Abstract
Metabolic aromatization of xenobiotics is an unusual reaction with some documented examples. For instance, the oxidation of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine to the neurotoxic pyridinium ion metabolite 1-methyl-4-phenylpyridinium by monoamine oxidase (MAO) B in the brain has been of interest to a number of investigators. It has also been reported that although the aromatization of N-methyl-tetrahydroisoquinoline occurs with MAO B, the metabolism does not proceed for its isomer, N-methyl-tetrahydroquinoline, by the same enzyme. The aromatization of an N-alkyl-tetrahydroquinoline substructure was identified during in vitro metabolite profiling of compound A, which was designed as a potent renin inhibitor for the treatment of hypertension. The N-alkylquinolinium metabolite of compound A was identified by liquid chromatography-tandem mass spectrometry of human liver microsomal incubates and proton NMR of the isolated metabolite. Further in vitro metabolism studies with a commercially available chemical (compound B), containing the same substructure, also generated an N-alkylquinolinium metabolite. In vitro cytochrome P450 (P450) reaction phenotyping of compound A revealed that the metabolism was catalyzed exclusively by CYP3A4. Although compound B was a substrate for several P450 isoforms, its quinolinium metabolite was also generated predominantly by CYP3A4. Neither compound A nor compound B was a substrate of MAOs. The quinolinium metabolites were readily produced by horseradish peroxidase, suggesting that aromatization of the N-alkyltetrahydroquinoline could occur via a mechanism involving single electron transfer from nitrogen. Although dihydro intermediates from the tetrahydroquinoline substrates were not observed in the formation of quinolinium metabolites, cyanide trapping results indicated the occurrence of iminium intermediates.
Compound A (Scheme 1) was a lead compound, consisting of a novel nonpeptidic ketopiperazine-based scaffold, that was designed as a potent renin inhibitor for the treatment of hypertension (Holsworth et al., 2005, 2006). The renin-angiotensin system (RAS) plays a key role in blood pressure regulation. Pharmacological blockade of the RAS cascade is usually achieved with angiotensin-converting enzyme inhibitors or angiotensin receptor blockers. Renin inhibition has been sought for decades, as it would inhibit the initial and rate-limiting step of the RAS and hence might offer a therapeutic profile distinct from angiotensin-converting enzyme inhibitors and angiotensin receptor blockers. Recent progress shows new potential for the treatment of hypertension by renin inhibition (Fisher and Hollenberg, 2005; Wood et al., 2005).
During in vitro metabolite profile studies on compound A, in support of lead optimization, an unusual metabolite was observed. It appeared to be derived from aromatization of the N-alkyltetrahydroquinoline ring, yielding an N-alkylquinolinium ion as a major metabolite in human liver microsomes. Similar metabolic oxidation of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) to the neurotoxic pyridinium metabolite, 1-methyl-4-phenylpyridinium, by monoamine oxidase B (MAO B) in the brain (Scheme 2), as well as the aromatization of a variety of MPTP derivatives, have been described previously (Maret et al., 1990; Gerlach et al., 1991; Di Monte et al., 1996; Castagnoli et al., 2002). Recently, a study of electrochemistry-electrospray mass spectrometry on the electrochemical oxidation of N-alkyl-4-phenyl-1,2,3,6-tetrahydropyridinyl derivatives and the chemical fate of the resulting aminyl radical cations has presented a unique perspective in understanding the metabolic oxidation mechanism for the tetrahydropyridines (Jurva et al., 2005). Also illustrated in Scheme 2 is the formation of N-methyl-isoquinolinium from N-methyl-tetrahydroisoquinoline by MAO B (Booth et al., 1989; Naoi et al., 1989). Two more examples of the similar aromatization have been reported in this journal: one involving an N-alkyl-5-tetrazolyl-1,2,3,6-tetrahydropyridine substructure (Christensen et al., 1999) and another with an N-alkyl-4,5,6,7-tetrahydro-thieno[3,2-c]pyridine substructure (Dalvie and O'Connell, 2004). The substrates in these reports all have remarkably similar substructures in that the cyclic tertiary nitrogen is distal to a carbon-carbon double bond or an aromatic ring by a methylene unit. It is fascinating that when the nitrogen atom was “moved” from this type of distal position (as in N-methyl-tetrahydroisoquinoline; Scheme 2) to the direct attachment on an aromatic ring (as in N-methyl-tetrahydroquinoline; Scheme 2), the latter isomer (N-methyl-tetrahydroquinoline) was no longer a substrate of the same enzyme (MAO B) that metabolized the former compound (N-methyl-tetrahydroisoquinoline) (Booth et al., 1989). Shaffer et al. (2001) have reported the horseradish peroxidase (HRP)-mediated conversion of N-cyclopropyl-N-methylaniline to N-methyl-quinolinium ion (Scheme 2), via a proposed intermediate of 1-methyl-3,4-dihydroquinolinium (Shaffer et al., 2001). In contrast, cytochrome P450 (P450)-catalyzed oxidation of N-cyclopropyl-N-methylaniline did not generate the quinolinium metabolite; rather, the P450s yielded common phase I metabolites (e.g., demethyl, decyclopropyl, and p-hydroxy metabolites) in liver microsomes from phenobarbital-pretreated rats (Shaffer et al., 2002).
In addition to the quinolinium metabolite of compound A, we have also investigated whether a similar N-alkylquinolinium metabolite could be generated in liver microsomes from a commercially available compound, 2-(3,4-dihydro-1(2H)-quinolinyl)-N-(4-ethoxyphenyl)acetamide (Scheme 1; referred to hereafter as compound B), that also contains an N-alkyltetrahydroquinoline substructure. Besides metabolite structural identification, enzyme phenotyping was also done on both compounds (A and B) using a panel of cDNA-expressed P450 isoforms, MAOs, and HRP. Furthermore, cyanide trapping experiments were performed to search for postulated iminium intermediates in the formation of the quinolinium metabolites.
Materials and Methods
Materials. Compound A was synthesized as described by Holsworth et al. (2005). Compound B, i.e., 2-(3,4-dihydro-1(2H)-quinolinyl)-N-(4-ethoxyphenyl)acetamide, was purchased from Sigma-Aldrich (St. Louis, MO) (catalog number R713775; CAS 303091-40-9). Human liver microsomes were purchased from In Vitro Technologies (Baltimore, MD) (catalog number HL-Mix-14). The following P450 Baculosomes were purchased from PanVera Corp. (Madison, WI) and assayed for reaction phenotyping: CYP1A2 (catalog number P2792), CYP2B6 (catalog number P3028), CYP2C9 (catalog number P2378), CYP2C19 (catalog number P2570), CYP2D6 (catalog number R1627), CYP2E1 (catalog number P2223), and CYP3A4 (catalog number P2377). The Baculosomes are microsomes prepared from insect cells infected with a recombinant baculovirus containing a human P450 isozyme and a rabbit NADPH-P450 reductase. In addition, the following P450 Supersomes were purchased through BD Gentest (Woburn, MA) and also assayed for reaction phenotyping: CYP2A6 (catalog number 456254) and CYP2C8 (catalog number 456252). The Supersomes are microsomes prepared from insect cells infected with a recombinant baculovirus containing a human P450 isozyme, NADPH-P450 reductase, and cytochrome b5. Horseradish peroxidase (catalog number P2088), catalase from bovine liver (catalog number C40), monoamine oxidase A (catalog number M7316), and monoamine oxidase B (catalog number M7441) were purchased from Sigma-Aldrich.
Human Liver Microsome Incubations. Each 3-ml human liver microsome (HLM) incubation was carried out in 50 mM potassium phosphate buffer pH 7.4. Incubation mixtures contained NADPH at 1 mM final concentration, 1 mg/ml final protein concentration, and 50 μM final substrate concentration (compound A or B). Negative controls were set up in the absence of NADPH. All components except substrate were added to reaction mixtures on ice. The mixtures were preincubated for 3 min at 37°C in a shaking water bath, and then substrate (5 mM stock solution in methanol) was added to initiate reactions. Reactions were terminated at 90 min by quenching with ice-cold acetonitrile at a 1:1 ratio.
To isolate metabolites using preparative LC, large-scale (100-ml) HLM incubations were carried out (10 Erlenmeyer flasks containing 10 ml each) using the same concentrations and incubation time as described above. Quenched reaction mixtures were concentrated down to ∼15 to 20 ml under vacuum at room temperature using a SpeedVac (Thermo Electron Corp., Waltham, MA).
Cyanide (CN–) Trapping of Iminium Intermediates. Each 1-ml incubation was carried out in 50 mM potassium phosphate buffer pH 7.4. Incubation mixtures contained 1 mg/ml final protein concentration of pooled HLMs, 1 mM NADPH, 50 μM substrate (compound A or B), and 1 mM potassium cyanide (KCN). Negative controls were set up in the absence of human liver microsomes and NADPH. The mixtures were incubated at 37°C in a shaking water bath for 60 min, and then the reactions were terminated by quenching with ice-cold acetonitrile at a 1:1 ratio.
Reaction Phenotyping with cDNA-Expressed Cytochrome P450 Enzymes. P450 Baculosome and Supersome incubations were carried out in 50 mM potassium phosphate buffer pH 7.4. Incubation mixtures contained NADPH at 1 mM final concentration, 0.5 mg/ml final total Baculosomes or Supersomes protein concentration, and 50 μM final concentration of substrate. The incubation volume was 1 ml. All components except drug were added to vials on ice; then, the vials were preincubated for 3 min at 37°C in a shaking water bath for 3 min, and substrate was added to vials to initiate reactions. Reactions were terminated at 90 min by quenching with ice-cold acetonitrile at a 1:1 ratio.
Horseradish Peroxidase Incubations. HRP incubations were carried out in 50 mM potassium phosphate buffer pH 7.4. Incubation mixtures contained HRP at 30 units/ml final concentration, 500 μM final hydrogen peroxide (H2O2) concentration, and 50 μM final substrate concentration (compound A or B). The incubation volume was 1 ml. Negative controls were set up in the absence of either HRP or H2O2. All components except substrate were added to reaction mixtures on ice. Reactions were preincubated for 3 min at 37°C in a shaking water bath, and then substrate was added to initiate reactions. Reactions were terminated at 45 min with the addition of catalase at 300 units/ml final concentration. After an additional 3 min at 37°C in a shaking water bath, reactions were quenched with ice-cold acetonitrile at a 1:1 ratio.
MAO A and MAO B Incubations. Each 1-ml incubation was carried out in 50 mM potassium phosphate buffer pH 7.4. Incubation mixtures contained 50 μM final substrate concentration (compound A or B) and MAO A or MAO B at 0.36 mg/ml or 0.83 mg/ml final concentration, respectively. Positive controls for both MAO A and MAO B were set up with kynuramine dihydrobromide at a final substrate concentration of 50 μM. Reactions were preincubated for 3 min at 37°C in a shaking water bath, and then substrate was added to initiate reactions. Reactions were terminated at 45 min by quenching with ice-cold acetonitrile at a 1:1 ratio.
LC-MS and Preparative LC. All quenched reaction mixtures (HLMs, P450s, MAOs, HRP, or CN– trapping) were vortexed and centrifuged at 14,000 rpm in an Eppendorf 5804R centrifuge for 20 min at 4°C to pellet proteins. Aliquots (20 μl) of the supernatants were injected for LC-MS or MS/MS analysis, as summarized in Table 1.
A 5-ml aliquot of the concentrated supernatant of the large-scale HLM incubation was injected into a 21.20-mm-i.d. preparative column. The metabolites were separated as described in Table 1 and collected by the fraction collector of the preparative LC system. Fractions of the same metabolite from multiple injections were combined and the solvent was evaporated until dry under vacuum at room temperature using a SpeedVac.
1H NMR. All samples were dissolved in 0.1 ml of methanol-d4“100%” (Cambridge Isotope Laboratories, Andover, MA). 1H and TOCSY spectra were referenced using residual methanol-d4 (δ = 3.31 ppm relative to tetramethylsilane, δ = 0.00).
All NMR spectra were recorded on a Bruker Avance 600 MHz (Bruker BioSpin Corporation, Billerica, MA) controlled by XWIN-NMR V3.5 and equipped with a 2.5 mm BBI probe. One-dimensional spectra were recorded using a sweep width of 8000 Hz and a total recycle time of 5 s. The resulting time-averaged free induction decays were transformed with an exponential line broadening of 0.3 Hz to enhance signal to noise.
The two-dimensional TOCSY data were recorded using the standard pulse sequence provided by Bruker. A 1K × 128 data matrix was acquired using 32 scans and 16 dummy scans with a spectral width of 6600 Hz. A mixing time of 80 ms was used. The data were zero-filled to a size of 1K × 1K. A relaxation delay of 2 s was used between transients.
Results
The pooled human liver microsomal incubation of compound A generated several metabolites (Fig. 1), including common phase I metabolites undergoing O-dealkylation (MA4 and MA6) or phenyl hydroxylation (MA5). An unusual metabolite (MA3), with a molecular ion m/z value at 4 Da less than the parent drug, was observed as the most abundant metabolite in both the mass spectrometric total ion chromatogram (TIC) and the UV chromatogram (Fig. 1). In addition, sequential metabolites undergoing the unusual metabolic pathway (4-Da loss) and a common phase I metabolic pathway (either O-dealkylation or phenyl hydroxylation) were also detected (MA1 or MA2, respectively; Fig. 1).
The LC-MS/MS product ion spectra of MA3 versus parent drug (Fig. 2, b versus a) indicated that the 4-Da loss occurred within the 1,7-disubstituted-1,2,3,4-tetrahydroquinoline substructure of compound A. Collective evidence was gathered from the unchanged fragments at m/z 383 (Fig. 2, a and b), a 4-Da shift to lower mass from the original fragments m/z 235 and 532 of the parent to m/z 231 and 528 of MA3 (Fig. 2, b versus a), and a new fragment ion of MA3 at m/z 146 (Fig. 2b). Desaturation of the tetrahydroquinoline ring by three hydrogen atoms and the concurrent formation of quinolinium, which eliminates the need for an ionizing proton, would yield a 4-Da loss (see chemical structures in Fig. 2). High confidence in the structural elucidation of the quinolinium metabolite by MS/MS (Fig. 2b) was built upon the reliable MS/MS spectral interpretation of the parent drug (Fig. 2a), which was verified beforehand with a common metabolite (MA5, formed via phenyl hydroxylation) exhibiting fragmentation patterns similar to those of the parent drug; i.e., the predicted fragments of the metabolite (MA5) all appeared in its MS/MS spectrum at expected m/z ratios (spectrum not shown). The diagnostic MS/MS spectra in Fig. 2 were acquired by a triple quadrupole instrument, whereas initial MS/MS experiments in an ion-trap mass spectrometer induced only a selective cleavage resulting in m/z 532 and 528 fragments for parent drug and MA3, respectively. Subsequent MS3 experiments of either m/z 532 or 528 fragments produced only a single further fragment, formed by a neutral loss of 108. In short, ion trap fragmentation was insufficient for the structural identification.
The NMR characterization of the isolated metabolite, MA3, confirmed the formation of the quinolinium structure. The aromatic regions of the one-dimensional 1H NMR spectra of the metabolite and the parent drug are provided in Fig. 3 (a versus b). Fourteen aromatic protons were observed in the spectrum of the metabolite, three greater than observed in the spectrum of the parent drug. The three new aromatic protons occur between 7.8 and 9.1 ppm (protons a, b, and c; Fig. 3a). In addition, the aromatic protons of the quinolinium that correlate with the aromatic protons of the tetrahydroquinoline are noticeably shifted toward the lower field (protons d, e, and f; Fig. 3a versus 3b). The 1H-1H coupling correlations observed in the TOCSY data (Fig. 3c) provide evidence for the chemical shifts assignments. The TOCSY data clearly reveal two independent spin systems, each composed of three protons: 7.86, 9.04, and 9.11 ppm (a, b, and c) and 7.64, 7.94, and 8.29 ppm (e, f, and d, respectively). The coupling patterns and coupling constants also support these spectral assignments. The observed chemical shifts of the quinolinium protons (especially a and c) of the metabolite are in agreement with what was reported 30 years ago for quinolinium ions generated by dissolving quinoline and seven individual monomethyl derivatives in DCl-D2O (Barbieri et al., 1975). Both the metabolite and the parent drug analyzed in the NMR experiments were isolated with formic acid present in the mobile phase of the preparative LC. As a result, the aldehyde proton of formic acid (and formate) was observed in the NMR spectra at δ ∼8.6 ppm with two 13C satellite peaks (Fig. 3).
The human liver microsomal incubation of the commercially available compound B, consisting of an N-substituted-tetrahydroquinoline substructure, also produced a metabolite that was characterized by a 4-Da loss from parent drug (MB1, Fig. 4), although it was a relatively minor metabolite as indicated by the UV signal response (UV chromatogram not shown). Dramatic changes in the MS/MS fragmentation patterns from parent compound to the metabolite can be readily explained by the formation of a quinolinium structure (Fig. 5, b versus a). Supporting evidence also came from active hydrogen/deuterium (H/D) exchange accomplished in an LC-MS experiment using D2O with deuterated acetic acid as the aqueous mobile phase. Compound A has one active hydrogen atom located at the acyl aniline nitrogen. The “molecular mass” of MB1 increased by only 1 Da after the H/D exchange (Table 2, row of MB1), which is consistent with MB1 not requiring an ionizing H/D in electrospray ionization. Otherwise a 2-Da increase, as noticed in the parent drug (Table 2), would have been observed.
In the presence of 1 mM KCN added to a human liver microsomal incubation with compound A or B, a cyano adduct was observed for each compound. For example, the ion chromatogram of the reaction mixture for compound B(Fig. 4b) indicates the formation of a cyano adduct (m/z 336) of compound B (m/z 311), coupled with the corresponding diminished abundances of metabolites MB1 (the N-alkylquinolinium) and MB8. LC-MS/MS spectra of the cyano adduct of compound A (not shown) or B (Fig. 5c) pinpointed the tetrahydroquinoline moiety as the location of the cyano group for both adducts. More specifically, the loss of a HCN neutral molecule in MS/MS fragmentation of both cyano adducts suggested that the cyano group was on the tetrahydro saturated portion of the substructure. As supporting evidence to the spectral interpretation in Fig. 5c, further fragmentation of fragment ion m/z 309, i.e., the fragment by the loss of a neutral HCN from the cyano-adduct of compound B, produced fragment ions m/z 144, 150, and 281 (spectrum not shown). It was not determined whether the loss of a C2H4 or a CO (e.g., via rearrangement), both having a nominal mass of 28 Da, contributed to the formation of fragment m/z 150 (Fig. 5, b and c). The cyano group is drawn on the carbon adjacent to the nitrogen in the chemical structure shown in Fig. 5c, as rationalized by the cyanide addition reaction to iminium, although its exact location on the ring cannot be determined by MS/MS spectra.
P450 reaction phenotyping was performed using a panel of cDNA-expressed enzymes: 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1, and 3A4. Among the P450s investigated, the metabolism of compound A was catalyzed exclusively by CYP3A4. In contrast, compound B was the substrate for a number of P450 isoforms. From the disappearance of parent drug (illustrated in Fig. 6), it is evident that multiple P450s, such as 2C19, 2D6, and 3A4, can extensively metabolize compound B. The reaction phenotyping results of compound B are also presented in the format of metabolite formation (Fig. 7). Chemical structures of the metabolites drawn in Fig. 7 are based on interpretation of LC-MS/MS product ion spectra (not shown, except MB1 in Fig. 5), as well as active H/D exchange results (Table 2). For example, a hydroxylation would result in the increase of an exchangeable hydrogen atom (e.g., M2 and 6, Table 2). The quinolinium metabolite (MB1) of compound B was generated predominantly by CYP3A4, with a minor contribution from CYP2C19 (Fig. 7), whereas common phase I metabolites of compound B were produced by different P450s, e.g., O-deethylation by 2D6 and 2C19 (M4, Fig. 7), and hydroxylation at two different sites by a number of P450s (M2 and 6, Fig. 7). The regioselectivity of CYP2D6 on metabolizing compound B appeared evident in that CYP2D6 produced only one metabolite, O-deethylation MB4 (Fig. 7). It has been known in the literature that drugs metabolized by CYP2D6 contain a basic nitrogen atom and a flat hydrophobic region coplanar to the oxidation site which is approximately either 5 Å (e.g., debrisoquine) or 7 Å (e.g., dextromethorphan; 7∼8 Å, to be more precise) away from the basic nitrogen atom (de Groot et al., 1996, 1997). It is uncertain, at present, whether the tetrahydroquinoline nitrogen atom is indeed the docking site of compound B, to fit into a 7-Å model within CYP2D6.
Incubations of either compound A or B with human recombinant MAO A or MAO B did not generate any metabolites. The incubation experiments with MAO enzymes were validated by a positive control, kynuramine, a known substrate for both MAO A and MAO B. LC-UV-MS indicated that both enzymes were active in generating kynuramine's metabolite, 4-hydroxyquinoline, at high yields.
Quinolinium metabolites of both compound A and compound B were generated in significant amounts from the HRP incubations. The quinolinium metabolite MA3 was virtually the only metabolite observed, with less than 1% parent drug remaining after the HRP incubation (chromatogram not shown). The HRP incubation results for compound B are displayed with the P450 reaction phenotyping results, in Fig. 6 and Fig. 7. Among the eight metabolites of compound B observed in human liver microsomes (Fig. 7), only the quinolinium metabolite (MB1) and another metabolite (MB7) were detected in the HRP incubation mixture. The sum of metabolites (MB1 and MB7) in HRP accounted for only ∼21% of compound B-derived LC-UV peak areas (Fig. 7), whereas more than 90% of the parent compound disappeared after the HRP incubation (Fig. 6). Discordant percentage values exist between the disappearance of parent drug and the production of metabolites because other HRP oxidation products are not exhibited in Fig. 7. More specifically, two unknown peaks in the LC-UV chromatogram, one eluting 3 min earlier than MB1 and the other 0.5 min later than MB1, were observed in the HRP incubate. Unfortunately, electrospray mass spectrometry did not reveal the chemical identities, not even the molecular masses for these two LC peaks. The unidentified LC-UV peaks were not detected in the incubations with human liver microsomes or P450 isoforms. Furthermore, the formation of the unidentified products could not occur in the absence of either HRP or H2O2.
Discussion
Two chemical structural features were introduced upon N-alkylquinolinium metabolite formation: 1) a fixed charge that could exhibit notable characteristics in mass spectrometry, and 2) a polycyclic aromatic structure containing a heteroatom nitrogen, which may affect electronic spectra such as UV, etc. As summarized in Table 2, electrospray ionization produced dimer ions (two neutral molecules plus an ionizing H+, Na+, or K+ ion) for compound B and its metabolites, except for MB1. The exception of MB1 can be easily rationalized, as Columbic repulsion would hinder two fixed-charged MB1 ions adhering together in the gas phase. A different characteristic for the fixed charge of MA3 was noted in the doubly charged ion [MA3 + H]2+ at m/z (613 + 1)/2 = 307, with an intensity approximately 15% relative to the singly charged ion (M+, m/z 613). The doubly charged ion was confirmed by LC-MS zoom scan around m/z 307 in an ion trap mass spectrometer, which revealed the expected half-dalton difference in m/z values between two adjacent isotopic ions. Doubly protonated ions did not occur for parent compound A and other common phase I metabolites (such as O-dealkyl and phenylhydroxyl metabolites). Apparently, an MA3 ion with the fixed charge is large enough to hold an additional proton at a basic site; reportedly, as did fixed-charged peptide derivatives consisting of 8 or 5 residues (Gu et al., 2000; Czeszak et al., 2004). Both compounds A and B have multiple chromophores; therefore, after the metabolic aromatization, the complete replacement of old UV absorption bands with new ones was not expected. Nonetheless, a new absorption band at λmax 315 was recorded for MB1 by a photodiode array UV detector. Likewise, a new band at λmax 350 nm appeared for MA3. These λmax values are in agreement with those previously reported for respective N-methylquinolinium or methoxy-N-methylquinolinium ions, e.g., 316 nm for N-methylquinolinium iodide in ethanol (Adams et al., 1955) or 342 nm for 6-methoxy-1-methylquinolinium bromide/iodide in water (Geddes et al., 2000). Fluorescence spectroscopy of the quinolinium metabolites was not investigated, although quinolinium is known to have fluorescent excitation/emission.
Ortiz de Montellano and De Voss have recently stated: “cytochrome P450 mechanisms continue to surprise and delight, although the field is growing to maturity and the completely unexpected is less frequently encountered” (Ortiz de Montellano and De Voss, 2005). Indeed, identifying unusual metabolites that are generated by plausible metabolic pathways is intriguing to scientists in the field. A postulated pathway for the formation of N-alkylquinolinium metabolites is proposed in Scheme 3. It has been reported that P450-mediated N-dealkylation of N,N-dialkylanilines occurs via single-electron transfer (SET) from the heteroatom nitrogen, e.g., based on the low kinetic isotope effect of deuterium versus hydrogen at the α-carbon on the N-dealkylation (Miwa et al., 1983; Hollenberg et al., 1985; Guengerich et al., 1996). Furthermore, the formation of the quinolinium metabolites of both A and B in HRP suggests that metabolic aromatization of the tetrahydroquinoline can be initiated via the SET mechanism, since the oxidation of aromatic amine by peroxidase is believed to undergo the amine free radical mechanism (Eling et al., 1991). It should also be noted that Shaffer et al. (2001) have previously discovered the HRP oxidation of N-methyl-1,2,3,4-tetrahydroquinoline to N-methylquinolinium during their investigation of the HRP oxidation mechanism of N-cyclopropyl-N-methylaniline to an quinolinium ion (Scheme 2; Shaffer et al., 2001). Therefore, it is assumed that the P450-catalyzed aromatization of the alkyltetrahydroquinolines undergoes an initial SET step to generate a radical nitrogen cation (II; Scheme 3). The postulation of the pathway from structure IV to structure VI (Scheme 3) was originated from the radical mechanism for the P450-catalyzed oxidation of 4-alkyl-1,4-dihydropyridine derivatives bearing various substitutions (Augusto et al., 1982; Böcker and Guengerich, 1986; Guengerich and Böcker, 1988; Lee et al., 1988). Augusto et al. (1982) reported the radical mechanism for the oxidation of the dihydropyridines on the basis of spin-trapping of the alkyl radical released from the 4-position, and the mechanism was further supported by more evidence, e.g., no significant isotope effect for the 4-position hydrogen on the dehydrogenation (Guengerich and Böcker, 1988). An alternative mechanism, in which the oxidation of the 1,4-dihydropyridines can be mediated by trace iron cations outside the P450 active site and dependent on the presence of H2O2 produced by P450 uncoupling, has been suggested by one group (Kennedy and Mason, 1990). However, CYP3A4-catalyzed aromatization of the 1,4-dihydropyridines has been reported by another group with purified P450 enzymes in systems where lipid peroxidation was not possible and effects of metal chelators were not observed (Guengerich et al., 1991).
The dihydro intermediates III and IV, which would provide direct evidence for the postulated mechanism in Scheme 3, were not observed for either of the substrates. Likewise, Shaffer et al. (2001) did not detect similar dihydro intermediates in their HRP incubation of N-cyclopropyl-N-methylquinoline (Scheme 2; Shaffer et al., 2001). Their proposal of an assumed 1-methyl-3,4-dihydroquinolinium intermediate was based on cyanide trapping of the iminium intermediate (Shaffer et al., 2001). On the contrary, dihydro intermediates in the formation of 1-methyl-4-phenylpyridinium (Scheme 2; Castagnoli et al., 2002), N-methylisoquinolium (Scheme 2; Booth et al., 1989), and thienopyridinium (Dalvie and O'Connell, 2004) from their tetrahydro analogs have been clearly observed. Metabolites MB2 and MB8, appearing at m/z 309 in the LC-MS spectra, were once suspected to be dihydro intermediates from compound B (m/z 311). However, it was later discovered that the m/z 309 ion of MB2 was an in-source fragment, formed by the loss of H2O from the [M + H]+ ion of the hydroxyl metabolite. Evidence included dimer ions of MB2 (Table 2), as well as intact [M–H]– ions of MB2 detected in negative ion electrospray LC-MS and the MS/MS spectrum of the [M–H]– ions (not shown). For MB8 structural identification, the H/D exchange results (Table 2) permitted only two possibilities: 1) an iminium (with a fixed positive charge) that does not require an ionizing proton in electrospray, which, however, would be contradictory to the MB8 retention time being slightly later than that of the parent B in reversed-phase LC and would also not be supported by MS/MS (spectrum not shown); or 2) a ring-closed structure that eliminated the active hydrogen at the acyl aniline nitrogen of compound B. In short, MB8is clearly not a dihydroquinoline intermediate. The most likely structure of MB8 (proposed in Fig. 7) is the result of intramolecular trapping of the iminium intermediate of compound B, since strong evidence for iminium intermediates was obtained by cyanide trapping in liver microsomal incubations (Scheme 3).
In addition to the absence of dihydro intermediates, the absence of liver microsomal generated N-oxide metabolites for the N-alkyltetrahydroquinolines in the present study was another notable metabolism difference from the previously reported MPTP derivatives (Castagnoli et al., 2002; Bissel and Castagnoli, 2005) and other substructures somewhat similar to MPTP (Christensen et al., 1999; Dalvie and O'Connell, 2004). The active H/D exchange experiment provided evidence that none of the detected mono-oxidation metabolites of compound B were N-oxides. This difference in N-oxygenation is not surprising, since it has been reported that the rate of N-dealkylation versus N-oxygenation is 940:1 in P450-catalyzed oxidation of N,N-dimethylaniline (Seto and Guengerich, 1993). On the other hand, the previously investigated substrates (cited above) have aliphatic tertiary amine nitrogen atoms which are prone to N-oxygenation.
Several relevant perplexing issues could not be addressed within the scope of this study. First, CYP3A4 is predominantly responsible for generating the quinolinium metabolites. It is unclear whether this is merely another example of CYP3A4's great promiscuity (Smith, 2003) or a reflection of CYP3A4 being the principal participant in oxidation of tertiary amines (e.g., N-demethylation; Smith, 2003). Second, whereas the iminium intermediate from MPTP or N-methyl-tetrahydroisoquinoline is relatively stable (Scheme 2), the iminium intermediate III might be converted into IV (Scheme 3). Could IV be more easily converted to quinolinium than III? Could the rapid conversions (III → IV → quinolinium) explain the absence of dihydro intermediates? Third, whereas MPTP and N-alkyltetrahydroisoquinoline have an aliphatic tertiary amine nitrogen (Scheme 2), the heteroatom in the N-alkyltetrahydroquinoline substructure is essentially a dialkylaniline nitrogen. It is relatively easier for an N,N-dialkylaniline to form a radical cation than it is for an aliphatic tertiary amine to do so. Evidence can be readily gathered from the ionization energy (IE) of the analogs, measured in the gas phase for M → M+·. For example, N,N-dimethylaniline has an IE of 7.12 ± 0.02 eV, whereas the IE is greater for N,N-dimethyl-N-benzylamine (7.69 ± 0.05 eV), which is very close to that of trimethylamine (7.85 ± 0.05 eV) (Lias, 2005). For comparison, toluene (absent of nitrogen) has an IE of 8.828 ± 0.001 eV (Lias, 2005). It is uncertain whether the low IE of N,N-dialkylaniline implies that an N-alkyltetrahydroquinoline could be more prone to the SET oxidation mechanism (I to II; Scheme 3) than an N-alkyltetrahydroisoquinoline or MPTP-like structure, an aliphatic tertiary amine with a benzylic or allylic methylene adjacent to the nitrogen. It would be our hope that the answers to these questions will be explored in the future.
Acknowledgments
We are grateful to the following Pfizer colleagues: Dr. Deepak Dalvie (La Jolla Laboratories), Dr. Benny M. Amore (Groton Laboratories), and Dr. Gwendolyn D. Fate (Ann Arbor Laboratories) for insightful discussions and helpful suggestions. We also thank Dr. Suzie A. Ferreira for requesting in vitro metabolite profiling of the renin discovery compounds including compound A. We are also indebted to Dr. Gwendolyn D. Fate for manuscript revision suggestions.
Footnotes
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.106.012286.
-
ABBREVIATIONS: RAS, renin-angiotensin system; P450, cytochrome P450; HLM, human liver microsome; HRP, horseradish peroxidase; IE, ionization energy; LC, liquid chromatography; MAO, monoamine oxidase; MS, mass spectrometry; MS/MS, tandem mass spectrometry; MPTP, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine; TOCSY, total correlated spectroscopy (a two-dimensional NMR); SET, single-electron transfer; KCN, potassium cyanide; TIC, total ion chromatogram; H/D, hydrogen/deuterium.
- Received July 28, 2006.
- Accepted September 13, 2006.
- The American Society for Pharmacology and Experimental Therapeutics