INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

ACELLULAR MODIFIED HEMOGLOBIN (Hb) solutions, presently in advanced stages of clinical trials, may soon be accepted as substitutes for blood transfusions in surgical procedures and treatment of trauma patients (49, 54). However, as clinical trials are extended to include larger numbers of individuals, it becomes apparent that the principal side effect consistently reported in the administration of acellular Hb solutions is hypertension presumably because of vasoconstriction.

Hypertension, a well-defined reaction of the acellular intramolecularly cross-linked Hb (XLHb), was proposed to be beneficial in the treatment of hypotension concomitant to hemorrhagic shock (33). However, vasoconstriction reduces blood flow, lowering functional capillary density, and therefore affecting tissue perfusion and oxygenation (10, 48). Furthermore, maintenance of functional capillary density per se, independently of tissue oxygenation, has been shown to be critical to tissue survival in hemorrhagic shock (19). Nitric oxide (NO) scavenging by Hb due to intrinsic high affinity of NO to Hb is the mechanism presumed to cause vasoconstriction and hypertension (7, 26, 45, 50). This theory was validated indirectly using exteriorized rabbit aortic rings in organ baths, where constriction was observed following the addition of acellular Hb solutions as well as an NO synthase inhibitor (10, 34, 36). Different modifications of the Hb molecule cause hypertension that is qualitatively and quantitatively different, and red blood cells (RBCs) and cellular liposome-encapsulated Hb (Hb vesicles) do not cause either vasoconstriction or hypertension (6, 36, 42).

Analysis of the reported data on different commercially developed Hb modifications suggests an inverse relationship between Hb molecular size and the extent of the pressor response, an effect that could be related to the dynamics of NO-to-Hb interactions. To systematically explore this relationship, we produced O₂-carrying Hbs of which their principal difference was their molecular sizes. Because molecular size is a determinant of viscosity, we infused the solutions as a bolus of 10% of blood volume, thus not significantly changing blood viscosity and shear stress.

Most evidence for the pressor response is obtained from measurements of systemic pressure, and direct evidence about the mechanism involved is scarce. In previous studies in conscious hamsters fitted with a dorsal skinfold, we found that small arteries of 130-160 µm diameter, termed resistance vessels, exhibit the greatest reactivity in hemorrhagic shock (39), playing a significant role in the regulation of blood flow. Constriction of these resistance vessels in our model was also directly correlated to the pressure response following administration of NO synthase inhibitor (38).

In this study we aim to determine whether resistance vessel constriction is related to hypertension after administration of acellular Hb and the extent to which the effect is dependent on the size of acellular Hb molecules modified by polymerization, polymer conjugation, and cellular liposome encapsulation.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Preparation of chemically modified acellular Hbs. Hb was purified from outdated donated blood obtained from the Hokkaido Red Cross Blood Center in Japan (40). The purification procedure included carbonylation, heat treatment at 60°C, ultracentrifugation (50,000 g, 40 min), ultrafiltration using 100-kDa membranes, and dialysis with 10-kDa membranes. Carbonylhemoglobin is stable at 60°C so that the heat treatment enables denaturation of other concomitant proteins and virus inactivation. Even though heating for 1 h is enough to remove other proteins (41), the current heating period is 10 h to ensure virus inactivation. This is the same condition for the clinically used human serum albumin. The purified Hb solutions were checked for the presence of residual phospholipids by extracting lipids with modified Bligh and Dyer's method and HPLC (41), where the removal efficiency was >99.4%, the remaining phospholipid being <4 µg/ml in the purified 5% Hb solution. Endotoxin content was measured by the conventional Limulus amoebocyte lysate test and was found to be <0.1 EU/ml in the Hb solution.

Physicochemical characterization of Hb products. The Hb preparations were characterized in terms of O₂-binding properties, viscosity, diameter, molecular weight, oncotic pressure, contents of metHb, and monomeric Hb. Oxygen affinities (P₅₀) were calculated from the O₂ equilibrium curve obtained with a Hemox Analyzer (TCS Medical Products). Particle sizes were measured by light scattering (model N4SD, Coulter particle analyzer). Oncotic pressure was measured with a model 4420, Wescor osmometer. Viscosity was measured with a capillary viscometer (Oscillatory Capillary Rheometer and Density Meter, OCR-D, Anton Paar, Austria) at 37°C. Number average molecular weights (M_n) were calculated from oncotic pressure measurements (52). Weight average molecular weights (M_w) were measured with a multiangle light scattering (miniDAWN DSP, Wyatt Technology) equipped with a size exclusion column (Shodex Protein KW-804). The presence of remaining monomeric XLHb in HES-XLHb and poly-XLHb, and PLP-Hb in PEG-PLP-Hb was detected by size exclusion chromatography (TOSO TSKgel G3000SWXL), with an eluent of phosphate-buffered saline (pH 7.4), a flow rate of 1 ml/min, and a detection wavelength at 419 nm.

Animal model and preparation. Experiments were carried out in 36 male Syrian golden hamsters weighing 66± 8 g (Simonsen, Gilroy, CA). All animals were housed in cages and provided with food and water ad libitum in a temperature-controlled room on a 12:12 h light-dark cycle. After the animals were anesthetized with intraperitoneally administered pentobarbital sodium (ca. 100 mg/kg body wt, Abbott, North Chicago, IL), the dorsal skinfold consisting of two layers of skin and muscle was fitted with two titanium frames with a 15-mm circular opening and surgically installed. A location that included a paired small artery and vein was selected. The resistance artery can be readily identified because a Y-shaped pair of artery and vein can be seen visually when the hamster dorsal skin is extended after the hair has been removed (39). Layers of skin muscle were separated from the subcutaneous tissue and removed until a thin monolayer of muscle, including the small artery and vein, and one layer of intact skin remained. A cover glass (diameter 12 mm) held by one frame covered the exposed tissue allowing intravital observation of the small artery (A₀, diameter 158 ± 21 µm), which is the main feeding vessel in this tissue, large feeding arterioles (A₁, 63 ± 12 µm), and small veins (V₀, 364 ± 73 µm).

Infusion of Hb samples. The infusion volume was 7 ml/kg, which is 10% of the total blood volume of a hamster (70 ml blood/kg). Therefore, the amount of Hb infused was 350 mg/kg because all Hb solutions had [Hb] = 5 g/dl ([heme] = 3.1 mM). Six experiments were made with each material. Human serum albumin solution (HSA, 5 g/dl, Bayer) was used as a control. Sample solutions of ~0.4 ml were infused through the jugular vein at a rate of 0.4 ml/min. The catheter was flushed with a small amount of saline.

Characterization of systemic conditions. Mean arterial pressure (MAP) and heart rate (HR) were recorded in analog format (Beckman R611, Beckman Instruments, Schiller Park, IL) from a transducer connected via an arterial catheter.

Microhemodynamic analysis. The microvasculature was observed with an inverted microscope (IMT-2, Olympus, Tokyo, Japan) using a ×10 objective (Olympus) and a ×40 water immersion objective (Olympus, Wplan) and transillumination. Microscopic images were recorded by video (Cohu 4815-2000, San Diego, CA) and transferred to a TV-VCR (Sony Trinitron PVM-1271Q monitor, Tokyo, Japan) and a Panasonic AG-7355 video recorder (Tokyo, Japan). Microvascular diameter and center-line RBC velocity were analyzed on-line in arterioles and venules (15). Diameter was measured with an image-shearing system (Digital Video Image Shearing Monitor 908, IPM, San Diego, CA), whereas RBC velocity was analyzed by photodiode and the cross-correlation technique (Velocity Tracker Mod-102 B, IPM). Blood flow rates () were calculated by means of Eq. 1

(1)

where R_v is the ratio of velocity in the middle of vessels to whole blood velocity. We selected R_v = 1.6 according to Lipowsky and Zweifach (22), who specified this value for vessels from 13 to 90 µm diameter, but not for larger vessels. Because all changes are normalized relative to the basal values, the results are not affected by the value of R_v selected.

Data analysis. Differences between treatment groups were analyzed using a one-way ANOVA followed by Fisher's protected least significant difference test. A paired t-test was used to compare the time-dependent changes within each group. The changes were considered statistically significant if P < 0.05. Changes in MAP and HR are expressed as differences from the basal value. Changes in microvascular diameter and blood flow rates were expressed as the percentage of basal values.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Physicochemical characteristics of Hb-based O₂ carriers. The physicochemical characteristics of the Hb molecules in solution used in these experiments are presented in Table 1. Their molecular diameters were 7 ± 2 for XLHb, 22 ± 2 for PEG-PLP-Hb, 47 ± 17 for poly-XLHb, 68 ± 24 for HES-XLHb, and 224 ± 76 nm for PEG-HbV. XLHb (5 g/dl) had an oncotic pressure of 15.8 mmHg. Polymerization of XLHb reduced this to 2.5 mmHg at the same concentration. PEG-PLP-Hb had the largest oncotic pressure (70.2 mmHg) and the highest viscosity (6.1 cP at 332 s¹) due to the highly hydrated PEG chains (52), whereas that of PEG-HbV was close to zero because the number of particles in suspension is significantly reduced. The larger particle size of HES-XLHb (68 ± 24 nm) and higher oncotic pressure than poly-XLHb are due to the additional highly hydrated HES chains and more expanded structure. The M_w and M_n of poly-XLHb and HES-XLHb are significantly different due to the wide molecular weight distribution.

Systemic responses. Control MAP for all groups was 106 ± 8 mmHg. The XLHb group increased blood pressure immediately after infusion (+12 ± 8 mmHg relative to baseline), which was sustained for 3 h (+34 ± 13 mmHg) (Fig. 1). This effect subsided after 24 h. The HSA and PEG-HbV group had no significant changes. The PEG-PLP-Hb and poly-XLHb groups had intermediate responses. Heart rates of all the groups tended to decline after infusion; the heart rate of the XLHb group decreased by 84 ± 44 beats/min 2 h after infusion. The PEG-HbV group exhibited the same level of bardycardia as the HSA group.

View larger version (39K): [in this window] [in a new window]   Fig. 1.   Changes in mean arterial pressure (MAP, A) and heart rate (HR, B) after infusion of different hemoglobin (Hb) products (HSA, human serum albumin; HES, hydroxymethyl starch; XLHb, cross-linked Hb; PEG, polyethylene glycol; HbV, Hb vesicles, PLP, pyridoxal 5'-phosphate). The time point 0 h indicates <30 s after completion of infusion. Values are means ± SE. *Significantly different from baseline. #Significantly different from the PEG-HbV group. The baselines of MAP and HR were 106 ± 8 mmHg and 414 ±47 beats/min, respectively.

Microvascular responses. Significant constriction was observed in A₀ of resistance arteries (Fig. 2) but not in smaller arterioles (A₁) nor capacitance venules (V₀). For example, the diameters (% of basal value) of A₁ and V₀ in each of the groups after 2 h were as follows: XLHb (97±5 and 99 ±4), PEG-PLP-Hb (101 ±4 and 102 ±1), poly-XLHb (106 ±10 and 100 ±3), HES-XLHb (101 ±2 and 104 ±2), and PEG-HbV (101 ±4 and 103 ±4). The changes in diameter of the A₀ vessels and MAP were plotted against the molecular sizes of Hb samples (Fig. 3). Vasoconstriction, which was most significant in the XLHb group, decreased with increasing molecular size; the HES-XLHb group showed results similar to those of the PEG-HbV group. Blood flow rates of the XLHb group declined significantly in A₀ vessels (36 ±17% of basal value at 3 h) and tended to return to control values after 24 h. The PEG-HbV and HSA groups showed small flow and A₀ diameter changes. The remaining Hb groups showed intermediate changes depending on their sizes. The A₀ vessels of the poly-XLHb group constricted reaching 87 ± 7% of the basal value at 3 h, and their flow reduced to 54 ±16% at 2 h. The HES-XLHb group showed the smallest changes of all modified Hbs, and effects were similar to those of the PEG-HbV group.

View larger version (40K): [in this window] [in a new window]   Fig. 2.   Changes in diameter (A) and blood flow rates (B) in the resistance artery (A0) after infusion of different Hb products. The time point 0 h indicates <30 s after completion of infusion. Values are means ± SE. *Significantly different from baseline. #Significantly different from the PEG-HbV group. The baselines of diameter and blood flow rate were 158 ± 21 µm and 316 ± 250 nl/s, respectively.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Influence of particle diamters of Hb products on MAP (A) and A0 diameter (B). The time point 0 h indicates <30 s after completion of infusion. The data are converted from Figs. 2 and 3.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Our principal finding is that the hypertensive effect following the administration of acellular Hb solutions is directly correlated to vasoconstriction of the resistance arteries (A₀ vessels) and that the magnitude of the effect is an inverse function of the molecular size of the material. Vasoconstriction in our model was circumscribed to the A₀ vessels and did not extend to the remainder of the microvasculature.

Our findings agree with the concept that the resistance arteries are as important as the arteriolar network in regulating peripheral blood flow (27). This is supported by studies in hypertensive rats where large arterioles and small arteries, and not small arterioles, are responsible for the increase in total organ vascular resistance (2, 25, 31). Similarly, in previous studies in our model of conscious hamsters we found that A₀ resistance arteries are the most reactive in hypotensive shock (39). In a recent study, we infused 10 mg/kg of the NO synthase inhibitor N-nitro-L-arginine methyl ester (L-NAME) and found the immediate onset of hypertension (+22 mmHg) and simultaneous constriction of A₀ vessels (30%), whereas the downstream arterioles A₁-A₃ did not constrict (38). Thus we can presume that the resistance arteries are the vessels most sensitive to NO and crucial in determining blood pressure and flow when NO production is inhibited. Vasoconstriction induced by the NO synthase inhibitor N^G-monomethyl-L-arginine was reported for dog coronary artery, human brachial artery, and hamster cheek pouch arterioles (26), all of which are resistance vessels. Stewart et al. (46) reported the constriction of coronary resistance vessels in the isolated rabbit heart by native Hb. Griffith et al. (12) showed that rabbit ear small arteries (diameter range 150-700 µm), where resistance and shear stress are maximal, have the highest sensitivity to endothelium-derived relaxation factor.

Although studies using aortic rings demonstrated vasoactivity of Hb and the NO synthase inhibitor (29, 36), it was recently reported (3) that the aorta in anesthetized rabbits did not constrict after the infusion of the NO synthase inhibitor L-NAME or dextran-conjugated Hb. Thus the peripheral arteries, and not the aorta, constrict to induce hypertension. Nolte et al. (32) reported that arterioles smaller than 60 µm in diameter did not remain constricted even though hypertension was long lasting after infusion of XLHb. These results indirectly support the concept that resistance arteries between the aorta and arterioles are crucial in Hb-induced vasoconstriction and resulting hypertension.

The vasoconstrictive responses in our study may be in part explained in terms of NO scavenging by Hb (13) and the relative capacity of the different molecules to extravasate. Nakai et al. (29) has shown that the molecular size of Hb products is a factor in determining their passage across the endothelial cell layer. In this context, the smallest sized XLHb would be the most permeable and would show a higher level of vasoconstriction and hypertension than PEG-Hb and liposome-encapsulated Hb. Smaller Hb molecules may extravasate in greater numbers and thus bind a greater amount of NO, whereas the larger HbV particles with diameters of about 250 nm would extravasate in significantly fewer numbers. Poly-XLHb and HES-XLHb, which are larger than XLHb, caused less vasoconstriction and correspondingly an intermediate level of hypertension, as shown in Fig. 3.

A large pressor effect was recently reported for 50% exchange transfusions of Dex-BTC-Hb, a combination of dextran and Hb with a 300-kDa mean molecular mass (8). This effect was found to correlate with the rapid penetration of Hb in the endothelium, although it was not determined whether Hb or Dex-BTC-Hb had penetrated these cells. It should be noted that the molecular weight of Dex-BTC-Hb is much smaller than that of HES-XLHb and poly-XLHb in our study. PEG conjugation of XLHb resulted in reduced hypertensive responses in rats according to Abassi et al. (1), although a similar product has been proposed for the treatment of septic shock to increase blood pressure (9). Glutaraldehyde-polymerized bovine Hb and o-raffinose-polymerized human Hb, which are currently under clinical trials, are also reported to induce hypertension (17, 35). The molecular dimensions of these products are much smaller than those of HES-XLHb and PEG-HbV in our study.

We measured the plasma retention half-life of XLHb, poly-XLHb, and PEG-HbV at the same dosages in Wistar rats (Table 1). XLHb had a half-life of 1.6 h, whereas poly-XLHb and PEG-HbV had half-lives of 3 and 4 h, respectively. The shorter half-life of XLHb suggests a greater tendency for extravasation and therefore an increased potential for inducing vasoconstriction. About 20% of XLHb remained in the circulation after 3 h and was not detectable after 8 h, which was consistent with the observation that hypertension lasted for 3 h and was not present after 24 h.

Duration and the level of vasoconstriction and hypertension may be partly related to the concentration of Hb in plasma (24). In some reports an immediate response after the infusion of Hb was observed, although the maximal effect occurred 0.5 or 1.0 h after infusion, even though the XLHb concentration was the highest just after the infusion (24, 45). In our experiments peak hypertension was seen at 3 h and peak vasoconstriction was at 2 h after infusion; however, there was no significant difference among 1, 2, and 3 h. Schultz et al. (45) have shown that significant hypertension was maintained over 1.5 h at 280 mg/kg infusion of XLHb in rats, which correlates with our findings in which 350 mg/kg of XLHb showed hypertension for 3 h. Our results agrees with those of Malcolm et al. (24), who indicated that the bolus infusion in 10 s induced a longer and higher level of hypertension than a slower 4-min infusion. We infused XLHb in 1 min, which may explain the duration of hypertension. Isolated femoral arterial strip experiments show that a Hb concentration of 1 µM is enough to induce vasoconstriction (18). In our experiment about 20% of infused XLHb remains in the blood after 3 h (~16 µM); therefore, the sustained hypertension and vasoconstriction may be related to the presence of Hb.

An alternative explanation is that the scavenging effect may be operational directly on the blood side of the endothelium, where the presence of molecular Hb in the RBC-free plasma layer could significantly distort the diffusion field from the endothelial cell, diverting NO from smooth muscle into blood according to Vaughn et al. (51). In normal blood there is an RBC-free layer plasma region next to the endothelium, and in this region NO is not consumed. For the vessels in our study, this region is ~2.5-µm thick (44), and the presence of this layer introduces a resistance to the diffusion of NO to the site of scavenging the RBCs. The presence of Hb in this plasma layer should significantly increase the rate of NO binding, lowering the availability of NO to the smooth muscle. Verification of this mechanism requires measurement of the NO reaction rate with the different Hbs used.

Measurements by the flash photolysis method show that modified Hb molecules with molecular dimensions ranging from normal Hb to that of PEG-conjugated Hb have similar Hb-NO binding rates (about 3.0×10⁷ M¹s¹ in Ref. 35); however, the binding rate of larger size molecules was not determined. The characteristic of PEG-conjugated Hb is that the PEG chains are highly hydrated, resulting in a large excluded volume and a significantly larger molecular diameter than XLHb. However, the NO-binding rate measurement indicated that NO molecules can diffuse in the PEG-water layer to bind Hb without any hindrance. In our study we compared the binding rate of unmodified Hb and Hb vesicles by the stopped flow method and we found the binding rate constants to be 3×10⁷ M¹s¹ and 4.8×10⁶ M¹s¹, respectively (37), indicating that the surface-to-volume ratio of the particles plays a role in determining the NO-trapping rate; thus Hb encapsulation may significantly contribute to retarding NO binding by the same mechanism that retards O₂ binding (5). The O₂ binding rate is related to particle size (53), and that of HbV is one order of magnitude slower than acellular Hbs, although faster than RBCs. Thus the vasoconstrictive effect does not appear to be explained solely on the basis of NO-binding rates, because the rates of HbV and RBC are different. However, neither of them causes vasoconstriction. Therefore, the relationship between the molecular dimension and permeability across the endothelial cell layer should be considered in parallel.

In our experiment the amount of the infused Hb was one-thirtieth of blood Hb, and the solution volume was 10% of blood volume to minimize the effects due to changes in viscosity, oncotic pressure, and O₂ affinity. Viscosity is governed by the RBC concentration, and this small dilution did not affect blood viscosity and therefore shear stress on the vascular wall. PEG-PLP-Hb has an oncotic pressure of 70.2 mmHg, which may have increased blood oncotic pressure by about 7 mmHg, possibly causing a small increase in vascular volume through autotransfusion.

Remaining phospholipids and endotoxin contamination during Hb purification may cause hemodynamic effects; however, endotoxin contamination reduces blood pressure (20). Macdonald et al. (23) reported unpurified stroma-free Hb induced more significant coronary vasoconstriction than purified Hb. We determined the phospholipid and endotoxin concentrations in the purified Hb solution but not in the final Hb solutions. However, all the materials were prepared from the same purified Hb; thus the response due to XLHb infusion is unlikely to be due to a different level of contamination. Gulati et al. (14) have shown that stroma free Hb with phospholipid contamination of more than 50 µg/ml induced renal failure, although hypertension was reduced; however, our samples should be significantly below this value.

Winslow (54) observed that when even relatively small amounts of Hb are dissolved in plasma by comparison to that present in RBCs (~0.01 g Hb/ml plasma in our case), the amount of O₂ present in plasma is significantly increased over that solely due to solubility of O₂ in plasma due to the chemical binding of Hb and O₂. This situation has a profound effect on the amount of O₂ that is transported from the RBCs to the vessel wall, across the plasma layer, because in the presence of dissolved Hb, O₂ is transported by both the concentration gradient of O₂ in the plasma layer and the concentration gradient of HbO₂, according to the process of facilitated diffusion (30). The additional O₂ flux due to facilitated diffusion is significant, even though the diffusion constant for the Hb molecule is much smaller than that of O₂ in plasma. As an example, in our experiments the amount of O₂ bound to free Hb in plasma is about 5 times that of the O₂ dissolved in plasma, whereas the diffusion constant of HbO₂ is 0.02 times of that of O₂ (21). The net result is that the presence of Hb in the plasma increases the delivery of O₂ to the vessels wall by about 20-10%, depending on the partial pressure of O₂ of blood in the vessels, a situation that could induce autoregulatory constrictive compensation. It should be noted that this effect is inversely proportional to molecular size and therefore appears to operate in accordance to our findings.

In summary, our study shows a direct correlation between the hypertensive response due to the infusion of acellular Hb solutions and the constriction of resistance arteries. This effect was decreased with increasing molecular size and not observed for Hb vesicles. The decrease of NO availability due to NO scavenging and increased O₂ delivery to the vessel wall due to facilitated diffusion are possible mechanisms that can account for the vasoconstrictive effects found with the smaller Hb molecules.