INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Control of blood pressure and resistance to blood flow is achieved by a dynamic constriction and relaxation of smooth muscle tissue which surrounds all blood vessels except capillaries. Vascular smooth muscle tension is continually adjusted by a complex system that causes either vasoconstriction or vasodilation, depending on metabolic need (1). Research performed over the last decade has established that endothelium-derived nitric oxide (^·NO)¹ can cause vasodilation. ^·NO is produced by endothelial cells that lie between the intravascular space and the surrounding smooth muscle. Among the findings was the demonstration that ^·NO donors (e.g. nitroprusside, nitroglycerin) lead to vasorelaxation through activation of guanylate cyclase, whereas inhibitors of ^·NO synthesis (e.g. N^G-monomethyl-L-arginine) or scavengers (e.g. hemoglobin) cause vasoconstriction (for reviews, see Refs. 2 and 3).

Since cell-free hemoglobin is being developed as a red cell substitute (4), reactions between hemoglobin and ^·NO are of potential importance in maintenance of microvascular blood flow and O₂ delivery. Despite the wide variation that exists in the physical properties (O₂ affinity, molecular mass, and solution properties) of different cell-free hemoglobins, it appears vasoconstriction is a feature common to many hemoglobin solutions (for reviews, see Refs. 2, 3, and 5). It is tempting to conclude that ^·NO scavenging is the principal, if not sole mechanism for vasoconstriction associated with cell-free hemoglobin. However, it is well established that multiple factors contribute to the physiological control of vascular smooth muscle tone, and other mechanisms may operate as well (6).

Recently, cell-free hemoglobin solutions that differ in blood pressure response were described by Migita et al. (7). These workers compared bovine hemoglobin chemically modified by surface conjugation to polyethylene glycol (PEG-Hb) with human hemoglobin cross-linked between the lysine 99 residues of the  subunits (-Hb). Isovolemic exchange transfusions in rats with a solution of -Hb resulted in a significant increase in MAP, whereas PEG-Hb solutions caused no significant change in blood pressure. If increases in MAP are due to ^·NO scavenging reactions, then these two hemoglobins would be expected to exhibit different reactivities with ^·NO. We have made a direct test of the hypothesis that the blood pressure response to cell-free hemoglobin is related to the reactivity of hemoglobin with ^·NO by studying the vasopressor response to hemoglobin solutions and the rate constants for reactions with ^·NO. In addition to -Hb and PEG-Hb, we have studied four other chemically modified hemoglobins (see Table I). The list includes examples of the three principle types of hemoglobin modification under development for use as a red cell substitute: intramolecular cross-linking, polymerization, and surface conjugation.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Hemoglobin Solutions-- Chemically modified hemoglobin preparations were dissolved in Ringer's lactate to make the solutions used for both exchange transfusion experiments and in vitro kinetic studies. Human hemoglobin cross-linked with bis-(3,5-dibromosalicyl) fumarate between the  chains at lysine 99 (-Hb, 7.9 g/dl, 4.9 mM heme) was supplied as a gift from the U.S. Army, Blood Research Detachment, Walter Reed Army Institute for Research (8). Purified native human hemoglobin (HbA₀, 9.0 g/dl, 5.5 mM heme) was supplied as a gift from Hemosol, Inc. Human hemoglobin cross-linked with trimesoyl tris (methyl phosphate) to make either a two-point intramolecular cross-link between the  chains at lysine 82 (82-Hb, 6.8 g/dl, 4.3 mM heme) or a three-point cross-link between the  chains at lysine 82 and valine 1 of one  chain (Tm-Hb, 7.0 g/dl, 4.4 mM heme) were prepared as described by Kluger et al. (9) and supplied as a gift from Hemosol, Inc. Human hemoglobin polymerized with ring-opened raffinose (o-R-poly-Hb, 9.4 g/dl, 5.9 mM heme) was also supplied as a gift from Hemosol, Inc. (10). Human hemoglobin modified by covalent attachment of pyridoxal-5'-phosphate and surface conjugation to -carboxymethyl, -carboxymethoxypolyoxyethylene (PHP, 7.7 g/dl, 4.8 mM heme) was supplied as a gift from Apex Biosciences, Inc. (11). Bovine hemoglobin surface conjugated to methoxypolyoxyethylene glycol (PEG-Hb, 5.5 g/dl, 3.4 mM heme) was supplied as a gift from Enzon, Inc. (12).

MAP Responses to 50% Isovolemic Exchange Transfusion-- Measurements of mean arterial pressure were conducted using groups of male Sprague-Dawley rats (250-350 g, Charles River Labs). All animal protocols were approved by the Animal Care Committee of the San Diego Veterans Affairs Medical Center. Animals were fed ad libitum prior to all experiments. Surgical preparation was performed 1 day prior to the exchange transfusion experiment. Rats were anesthetized with 250 µl of a mixture of ketamine (71 mg/ml; Aveco Co., Fort Dodge, IA), acepromazine (2.85 mg/ml; Fermenta, Kansas City, MO), and xylazine (2.85 mg/ml; Lloyd Laboratories, Shenandoah, IA). The areas of the femoral arteries and veins were exposed by blunt dissection. A specially constructed polyethylene catheter (PE-10 connected to PE-50) was placed into the abdominal aorta via one femoral artery to allow rapid withdrawal of arterial blood. An identical catheter was placed in the femoral artery of the opposite leg to monitor blood pressure. A PE-50 catheter was placed in the inferior vena cava via a femoral vein to allow infusion of test solutions. Catheters were tunneled subcutaneously, exteriorized through the tail, held in place by a plastic sheath assembly, and flushed with approximately 100 µL of normal saline. Animals were allowed to recover from the surgical procedure and remained in their cages for an additional 24 h before being used in experiments.

Measurement of Oxygen Equilibrium Curves-- Hemoglobin-oxygen equilibrium curves were measured as described by Vandegriff et al. (13). Hemoglobin solutions (0.1 M bis-Tris propane, 0.1 M Cl, pH 7.4) were rapidly deoxygenated using the protocatechuic acid/protocatechuic acid 3,4-dioxygenase system (14). Both the enzyme and substrate were obtained from Sigma. The deoxygenation reaction was followed by simultaneous measurements of hemoglobin visible spectra and pO₂ using a Milton Roy 3000 diode array spectrophotometer (SLM Instruments, Inc., Urbana, IL) and a micro-oxygen electrode (Microelectrodes, Inc., Londonderry, NH). Data were analyzed as hemoglobin fractional saturation versus of pO₂ and fitted for Adair constants, p₅₀, and Hill's coefficient at 50% saturation (n₅₀).

Sample Preparation for in Vitro ^·NO Reaction Studies-- All gases were obtained from Matheson. Sodium hydrosulfite (sodium dithionite) was provided as a gift from Hoechst-Celanese. Sodium chloride, and bis-Tris propane were obtained from Sigma. Anaerobic buffers were prepared by bubbling the solutions with either nitrogen or argon. ^·NO gas was scrubbed before use by bubbling through anaerobic 2 M potassium hydroxide solution to remove contaminating products of ^·NO autoxidation. The scrubbed ^·NO gas was flushed through a glass tonometer equipped with input and output stopcocks and a side port that was sealed with a rubber septum. After flushing the tonometer with ^·NO gas, anaerobic 0.1 M bis-Tris propane, 0.1 M Cl, pH 7.4, buffer was added to the tonometer through the septum. The buffer was equilibrated under 1 atm of ^·NO gas to produce [^·NO] = 1.97 mM stock solutions. Solutions at lower ^·NO concentrations were prepared by dilution of a stock ^·NO solution with the appropriate volume of anaerobic buffer.

Kinetic Measurements-- Bimolecular association kinetics were measured by laser flash photolysis, a technique which takes advantage of the photosensitivity of the bond between ferrous heme protein and ligand (15). A Lumonics XeCl Excimer (308 nm) laser pumped a dye laser to produce a light pulse 4 ns in duration that averages 3 mJ per pulse at 540 nm. The observation light was provided by a tungsten-halogen lamp, and both the photolysis and observation beams impinged from counterpropagating directions onto a 2-mm path length cuvette containing the Hb-NO sample. ^·NO recombination kinetic transients were monitored at 435 nm. Substantially less than 1% photolysis was produced by 100% laser light intensity due to the extremely low quantum yield for dissociation of Hb-NO complexes (15), and 1000 transients were averaged for each sample to improve the signal-to-noise ratio and produce usable ^·NO recombination time courses. The time courses were fitted to a single exponential expression, taking into account the shape of the laser pulse and the response time of the instrument, using an iterative non-linear least squares algorithm to obtain the observed rate constants (k_obs). Under these conditions, the reaction is pseudo first order and the bimolecular association rate constants were obtained by dividing the observed rate constants by the ^·NO concentration (k' = k_obs/[^·NO]).

(Eq. 1)

1

1

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Effect of Hemoglobin Solutions on Blood Pressure-- The mean arterial pressure responses to a 50% isovolemic exchange transfusion with the six different cell-free hemoglobin solutions are shown in Fig. 1. The data are divided into two panels and plotted as the percent change in blood pressure during the base-line and exchange periods. Fig. 1 shows the final 10 min of the 30-min base-line period followed by the 30-min exchange transfusion period. The x-axis is scaled such that the exchange begins at time = 0 min. The different types of MAP response observed with the different hemoglobin solutions used in these experiments can be put into three categories based on their changes from base line (Table I): 1) an immediate and sustained increase (-Hb, Tm-Hb, and o-R-poly-Hb); 2) an immediate but transient increase (PHP and 82-Hb); and 3) no significant (p > 0.05) increase (PEG-Hb). The time period between exposure to cell-free hemoglobin and appearance of a significant increase in MAP is approximately 8-10 s for the five groups of rats that show blood pressure increases.

View larger version (45K): [in this window] [in a new window]   Fig. 1.   Arterial blood pressure response to 50% isovolemic exchange transfusion with cell-free hemoglobin solutions. The percent change in mean arterial pressure is plotted versus time with the x-axis scaled such that the last 10 min of the 30-min base-line period are shown, and the exchange transfusion begins at time = 0 min. The symbols represent the average value at a given time point for each group of animals. The error bars represent the standard error of the mean. Results for the six groups are divided into two panels. A, MAP response for animals receiving -Hb (, n = 6), PHP (, n = 16), and PEG-Hb (, n = 5). B, MAP response for animals receiving Tm-Hb (, n = 9), o-R-poly-Hb (, n = 5), and 82-Hb (, n = 4). Data for -Hb and PEG-Hb are taken from Migita et al. (7).

View this table: [in this window] [in a new window]   Table I Properties of modified hemoglobins Types of hemoglobin modifications are described in Winslow (4). Molecular mass values for the hemoglobin cross-linked tetramers (-Hb, Tm-Hb, and 82-Hb) are not significantly different than unmodified human hemoglobin. Molecular mass values for PHP, PEG-Hb, and o-R-poly-Hb were taken from Vandegriff et al. (25). Oxygen equilibrium binding parameters (p50, n50) were measured in 0.1 M bis-Tris propane, 0.1 M Cl, pH 7.4, 37 °C by the method described in Vandegriff et al. (13). The errors represent the standard error of the mean. Entries in the column labeled "Vasopressor Effect" are based on statistical analyses of the data shown in Fig. 1. MAP responses that increased and remained significantly above base line (p < 0.05) throughout the exchange transfusion period are considered "Sustained." MAP values that initially increase significantly above base line, then fall to values not significantly different from baseline during the exchange transfusion are considered "Transient." MAP responses not significantly different from baseline are labeled as "None" since no vasopressor effect is observed.

^·NO Reactions at the Heme-- To test the hypothesis that different patterns of MAP response can be explained by different degrees of ^·NO scavenging, the reaction of ^·NO at the hemes of these six hemoglobins was studied in vitro. Kinetic and equilibrium binding constants were determined for the reaction between ^·NO and the reduced, unliganded form of these hemoglobins. The reaction between ^·NO and the oxygenated forms of these hemoglobins was also studied.

Bimolecular Association Kinetics-- The bimolecular recombination time courses for ^·NO binding to all six chemically modified hemoglobins studied, and native human HbA₀ as a control, are identical within experimental error (Table II). Recombination time courses subsequent to the laser pulse were best described by a single exponential function. The concentration of free ^·NO is in vast excess of the concentration of photolysed hemes (<1 µM) and a pseudo-first order approximation is valid for the recombination reaction. The overall bimolecular association rate constants for these reactions are obtained by dividing the fitted value of the observed rate constant by the concentration of ^·NO. All of the hemoglobins studied exhibited values for the association rate constant that are, within experimental error, equal to 30 µM¹ s¹. This value is essentially the same as the value previously reported for ^·NO binding to native human hemoglobin (25 µM¹ s¹) (21) and indicates that the various types of chemical modification (i.e. cross-linking, surface modification, polymerization) have no measurable effect on the room temperature association kinetics of ^·NO binding to the heme of hemoglobin.

View this table: [in this window] [in a new window]   Table II Nitric oxide reaction parameters Each parameter listed for each hemoglobin represents the mean of at least three separate determinations, all determined in 0.1 M bis-Tris propane, 0.1 M Cl, pH 7.4, 23 °C. The errors represent the standard error of the mean. The bimolecular association rate constants (k') were measured by flash photolysis of nitrosylhemoglobin samples prepared in the presence of 1.97 mM (1 atm) ·NO and 0.197 mM (0.1 atm) ·NO. Dissociation rate constants were determined by ligand displacement with CO in the presence of excess sodium dithionite ([dithionite] = 1-2 mM). Rate constants (krapid, kslow) and amplitudes (frapid, %) for the rapid and slow kinetic phases of the displacement reaction were obtained from biexponential fits of the time courses (Equation 1) (fraction of the slow kinetic phase equals 1  fraction rapid). The overall dissociation rate constants (k) represent the weighted mean of both kinetic phases based on the fractional amplitudes for each kinetic phase. Equilibrium constants (Kd) were calculated as the ratio of the dissociation rate constant to the association rate constant (Kd = k/k'). Bimolecular rate constants for the reaction of ·NO with oxyhemoglobin were determined by stopped-flow rapid mixing.

^·NO Oxidation of Oxyhemoglobin-- Bimolecular rate constants for the reaction of ^·NO with the oxygenated forms of the chemically modified hemoglobins were measured by rapid mixing techniques. This irreversible reaction proceeds by direct bimolecular combination of ^·NO with O₂ bound to hemoglobin and is not an O₂ displacement reaction (18, 19). Time courses for the ^·NO reaction with each of the six chemically modified hemoglobins studied, and native human HbA₀ as a control, are identical within experimental error (Table II). Overall bimolecular rate constants for these reactions were obtained by dividing the fitted value of the observed rate constants by the concentration of ^·NO (k_ox' = k_obs/[^·NO], Table II). As observed for the ^·NO bimolecular recombination reaction, there is no difference in the reaction kinetics between any of the hemoglobins studied, within experimental noise limits. All hemoglobins studied yielded values of approximately 30 µM¹ s¹ for k_ox'.

Unimolecular Dissociation Kinetics-- The dissociation of ^·NO from each of the hemoglobins was studied by CO replacement in the presence of excess sodium dithionite. In all cases, the time courses for the conversion of nitrosylhemoglobin to carbonylhemoglobin displayed distinctly biphasic kinetic behavior and were fitted to a double exponential expression (Equation 1) to obtain observed rate constants. Under the conditions used for these reactions, the values of the dissociation rate constants were assumed to be equal to the values of the observed rate constants (Table II). The values of the rate constants for the slow kinetic phase vary by less than a factor of two, ranging from 3.6 × 10⁵ s¹ for 82-Hb to 6.1 × 10⁵ s¹ for Tm-Hb. By contrast, the values of the rate constants for the rapid kinetic phase vary by more than a factor of five, ranging from 19 × 10⁵ s¹ for 82-Hb to 104 × 10⁵ s¹ for o-R-poly-Hb.

Equilibrium Constants for ^·NO Binding-- Due to the extremely high affinity of hemoglobin binding to ^·NO, it is generally not possible to determine the values of overall dissociation equilibrium constants by titration methods. In such cases, the equilibrium dissociation constants are calculated as the ratio of the respective unimolecular dissociation rate constants to the bimolecular association rate constants (K_d = k/k', Table II). The values of the equilibrium constants calculated in this manner vary by a factor of six, ranging from 2.3 pM for 82-Hb to 14 pM for o-R-poly-Hb. The ^·NO affinities for these hemoglobins parallel their respective oxygen affinities (see Tables 1 and 2 and Fig. 2). The overall affinity of these chemically modified hemoglobins for ^·NO is principally controlled by the rate of dissociation from T-state hemoglobin (see "Discussion"). In the case of 82-Hb and HbA₀, both the small values of the slow kinetic phase rate constants and the relatively small fractions of fully liganded T-state contribute to the low value of their overall dissociation rate constants. The hemoglobins that show an immediate and sustained increase in mean arterial blood pressure in response to 50% isovolemic exchange transfusion (-Hb, Tm-Hb, o-R-poly-Hb, Table I, Fig. 1) exhibit the weakest ^·NO binding affinity (Table II). The hemoglobins that show either a transient MAP increase (82-Hb, PHP) or no significant increase (PEG-Hb) display the tightest ^·NO binding.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Correlation between ·NO and O2 affinities. Equilibrium dissociation constants for ·NO binding to chemically modified hemoglobins (Kd)are plotted versus the corresponding partial pressures of oxygen at half saturation (p50). The error bars represent the standard error of the mean. Values are taken from Tables I and II. Cell-free hemoglobins fall into two groups: 1) those that elicited an immediate and sustained MAP increase (-Hb, Tm-Hb, and o-R-poly-Hb) (), and 2) hemoglobins that caused either a transient (PHP and 82-Hb) or no significant MAP increase (*PEG-Hb) (). The numbers in parentheses are the values (mm Hg) for the maximum increase in MAP over base line during exchange transfusion (Fig. 1). The symbols are within a 95% confidence level of the fitted linear regression (solid line).

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

A hypothesis has been advanced by several researchers that the vasopressor response usually observed upon administration of cell-free hemoglobin solutions is due to ^·NO scavenging by hemoglobin (for a review, see Ref. 5). Physiologic control of vascular smooth muscle tone is achieved by a dynamic balance of factors that cause vasoconstriction and vasorelaxation. A reduction in concentration of EDRF (^·NO) in vivo is expected to result in an increase in vascular smooth muscle tension since a factor believed to counter vasoconstrictive processes is lost. Furthermore, the time period between exposure to hemoglobin and appearance of a significant increase in MAP is approximately 8-10 s for the five groups of rats that show blood pressure increases. This implies that only a small amount of cell-free hemoglobin (<10 mg) is required to elicit the blood pressure responses, and suggests that the rates of the reaction(s) that cause these responses are very rapid. ^·NO binds to the deoxy-form of hemoglobin with picomolar affinity, and reacts irreversibly with oxyhemoglobin. Both of these reactions occur with bimolecular rate constants in excess of 30 µM¹ s¹ at 37 °C. Thus it is easy to assume that cell-free hemoglobin solutions cause vasoconstriction by rapid removal of EDRF.

Recent evidence in support of this theory comes from experiments that used site-directed mutants of recombinant human hemoglobin designed such that the rates of ^·NO reaction with these mutant hemoglobins were reduced relative to that of native hemoglobin (19, 22). The bimolecular rate constants for ^·NO-induced oxidation for the resulting mutant oxyhemoglobins were up to 20-fold lower than the rate observed for wild-type Hb-O₂. These recombinant hemoglobins elicited vasopressor responses in rats that were significantly diminished relative to that observed with recombinant hemoglobin exhibiting a ^·NO reaction rate constant equal to native human hemoglobin. There was a direct correlation between the increase in MAP in vivo and the in vitro rate of ^·NO-induced oxidation. These results suggest a causal relationship exists between ^·NO reactivity and the vasopressor response, and by modulating the reactivity of hemoglobin toward ^·NO, one can reduce the hemoglobin-induced vasopressor effect.

We have studied six different cell-free hemoglobin preparations and have found three distinct types of MAP response during isovolemic 50% exchange transfusion in rats (Fig. 1 and Table I): 1) an immediate and sustained increase (-Hb, Tm-Hb, o-R-poly-Hb); 2) an immediate and transient increase (82-Hb, PHP); and 3) no significant increase (PEG-Hb). Three MAP responses imply that differences in one or more physical properties of these hemoglobin solutions must exist. If the differences in MAP response are due to differences in rates of ^·NO scavenging, then these hemoglobins are expected to display different ^·NO reaction rates and/or affinities that correlate with the different MAP responses. The value of the rate constant, or affinity, for the reaction of PEG-Hb with ^·NO would be expected to be less than the corresponding values for any of the other hemoglobins in this study. The transient MAP increases seen with PHP and 82-Hb would be explained by intermediate rates or levels of ^·NO scavenging, and the large or sustained MAP increases seen with -Hb, Tm-Hb, and o-R-poly-Hb would correspond to the largest rates of reaction, or overall affinities with ^·NO. The results of this study, however, show that differences in MAP response (Fig. 1) do not correlate with ^·NO reaction rates and inversely correlate with ^·NO affinities (Table II, Figs. 1 and 2).

The values of the rate constants determined for the ^·NO-induced oxidation reaction are consistent with those obtained for the bimolecular association reaction with the deoxyhemoglobins. In their recent study of the mechanism of ^·NO-induced oxidation of hemoglobin, Eich et al. (19) reported bimolecular rate constant values in the range 30-50 µM¹ s¹ for the reaction of ^·NO with wild-type human Hb-O₂, which are slightly larger than the values obtained in this study. We cannot account for the minor discrepancy in these two studies. However, it is apparent that under identical reaction conditions, the rate constants for the reaction of ^·NO with both the deoxy- and oxy-forms of all of the chemically modified hemoglobins examined in this study, and unmodified human hemoglobin, are the same.

The dissociation rate constants are the only measurable difference in ^·NO binding parameters (Table II). The ^·NO dissociation time courses are biphasic with unequal amplitudes for the rapid and slow kinetic phases (Table II). This behavior has been reported for native human hemoglobin under conditions that decrease oxygen affinity (i.e. low pH, the presence of inositol hexaphosphate) and have been interpreted as being due to cooperative dissociation of ^·NO (16, 23, 24). The kinetic heterogeneity observed in this study suggests cooperative dissociation of ^·NO and the existence of significant levels of both high affinity (R) and low affinity (T) Hb-NO conformational states which coexist in equilibrium. As ^·NO dissociates from hemoglobin in the presence of CO, a solution of Hb-NO that is partially in the high-affinity (R) state and partially in the low-affinity (T) state is converted to carbonomoxyhemoglobin which is entirely in the high-affinity (R) state.

The observed rate constants for the slow kinetic phase of the ^·NO dissociation time courses are approximately the same for all the hemoglobins studied suggesting that there is little difference in the rates of ^·NO dissociation from R-state Hb-NO. Differences in overall ^·NO dissociation rate constants appear to be controlled by the differences in the rate constant for ^·NO dissociation from T-state Hb-NO. The lowest values for overall ^·NO dissociation rate constants are observed with HbA₀ and 82-Hb, and are due to a combination of both a smaller fraction of T-state Hb-NO and a relatively low value for the rate constant for ^·NO dissociation from T state. It is also the case that there is no difference in the association rate constants among the hemoglobins studied. Differences in the overall equilibrium dissociation constants, therefore, are also due to differences in the rate constants for ^·NO dissociation from T-state Hb-NO.

The correlation between overall ^·NO affinity in vitro and MAP response in vivo appears to be the opposite of what is expected, based on the hypothesis that an ^·NO scavenging mechanism accounts for MAP responses and vasoconstriction. Hemoglobins that are observed to cause the largest and/or most persistent increases in MAP (-Hb, Tm-Hb, and o-R-poly-Hb) exhibit the weakest ^·NO binding. Hemoglobins that exhibit either transient or no MAP increase exhibit the highest ^·NO affinity. However, the differences observed upon 50% exchange transfusion must arise from differences in one or more physical properties exhibited by these solutions. For example, these hemoglobin solutions display different O₂ binding parameters, molecular weights (Table I) and different solution properties including viscosity and colloid osmotic pressure (25, 26).

Because of their size and solution properties, the surface modified hemoglobins (PEG-Hb and PHP) occupy a much greater molecular volume in solution than do cross-linked tetrameric hemoglobins (25). This may be important if the mechanism for cell-free hemoglobin-induced vasoconstriction requires movement of the hemoglobin from the circulation to the extravascular space. It is possible that ^·NO scavenging is important only when hemoglobin enters the interstitial space between endothelial cells and smooth muscle cells. If this is the case, then the lack of MAP increase observed with PEG-Hb and the reduced MAP response observed with PHP may be due to molecular size which may prevent passage through endothelial cell junctions.

There is little experimental work that directly assesses the ability of cell-free hemoglobin to diffuse out of the lumen. Differences in rates of disappearance of Evans blue dye from the plasma of rats treated with PEG-Hb or -Hb suggest that greater vascular leaking is induced by -Hb than by PEG-Hb (7). Fluorescently labeled PEG-Hb has been shown to extravasate within minutes in the intestine of anesthetized rats despite observations that suggest that this type of junction is too small to permit passage of PEG-Hb (27). However, in both these cases, the time for extravasation appears to be too long to account for the immediate MAP increases seen in this study. Additionally, it is unlikely that the difference in MAP response between 82-Hb and the low O₂ affinity tetramers -Hb and Tm-Hb (Fig. 1) is due to differences in extravasation since these hemoglobin molecules are structurally so similar. The differences in MAP response between these hemoglobin solutions are probably not due to differences in ability to diffuse out of the lumen. However unlikely, we cannot discount the possibility that the differences in MAP response are due to differences in extravascular ^·NO scavenging, and further study of the interaction between cell-free hemoglobin and the endothelium is required.

It has also been suggested that oxygen-linked S-nitrosylation reactions of hemoglobin may be important in regulation of vascular tone (28, 29). It is possible that different MAP responses may be due to differences in the abilities of chemically modified hemoglobins to participate in these reactions. However, it is not clear whether S-nitrosylation reactions are relevant to extracellular hemoglobin. It also appears that even in the presence of low molecular weight thiols, transnitrosylation reactions occur on time scales that are too slow to account for the immediate MAP increases observed in this study.

Mechanical forces associated with blood flow may play important roles in the control of vascular tone (30). When flow increases, shear stress increases, causing arteries to dilate by endothelial-dependent, nervous-system-independent, relaxation of the smooth muscle cells. For a Newtonian fluid, shear stress is directly proportional to viscosity. In the case of blood (non-Newtonian), or mixtures of blood and cell-free hemoglobin solutions which are Newtonian (26), shear stress increases with increasing viscosity. The effects of solution viscosity on O₂ delivery have been observed by comparing tissue oxygenation following hemodilution with either colloids or with crystalloids (31). The differences in MAP responses observed in this study are consistent with this model of shear stress-mediated vasoregulation. PEG-Hb and PHP, which elicit either transient or no MAP increase, exhibit significantly higher viscosities than the other hemoglobin solutions when measured at the same concentration and shear rate (26).

We find a direct correlation between ^·NO and O₂ affinities (Fig. 2). Interestingly, it has been suggested that the presence of a low oxygen affinity cell-free hemoglobin solution in vivo may not necessarily lead to a corresponding high level of O₂ delivery to the tissues (32-34). Cell-free hemoglobin may overcome a diffusive limitation to O₂ delivery (35) resulting in abnormally high O₂ levels in the regulatory arterioles. This has the potential to produce an autoregulatory response in which vasoconstriction causes increased vascular resistance and reduced flow. In this model, the shape and position of the O₂ equilibrium binding curve, the total O₂ carrying capacity, and the fluid viscosity are crucial in determining the amount of O₂ delivered to the regulatory regions of the microcirculation (6).

The differences in MAP response observed in this study are consistent with the hypothesis that MAP responses are, in part, dictated by the O₂ binding properties. A comparison of the intramolecularly cross-linked hemoglobin tetramers illustrates the effect of O₂ binding properties. Solutions of Tm-Hb and -Hb both produce an immediate and sustained increase in MAP (Table I, Fig. 1). These hemoglobins are structurally very similar, exhibit identical solution properties, at the same concentration (25, 26), and similar O₂ affinity and cooperativity (Table I). The physical properties of 82-Hb are also nearly identical to those of Tm-Hb and -Hb, with the exception of the O₂ binding properties. 82-Hb binds O₂ with higher affinity and less cooperativity than either Tm-Hb or -Hb (Table I). Solutions of 82-Hb elicit a diminished and transient MAP response relative to Tm-Hb and -Hb. Thus we find a direct inverse correlation between O₂ affinity and vasopressor response. The hemoglobins that cause sustained MAP increases exhibit the highest p₅₀ values (lowest O₂ affinity) whereas the hemoglobins that cause either transient or no MAP increase exhibit the lowest p₅₀ values (highest O₂ affinity, Table I and Fig. 2).

Physiological control of vascular smooth muscle tone relies on many factors, any or all of which may be affected by the presence of a cell-free hemoglobin. This study has addressed only one such factor: the ability of a cell-free hemoglobin to react with ^·NO at the heme ligand binding site, and has found no correlation between ^·NO reactivity and MAP response. It is more likely that the O₂ affinities and solution properties of hemoglobin solutions are more important in determining the vasopressor effect. The full physiological relevance of ^·NO scavenging reactions will only be discovered through continued investigations of the mechanisms involved in regulating blood pressure.