Generation of Quinoneimine Intermediates in the Bioactivation of 3-(N

Eosinophilia-myalgia syndrome (EMS) was an intoxication episode that occurred in the US in 1989 and affected 1,500 people. EMS was associated with the...
0 downloads 0 Views 166KB Size
Download PDF
1556

Chem. Res. Toxicol. 2007, 20, 1556–1562

Generation of Quinoneimine Intermediates in the Bioactivation of 3-(N-Phenylamino)alanine (PAA) by Human Liver Microsomes: A Potential Link Between Eosinophilia-Myalgia Syndrome and Toxic Oil Syndrome Anna Martínez-Cabot and Angel Messeguer* Department of Biological Organic Chemistry, Institut d’InVestigacions Químiques i Ambientals de Barcelona (CSIC), Barcelona, Spain ReceiVed July 13, 2007

Eosinophilia-myalgia syndrome (EMS) was an intoxication episode that occurred in the US in 1989 and affected 1,500 people. EMS was associated with the ingestion of manufactured L-tryptophan, and 3-(N-phenylamino)alanine (PAA) was identified as one of the contaminants present in the L-tryptophan batches responsible for intoxication. In previous studies (Martínez-Cabot et al., Chem Res. Toxicol., in press), we have shown that the incubation of 3-(N-phenylamino)propane-1,2-diol (PAP), a toxic biomarker of the oil batches that caused Toxic Oil Syndrome in Spain, with human liver microsomes generates a reactive quinoneimine intermediate. The structural similarity between PAA and PAP led Mayeno and co-workers (Mayeno et al. (1995) Chem. Res. Toxicol. 8, 911–916) to hypothesize that both xenobiotics could be linked to a common etiologic agent. We thus set about to study the bioactivation of PAA by human liver microsomes. Under these conditions, PAA is converted to its 4′-hydroxy derivative, an unstable intermediate that is rapidly transformed into the final metabolites 4-aminophenol and formylglycine, which were identified in the incubations by GC/MS using the H218O-labeled medium. We also provide evidence that 4-aminophenol and formylglycine are formed from a quinoneimine intermediate via a pathway similar to that demonstrated for PAP bioactivation. This quinoneimine, in the absence of nucleophiles in the incubation medium, could isomerize to give the corresponding imine, which could undergo hydrolysis to yield the aforementioned final products. These findings establish that EMS and TOS are linked by a common toxic metabolite (4-aminophenol) and that they may be further linked by the concomitant release of potentially hazardous carbonyl species. Introduction 1

Eosinophilia-myalgia syndrome (EMS) was an intoxication episode that occurred primarily in the US in 1989 and affected 1,500 people, causing over 30 deaths (1, 2). Cases of EMS were also described in Europe and Canada (3). EMS was attributed to the consumption of tainted L-tryptophan from a specific manufacturer in Japan (4) and was shown to have been triggered by a contaminant rather than the amino-acid itself. Initially, three contaminants were implicated: 1,1′-ethylidenebis[L-tryptophan], 3-(N-phenylamino)-L-alanine (PAA), and 2-(3-indolylmethyl) tryptophan (5, 6). Online HPLC-tandem mass spectrometry has recently permitted the identification of additional contaminants (7). Although neither the compound responsible for EMS nor the molecular mechanisms responsible for its toxicology have been conclusively established, many studies have focused on PAA and its bioactivation products. Specifically, in ViVo assays on the urine of PAA-treated rats have revealed para-hydroxyanilino derivatives of PAA (8) (Scheme 1).

Our previous experience (Martínez-Cabot et al., Chem Res. Toxicol, in press) with metabolism studies on 3-(N-phenylamino)propane-1,2-diol (PAP), a toxic biomarker of toxic oil syndrome (TOS) (9, 10), and the structural similarity between PAA (Scheme 1) and PAP (Scheme 2a), raised our attention to PAA. Because the presence of hydrogens at the carbon atom bonded to the anilino nitrogen of PAP directs its bioactivation, we sought to establish whether the same is true for the bioactivation of PAA. Herein, we report the results of studies on the bioactivation of PAA promoted by human liver microsomes (HLM). We show that under these conditions, PAA is converted into an unstable phenol derivative that undergoes a sequence of reactions to give potentially toxic compounds as intermediates and as final products. In view of the results obtained and taking into account previous suggestions that EMS and TOS may share a common etiologic agent (11, 12), we also discuss a new potential link between the toxicological mechanisms of each syndrome.

Experimental Procedures * To whom correspondence should be addressed. Dr. Angel Messeguer, Institut d’Investigacions Químiques i Ambientals de Barcelona (CSIC), Jordi Girona 18-26, 8034 Barcelona, Spain. Tel: +34-93-400 61 21. Fax: +3493-204 59 04. E-mail: [email protected]. 1 Abbreviations: BSTFA, bis-(trimethylsilyl)trifluoroacetamide; 2,4DNPH, 2,4-dinitrophenylhydrazine; EMS, Eosinophilia-Myalgia Syndrome; ESI-Hrms, electrospray high resolution mass spectrometry; HLM, human liver microsomes; NAC, N-acetylcysteine; PAA, 3-(N-phenylamino)alanine; PAP, 3-(N-phenylamino)propane-1,2-diol; SIM, selected ion monitoring; TOS, Toxic Oil Syndrome.

Chemicals Reagents and Suppliers. Glycidol (Merck, 98%) was distilled, and then stored at -20 °C. Aniline, 4-aminophenol, 4-benzyloxyaniline hydrochloride, silver oxide, D,L-glyceraldehyde and ZnSO4 were obtained from Aldrich-Chemical Co (Milwaukee, WI). PAP was synthesized as previously described (13). Ammonium acetate and 10% palladium/charcoal were from Merck (Darmstadt, Germany). β-Chloro-L-alanine hydrochloride, glutathione and NADP(H) were from Sigma Chemical Co. (St. Louis, MO). 2,4-

10.1021/tx700256v CCC: $37.00  2007 American Chemical Society Published on Web 09/25/2007

Quinoneimine Intermediates in PAA BioactiVation

Chem. Res. Toxicol., Vol. 20, No. 10, 2007 1557

Scheme 1. Proposed Steps for the Generation of Two of the Metabolites Detected in PAA-Treated Rats (8)

Scheme 2. (a) Formation of NAC Conjugate 2 from PAP (15) (Also See Martínez-Cabot et al., Chem Res. Toxicol, in Press) and (b) Generation of Glyceraldehyde from 2 and Identification of Its 2,4-DNPH Derivative

Figure 1. (A) HPLC profile of the crude reaction mixture resulting from the oxidation of NAC conjugate 2 and further reaction with 2,4DNPH. (B) HPLC profile of the standard hydrazone derivative 3. The UV spectra a and b correspond to the peaks marked with arrows.

DNPH was from Fluka (Buchs SG, Switzerland). HLM were obtained from Advancell (Barcelona, Spain), and water (H218O, 95.4%) was from Eurisotop (Saint Aubin, France). Acetonitrile was of HPLC grade and purchased from VWR Prolabo (Leuven, Belgium). All steps involving quinoneimines were carried out under argon atmosphere and protected from light. General Methods and Instruments. The products were purified on a semipreparative RP-HPLC Waters (Milford, MA, U.S.A.) system using an XTerra Prep. MS C18 column (15 × 1 cm, 5 µm, Waters, Milford, USA). HPLC analyses were performed on a Hewlett-Packard 1100 system equipped with a DAD detector and employing a Kromasil 100 C18 column (25 × 0.4 cm, 5 µm, Scharlab, Spain). Eluents consisted of 10 mM NH4OAc (pH 7.0) in water (A) and CH3CN (B) at a flow rate of 1 mL/min, and the gradient used, unless another is indicated, was an isocratic step at 100% (A) for 1 min, then, a linear ramp from 100% (A) to reach 95% (A) in 14 min, a second ramp from 95% (A) to reach 60% (A) in 5 min, and finally a return to initial conditions. The detection was set at 220, 245, or 300 nm as specified. Microwave-assisted reactions were performed in a Discover (North Carolina, USA) monomode CEM microwave reactor. Incubation samples were centrifuged in a Biofuge pico centrifuge (Heraeus Instruments, Osterode, Germany). The 1H-NMR spectra (500 MHz) were recorded on a Varian INOVA 500 spectrometer and the 13C-NMR spectra (100 MHz) in a Varian Mercury 400 spectrometer. Spectra were taken in neutralized CDCl3 or perdeuterated DMSO as indicated. Chemical shifts are given in ppm related to TMS for 1 H. ESI-Hrms data were obtained with a UPLC Acquity (Waters, Mildford, USA) coupled to a mass spectrometer model LCT Premier XE (Waters) provided with a TOF analyzer in a positive or negative mode as indicated. An Acquity UPLC BEH C18 column (1.7 µm 2.1 × 50 mm, Waters) was employed for the UPLC analyses. ESIMS were obtained with a Hewlett-Packard 1100 LC/MS system in

the positive ionization mode. GC-MS analyses were carried out with a TRACE (Thermo Flinnigan, Manchester, UK) GC apparatus. The column used was an Agilent HP-MS (30 m × 0.25 mm, 0.25 µm phase thickness). The apparatus conditions were as follows: helium as carrier gas at a flow rate of 1 mL/min, temperature gradient from 100 to 300 °C at a rate of 4 °C/min, splitless injection mode at 250 °C, and electron-impact mass spectra using SIM mode. Samples were lyophilized in a LABCONCO (Kansas City, MO) unit. 1. Synthesis of the Standard of the 2,4-Dinitrophenylhydrazone of Glyceraldehyde (3). The standard was synthesized according to a literature protocol (14). A mixture of glyceraldehyde (1 mg, 0.005 mmol) and 2,4-DNPH (3.3 mg, 0.025 mmol) in 2 N HCl (6.6 mL) was allowed to react for 15 min at 20 °C under protection from light. The reaction was monitored by HPLC using the following gradient: an isocratic step (from 0 to 10 min) at 1 mL/min with 62% A (0.1% TFA in water) and 38% B (0.1% TFA in CH3CN), followed by a linear ramp to reach 25% A (from 10 to 20 min). The UV detector was set at 365 nm. The HPLC profile of the crude reaction mixture showed a major peak at 4.7 min that was identified by LC/MS as the expected hydrazone derivative 3: UVmax at 238, 260, and 365 nm, ESI-MS 271(M + H)+. 2. Detection of 2,4-Dinitrophenylhydrazone 3 in the Decomposition of 3-[4′-Amino-3′-(N-acetylcystein-S-yl)hydroxyphenyl]-propane-1,2-diol (2). Adduct 2 was synthesized according to the procedure of Martínez-Cabot et al. (15). A solution of pure 2 (0.5 mg/mL) in 50 mM, pH 7.4 phosphate buffer (200 µL) was maintained for 15 h at 37 °C. An excess of 2,4-DNPH was then added, and the mixture was allowed to react for 15 min at 20 °C (see Scheme 2b). HPLC analysis of the crude reaction mixture using the conditions described above showed the formation of a new product with the same retention time and UV spectrum as those of standard 3 (see Figure 1). 3. Synthesis of 3-(4′-(Benzyloxy)Phenylamino)-2-aminopropanoic Acid (4). The compound was synthesized according to a literature protocol (11). A mixture of β-chloro-L-alanine (10 mg, 0.06 mmol), 4-benzyloxyaniline (12 mg, 0.06 mmol), and NaHCO3 (6.4 mg, 0.06 mmol) in 4:1 water/CH3CN (2.4 mL) was stirred for 5 min at 155 °C under microwave irradiation (250 W) (see Scheme 3). The crude mixture was allowed to cool, and the resulting solid was filtered through Celite and washed with water and CH3CN to obtain a white powder that was identified as the benzylic ether 4 (9% yield, 97% purity by HPLC). 4: 1H NMR (CD3OD) δ: 7.39 (d, 2 H; J ) 8; H-3′′, H-5′′), 7.28 (t, 1 H; J ) 7.5; H-4′′), 6.84 (d, 2 H; J ) 9; H-2′′, H-6′′), 6.84 (d, 2 H; J ) 9; H-3′, H-5′), 6.71 (d,

1558 Chem. Res. Toxicol., Vol. 20, No. 10, 2007

Martínez-Cabot and Messeguer

Scheme 3. (a) Synthesis of PAA and Its 4′-Hydroxy Derivative 5 and (b) Oxidation of Phenol Derivative 5 to Give 4-Aminophenol and Formylglycine

Figure 2. HPLC profiles of the oxidation of phenol 5 by silver oxide at (A) t ) 0 min, (B) t ) 5 min, and (C) t ) 20 min. (D) HPLC profile of standard 4-aminophenol. The inset corresponds to the ESI-Hrms spectrum of the new peaks formed, identified as 4-aminophenol.

2 H; J ) 9; H-2′, H-6′), 4.98 (s, 2 H, CH2-Ph), 3.84 (dd, 1H, J1 ) 4, J2 ) 9, CH2), 3.62 (dd, 1 H; J1 ) 4, J2 ) 13.5, CH2), 3.45 (m, 1 H; CH); 13C NMR (DMSO) δ: 169 (C-1), 150 (C-4′), 141 (C1′′), 128.3–127.3 (C-2′′, C-3′′, C-4′′, C-5′′, C-6′′), 115.8–113.9 (C2′, C-3′, C-5′, C-6′), 70.3 (CH2-O), 51.9, (C-3), 44.2 (C-2). ESIHrms positive mode: m/z 287.1406; m/z for (M + H)+C16H19N2O3 requires 287.1396. 4. Synthesis of 3-(4′-Hydroxyphenylamino)-2-aminopropanoic Acid (5). The phenol derivative of PAA was synthesized according to a literature protocol (15). A mixture of 4 (10 mg, 0.035 mmol) and 10% Pd/C (3.7 mg, 0.003 mmol) in methanol (3 mL) was stirred for 2 h at 20 °C and then filtered through Celite. The HPLC profile of the crude reaction mixture showed a peak that was identified as the expected phenol 5. 5: 1H NMR (CD3OD) δ: 7.34 (m, 1 H, Har), 7.28 (m, 1 H, Har), 7.03 (d, 1 H, J ) 9, Har), 6.82 (d, 1 H, J ) 9, Har) 3.79 (dd, 1 H, J1 ) 4, J2 ) 8, CH2), 3.59 (dd, 1 H, J1 ) 4, J2 ) 13.5, CH2), 3.36 (dd, 1 H, J1 ) 8.5, J2 ) 14, CH); 13C NMR (CD3OD) δ: 172 (C-1), 150.9 (C-4′), 142.2 (C-1′), 116.9–116.2 (C-2′, C-3′, C-5′, C-6′), 55.6 (C-3′), 47.2 (C-2). ESI-Hrms positive mode: m/z 197.0919; m/z for (M + H)+C9H13N2O3 requires 197.0926. 5. Generation of Quinoneimine 6. Following a literature procedure (15), phenol 5 was oxidized by silver oxide in a methanol/ CH3CN (1:1) mixture to obtain quinoneimine 6. The reaction course was monitored by HPLC; unfortunately, all efforts to detect 6 in the crude reaction mixture were unsuccessful. The predominant product in the crude reaction mixture was identified as 4-aminophenol by Hrms and by comparison with a standard (see Figure 2). ESI-Hrms positive mode: m/z 110.0610; m/z for (M + H)+C6H8NO requires 110.0606. 6. Synthesis of 3-(N-Phenylamino)alanine (PAA). A mixture of β-chloro-L-alanine (120 mg, 0.75 mmol), aniline (70 mg, 0.75 mmol), and NaHCO3 (63 mg, 0.75 mmol) in water (3 mL) was stirred for 25 min at 105 °C under microwave irradiation (250 W). The reaction was monitored by HPLC using the following gradient: a linear ramp at 80% A (10 mM, pH 7.0 NH4OAc) and 20% B (CH3CN) to reach 50% A in 15 min at 1 mL/min. The UV detector was set at 220 nm. The crude reaction mixture was purified by semipreparative HPLC using 10 mM, pH 7.0 NH4OAc buffer (A) and CH3CN (B). The gradient used consisted of a linear ramp starting at 98% A to reach 90% A in 25 min at 5 mL/min. PAA

was isolated in 9% yield and 98% purity as determined by HPLC. ESI-Hrms and NMR spectral data for PAA were in agreement with those described in the literature (11). In Vitro Assays. 1. Incubations with PAA. Incubations with HLM were performed in 50 mM, pH 7.4 phosphate buffer with 5 mM MgCl2. The incubations contained 5 mM substrate (PAA) dissolved in 0.05% acetic acid solution and HLM (1.57 mg/mL) and were initiated by the addition of the NADP(H)-regenerating system (1 mM). Samples were incubated at 37 °C, the reaction was stopped by cooling 150 µL aliquots of each sample on ice and then adding 50 µL of 125 mM ZnSO4 solution. The mixtures were centrifuged at 10,000 rpm for 10 min, and the supernatants were extracted for HPLC analysis. The UV detector was set at 300 nm. Under these conditions, a new peak appeared at 4.7 min, and it was identified as phenol 5 by Hrms in negative mode and by comparison with the corresponding standard previously prepared (see Figure 3). 5: UVmax at 210, 245, and 305 nm. ESI-Hrms negative mode: m/z 195.0781; m/z for (M + H)+C9H11N2O3requires 195.0771. For GC-MS analysis, samples were lyophilized, redissolved in CH3CN, filtered through a 0.45 µm PVDF filter, and then treated with BSTFA prior to injection. The SIM mode was used to detect 4-aminophenol and formylglycine by employing combinations of peaks at m/z 149, 166, and 181 and m/z 147, 175, and 177, respectively. Aqueous (H218O) in Vitro assays were performed following the same procedure described above but using a 1:1 H216O/H218O buffer solution to obtain a final H216O/H218O ratio of 3:2 in each sample incubation. 2. Incubations with Phenol 5. Incubations of PAA metabolite 5 (250 µM) with HLM (1.53 mg/mL) in the presence and absence of NADP(H) (1 mM) were performed in 50 mM, pH 7.4 phosphate buffer with 5 mM MgCl2. Incubations of 5 in absence of microsomes were also carried out. In all cases, samples were incubated at 37 °C, and aliquots of 150 µL were drawn at different times. The reactions were stopped by cooling the aliquots of each sample on ice and then adding 50 µL of 125 mM ZnSO4 solution. The mixtures were centrifuged at 10,000 rpm for 10 min, and the supernatants were extracted for HPLC analysis. The UV detector was set at 300 nm. Using these conditions, the formation of a new

Quinoneimine Intermediates in PAA BioactiVation

Figure 3. Phenol 5 is a metabolite of PAA. HPLC profiles of (A) in Vitro incubation of PAA with HLM and NADP(H) after 15 min and (B) the standard 4′-hydroxy metabolite 5. The insets show the corresponding UV spectra of the peaks marked with arrows. The ESIHrms spectrum of peak (a) is also shown.

peak at 10.5 min was detected in all cases, and it was identified as 4-aminophenol by Hrms and by comparison with the corresponding standard. All incubations were performed in triplicate. Hrms positive mode: m/z (HLM with NADP(H)) 110.0603, m/z (HLM without NADP(H)) 110.0601, and m/z (buffer with NADP(H)) 110.0606; (M + H)+C6H8NO requires 110.0606. Phenol 5 (1.25 mM) was also incubated in 50 mM, pH 7.4 phosphate buffer with 5 mM MgCl2 in the presence of glutathione (2.5 mM) at 37 °C. Aliquots of 150 µL were drawn at different times, and were analyzed by HPLC. The UV detector was set at 245 nm. Under these conditions, the formation of a new peak at 8 min was detected. This peak showed UV spectra typical of aminophenol-glutathione conjugates (15) (see Figure 8 in Supporting Information) and was identified as conjugate 8 by Hrms in positive mode: m/z 502.1604, m/z for (M + H)+C19H28N5O9S requires 502.1608. UVmax at 210, 245, and 320 nm.

Results Detection of Glyceraldehyde from the Conversion of an N-Acetylcysteinyl Quinoneimine Conjugate of PAP. We recently reported (15) that conjugates related to the glutathione family could be generated by trapping the quinoneimine intermediate that results from the oxidative activation of 3-(Nphenylamino)propane-1,2-diol (1), a PAP bioactivation product (see Scheme 2a and Martínez-Cabot et al., Chem Res. Toxicol., in press). We also showed that these conjugates are unstable, and readily oxidize to give new quinoneimine adducts that can ultimately be converted into glyceraldehyde and the corresponding 4-aminophenol conjugates (Scheme 2b). It was anticipated that HPLC detection of simple aldehyde derivatives such as glyceraldehyde (for TOS studies) or formylglycine (for EMS

Chem. Res. Toxicol., Vol. 20, No. 10, 2007 1559

studies) in this reaction media could be troublesome because of their potential instability and lack of strong chromophores. We therefore decided to identify the derivatives indirectly, by converting them into their respective 2,4-dinitrophenylhydrazones (14, 16). Toward this aim, the sequence involving the decomposition of the N-acetylcysteine (NAC) conjugate of PAP to give ultimately glyceraldehyde and the corresponding 4-aminophenol conjugate (Scheme 2b) was selected as a model process before studying the case of PAA. The 4-aminophenol conjugate of PAP had been previously identified by HPLC and characterized (15). Thus, to the crude reaction mixture resulting from the oxidation and further hydrolysis of the NAC conjugate of PAP 2 was added an excess of 2,4-DNPH. As expected, in addition to the detection of the respective 4-aminophenol conjugate, the formation of hydrazone 3 in the crude reaction mixture was confirmed by comparing its HPLC retention time with that of an independently prepared standard (Figure 1). This result enabled us to undertake an analogous study on the in Vitro bioactivation pathway followed by PAA. Generation of the Quinoneimine Derivative of PAA. In PAA and PAP, the carbon atom contiguous to the aniline nitrogen bears hydrogens. We anticipated that quinoneimine intermediate 6 would be unstable compared to the quinoneimine derived from PAP and thus follow a similar metabolic pathway to give 4-aminophenol and the corresponding aldehyde, formylglycine (15). To test this hypothesis, we acquired phenol derivative 5, which was synthesized via microwave-assisted reaction of 4-benzyloxyaniline with β-chloro-L-alanine to give ether 4, followed by hydrogenolysis to remove the benzyl protecting group (Scheme 3a). In order to generate quinoneimine derivative 6, phenol 5 was treated with silver oxide (Scheme 3b). Unexpectedly, quinoneimine 6 could not be detected by HPLC-MS; instead, rapid formation of 4-aminophenol was observed (Figure 2). Conversion of phenol 5 to 4-aminophenol was practically quantitative within 20 min of reaction. This result indicated that quinoneimine 6 was more reactive than the quinoneimine derived from PAP, a species that could be detected by HPLC-MS (15). Nevertheless, the detection of 4-aminophenol suggested that 6 had indeed been generated and had gone on to isomerize to the corresponding Schiff base 7, and, finally, had undergone hydrolysis. As also expected, formylglycine, the product accompanying the formation of 4-aminophenol, could not be detected by HPLC. All efforts to synthesize this aldehyde (not reported to date) for use as a standard were unsuccessful, probably because of the inherent instability of this kind of R-aminoaldehyde (17–19). In Vitro Assays of PAA with Human Liver Microsomes. Adachi et al. reported that in ViVo metabolism of PAA in rats produced 4-aminophenol derivatives, among other compounds. These metabolites could be generated upon the formation of phenol 5 as the intermediate (8) (Scheme 1). However, we have shown that phenol 5 could undergo silver oxide-promoted oxidation followed by transformation to give 4-aminophenol. In addition, we have previously shown that HLM converts PAP into the corresponding phenol derivative by hydroxylation at the activated para position of the aromatic ring (Scheme 2a, Martínez-Cabot et al., Chem Res. Toxicol., in press). These findings suggested that PAA could follow bioactivation similar to that observed for PAP. We thus set-up in Vitro incubations of PAA (prepared by the sequence depicted in Scheme 3a) with HLM and NADP(H). After 15 min, the formation of phenol derivative 5 was detected by HPLC and confirmed by Hrms (Figure 3). It should be noted that compound 5 is highly sensitive to oxidation, the major product of which is most likely the

1560 Chem. Res. Toxicol., Vol. 20, No. 10, 2007

Martínez-Cabot and Messeguer

Figure 4. GC-MS profile using SIM mode (m/z: 147, 175, 177) of a sample corresponding to the in Vitro incubation of PAA with HLM and NADP(H) in the presence of a 3:2 H216O/H218O mixture in the buffer solution for 15 h at 37 °C and subsequent silylation with BSTFA. An extract ion application was used to search for peaks with m/z 177, m/z 175, and m/z 147. The peak at 6.8 min in the three profiles shows the mass spectra containing the three m/z values searched as well as the anticipated 3:2 ratio for m/z 177 and 175.

corresponding quinoneimine 6. These results confirmed that PAA, analogous to PAP, is transformed by HLM to give the corresponding 4′-hydroxy derivative. Consequently, phenol derivative 5 could be considered as a metabolite of PAA bioactivation, in addition to being capable of generating potentially toxic intermediates through the formation of the corresponding quinoneimine oxidation product 6. Detection of 4-Aminophenol in the Incubations of PAA with Human Liver Microsomes. Assuming the high instability of phenol derivative 5 and quinoneimine intermediate 6, an incubation of PAA with HLM and NADP(H) was performed in which the crude mixture was allowed to stand overnight. It was anticipated that the absence of nucleophiles in the incubation medium would ultimately favor the conversion of 6 into 4-aminophenol and formylglycine, as observed in the assays performed with silver oxide. GC/MS analysis using the SIM mode (m/z 149, 166, and 181) of a sample (presilylated with BSTFA) showed the presence of the O-silylated derivative of 4-aminophenol (see Figure 7 in Supporting Information). Identification of this end metabolite from the bioactivation of PAA prompted us to detect the formation of the concomitant metabolite formylglycine. Detection of Formylglycine by GC-MS Using H218O in the Incubation Medium. Since we were unsuccessful in our efforts at synthesizing formylglycine for use as a standard, we required an alternate strategy for its identification in the incubation experiments. According to the pathway for the evolution of quinoneimine 6 in the absence of nucleophiles (Scheme 3b), by using a mixture of H216O and H218O as the solvent for the incubation medium, we would be able to unambiguously identify formylglycine by GC/MS. Thus, incubations of PAA with HLM and NADP(H) using a 3:2 H216O/ H218O mixture of water were carried out under the same conditions as those described above. The crude reaction mixtures were lyophilized, treated with BSTFA, and then analyzed by GC-MS. We reasoned that if water from the incubation medium was responsible for the hydrolysis, then two isotopomers of formylglycine, with different molecular mass peaks, would be obtained in an approximately 3:2 ratio. To increase detection sensitivity, the SIM mode (m/z 147, 175, and 177) was used,

Figure 5. Time-course formation of 4-aminophenol from incubations of phenol 5 (250 µM) at 37 °C. Incubations were performed in three different conditions: in the presence of HLM without NADP(H), in the presence of HLM with NADP(H), and in buffer solution with NADP(H). The bars indicate average amounts of 4-aminophenol from triplicate measurements.

whereby 175 and 177 correspond to the mass peaks of the 16Oand the 18O-products, respectively, and 147 to the mass of either product after decarbonylation, which is a typical MS fragmentation for aldehydes. As shown in Figure 4, a peak at 6.8 min was observed in three chromatograms, all of which also exhibited mass spectra containing the expected 147, 175, and 177 m/z values. In addition, the ratio of peaks at 175 to those at 177 was approximately 3:2. These results confirmed that formylglycine is indeed the co-metabolite of 4-aminophenol in the bioactivation of PAA by HLM. Formation of 4-Aminophenol in Incubations of Phenol 5. To confirm the generation of 4-aminophenol through the participation of phenol 5 as the intermediate in PAA incubations and also to study the enzymatic role in the oxidation step in this pathway, incubations of 5 with HLM in the presence and absence of NADP(H) and also in buffer medium with NADP(H) were carried out. The time course study showed the increase in the 4-aminophenol formation rate in all cases, especially in HLM in the absence of NADP(H) conditions (see Figure 5). This result could be explained by the reductive properties of NADP(H), which may protect phenol 5 from its oxidation. In any case, oxidation of 5 to the corresponding quinoneimine intermediate 6 might be mainly a nonenzymatic process.

Quinoneimine Intermediates in PAA BioactiVation

Chem. Res. Toxicol., Vol. 20, No. 10, 2007 1561

Scheme 4. Proposed Pathway for the Generation of Glutathione (GSH) Conjugate 8 from Phenol 5 via Quinoneimine Intermediate 6

Detection of Conjugate 8 Formation. As described, isolation of quinoneimine 6 from phenol 5 oxidation in the presence of silver oxide was not achieved because of its high instability. Actually, this intermediate generation could not be even detected by HPLC because it is rapidly transformed to 4-aminophenol (see Figure 2). However, quinoneimine 6 formation could be confirmed indirectly by trapping as a thiol conjugate. Thus, considering the phenol 5 autoxidation previously observed, we attempted to generate the corresponding glutathione conjugate 8 by incubating 5 with glutathione in a buffer medium at 37 °C (Scheme 4). The analysis of the samples by HPLC showed the formation of a new peak (data not shown) with a UV spectrum typical of aminophenol-glutathione conjugates. The Hrms analysis of this peak allowed its identification as the expected conjugate 8 (see Figure 8 in Supporting Information). Consequently, the detection of this thio-conjugate may confirm the generation of the highly reactive quinoneimine 6 from phenol 5 oxidation. It is noteworthy that in addition to conjugate 8, a significant generation of 4-aminophenol was detected in the incubation medium, which confirms the high lability of quinoneimine 6, even in the presence of a trapping agent such as glutathione.

Discussion Taken together, the results reported herein show that PAA, one of the contaminants present in the L-tryptophan batches responsible for EMS, can be bioactivated by HLM to yield the phenol derivative 5 as the major metabolite. This oxidation metabolite is unstable and undergoes further oxidation to give quinoneimine 6. Although this quinoneimine is too reactive to be characterized, its involvement in the bioactivation of PAA was inferred through the unambiguous identification of 4-aminophenol and formylglycine in the incubation medium. The formation of these compounds was rationalized through a sequence (Figure 6, right) analogous to that demonstrated previously in our laboratory for the bioactivation of PAP under the same conditions (Figure 6, left). PAP, a toxic biomarker of the oil batches that caused TOS, undergoes bioactivation by HLM to generate phenol derivative 1 that is further oxidized to a quinoneimine intermediate. In contrast to compound 6, this intermediate could be detected by chromatographic and spectral means, although it was highly reactive with nucleophiles that are potentially present in biological media (15). In the absence of nucleophiles, the quinoneimine is in equilibrium with the corresponding imine. Finally, this equilibrium is shifted by hydrolysis of the imine to yield 4-aminophenol and glyceraldehyde as final metabolites. In addition, previous in ViVo studies had shown that intraperitoneal administration of PAP to two mouse strains produced the 4′-hydroxy derivative 1 and paracetamol, among other metabolites (20, 21). Thus, as occurred with PAP, the bioactivation of PAA by HLM involves the generation of a highly reactive quinoneimine intermediates. Concerning the potential links between EMS and TOS, Mayeno and co-workers reported that the bioactivation of PAP

Figure 6. New link proposed between EMS and TOS, whereby the bioactivation of similar toxic biomarkers (PAA and PAP, respectively) by HLM yields 4-aminophenol and the corresponding aldehyde derivatives (formylglycine and glyceraldehyde, respectively).

by rat hepatocytes and human liver tissue afforded PAA, among other metabolites. The authors suggested a bioactivation pathway comprising three oxidation steps followed by transamination (12). The results presented herein propose yet another link between EMS and TOS. As illustrated in Figure 6, the compounds implicated in each intoxication episode (i.e., PAA, in the case of EMS, and PAP, in the case of TOS) are converted by HLM into metabolites bearing a hydroxy group para to the amine (compounds 5 and 1, respectively). Interestingly, these compounds are unstable and readily undergo oxidation, although not necessarily by enzymatic action, to generate quinoneimine intermediates. It is generally accepted that these compounds are highly reactive, particularly towards nucleophiles (22). Our group previously demonstrated this behavior in PAP-derived quinoneimine (15). It should be noted that in their EMS study of the effects of PAA on binding to rat hepatic and nuclear envelopes, Sidransky et al. proposed that PAA or a metabolite may become incorporated into proteins, possibly replacing L-tryptophan or L-alanine residues (23). We suggest that the possibility of the covalent binding of quinoneimine intermediate 6 to nucleophilic sites of proteins might be also contemplated. In this context, we have detected the spontaneous formation of the corresponding glutathione conjugate 8. However, the low stability of quinoneimine 6 is also manifested as much in the presence as in the absence of nucleophiles via isomerization and subsequent hydrolysis of the imine intermediate to give formylglycine and 4-aminophenol. Moreover, the identification of paracetamol as a metabolite in two murine species that had been treated with 14C-labeled PAP may stimulate researchers to seek this compound or even 4-aminophenol in biological samples related to EMS. If either of said compounds were to be found in such samples, then EMS and TOS would be linked not only by the generation of a final common toxic compound (4-aminophenol) (24–28) but also by the concomitant release of carbonyl species that could represent an additional risk factor.

1562 Chem. Res. Toxicol., Vol. 20, No. 10, 2007

Martínez-Cabot and Messeguer

Finally, the similarities found in the bioactivations of PAA and PAP provide additional support to the existence of a common pathway for all aniline xenobiotics that bear hydrogen atoms at the carbon atom contiguous to the amino moiety.

(12)

Acknowledgment. Financial support from the WHO TOS Committee is gratefully acknowledged.

(13)

Supporting Information Available: GC-MS profile using the SIM mode of a sample corresponding to the in Vitro incubation of PAA and subsequent silylation with BSTFA compared with the GC-MS profile of a standard silylated 4-aminophenol (Figure 7). Hrms spectra of the new peak formed in phenol 5 incubations with glutathione and its corresponding UV spectra compared with the UV spectra of the glutathione conjugate of the quinoeimine derived from phenol 1 (Figure 8). This information is available free of charge via the Internet at http://pubs.acs.org.

(14)

(15)

(16) (17)

References (1) Varga, J., Jimenez, S. A., and Uitto, J. (1993) L-Tryptophan and the eosinophilia-myalgia syndrome: current understanding of the etiology and pathogenesis. J. InVest. Dermatol. 100, 457–465. (2) Silver, R. M., Ludwika, A., Hampton, M., Ohba, T., Bingel, S. A., Smith, T., Harley, R. A., Maize, J., and Heyes, M. P. (1994) A murine model of the eosinophilia-myalgia syndrome induced by 1,1′-ethylidenebis (L-tryptophan). J. Clin. InVest. 93, 1473–1480. (3) Hertzman, P. A., Falk, H., Kilbourne, E. M., Page, S., and Shulman, L. E. (1991) The eosinophilia-myalgia syndrome: The Los Alamos Conference. J. Rheumatol. 18, 867–873. (4) Belongia, E. A., Hedberg, C. W., Gleich, G. J., White, K. E., Mayeno, A. N., Loegering, D. A., Dunnette, S. L., Pirie, P. L., McDonald, K. L., and Osterholm, M. T. (1990) An investigation of the cause of the eosinophilia-myalgia syndrome associated with tryptophan use. N. Engl. J. Med. 323, 357–365. (5) Goda, Y., Suzuki, J., Maitani, T., Yoshihira, K., Takeda, M., and Uchiyama, M. (1992) 3-Anilino-L-alanine, structural determination of UV-5, a contaminant in EMS-associated L-tryptophan samples. Chem. Pharm. Bull. 40, 2236–2238. (6) Hill, R. H., Caudill, S. P., Philen, R. M., Bailey, S. L., Flanders, W. D., Driskell, W. J., Kamb, M. L., Needham, L. L., and Sampson, E. J. (1993) Contaminants in L-tryptophan associated with eosinophiliamyalgia syndrome. Arch. EnViron. Contam. Toxicol. 25, 134–142. (7) Williamson, B. L., Benson, L. M., Tomlinson, A. J., Mayeno, A. N., Gleich, G. J., and Naylor, S. (1997) On-line HPLC-tandem mass spectrometry analysis of contaminants of L-tryptophan associated with the onset of the eosinophilia myalgia syndrome. Toxicol. Lett. 92, 139– 148. (8) Adachi, J., Mio, T., Ueno, Y., Naito, T., Nishimura, A., Fujiwara, S., Sumino, K., and Tatsuno, Y. (1994) Identification of four metabolites of 3-(phenylamino)alanine, a constituent in L-tryptophan products implicated in eosinophilia-myalgia syndrome, in rats. Arch. Toxicol. 68, 500–505. (9) Posada de la Paz, M., Philen, R. M., and Abaitua, I. (2001) Toxic oil syndrome: The perspective after twenty years. Epidemiol. ReV. 23, 231–247. (10) World Health Organization. (2004) Toxic Oil Syndrome. Ten Years of Progress, Regional Office for Europe. (11) Mayeno, A. N., Belongia, E. A., Lin, F., Lund, S. K., and Gleich, G. J. (1992) 3-(Phenylamino)alanine, a novel aniline-derived aminoacid

(18) (19)

(20)

(21)

(22) (23) (24) (25) (26) (27) (28)

associated with eosinophilia-myalgia syndrome: a link to toxic oil syndrome. Mayo Clin. Proc. 67, 1134–1139. Mayeno, A. N., Benson, L. M., Naylor, S., Colberg-Beers, M., Puchalski, J. T., and Gleich, G. J. (1995) Biotransformation of 3-(phenylamino)-1,2-propanediol to 3-(phenylamino)alanine: a chemical link between toxic oil syndrome and eosinophilia-myalgia syndrome. Chem. Res. Toxicol. 8, 911–916. Ferrer, M., Galceran, M., Sanchez-Baeza, F., Casas, J., and Messeguer, A. (1993) Synthesis of aniline derivatives with potential toxicological implications to the Spanish toxic oil syndrome. Liebigs Ann. Chem., 5, 507–511. Santos de Oliveira, F., Leite, B. C. O., Santana de Andrade, M. V. A., and Korn, M. (2005) Determination of total aldehydes in fuel ethanol by MBTH method-sequential injection analysis. J. Braz. Chem. Soc. 16, 87–92. Martínez-Cabot, A., Morató, A., and Messeguer, A. (2005) Synthesis and stability studies of the glutathione and N-acetylcysteine adducts of an iminoquinone reactive intermediate generated in the biotransformation of 3-(N-phenylamino)propane-1,2-diol: implications for toxic oil syndrome. Chem. Res. Toxicol. 18, 1721–1728. Sai Hang Ho, S., and Zhen Yu, J. (2004) Determination of airborne carbonyls: comparison of a thermal desorption/GC method with the standard DNPH/HPLC method. EnViron. Sci. Technol. 38, 862–870. Pappo, D., Vartanian, M., Lang, S., and Kashman, Y. (2005) Synthesis of cyclic endiamino peptides. J. Am. Chem. Soc. 127, 7682–7683. Nakazawa, T., Suzuki, T., and Ishii, M. (1997) Synthesis and characterization of β-O-tosyldehydroserine as a precursor of dehydroamino acids. Tetrahedron Lett. 38, 8951–8954. Cheung, K.-M., and Shoolingin-Jordan, P. M. (1997) Synthesis of 3-amino-3-vinyl propanoic acid and its conversion to 4-amino-5hydroxy-4,5-dihydrofuran-2-one hydrochloryde (HAD), a cyclic stabilised form of aspartate 1-semialdehyde hydrochloryde. Tetrahedron 53, 15807–15812. Ladona, M. G., Bujons, J., Messeguer, A., Ampurdanés, C., Morató, A., and Corbella, J. (1999) Biotransformation and clearance of 3-(phenylamino)propane-1,2-diol, a compound present in samples related in samples related to toxic oil syndrome, in C57BL/6 and A/J mice. Chem. Res. Toxicol. 12, 1127–1137. Bujons, J., Ladona, M. G., Messeguer, A., Morató, A., and Ampurdanés, C. 2001 Metabolism of (R)- and (S)-3(phenylamino)propane-1,2-diol in C57BL/6- and A/J-strain mice. Chem. Res. Toxicol. 14, 1097–1106. Identification of new metabolites with toxicological significance to the toxic oil syndrome. Monks, T. J., and Jones, D. C. (2002) The metabolism and toxicity of quinones, quinonimines, quinone methides, and quinone-thioethers. Curr. Drug. Metab. 3, 425–438. Sidransky, H., Verney, E., Cosgrove, J. W., and Latham, P. S. (1994) Effect of 3-phenylamino-L-alanine on tryptophan binding to rat hepatic nuclear envelopes. Toxicology 86, 135–145. Klos, C., Koob, M., Kramer, C., and Dekant, W. (1992) p-aminophenol nephrotoxicity: biosynthesis of toxic glutathione conjugates. Toxicol. Appl. Pharmacol. 115, 98–106. Fowler, L. M., Foster, J. R., and Lock, E. A. (1994) Nephrotoxicity of 4-amino-3-S-glutathionylphenol and its modulation by metabolism or transport inhibitors. Arch. Toxicol. 68, 15–23. Lock, E. A., Cross, T. J., and Schnellmann, R. G. (1993) Studies on the mechanism of 4-aminophenol-induced toxicity to renal proximal tubules. Hum. Exp. Toxicol. 12, 383–388. Fu, X., Chen, T. S., Ray, M. B., Nagasawa, H. T., and Williams, W. M. (2004) p-Aminophenol-induced hepatotoxicity in hamsters: role of glutathione. J. Biochem. Mol. Toxicol. 18, 154–161. Yoshida, R., Oikawa, S., Ogawa, Y., Miyakoshi, Y., Ooida, M, Asanuma, K., and Shimizu, H. (1998) Mutagenicity of p-aminophenol in E. coli WP2uvrA/pKM101 and its relevance to oxidative DNA damage. Mutat. Res. 415, 139–150.

TX700256V