protein on the filter paper was then dissolved in 10% sodium dodecyl sulfate overnight at 55°C. An aliquot was analyzed for protein concentration with the bicinconinic acid protein assay kit; another aliquot was mixed with ScintiSafe Gel and counted for radioactivity in a Beckman liquid scintillation counter (model LS6500; Beckman Coulter, Fullerton, CA). The irreversible binding to proteins was expressed as picomole equivalents per milligram of protein from duplicate or triplicate incubations using the specific activity of the substrate in the incubation mixture and the amount of protein determined for each sample.

In Vitro Incubation Conditions. Rat recombinants CYP3A1, 3A2, 2B1, 2D1, 1A1, 1A2, and 2A2 were purchased from BD Gentest (Woburn, MA). The incubation mixture contained MRL-A or MRL-B (2, 5, 10, 20, 50, and 100 μM), 0.1 M phosphate buffer (pH 7.4), 2 mM MgCl₂, 5 mM GSH or NAC, 1 mg/ml NADPH, and 1 mg/ml microsomal protein or 200 pmol/ml P450 isoform in a total volume of 0.5 ml. Microsomes containing an empty cDNA expression vector were used as control in studies with recombinant P450 isozymes. In the trapping experiments with semicarbazide, semicarbazide (5 mM) was added to the incubations. In the experiments with epoxide hydrolase, microsomal epoxide hydrolase (1 mg of protein/ml) was added to the incubations of MRL-A (50 μM) with rat liver microsomes (1 mg/ml) in the presence of 1 mg/ml NADPH and 5 mM NAC. The reaction was initiated by adding NADPH to the mixture and stopped by adding an equal volume of ice-cold acetonitrile. Samples were centrifuged at 1850g for 15 min and supernatants were dried under N₂. The residue was reconstituted with mobile phase and analyzed by LC-MS.

Mass Spectrometric Analysis. LC-MS/MS was carried out on either a PerkinElmer Sciex API-3000 (PerkinElmerSciex Instruments, Boston, MA) triple quadrupole mass spectrometer or a Finnigan LCQ Deca XP mass spectrometer (Thermo Finnigan, San Jose, CA) interfaced to a PerkinElmer HPLC system (PerkinElmer Life and Analytical Sciences, Boston, MA) equipped with two Series 200 pumps and a PerkinElmer Series 200 autosampler. For metabolite identification, a YMC ODS-AQ column (4.6 × 250 mm, 4.0 μm; Waters, Milford, MA) was used. Mobile phases were A, 5 mM ammonium acetate in water with 0.1% formic acid and B, 70% acetonitrile/30% methanol (v/v) with 0.1% formic acid. The flow rate was 1 ml/min. A linear gradient from 10% B to 70% B in 40 min was applied. The column was washed with 95% B for 5 min before equilibrating for 8 min with starting conditions. The LC effluent was split with 1:25 directed into the mass spectrometer. LC-MS/MS experiments were conducted using a Turbo-ionspray interface (API-3000) or an electrospray ionization probe (Deca XP) in the positive ion mode. For determining the relative amount of NAC adduct of MRL-A from different incubations, a DASH 8 column (2.1 × 50 mm, 5 μm; Thermo Hypersil, Keystone Scientific Operations, Bellefonte, PA) was used. Mobile phases were A, 5% acetonitrile/95% water (v/v) in 0.1% acetic acid and B, 5% water/95% acetonitrile (v/v) in 0.1% acetic acid. The flow rate was 1.5 ml/min. A steep linear gradient from 5% B to 90% B in 1.9 min was applied. The multiple reaction monitoring (MRM) transitions monitored were m/z 627 → m/z 164 for the NAC adduct and m/z 468 → m/z 351 for the parent compound MRL-A. The retention time for the NAC adduct was 1.1 min, and the retention time for the parent compound was 1.2 min. The peak area ratio of the NAC adduct over internal standard was calculated to indicate the extent of NAC adduct formation.

Accurate mass measurements were performed on a Micromass tandem quadrupole time-of-flight mass spectrometer (Q-TOF; Waters). The instrument was operated in the positive ion mode. Mass measurements were based on a lock mass of m/z 609.2812 (the MH⁺ ion of reserpine). Samples were introduced by a Hewlett Packard Series 1100 HPLC (Hewlett Packard, Palo Alto, CA) via a Zorbax SB-phenyl column (4.6 × 150 mm, 5 μm; Agilent Technologies, Wilmington, DE). The mobile phase consisted of A, 5 mM ammonium acetate in water and B, 70% acetonitrile/30% methanol (v/v). The flow rate was 1 ml/min. A linear gradient from 10% B to 95% B in 40 min was applied. The column was washed with 95% B for 5 min before equilibrating for 8 min with initial conditions.

Results

Irreversible Binding of [³H]MRL-A- and [³H]MRL-B-Related Radioactivity to Rat Liver Microsomes. Incubation of [³H]MRL-A or [³H]MRL-B with rat liver microsomes in the presence of NADPH resulted in irreversible binding of radioactivity to microsomal protein. Levels of irreversible binding were 137 and 215 pmol Eq/mg of protein for [³H]MRL-A and 71 and 135 pmol Eq/mg of protein for [³H]MRL-B after 30 and 60 min incubations, respectively (Fig. 2). These levels were higher than what is considered an acceptable threshold at MRL (Evans et al., 2004). This time- and NADPH-dependent binding of radioactivity to microsomal protein indicated that both compounds underwent metabolic activation to reactive intermediates that bind to protein irreversibly. Addition of GSH (5 mM) to the incubation did not affect the turnover of MRL-A and MRL-B by rat liver microsomes (data not shown) but significantly lowered the in vitro irreversible binding of MRL-A and MRL-B to 62 and 27 pmol Eq/mg of protein, respectively, after 60-min incubations (Fig. 2), suggesting that the reactive intermediate(s) formed in the presence of NADPH had been trapped by GSH. NAC, a thiol-based nucleophile that can scavenge electrophilic intermediates, was used in our later studies to trap the reactive intermediate.

Characterization of GSH and NAC Adducts Formed in Rat Liver Microsomes by LC-MS/MS. The GSH and NAC adducts detected in rat liver microsomal incubations were characterized by comparing their product ion spectra to that of the corresponding parent. The product ion spectra of the parent MRL-A and MRL-B are shown in Fig. 1, where fragments at m/z 351 and 242 from MRL-A are associated with the trifluoromethoxyphenyl sulfonamide moiety, and fragments at m/z 303 and 194 from MRL-B are associated with the difluorophenyl sulfonamide moiety; the fragment at m/z 90 is assigned as the protonated fluoropyrrolidine.

The presence of a MH⁺ ion at m/z 771 (addition of 303 Da to parent) was observed following full-scan analysis of rat liver microsomal incubations of MRL-A in the presence of GSH. Subsequent collision-induced dissociation of m/z 771 (Fig. 3A) produced product ions at m/z 696 and 642, resulting from typical neutral losses of glycine (75 Da) and pyroglutamate (129 Da), respectively (Baillie and Davis, 1993). The fragment ion at m/z 753, derived from loss of water, underwent further losses of glycine and pyroglutamate to give rise to the ions at m/z 678 and 624, respectively. Fragment ions at m/z 308 and 179, corresponding to protonated GSH and subsequent neutral loss of pyroglutamate, respectively, also were observed. The fragment ion at m/z 351 in the spectrum indicated that the trifluoromethoxyphenyl sulfonamide cyclohexyl moiety remained intact, suggesting that GSH conjugation had occurred at the fluoropyrrolidine ring. The fragment at m/z 446 originated from combined losses of GSH and water.

The corresponding NAC adduct of MRL-A exhibited a protonated molecular ion at m/z 627 (addition of 159 Da to parent). The product ion spectrum of the MH⁺ at m/z 627 (Fig. 3B) showed fragment ions at m/z 164 and 122, corresponding to protonated NAC and subsequent neutral loss of ketene (42 Da), respectively. The presence of the fragment at m/z 351 further established that NAC was added to the fluoropyrrolidine ring. The fragment ion at m/z 231 was assigned as the product arising from cleavage of the amide bond from the parent moiety, resulting in a protonated NAC conjugate of the fluoropyrrolidine moiety followed by loss of water. The signal at m/z 68 could be derived from the ion at m/z 231 with a cleavage at the thioether bond (Fig. 3B).

The GSH adduct of MRL-B had a MH⁺ ion at m/z 723 (addition of 303 Da to parent). The product ion spectrum of the GSH adduct (Fig. 4A) gave rise to characteristic neutral losses of glycine and pyroglutamate, in addition to loss of water. The fragment ion at m/z 303 was consistent with an intact difluorophenyl sulfonamide cyclohexyl moiety in the conjugate, whereas the signal at m/z 398 was assigned as loss of GSH coupled with loss of water. The MH⁺ ion of the NAC adduct of MRL-B was at m/z 579 (addition of 159 Da to parent). Its product ion spectrum included signals at m/z 164, 122, 303, 398, 68, and 231 (Fig. 4B). As discussed above, product ions at m/z 303, 68, and 231 indicated that the site of conjugation was the fluoropyrrolidine ring.

Accurate Mass Determination. Based on the masses of the GSH and NAC adducts, GSH or NAC had added to the fluoropyrrolidine ring of MRL-A or MRL-B with a loss of 2 Da. This was rationalized to occur via two routes: route 1 was postulated to involve dehydrogenation of the fluoropyrrolidine ring (fluorine still present), whereas route 2 was considered to involve defluorination with addition of a hydroxyl group. To distinguish between these two routes, the accurate mass of the MRL-A-NAC adduct was measured to confirm its elemental composition. The MH⁺ ion was observed at m/z 627.1746, which was in good agreement with the calculated value of m/z 627.1770 when assuming fluorine had been eliminated (-3.8 ppm error). Conversely, dehydrogenation would have resulted in a product of m/z 627.1570, leading to a larger error (28 ppm).

Trapping Experiments with Semicarbazide. Following incubations of MRL-A with rat liver microsomes in the presence of NADPH, an oxidative metabolite (M1) was detected. The product ion scan of m/z 484 (addition of 16 Da to parent) gave rise to a fragment ion at m/z 466 derived from loss of water and a fragment ion at m/z 351 consistent with an intact trifluoromethoxyphenyl sulfonamide cyclohexyl moiety in the metabolite (Fig. 5A), collectively suggesting that hydroxylation occurred on the fluoropyrrolidine moiety. Addition of semicarbazide (5 mM), a trapping agent for aldehydes, to the incubation resulted in the formation of a semicarbazide adduct (MH⁺ at m/z 541) with significant reduction in the formation of M1 (Fig. 6), suggesting that M1 was converted to the semicarbazide adduct. This adduct was not detected in the incubations in the absence of semicarbazide. The MS/MS spectrum of the MH⁺ ion at m/z 541 consisted of fragment ions at m/z 521, 466, and 446 derived from neutral losses of 20 Da (-HF) and 75 Da (-semicarbazide) and combined losses (-HF-semicarbazide), respectively, and the fragment ion at m/z 351 (Fig. 5B). It is very likely that the product (M1) was hydroxylated at C-5 or C-2, which would allow it to be in equilibrium with a ring-opened aldehyde that can be trapped by semicarbazide (Scheme 1A).

Incubations of MRL-A with rat liver microsomes (in the presence of NADPH) enriched with both NAC and semicarbazide generated a semicarbazide adduct of NAC conjugate (MH⁺ at m/z 684). Collision-induced dissociation of the MH⁺ at m/z 684 resulted in fragment ions at m/z 231, 446, and 609 (Fig. 5C), suggesting the presence of an aldehyde group trapped by semicarbazide in the NAC conjugate.

Identification of P450 Enzymes Involved in the Formation of the NAC Adduct of MRL-A. The extent of formation of the NAC adduct of MRL-A was semiquantified as peak area ratio of the NAC adduct of MRL-A over internal standard. No attempts were made in the present studies to quantify accurately the NAC adduct formed, since the turnover of substrate to the detectable product was low. When MRL-A was incubated with rat recombinants CYP1A1, 1A2, 2B1, 2D1, 2A2, 3A1, and 3A2 in the presence of NADPH and NAC, the NAC adduct was formed most abundantly from incubations that contained CYP3A1 or CYP3A2 (Table 1), indicating that rat CYP3A1 and CYP3A2 are the major isoforms catalyzing the metabolic activation of MRL-A. The extent of formation of the NAC adduct by CYP3A1 was dependent on substrate concentration investigated (2-100 μM) with a maximal rate being observed at 50 μM (data not shown).

Irreversible Binding of [³H]MRL-A-Related Radioactivity to Liver Microsomes from Rats Pretreated with Different P450 Inducers. The irreversible binding of [³H]MRL-A-related radioactivity to rat liver microsomes in the presence of NADPH was increased about 5- or 9-fold when the liver microsomes used had been prepared from rats treated with PCN or dexamethasone, respectively (Fig. 7). A 3-fold increase in the irreversible binding was observed in liver microsomes prepared from rats pretreated with phenobarbital (Fig. 7). Pretreatment of rats with 3-MC or clofibric acid had little or no effect on the level of the in vitro irreversible binding determined for MRL-A (Fig. 7), whereas treatment of rats with isoniazid seemed to decrease these levels (Fig. 7). The irreversible binding to the above rat liver microsomes was significantly reduced by the presence of NAC (5 mM) in the incubation mixture (Fig. 7).

Formation of the NAC Adduct of MRL-A in Rat Liver Microsomes from Rats Pretreated with Different Inducers. The NAC adduct of MRL-A was formed most abundantly in liver microsomes from PCN- or dexamethasone-treated rats, followed by liver microsomes from phenobarbital-treated rats (Table 2). The yield of the NAC adduct was the lowest in liver microsomes from rats treated with isoniazid, where the lowest in vitro irreversible binding (50 pmol/mg of protein) was also observed (Table 2). There was a good correlation between the formation of an aldehyde group trapped by semicarbazide as a semicarbazide adduct and the formation of NAC conjugate of MRL-A in the liver microsomes prepared from rats treated with different inducers (Fig. 8).

Discussion

The current studies evaluated the metabolic activation potential of two compounds from a series of fluorinated pyrrolidines with 4-substituted cyclohexylglycine subunits, MRL-A and MRL-B, which had been optimized for DPP-IV inhibition and selectivity with good pharmacokinetic properties in rats and significant efficacy in an oral glucose tolerance test in lean mice. Our studies demonstrated that incubations of MRL-A and MRL-B with rat liver microsomes resulted in irreversible binding levels in excess of the 50 pmol Eq/mg of protein typically considered acceptable at MRL (Evans et al., 2004). In addition, substrate turnover in rat liver microsomes was low, suggesting that most of the turnover of the compound was being metabolically activated.

The similar fragmentation patterns observed from the GSH/NAC adducts of MRL-A and MRL-B (Figs. 3 and 4) suggested the same bioactivation mechanism was involved with the fluoropyrrolidine ring as the labile site. It has been shown that compounds containing a pyrrolidine ring can be bioactivated via formation of an iminium ion intermediate, which, in most cases, can be trapped by cyanide (Streeper et al., 1997). In our studies, cyanide had little or no effect on the irreversible binding (data not shown), suggesting that this metabolic activation pathway may not play an important role in the irreversible binding to protein.

The accurate mass determined for the NAC adduct of MRL-A indicated that defluorination and oxidation events were involved in the metabolic activation of the fluoropyrrolidine ring. Furthermore, the presence of an aldehyde group in the NAC conjugate and the positive correlation observed between the formation of an aldehyde group (trapped by semicarbazide) and the formation of NAC conjugate (Fig. 8) indicated that aldehyde formation played a role in the bioactivation. One of the proposed mechanisms involved hydroxylation at C-5 leading to a ring-opened β-fluoroaldehyde, which in turn may undergo elimination of HF to form an α, β-unsaturated aldehyde as the electrophilic intermediate. Nucleophiles such as GSH or NAC would then attack the intermediate via a Michael addition mechanism to form the GSH or NAC conjugate, in which the aldehyde group could be trapped by semicarbazide. The aldehyde group may also cyclize with the amide nitrogen to form the ring-closed hydroxylated derivative in a reversible manner (Scheme 1A).

An alternative mechanism for the bioactivation of the fluoropyrrolidine that was taken into consideration (Scheme 1B) involved initial hydroxylation β to the carbon (C-4) attached to fluorine, followed by elimination of the fluoride anion upon intramolecular attack by the β-hydroxyl group resulting in the formation of an epoxide. GSH or NAC then would attack the epoxide yielding the GSH or NAC adducts. However, when microsomal incubations enriched with NAC were conducted in the presence or absence of microsomal epoxide hydrolase (1 mg of protein/ml), no significant alteration in the formation of NAC adduct was observed with or without epoxide hydrolase (data not shown). Possible reasons for this observation could be that the epoxide pathway did not play a role in the metabolic activation process or that the epoxide intermediate proposed here was not a substrate of or not accessible to the epoxide hydrolase used in this study. However, the current data indicated the formation of a semicarbazide conjugate, which cannot be accommodated by the pathway illustrated in Scheme 1B, further supporting the conclusion that the bioactivation of MRL-A and MRL-B occurred via pathway A.

NMR analysis was attempted on the isolated NAC adduct of MRL-A. The ¹H NMR spectrum was, however, too complex to interpret because of the presence of potential multiple species arising from multiple stereoconfigurations existing as two rotamers around the pyrrolidine N-CO amide bond.

The formation of the reactive electrophilic intermediate trapped by NAC was catalyzed primarily by rat recombinant CYP3A1 and CYP3A2 (Table 1). Dexamethasone and PCN are potent inducers of rat CYP3A1 and CYP3A2 (Parkinson, 2001). Accordingly, the bioactivation potential of MRL-A in liver microsomes from rats pretreated with these inducers was increased significantly, as indicated by the high level of the in vitro irreversible binding and the high level of the NAC adduct detected (Table 2). The moderate increase in the irreversible binding of radioactivity to liver microsomes from rats treated with phenobarbital could be attributed to the moderate inductive effect of phenobarbital on CYP3A2 (Parkinson, 2001). CYP1A1 and CYP1A2 had little or no role in the metabolic activation of MRL-A (Table 1). Consistent with that, treatment of rats with 3-MC, a potent inducer of CYP1A1 and CYP1A2 (Parkinson, 2001), had little or no effect on the in vitro irreversible binding determined for [³H]MRL-A (Fig. 7). No increase in the in vitro irreversible binding was observed with liver microsomes from rats pretreated with clofibric acid (Fig. 7), an inducer of CYP4A (Parkinson, 2001). This lack of effect could be due to the fact that CYP4A enzymes are not responsible for the bioactivation or that there is competition between the bioactivation pathway and alternative metabolic pathways for MRL-A, as clofibric acid is also an inducer of other enzymes (Fournel et al., 1983). Treatment of rats with isoniazid decreased the in vitro irreversible binding to some extent (Fig. 7). It is possible that CYP2E1, which is induced by isoniazid (Parkinson, 2001), might increase alternative metabolic pathways, thus decreasing the substrate availability for bioactivation. A good correlation was observed between the formation of the NAC adduct of MRL-A and the in vitro irreversible binding of [³H]MRL-A in liver microsomes from rats pretreated with different inducers (Table 2) suggesting that, in the context of the present study, formation of reactive intermediates identified as GSH or NAC adducts could serve as an index for the potential of irreversible binding.

The chemically reactive intermediates studied herein are responsible for the irreversible binding to microsomal protein in vitro, implying that these fluoropyrrolidine derivatives might have the potential to covalently bind to proteins or other macromolecules in vivo. It has been shown that covalently modified protein may interrupt normal cellular functions or function as haptens causing idiosyncratic drug reactions (Uetrecht, 1999). Moreover, the reactive metabolites that can be scavenged by GSH may deplete cellular GSH resulting in alterations in the redox state of the cell and subsequently causing oxidative stress in cells with progression to cell injury (Nelson and Pearson, 1990). Given the potential liability of chemically reactive electrophilic metabolites, it is reasonable to attempt to minimize the potential of bioactivation through structural modification at the early stage of drug discovery (Baillie and Kassahun, 2001; Evans et al., 2004).