JOURNAL OF

InorganicBiochemistry

Journal of Inorganic Biochemistry 98 (2004) 11511159

www.elsevier.com/locate/jinorgbio

Equilibrium characterization of the As(III)cysteine and the As(III)glutathione systems in aqueous solution

Nicol_as A. Rey a, Oliver W. Howarth b, Elene C. Pereira-Maia a,*

a Departamento de Qu_mica ICEx, Universidade Federal de Minas Gerais 31.270-901 Belo Horizonte MG, Brazil b Centre for NMR, Department of Chemistry, University of Warwick, Coventry CV4 7AL, UK

Received 27 November 2003; received in revised form 19 March 2004; accepted 24 March 2004

Available online 21 April 2004

Abstract

Some arsenic compounds were the first antimicrobial agents specifically synthesized for the treatment of infectious diseases such as syphilis and trypanosomiasis. More recently, arsenic trioxide has been shown to be ecient in the treatment of acute promye-locytic leukemia. The exact mechanism of action has not been elucidated yet, but it seems to be related to arsenic binding to vicinal thiol groups of regulatory proteins. Glutathione is the major intracellular thiol and plays important roles in the cellular defense and metabolism. This paper reports on a study of the interactions between arsenic(III) and either cysteine or glutathione in aqueous solution.

The behavior observed for the As(III)glutathione system is very similar to that of As(III)cysteine. In both cases, the formation of two complexes in aqueous solution was evidenced by NMR and electronic spectroscopies and by potentiometry.

The formation constants of the cysteine complexes [As(H_1Cys)3], log K 29:846, and [As(H_2Cys)(OH)2]_, log K 12:019, and of the glutathione complexes [As(H_2GS)3]3_, log K 32:06, and [As(H_3GS)(OH)2]2_, log K 103 were calculated from potentiometric and spectroscopic data.

In both cases, the [As(HL)3] species, in which the amine groups are protonated, predominate from acidic to neutral media, and the [As(L)(OH)2] species appear in basic medium (the charges were omitted for the sake of simplicity). Spectroscopic data clearly show that the arsenite-binding site in both complexes is the sulfur atom of cysteine. In the [As(L)(OH)2] species, the coordination sphere is completed by two hydroxyl groups. In both cases, arsenic probably adopts a trigonal pyramidal geometry. Above pH 10, the formation of [As(OH)2O]_ excludes the thiolates from arsenic coordination sites. At physiological pH, almost 80% of the ligand is present as [As(HL)3].2004 Elsevier Inc. All rights reserved.

Keywords: Arsenic(III); Glutathione; Cysteine; Equilibrium studies; Spectroscopy

1. Introduction

The discovery of an organoarsenic compound by Ehrlich and co-workers in 1909, salvarsan (arsphena-mine), proven to be eective in the treatment of syphilis, stimulated the development of a wide range of arsenicals for the treatment of infectious diseases [1]. More re-cently, As2O3 has been reported to induce complete re-mission in patients with acute promyelocytic leukemia

* Corresponding author. Tel.: +55-31-34226506; fax: +55-31-34995700.E-mail address: elene@apolo.qui.ufmg.br (E.C. Pereira-Maia).

0162-0134/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jinorgbio.2004.03.010

[2]. Like other antitumoral drugs, arsenic has been used in the clinical practice before its mechanism of action being completely elucidated. Arsenic trioxide was shown to inhibit cell growth and induce apoptosis in several cell lines [36]. However, the mechanism of arsenic-induced apoptosis in tumor cells remains unclear. Arsenite in-duces the breaking of DNA strands, stimulates poly-(ADP-ribosylation), induces an increase in cellular levels of nitric oxide and superoxide, and aects protein phosphorylation by binding to vicinal thiols [7].

The eects of arsenic can be modulated by combining it with other compounds, such as thiol-containing mol-ecules. For example, Watson et al. [8] found that some1152N.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 11511159

thiols protect cells from the toxic eects of arsenite. Accordingly, Dai and co-workers [4] observed that the As2O3-induced apoptosis was inversely related to in-tracellular glutathione (GSH) concentration. In con-trast, Gurr et al. [7] found that dithiothreitol enhances arsenic trioxide-induced apoptosis. The authors pro-posed that arsenite can complex dithiothreitol produc-ing a new compound that has a higher potency in inducing apoptosis. St_yblo et al. [9] have shown that arsenothiols were more potent inhibitors of the gluta-thione reductase than arsenicals.

The biometabolism of arsenic involves a redox cycle between As(V) and As(III) with subsequent methylation of As(III), giving rise to the mono, di- and trimethylated derivatives. The reduction of arsenate to arsenite can be promoted nonenzymatically by glutathione, however, in physiological conditions, the reduction by As(V) re-ductases may predominate. Arsenic(III) is methylated in the liver during its hepato-enteric circulation by meth-yltransferases [10]. Arsenic is mostly excreted in urine in the form of methylated metabolites, being dimethylated arsenic the major form [11]. Because of this fact, bio-transformation was during a long time considered a detoxification mechanism. But recently, it was proposed that, in fact, methylation is a pathway for arsenic acti-vation because methylated forms can persist in tissues and are more toxic than inorganic arsenic(III) [12].

It is well known that arsenic(III) has a high anity for sulfur containing molecules such as dithiothreitol [13] and glutathione [14]. Scott et al. [15] isolated and characterized an arsenite complex of glutathione as As(SG)3 by mass spectrometry. Delnomdedieu et al. [16] found that glutathione reduces arsenate to arsenite and forms a (glutathione)3arsenite complex.

Another point of interest when considering the clin-ical use of arsenicals is the resistance to metalloid salts found in bacteria, fungi, parasites and animals. In bac-teria, the resistance system responsible for detoxification of metalloids transports arsenite out of the cell. This arsenite-eux system can be conferred by a carrier protein (ArsB) or an anion-translocating ATP-ase (Ar-sAB) [17]. If arsenic is present as arsenate, it must be reduced to arsenite prior to extrusion. The reduction is catalized by thiol-linked reductases that use glutare-doxin, glutathione or thioredoxin as reductants [18]. In eukaryotes, another transmembrane protein, MRP1, functions as an eux pump. It has been shown that tumor cells overexpressing MRP1 exhibit cross-resis-tance to arsenite [19]. Resistance mediated by MRP1 requires intracellular GSH [20], suggesting that either GSH is a co-transported substrate or the transported substrate is an arsenite complex of GSH. There is a controversy in current literature about the formation of a complex between arsenic and GSH at physiological pH values. The formation of an As(glutathione)3 com-plex has already been evidenced, but there is a lack of information about the stability of this species. So there is a need for a better quantitative understanding of the equilibrium between As(III) and glutathione. Once the stability constants of all the relevant complexes formed have been evaluated, one could simulate spe-cies distribution under physiological pH and GSH concentrations.

2. Experimental

2.1. Reagents

Oxidized and reduced forms of glutathione and DL-cysteine hydrochloride were used as obtained from Sigma. Stock solutions of glutathione and cysteine were prepared just before use under nitrogen atmosphere to prevent ligand oxidation. For NMR experiments, the ligands were dissolved in D2O. Stock solutions of so-dium metaarsenite, also from Sigma, contained 0.1 M of perchloric acid.

2.2. Spectroscopic measurements

A Diode Array Hewlett Packard 8451 A spectrometer equipped with a Masterline 2095 thermostat at 25 LC was used for UV and visible absorption measurements. Re-sults are expressed in terms of the molar absorption co-ecient e related to the total concentration of the ligand. The concentration of ligand used in the spectroscopic measurements varied from 2.0 _ 10_4 to 1.5 _ 10 _2 M. The stability constants were calculated from the spec-trophotometric data with the SQUAD algorithm [21].

NMR spectra were obtained using a Bruker Avance DRX 400 spectrometer with tetramethylsilane as an internal standard. A small quantity of acetone was ad-ded to the samples and used as a secondary reference (d 2.05). The concentration of ligand used in the mea-surements was 1.5 _ 10_2 M.

2.3. Potentiometric studies

Potentiometric titrations were performed by mea-suring the electromotive force (emf) using the Metrohm glass electrode 6.0102.100 (PB) as indicator and the Metrohm calomel electrode 6.0702.100 (0.1 M NaCl solution) as reference. A titroprocessor Metrohm 670, coupled to a Metrohm Dosimat 665 autoburette, was used to measure the emf. Experiments were carried out under nitrogen atmosphere and the temperature was kept constant at 25 LC. The ionic strength was main-tained at 0.1 M with sodium perchlorate. The ligand concentration lay in the range 4 _ 10_ 3 M to 1 _ 10_2 M and the ligand-to-metal molar ratios were in the range 50.5. The electrode system was calibrated for hydrogen ion concentration by perchloric acid titration (or sodiumN.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 115111591153

hydroxide) with a standardized sodium hydroxide so-lution (or a perchloric acid solution) at ionic strength of 0.1 M. The standard electrode potential E L and the calculated water dissociation constant (Kw) under the experimental conditions employed were used to calcu-late hydrogen ion concentration (pH _ logH&) from measured potentials. Stability constants were calculated from the potentiometric titration data with the SU-PERQUAD [22] algorithm.

3. Results and discussion

3.1. Proton complexes

3.1.1. Potentiometric studies

Cysteine possesses three ionizable protons: one in the carboxylic acid, another in the thiol group and the other in the ammonium group. The pK of the carboxyl group can be easily determined and identified because it is much more acid than the other groups. Inasmuch as the acid strengths of the thiol and ammonium groups are similar, it is possible that there is an overlap be-tween their dissociation. Thus, the macroscopic con-stants determined (Table 1) are composite and cannot be assigned to individual groups. Our potentiometric

Table 1Macroscopic dissociation constants (pKa) of cysteine and glutathione and complex stability constants for the As(III)Cys and As(III)GSH systems

(a) Cys (H3Cys)

pK1pK2pK3

2.44a8.78a10.71a

8.33b10.50b

1.88c8.15c10.29c

1.69(2)d8.17(1)d10.30(1)d

(b) As(III)Cys

Specieslog K

[As(H_1Cys)3]29.84(6)d

[As(H_2Cys)(OH)2]_12.01(9)d

(c) GSH (H4GS)

pK1pK2pK3pK42.60e3.82e9.16e9.88e3.59f8.75f9.65f2.04g3.54g8.54g9.42g2.10(3)d3.53(3)d8.65(2)d9.52(2)d(d) As(III)GSH

Specieslog K

[As(H_2GS)3]3_32.0(6)d

[As(H_3GS)(OH)2]2_10(3)d

Standard deviations are given in parentheses. a Ref. [27].b Ref. [23].

c Ref. [24]. d This work. e Ref. [27].f Ref. [28].g Ref. [26]. data are in a good agreement with other values re-ported in the literature [2325]. Clement and Hartz [25] also did spectrophotometric measurements aiming the calculation of the microscopic constants. They con-cluded that both the ionization of the thiol and the ammonium groups contribute to the second deproto-nation. Since the thiol is more acidic it makes a larger contribution.

The ionization pattern of glutathione is more com-plex than that of cysteine. Glutathione (c-L-glutamyl-L-cysteinylglycine) contains four ionizable protons: two in the carboxylic acids of the L-glutamyl and the glycyl groups, one in the L-cysteinyl sulfhydryl group and the other in the L-glutamyl ammonium group. Since the ionizations of both carboxylic acid protons are close together and so are the ionizations of the sulfhydryl and the ammonium groups, the macroconstants determined are composite of the microscopic constants for ioniza-tion from the individual groups. Our macroscopic con-stants (Table 1) are in accordance with those obtained by other authors [2628].

3.1.2. Spectroscopic measurements

The 1H NMR spectrum of cysteine at pH 7.0 consists of three doublets of doublets: one of them due to the a-methine proton at d 3.79, and the two others to the non-equivalent b-methylene protons at d 2.84 and 2.90. The resonance frequencies, their assignments and the cou-pling constants are shown in Table 2.

For GSH at pH 7, the two-cysteinyl methylene pro-tons give rise to closely spaced doublets of doublets at d 2.74 and 2.80. At pH 9.5 these doublets are shifted to-wards lower frequencies and are more separated, d 2.61 and 2.72. The deprotonation, which occurs mainly at the thiol group, produces a shielding eect on the protons. At pH 7.0, the cysteinyl a-methine proton is overlapped by the solvent signal around d 4.4, but at pH 9.5 it ap-pears as a doublet of doublets at d 4.09. At pH 7.0, the glycyl methylene protons give rise to a single signal at d 3.59, but at pH 9.5 the signal splits into two doublets at d 3.55 and 3.60. The assignments shown on Table 2 were confirmed by 2D COSY correlation (data not shown). The analysis of the 1H NMR spectrum of the glutathi-one molecule has already been reported [15,16,29] but the analysis presented here is more complete.

The ionization of the sulfhydryl proton can also be followed by ultraviolet absorption. Until pH 6.5, a so-lution of cysteine practically does not absorb in the in-vestigated region 220290 nm. From pH 7, the appearance of a broad band centered at 230 nm indi-cates the ionization of the thiol. As the pH is increased, this absorption, assigned to an n ! r_ transition of thiolate, is intensified and undergoes a bathochromic shift to 236 nm. Above pH 11, the absorbance is maxi-mal (e 3:95 _ 10 3 M _1 cm _1) indicating the complete deprotonation of the thiol. For glutathione the same1154N.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 11511159Table 2

1H NMR data for cysteine and glutathione in D2O at pD 7

dMultiplicityJ (Hz)

Cysteine

Cys Ha3.79ddHaHb(1) 5.60; HaHb(2) 4.20Cys Hb(1)2.90ddHb(1)Hb(2) 14.8; Hb(1)Ha 5.60Cys Hb(2)2.84ddHb(2)Hb(1) 14.8; Hb(2)Ha 4.20Glutathione

Glu Ha3.62apptGlu HaGlu Hb 6.30Glu Hb2.021.96mGlu Hc2.452.31mCys Ha###Cys Hb(1)2.80ddCys Hb(1)Cys Hb(2) 14.2

Cys Hb(1)Cys Ha 5.12Cys Hb(2)2.75ddCys Hb(2)Cys Hb(1) 14.2

Cys Hb(2)Cys Ha 7.24Gly Ha3.59s

holds, being the maximal value of e 5:80 _ 103 M_1 cm_1 at 232 nm.

3.2. Arsenic complexes

3.2.1. Potentiometric studies

Acidified mixtures of cysteine and sodium metaarse-nite were titrated with NaOH. Data ranging from pH 2.0 to 11.0, shown in Fig. 1, correspond to solutions with four metal-to-ligand molar ratios, 1:5, 1:2, 1:1 and 2:1. This set of data was used to calculate, simulta-neously, the acid dissociation constants of cysteine and the metalcomplex formation constants.

Ligand protonation, arsenic(III) hydrolysis and complexation can all be represented by the following general equation:

pH& qAsOH3& rH_2Cys2_&

Hp_r AsOH3_r qH_2Cys2_r &p_2r rH2O& 1

where p, q and r represent the stoichiometric coecients and H_2Cys2_ represents the completely deprotonated

cysteine. According to this notation, protonation of this form of cysteine leads to the species H_1Cys_, Cys, and HCys. The corresponding formation constants of these species are shown in Table 1.

The titration curves were divided in two sections, which were separately analysed: the first one from pH 2.0 to 8.4 and the second one from pH 8.5 to 11.0.

Up to pH 8.4, the best fit between experimental and calculated titration curves was attained assuming the formation of the complex [As(H_1Cys)3], where the carboxylic group is deprotonated, according to:

AsOH3 3Cys AsH_1Cys3& 3H2O

Above pH 8.5, calculations from potentiometric data lead to the proposition of another complex species [As(Cys)(OH)2]_, where Cys is completely deproto-nated. The following equation represents its formation AsOH3 H_1Cys_ AsH_2CysOH2&_ H2OIn these species, the p, q and r coecients of Eq. (1) are 6, 1 and 3 in the first case and 1, 1 and 1, in the second one. The species [As(H_1Cys)3] should be written more accurately as [As(HCys)3] and the species

Fig. 1. Titration curves of solutions containing: 1, [Cys] 10 mM and [As(III)] 2 mM; 2, [Cys] 10 mM and [As(III)] 5 mM; 3, [Cys] 5 mM and [As(III)] 5 mM; 4, [Cys] 5 mM and [As(III)] 10 mM at 25 LC (I 0:1 M) with a solution of NaOH (0.1 M).N.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 115111591155

[As(H_2Cys)(OH)2]_, as [As(Cys)(OH)2]_. Thus, during the formation of [As(Cys)(OH)2]_ from [As(HCys)3] the number of protons lost depends on the ligand excess. The formation constants of the arsenic(III) hydroxo complexes [As(OH)2O]_ and [As(OH)O2]2_ were deter-mined in a separate experiment and were fixed during the calculations of the equilibrium constants involving cysteine or glutathione.

The solution properties of arsenous acid resemble that of boric acid. From acid to neutral media, ar-senic(III) exists in solution mainly as As(OH)3. When a solution of arsenic at pH 7.0 is added to a solution of cysteine, also at pH 7.0, the pH of the resultant solution does not change after complexation. Thus, the forma-tion of the species [As(HCys)3] does not involve the liberation of protons to the medium. Evidently arsenic coordination to cysteine occurs concomitantly with the deprotonation of the thiol group and the binding of the dissociated proton to a hydroxyl group of As(OH)3, releasing a water molecule. The assignment of the co-ordination site was made from spectroscopic data and not from potentiometry.

Several other putative species such as [As(H-Cys)2(OH)] and its deprotonated forms were tested for the equilibrium model but the SUPERQUAD algorithm rejected all.

It should be emphasized that the same set of titration data was used to calculate both the acid dissociation constants of cysteine and the arseniccysteine complex formation constants. Furthermore, the pKa values of the ligand measured in the presence of arsenic are in agreement with the data given in the literature indicating the validity of the model. In the absence of arsenic, but keeping the same physicalchemical conditions, the es-timated pKa values are 1.69, 8.17 and 10.30 for the first, second and third deprotonation of HCys, respectively. These values may be compared with those estimated in the presence of arsenic. From the section of the titration

curves up to pH 8.4, we calculated the first and second pKa values of 1.69 and 8.16, respectively. From the data above pH 8.5, we obtained the values of 8.17 and 10.30 for the second and third pKa. All values agreed closely with those determined in the absence of arsenic. Fig. 2(a) shows the species distribution of the AsCys system calculated with the program SCECS [30]. The [As(HCys)3] species appears in the range of pH 1.58.5, competing with the As(OH)3 species. From pH 6.0, the [As(Cys)(OH)2]_ appears and predominates between pH 8.0 and 9.5. With the increase of the pH, the hydroxide ions exclude two cysteine molecules from the arsenic coordination sphere. Above pH 10, the formation of [As(OH)2O]_ leads to ligand dissociation from the ar-senic coordination sphere.

The co-ordination pattern of As(III) to glutathione is similar to that of cysteine, in spite of the complexity of glutathione molecule. Glutathione contains four ioniz-able protons instead of three in cysteine. At pH 7, glu-tathione is negatively charged, being deprotonated at the two carboxyl oxygens and protonated at the thiol and ammonium groups.

As done for cysteine, acidified mixtures of glutathione and sodium metaarsenite were titrated with NaOH. Data ranging from pH 2.0 to 8.4 were used to calculate, simultaneously, the first and second ligand acid disso-ciation constants and the metalloid complex formation constant. From the part of the titration curves above pH 8.5, the second and third ligand deprotonations and the metalloid-complex formation constant were calculated (data not shown). Despite the dierences between glu-tathione and cysteine, their interactions with arsenic follow the same coordination patterns and the conclu-sions will be presented concisely.

The best fit between experimental and calculated ti-tration curves was attained assuming the formation of two complexes [As(HGS)3]3 _ and [As(GS)(OH)2]2_. In the former, the ammonium group remains protonated

Fig. 2. Species distribution curves for the systems: (a) AsCys (1) As(OH)3; (2) As(OH)2O_; (3) As(OH)O22_; (4) [As(HCys)3]; (5) As(Cys)(OH)2]_; and (b) AsGSH (1) As(OH)3; (2) As(OH)2O_; (3) As(OH)O22_; (6) [As(HGS)3]3_; (7) [As(GS)(OH)2]2_. In both cases the total ligand concentration is 15 mM and the total As(III) concentration is 5 mM.1156N.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 11511159

and in the later, it is deprotonated. In both cases, GSH is deprotonated in the carboxylic groups and in the thiol. Once again, other species such as [As(HGS)2-(OH)]2 _ and its deprotonated forms were also included in the calculations but rejected by the computer pro-gram. The pKa values of GSH determined in the absence of arsenic (2.10, 3.53 and 8.65) are in good agreement with those determined in its presence (2.10, 3.59 and 8.68). The similarities with the AsCys system indicate that the binding site is the cysteinyl sulfur atom, in ac-cordance with the spectroscopic data. The species dis-tribution curves for this system were calculated by the SCECS program and are shown in Fig. 2(b).

3.3. Spectrometric studies

The interactions of As(III) and the two ligands were followed by UVvisible absorption and 1H nuclear magnetic resonance spectroscopies at two fixed pH values: 7.0 and 9.5.

1H NMR spectra were used to assign the binding site and follow the binding ability of the ligands studied.

1H NMR spectra of solutions containing 1.5 _ 10 _2 M of cysteine and various concentrations of As(III) at pH 7.0 have been recorded (Fig. 3). By increasing the concentration of As(III) the b-methylene doublets of doublets overlap, become broad and deshielded, indi-cating coordination to the sulfur atom. The a-methine proton is also deshielded but to a lesser extent. The broadening of the peaks probably results from the quadrupole moment of the 75As nucleus (I 3=2). The presence of quadrupolar nuclei can result in very eec-tive relaxation of any spins to which they couple. A plot of the methylene proton chemical shift in function of the arsenic concentration indicates a stoichiometry equal to 1:3 metalloid-to-ligand (inset Fig. 3). The curve reaches a plateau at arsenic-to-cysteine molar ratio equal to 2.

A similar experiment was carried out with GSH, and Fig. 4 shows the variation of the cysteinyl b-methylene resonance frequencies of solutions containing 1.5 _ 10_2 M of GSH and dierent concentrations of As(III), at pH 7.0. As the amount of As(III) is increased, all peak po-sitions are shifted towards higher frequencies, the most aected signals being those of the cysteinyl b-methylene protons, indicating coordination of arsenic via sulfhy-dryl group. A plot of the methylene chemical shift against metal-to-ligand molar ratio indicates a stoichi-ometry of 1:3.

At pH 9.5, arsenic addition causes a cysteinyl protons shift to higher frequencies indicating complexation (data not shown). At this pH value, it was necessary to add an excess of arsenic over the ligand in order to attain sat-uration. The results suggest a stoichiometry of 1:1.

In the UVvisible range, both ligands practically do not absorb when the thiol group is protonated. The coordination of arsenic(III) to Cys or to GSH strongly

Fig. 3. 1H NMR spectra of solutions containing 15 mM Cys and dierent As(III) concentrations. Metalloid-to-ligand ratios of 0; 0.2; 0.33; 1.0 and 2.0 in D2O (I 1:5 M) at pH 7.0. Inset: Methylene protons chemical shift against arsenic concentration.

modifies their spectra. As observed in the potentiometric study, the results are very similar for both ligands and only the spectrophotometric study of the interactions with glutathione is shown in Fig. 5. A 1.5 _ 10_2 M aqueous solution of GSH at pH 7.0 practically does not absorb in the range 250280 nm, indicating a low degree of ionization of the thiol group. In a 1.5 _ 10_2 M aqueous solution of GSH at pH 7.0, 97% of the ligand exists as H2GS_ and 3% as HGS2_. The addition of increasing concentrations of As(III) induces the ap-pearance of a shoulder around 280 nm, assigned to a sulfur-to-arsenic charge transfer transition. The satura-tion is attained at arsenic concentration equal to 3.2 _ 10_2 M. These data and also that obtained for theN.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 115111591157

Fig. 4. 1H NMR spectra of solutions containing 15 mM GSH and dierent As(III) concentrations showing the variation of the cysteinyl b-methylene protons resonance frequencies. Metalloid-to-ligand ratios of 0; 0.2; 0.33; 0.5; 1.0 and 3.0 in D2O (I 1:5 M) at pH 7.0. Inset: Cysteinyl b-methylene protons chemical shift against arsenic concentration.

As(III)cysteine system were treated by means of the SQUAD program, which searches for the best combi-nation of stability constants of the species to fit the data and simultaneously calculates the molar absorptivities based on the current value of bpqr . Not all stability constants and molar absorptivities relevant to the sys-tem need to be varied at once. The constants of ligand ionization and of arsenic hydrolysis determined sepa-rately by potentiometric titration were used and fixed

during the calculations. We tested several equilibrium models and the best fit allowed us to calculate the stability constants (log b) of the complex species [As(HGS)3]3_ as equal to 33.28 _ 0.01 and [As(HCys)3] as equal to 30.92 _ 0.02. The dierences with respect to that estimated from potentiometric experiments (Table 1) are 1.28 and 1.08, respectively. The estimated molar absorptivities of [As(HGS)3]3_ can be seen in the inset of Fig. 5(a). The molar absorptivities of [As(HCys)3] and Hcys_ calculated by the program are also in a good agreement with earlier data and are not shown.

As already said, As(III) in acid to neutral aqueous solution is already strongly complexed by hydroxide ions, existing as As(OH)3. Coordination to the SH group occurs without liberation of protons to the me-dium. Therefore, the three protons of the thiol groups bind to three hydroxyl groups liberating three water molecules, with subsequent binding of arsenic to thio-late. As a consequence, the calculated stability constant does not correspond to the formation of the complex from free arsenic(III) concentrations but from As(OH)3 concentration.

Similar experiments were carried out at pH 9.5 (Fig. 5(b)) with both cysteine and glutathione. As in the previous case, due to the similarity of the results, only those obtained for GSH are shown. In a 1.5 _ 10_2 M aqueous solution of GSH at pH 9.5, 6% of the ligand exists as H2GS_, 48% as HGS2_ and 46% as GS3_. As described for pH 7.0, As(III) addition induces the ap-pearance of a shoulder around 280 nm, assigned to a sulfur-to-arsenic charge transfer transition. However, the absorption spectra change further on adding an excess of As(III). A molar ratio of metalloid-to-ligand of 8:1 is required to attain saturation. The absorption data in the range 260320 nm were used in the calculation of the complex formation constant by means of the SQUAD program. The best fit between experimental and calculated spectra were obtained assuming the

Fig. 5. UVVis spectra of solutions containing 15 mM GSH and As(III) concentrations varying from 3 to 120 mM at 25 LC l 0:1 cm (I 1:5 M).

(a) pH 7.0. Inset: Molar absorptivities of the complex species [As(HGS)3]3_. (b) pH 9.5. Inset: Molar absorptivities of the complex species [As(GS)(OH)2]2_.1158N.A. Rey et al. / Journal of Inorganic Biochemistry 98 (2004) 11511159

Fig. 6. Proposed molecular structures of the As(III)GSH complexes.