Mechanism for the Anti-Thyroid Action of Minocycline - American

MN with thyroid peroxidase (TPO), the key enzyme in thyroid hormone synthesis. In the present study, the mechanisms for inhibition of TPO-catalyzed io...
0 downloads 0 Views 319KB Size
Download PDF
Chem. Res. Toxicol. 1997, 10, 49-58

49

Mechanism for the Anti-Thyroid Action of Minocycline Daniel R. Doerge,*,† Rao L. Divi,† Joanna Deck,† and Alvin Taurog‡ Food and Drug Administration, National Center for Toxicological Research, Jefferson, Arkansas 72079-9502, and Department of Pharmacology, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75235-9041 Received August 19, 1996X

Administration of minocycline (MN), a tetracycline antibiotic, produces a black pigment in the thyroids of humans and several species of experimental animals and antithyroid effects in rodents. We have previously shown that these effects appear to be related to interactions of MN with thyroid peroxidase (TPO), the key enzyme in thyroid hormone synthesis. In the present study, the mechanisms for inhibition of TPO-catalyzed iodination and coupling reactions by MN were investigated. MN was stable in the presence of TPO and H2O2, but adding iodide or a phenolic cosubstrate caused rapid conversion to several products. TPO-dependent product formation, characterized by on-line LC-APCI/MS and 1H-NMR, involved oxidative elimination to form the corresponding benzoquinone with subsequent dehydrogenation at the aliphatic 4-(dimethylamino) group. Addition of thiol-containing polymers (bovine serum albumin or thiol-agarose chromatographic beads) had a minimal effect on MN oxidation by TPO, but substantially reduced product formation and produced concomitant losses in free thiols. Covalent bonding through a thioether linkage of a reactive intermediate, the benzoquinone iminium ion, was inferred from these findings. Iodide- and phenolic cosubstrate-dependent oxidation of tetracycline to demethylated and dehydrogenated products was also observed, although at a slower rate than MN. The products and kinetics observed with MN were consistent with oxidation of MN by either the enzymatic iodinating species formed by reaction of TPO compound I with iodide or phenoxyl radicals/cations generated by TPO-mediated oxidation of a phenolic cosubstrate. The proposed reaction mechanism is consistent with alternate substrate inhibition of TPO-catalyzed iodination of tyrosyl residues in thyroglobulin (Tg) by MN, as previously reported. Furthermore, the observed phenoxyl radical-mediated oxidation of MN is consistent with its previously reported potent inhibition of the coupling of hormonogenic iodotyrosine residues in Tg in the reaction that forms thyroid hormones. The proposed reaction mechanism also implicates a reactive benzoquinone iminium ion intermediate that could be important in toxicity of MN.

Introduction A substantial body of evidence indicates the potential for disruption of thyroid function by minocycline (MN)1 (see Figure 1 for structure), a tetracycline antibiotic used to treat acne (1), infectious diseases (2), and rheumatoid arthritis (3). MN administration and uptake into the thyroid (4) results in deposition of a black pigment in the thyroid of several species of experimental animals (5) and humans (6). Anti-thyroid effects observed in rodents include goiter and inhibition of thyroid hormone synthesis (7). MN use has also been associated with induction of thyroid cancer in humans (8-11). In a previous study (12), we described a causative role for thyroid peroxidase (TPO), the key enzyme in thyroid hormone synthesis, in formation of the MN-derived black pigment in vitro. That study also described the inhibitory effect of MN on TPO-catalyzed iodination of tyrosine moieties in thyroglobulin (Tg) and a highly potent inhibition of the subsequent coupling of hormonogenic iodotyrosine resi* Corresponding author. Tel: (501) 543-7943; FAX: (501) 543-7720; E-Mail: [email protected]. † National Center for Toxicological Research. ‡ University of Texas Southwestern Medical Center. X Abstract published in Advance ACS Abstracts, December 15, 1996. 1 Abbreviations: APCI, atmospheric pressure chemical ionization; BSA, bovine serum albumin; DIT, 3,5-diiodotyrosiine; DTT, dithiothreitol; HRP, horseradish peroxidase; LPO, lactoperoxidase; MES, morpholinoethanesulfonic acid; MIT, 3-iodotyrosine; MMI, methimazole; MN, minocycline; Tg, thyroglobulin; TPO, thyroid peroxidase.

S0893-228x(96)00150-6 CCC: $14.00

Figure 1. Structures for minocycline and tetracycline.

dues to form Tg-bound thyroid hormones (12). In the present report, we propose mechanisms by which MN blocks these TPO-mediated reactions that are essential to formation of thyroid hormones.

Experimental Procedures Reagents. Bovine serum albumin (BSA), dithiothreitol (DTT), 3,5-diiodotyrosine (DIT), glucose, glucose oxidase, guaiacol, hydrogen peroxide, horseradish peroxidase (HRP), lactoperoxidase (LPO), methimazole (MMI), metallothionein, 3-iodotyrosine (MIT), minocycline (MN), morpholinoethanesulfonic acid (MES), poly(ethylene glycol) (PEG), potassium iodide, tetracycline, and tyrosine were purchased from Sigma Chemical Co. (St. Louis, MO). Bio-Gel P-6DG was obtained from Bio-Rad Laboratories (Hercules, CA), and thionitrobenzoic acid-modified (TNB-thiol) agarose was obtained from Pierce (Rockford, IL). Porcine thyroid peroxidase (TPO) used in the present study was purified and quantified spectrophotometrically as described earlier (13). Iodine was obtained from Aldrich Chemical Co. Solutions of triiodide (1.0 mM) were prepared just before use by mixing iodine (1.0 mM) and excess iodide ion and diluting as required.

© 1997 American Chemical Society

50

Chem. Res. Toxicol., Vol. 10, No. 1, 1997

Inhibition of Tyrosine Iodination by MN. Tyrosine (125 µM), iodide (250 µM), and TPO (10 nM) were incubated with and without MN (25 µM) at pH 6.5 in the presence of H2O2 (250 µM) or 2.5 mM glucose/glucose oxidase (9.6 mU/mL as determined from HRP-catalyzed scopoletin oxidation assay of H2O2 production, see below) for 5 min at 37 ( 0.1 °C. An aliquot was withdrawn from the reaction mixture, and iodotyrosines were separated by HPLC using NovaPak C18 cartridge (5 × 100 mm, 4-µm particle size, Waters Associates, Milford, MA) with the following linear gradient: A ) acetonitrile, B ) acetonitrile/ trifluroacetic acid/water (5:0.2:94.8 v/v); 5% A in B (v/v) to 45% A in B (v/v) in 30 min. Peaks for 3-iodotyrosine (MIT) and 3,5diiodotyrosine (DIT) were monitored by UV absorbance at 230 nm. To eliminate the possibility that MN inhibited glucose oxidase-catalyzed H2O2 generation, the following experiment was carried out. Glucose oxidase (9.6 mU/mL), glucose (2.5 mM), and MN (0-128 µM) were incubated at 37 ( 0.1 °C for 5 min in 0.1 M MES buffer (pH 6.5). An aliquot of the reaction mixture was passed through C18 SepPak cartridge to remove MN and related compounds. The amount of H2O2 in the effluent was measured spectrofluorometrically (Ex 360 nm, Em 460 nm) using scopoletin (2.0 µM) oxidation by HRP (20 nM) at 25 ( 0.1 °C in 0.1 M MES buffer, pH 6.5 (14). No effect of MN on H2O2 production was observed. Effect of Cosubstrates on TPO-Mediated MN Oxidation. MN (25 µM) was incubated with iodide (100 µM), guaiacol (100 µM), tyrosine (100 µM), MIT (100 µM), or DIT (100 µM) in the presence TPO (10 nM) for 3 min in 0.1 M MES buffer (pH 6.5) at 25 ( 0.1 °C. The reaction was started by adding H2O2 (100 µM) to the reaction mixture in the flow cell of an SFA-11 stopped-flow apparatus (HiTech Instruments, Salisbury, U.K.) fitted with an external constant temperature circulating bath ((0.1 °C). The disappearance of MN was followed spectrophotometrically at 340 nm using a Hewlett Packard 8452 diode array spectrophotometer equipped with a temperature-controlled cell compartment. MN disappearance at varying concentrations of iodide (0-500 µM) was determined in a similar manner. MN and its reaction products were separated by HPLC using a NovaPak C18 cartridge with a linear gradient of A: acetonitrile, B: 0.01 M oxalic acid; 10% A in B (v/v) to 40% A in B (v/v) in 30 min. The peaks were detected by UV absorbance at 265 nm. Hydrogen Peroxide Concentration Dependence of MN Disappearance. Reaction mixtures containing MN (50 µM), TPO (20 nM), and iodide (100 µM) in 0.05 M MES buffer (pH 7.0) were incubated at 37 ( 0.1 °C for 3 min, and the reaction was started by addition of varying concentrations of H2O2 (0, 12.5, 25, 37.5, 50 µM). The reactions were complete after 3 min of reaction, and MN remaining in the reaction mixture was determined spectrophotometrically at 340 nm using second derivative spectrophotometry. Measurement of Simultaneous Oxidation of Guaiacol and MN Disappearance. Reaction mixtures containing guaiacol (2.5 mM), MN (25 µM), and TPO (10 nM) were incubated in 0.1 M MES buffer (pH 6.5) at 37 ( 0.1 °C for 3 min. The reaction was initiated by addition of H2O2 (100 µM) using the stopped-flow apparatus. Guaiacol oxidation was monitored spectrophotometrically at 470 nm (15) and MN at 340 nm simultaneously by repetitive scanning at 10 s intervals. Measurement of Dityrosine Formation. Reaction mixtures consisting of tyrosine (350 µM), HRP (20 nM), and glucose (6.8 mM) were incubated with varying concentrations of MN at 22 ( 0.1 °C in 0.05 M MES buffer (pH 7.0). Reaction was initiated by addition of glucose oxidase (50 µM). Dityrosine, the reaction product of tyrosine oxidation, was monitored spectrofluorometrically (16) using a Perkin Elmer LS 50 B luminescence spectrometer (excitation wavelength 315 nm, emission 408 nm). Time Course for MN Product Formation. TPO (20 nM) was incubated with MN (50 µM) in the presence of iodide (100 µM) at 22 ( 0.1 °C in 0.05 M MES (pH 7.0) for 3 min. The reaction was started by addition of H2O2 (100 µM). After

Doerge et al. reaction times of 30 s, 1.0 min, 2.0 min, 3.0 min, and 8.0 min in separate incubations, HPLC analysis of products was carried out as described above. Alternatively, the MN (50 µM) was incubated with: (a) tyrosine (100 µM), TPO (20 nM), and H2O2 (100 µM); (b) triiodide (100 µM); (c) HRP (50 nM) in the presence of H2O2 (100 µM) without added iodide. MMI (0.5 mM final concentration) was added to incubations to terminate TPOmediated reactions. TPO-Catalyzed Tetracycline Oxidation. Tetracycline (50 µM), TPO (20 nM), iodide (100 µM), and H2O2 or glucose (5.0 mM)/glucose oxidase (10 mU) were incubated for 30 min, and the products were separated by HPLC essentially as described for MN products. Modification of Thiol Groups during TPO-Mediated MN Oxidation. Incubation samples contained metallothionein (1 mg/mL) or BSA (16-32 µM), TPO (20 nM), MN (50 µM), iodide (100 µM), and H2O2 (100-200 µM) in 0.05 M MES buffer (pH 7.0) at 22 ( 0.1 °C. After 3 min of incubation, MN-derived products were separated by HPLC. For BSA incubations, an aliquot of the reaction mixture was passed through a Bio-Gel P6-DG chromatocentrifugation column. The protein-containing eluate was treated with DTT (100 µM) to reduce any disulfide groups present and again passed through a fresh chromatocentrifugation column to separate the protein from unreacted DTT. Thiol content of the protein was measured spectrophotometrically at 340 nm using Ellman’s reagent with reduced glutathione as a reference. Alternatively, thionitrobenzoic acid (TNB) thiolmodified agarose chromatographic beads containing 20 nmol of thiol/mg of beads were activated to the thiol form by treatment with DTT (10 mM). Following extensive washing with buffer, the beads were incubated with gentle shaking (20-40 µM free thiol groups per incubation) with MN (50 µM), TPO (20 nM), and iodide (100 µM), with and without H2O2 (250 µM), in buffer at 22 ( 0.1 °C for 5 min. The beads were washed thoroughly with buffer and the thiol content was measured. Beads from a similar incubation were further treated with DTT (100 µM) and thoroughly washed with buffer, and the thiol content was determined. Mass Spectrometry. MN was incubated with TPO, iodide, and H2O2 as described above, and at the end of the reaction interval, a 20 µL aliquot was injected directly into the NovaPak HPLC column for on-line LC/MS analysis. MS experiments were performed using a VG Platform single quadrupole mass spectrometer (Micromass, Altrincham, U.K.) equipped with an atmospheric pressure chemical ionization (APCI) interface. The total LC column effluent (1.0 mL/min) was delivered into the atmospheric pressure ion source (150 °C) through a heated nebulizer probe (500 °C) using nitrogen as the probe and bath gas (275 L/h). At a low sampling cone-skimmer voltage (15 V), mass spectra for MN and tetracycline consisted predominately of the respective protonated molecule; however, at higher voltages, diagnostic fragment ions were observed, and such a value (20 V) was used throughout these studies. Positive ions were acquired in full scan mode (m/z 100-750, 2.1 s cycle time) in series with a UV detector set at 275 nm. Backgroundsubtracted mass spectra were obtained by averaging spectra across the respective chromatographic peak and subtracting average background spectra immediately before and after this peak. The mass spectrometer was calibrated over the range m/z 85-1200 using a solution of poly(ethylene glycols) [PEG 200 (25 µg/mL), 300 (50 µg/mL), 600 (75 µg/mL), 1000 (250 µg/ mL)] in 50% acetonitrile in aqueous ammonium acetate (5 mM). A switching valve (Rheodyne Model 7030, Cotati, CA) was used to divert chromatographically unretained compounds to waste during the initial 3.5 min of each run. During this interval, a constant flow of the same mobile phase was delivered to the APCI probe by a Waters M6000 pump. NMR Characterization of MN P2. MN was incubated with TPO, iodide, and H2O2 as described above with the reaction volume scaled up to 250 mL. The reaction mixture was extracted 3 times with a 250 mL portion of ethyl acetate. The organic layers were then combined and reduced in volume in vacuo. The residue was dissolved in 2 mL of 10% aqueous

Anti-Thyroid Mechanism for Minocycline

Chem. Res. Toxicol., Vol. 10, No. 1, 1997 51

Table 1. Disappearance of MN in the Presence of Enzymatic and Nonenzymatic Oxidative Reagentsa reaction conditions

% minocycline consumed

TPO + H2O2 + MN TPO + I- + H2O2 + MN HRP + MN HRP + H2O2 + MN LPO + H2O2 + MN LPO + I- + H2O2 + MN I3- + MN

2 ( 0.9 99 ( 5 1 ( 0.9 55 ( 6 1 ( 0.8 100 ( 3 57 ( 4

a All reactions were carried out at 25 ( 0.1 °C for 3 min in 0.1 M MES buffer (pH 6.5) containing MN (25 µM), and MN absorbance at 340 nm was monitored. The concentrations of reagents used were: TPO (10 nM), LPO (5 nM), H2O2 (100 µM), HRP (100 nM), and triiodide (100 µM). Values shown are the mean ( standard deviation of UV absorbances from at least 3 determinations within the linear response range for MN.

Figure 3. Simultaneous oxidation of guaiacol and MN catalyzed by TPO/H2O2. MN (25 µM) was incubated with guaiacol (2.5 mM) and TPO (10 nM), and the reaction was initiated by addition of H2O2 (100 µM). Spectrophotometric measurements were made at various times to determine the relative disappearance of MN (340 nm absorbance, squares) and the oxidation of guaiacol (470 nm absorbance, circles). The amount of MN remaining is expressed as the percent (%) of the original concentration, and guaiacol-derived product formation is expressed as the percent (%) of maximal. Oxidation of guaiacol in the absence of MN produced maximal absorbance (ca. 30% higher than that shown with MN) after ca. 15 s (data not shown).

Figure 2. Cosubstrate-dependent oxidation of MN. MN (25 µM) was incubated with TPO (10 nM) in the presence of various effectors (100 µM) in MES buffer (0.1 M, pH 6.5). The reaction was initiated by addition of H2O2 (100 µM), and the disappearance of MN was monitored spectrophotometrically at 340 nm. (A) Control containing MN, TPO, and H2O2; (B) +DIT; (C) +tyrosine; (D) +guaiacol; (E) +KI. acetonitrile, and 20 µL aliquots were injected into the HPLC column. The fraction corresponding to P2 was collected as it eluted from the column. Several aliquots of D2O were added during evaporation of solvent in vacuo to minimize the presence of exchangeable protons. MN was dissolved in methanol-d4 or acetonitrile-d3, and P2 was dissolved in acetonitrile-d3 for 1HNMR measurements using a Bruker AM500 spectrometer operating at 500.13 MHz.

Results Iodide-Dependent Oxidation of MN by TPO. We previously reported the formation of a black pigment upon incubation of high concentrations of MN (547 µM) with TPO and a H2O2-generating system (glucose + glucose oxidase) for an extended time period (90 min, see ref 12). In the current study, reaction of MN with TPO and other oxidative reagents was studied at much lower concentrations of MN and enzyme, and at much shorter incubation times. No black color formation was observed under these conditions. MN was stable in the presence of TPO/H2O2 for periods up to 10 min at concentrations of TPO up to 50 nM; however, in the presence of iodide, MN was rapidly oxidized (i.e., 80% loss of MN absorbance after ca. 20 s under the conditions described in Figure 2 and Table 1). The TPO-mediated oxidation of MN was greatly enhanced by increasing iodide concentrations, and half-maximal MN oxidation at 1 min was observed with

ca. 2 µM KI (data not shown). LPO-catalyzed destruction of MN showed a similar absolute requirement for iodide, but HRP required only H2O2. Addition of iodide to HRP incubations had minimal effects on the reaction with MN. MN reacted with triiodide, a potential product from incubations of TPO and H2O2 with iodide; however, the rate was much slower than that observed with TPO or LPO and H2O2 in the presence of iodide. The concentration of triiodide (100 µM) used was the theoretical maximum possible from the amount of H2O2 used throughout (see Table 1). Inhibition of TPO-Catalyzed Tyrosine Iodination by MN. MN inhibited TPO-catalyzed iodination of tyrosine, a model iodination substrate, in a manner similar to that previously observed for TPO-mediated iodination of Tg (12). Addition of 25 µM MN to an incubation mixture containing TPO/H2O2, iodide, and tyrosine decreased the formation of MIT and DIT, measured using HPLC after a 5 min incubation, to 29% and 7%, respectively, of control values in the absence of MN. TPO-Mediated Oxidation of MN in the Presence of Phenolic Cosubstrates. A. Guaiacol. Incubation of TPO/H2O2 with guaiacol led to formation of the colored coupling product, 3,3′-dimethoxybiphenylquinone (15, 16). When MN was included in the incubation, the rate of formation of this product was substantially diminished and concomitant oxidation of MN was observed (see Figure 3). The loss of MN (25 µM initial concentration) was dependent on the concentration of guaiacol, and halfmaximal inhibition was seen at ca. 10 µM guaiacol. Guaiacol oxidation catalyzed by TPO and H2O2 was inhibited competitively by MN (intersecting 1/rate vs 1/[guaiacol] plots with a linear slope replot, data not shown), and MN destruction occurred simultaneously (see Figure 3). No lag phase was observed. B. Tyrosine. Tyrosine oxidation to dityrosine (17), catalyzed by either TPO or HRP, was much slower than guaiacol oxidation. It was necessary to use a hydrogen

52

Chem. Res. Toxicol., Vol. 10, No. 1, 1997

Doerge et al.

Figure 4. Inhibition of HRP-catalyzed dityrosine formation by MN. HRP (50 nM) was incubated with various concentrations of MN (A, 0; B, 5 µM; C, 15 µM; D, 25 µM; E, 50 µM), tyrosine (350 µM), and glucose (6.8 mM). The reactions were initiated by addition of glucose oxidase (50 mU/mL). At various times, aliquots were removed and analyzed fluorometrically for dityrosine formation (Ex 315 nm, Em 408 nm).

peroxide-generating system (glucose + glucose oxidase) to determine the apparent alternate substrate behavior discussed below. Similar dityrosine formation kinetics were observed with TPO and HRP, and in order to minimize consumption of TPO, HRP was used for the complete kinetic study shown in Figure 4. Inhibition of dityrosine formation was characterized by a lag phase, followed by product formation at rates somewhat lower than the control rate. The length of the lag phase increased as the concentration of MN was increased. C. MIT and DIT. Incubation of TPO/H2O2 with DIT (100 µM) led to a slow decrease in MN concentration (5 min), a third product (P3) was also observed. The H2O2 stoichiometry for TPO-mediated loss of MN in the presence of iodide was determined by adding limiting amounts of H2O2 to separate incubation mixtures. Under conditions where P1 was the predominant product, 1 mol of MN was lost for each mole of H2O2 added (slope ) 0.95 mol of MN lost per mole of H2O2 added, correlation coefficient ) 0.99, data not shown). Similar products were observed when triiodide or HRP/ H2O2 was used as the oxidant, albeit with different kinetics and relative yields. After 1 min of incubation,

Figure 5. HPLC analysis of MN-derived oxidation products. MN was incubated in the presence of iodide, TPO, and H2O2 as described in Figure 2, and at various times, 10 µL aliquots were injected onto the HPLC column and products monitored using UV detection (260 nm). (a) TPO/H2O2 + MN, 10 min; (b) TPO/ H2O2/I- + MN, 30 s; (c) as in (b), 2 min; (d) as in (b), 8 min; (e) as in (d) after addition of 250 µM H2O2.

100 µM triiodide produced a mixture of P1, P2, and P3 in approximately equal amounts, with ca. 50% conversion of MN. HRP (100 nM) plus H2O2 also produced a mixture of P1, P2, and P3 with conversion of ca. 50% of the MN after 1 min. Complete loss of MN and the exclusive formation of P2 were observed after 1 min incubation of MN with TPO plus H2O2 in the presence of tyrosine (100 µM). To determine whether the conversion of P1 to the other products was enzyme-mediated, MN was incubated with TPO/H2O2 in the presence of iodide for 30 s, conditions where essentially complete conversion to P1 occurred (see Figure 5). At this time, a high concentration of MMI was added to the mixture to inhibit all TPO-mediated reactions. This treatment resulted in a greatly slowed conversion of P1 to P2, although at long incubation times P2 and P3 did form in similar amounts compared to the normal incubation conditions (data not shown). Further evidence for a spontaneous, nonenzymatic conversion of P1 to P2 and P3 was observed following collection of P1 from the LC column effluent. It was also observed that addition of H2O2 to solutions containing P3 converted it to P2. Product Structure Characterization. MN and the resultant oxidation products were characterized using online LC-APCI/MS (Figure 6). Molecular species (MH+ or M+ for P2) were observed for MN, P1, P2, and P3 at m/z 458, 429, 427, and 431, respectively, as was the respective fragment ion due to loss of 17 amu (-NH3). Essentially identical mass spectra were observed for MN and P2 using off-line electrospray ionization/MS. The presence of m/z 431 (ca. 40% of base peak intensity) in the mass spectrum of P1 probably arises from in-source reduction to P3, a common reaction we have observed for quinones

Anti-Thyroid Mechanism for Minocycline

Chem. Res. Toxicol., Vol. 10, No. 1, 1997 53

Figure 6. Mass spectra for MN and its oxidation products. Background-subtracted mass spectra were obtained from on-line LCAPCI/MS analysis of reaction mixtures as described in the Experimental Procedures.

during APCI analysis.2 No evidence for iodination of MN was seen in the mass chromatograms obtained from online LC/MS acquisitions (data not shown). In order to obtain structural information to complement the LC/MS results, P2 was collected from the LC effluent, concentrated, and analyzed by 1H-NMR. All resonances in MN were assigned using nuclear Overhauser effect (NOE) and decoupling experiments and were in accord with the previous assignments (18), with the exception of the aromatic protons attached to C8 and C9 (see Figure 1 for numbering). Irradiation of the doublet at 7.44 ppm produced an NOE of the singlet at 2.62 ppm, the N-dimethyl resonance. For this reason, it appears that the previous assignments for H8 and H9 are reversed. Nilges et al. demonstrated formation of the semiquinone radical corresponding to P1 upon mixing MN and ferricyanide ion under stopped-flow conditions (18). Subsequent addition of ascorbate produced a compound with 1H-NMR and FAB-mass spectra consistent with the structure for P3 in the present study. As described by Nilges et al. for the hydroquinone, P3, the proton resonance corresponding to the aromatic Ndimethyl group was also lost in P2 (data not shown). Moreover, the aromatic protons in MN showed a typical AB pattern of two doublets with chemical shifts of 7.44 (C8) and 6.81 ppm (C9). Upon oxidation to P2, this changed to a pseudo-quartet with a chemical shift of 7.01 ppm. This demonstrates a change from magnetically nonequivalent protons in MN to essentially equivalent ones in P2 as predicted for changing a p-(dimethylamino)phenol to a p-quinone. Because of a broad interfering signal, it was not possible to get NOE or decoupling results of sufficient quality to unambiguously assign all resonances in P2, particularly the aliphatic methine proton attached to C4 that would be lost during oxidation to the proposed 2

D. R. Doerge and M. I. Churchwell, unpublished observations.

conjugated iminium ion. Therefore, the structural assignment of the conjugated iminium ion in P2, based solely on the mass spectrum and predicted chemical reactivity, is tentative. Trapping Reactive Species Derived from MN Using Thiol-Containing Polymers. In order to detect the presence of reactive species derived from MN oxidation by TPO in the presence of H2O2 and iodide, various thiol-containing reagents were added to incubation mixtures. Rates of MN disappearance and P1 formation were monitored using HPLC. It was not possible to use low molecular weight thiols (e.g., DTT or cysteine) because these compounds are substrates for TPO-mediated iodination (19) and oxidation (20). Similarly, it was observed that metallothionine, a low molecular weight protein that contains many free cysteine residues, blocked all MN oxidation in the presence of iodide, presumably by competing with MN for oxidized iodine (data not shown). However, addition of BSA (see Table 2) or thiolagarose beads had only a small inhibitory effect on MN disappearance. Moreover, the presence of BSA caused substantial inhibition of P1 formation as well as subsequent conversion to P2 and P3 (see Table 2). These effects were more pronounced at 32 µM BSA than at 16 µM (data not shown). Changes in thiol content of BSA were monitored under the same conditions (see Table 2). Native BSA contained 0.70 mol of free thiols/mol of protein, as previously observed (21). TPO in the presence of H2O2 or H2O2 + iodide effected extensive oxidation of free thiols, but these changes were readily reversed by treatment of the protein with DTT. These observations are consistent with the formation of disulfides. When TPO-mediated MN oxidation was induced in the presence of iodide and H2O2, thiol group content of added BSA was eliminated, and only about 27% of free thiols were restored upon treatment with DTT. This corresponds to a loss of about 15 nmol of thiol/mL.

54

Chem. Res. Toxicol., Vol. 10, No. 1, 1997

Doerge et al.

Table 2. Inhibition of TPO-Mediated Product Formation from MN and Concomitant Loss in BSA Thiol Groupsa reaction conditions

MN remaining (%)

P1 + P2 (%)

BSA thiols before DTT (nmol/nmol of BSA)

BSA thiols after DTT (nmol/nmol of BSA)

TPO/I-/BSA + MN TPO/H2O2/I-/BSA TPO/H2O2/BSA + MN TPO/H2O2/I- + MN TPO/H2O2/I-/BSA + MN

100 ND 100 2 14

0 ND 0 100 48

0.70 0.08 0.12 ND 0.00

ND 0.73 0.68 ND 0.19

a Incubations containing TPO (10 nM), iodide (100 µM), and H O (100 µM) were carried out for 3 min in the presence of 50 µM MN 2 2 and 20 µM BSA. The amounts of MN remaining (based on a 50 µM standard) and P1 + P2 formed (based on the levels formed in the presence of TPO/H2O2/I-) were determined using HPLC as described in the Experimental Procedures and are expressed as a percent of the peak areas from the respective UV 265 nm chromatogram. Free thiol groups were determined using Ellman’s reagent before and after treatment of the protein samples with DTT (100 µM). The values shown are averages of duplicate incubations that agreed closely. Thiol determinations were performed in another experiment, and similar results were obtained. ND ) not done.

Table 3. Changes in Thiol Content of Thiol-Agarose Beads during TPO-Mediated MN Oxidationa

reaction conditions

thiol groups before DTT (nmol)

thiol groups after DTT (nmol)

beads beads + H2O2 beads + TPO/H2O2 beads + TPO/I-/H2O2 beads + TPO/MN/H2O2 beads + TPO/H2O2/I- + MN

38.2 ( 4.6 38.9 ( 4.0 25.9 ( 7.3 6.4 ( 0.5 3.5 ( 1.2 7.1 ( 0.6

39.5 ( 3.7 37.8 ( 5.3 37.3 ( 1.7 30.7 ( 2.9 38.6 ( 4.6 6.9 ( 3.9

a Agarose chromatographic beads containing 20 nmol of thiols/ mg of beads were incubated with TPO (10 nM) and H2O2 (250 µM) as described in the Experimental Procedures. Free thiols were measured after extensive washing with buffer before and after treatment with DTT (100 µM) as described in the Experimental Procedures. The values shown are means ( SD for 6 determinations.

Thiol-modified agarose beads were also used as a potential trapping agent that does not contain other potentially reactive functional groups for TPO-mediated iodination that are present in BSA (i.e., tyrosyl residues). Adding thiol-agarose to incubations containing TPO/ H2O2/I- did not affect the oxidation of MN as measured spectrophotometrically (data not shown). In the absence of MN, TPO/H2O2/I- effected extensive thiol oxidation (83%) which was mostly, but not completely, reversed by subsequent treatment of the beads with DTT (see Table 3). Similarly, when thiol-agarose was incubated in the presence of MN and TPO/H2O2/I-, thiol content was substantially reduced (81%). Even after extensive washing, the beads had acquired a noticeable yellow color that turned black upon standing overnight. Moreover, treatment of the yellow beads with DTT followed by extensive washing did not restore any of the lost thiol content. Unexpectedly, the thiol content of beads was also greatly reduced by treatment with TPO/H2O2/MN in the absence of iodide. However, in this case, subsequent DTT treatment restored thiol content to control values (see Table 3). TPO-Mediated Oxidation of Tetracycline. The oxidation of tetracycline by TPO, using glucose + glucose oxidase as an H2O2-generating system, also required the presence of iodide although the reaction was much slower than that observed for MN (30 min for ca. 50% conversion as measured by the change in absorbance at 402 nm, data not shown). In the presence of iodide, TPO/H2O2-mediated oxidation of tetracycline gave two products in approximately equal amounts with retention times of 6.4 and 12.0 min. LC-APCI/MS analysis of these peaks showed molecular species consistent with dehydrotetracycline at m/z 443 and desmethyltetracycline at m/z 431, respectively, as well as the corresponding fragment ions corresponding to loss of water (see Figure 7).

Discussion The observation that iodide was required for MN oxidation is consistent with the view that the active species in TPO-containing incubations with MN is the enzymatic iodinating intermediate previously described for TPO (22) and LPO (23). This form of the enzyme, designated [TPO-OI]-, is the equivalent of ferric ironbound hypoiodite ion (22) and is formed from reaction of TPO π-cation radical compound I with iodide. Our results in Figure 1 and Table 1 show that although TPO compound I cannot oxidize MN under these conditions, the corresponding hypoiodite complex does. Furthermore, the presence of iodide in the normal thyroid gland underscores the physiological relevance of this enzymatic intermediate as a potential drug metabolizing system. HRP, in the presence of H2O2, catalyzed MN oxidation at pH 6.5 without added iodide, and addition of iodide had only a minimal effect. The observation that MN was directly oxidized by HRP compound I but not LPO or TPO compounds I suggests differences in reactivity between the respective enzymatic intermediates. The rapid isomerization of TPO and LPO compounds I, presumably from a π-cation radical to a protein radical form (24, 25), differs significantly from the stability observed for HRP compound I (26). It is possible, in the case of LPO or TPO, that oxidation of MN cannot compete with the fast isomerization to a less reactive enzymatic oxidant. A predicted consequence of the observed rapid reaction between MN and [TPO-OI]- is the inhibition of enzymemediated iodination. We previously described the inhibition of Tg iodination by MN and other tetracyclines (12). These findings were confirmed in the present study by demonstrating that MN also inhibits the iodination of free tyrosine to MIT and DIT. The observed lag phase in tyrosine iodination, followed by resumption of iodination upon destruction of the MN, is consistent with a mechanism of alternate substrate inhibition, as previously described (12). The higher reactivity of MN, relative to tyrosine (free in solution or residues in Tg), makes it the preferred substrate. The minimal loss in enzymatic activity observed during the reaction (12) is consistent with the proposed mechanism. Adding a phenolic cosubstrate to TPO, LPO, or HRP in the presence of H2O2 also led to oxidation of MN. In the absence of MN, guaiacol or tyrosine was enzymatically oxidized to the respective dimeric product (15-17). These reactions reflect the ortho-coupling of phenoxy radicals, although TPO catalyzes tyrosine oxidation by a two-electron mechanism (27). In the presence of MN, guaiacol oxidation by TPO was inhibited competitively, and destruction of MN occurred simultaneously (see Figure 3). In contrast, the kinetics of HRP-catalyzed

Anti-Thyroid Mechanism for Minocycline

Chem. Res. Toxicol., Vol. 10, No. 1, 1997 55

Figure 7. LC-APCI/MS and LC-UV chromatograms and mass spectra for tetracycline and its TPO-mediated oxidation products. The left panel shows 275 nm absorbance and total ion chromatograms (TIC) acquired from UV and MS detectors in series as described in the Experimental Procedures, and the background subtracted mass spectra for designated peaks are shown in the right panel.

Scheme 1. Proposed Mechanism for Phenolic Cosubstrate-Dependent Oxidation of MN

dityrosine formation (see Figure 4) and the concomitant destruction of MN are consistent with alternate substrate inhibition. The inhibition cannot involve competition between MN and the phenolic cosubstrates for TPO compound I, since MN is stable in the presence of TPO and H2O2 under the conditions used here. Therefore, the difference in inhibitory mechanism between guaiacol and tyrosine presumably results from the different reactivity of MN with the radical species formed upon oxidation of the respective phenolic compound by peroxidase compound I (see Scheme 1). In the case of guaiacol, the resulting radical appears to be more reactive (see Figure 2) and dimerizes at a rate comparable to the rate of reaction with MN. This leads to simultaneous loss of guaiacol and MN and competitive inhibition kinetics. On the other hand, dimerization of the tyrosine radicals is slow with respect to reaction with MN and leads to preferential destruction of MN and alternate substrate inhibition. Use of DIT and MIT as phenolic cosubstrate for TPO was much less effective in catalyzing MN oxidation than tyrosine or guaiacol (