Albumin Thiolate Anion Is an Intermediate in the Formation of Albumin-S–S-Homocysteine*

An elevated concentration of plasma total homocysteine is an independent risk factor for cardiovascular disease. Greater than 80% of circulating homocysteine is covalently bound to plasma protein by disulfide bonds. It is known that albumin combines with cysteine in circulation to form albumin-Cys 34 -S–S-Cys. Studies are now presented to show that the formation of albumin-bound homocysteine proceeds through the generation of an albumin thiolate anion. Incubation of human plasma with L - 35 S-homocysteine results in the association of > 90% of the protein-bound 35 S-homocysteine with albumin as shown by nonreduced SDS-polyacryl-amide gel electrophoresis. Treatment of the complex with (cid:1) -mercaptoethanol results in near quantitative release of the bound L - 35 S-homocysteine, demonstrating that the binding of homocysteine to albumin is through a disulfide bond. Furthermore, using an in vitro model system to study the mechanisms of this disulfide bond formation, we show that homocysteine binds to albumin in two steps. In the first step homocysteine rapidly displaces cysteine from albumin-Cys 34 thiols ( e.g. glutathione and cysteinylglycine) along with other metabolites ( e.g. nitric oxide) on Cys 34 ; however, the concentrations of these compounds were not determined in this study. Preparation of L - 35 S-Homocysteine— L - 35 S-Homocysteine was prepared from L -[ 35 S]methionine as described by Mudd et al. (25) with slight modifications. Briefly, 0.02 mmol of L -methionine was mixed with 1 nmol of L -[ 35 S]methionine (1 mCi) and refluxed with 5 ml of hydriodic acid for 18–20 h under argon atmosphere to produce L - 35 S-homocys-teine thiolactone. The solution was evaporated to dryness under flowing argon for 24 h. The resulting yellow oily reaction mixture was dissolved in 0.5 ml of water and subjected to descending paper chromatography (Whatman 3 MM), using the solvent system isopropyl alcohol:formic acid:water (70:10:20). The standards of methionine and homocysteine thiolactone were run adjacent to the reaction mixture. Before developing the chromatogram with ninhydrin to locate the amino acid stand- ards, the central portion of the chromatogram was removed. After developing the standard spots, it was repositioned, and the area corre- sponding to homocysteine thiolactone was marked, cut out, and eluted from the paper with water. The resulting aqueous solution was dried under vacuum and resuspended in water. The concentration of L - 35 S-homocysteine thiolactone was determined spectrophotometrically at 243 nm using a molar absorptivity of 2.50 M (cid:1) 1 cm (cid:1) 1 . The samples were then transferred to injector vials for automated HPLC analysis. Standard curves were generated with known amounts of cysteine and homocysteine to calculate the concentrations of the two thiols in the reaction mixture. Albumin concentration was determined by the bicinchoninic acid method

Homocysteine is a sulfur-containing amino acid formed during methionine metabolism (1). It is catabolized to cysteine through the transsulfuration pathway, or it may be remethylated back to methionine (2). An elevated level of plasma total homocysteine (tHcy) 1 is a strong independent risk factor for cardiovascular disease (3,4) and an emerging risk factor for Alzheimer's disease (5,6). tHcy is the sum of free homocysteine and protein-bound homocysteine. Free homocysteine is made up of reduced homocysteine (ϪSH) (Ͻ1% of tHcy), and low molecular weight oxidized disulfide (-S-S-) forms including homocystine (5-10% of tHcy) and homocysteine-cysteine mixed disulfide (5-10% of tHcy). Greater than 80% of tHcy in circulation is bound to protein by disulfide bonds (7)(8)(9). A small amount of homocysteine may also be bound to plasma proteins via amide linkage as a result of homocysteine thiolactone reacting with the ⑀-amino group of protein lysine residues (10). The upper limit of normal tHcy is Յ0.012 mM (11,12). However, in patients with homocystinuria, tHcy levels approach 0.5 mM (13). The overall in vitro binding capacity of human plasma proteins for homocysteine is Ͼ0.4 mM (14). Almost all pathophysiology studies utilize free reduced homocysteine (reviewed in Ref. 15), whereas little or no attention has been paid to protein-bound homocysteine, despite the fact that it is the most abundant form of circulating homocysteine both in normal and hyperhomocysteinemic subjects.
Albumin is the most abundant protein in plasma. Typical plasma concentrations range from 0.6 to 0.75 mM, and albumin makes up more than 50% of the total plasma protein (16). It is a nonglycosylated, single-chain polypeptide tightly folded into three domains that are structurally defined by 17 intrachain disulfide bonds formed between 34 cysteine residues. Albumin contains one additional cysteine residue at Cys 34 that does not participate in intrachain disulfide bonding. Albumin Cys 34 accounts for the bulk of free thiol (ϪSH) in plasma (17). The crystal structure of human serum albumin shows that Cys 34 is situated in a partially protected site in a seven residue turn between helices h2 and h3 of subdomain 1A (18) and sits in a crevice 9.5-10 Å deep (17).
The pK a of the thiol group of Cys 34 is abnormally low (ϳ5) (19). This is in contrast to the pK a of most of the low molecular weight aminothiols present in plasma. Thus, at physiological pH, albumin-Cys 34 exists primarily as thiolate anion and is highly reactive with metals, thiols, and disulfides (20). In fact, about one-third of the albumin molecules in the plasma carry disulfide-bonded thiols at this Cys 34 residue (20). These ligands probably become disulfide bonded in the plasma, because the albumin that is formed and secreted from the liver is in the free thiol form (17). Certain drugs containing thiol groups also bind to Cys 34 of albumin (19,21,22). Thus, Cys 34 of albumin seems * This work was supported by National Institutes of Health Grant RO1 HL 52234 (to D. W. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ to be the most probable binding site for low molecular weight thiols including homocysteine. In an earlier study where plasma proteins were resolved by gel filtration chromatography, it appeared that homocysteine was associated with albumin; however, the mechanism of homocysteinylation was not addressed (23). In this study we show that albumin is homocysteinylated when its thiolate anion attacks homocysteinecysteine mixed disulfide or homocystine.
Human Serum Albumin-Crystalline human serum albumin (Sigma; item number A-1653 and lot number 88H7610) was used in these studies. We determined that this albumin preparation contained 0.23 mol ϪSH/mol protein, 0.33 mol -S-S-cysteine/mol protein and 0.015 mol -S-S-homocysteine/mol protein. This human serum albumin had 1.5 mol fatty acids/mol albumin. The metal content of this albumin was also determined using inductively coupled plasma mass spectrometry (24). The samples were digested with nitric acid in polytetrafluoethylene test tubes with 71 Ga as an internal standard. The albumin was found to contain 3.62 ppm of copper, 191.5 ppm of calcium, 12.96 ppm of iron, 0.015 ppm of cobalt, and 0.35 ppm of nickel. Albumin is also known to carry other thiols (e.g. glutathione and cysteinylglycine) along with other metabolites (e.g. nitric oxide) on Cys 34 ; however, the concentrations of these compounds were not determined in this study.
Preparation of L- 35 S-Homocysteine-L- 35 S-Homocysteine was prepared from L-[ 35 S]methionine as described by Mudd et al. (25) with slight modifications. Briefly, 0.02 mmol of L-methionine was mixed with 1 nmol of L-[ 35 S]methionine (1 mCi) and refluxed with 5 ml of hydriodic acid for 18 -20 h under argon atmosphere to produce L-35 S-homocysteine thiolactone. The solution was evaporated to dryness under flowing argon for 24 h. The resulting yellow oily reaction mixture was dissolved in 0.5 ml of water and subjected to descending paper chromatography (Whatman 3 MM), using the solvent system isopropyl alcohol:formic acid:water (70: 10:20). The standards of methionine and homocysteine thiolactone were run adjacent to the reaction mixture. Before developing the chromatogram with ninhydrin to locate the amino acid standards, the central portion of the chromatogram was removed. After developing the standard spots, it was repositioned, and the area corresponding to homocysteine thiolactone was marked, cut out, and eluted from the paper with water. The resulting aqueous solution was dried under vacuum and resuspended in water. The concentration of L-35 Shomocysteine thiolactone was determined spectrophotometrically at 243 nm using a molar absorptivity of 2.50 M Ϫ1 cm Ϫ1 . L- 35 S-Homocysteine was prepared from the L-35 S-homocysteine thiolactone by the method of Duerre and Miller (26) with slight modification. Briefly, L-35 S-homocysteine thiolactone was hydrolyzed with NaOH (5 M) for 5 min at 37°C to open the thiolactone ring. The solution was then neutralized with 2 N HCl and diluted with TES buffer (0.05 mM, pH 7.4) to obtain the desired stock concentration of L-homocysteine. The concentration of L-homocysteine was determined by the method of Ellman (27). For nonradioactive experiments L-homocysteine was prepared from L-homocysteine thiolactone in the same manner.
Preparation of Homocysteine-Cysteine Mixed Disulfide-Homocysteine-cysteine mixed disulfide was prepared as described by Bü dy et al. (28). Briefly, homocysteine (20 mM) and cysteine (20 mM) were incubated in the presence of the catalyst, diaquocobinamide, in TES buffer (1 mM, pH 7.4) for 40 min. The reaction mixture was then separated by preparative paper chromatography using the solvent system isopropanol:formic acid:water (70: 10:20), and the band corresponding to homocysteine-cysteine mixed disulfide was cut from the paper and eluted with water.
Preparation of Albumin Thiolate Anion-Human albumin thiolate anion (mercaptalbumin) was prepared as described by Sogami et al. (29). Briefly, human serum albumin (1 mM) in 0.1 M sodium phosphate buffer (0.3 M NaCl, pH 6.86), was treated with dithiothreitol (final concentration, 5 mM) at 25°C for 45 min. It was then dialyzed extensively at 4°C against 0.05 M TES buffer, pH 7.2, to remove excess dithiothreitol. Using the method of Ellman (27), the dialyzed albumin contained 1 mol ϪSH/mol protein suggesting that, under these condi-tions, none of the 17 intrachain disulfide bonds of human albumin were reduced.
Binding of L- 35 S-Homocysteine to Human Plasma-Human plasma (0.1 ml) was diluted with 0.1 ml of 0.1 M TES buffer (pH 7.2) in a microcentrifuge tube and preincubated for 10 min at 37°C before the addition of L-35 S-homocysteine (final concentration, 0.5 mM). The reaction mixture was incubated at 37°C with continuous mixing for 5 h. Plasma proteins were then precipitated using 0.02 ml of 1.5 M perchloric acid. After centrifugation (10 min, 12,000 rpm), the protein pellet was washed three times with 0.1 ml of 1.5 M perchloric acid, and the pellet was dissolved in 0.1 ml of nonreducing SDS-polyacrylamide gel electrophoresis sample buffer (0.0625 M Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 0.02% bromphenol blue). To one half of the sample, 0.003 ml of ␤-mercaptoethanol was added and the sample was boiled for 5 min at 100°C to reduce disulfide bonds. Aliquots of the ␤-mercaptoethanoltreated and untreated samples (0.02 ml) were applied to a 10% SDSpolyacrylamide gel and electrophoresed according to the method of Laemmli (30). Gels were dried and analyzed by phosphorimaging to identify the bands corresponding to protein-bound homocysteine.
Binding of Homocysteine to Albumin-L-Homocysteine (final concentrations, 0.25-1 mM) was added to 0.75 mM human serum albumin in TES buffer (0.05 M, pH 7.2), and the reaction mixture was incubated at 37°C in a shaking water bath. Aliquots were withdrawn at various time points and added directly to tubes containing 0.1 ml of 1.5 M perchloric acid to precipitate albumin. The tubes were vortexed, incubated for 10 min on ice, and centrifuged for 10 min at 12,000 rpm. The protein pellet was washed three times with 0.10 ml of 1.5 M perchloric acid. The washed pellet was then solubilized in 0.10 ml of Tris buffer (0.5 M, pH 8.5), and the concentrations of albumin-bound thiols were estimated by HPLC, as described below. The perchloric acid soluble fraction was immediately stored at Ϫ20°C. The amount of total free thiol in this fraction was determined using the method of Ellman (27).
Quantification of Homocystine and Homocysteine-Cysteine Mixed Disulfide-To specifically determine the amount of homocystine and homocysteine-cysteine mixed disulfide formed during the reaction of human serum albumin with homocysteine, 0.75 mM human serum albumin was incubated with 0.5 mM 35 S-L-homocysteine. After 3 h of the reaction 50-l aliquots were withdrawn from the reaction mixture, and albumin was precipitated by adding 1.5 M perchloric acid. The supernatant was subjected to descending paper chromatography using the same conditions as mentioned above. The standards used were homocystine and homocysteine-cysteine mixed disulfide. The areas corresponding to the individual disulfides were cut from the paper and eluted with water, and their radioactivity was determined by counting in a liquid scintillation counter.
Reaction of Albumin Thiolate Anion with Low Molecular Weight Oxidized Thiols-Homocystine (0.125 mM), cystine (0.125 mM), homocysteine-cysteine mixed disulfide (0.25 mM), or a mixture of homocystine (0.0625 mM) and cystine (0.0625 mM) was added to 0.25 mM albumin thiolate anion at 37°C in a shaking water bath. Aliquots were withdrawn at various time points and added directly to tubes containing 0.1 ml of 1.5 M perchloric acid to precipitate the protein.
HPLC Determination of Thiols-Albumin-bound homocysteine and albumin-bound cysteine were determined by HPLC with fluorescence detection as described by Jacobsen et al. (31). Briefly, 0.1 ml of the solubilized albumin pellet (obtained after precipitating the reaction mixture with perchloric acid as mentioned above) was treated with 0.035 ml of 1.43 M sodium borohydride in 0.10 M sodium hydroxide followed immediately by the addition of 0.035 ml of 1.0 M HCl. After addition of 0.05 ml of 7 mM monobromobimane in 4 mM sodium EDTA (pH 7.0), the solution was incubated at 42°C for 12 min. Albumin was precipitated by the addition of 0.05 ml of 1.5 M perchloric acid. After centrifugation (12,000 rpm, 10 min), the supernatant was adjusted to pH 4 by the addition of 0.025 ml of 2.0 M Trizma® base. The samples (0.10 ml) were then transferred to injector vials for automated HPLC analysis. Standard curves were generated with known amounts of cysteine and homocysteine to calculate the concentrations of the two thiols in the reaction mixture. Albumin concentration was determined by the bicinchoninic acid method (32).

Identification of Albumin as a Binding
Protein for Homocysteine in Human Plasma-We recently determined the equilibrium binding capacity of plasma proteins for homocysteine but did not identify the specific proteins responsible for the binding (14). To identify specific homocysteine-binding proteins, human plasma was incubated with 35 S-L-homocysteine. Albumin was found to be the predominant (Ͼ90%) homocysteine-binding protein in plasma (Fig. 1). Treatment with ␤-mercaptoethanol resulted in the near quantitative removal of the bound homocysteine, indicating that the binding of homocysteine to albumin was through a disulfide linkage.
Binding of Homocysteine to Human Serum Albumin-Binding of homocysteine to human serum albumin was studied as a function of both time and concentration. Time course studies ( Fig. 2A) revealed that the binding of homocysteine to albumin increases with time. With concentrations of homocysteine greater than 0.1 mM, an equilibrium of homocysteine binding is reached in about 10 h. The formation of albumin-bound homocysteine also increases with increasing homocysteine concentrations (0.025-1 mM), and saturation of binding is achieved between 0.5 and 1 mM homocysteine. Concomitantly, the concentration of free reduced thiol in the medium decreased, and no free reduced thiol could be detected after 3 h of reaction (Fig.  2B), when 1 mM homocysteine was used.
As mentioned under "Experimental Procedures", the albumin used in these studies contained 0.33 mol cysteine/mol albumin. Interestingly, upon the addition of homocysteine, this cysteine was rapidly displaced from albumin. The amount of cysteine displaced from albumin-Cys 34 -S-S-Cys was dependent on the amount of homocysteine added (Fig. 3). The addition of 0.025 mM homocysteine reduced the albumin-Cys 34 -S-S-Cys concentration from 0.33 mol/mol albumin to 0.31 mol/mol albumin after 2 h, whereas treatment with 0.5 mM homocysteine reduced the concentration of bound cysteine to 0.047 mol/mol albumin (Fig. 3). After 2 h of reaction, the concentration of albumin-Cys 34 -S-S-Cys began to increase slowly until it reached equilibrium at 6 h where the final concentration of albumin-bound cysteine was 0.12 mol/mol albumin in the reaction using 0.5 mM homocysteine (Fig. 3).
Kinetics of Homocysteine Binding to Albumin-Additional studies on the homocysteinylation of albumin were carried out using 0.5 mM homocysteine because the binding of homocysteine to albumin saturates at about 0.5 mM (Fig. 2A). Moreover, in patients with homocystinuria, the tHcy concentration can approach 0.5 mM. The rates of reaction were studied using 0.5 mM homocysteine and 0.75 mM albumin. During the first 3 min of reaction, there was a rapid binding of homocysteine to albumin with a rate constant of 0.26 M Ϫ1 s Ϫ1 (Fig. 4A, inset). The rate of albumin-Cys 34 -S-S-Hcy formation then briefly plateaued, followed by a slower increase (0.045 M Ϫ1 s Ϫ1 ) over the next 10 h, after which it reached equilibrium (Fig. 4A). However, the formation of albumin-bound homocysteine did not equal the release of cysteine from albumin (Figs. 2 and 3). After 1 h of reaction, only about 0.067 mol of homocysteine was bound per mol albumin, whereas during the same time period 0.27 mol of cysteine was liberated per mol of albumin. To account for this difference, we hypothesized that albumin thiolate anion was being formed during the reaction. To test this, we measured the formation of albumin thiolate anion in the reaction mixture using the method of Ellman (Fig. 4B). We found that the formation of albumin thiolate anion followed a bell-shaped curve. It steadily increased from an initial concentration of 0.23 mol/mol albumin to 0.43 mol/mol albumin during the first hour of reaction and then decreased (Fig. 4B) as the concentration of albumin-Cys 34 -S-S-Hcy increased.
The homocysteine in the reaction and the cysteine released from albumin undergo oxidation to homocystine, cystine, and homocysteine-cysteine mixed disulfide and after 3 h of reaction, free reduced thiol could not be detected in the system (Fig. 2B).
To study the fate of homocysteine in the reaction mixture, we reacted 0.75 mM albumin with 0.5 mM L- 35 S-homocysteine and found that after 3 h of reaction, 0.14 mM homocystine and 0.095 mM homocysteine-cysteine mixed disulfide were formed in addition to the formation of 0.21 mol albumin-Cys 34 -S-S-Hcy per mol albumin (Fig. 4A). Because the formation of albumin- bound thiol continues to increase even in the absence of detectable free reduced thiol (Figs. 2 and 3), we hypothesized that the oxidized forms of homocysteine and cysteine were participating in the overall reaction. Therefore, we decided to test the reaction of albumin with homocystine. We incubated 0.75 mM albumin with 0.25 mM L-homocystine (equivalent to 0.5 mM homocysteine) and found that albumin did indeed react with homocystine (Fig. 5A), but the reaction rate was much slower compared to that with homocysteine. Moreover, homocystine did not displace cysteine from albumin-Cys 34 -S-S-Cys (data not shown). Thus, the reaction of homocystine with albumin alone could not explain the data shown above. Another possibility was that the albumin thiolate anion formed during the initial stage of the reaction could react with the oxidized forms of the thiols that were concurrently being produced during the course of the reaction.
To test this hypothesis, albumin thiolate anion was prepared by treating albumin with dithiothreitol (5 mol/mol albumin). Albumin thiolate anion was then reacted either with homocysteine or homocystine. With 0.5 mM homocysteine, only 0.02 mol albumin-Cys 34 -S-S-Hcy was formed/mol albumin after 1 h of reaction. However, when 0.25 mM homocystine was used, 0.12 mol of albumin-Cys 34 -S-S-Hcy was formed/mol albumin (Fig.  5B). The rate constant for the reaction with homocystine was 0.175 M Ϫ1 s Ϫ1 . This suggests that initially homocysteine reacts with albumin forming albumin thiolate anion, which then reacts with the oxidized form of the thiols to yield the final products.
To investigate whether trace metals associated with albumin might play a role in homocysteinylation, we treated albumin with the metal chelator DTPA. DTPA treatment and dialysis removed Ͼ97% of the metals from the albumin preparation as determined by inductively coupled plasma mass spectrometry. In the absence of trace metals, formation of homocystine or cystine by autooxidation was inhibited. As shown in Fig. 6, the initial rate of the reaction was almost identical when native and DTPA-treated albumin was used. However, with DTPAtreated albumin, the formation of albumin-Cys 34 -S-S-Hcy slowed considerably after the initial 30 min of reaction. This was in contrast to the reaction with native albumin, where the levels of albumin-Cys 34 -S-S-Hcy continuously increased up to 10 h. This suggests that, in the absence of homocystine, which is normally formed as a result of trace metal-catalyzed autooxidation of homocysteine, albumin thiolate anion lacks a suitable target for nucleophilic attack.
Reaction of Albumin Thiolate Anion with Cystine, Homocystine, and Homocysteine-Cysteine Mixed Disulfide-When albumin enters circulation it is likely that most, if not all of the Cys 34 is in the thiolate anion form (17). Once entering the circulation, albumin thiolate anion can react with either cystine, homocystine, or homocysteine-cysteine mixed disulfide. To mimic this situation in vitro, 0.25 mM albumin thiolate anion was reacted with 0.125 mM cystine or 0.125 mM homocystine. During the reaction of albumin thiolate anion with cystine, 0.69 mol of albumin-Cys 34 -S-S-Cys/mol of albumin was formed after 24 h (Fig. 7A). In the same time period 0.43 mol of albumin-Cys 34 -S-S-Hcy/mol albumin was obtained when homocystine was used (Fig. 7B). However, when both homocystine and cystine were present in equal concentrations (0.0625 mM) in the same mixture, albumin thiolate anion (0.25 mM) preferentially attacked homocystine to form albuminbound homocysteine (Fig. 7C). In fact, after 24 h of reaction, formation of albumin-bound homocysteine (0.5 mol/mol albumin) exceeded the formation of albumin-bound cysteine (0.19 mol/mol albumin) by about 2.6-fold (Fig. 7C). In homocystinurics, the concentration of oxidized homocystine is reported to be slightly greater than oxidized cystine (13). Thus, albumin thiolate anion reacts efficiently with cystine alone. However, when both homocystine and cystine are present in the reaction mixture, formation of albumin-bound homocysteine predominates. Similarly, when albumin thiolate anion was incubated with homocysteine-cysteine mixed disulfide, 0.518 mol of albumin-Cys 34 -S-S-Hcy/mol albumin was formed after 24 h of reaction, whereas during the same time only 0.175 mol of albumin-Cys 34 -S-S-Cys/mol of albumin was obtained (Fig. 7D). It is obvious that albumin thiolate anion preferentially attacks the sulfur atom of homocysteine in homocysteine-cysteine mixed disulfide.

DISCUSSION
Our studies using 35 S-L-homocysteine clearly indicate that homocysteine binds to albumin through a disulfide bond in human plasma. The binding is saturable, and maximal binding is found at 0.5 mM homocysteine. The formation of albumin- Cys 34 -S-S-Hcy appears to be biphasic. Within the first 3 min there is a rapid formation of albumin-Cys 34 -S-S-Hcy, followed by a short plateau (Fig. 4A, inset) that is then followed by the slow second phase of albumin-Cys 34 -S-S-Hcy formation. After 10 h, the reaction reaches equilibrium (Fig. 4A). Jakubowski (10) also found that albumin was homocysteinylated by homocysteine. In his study 35 S-L-homocysteine thiolactone was used, but a portion of the thiolactone undergoes hydrolysis to 35 S-L-homocysteine, which in turn can form albumin-bound homocysteine.
Mechanistically, homocysteine first displaces cysteine from albumin-Cys 34 -S-S-Cys, but this displacement can occur by two different pathways (Reactions 1 and 2). The homocysteine sulfhydryl group can react with the sulfur of albumin-bound cysteine forming homocysteine-cysteine mixed disulfide and free albumin thiolate anion (Reaction 1), and/or it can react with the sulfur of albumin Cys 34 , forming albumin-bound homocysteine and free cysteine thiolate anion (Reaction 2). Based on our experimental results, we propose that Reaction 1 predominates under the experimental conditions. The amount of homocysteine-cysteine mixed disulfide formed is also consistent with this observation.
The predominance of Reaction 1 can be explained by the pK a values of the two thiolate anions formed during Reaction 1 (albumin thiolate anion) and Reaction 2 (cysteine thiolate anion). The pK a of albumin thiolate anion is much lower (ϳ5) than cysteine thiolate anion (8.15) (33). Thus, at pH 7.4 albumin thiolate anion will be much more thermodynamically stable than cysteine thiolate anion, favoring Reaction 1. Moreover, as proposed by Christadoulou et al. (34,35), albumin thiolate anion is further stabilized by forming a salt bridge with His 39 .
These results are in agreement with the in vivo observations of Mansoor et al. (36), who found that after intravenous admin-istration of homocysteine into healthy individuals, there was a decrease in protein-bound cysteine. They also observed that the decrease in protein-bound cysteine exceeded the increase in protein-bound homocysteine. Our proposed mechanism explains these observations. In the initial phase of the reaction, homocysteine preferentially attacks the cysteine sulfur of albumin-Cys 34 -S-S-Cys, generating albumin thiolate anion and homocysteine-cysteine mixed disulfide (Reaction 1).
In the second phase of the reaction, albumin thiolate anion attacks homocysteine-cysteine mixed disulfide preferentially on the homocysteine sulfur (Reaction 3) to form albumin-bound homocysteine and cysteine thiolate anion.
Albumin thiolate anion can also attack the cysteine sulfur of homocysteine-cysteine mixed disulfide to form albumin-bound cysteine and homocysteine thiolate anion (Reaction 4).
However, because the pK a of free homocysteine (8.7) 2 is higher than that of free cysteine (8.15) (33), cysteine thiolate anion FIG. 8. Buried and exposed conformations of albumin-Cys 34 . Cys 34 exists in buried and exposed conformations as proposed by Christodoulou et al. (35). When albumin-Cys 34 thiolate anion (exposed) attacks the mixed disulfide, the homocysteinylated product is stabilized in the exposed conformation. This figure is modified from Christadoulou et al. (35). REACTION 3 REACTION 4 Mechanism of Albumin-S-S-Homocysteine Formation (formed in Reaction 3) will be thermodynamically more stable at pH 7.4 than homocysteine thiolate anion (formed in Reaction 4). Thus, as shown in Fig. 7D, the formation of albumin-bound homocysteine (Reaction 3) is preferred over formation of albumin-bound cysteine (Reaction 4). This was also confirmed by the fact that the amount of albumin-bound cysteine increased by only 0.05 mol/mol albumin after 24 h of reaction (Fig. 3A). Moreover, it has been reported that when a thiolate anion attacks an unsymmetrical disulfide, the thiol that leaves will be the one having the lowest pK a (37). Thus, we conclude that when treated with mixed disulfide, albumin thiolate anion preferentially attacks the sulfur of homocysteine in the mixed disulfide.
In addition to reacting with homocysteine-cysteine mixed disulfide, albumin thiolate anion also reacts with homocystine and cystine. Reaction with homocystine results in the formation of albumin-bound homocysteine (Reaction 5), while reaction with cystine will lead to the formation of albumin-bound cysteine (Reaction 6). Cys 34 is situated in a pocket consisting of 4 amino acids and is protected from solvent in a crevice. Our results fit the model proposed by Christodoulou et al. (34,35), where the Cys 34 exists in exposed and buried forms (Fig. 8). The thiolate anion form of Cys 34 is primarily in the buried form and is in close proximity to His 39 with which it can form a salt bridge for stabilization. Disulfide bond formation of Cys 34 with homocysteine (or cysteine) would lead to the stabilization of the exposed form. Thus, when the exposed thiolate anion undergoes a thiol/ disulfide exchange reaction, it forces the equilibrium to shift to the exposed form (34). In this study, the majority of Cys 34 in the native albumin preparation was oxidized (only 23% was determined to be in the reduced form) and, thus, predominantly in the exposed form. Upon the addition of homocysteine, Cys 34bound cysteine is removed, resulting in the probable conversion of the thiolate anion to the buried conformation. Further reaction of homocysteine with the thiolate anion is less facile because the interaction between ionic charges controls entry into the crevice. This accounts for the fact that once the thiolate anion is formed, homocysteine-cysteine mixed disulfide (or homocystine) reacts with exposed Cys 34 much more efficiently than homocysteine, and after the mixed disulfide (or homocystine) reacts with the thiolate anion, the equilibrium shifts toward the exposed form.
Does albumin-bound homocysteine play a role in the pathogenesis of vascular disease? Because albumin is known to bind to at least three endothelial cell membrane proteins (39) and is either transcytosed via plasmalemmal vesicles (gp 60) or endocytosed via receptor-mediated endocytosis (gp30, gp18) (40,41), we propose that homocysteine can be delivered to cells via its association with albumin. The proteins, gp30 and gp18, bind to conformationally modified albumin (e.g. albumin-gold conjugates) and behave like scavenger receptors to internalize and degrade albumin (41,42). The major binding site on albumin for gold is known to be Cys 34 (35). Thus, we propose that binding of homocysteine to albumin via Cys 34 may result in the same conformational modification that results when gold binds to albumin. Albumin-bound homocysteine could then be internalized by endothelial cells, degraded in lysosomes, and released to the cytosol, where it could alter the intracellular redox potential or modify intracellular proteins resulting in endothelial dysfunction. Studies addressing the role of albumin-bound homocysteine on endothelial cell function are currently in progress.