Catalysis of S-nitrosothiols formation by serum albumin: The ...

Catalysis of S-nitrosothiols formation by serum albumin: The mechanism and implication in vascular control Olga Rafikova, Ruslan Rafikov, and Evgeny Nudler* Department of Biochemistry, New York University Medical Center, New York, NY 10016

O䡠 is synthesized by various types of cells and involved in numerous biological functions, including vasodilation, platelet aggregation, neurotransmission, and inflammation (1, 2). Heme proteins such as guanylyl cyclase and free radical species—e.g., 䡠 (䡠O⫺ 2 )—serve as NO primary targets (1–3). Another physiologically significant component of NO䡠 biochemistry involves the formation of thionitrite esters with cysteine (Cys) or Cys residues (Snitrosothiols; RS-NO). Low-molecular-weight (LMW) RS-NO (e.g., S-nitrosoglutathione) and nitroso-derivatives of proteins such as albumin and hemoglobin exert NO-like activity in vivo. They cause arterial and venous smooth muscle relaxation, inhibit platelet aggregation, and activate guanylyl cyclase (4–8). Vasoactive S-nitrosothiols are known to be generated in vivo (6–10), although the actual amount of these species in circulation is debated (refs. 11 and 12, and references therein). Because RS-NO are relatively stable and can release NO䡠 when required, via reactions with transition metal ions or other reducing agents (13–15), they are envisioned as a buffering system that controls intra- and extracellular activities of NO䡠 and magnify the range of its action. Once formed, circulating RS-NO can deliver NO into cytosol via specific mechanisms (16) or directly transfer the nitrosyl cation (NO⫹) to another thiol via the so-called transnitrosation reaction that ensures the dynamic state of RS-NO in vivo (7, 17). S-nitrosylation of protein free Cys residues modulates activities of various regulatory factors and enzymes and represents a widespread signaling mechanism (refs. 18 and 19, and references therein). Despite the growing interest in the role of nitrosothiols in biological systems, there is still uncertainty about how they form in vivo. In the absence of an electron acceptor, NO䡠 is unable to react with nucleophiles under oxygen free conditions, implying that metabolites of NO䡠 oxidation, such as N2O3, are actual nitrosating agents (20–23). However, considering the low concentration and short life span of NO䡠 in vivo, the third-order reaction of NO䡠 with O2 (Eq. 1) (k ⬇ 4 ⫻ 106 M⫺2䡠sec⫺1) (20, 21) seems to be too slow to account for any detectable amount of circulating RS-NO.

N

2NO䡠 ⫹ O2 3 2NO2䡠

[1]

NO2䡠 ⫹ NO䡠 ^ N2O3

[2]

N2O3 ⫹ H2O ^ NO⫹ ⫺ NO⫺ 2 ⫹ H2O 3 2HNO2

[3]

www.pnas.org兾cgi兾doi兾10.1073兾pnas.092048999

High instability of N2O3 in aqueous solution (Eq. 3) further supports this notion. These apparent theoretical constraints can be resolved, however, owing to remarkable properties of NO䡠 in multiphase systems (19, 24). NO䡠 and O2 are both hydrophobic molecules, and areas of high hydrophobicity should act to increase their local concentration by sequestering them from the surrounding aqueous phase. Under aerobic conditions, high local concentrations of NO䡠 and O2 in the hydrophobic phase—e.g., within lipid membranes or protein interiors—can significantly accelerate NO䡠 oxidation and N2O3 formation (19, 25, 26). In other work, we have shown that such micellar catalysis of NO䡠 oxidation can be efficiently mediated by serum albumin (19). Albumin is the most abundant transport and depot protein in the mammalian vasculature. Considering the large (⬇0.8 mM) concentration of albumin in plasma, we assumed that its hydrophobic core can be the major absorber of free NO䡠 and catalyst of N2O3 (19). Here we designed experiments to estimate the impact of albumin-mediated catalysis of NO䡠 oxidation on the pool of circulating LMW RS-NO. We suggest the mechanism by which NO⫹ is generated in the albumin hydrophobic interior and transferred to various LMW thiols (RSH), and show how this mechanism can control the vascular tone in mammals. Experimental Procedures Chemicals and Reagents. NO䡠兾H2O solution was prepared as in ref.

19. BSA (Sigma) was ⬎99% pure and free of fatty acids and globulins. BSAN2O3 was prepared by adding 3.5 ml of aqueous NO䡠 solution to 10 ml of BSA (17 ␮M) in 30 mM Tris䡠HCl (pH 7.9) so that the concentration of NO䡠 was ⬇100 ␮M. As soon as the NO䡠 concentration decreased from 100 ␮M to less than 1 ␮M, 1 ml of 3% ammonium sulfamate was added to remove nitrite. WBSA and CWBSA were prepared by treating BSA (1 mM) or carboxymethylated BSA (CBSA, 1 mM, Sigma) with 5 mM Nbromosuccinimide (Aldrich), which selectively modifies Trp residues, for 10 min at 25°C in NaCH3COO兾CH3COOH (pH 5.0), followed by dialysis (MW cut off 12–14.000 membrane) in 2 liters of 20 mM Tris䡠HCl (pH 7.8), 50 mM NaCl, 0.1 mM EDTA, 1 mM C2H5SH, and 5% glycerol. Inability of modified Trp-214 and Cys-34 of CWBSA to undergo nitrosation was evaluated spectrophotometrically at 330 nm (Fig. 3B). Specifically, 0.2 mM CWBSA or BSA (20 ␮l) was treated with 0.1 M NaNO2⫹0.1 M HCl for 10 min, followed by addition of ammonium sulfamate to 3% and 80 mM Tris䡠HCl (pH 7.9) to remove HNO2. UV-visible spectra were recorded by using an Ultrospec 3000 spectrophotometer (Pharmacia). Spectra were digitized and analyzed by WIN DIG and ORIGIN 6.0 software (Microcal, Northampton, MA). This paper was submitted directly (Track II) to the PNAS office. Abbreviations: GSH, glutathione; RSH, thiols; LMW, low-molecular-weight; MAP, mean arterial pressure. *To whom reprint requests should be addressed. E-mail: [email protected]. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9 兩 5913–5918

BIOCHEMISTRY

Nitric oxide (NO䡠) is a short-lived physiological messenger. Its various biological activities can be preserved in a more stable form of S-nitrosothiols (RS-NO). Here we demonstrate that at physiological NO䡠 concentrations, plasma albumin becomes saturated with NO䡠 and accelerates formation of low-molecular-weight (LMW) RS-NO in vitro and in vivo. The mechanism involves micellar catalysis of NO䡠 oxidation in the albumin hydrophobic core and specific transfer of NOⴙ to LMW thiols. Albumin-mediated S-nitrosylation and its vasodilatory effect directly depend on the concentration of circulating LMW thiols. Results suggest that the hydrophobic phase formed by albumin serves as a major reservoir of NO䡠 and its reactive oxides and controls the dynamics of NO䡠-dependant processes in the vasculature.

PHARMACOLOGY

Edited by Louis J. Ignarro, University of California School of Medicine, Los Angeles, CA, and approved March 5, 2002 (received for review January 28, 2002)

a buffer in the mixture with pH 4.7. The solution remained transparent during the experiment, indicating that no significant protein denaturation and aggregation occurred. Determination of Thiols, S-Nitrosothiols, and Nitrite. Total level of

Fig. 1. Micellar catalysis of NO䡠 autooxidation by serum albumin. (A) The model. If partitioning (Q) of NO䡠 and O2 in the albumin兾H2O system is ⬎1, the protein hydrophobic core (dark gray) would increase the local concentration of both gases by sequestering them from the aqueous phase and accelerate NO䡠 autooxidation and N2O3 formation. (B) Determination of QNO for albumin. On top, the equation and parameters used for QNO calculation: [NO]H2O, NO䡠 concentration in the aqueous phase; [NO]h, NO䡠 concentration in the hydrophobic phase; [NO], NO䡠 concentration in the aqueous solution of albumin (4.8 ␮M); V, volume of the sample (10 ml); Vh, volume of the hydrophobic phase (⬇28 ⫻ 10⫺3 ml, as determined by WEBLAB VIEWER PRO 3.5 software that calculates an approximate volume of hydrophobic compartments of a protein with known atomic structure). ⌬ ⫽ [NO]H2O ⫺ NO is determined experimentally. Infusion of 100 mg of BSA into 10 ml of NO䡠 solution under O2-free argon conditions results in decreasing [NO]H2O from 6.5 ␮M (bar 1) to 4.8 ␮M (bar 2) (⌬ ⫽ 1.7 ␮m) as detected electrochemically. Similar ⌬ values were obtained with rat plasma (10 mg兾ml) (bar 3) and chemically modified BSA, CWBSA, in which Cys-34 and Trp-214 were covalently blocked (see Experimental Procedures) and resistant to nitrosation (bar 4). Because Vh was not determined experimentally but estimated using computer software, actual QNO for BSA may be subject to some variation. (C) Time course of nitrite formation in the absence (black dots) or presence of 0.8 mM BSA (squares) or CWBSA (triangles). Nitrite concentration, [NO2⫺], was measured by Griess reagent immediately after addition of NO䡠 to 25 ␮M. (D) Effect of proteolysis on albumin-mediated nitrite formation. At indicated intervals, proteinase K (0.25 mg兾ml; Sigma) was added to the NO䡠-treated BSA solution (0.2 mM). In each case, rapid increase of [NO2⫺] occurred (arrows) as measured by Griess reagent 1 min after the protease challenge. Each time point was processed as an independent experiment. The difference in [NO2⫺] between protease-treated and nontreated BSAN2O3 was plotted (squares) and used for calculation of the ‘‘half-life’’ of N2O3 (t1/2 ⬃ 7 min) in BSA. Schematic interpretation of results is shown at the bottom. t1/2(N2O3) is much longer in the albumin interior than in the aqueous phase. Enzymatic disruption of the protein hydrophobic core instantly releases N2O3 into the aqueous phase, resulting in eruption of nitrite.

Determination of NO䡠 Partitioning (Q). To determine QNO for albu-

min, the ice-cold calibrating solution (10 ml, 0.1 M H2SO4 ⫹ 0.1 M KI) was degassed in the air-tight device by bubbling with argon for 40 min. BSA, CWBSA (0.1g dry powder), or rat plasma (0.1 g dry powder, Sigma) was added to the solution and dissolved in the presence of argon. To minimize protein denaturation, the mixture was prepared initially at 0°C. Temperature was then adjusted to 20°C and the ISO-NO Mark II electrode (WPI Instruments, Waltham, MA) was inserted into the solution under argon and sealed. Released NO䡠 was detected on injection of 0.4 ml of 50 mM NaNO2. QNO is calculated as described in Fig. 1. Albumin serves as 5914 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.092048999

thiols in plasma was determined by Ellman’s reagent (27). Plasma glutathione (GSH) was detected at 230 nm by HPLC (Waters) by using C18 column with a mobile phase of 99% 0.1 M monochloroacetate (pH 3) and 1% methanol. Total RS-NO was determined in vitro or in plasma by using CuCl2 (1.5 mM) to displace NO䡠 from thiol residues. Released NO䡠 was detected electrochemically with the NO electrode. GS-NO was used as a standard. It was prepared by incubating equimolar amounts of GSH and NaNO2 in acidified water on ice. LMW RS-NO in plasma was determined by incubating fresh plasma with CuCl2 for 3 min while conducting electrochemical detection of NO䡠. To prepare plasma samples (100 ␮l), 0.5 ml of arterial blood was collected into heparinized plastic tubes and centrifuged at 4,500 rpm for 3 min. Plasma was processed immediately after preparation. In the experiment of Fig. 5A, 20 ␮l of GSH (100 mM) in 80 mM Tris䡠HCl (pH 7.9) and 10 ␮l of CuCl2 (100 mM) were added to 100 ␮l of plasma, followed by a detection of the level of liberating NO䡠 with the NO electrode. Nitrite was detected with the Griess reagent kit at 540 nm as specified by the manufacturer (Cayman Chemical, Ann Arbor, MI). In the case of Fig. 1C, NO䡠 (25 ␮M) was added to 0.8 mM BSA or CWBSA in 80 mM Tris䡠HCl (pH 7.9). In the case of Fig. 1D, NO䡠 (25 ␮M) was added to 0.2 mM BSA in 80 mM Tris䡠HCl. After indicated time intervals, proteinase K (0.25 mg兾ml; Sigma) was added for 1 min at 36°C. Animals. Adult male Wistar rats (300–400 g) were used. Both femoral arteries and left femoral vein of anesthetized animals (urethane 1.2 g兾kg i.p.) were cannulated using polyethylene tubing (PE-50, WPI). Arterial catheter was used for continuous monitoring of aorta blood pressure and heart rate, and for withdrawing blood samples. Venous catheter was used for drug administration. The same infusion rate of drugs or saline (0.2 ml兾min) was applied for all animals. Drugs were used as follows: BSA, BSAN2O3, or CMBSA, 0.05 g兾ml; saline, 0.9% NaCl; N-nitro-L-arginine methyl ester (L-NAME), 20 mg兾kg; GSH, 1 ml to the final concentration 5 mM, 12 mM, or 20 mM, based on the total blood volume (Fig. 5C). Before drug administration, mean arterial pressure (MAP) was recorded and the initial level of RS-NO was determined in plasma as described above. MAP and the heart rate were detected with the pressure sensor connected to the amplifier. The signal was digitally converted at the PowerLab 200 workstation (A. D. Instruments, Milford, MA) and transmitted to PC.

Results Acceleration of NO䡠 Oxidation by Serum Albumin: Quantitative Assessments. Our previous study suggests that NO䡠 oxidation occurs much

faster in the hydrophobic interior of albumin than in surrounding aqueous media, leading to effective nitrosylation of the only free thiol, Cys-34, and the only Trp-214 residues of the protein (19). To calculate the maximum acceleration value (H) of NO䡠 oxidation by the hydrophobic phase of albumin, we first determined the partition coefficient (Q) of NO䡠 for the albumin兾H2O system (Fig. 1). For the 2 third-order reaction of NO䡠 with O2 (1), H ⫽ QNO ⫻ QO2—i.e., acceleration of NO䡠 oxidation is primarily determined by QNO. To calculate QNO, which is described by the equation shown in Fig. 1B, we used a Clark-type NO䡠-electrode attached to an airtight gas兾 liquid mixing device to determine changes of the NO䡠 concentration (⌬) upon addition of the BSA under O2-free conditions (Fig. 1B). ⫺ ⫹ 䡠 NO䡠 was generated in the reaction: NO⫺ 2 ⫹ I ⫹ 2H 3 NO ⫹ 1兾2I2 ⫹ H2O, as described in Experimental Procedures. QNO(BSA) appears to be ⬇120 (Fig. 1B), implying that albumin accelerates the reaction in Eq. 1 at least 1202 or 1.4 ⫻ 104 times. A similar ⌬ value was obtained for rat plasma (Fig. 1B), suggesting that the hydroRafikova et al.

PHARMACOLOGY

phobic phase formed by albumin is a powerful absorbent of free NO䡠 and a catalyst of its oxidation in the mammalian vasculature. To provide independent evidence for albumin-mediated NO䡠 absorption and to estimate the half-life (t1/2) of N2O3 in the protein interior, we used enzymatic proteolysis as a tool for rapid disruption of the protein hydrophobic core and release of N2O3 into the aqueous phase (Fig. 1 C and D). In this and other in vitro studies we used a freshly prepared oxygen-free water solution of NO䡠. Addition of NO䡠 to an aerobic BSA solution resulted in nitrite formation, which accumulated much slower than in the control, without BSA (Fig. 1C). This result suggests that a significant proportion of N2O3 was trapped by albumin and reacted with water only upon slow Rafikova et al.

release. Addition of proteinase K to the BSA probe shortly after it was treated with NO䡠 caused an immediate nitrite accumulation, almost to the plateau level (Fig. 1D), indicating that most of the trapped N2O3 was abruptly released into the aqueous phase. As judged by SDS兾PAGE analysis, more than 99% of BSA was degraded within 1 min of proteolysis (not shown) thus eliminating most, if not all, protein hydrophobic pockets. The burst of nitrite in response to the protease declined with time so that after 40 min no detectable nitrite accumulation was observed upon protease challenge. At this time, the overall level of nitrite had already reached its plateau, suggesting that no N2O3 remained inside the protein. Based on these data, we estimate t1/2 (N2O3) in BSA to be ⬇7 min (Fig. 1D). Such a dramatic stabilization of N2O3 by albumin may be due to its potentially high Q value. Additionally, slow diffusion N2O3 within the albumin interior can be attributed to formation of ␲ and ␴ complexes with aromatic amino acid residues and nucleophiles. Addition of Cu⫹⫹ to 1.5 mM, which induces cleavage of S-NO PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9 兩 5915

BIOCHEMISTRY

Fig. 2. Generation of LMW RS-NO by albuminN2O3. (A) BSAN2O3-mediated Cys-NO formation as a function of Cys concentration. (Left) A representative NO䡠-electrode tracing of Cu⫹⫹-dependant NO䡠 production after addition of 1 mM Cys to 17 ␮M BSAN2O3 (‘‘BSA’’ trace) or just Tris䡠HCl buffer (80 mM, pH 7.9) (‘‘buffer’’ trace). Cu⫹⫹ was added as the NO䡠 concentration in the probe reached 1 ␮M after initial 100 ␮M. (B) BSAN2O3-mediated GS-NO formation as a function of GSH concentration. (Left) Experiment is the same as in A, except that 2 mM GSH was used instead of Cys. (C) BSAN2O3-mediated GS-NO formation as a function of BSA concentration (experimental design as in B). (D) BSAN2O3-mediated GS-NO formation as a function of time (experimental design as in B, except that 30 ␮M GSH was added after indicated time intervals). (E) Effect of proteolysis on BSAN2O3-mediated GS-NO formation [experimental design as in B, except that proteinase K (0.25 mg兾ml) was added before the addition of NO䡠 to BSA (Left), or after that, when the NO䡠 concentration dropped to ⬇1 ␮M (Right)].

Fig. 3. Mechanism of albumin-mediated catalysis of LMW RS-NO. (A) The model. AlbuminN2O3 can transfer the nitrosonium cation (NO⫹) to LMW RSH via three pathways: (I) directly, (II) via Cys-34, and (III) via Trp-214. (B) Blockage of albumin Cys-34 and Trp-214 nitrosation by chemical modification. UV-visible absorption spectra of newly generated chromophore during the reaction of BSA or CWBSA with 0.1 M NaNO2兾HCl. D, absorbance (OD units) with the maximum at ␭ ⫽ 330 nm for N-Trp-NO and 346 nm for S-Cys-NO (19). (C) Role of albumin Cys-34 and Trp-214 in catalysis of LMW RS-NO. Chemically modified BSA that carry S-carboxymethylated Cys-34 (CBSA), N-oxidized Trp-214 (WBSA), or both (CWBSA) (see Experimental Procedures) were used in the experiment similar to that described in Fig. 2B.

Fig. 4. Effect of albumin and albuminN2O3 on MAP of anaesthetized rats. (A) Sample tracing (top) and time course (bottom) of the hemodynamic effect of BSA (light gray column) or CWBSA (black column). Infusion of 0.12 g兾kg albumin produced a steep increase in blood pressure followed by a gradual and prolong asymptotic recovery. Mean ⫾ SE from ten independent experiments (see Experimental Procedures). (B) Sample tracings (top) and time course (bottom) of the hemodynamic effect of CWBSAN2O3 administered after infusion of L-NAME (light gray column) or without L-NAME (control). CWBSAN2O3 was prepared as in Fig. 2B and infused at 0.12 g兾kg. L-NAME (20 mg兾kg) causes prolonged hypertension (compare left and right tracings). In this case, CWBSAN2O3 induces a rapid fall in MAP followed by a gradual asymptotic return to the L-NAME-dependent base line (light gray column). If saline is used instead of L-NAME (dark gray column), CWBSAN2O3 does not change MAP significantly. Mean ⫾ SE from ten independent experiments. (C) Schematics explaining hemodynamic effects of albumin and albuminN2O3. Exogenous albumin (open circle) rapidly absorbs free NO䡠 thus causing vasoconstriction (left drawing). On saturation with NO䡠, albumin generates NO⫹ via micellar catalysis of NO䡠 oxidation (filled circles). NO⫹ then acts as a vasodilator to recover MAP. Exogenous albuminN2O3 does not change the NO䡠兾NO⫹ balance and thus has no hemodynamic effect (middle drawing). Inhibition of NOS by L-NAME exhausts plasma albumin with NO⫹ causing vasoconstriction (right drawing). In this case, exogenous albuminN2O3 works as a vasodilator by supplying NO⫹.

bond, to BSA 40 min after it was treated with NO䡠 results in ⬇800 nM NO䡠 release as detected electrochemically (see Experimental Procedures), indicating that 5–6% of BSA was S-nitrosated during the experiment. The only free thiol of albumin, Cys-34, and Trp-214 were nitrosated as confirmed by spectroscopic measurements (see Fig. 3B). Other amino acid residues undergo nitrosation only to a ⫹ small extent under these conditions (19, 28). Indeed, NO⫺ 2 兾H , 䡠 which is a more aggressive nitrosating agent than NO , at 1:1 ratio to BSA generated 0.6 nitrosating amino acid residues per BSA molecule (28). Thus, even theoretically, if all NO䡠 would act as ⫹ NO⫺ 2 兾H (in our experiment it was not the case as the pH was neutral), we could expect no more than one to two nitrosated residues per BSA molecule (most likely Cys and Trp, because they are the strongest nucleophiles). Subsequently, we will refer to albumin saturated with NO䡠兾N2O3 as BSAN2O3 and the fraction that underwent nitrosation as NOBSAN2O3 or NO-BSA. Albumin Is a Catalyst of LMW RS-NO Formation. S- and N-nitrosation of albumin with NO䡠 is much more efficient than that of LMW thiols or amines (19). This phenomenon is readily explained by the ability of albumin to catalyze and preserve nitrosating species, N2O3, in its own hydrophobic compartments (ref. 19; Fig. 1). We reasoned that albumin may also be the micellar catalyst of nitrosation of external molecules such as LMW RSH. To test this hypothesis, we compared the rate of LMW RS-NO formation in the presence or absence of albumin. We chose two bioactive LMW RSH: GSH and L-cysteine (Cys) (Fig. 2). Nitrosothiols were detected by employing Cu⫹⫹, followed by electrochemical detection of released NO䡠 (see Experimental Procedures). Each experiment began with generation of BSAN2O3, that is saturation of BSA with NO䡠 under aerobic conditions by using water solution of NO䡠 (100 ␮M). As soon as the NO䡠 concentration in the BSA solution or control probe dropped from 100 ␮M to ⬇1 ␮M, GSH or Cys were added. In each case, the presence of BSA had a great stimulating effect on the rate and efficiency of LMW RS-NO formation (Fig. 2 A and B). The ability of BSAN2O3 to generate GS-NO declined with time (Fig. 2D). The rate of decline (t1/2(activity) ⬃ 10 min) was close to what one can expect from the estimated ‘‘half-life’’ of N2O3 in albumin (Fig. 1D). The efficiency of albumin-mediated S-nitrosation also depended on the amount of albumin in the probe. The maximum effect was 5916 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.092048999

observed at ⬇25 ␮M BSA (Fig. 2C). This fact is readily explained in terms of micellar catalysis of NO䡠 oxidation (24) when the relatively small ‘‘optimal’’ volume of the hydrophobic phase (in this case albumin) generates the maximum acceleration of the reaction. Mechanism of Albumin-Mediated Catalysis of LMW RS-NO Formation.

To confirm that micellar catalysis of 䡠NO oxidation is responsible for observed results, we studied the effect of proteinase K on albumin-mediated GSH nitrosation. As described above, proteinase K rapidly converts albumin into short peptides incapable of forming appropriate micelles. Addition of this protease to BSAN2O3 causes an immediate release of N2O3 into the aqueous phase, resulting in a burst of nitrite formation (Fig. 1D). Fig. 2E demonstrates that protease-treated BSA was unable to potentiate GS-NO formation, apparently because of its inability to accumulate 䡠NO兾 N2O3. On the other hand, if the protease was added to BSAN2O3, the abrupt accumulation of GS-NO occurred (Fig. 2E), indicating that the large pool of NO⫹ became available for GSH upon proteolysis. GSH itself is resistant to proteinase K. These results directly implicate micellar catalysis of NO䡠 oxidation as a mechanism for albumin-mediated LMW RSH nitrosation. NO⫹ originated from N2O3 in the albumin hydrophobic core can be transferred to LMW RSH via at least three possible pathways (Fig. 3A). Direct attack on RSH is possible if RSH enters the hydrophobic interior of BSA (pathway I); alternatively, NO⫹ can reach RSH outside the protein hydrophobic core through the intermolecular transport to Cys-34 or Trp-214 followed by transnitrosation (pathways II and III). To determine which pathway plays the most important role in formation of LMW RS-NO, we prepared chemically modified BSA, where either Cys-34 or Trp-214 or both were covalently blocked and unable to undergo S- or N-nitrosation as confirmed by UV-Vis spectroscopy (Fig. 3B). Corresponding derivatives are designated as CBSA, WBSA, and CWBSA. We next compared the ability of BSAN2O3, CBSAN2O3, WBSAN2O3, and CWBSAN2O3 to stimulate nitrosation of RSH. Covalent modification of Cys-34 or Trp-214 or both should exclude pathway II and兾or III from LMW RSH nitrosation. The experiment with CWBSAN2O3 shows that this derivative potentiates formation of GS-NO with ⬇35% efficiency that of native BSAN2O3, implying that pathway I plays a substantial role in LMW RSH nitrosation (Fig. 3C). CBSAN2O3 and WBSAN2O3 exhibit ⬇60% and ⬇70% the activity of intact BSAN2O3, respectively (Fig. 3C), sugRafikova et al.

gesting that these pathways contribute roughly 40% and 30% to the full effect of intact BSAN2O3. Notably, at lower, close to physiological GSH concentrations, the efficiency of S-nitrosation mediated by BSAN2O3 and CWBSAN2O3 was virtually the same, suggesting that pathway I may be predominant in vivo.

Micellar Catalysis of LMW RS-NO Formation in Vivo and Regulation of Blood Pressure. Our in vitro and in vivo results with albumin suggest

that plasma protein absorbs NO䡠 under normal conditions in vivo because of its high QNO. Since free O2 is dissolved in mammalian blood in micromolar concentration, one can expect that because of micellar catalysis, a large portion of circulating albumin exists as albuminN2O3. Such albumin is also expected to serve as a catalyst of LMW RS-NO formation in vivo. To address this issue directly, we first tested rat blood plasma for its ability to potentiate GS-NO formation. An experiment similar to that described in Fig. 2B was performed in which freshly prepared rat plasma was used instead of BSAN2O3 (Fig. 5A). Although we did not add any exogenous NO䡠 to plasma, a significant amount of GS-NO was formed on addition of GSH. The amount of newly formed GS-NO exceeded the amount of preexisting RS-NO in plasma by more than 2-fold, suggesting that much of the newly formed GS-NO was not a result of transnitrosation. Furthermore, if plasma was preincubated for 60 min before addition of GSH, two times less GS-NO was formed in comparison with the fresh plasma situation (Fig. 5A). A similar rate of time-dependent decline in the ability to induce S-nitrosation was observed in vitro with pure albuminN2O3 (Fig. 2D). These results strongly suggest that plasma is indeed saturated with NO䡠兾N2O3 and serves as a catalyst of LMW RS-NO formation in vivo. To provide further support for this conclusion, we examined the effect of GSH on LMW RS-NO formation and vasodilation in rats. In vitro, the yield of LMW RS-NO in the presence of albuminN2O3 or fresh plasma directly depends on the concentration of LMW RSH (Fig. 2B). To test whether a similar relationship occurs in vivo, we measured the level of circulating LMW RS-NO and MAP in parallel after i.v. administration of various amounts of GSH to rat. As Fig. 5C demonstrates, infusion of 0.4 g兾kg GSH resulted in a more than 4-fold increase in the amount of LMW RS-NO in blood. Rafikova et al.

Fig. 5. Catalysis of RS-NO in vivo and its effect on systemic blood pressure. (A) Catalysis of GS-NO by rat plasma. Cu⫹⫹-dependant NO䡠 production was detected electrochemically in 100 ␮l of fresh plasma on addition of 10 ␮l 100 mM GSH (dark gray column) or 10 ␮l buffer (light gray column). The same experiment was done with plasma after 1 h at 25°C in air (right columns). (B) Schematics explaining the vasodilating effect of exogenous RSH. Increased concentration of LMW RSH in rat vasculature leads to a higher production of RS-NO via albuminN2O3. (C) Systemic vascular response to i.v. bolus of GSH produced a dose-dependent decrease in MAP (Left), which correlated with the increasing level of plasma LMW RS-NO (Right). 5 mM GSH, 12 mM GSH, and 20 mM GSH stand for the expected initial plasma GSH concentration upon GSH infusion. (Inset) The change of plasma GSH concentration as function of time as detected by HPLC analysis. The first plasma sample was ready for analysis 5 min after GSH administration. Blood pressure was measured ⬇10 min after GSH administration.

It was accompanied by a substantial decrease of MAP, which then remained at the same low level for more than an hour. Large doses of GSH were administered to compensate for the highly efficient import of GSH by various tissues (e.g., refs. 29 and 30). Concentrations shown in Fig. 5C (5, 12, and 20 mM) are estimated to be initial concentrations in blood on GSH injection. After that, plasma GSH decreased exponentially (Fig. 5C) so that at the moment of data collection GSH was detected within a micromolar range, yet significantly above the basal level (Fig. 5C). Effect of GSH on MAP and formation of RS-NO was dosedependant and correlated well with the dose-dependent curve of GS-NO formation in vitro (Fig. 2B). These results further suggest that the micellar catalytic mechanism of S-nitrosylation operates in mammalian circulation and is involved in control of vascular tone. Discussion The present study demonstrates an important role of the hydrophobic phase formed by plasma protein in the biochemistry of NO䡠. A soluble protein molecule, such as albumin, is envisioned as a micelle that is capable of concentrating nonpolar molecules of NO䡠 and O2 in a small volume of its hydrophobic compartments. We estimate that the concentration of NO䡠 inside the albumin globule must be ⬇120 times higher than in the surrounding aqueous phase (QNO ⬃ 120). Even supposing that QO2 for albumin is only 1, our results suggest that the acceleration of NO䡠 oxidation due to the increased local concentration of NO䡠 by albumin is several thousand-fold. Furthermore, reactive products of this reaction, most likely N2O3, appear to be much more stable within the protein PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9 兩 5917

BIOCHEMISTRY

compartments of albumin to absorb NO䡠 in vivo and the physiological role of this phenomenon, we studied the effect of exogenous albumin and albuminN2O3 on MAP in rats. To exclude the potential hemodynamic effect of S- and N-nitroso-derivatives of BSA, we used CWBSA. Intravenous infusion of CWBSA (⬇0.12 g兾kg body weight; see Experimental Procedures) was characterized by extensive hypertension (Fig. 4A), whereas the administration of the same amount of CWBSAN2O3 had only a slight effect on MAP (Fig. 4B). Moreover, the same amount of CWBSAN2O3 induced a substantial decrease of MAP if it was administered after animals were i.v. pretreated with the inhibitor of NO䡠 synthesis, N-nitro-L-arginine methyl ester (L-NAME) (Fig. 4B). Experiments with native BSA and BSAN2O3 gave similar results (Fig. 4). These observations are consistent with the model (Fig. 4C) in which the hydrophobic phase of plasma protein serves as a NO䡠兾N2O3 reservoir. Exogenous albumin, which has not been saturated with NO䡠, rapidly absorbs free circulating NO䡠, thus causing the MAP increase. MAP eventually recovers because newly synthesized NO䡠 apparently restores the equilibrium between the pool of free NO䡠 and that solubilized by plasma protein. On the other hand, exogenous CWBSAN2O3 or BSAN2O3 did not change MAP because they have already been saturated with N2O3 so that the pool of NO䡠兾NO⫹ in blood was not significantly perturbed. It should be noted that the solution of albumin that we used was isooncotic. Also, the amount of exogenous albumin was only 1% of the total plasma albumin and the rate of its infusion was slow (0.2 ml兾min). This rules out the possibility that exogenous albumin affected MAP by changing the osmotic status.

PHARMACOLOGY

Albumin as a NO䡠-Sink in Vivo. To evaluate the ability of hydrophobic

interior than in the surrounding aqueous phase (Fig. 1 C and D). Both these features, the micellar catalysis of NOx formation and preservation of this otherwise short-lived nitrosating agent, make albumin a potent catalyst of nitrosylation of its own Cys and Trp residues (19), as well as of external thiols (Figs. 2 and 3). The micellar catalytic mechanism of albumin-mediated NO䡠 oxidation may resolve an apparent paradox between the substantial amount of S-nitrosothiols observed in vivo (5–10) and limited concentration, life span, and reactivity of circulating NO䡠. Reaction of NO䡠 with O2 is third order (k ⬃ 4 ⫻ 106 M⫺2䡠sec⫺1) (20, 21), and under normal noninflammatory conditions, formation of nitrosating species, such as N2O3, would be too slow to account for any traceable amount of S-nitrosothiols (31). However, in this study we detect the basal level of RS-NO in rat plasma ranging from 50 nM to ⬇1 ␮M, which is consistent with earlier related estimates (5–10). The actual amount of circulating RS-NO is apparently higher, because some LMW RS-NO—e.g., S-nitroso-Cys and S-nitroso-␣lipoic acid—are highly unstable and at least partially decompose while blood samples were processed (O.R. and E.N., unpublished observations). Our in vitro and in vivo results correlate well with each other and support the model, in which plasma protein, predominantly albumin, is normally saturated with N2O3 (NO⫹-NO⫺ 2 ) and other NOx species in vivo and transfers NO⫹ to LMW RSH. We show that the level of circulating RS-NO can be increased severalfold in response to iv infusion of GSH (Fig. 5). A reasonable explanation for this result is that endogenous and exogenous GSH acquire NO⫹ not only via transnitrosation from albumin S-NO-Cys-34, which is limited to a basal level of detected RS-NO, but also from albumin N-NO-Trp-214 (ref. 32; Fig. 3) and directly from the pool of NOx in the protein interior (Fig. 3). This conclusion is further supported by our experiment in which fresh rat plasma is shown to possess a strong nitrosating activity toward exogenous GSH (Fig. 5A). Freshly prepared plasma was able to produce GS-NO to the level that is significantly higher than that of preexisting RS-NO in plasma, indicating that there was a larger source of NO⫹ in vivo unrelated to RS-NO. The nitrosation activity of fresh plasma lasts for about as long as that of albuminN2O3 (Fig. 2D) and correlates well with the estimated half-life of N2O3 in the albumin interior (Fig. 1D). Additionally, QNO for plasma and pure albumin are virtually identical (Fig. 1B). Taken together, these data argue that much of NO⫹ in circulation originates directly from N2O3 that accumulates in hydrophobic compartments of albumin. Another major source of NO⫹ in blood is S-nitroso-hemoglobin (SNO-Hb) (8). Like albumin, Hb is a globular protein, implying that its hydrophobic core is likely to be a micellar catalyst of N2O3

formation and S-nitrosylation. Unlike albuminN2O3, which rapidly and directly transfers NO⫹ to plasma thiols, SNO-Hb employs a relatively slow system, which relies on the membrane anionexchange protein 1, to ‘‘pump’’ its NO⫹ out of red blood cells (33). Additionally, the ability of Hb to exchange its NO⫹ is highly sensitive to pO2 (34). Thus, Hb and albumin may represent two major generators and transporters of NO⫹ but for different functioning. Whereas NO-derived vasoactivity of Hb matches ventilation to perfusion in lungs (35), such an activity of albumin may be of a more general nature in that it continuously adjusts the vascular response to rapidly changing NO䡠 concentrations, causing the systemic blood pressure to change smoothly. Formation of vasoactive nitrosothiols in response to i.v. infusion of GSH apparently explains dose-dependent hypotention induced by GSH (Fig. 4). This result suggests that one can regulate the level of circulating RS-NO and vascular tone simply by administering LMW RSH. We believe that this approach may have pharmacological implications. Because nitrosothiols are potent vasodilators and antiplatelet agents, they are considered promising therapeutics for various cardiovascular complications (36, 37). However, one major drawback of nitrosothiols as a commercial drug is their instability in physiological vehicles. Use of stable and nontoxic RSH, even at higher concentrations, to achieve the same effect as that of corresponding RS-NO seems to be advantageous. Another approach to control blood pressure can be envisioned based on our studies. If hydrophobic compartments of albumin serve as a powerful NO䡠 sink and reservoir for NOx, one can control vascular tone by changing the volume of circulating hydrophobic phase—e.g., filling of albumin with NO䡠兾NOx. Indeed, ‘‘empty’’ albumin (1% of total rat plasma albumin) administered i.v. to rats works as a vasoconstrictor, apparently because of its ability to sequester free circulating NO䡠, whereas ‘‘full’’ albumin was neutral in this sense (Fig. 4). In those experiments, chemically modified albumin, which could not undergo S- and N-nitrosation, produced the same hemodynamic response as that of unmodified albumin, arguing that Q-driving NO䡠 adsorption, rather than transnitrosation, had the major impact on vascular tone in this case. Apart from albumin, there are many natural and synthetic compounds that can form appropriate micelles in blood with high Q for NO䡠 and thus potentially serve as hemodynamic regulators. Exploring these as a tool to manipulate NO䡠 bioactivity should open a way for conceptionally new and exiting therapies in vascular pathophysiology.

Moncada, S. & Higgs, A. (1993) N. Engl. J. Med. 329, 2002–2012. Bredt, D. S. & Snyder, S. H. (1994) Annu. Rev. Biochem. 63, 175–195. Beckman, J. S. & Koppenol, W. H. (1996) Am. J. Physiol. 271, C14241–C1437. Ignarro, L. J., Lippton, H., Edwards, J. C., Baricos, W. H., Hyman, A. L., Kadowitz, P. J. & Gruetter, C. A. (1981) J. Pharmacol. Exp. Ther. 218, 739–749. Keaney, J. F., Jr., Simon, D. I., Stamler, J. S., Jaraki, O., Scharfstein, J., Vita, J. A. & Loscalzo, J. (1993) J. Clin. Invest. 91, 1582–1589. Stamler, J. S., Jaraki, O., Osborne, J., Simon, D. I., Keaney, J., Vita, J., Singel, D., Valeri, C. R. & Loscalzo, J. (1992) Proc. Natl. Acad. Sci. USA. 89, 7674–7677. Scharfstein, J. S., Keaney, J. F., Jr., Slivka, A., Welch, G. N., Vita, J. A., Stamler, J. S. & Loscalzo, J. (1994) J. Clin. Invest. 94, 1432–1439. Jia, L., Bonaventura, C., Bonaventura, J. & Stamler, J. S. (1996) Nature (London) 380, 221–226. Gaston, B., Sears, S., Woods, J., Hunt, J., Ponaman, M., McMahon, T. & Stamler, J. S. (1998) Lancet 351, 1317–1319. Wink, D. A., Kim, S., Coffin, D., Cook, J. C., Vodovotz, Y., Chistodoulou, D., Jourd’heuil, D. & Grisham, M. B. (1999) Methods Enzymol. 301, 201–211. Marley, R., Patel, R. P., Orie, N., Ceaser, E., Darley-Usmar, V. & Moore, K. (2001) Free Radical Biol. Med. 31, 688–696. Gladwin, M. T., Shelhamer, J. H., Schechter, A. N., Pease-Fye, M. E., Waclawiw, M. A., Panza, J. A., Ognibene, F. P. & Cannon, R. O., III (2000) Proc. Natl. Acad. Sci. USA 97, 11482–11487. Singh, R. J., Hogg, N., Joseph, J. & Kalyanaraman, B. (1996) J. Biol. Chem. 271, 18596–18603. Aleryani, S., Milo, E., Rose, Y. & Kostka, P. (1998) J. Biol. Chem. 273, 6041–6045. Nikitovic, D. & Holmgren, A. (1996) J. Biol. Chem. 271, 19180–19185. Ramachandran, N., Root, P., Jiang, X. M, Hogg, P. J. & Mutus, B. (2001) Proc. Natl. Acad. Sci. USA 98, 9539–9544. Jourd’heuil, D., Hallen, K., Feelisch, M. & Grisham, M. B. (2000) Free Radical Biol. Med. 28, 409–417.

18. Stamler, J. S., Toone, E. J., Lipton, S. A. & Sucher, N. J. (1997) Neuron 18, 691–696. 19. Nedospasov, A., Rafikov, R., Beda, N. & Nudler, E. (2000) Proc. Natl. Acad. Sci. USA 97, 13543–13548. 20. Wink, D. A., Nims, R. W., Darbyshire, J. F., Christodoulou, D., Hanbauer, I., Cox, G. W., Laval, F., Laval, J., Cook, J. A., Krishna, M. C., et al. (1994) Chem. Res. Toxicol. 7, 519–525. 21. Lewis, R. S. & Deen, W. M. (1994) Chem. Res. Toxicol. 7, 568–574. 22. Kharitonov, V. G., Sundquist, A. R. & Sharma, V. S. (1995) J. Biol. Chem. 270, 28158–28164. 23. Williams, D. L. (1997) Nitric Oxide 1, 522–527. 24. Beda, N. V. & Suntsova, T. P. (1999) FEBS Lett. 453, 229–235. 25. Liu, X., Miller, M. J., Joshi, M. S., Thomas, D. D. & Lancaster, J. R., Jr. (1998) Proc. Natl. Acad. Sci. USA 95, 2175–2179. 26. Goss, S. P., Singh, R. J., Hogg, N. & Kalyanaraman, B. (1999) Free Radical Res. 31, 597–606. 27. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70. 28. Simon, D. I., Mullins, M. E., Jia, L., Gaston, B., Singel, D. J. & Stamler, J. S. (1996) Proc. Natl. Acad. Sci. USA 93, 4736–4741. 29. Garcia-Ruiz, C., Fernandez-Checa, J. C. & Kaplowitz, N. (1992) J. Biol. Chem. 267, 22256–22264. 30. Lash, L. H. & Jones, D. P. (1984) J. Biol. Chem. 259, 14508–14514. 31. Keshive, M., Singh, S., Wishnok, J. S., Tannenbaum, S. R. & Deen, W. M. (1996) Chem. Res. Toxicol. 9, 988–993. 32. Zhang, Y. Y., Xu, A. M., Nomen, M., Walsh, M., Keaney, J. F., Jr., & Loscalzo, J. (1996) J. Biol. Chem. 271, 14271–14279. 33. Pawloski, J. R., Hess, D. T. & Stamler, J. S. (2001) Nature (London) 409, 577–578. 34. Lipton, A. J., Johnson, M. A., Macdonald, T., Lieberman, M. W., Gozal, D. & Gaston, B. (2001) Nature (London) 413, 171–174. 35. Lipton, S. A. (2001) Nature (London) 413, 118–121. 36. Hogg, N. (2000) Free Radical Biol. Med. 28, 1478–1486. 37. Al-Sa’doni, H. & Ferro, A. (2000) Clin. Sci. 98, 507–520.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

5918 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.092048999

These studies were supported by the Searle Scholar Award, the Irma T. Hirschl Career Award, and a grant from the National Institutes of Health (to E.N.).

Rafikova et al.