Targeted O₂ Delivery by Low-p50 Hemoglobin: A New Basis for Hemoglobin-based Oxygen Carriers

Abstract

We have proposed new criteria for a successful cell-free, hemoglobin-based O₂ carrier. These include increased molecular radius, increased viscosity, increased oncotic pressure, and reduced p50. A new molecule, MalPEG-Hb, formulated at 4.2 g/dL in lactated Ringer's solution (MP4), has been produced according to these new criteria. MP4 has a p50 of 5–6 mm Hg, oncotic pressure of 49 mm Hg and viscosity of 2.2 cPs. After 50% exchange transfusion with MP4, rats survive a 60% controlled hemorrhage in spite of total hemoglobin of 7.8 g/dL and plasma hemoglobin concentration of 1.6 g/dL. This model results in 50% mortality in control animals and 100% mortality in animals exchange-transfused with either crosslinked or polymerized hemoglobin. Oxygen supply to tissue was measured directly in the hamster skinfold model, in which O₂ release in precapillary and capillary vessels can be quantified. The data demonstrate that the effectiveness of MP4 results from its ability to conserve O₂ in precapillary vessels and release O₂ in capillaries, thereby “targeting” O₂ to hypoxic tissue. Preservation of functional capillary density and prevention of vasoconstriction further contribute to the effectiveness of this new formulation.

BACKGROUND: TRADITIONAL DESIGN STRATEGIES

This presentation will describe a new approach to the design of oxygen carriers based on cell-free hemoglobin, particularly the concept of low p50 and the avoidance of vasoconstriction. The work to be described is the result of collaborations between our group at Sangart, Inc. (Drs. Vandegriff and Young), the Department of Bioengineering at UCSD (Drs. Intaglietta, Tsai and Johnson), Albert Einstein College of Medicine (Drs. Acharya and Manjula), and The Karolinska Institute, Stockholm (Drs. Fagrell, Kjellstrom, Drobin and Hahn).

Clinical deployment of a cell-free oxygen carrier is, of course, a long-held goal, and enormous strides have been made since the early 1980s when fears were raised about contamination of the blood supply with HIV. Efforts such as those by the U.S. Army, certain companies and some academic centers led to the state of the art a decade later, when certain basic elements were held as dogmatic [[16]]. These were that:

1. Hemoglobin must be scrupulously purified to free it from all red cell enzyme and membrane components. This concept was based on the work primarily of Rabiner [[10]] and others who showed that the toxicity of cell free hemoglobin could be reduced by scrupulous purification.

2. Hemoglobin must be chemically modified to prevent subunit dissociation and subsequent renal clearance, which could result in further toxicity [[2]].

3. Modified hemoglobin should have the same oxygen affinity as cellular hemoglobin to ensure tissue oxygen delivery. The basis for this concept is not easily traced, but appears to be rooted in the belief that the oxygen equilibrium curve (OEC) shifts right (lower oxygen affinity, higher p50) in hypoxia, and that this regulation is by 2,3-DPG, a molecule that is present only within red blood cells. Furthermore, carriers of high oxygen affinity hemoglobin mutants develop polycythemia, resulting from inadequate kidney oxygenation.

4. Hemoglobin solutions should have the lowest possible viscosity in order to minimize vascular resistance and permit maximal blood penetration into capillaries. This concept is based on early studies in which cardiac output increased with peripheral resistance was lowered by hemodilution [[4]].

5. Hemoglobin solutions should have an oncotic pressure similar to that of plasma or normal blood in order to avoid overexpansion of the blood volume, particularly in elderly patients.

6. Hemoglobin solutions should, above all, not be vasoactive. By 1992, vasoactivity was thought to result from nitric oxide scavenging by hemoglobin, and therefore any successful hemoglobin-based product must not bind nitric oxide.

EVIDENCE THAT A NEW PARADIGM IS NEEDED

In the early 1990s, our research group at the University of California San Diego began to study vasoactivity of a number of modified hemoglobins that had been proposed as blood substitutes. We devised a particularly severe model of rat hemorrhage [[17]] with 50% untreated mortality (Figure 1). Rats were first exchange-transfused 50% of the estimated blood volume with test material. After a short equilibration (∼30 minutes), an exponential, controlled hemorrhage of 60% of estimated blood volume via an arterial catheter was carried out over 1 hour. Animals were observed for an additional hour before euthanasia of the survivors.

Figure 1 A stringent model of efficacy. Rats are exchange-transfused 50% of blood volume, then a controlled, exponential 60% hemorrhage is done over 60 minutes. Animals are observed for an additional hour. The hemoglobin concentrations at the right of the figure are at the start of the 60% hemorrhage. Modified from [[17]].

In control animals (sham), the hemoglobin concentration was 13.8 g/dL at the start of hemorrhage, and 50% of these animals died before the end of the observation period. Animals that received Pentastarch^®, a hyperoncotic non-oxygen carrier, had only 6.8 g/dL hemoglobin at the start of hemorrhage, and had worse survival, as expected. Animals that received the Army's αα-crosslinked hemoglobin (αα-Hb) had a total hemoglobin of 10.2 g/dL at the start of hemorrhage, of which 3.6 g/dL was contained in the plasma. Survival of these animals was worse than either the controls or Pentastarch^® animals, confirming our supposition that vasoconstriction offsets any advantage of increased oxygen capacity. Hemoglobin polymerized with ring-opened raffinose had a total hemoglobin of 11 g/dL per deciliter after the exchange, 4.9 g/dL of which was free in the plasma. These animals seemed to show slightly better short–term survival compared with Pentastarch^® or ααHb, but all animals were dead by the end of the observation period. In contrast to all these groups of animals, MalPEG-Hb (MP4) was spectacularly successful: all animals survived the exchange and observation periods, in spite of a total hemoglobin concentration of 7.8 g/dL, with only 1.6 g/dL as plasma hemoglobin.

We postulated that the basis of this success was lack of a hypertensive response to MP4. The popular explanation for hypertension and vasoconstriction following hemoglobin infusion is that hemoglobin avidly scavenges nitric oxide (NO), which is a natural endothelium-derived relaxing factor. Figure 2 compares three representative hemoglobin solutions infused into the rat as top loads. The greatest increase in blood pressure occured in animals that received αα-Hb. An intermediate response was seen in animals that received o-raffinose polymerized hemoglobin, and little if any blood pressure response was seen in animals that received hemoglobin surface modified with polyethylene glycol (MP4). Drs. Rohlfs and Vandergriff in our laboratory found the NO reactivity of these three molecules to be identical. Thus the different blood pressure responses could not be explained on the basis of the NO scavenging alone, and a new theory was needed.

Figure 2 Hypertensive response in the rat to exchange transfusion with three representative modified hemoglobins. The molecules are drawn to scale at the right and the reaction rate constants for NO binding are shown for each. Data from [[11]].

THE NEW PARADIGM: FACILITATED DIFFUSION AND AUTOREGULATION

One basis of the new paradigm that emerged from these findings was the concept of “autoregulatory” vasoconstriction: as is well known, microcirculatory vasoconstriction is triggered by excessive oxygen supply [[8]]. The second basis of the new paradigm is the extensive literature on so-called “facilitated diffusion” of myoglobin and hemoglobin. We were surprised to calculate that by adding even small amounts of hemoglobin to plasma, the O₂ concentration increased markedly. In fact, as little as 1 g/dL of plasma hemoglobin doubles the concentration of oxygen in plasma at a pO₂ 100 mm Hg. This startling result is a consequence of exceedingly low solubility of oxygen in plasma. Since the transfer of oxygen from plasma into tissue is a diffusive process, dependent only upon the O₂ concentration gradient of oxygen between plasma and tissue (not red cells and tissue) we became intrigued with the idea that hemoglobin-induced vasoactivity was basically a consequence of oxygen oversupply to arterial regulatory vessels.

In order to better understand the difference between O₂ delivery by red blood cells and cell-free hemoglobin, it is helpful to briefly review the normal mechanisms in place to assure adequate tissue oxygenation. Mitochondria can function aerobically to a pO₂ in the range of a few mm Hg. In the body, mechanisms must increase or decrease blood flow to tissues, depending on the O₂ need; however, a metabolic reaction with a product, like H⁺or lactate, when the pO₂ drops below 2–3 mm Hg, would not be a good control because at this pO₂ it is already too late, and tissue death will occur very quickly.

Therefore, a mechanism is needed that detects impending O₂ lack with a cushion of O₂ supply. Such a mechanism can be engaged at the arteriolar level, where pO₂ is normally in the range of 30–40 mm Hg [[7]]. Terminal arterioles are highly innervated, and are good candidates for the O₂ sensors that regulate blood flow to capillaries [[5], [6]]. Coincidentally, arteriolar pO₂ is on the steep portion of the O₂ equilibrium curve, such that small changes in pO₂ result in large changes in O₂ release. Thus, a regulatory sensor at this point in the circulation would be sensitive and fulfill the need to operate well above critical tissue pO₂ levels.

Figure 3 superimposes the human OEC and data from Lindbom and coworkers [[8]] in the rabbit tenuismus muscle, showing that functional capillary density (the number of capillaries per unit tissue mass that permit blood flow) is regulated by O₂ supply. Furthermore, the midpoint in the FCD curve falls approximately at the mid portion of the OEC, strengthening our postulate that the key regulatory point for control for capillary blood flow lies at the terminal arterioles.

Figure 3 Coincidence of the blood oxygen saturation and functional capillary density as a function of pO₂. Data from [[8], [18]].[11]].

A MOLECULE BASED ON THE NEW PARADIGM: MALPEG-HEMOGLOBIN

Our new criteria for the design of a cell-free O₂ carrier, based on prevention of vasoconstriction, involve optimization of facilitated diffusion by control of molecular size, viscosity and oxygen affinity [[9], [12]]. We then needed to implement a modification strategy that was simple, inexpensive, non-toxic and with high chemical yield. Such chemistry was available through the work of Acharya, Manjula and Smith (1), who described the very elegant coupling of PEG-5000 to surface sites on hemoglobin (Figure 4). The modification is carried out as a 2-step procedure, first modifying surface Lysine groups with 2-Iminothiolane to add new SH groups, then coupling maleimide-activated PEG-5000 to these new sites. The reaction is highly specific and reproducible, and there are no toxic side products. Size-exclusion chromatograms, also shown in Figure 4, demonstrate the striking homogeneity of the MalPEG-Hb, and that it is free of any detectable low molecular weight hemoglobin. The process is very economical, since no chromatographic purification is necessary.

Figure 4 Schematic drawing of MalPEG-Hemoglobin. MP4 is MalPEG-Hemoglobin formulated for use as a blood substitute (4.2 g/dL in Lactated Ringer's solution). Size exclusion chromatograms comparing elution behavior of MP4 and stroma-free hemoglobin are shown at right. Modified from [[14]].

MalPEG-Hb, as formulated in lactated Ringer's solution at 4.2 g/dL, is MP4, the product that is now in clinical trials [[14]]. The properties of MP4 (Table 1) include a viscosity of 2.2 centipoise that is somewhat higher than conventional oxygen carriers but slightly less than the viscosity of blood. The oncotic pressure is approximately 49 mm Hg, higher than the oncotic pressure of blood, but in the range of the commonly used plasma expander Pentastarch^®. The p50 of MP4 is 5–6 mm Hg, and it is essentially devoid of cooperativity. The molecular radius is increased by factor of four to approximate 10 nm as a consequence of the surface modification and the layer of water that surrounds the molecule.

Table 1 Properties of MP4 compared with blood and hemoglobin [[14]]

	Blood	Hemoglobin	MP4
Viscosity (cPs)	4.0	1.0	2.2
COP (mm Hg)	27	16	49
Hb (g/dL)	14.0	4.2	4.2
p50 (mm Hg)	28.0	15.0	5.4
n (Hill's parameter)	2.6	2.9	1.2
MW (kDa)	–	65	90
Radius (nm)	–	2.7	10

MP4 IN THE MICROCIRCULATION

The most striking departure from dogma for MP4 is its low p50. In order to test the ability of this molecule to oxygenate tissue we turned to the microcirculation laboratory of Drs. Tsai and Itaglietta at UCSD [[13]]. In their model, blood is hemodiluted, stepwise with dextran-70, then finally with test article. In our experiments, we compared MP4 with the commercially available polymerized bovine hemoglobin (PolyBvHb, Oxyglobin^®). The OECs of the 2 oxygen carriers were measured under physiological conditions [[15]] (Figure 5) and compared with hamster blood.

Figure 5 Oxygen equilibrium curves for MP4, hamster blood and PolyBvHb. From [[13]].

Although the hematocrit values in the 3 groups were the same (Table 2), the total and plasma hemoglobin differ because concentration of the 2 oxygen carriers are different. Thus, the animals that received PolyBvHb had the highest total and plasma hemoglobin, the MP4 animals less, and the dextran animals had the lowest values.

Table 2 Hematologic measurements after hemodilution in the hamster [[13]]

	n	Total Hb (g/dL)	Plasma Hb (g/dL)	Hct (%)
MP4	6	4.8 ± 0.1	1.1 ± 0.0	11.2 ± 0.2
PolyBvHb	5	6.7 ± 0.2	3.7 ± 0.3	11.2 ± 0.4
Dextran	7	3.4 ± 0.1	0.0 ± 0.0	11.2 ± 0.5

The parameters of O₂ transport that were measured in the experiments (Figure 6) included systemic blood gases and acid-base status, as well as pO₂ measured by the phosphoresecence decay method [[3]]. Using these data, and the O₂ saturation curves (Figure 5), it is possible to calculate the O₂ content of arteriolar and venular blood, in all three compartments (plasma dissolved, plasma hemoglobin-bound and red cell hemoglobin-bound). The differences between arteriolar and venular O₂ content are the amounts of O₂ released across capillary beds. Similarly, the differences between O₂ content in systemic blood and arteriolar blood are the amounts of O₂ released in precapillary vessels.

Figure 6 Microcirculation measurements in the hamster skinfold model. Hemodynamic measurements include flow velocity, vessel diameter and functional capillary density. Oxygen measurements include systemic (arterial) blood gases, and pO₂ measured by the phosphorescence decay technique in arterioles and venules. Modified from [[13]].

The results of the experiments (Figure 7) demonstrate a clear difference between the O₂ carriers, and show the effectiveness of the low p50 formulation. Approximately equal amounts of oxygen are released in the precapillary and capillary vessels in animals hemodiluted with dextran-70. In the animals exchange transfused with PolyBvHb, more O₂ is delivered in precapillary than in capillary vessels. This applies not only to the plasma hemoglobin, but also to red cell hemoglobin. In contrast, the animals that received MP4 showed more O₂ release in capillaries, both from red cell and plasma hemoglobin, compared with precapillary vessels, suggesting that O₂ delivery is more efficient in this case. It appears that the extreme right shift of the oxygen equilibrium curve with PolyBvHb results in premature release of O₂ prior to arriving in capillary beds.

Figure 7 Oxygen release in precapillary compared to capillary vessels in the hemodiluted (hematocrit 11%) hamster using the skinfold observation technique. Hemodilution with dextran-70 results in about equal amounts of O₂ released at both levels of the circulation. In PolyBvHb animals O₂ release in precapillary vessels predominates, while the reverse is true in MP4 animals. From [[13]].

The question of the relevance of these observations can be addressed by examining the base excess as a reflection of global oxygen transport in these animals (Figure 8). Total arterial O₂ content in the three groups of animals does not correlate with base excess, but capillary O₂ release is highly correlated. Thus we conclude that delivery of O₂ into capillary beds is critical to the success of a cell-free oxygen carrier. Furthermore, this effectiveness at low concentration will result in lower doses being necessary, reducing both the risk of toxicity and cost.

Figure 8 Consequences of the location of O₂ release. There is no significant correlation between arterial O₂ content and base excess, while the correlation between capillary release and base excess is highly significant. From [[13]].

SUMMARY

We have proposed, on theoretical grounds, that in order to minimize vasoconstriction and to assure oxygenation of at-risk tissue (low pO₂), a cell-free O₂ carrier should have a low p50, it should be larger than native hemoglobin, and it should have elevated viscosity. These properties serve to control facilitated diffusion such that the autoregulatory mechanism will be defeated and O₂ delivery will be highly efficient. A new molecule, MalPEG-Hb (MP4, as formulated at 4.2 g/dL in lactated Ringer's solution) has been produced according to this new paradigm and is now being tested in clinical trials.

Dr. Winslow is President and CEO of Sangart, Inc., a company involved in the research and development of O₂ carriers. This work was supported, in part, by grants RO1 HL64579, R43 HL64996, R44 HL62818, R24 64395, R01 40696, R01 62354, R01 62318 from the NHLBI, NIH, to Sangart, Inc.