We used bovine Hb (Serva, Germany), Na₂HPO₄∙6H₂O and NaH₂PO₄∙H₂O (MP Biomedicals, Germany). The water was purified by distillation in a Bi/Duplex distiller (Germany). (II) was synthesized using the known method [14] . Synthesis of (I) is described in [15] . It has been obtained by reaction of dissolved in water ferrous

sulphate (II) with an aqueous solution of D-penicillamine in the molar ratio 1:3. The reaction was per- formed using standard vacuum line and Schlenk technology under argon. Previously, oxygen has been removed from the water by triple freezing and vacuum pumping. To the dry mixture, containing 0.42 g (1.5 mmol) of FeSO₄∙7H₂O and 0.68 g (4.5 mmol) D-penicillamine poured 10 ml water, prepared as described above, and nitric oxide have been passed through the resulting deep purple solution at room temperature. The fine red needles have been appeared on the walls of the reaction vessel after 10 - 12 min, gradually filling the entire volume of the solution. The mixture was kept 3 days at 6˚C - 8˚C, filtered and dried in vacuum under argon. The filtered product is stable in the solid phase in the air for a long time (months). Yield is 198 mg (20%). Elemental analysis of the polycrystalline powders obtained was performed at the Analytical Center of the Institute of Problems of Chemical Physics of the Russian Academy of Sciences. For (I): Fe₂S₃N₆C₁₀H₃₂O₁₇. Found %: Fe, 15.62; S, 13.34; N, 11.89; C, 16.80, H, 4.62. Calculated, %: Fe, 15.64; S, 13.42; N, 11.72; C, 16.76, О, 37.99; H, 4.47; IR: ν/cm⁻¹ = 1771 (s, NO); 1723 (s, NO).

Operation technique in inert gas atmosphere has been described earlier [8] .

A homogeneous solution of bovine Hb was prepared from commercial Hb (a mixture of oxygenated hemoglobin (HbO₂) and methemoglobin according to a known procedure [9] .

To a weighed sample of the (I) in a vessel filled with nitrogen, an anaerobic 0.05 M phosphate buffer, pH 7.0, was added so as to prepare a solution of the complex with a concentration of 6 × 10⁻⁴ M. Then the solution was stirred for 15 min under a nitrogen stream until the complex was completely dissolved. An aliquot (0.1: 0.5; 0.75; 1 ml) of the solution was drawn under a nitrogen stream and transferred to the anaerobic sample cell and the reference cell (the volume 4 ml, the optical path length 1 cm) containing an anaerobic phosphate buffer, pH 7.0 in an amount sufficient for obtaining the final (I) concentration 2 × 10⁻⁵; 10⁻⁴; 1.5 × 10⁻⁴ and 2 × 10⁻⁴ M, respectively. Then the phosphate buffer, pH 7.0, was added to the reference and sample cells in amounts required to make the volume of the reaction solution after the insertion of Hb into the sample cell equal to 3.0 ml. The reaction was initiated by adding an Hb solution with the initial concentration of 4.5 × 10⁻⁴ M into the sample cell. The concentration of Hb in the sample cell was (1.94 - 2.1) × 10⁻⁵ M. Then the difference absorption spectra were recorded at certain time intervals. The absorption spectra were measured until Hb was completely transformed into HbNO, i.e., until the spectra stopped to change.

Absorption spectra were recorded at 25˚С using a Specord M-40 spectrophotometer equipped with an interface for computer-aided registration of spectra and thermostatic cuvette holder. Amount of Hb and HbNO was evaluated spectrophotometrically. For this purpose absorption spectra were factored by components using program Mathcad 11 Enterprise Edition as described in the paper [8] .

We considered the assumed reaction scheme describing the interaction of (I) with Hb. The rate constants were determined by the least squares method based on the numerical solution of the corresponding system of differential equations. The concentrations of NO or the HbNO were determined after the resolution of the absorption spectra into components (the spectra of Hb and HbNO), and used as experimental data.

The analysis of Hb surface was performed using the program PyMOL [16] . To view and analyze the surface of Hb, we used X-ray data of bovine Hb [17] , retrieved from the database PDB, access code 1HDA (see Ref. [18] ). This method allowed us to determine the total electrostatic surface charge. The density of the color depends on its magnitude. Dimensions can be determined with an accuracy of ~10%.

In the present work we studied the interaction of (I) with Hb. All reactions with complex (I) were carried out under nitrogen because NO rapidly reacts with O₂ to give nitrogen oxides (the rate constant is 2 × 10⁶ M⁻²∙s⁻¹) [19] . Hb is a trap for NO: the binding rate is close to the diffusion rate [20] , the equilibrium constant is 3 × 10¹⁰ M⁻¹ [21] . Hb gives a characteristic absorption spectrum, which changes in the course of NO binding. Hence, as was reported in [8] [9] , the NO release can be followed from the formation of HbNO. Since all sulfur-nitrosyl iron complexes show absorption in the visible region, we recorded the difference absorption spectra of the control and test systems with Hb containing the complex (I) at the same concentrations (see the Experimental section). The changes in the difference absorption spectra with time in the course of the reaction of Hb with the complex (I) are displayed in Figure 2. We stopped recording the spectra after the latter ceased to change. This was accompanied by a decrease in the absorbance at the maximum at 556 nm in the absorption spectrum of Hb and an increase in the absorbance at 545 and 575 nm suggesting the formation of the HbNO complex. These

spectra have three isosbestic points at 551, 570, and 595 nm. This is evidence that only Hb and HbNO contribute to the absorption spectra, as in the case of the reactions of this class of NICs with Hb investigated in our earlier research [8] . We measured the kinetics of the formation of HbNO by recording the accumulation of HbNO and the deconvolution of the absorption spectra with the use of the MathCad program (Figure 3). The rate constants k (Table 1) is somewhat less than the constant rate of NO release from (I), defined by the sensor electrode in 0.05 M phosphate buffer pH 7.0 and a temperature 25˚C in a nitrogen atmosphere: k₁ = (4.6 ± 0.1) × 10⁻³ s⁻¹ [22] . In [8] the authors explain why the HbNO formation reaction should “track” the NO released from NIC. We firmly established that Hb usually stabilizes the NIC and the reaction of NO release in the presence of Hb is slower than without Hb, as was found in the determination of the reaction rate of NO release from NIC using the sensor electrode [12] . We have determined that sulfur ligands, which are part of the NICs with excess electron density, contribute to the stabilization of complexes in solution [9] . The greater the electron density on the NIC ligand, the more stable it is in the presence of Hb [23] . This stabilization is due to a known anion-binding center in Hb, located in the cavity between Hb subunits [11] . Recently, we described the presence of cations binding sites on Hb. In the case of the cationic NIC (II), having thiol ligands cysteamine instead of penicillamines there was significant stabilization of the complex in the presence of Hb. This was attributed to the presence of the cationic sites, negative charge centers, reversible binding (II), and that these centers were visualized [12] . The size of (I) was greater than the size of (II). Thus, using the program PyMOL [16] , we found that the size of the (I) dication is ~375 ų, while the size of the (II) dication is equal to ~265 ų [12] .

Consider the reaction of (I) with Hb. If Hb is present in the medium, it seems, due to the large value of the equilibrium constant of, equal to 3 × 10¹⁰ М⁻¹ [21] , it shifts the equilibrium decomposition reaction of (I) towards the formation of HbNO and, being a depot for NO, carries it to the respective targets.

Without the Hb equilibrium decomposition reaction of (I), evidently biased toward (I) decomposition. It is consumed in the reactions of NO, known as the universal controllers necessary for functions of cellular metabolism [1] [2] .

In [12] , the authors observed a decrease in the processing rate for NO release for cysteamine cationic NIC (II) in the presence of Hb, which was explained by adsorption of the complex by the Hb molecule, resulting in a substantially reduced capacity of (II) to evolve NO. For (I), a structural analog of (II), the reaction scheme corresponds to a model of the given process:

Here S_Hb are the active binding site of (I) on Hb surface.

In accordance with this scheme, the free molecules of complex (II) evolve NO in solution, with reversible reaction rate constants k₁ and k₋₁. Decomposition of (I) in water is followed by separation of the penicillamine ligand [22] from molecule (I) and the product P₁, which is obtained after the separation of NO from (I), as seen in reactions (2) and (3). Simultaneously, the molecule is (I) adsorbed on an Hb macromolecule. It is assumed that the binding sites of (I) exist on Hb surface (in a certain amount, n_s per Hb molecule), with which the molecule of complex (I) can reversibly interact with an equilibrium constant, K_s. In the bound state of the complex, molecules evolve NO with a rate constant of k₄. The separated NO molecules reversibly interact with the heme iron of Hb to form the experimentally measured HbNO product.

He values of rate constants k₁ = 4.6 × 10⁻³ s⁻¹, k₋₁ = 9.7 × 10³ M⁻¹∙s⁻¹, k₂ = 5.3 × 10⁻⁴ s⁻¹, k₋₂ = 0.1 M⁻¹∙s⁻¹, k₃ = 6.6 × 10⁻⁶ s⁻¹ we determined in paper [22] . The literature contains information on the value of the constant k₅ = 10⁸ s⁻¹ [19] and the equilibrium constant K₅, k₅/k₋₅, equal to 3 × 10¹⁰ M⁻¹ [21] . The unknown parameters are K_s, n_s and k₄. Analysis of the inverse problem has shown that by available experimental data, these parameters cannot be uniquely determined. Therefore, in a first approximation, we considered the limiting case when NO release by the bound complex (I)―S_Hb is negligible. This regime corresponds to the value of k₄ ~ 10⁻⁸ s⁻¹ [12] , as in the case of (II). The problem reduces to the determination with the experimental data [HbNO](t), the equilibrium constant K_s, and the number of active sites on the surface of the hemoglobin, n_s.

The corresponding system of equations is as follows:

Here;.

The initial conditions, taking into account the detailed balance in the reaction (4):

,

where

;

;

;

;

Here, , are the initial concentrations of the complex (I), Hb, and the active sites on the surface S_Hb, respectively.

Unknown kinetic parameters K_s and n_s were determined by numerical solution of the inverse task, using the kinetics of accumulation of HbNO in four experiments of the (I) interaction with Hb (Table 1).

Calculations were carried out at the value of the equilibrium constant K₅ = 0.5 × 10⁹ М⁻¹, which provides a better match of theory and experiment than K₅ = 3 × 10¹⁰ M⁻¹. This is justified, considering the fact that, in the presence of salts, K₅ decreases, wherein the salt effect increases in the following order: NaCl, KCl, sodium citrate, sodium phosphate [24] . In our case, the experiments were performed in 0.05 M phosphate buffer, pH 7.0. Point values (Table 1) satisfactorily describe the experimental data (Figure 3). As seen from Table 1, increasing the concentration of the complex causes value n_s to increase and K_s to decrease. Apparently, there are various regions of different binding strength on the surface of Hb for binding of the cation complex. At low concentrations of complex (I) (number 1 in Table 1), its molecules primarily bind with regions, providing a stronger link (large K_s), and while increasing the concentration of [(I)]₀ during adsorption, they begin to participate in other regions on the Hb surface with lower bond strength. In processing of the experimental data, this leads to a change in the effective values of the parameters: increase of n_s and decrease of K_s.

It is known that Hb is composed of four subunits: two α (A, C) and two β (B, D). The α subunit consists of 141 amino acid residues, and the β subunit contains 145 residues. The α subunits are in contact with each other and form a narrow entrance of the through-channel. The β subunits are in contact with each α subunit, but are not in contact with each other and form a wide entrance of this channel. On the whole, Hb is a symmetrical macromolecule with a twofold symmetry axis and a channel running through the structure. It is known that the anion binding site, where DPG is bound, is located in this through-channel [11] . In the through-channel, 8 cationic and 3 anionic pairs of amino acids were located [12] ; in general it is positively charged and directly enters the channel, which is also positively charged. Therefore, the complexes (I) and (II) can not penetrate to this channel. It has been shown in [12] , using the program PyMOL, that the surface area of Hb has a local concentration of both negatively and positively charged amino acids , as well as more or less than the neutral portion of the surface. Among various roughnesses on the Hb surface, there are four large cavities (see Figure 4). Two of them (cavities 1) are located in the region of the negatively charged surface from the side of the wide entrance at equal distances (~13 Å) from the hemes of the А and D subunits. The volume of the cavity 1 is ~17 ų. Two other large cavities (cavities 2) with volumes of ~850 ų are also negatively charged and are located symmetrically with respect to the cavity 1 at equal distances from the hemes of the А and D subunits (see Figure 4). The negative potential of the cavities 1 and the adjacent surfaces is formed by the negatively charged carboxyl groups of the following amino acids: cavity 1 (subunits A, B, D), by A-Glu23, A-Glu27, A-Glu30, A-Asp47, B-Glu5, B-Glu6, B-Glu120, B-Asp128, D-Glu89, D-Asp93, and the terminal aminoacid D-His145; cavity 1 (subunits C, D, B), by C-Glu23, C-Glu27, C-Glu30, C-Asp47, D-Glu5, D-Glu6, D-Glu120, D-Asp128, B-Glu89, B-Asp93 and the terminal amino acid B-His145. The negative potential of the cavities 2 and the adjacent surfaces is formed by the negatively charged carboxyl groups of the following amino acids: cavity 2 (subunits A, C, D), by A-Asp85 and the terminal amino acid A-Arg141, C-Asp6, D-Glu42, D-Asp46, D-Asp51; cavity 2 (subunits C, A, B), by C-Asp85 and the terminal amino acid C-Arg141, A-Asp6, B-Glu42, B-Asp46, B-Asp51. Most likely, the complex (II) can be located in these cavities and can be bound to the negatively charged functional groups of Hb. Correspondingly, it can be suggested that at least one (II) molecule can be bound in the cavity 1, and at least three (II) molecules can be bound in the cavity 2. In total, each Hb molecule can bind at least eight (II) molecules. Taking into account the results of the kinetic modeling [12] , which gave approximately 12 (II)-binding sites per Hb molecule, it can be concluded that there are from 8 to 12 such sites. Figure 4 shows that, in addition to the cavities 1 and 2, there are adjacent surfaces bearing a negative charge. Apparently, they are also involved in the binding of (II). The surfaces in the direct vicinity of the hemes are positively charged, the charge gradually decreasing with increasing distance from the heme. In this study, we had a different situation. The volume of (I) is ~375 ų, and only one molecule of (I) can fit into the large, negatively charged cavity, i.e., to cavity 2. Moreover, in this case, the binding is stronger: K_s = 4.3 × 10⁷ M⁻¹. Therefore, the (I) molecules bind more weakly. The PyMOL program identifies shallow and indistinct regions with mixed charge from a small positive to a negative charge on the surface of the β-subunit at a distance of approximately 10 Å from the heme. On such areas, the binding of (I) probably occurs in this case, starting from a negative area. So K_s decreases with increasing concentrations of (I).