Inhibition of 2, 5-Hexanedione-Induced Protein Crosslinking by

Jun 15, 1995 - State University of New York, and Laboratory of Human Toxicology and Molecular Epidemiology,. Wadsworth Center for Laboratories and ...
2 downloads 0 Views 2MB Size
Chem. Res. Toxicol. 1995,8, 764-771

764

Inhibition of 2,5=Hexanedione=Induced Protein Cross-Linking by Biological Thiols: Chemical Mechanisms and Toxicological Implications Mingshe Zhu,tJ David C. Spink,t9$Bin Yan," Shelton Bank," and Anthony P. DeCaprio*3tJ Departments of Environmental Health and Toxicology and Chemistry, The University at Albany, State University of New York, and Laboratory of Human Toxicology and Molecular Epidemiology, Wadsworth Center for Laboratories and Research, New York State Department of Health, Albany, New York 12201-0509 Received January 12,1995@ n-Hexane is metabolized to the y-diketone 2,5-hexanedione (2,5-HD), a derivative that covalently binds to lysine residues in neurofilament (NF) protein to yield 2,5-dimethylpyrrole adducts. Studies comparing the pyrrole-forming potential and neurotoxic potency of y-diketones have demonstrated that pyrrolylation is a n absolute requirement in the neuropathogenesis. Autoxidative cross-linking of pyrrolylated NF proteins occurs and is proposed as a second required event. In the present study, the role of nucleophilic thiols and amines in the pyrrolemediated cross-linking reaction was investigated. When pyrrolylated ribonuclease was incubated with N-acetyllysine, N-acetylcysteine, or glutathione in physiologic buffer (pH 7.4) under air, pyrrole-to-pyrrole cross-linking was inhibited only by the thiol-containing compounds. Stable thiol-pyrrole conjugates containing a bridge from the pyrrole ring a t C-3 to the sulfur atom of the thiol were characterized by thermospray LCMS and lH-NMR spectroscopy. In contrast to low-molecular-mass thiols, SDS-PAGE studies indicated that, under the same incubation conditions, free thiols present in proteins did not undergo reaction with pyrrole adducts to form cross-links. Further experiments using a low-molecular-mass pyrrole derivative indicated that glutathione may also able to suppress pyrrole dimerization without conjugate formation, possibly via inhibition of a free radical-dependent mechanism. The results suggest the following: (1)2,5-HD-induced protein cross-linking is mediated primarily by pyrrole-topyrrole bridging under physiologic conditions, and (2) glutathione and other low-molecularmass thiols may inhibit the pyrrole dimerization reaction by two distinct pathways. These findings have significant implications for the mechanism of y-diketone neuropathy.

Introduction n-Hexane is a widely used industrial solvent that causes peripheral neuropathies upon chronic exposure in humans and experimental animals (11. Metabolism studies in vitro and in vivo have revealed that n-hexane is bioactivated by cytochromes P-450 and cytoplasmic dehydrogenases to form the y-diketone, 2,5-hexanedione (2,5-HD)l (2). This ultimate neurotoxic metabolite covalently binds to lysine residues in neurofilament (NF) proteins in vitro (3)and in vivo ( 4 ) to yield 2,bdimethylpyrrole adducts. Several studies comparing the pyrrole-forming potential and neurotoxicity of y-diketone derivatives demonstrated that pyrrolylation is an abso* To whom correspondence should be addressed at the New York State Department of Health, P.O. Box 509, Albany, NY 12201-0509. + Department of Environmental Health and Toxicology, The University at Albany, SUNY. Present address: Department of Medicinal Chemistry, University of Washington, Seattle, WA 98195. Laboratory of Human Toxicology and Molecular Epidemiology, New York State Department of Health. ' 1 Department of Chemistry, The University at Albany, SUNY. Abstract published in Advance ACS Abstracts, June 15, 1995. Abbreviations: Ac-Cys, N a-acetylcysteine; Ac-Lys, N a-acetyllysine; BSA, bovine serum albumin; DMAB, p(dimethy1amino)benzaldehyde; GSH, glutathione (reducedform);GSSG, glutathione (oxidized form); 2,5-HD, 2J-hexanedione; NF, neurofilament; OA, ovalbumin; PAGE, polyacrylamide gel electrophoresis; pyrrole-OA, pyrrolylated ovalbumin; pyrrole-RN, pyrrolylated ribonuclease A; RNase, ribonuclease A; SDS, sodium dodecyl sulfate; TFA, trifluoroacetic acid.

*

@

0893-228x/95l27Q8-Q764$09.00/Q

lute requirement in the neuropathogenetic mechanism (5-7). Pyrrole formation in NF proteins has been proposed to result in physicochemical changes in the proteins sufficient to cause NF accumulation and nerve degeneration (4, 8). As with most alkylpyrroles (91, pyrrolylated protein is susceptible to attack by molecular oxygen. It was observed that autoxidation of alkylpyrrole adducts in vivo leads to the cross-linking of numerous proteins, including NF proteins (4, 10-12). Pyrrole oxidation-mediated covalent cross-linkingof NF proteins has been proposed as the second obligatory step in y-diketone neuropathy (13,14). A clear understanding of the chemistry of autoxidation of N-substituted 2,5-dimethylpyrroles is a prerequisite for the elucidation of a role for protein cross-linking in 2,5-HD neurotoxicity. Our previous studies showed that 2,5-dimethyl-N-alkylpyrroles are chemically reactive in aqueous solutions at physiologic temperature and undergo spontaneous and rapid autoxidation to form pyrrole dimers and trimers (15). These oligomerized products contain methylene bridges between C-2 of one pyrrole ring and C-3 of a second ring. Results of additional investigations indicated that the same autoxidative dimerization occurred when N-acetyllysine or lysinecontaining dipeptides are incubated with 2,5-HD (16). The pyrrole-to-pyrrole linkage was also elucidated in autoxidized, pyrrolylated protein. Thus, these studies support an earlier hypothesis that pyrrole-to-pyrrole

0 1995 American Chemical Society

Chem: Res. Toxicol., Vol. 8, No. 5, 1995 766

Thiol -Pyrrole Conjugates and 2,5-HD Neuropathy

Chart 1. Structures of Compounds under Study H

1

2

3

4

\

A

H

5

6

linkages mediate 2,B-HD-derived protein cross-linking (17). In addition to autoxidative pyrrole-pyrrole dimerization, alternative cross-linkingreactions involving oxidized pyrrole rings and protein nucleophiles, such as thiols and amines, have been postulated (18). Recently, a pyrrolethiol derivative, resulting from the reaction of 2,5dimethyl-N-isobutylpyrrolewith 2-mercaptoethanol,was isolated and structurally characterized (19). This model alkylpyrrole also reacted slowly with butylamine to yield several conjugated products under air atmosphere. However, because these oxidative products were formed with high concentrations of reactants and in an organic solvent (acetonitrile), it is difficult to predict whether the same reactions would occur between trace amounts of pyrrole adducts in protein and protein thiols or amines under aqueous conditions. In the present study, the involvement of biological nucleophiles in pyrrole-mediated protein cross-linking under physiological conditions was investigated. The reactivity of a number of species, including N-acetyllysine, N-acetylcysteine, and glutathione, was examined, in addition to the reactivity of proteins containing free thiols and amines. Nucleophiles that exhibited high reactivity toward pyrrole adducts were identified, and the reaction products and mechanisms were investigated. On the basis of these results, we propose that the major role played by biological nucleophiles in 2,5-HD neuropathy may be the inhibition of pyrrole-mediated protein crosslinking, rather than the mediation of protein crosslinking via pyrrole-thiol or pyrrole-amine linkages.

Experimental Procedures Chemicals. Caution: 2,5-Hexanedione is a mammalian neurotoxicant. It should be handled with gloves under a fume

H

H 8

7

hood. 2,5-HD (1; Chart 1) was purchased from Eastman Organic Chemicals (Rochester, NY).N a-Acetyllysine (Ac-Lys; 3) was obtained from Bachem Bioscience (Philadelphia, PA). Ribonuclease A (RNase), ovalbumin (OA),N a-acetylcysteine (AcCys; 2), reduced glutathione (GSH; 41, and p(dimethylamino1benzaldehyde (DMAB) were from Sigma Chemical Co. (St. Louis, MO). Oxidized glutathione (GSSG) was obtained from Aldrich Chemical Co. (Milwaukee, WI). Acrylamidehis(acry1amide) solutions for SDS-PAGE and Dye Reagent Concentrate for protein assay were from Bio-Rad Laboratories (Richmond, CA). Syntheses. N a-Acetyl-6-(2,5-dimethylpyrrol-l-yl)norleucine (5) was prepared as previously reported (26).The structure of the pyrrolylated lysine derivative was confirmed by lH-NMR: 6 1.50 (2H, m), 1.6-1.9 (4H, m), 2.08 (3H, s ) , 2.27 (6H, s), 3.83 (3H, t, J = 7.9 Hz), 4.38 (lH, m), 5.74 (2H, s), and thermosprayMS: [M HI+ at m l z 267, [M H - HzO]+ a t m l z 249, and DMAl3-based pyrrole assay. The dimer of N a-acetyl-6-(2,5dimethylpyrrol-1-y1)norleucine(6)was synthesized by autoxidation of 5 in phosphate buffer (pH 7.4) under air at 37 "C for 2 days. A nonpolar compound present in the oxidized sample was isolated by semipreparative HPLC. Thermospray-MS and lH-NMR analysis of the product gave results identical t o those previously reported for 6 (16). N a-Acetyl-6-[3-(Na-acetylcystein-S-yl)-2,5-dimethylpyrrol-lyllnorleucine (7) was synthesized from 2 (16 mg), 3 (19 mg), and 1 (33 pL). These compounds were dissolved in 1mL of 200 mM sodium phosphate buffer (pH 7.4), and the mixture was incubated at 37 "C under air (standard autoxidation conditions) for 4 days. The product was isolated by semipreparative HPLC, dried under Nz, dissolved in MeOD, and analyzed by lH-NMR spectroscopy: 6 1.54-1.89 (Lys-/?,G CHz; 4H, m), 1.40 (Lys-y CHz; 2H, m), 3.76 (Lys-E CHz; 2H, t, J = 7.5 Hz), 4.25-4.33 (Lys-a, Cys-a CH; 2H, m), 2.72 (Cys-/?CH; lH, dd, J M = 9.3, Jpp = 13.2 Hz), 3.03 (cYs-~' CH; l H , dd, Jqj = 3.9, Jflp = 13.2 Hz), 1.93, 1.96 (RCOCH3; 3H, s), 2.16, 2.23 (pyrrole-CH3; 3H, s ) , 5.81 (pyrrole-H; lH, s). Thermospray-MS analysis revealed a n ion for the protonated molecule ([M + HI+) at m l z 428.

+

+

766 Chem. Res. Toxicol., Vol. 8, No. 5, 1995 Pyrrolylated proteins were prepared as previously reported (16,171.Briefly, the proteins (RNase or OA) were dissolved (10 mg/mL) in 200 mM sodium phosphate buffer (pH 7.4 or 9.4), and 1 was added to give a 10- to 50-fold molar excess over the concentration of protein lysine residues (20,21).Pyrrolylation was achieved under argon a t 37 "C for 12-36 h in the dark. Unreacted 1 was removed by exhaustive dialysis against phosphate buffer (pH 7.4)under argon at 4 "C. HPLC and Thermospray-MS. Separation and quantitation of the pyrrole derivatives and their oxidation and conjugation products were performed on a Waters HPLC system (Waters Associates, Milford, MA) with a Nova-Pak (Waters) C18 column (0.39x 15 cm, 4 pm particle size) a t a flow rate of 1mumin. Details of analytical methods and instrumentation were previously described (15,16). For most HPLC analyses, including HPLC/thermospray-MS, a gradient system consisting of eluent A (0.1 M ammonium acetate in water) and eluent B (0.1 M ammonium acetate in acetonitrile-water, 1:l) was used as follows: from 0% B to 20% B over 10 min, then to 100% B over 15 min, hold for 5 min, to 0% B over 5 min. For semipreparative isolation, a 1.0 x 25 cm CIS column (Supelco, Bellefonte, PA) was used a t a flow rate of 4 mumin. HPLC separation of incubation products of 1,3,and 4 was achieved using a TFAbased elution system (eluent A 0.06% TFA in water, eluent B: 0.06% TFA in acetonitrile-water, 1:l) with the following gradient 0-30% B over 7 min, 30-55% B over 13 min, to 70% B over 10 min, to 80% B over 5 min, to 0% B over 5 min. The mass spectrometer used in this study was a HP Model 5970 quadrupole mass analyzer (Hewlett-Packard, Palo Alto, CA), equipped with a Vestec Model 101 thermospray interface (Vestec, Houston, TX). Samples were introduced by on-line HPLC or loop injection. Operation of the mass spectrometer was previously described in detail (15). Other Analytical Methods. A Varian XL-300NMR spectrometer was used to obtain 'H-NMR spectra of synthetic 2 and 7. HPLC-isolated and lyophilized samples were redissolved in CD30D and spectra recorded a t 75.4MHz. Chemical shifts were relative to tetramethylsilane. SDS-PAGE was carried out as previously reported (16).Approximately5-10 pg of protein was loaded per lane for each analytical gel (7.5% acrylamide separating gel, 5% stacking gel). The gel was run under constant current (30 mA) a t 10 "C. Separated proteins were fmed in 10% TCA and stained by Coomassie Blue G-250. The quantitation of 5 and pyrrole adducts in protein employed the DMAB-based colorimetric assay (3). Protein determination utilized the Bio-Rad protein assay reagent.

Results Effects of Thiols and Amines on Pyrrole-to-Pyrrole Protein Cross-Linking. The chemical reactivities of low-molecular-mass nucleophilic thiols and amines toward pyrrole adducts were tested by determining their ability to inhibit pyrrole-to-pyrrole cross-linking of model proteins. Pyrrolylated RNase (10 mg/mL; 0.68 nmol of pyrrole adductlpg of protein) was incubated without or with Ac-Cys (2), Ac-Lys (3),or 5 (25-fold molar excess over protein pyrrole adducts) for 40 h. The formation of protein polymers was assessed by SDS-PAGE (Figure 1A). Extensive protein dimer, trimer, and tetramer formation was observed with incubation times of up to 40 h without any additions (Figure lA, lane 2). In contrast, the addition of pyrrole 5 inhibited protein crosslinking (Figure lA, lane 3). Adduction of 5 to pyrrolylated RNase, presumably via pyrrole-to-pyrrole binding, resulted in a broadened monomeric band. The inclusion of Ac-Cys also caused a time-dependent inhibition of the formation of RNase polymers, but the inhibition was not as complete as with equimolar levels of 5 (Figure lA, lane 5). The band of monomeric RNase in this sample was only slightly broadened. In contrast to the inhibitory

Zhu et al.

A

B

I*.2- 3 4

f

f

2- 3

4

-5

-P

Figure 1. SDS-PAGE analysis of autoxidative cross-linking of pyrrolylated RNase. Gel A. Protein cross-linkingafter 0 (lane 1)or 40 h (lane 2) under oxidative conditions, or after 40 h in the presence of 5 (lane 31,Ac-Lys (lane 4),or Ac-Cys (lane 5). Gel B: Protein cross-linking after 0 h (lane 11,or 24 h (lane 5), or after 24 h in the presence of a 50-fold (lane 2),25-fold (lane 31,or 10-fold (lane 4)molar excess of GSH to pyrrole adducts.

effects of 5 and Ac-Cys, inclusion of Ac-Lys showed no effect on pyrrole-mediated protein cross-linking (Figure lA, lane 4). Data from this experiment provide strong evidence for interactions between pyrrole adducts and thiol groups, but not amino groups, during the process of pyrrole oxidation under physiologic conditions. The interactions of biological thiols with pyrroles were h r t h e r investigated by using GSH (4). Pyrrolylated RNase (10 mg/mL; 0.062 nmol of pyrrole adductdpg of protein) was allowed to incubate under standard autoxidation conditions for 24 h without additions, or with a 2.5, lo-, or 50-fold molar excess of GSH over pyrrole adducts. SDS-PAGE showed that RNase dimers and trimers were formed during the incubation of pyrrolylated RNase alone (Figure lB, lane 5). As reflected by the decrease in levels of the dimers and trimers, GSH inhibited protein cross-linking in a concentration-dependent manner (Figure lB, lanes 2-4). The reactivity of free thiols and amines present in proteins in cross-linking reactions was also explored using RNase and OA as model proteins. RNase contains no free thiols, while OA has four cysteine residues per molecule (20,21).These proteins contain 10 and 20 mol of lysine/mol of protein, respectively. Various combinations of pyrrolylated RNase (pyrrole-RN) (0.43 nmol of pyrroldpg of protein; 5.89 nmol of pyrroldnmol of protein) and pyrrolylated ovalbumin (pyrrole-OA) (0.13 nmol of pyrrolelpg of protein; 5.47 nmol of pyrrole/nmol of protein) were incubated under standard conditions. The resulting intermolecular protein cross-links were analyzed by SDS-PAGE as presented in Figure 2. After 24 h of oxidation, dimers of OA as well as RNase polymers became evident (Figure 2, lanes 3 and 4). The formation of RN-OA heterodimers was observed when an equimolar mixture of pyrrole-RN and pyrrole-OA was incubated (Figure 2, lane 7, arrow). In contrast, the incubation of RN with pyrrole-OA or the incubation of OA with pyrroleRN did not result in the formation of heteropolymeric protein (Figure 2, lanes 5 and 6). Formation and Structural Characterization of Thiol-Pyrrole Conjugates. In order to examine possible products of the interactions between pyrroleprotein adducts and low-molecular-weight biological thiol compounds, Ac-Cys (10 mM) or GSH (10 mM) was incubated with mixtures of 2,5-HD (100 mM) and AcLys (100 mM). Products were separated by reversedphase HPLC and characterized by on-line thermospray-

Chem. Res. Toxicol., Vol. 8, No. 5,1995 767

Thiol-Pyrrole Conjugates and 2,5-HD Neuropathy

1 2 3 4 5 6 7

+

+

c

U

PRlD

Figure 2. SDS-PAGE analysis of autoxidative cross-linking of pyrrolylated protein mixtures. Control OA and RNase (lanes 1 and 2); pyrrolylated OA (lane 3); pyrrolylated RNase (lane 4); equimolar mixture of pyrrolylated OA and control RNase (lane 5); equimolar mixture of control OA and pyrrolylated RNase (lane 6);equimolar mixture of pyrrolylated OA and pyrrolylated RNase (lane 7). Arrow shows presence of pyrrolylated OApyrrolylated RNase heterodimer in lane 7. All incubationswere for 24 h with standard autoxidative conditions.

70

2.0

25

1.5

Q)

0

g

1.0

U

c

3 0.5

P 0

5

10

15

assigned to pyrrole 5 based on its thermospray mass spectrum ([M HI+ at m l z 267) and DMAB-based pyrrole assay of the recovered fraction. The mass spectrum of a new product, represented by the peak at 16.5 min (Figure 3A, arrow), is shown in Figure 3B. On the basis of the observed peak for the protonated molecule ([M H]+ at m l z 428) and the identification of several fragment ions, this product was assigned to thiol conjugate 7, as illustrated in Figure 3B. Ions at m l z 299 and 267 were generated from the cleavage of the two C-S bonds, and ions at m l z 410,281, and 249 are proposed to result from the loss of water molecules from the ions at m l z 428, 299, and 267, respectively. The structure of conjugate 7 was further confirmed by lH-NMR analysis following semipreparative HPLC purification (see under Experimental Procedures). In addition to the peaks representing Ha of cysteine and lysine and the peaks representing methylene Hp, H,, and Hd of lysine (these proton signals were also present in the lHNMR spectrum of 5; see ref 161, there were several characteristic peaks consistent with a bond from the sulfur atom of Ac-Cys to the pyrrole ring at the C-3 position. The two diastereoisotopic Hp of cysteine exhibited a characteristic pattern of resonances and splitting in the NMR spectrum, analogous to those of previously reported GSH conjugates (22,23). Incubation of GSH with 2,5-HD and Ac-Lys under air for 68 h resulted in the formation of several products. In this case, a TFA-based HPLC gradient was used for analysis, as the ammonium acetate system could not effectively separate Ac-Lys, GSH, and GSSG. The W chromatogram (maxplot mode, 225 and 265 nm) from analysis of this incubation sample is shown in Figure 4A-C. Peaks 1 and 2 of Figure 4A represent unreacted Ac-Lys and GSH, respectively. The compound corresponding to peak 3 is unreacted 2,5-HD. Pyrrolylation of Ac-Lys led to the formation of 5 (Figure 4B, peak 6), which underwent subsequent autoxidative reaction to yield dimer 6 (Figure 4B, peak 7). The formation of these derivatives has been previously reported (16). The decrease in GSH levels and the appearance of GSSG after 28 h, represented by peak 4 in Figure 4B,C ([M HI+ a t m l z 613), indicate that GSH was oxidized rapidly under these reaction conditions. By 68 h, GSH (peak 2) was totally consumed, probably by a combination of oxidation and pyrrole conjugation pathways (Figure 4C). The compound corresponding to peak 5 was an unknown product that was not generated when 2,5-HD and Ac-Lys were incubated in the absence of GSH (data not shown). Thermospray-MS analysis of the product represented by peak 5, following semipreparative HPLC isolation using an ammonium acetate gradient system, allowed its characterization as GSH conjugate 8. The mass spectrum and assigned structure of peak 5 are illustrated in Figure 4D. The spectrum exhibits a prominent peak for the protonated molecule at m l z 572. Cleavage at either of two C-S bonds led to the formation of protonated pyrrole ions at m l z 267 and 249 ( m l z 267 - H20) as well as of sulfur adduct ions at m l z 299 and 281 ( m l z 299 - H2O). These fragments ions are identical to those of thiol conjugate 7 (Figure 3B), suggesting that a sulfur atom of GSH was directly bound to the pyrrole ring of 5. There are several “classicalions” resulting from thermospray ionization of GSH conjugates, including those at m l z 129, 147,443 ([M + H - 129]+),425 ([M H - 147]+), and 407 ([M H - 147 - H20]+>. The formation of these ions during thermospray ionization

20

25

Time (min)

+

c

I

267

a> 40

.-> Y

ca 5 w

20

0 100

200

300

400

500

mlz Figure 3. HPLC/thermospray-MS analysis of reaction products of 2,5-HD (l),Ac-Cys (2), and (Ac-Lys)(3) under autoxidative conditions. Panel A Total-ion chromatogram of HPLChhermospray-MS analysis of the reaction mixture after 24 h. Arrow indicates the peak for the thiol-pyrrole conjugate (7). Panel B: thermospray mass spectrum of thiol-pyrrole conjugate eluting at 16.5 min in panel A, with assigned structure.

MS. The utility of this analytical system was previously demonstrated in the study of pyrrole dimerization reactions (15).Figure 3A shows the total-ion chromatogram of an incubation mixture of 2,5-HD, Ac-Cys, and Ac-Lys a t 37 “C under air for 24 h. The broad peak at 3.5 min corresponds to both unreacted Ac-Cys ([M HI+ at m l z 164) and Ac-Lys ([M H]+ at m l z 189), and the peak at 12.2 min is unreacted 2,5-HD. The peak at 19.7 min was

+

+

+

+

Zhu et al.

768 Chem. Res. Toxicol., Vol. 8, No. 5, 1995 loo

90

4 W

5

0

5

10

15

20

25

30

15

20

Time (h)

I 0

10

Figure 5. 5. Time course of inhibition of autoxidative dimer-

35

ization of pyrrole 5 by GSH. Purified 5 (1.27 mM) was incubated with or without GSH (6.35 mM) under autoxidative conditions, followed by HPLC analysis of relative levels of reaction products. Panel A Decrease in levels of 5 in the absence ( 0 )and presence (0) of GSH over time. Panel B: Time course of appearance of GSSG (W), pyrrole-thiol conjugate 8 (A), pyrrole dimer 6 in the presence of GSH (O),and pyrrole dimer 6 in the absence of GSH

Time (min)

(0). 1

200

300

400

500

1

'

I

'

1

'

I

'

I

800

mh Figure 4. HPLC/thermospray-MSanalysis of reaction products of 2,5-HD (l),Ac-Lys (3),and GSH (4) under autoxidative conditions. Panels A-C: HPLC chromatogram (with TFA Nmaxplot detection at 225 and 265 gradient elution system; l nm) of reaction mixture a t 0, 28, and 68 h. Identities of peaks 1-7 are discussed under Results. Panel D: Thermospray mass spectrum of thiol-pyrrole conjugate ( 8 ) (peak 5; isolated by semipreparative HPLC with ammonium acetate gradient elution system), with assigned structure.

has been previously described in detail (24,25). The GSH conjugate 8 was not produced under argon, suggesting that thiol-pyrrole conjugation requires molecular oxygen, as does pyrrole dimerization (15). Kinetic Studies of GSH Conjugation and the Inhibition of Pyrrole Dimerization by GSH. The time-dependent inhibition of pyrrole dimerization by GSH is shown in Figure 5. When pyrrole 6 was incubated (1.27 mM) in phosphate buffer under air, its concentration was found to decrease linearly over time (Figure 5A). Pyrrole levels also decreased but were slightly higher after 20 h when GSH (6.35 mM) was present in the incubation mixture. When 5 was incubated without addition of GSH, the formation of dimer 6 was directly proportional to the reaction time (Figure 5B). In contrast, in the presence of GSH, the rate of pyrrole dimerization was very slow in the first 10 h, while the level of pyrrole-thiol conjugate 8 steadily increased. After this initial period, the rate of pyrrole dimerization

C

.-p0

't

40

O L 0

20

40

60

80

Time (h)

Figure 6. Relative stability of pyrrole 5 ( 0 )and pyrrole-thiol conjugate 8 (W) under autoxidative conditions. Purified derivatives (6.4 mM) were incubated in 200 mM phosphate buffer (pH 7.4) under air, and levels were determined by HPLC analysis. w a s accelerated, despite the relatively lower level of 5

at this time (Figure 5B). Conversely, GSSG was rapidly produced in the first 10 h but at a reduced rate thereafter, consistent with a depletion of GSH in the incubation mixture. Results of stability studies of pyrrole 5 and the thiol conjugate 8 under physiologic conditions are shown in Figure 6. Relative levels of these compounds as a function of incubation time were determined as peak areas by HPLC, compared with zero-time concentrations. Levels of pyrrole 6 rapidly decreased with time, consistent with instability due to autoxidative reaction. In contrast, more than 80% of 8 remained after 70 h of incubation, indicating a higher stability of the thiol

Thiol-Pyrrole Conjugates and 2,5-HD Neuropathy conjugate in aqueous medium. In addition, the reactivity of thiol conjugates 7 and 8 toward DMAB was examined and found to be negligible, suggesting deactivation of the pyrrole ring (data not shown).

Discussion In our previous studies of 2,5-HD-treated amines, peptides, and proteins, the pyrrole-to-pyrrole cross-linking reactions and the structures of the resulting dimers were determined by using HPLUthermospray-MS and NMR spectroscopy (15,161. Since it was recognized that 2,5-HD-induced protein cross-linking is mediated by the autoxidation of pyrrole adducts, it has also been proposed that covalent reaction resulting in protein cross-linking might also occur between an oxidized pyrrole adduct in one protein and a nucleophilic thiol or amine of a second protein (13). However, owing to difficulties in the analysis of products of pyrrole-mediated protein crosslinking, few studies that address this hypothesis have been published (18,19). In these studies, low-molecularmass thiols and amines were found to inhibit covalent binding of pyrrolylated lysines to ovalbumin. In addition, a pyrrole-thiol adduct (sulfur bound at the C-3 position of the pyrrole ring) was characterized when high concentrations of 2-mercaptoethanol and a model 2,5-dimethyl-N-alkylpyrrole were incubated under organic reaction conditions. Our previous studies and technical developments provided an opportunity to further explore the role of biological thiols and amines in the protein cross-linking reaction and the chemical mechanisms involved. In the present investigation, the reactivity of small nucleophiles, including N a-acetyllysine, N “-acetylcysteine, and GSH, toward pyrrole adducts in protein was investigated, and the ability of these compounds to inhibit pyrrole-topyrrole cross-linking of pyrrolylated RNase was determined. Results of these experiments demonstrated that free thiols, but not free amines, can influence pyrrole autoxidation under model physiologic conditions. The latter result is in conflict with published results of similar experiments conducted under nonaqueous conditions (19). The finding of the inactivity of low-molecular-mass lysyl amines supports our previous observations that free amino groups in protein are not major participants in the mechanism of cross-linking of pyrrolylated proteins (17). Once the interactions of low-molecular-massthiols with pyrrole adducts in protein were established, the question of whether similar reactions occur between protein-bound thiols and pyrrole adducts was examined. Experiments employing OA (which contains four free thiol residues and 20 lysines per molecule) and RNase (no free thiols, ten lysines) demonstrated that, at least for these proteins, cross-linking was mediated by pyrrole-to-pyrrole and not pyrrole-to-thiol or pyrrole-to-lysyl bridging. These results are consistent with previous in vitro findings that the single free thiol in bovine serum albumin (BSA) was not capable of covalent binding to adducts in pyrrolylated RNase (17). It is likely that cysteinyl thiols in BSA and other proteins are, for the most part, buried in less accessible pockets within the hydrophobic interior of the macromolecule (26, 27) and are therefore not readily accessible for reaction with pyrrole adducts bound to a second protein molecule. If particular cysteinyl thiols were located on the surface of a protein, cross-links would probably be generated via disulfide bridging instead of

Chem. Res. Toxicol., Vol. 8, No. 5, 1995 769 pyrrole-to-thiol bridging, since our findings indicate that the oxidation of thiols to disulfides appears to be the more rapid process. The important molecular targets of the y-diketones are thought to be the NF proteins, which have high molecular mass and extremely low turnover rates in vivo (28, 29). The rat NF-L, -M, and -H subunit proteins contain only 1,4,and 5 cysteine residues, respectively, some or all of which may be tied up in disulfide linkages (30-32). In addition, the few cysteines in these proteins are located in regions involved in coiled-coil formation and are thus unlikely to be available for covalent reaction. Following in vitro 2,5-HD exposure, few pyrrole adducts appear to be localized in these domains (3). On the basis of these experimental results and structural considerations, we propose that pyrrole-to-pyrrole and not pyrrole-to-thiol linkages mediate NF protein cross-linking in vivo. Although it is known that GSH reacts covalently with a variety of xenobiotic species, owing to the nucleophilicity of its thiol group (331, reactions of GSH and other thiols with N-substituted 2,5-dimethylpyrrole adducts in aqueous solution have not been documented. Our in vitro experiments using low-molecular-mass pyrroles revealed that GSH and N “-acetylcysteine react with 2,5-dimethylN-alkylpyrroles under model physiologic conditions to yield pyrrole-thiol conjugates. Further studies showed that the GSH conjugation reaction is oxygen-dependent and is a major pathway of pyrrole oxidation when present. GSH is an important thiol-containing molecule in humans. It is present in most cells of many tissues, including liver, kidney, and brain, at concentrations up to 10 mM (33). The small size of GSH should allow it much more facile access to protein-pyrrole adducts than thiols present in proteins would have. Therefore, the formation of the GSH conjugates of pyrrolylated protein in vivo may represent a potentially important alternative reaction pathway for pyrrole adducts. Evidence from the present study also suggests that the inhibition of pyrrole dimerization by thiols could involve a second mechanism, distinct from simple competitive reaction with the pyrrole. As demonstrated in these experiments, the addition of GSH may change the kinetics of pyrrole dimerization, since the relative rate of formation of pyrrole dimer was not proportional to the decrease in the levels of unoxidized pyrrole. Furthermore, addition of GSH did result in a modest increase in the levels of unoxidized pyrrole, a result that would be expected with a direct inhibition mechanism. In addition, the decreased cross-linking of pyrrolylated RNase by N “-acetylcysteine was not associated with a broadened monomeric protein band in SDS-PAGE, as was observed for low-molecular-mass pyrroles. This observation may reflect either a low level of pyrrole-thiol adduct or a lesser conformational change induced by the more polar thiol adduct. These observations are not totally consistent with a mechanism based only on competitive pyrrole-to-thiol conjugation. Alternatively, the inhibition of protein cross-linking may be associated with free radical scavenging. GSH is known to be a good hydrogen donor for free radicals, since the S-H bond is relatively weak (34). For instance, GSH has an important role in protection against the deleterious effect of radiation, where it reacts with hydroxyl, peroxyl, and other radicals that are generated during this process (35). The resulting thiol radical, GS’, is relatively stable and can be further converted to GSSG. Our previous studies suggested that

Zhu et al.

770 Chem. Res. Toxicol., Vol. 8, No.5, 1995

Scheme 1. Proposed Mechanisms of Inhibition of Autoxidative Pyrrole-to-PyrroleCross-Linkingby GSH,via Thiol-to-PyrroleConjugation and Direct Antioxidant Action 0

1

\ /\

t O2

‘OSH\

lnhlbklon

-

i

\

fraction of globin-pyrrole adducts would form thiol conjugates. We propose that GSH-pyrrole conjugates in globin should be evaluated as possible stable biomarkers of 2,5-HD exposure. Finally, it has been reported that GSH is present at relatively high levels in astroglial.and ependymal cells within the brain (39,40).Although its concentration in the microenvironment of N F s is u r h q m n , histochemical evidence indicates the presence of GSH within internodal regions of peripheral axons (41). If, as has been proposed, covalent cross-linking of NF proteins is critical to the mechanism, it is conceivable that biological thiols could exert protective effects against y-diketone neuropathy by interrupting the cross-linking process. Alternatively, a lack of effect for GSH would be consistent with the hypothesis that pyrrole formation itself is sufficient to cause neuropathy. The induction of GSH synthesis in nerve axons or the direct application of GSH to nerves exposed to 2,5-HD may be additional ways to examine the relevance of each of these hypotheses.

Acknowledgment. This work was supported by a research grant (ES-05172) from the National Institute for Environmental Health Sciences (to A.P.D.). In addition, support from the National Science Foundation for the purchase of NMR instrumentation used in this research is gratefully acknowledged.

SG

References GSH-conjugated adduct

crosslinked adducts

pyrrole dimerization may involve free radical chain reactions (15,161, and GSH could directly affect the formation of pyrrole dimer by inhibiting this process. Further experiments in which quantitative product analyses are performed and rate constants are determined for pyrrole oxidations and related free radicals are required to confirm the action of GSH as a radical inhibitor. The two potential pathways of inhibition of autoxidative pyrrole-to-pyrrole cross-linking by GSH are summarized in Scheme 1. The findings of in vitro inhibition of pyrrole-mediated protein cross-linking by GSH and the GSH-pyrrole conjugation reaction have several important implications for the mechanistic study of 2,5-HD neuropathy. For example, the DMAB-based pyrrole assay is widely used for the quantitation of the protein adducts in vivo (4,11, 14). Pyrrole levels in protein determined by this method have been considered to represent the total concentration of pyrrole adducts. However, the present study has shown that, if pyrrole adducts are converted to thiol conjugates in vivo, they would not be detected by the assay. Therefore, the DMAB assay could underestimate total pyrrole adducts under these conditions, leading to incorrect conclusions regarding relative adduct levels between various y-diketones. In addition, stable xenobiotic-protein adducts have been considered as good biological markers of chronic chemical exposures (36). The possible use of pyrrole adducts in rat globin as biomarkers for n-hexane exposure has been suggested (37). However, as indicated in our studies, thiol-pyrrole conjugates are much more stable than native pyrrole adducts in aqueous solution. The conjugation also occurs in the absence of any specific transferase enzymes. Since erythrocytes can contain as much as 3 pmol of GSWmL of packed cells (38),it is therefore reasonable to expect that at least a certain

Arlien-Soborg, P. (1992)Solvent Neurotoxicity, pp 155-183, CRC Press, Boca Raton, FL. Perbellini, L., Amantini, M. C., Brugnone, F., and Frontali, N. (1982)Urinary excretion of n-hexane metabolites: A comparative study in rat, rabbit, and monkey. Arch. Toxicol. 50, 203-215. DeCaprio, A. P., and Fowke, J. H. (1992) Limited and selective adduction of carboxyl-terminal lysines in the high molecular weight neurofilament proteins by 2,5-hexanedionein vitro. Brain Res. 566,219-228. DeCaprio, A. P., and O’Neill, E. A. (1985) Alterations in rat axonal cytoskeletal proteins induced by in vitro and in vivo 2,5-hexanedione exposure. Toxicol. Appl. Pharmacol. 78, 235-247. Sayre, L. M., Shearson, C. M., Wongmongkolrit, T., Medori, R., and Gambetti, P. (1986) Structural basis of gamma-diketone neurotoxicity: Non-neurotoxicity of 3,3-dimethyl-2,5-hexanedione, a gamma-diketone incapable of pyrrole formation. Toxicol. Appl. Pharmacol. 84,36-44. Genter, M. B., Szakal-Quin, G., Anderson, W., Anthony, D. C., and Graham, D. G. (1986) Evidence that pyrrole formation is a pathogenetic step in gamma-diketone neuropathy. Toxicol. Appl. Pharmacol. 87, 351-362. DeCaprio, A. P., Briggs, R. G., Jackowski, S.J., and Kim, J. C. S. (1988) Comparative neurotoxicity and pyrrole-forming potential of 2,5-hexanedione and perdeuterio-2,5-hexanedionein the rat. Toxicol. Appl. Pharmacol. 92, 75-85. DeCaprio, A. P. (1985) Molecular mechanisms of diketone neurotoxicity. Chem.-Bwl. Interact. 64, 257-270. Jones, R. A. (1990)Pyrroles. Part One,John Wiley and Sons, New York. Lapadula, D. M., Irwin, R. D., Suwita, E., and Abou Donia, M. B. (1986) Cross-linking of neurofdament proteins of rat spinal cord in vivo after administration of 2,5-hexanedione. J. Neurochem. 48,1843-1850. Anthony, D. C., Boekelheide, K, Anderson, C. W., and Graham, D. G. (1983) The effect of 3,4-dimethyl substitution on the neurotoxicity of 2,5-hexanedione. 11. Dimethyl substitution accelerates pyrrole formation and protein crosslinking. Toxicol. Appl. Pharmacol. 71,372-382. Carden, M. J., Lee, V. M. Y., and Schlaepfer, W. W. (1986) 2,5Hexanedione neuropathy is associated with the covalent crosslinking of neurofilament proteins. Neurochem. Pathol. 5, 25-35. Graham, D. G., Anthony, D. C., Boekelheide, K , Maschmann, N. A., Richards, R. G., Wolfram, J. W., and Shaw, B. R. (1982) Studies of the molecular pathogenesis of hexane neuropathy. 11. Evidence that pyrrole derivatization of lysyl residues leads to protein crosslinking. Toxicol. Appl. Pharmacol. 64, 415-422.

Thiol-Pyrrole Conjugates and 2,5-HD Neuropathy Genter St. Clair, M. B., Amamath, V., Moody, M. A., Anthony, D. C., Anderson, C. W., and Graham, D. G. (1988) Pyrrole oxidation and protein cross-linking as necessary steps in the development of y-diketone neuropathy. Chem. Res. Toxicol. 1, 179-185. Zhu, M., Spink, D. C., Bank, S., Chen, X., and DeCaprio, A. P. (1993) Analysis of alkylpyrrole autoxidation products by highperformance liquid chromatography with thermospray mass spectrometry and U V photodiode-array detection. J. Chromatogr. 628, 37-47. Zhu, M., Spink, D. C., Yan, B., Bank, S., and DeCaprio, A. P. (1994) Formation and structure of crosslinking and monomeric pyrrole autoxidation products in 2,5-hexanedione-treated amino acids, peptides, and protein. Chem. Res. Toxicol. 7 , 551-558. DeCaprio, A. P. (1986) Mechanisms of in vitro pyrrole adduct autoxidation in 2,5-hexanedione-treated protein. Mol. Pharmacol. 30,452-458. Anthony, D. C., Amamath, V., and Graham, D. G. (1990) Pyrrolemediated protein crosslinking: Toward an understanding of the molecular mechanism. Toxicologist 10, 183. Amamath, V., Valentine, W. M., Amamath, K., Eng, M. A., and Graham, D. G. (1994) The mechanism of nucleophilic substitution of alkylpyrroles in the presence of oxygen. Chem. Res. Toxicol. 7 , 56-61. Mitra, S., and Lawton, R. G. (1979) Reagents for the crosslinking of proteins by equilibrium transfer alkylation. J . Am. Chem. Socy 101, 3097-3110. Nisbet, A. D., Saundry, R. H., Moir, A. J. G., Fothergill, L. A., and Fothergill, J. E. (1981) The complete amino-acid sequence of hen ovalbumin. Eur. J . Biochem. 115,335-345. Rosen, G. M., Rauckman, E. J., Ellington, S. P., Dahlin, D. C., Christie, J. L., and Nelson, S.D. (1984) Reduction and glutathione conjugation reactions of N-acetyl-p-benzoquinone imine and two dimethylated analogues. Mol. Pharmacol. 25, 151-157. Ellis, M. K., Hill, S., and Foster, P. M. (1992) Reactions of nitrosonitrobenzenes with biological thiols: Identification and reactivity of glutathion-S-yl conjugates. Chem.-Biol.Interact. 82, 151-163. Dulik, D. M., Colvin, 0. M., and Fenselau, C. (1990) Characterization of glutathione conjugates of chlorambucil by fast atom bombardment and thermospray liquid chromatography/mass spectrometry. Biomed. Environ. Mass Spectrom. 19, 248-252. Baillie, T. A,, and Davis, M. R. (1993) Mass spectrometry in the analysis of glutathione conjugates. B i d . Mass Spectrom. 22,319325. Peters, T. (1985) Serum albumin. In Advances in Protein Chemistry, Anfinsen, C. B., Edsall, J. T., and Richards, F. M. Eds.) Vol. 37, pp 161-245, Academic Press, New York. Rose, G. D., Geselowitz,A. R., Lesser, G. J., Lee, R. H., and Zehfus, M. H. (1985) Hydrophobicity of amino acid residues in globular proteins. Science 229, 834-838.

Chem. Res. Toxicol., Vol. 8, No. 5, 1995 771 (28) Shaw, G. (1991) Neurofilament proteins. In The Neuronal Cytoskeleton (Burgoyne, R. D., Ed.) pp 185-214, Wiley-Liss, New York. (29) Nixon, R. A. (1991) Axonal transport of cytoskeletal proteins. In The Neuronal Cytoskeleton (Burgoyne, R. D., Ed.) pp 283-307, Wiley-Liss, New York. (30) Chin, S. S. M., and Liem, R. K. H. (1989) Expression of rat neurofilament proteins NF-L and NF-M in transfected nonneuronal cells. Eur. J. Cell Biol. 50, 475-490. (31) Napolitano, E. W., Chin, S. S. M., Colman, D. R., and Liem, R. K. H. (1987) Complete amino acid sequence and in vitro expression of rat NF-M: The middle molecular weight neurofilament protein. J.Neurosci. 7 , 2590-2599. (32) Chin, S. S. M., and Liem, R. K. H. (1990) Transfected rat highmolecular-weight neurofilament (NF-H) coassembles with vimentin in a predominantly nonphosphorylated form. J.Neurosci. 10, 3714-3726. (33) Reed, D. J. (1990) Glutathione: toxicologicalimplications. Annu. Rev. Pharmacol. Toxicol. 30, 603-631. (34) Diplock, A. T. (1994) Antioxidants and free radical scavengers. In Free Radical Damage and its Control (Rice-Evans, C. A., and Birdon, R. H., Eds.) pp 113-130, Elsevier, New York. (35) Biaglow, J. E., Varmes, M. E., Epp, E. R., Edward, R., Tuttle, S. W., and Held, K. D. (1989) Role of glutathione and other thiols in the cellular response to radiation and drugs. Drug Metab. Rev. 20, 1-12. (36) Skipper, P. L., and Tannenbaum, S. R. (1991) Protein adducts in the molecular dosimetry of chemical carcinogens. Carcinogenesis 10, 507-518. (37) Pyle, S. J.,Lee, A., and Anthony, D. C. (1994) Correlation of globin adducts with target-site adducts during hexanedione exposure. Toxicologist 14, 361. (38) Griffith, 0. W. (1981) Glutathione turnover in human erythrocytes. Inhibition by buthionine sulfoximine and incorporation of glycine by exchange. J. Biol. Chem. 256,4900-4904. (39) Slivka, A., Mytilineou, C., and Cohen, G. (1987) Histochemical evaluation of glutathione in brain. Brain Res. 409, 275-284. (40) Cooper, A. J. L., and Meister, A. (1993) Glutathione in the brain: disorders of glutathione metabolism. In The Molecular and Genetic Basis of Neurological Disease (Rosenberg, R., Prusiner, S.,DiMauro, S.,Barchi, R., and Kunkel, L., Eds.) pp 209-238, Butterworths, Stoneham, MA. (41) Philbert, M. A., Beiswanger, C. M., Waters, D. K., Reuhl, K. R., and Lowndes, H. E. (1991) Cellular and regional distribution of reduced glutathione in the nervous system of the rat histochemical localization by mercury orange and o-phthaldialdehydeinduced histofluorescence. Toxicol.Appl. Pharmucol. 107,215-227.

TX950004V