A recombinant human hemoglobin with anti-sickling properties greater than fetal hemoglobin.

A new recombinant, human anti-sickling beta-globin polypeptide designated beta(AS3) (betaGly(16) --> Asp/betaGlu(22) --> Ala/betaThr(87) --> Gln) was designed to increase affinity for alpha-globin. The amino acid substitutions at beta22 and beta87 are located at axial and lateral contacts of the sickle hemoglobin (HbS) polymers and strongly inhibit deoxy-HbS polymerization. The beta16 substitution confers the recombinant beta-globin subunit (beta(AS3)) with a competitive advantage over beta(S) for interaction with the alpha-globin polypeptide. Transgenic mouse lines that synthesize high levels of HbAS3 (alpha(2)beta(AS3)(2)) were established, and recombinant HbAS3 was purified from hemolysates and then characterized. HbAS3 binds oxygen cooperatively and has an oxygen affinity that is comparable with fetal hemoglobin. Delay time experiments demonstrate that HbAS3 is a potent inhibitor of HbS polymerization. Subunit competition studies confirm that beta(AS3) has a distinct advantage over beta(S) for dimerization with alpha-globin. When equal amounts of beta(S)- and beta(AS3)-globin monomers compete for limiting alpha-globin chains up to 82% of the tetramers formed is HbAS3. Knock-out transgenic mice that express exclusively human HbAS3 were produced. When these mice were bred with knock-out transgenic sickle mice the beta(AS3) polypeptides corrected all hematological parameters and organ pathology associated with the disease. Expression of beta(AS3)-globin should effectively lower the concentration of HbS in erythrocytes of patients with sickle cell disease, especially in the 30% percent of these individuals who coinherit alpha-thalassemia. Therefore, constructs expressing the beta(AS3)-globin gene may be suitable for future clinical trials for sickle cell disease.

Sickle cell disease (SCD) 1 results from an A to T transversion at the sixth codon of the human ␤-globin gene on chromosome 11 (1,2). The mutation of a single DNA base leads to the substitution of a valine for a glutamic acid in the ␤-globin polypeptide of sickle hemoglobin (HbS). The positioning of a hydrophobic residue at this position permits an interaction with a hydrophobic pocket on another hemoglobin tetramer (see Fig. 1). This interaction allows deoxy-HbS to polymerize in an entropy-driven process (3)(4)(5). The polymerization of deoxy-HbS leads to erythrocyte deformation from a biconcave morphology into the sickle shapes for which SCD is named.
Polymerization of deoxy-HbS is effectively inhibited by fetal hemoglobin (HbF), and individuals who are homozygous for the sickle mutation but also express high levels of HbF are typically asymptomatic (6 -8). The level of HbF necessary to significantly reduce the symptoms of SCD ranges from about 20 to 25%; however, data showing an enhanced red cell survival with as little as 9% HbF have been reported (9). The efficacy of HbF in inhibiting HbS polymerization suggests that transduction of fetal globin genes into hematopoietic stem cells might be an effective strategy for sickle cell disease gene therapy. However, high levels of ␥-globin gene expression are difficult to achieve in adult erythroid cells even in the absence of competition with the ␤-globin gene for locus control region (LCR) interactions (10,11). Low level expression of the ␥-globin gene apparently results from the absence of fetal-specific positive regulatory factors in adult cells. Our approach to overcome this deficiency has been to utilize the ␤-globin gene as a backbone and to introduce ␥-globin amino acid substitutions into this construct. We demonstrated previously that a ␤-globin gene containing alanine at position 22 and glutamine at position 87 (␤ AS2 ) significantly inhibited HbS polymerization (12). We have now introduced an additional modification into ␤ AS2 to form ␤ AS3 ; this modification increases the affinity of ␤-globin subunits for ␣-globin polypeptides. In this paper we demonstrate that ␤ AS3 has a competitive advantage over ␤ S polypeptides for interaction with limiting ␣-globin subunits and that HbAS3 dramatically inhibits deoxy-HbS polymerization.
Fragment preparation and microinjection were as described previously (16). Transgenic animals expressing high levels of human hemoglobin were identified by isoelectric focusing of hemolysates. Isoelectric focusing was performed using the Isothermal Controlled Electrophore-sis system (Fisher) with precast agarose isoelectric focusing gels (Isolab, Akron, OH).
Functional Analysis of Recombinant Human Hemoglobins-Oxygen equilibrium curves were measured with a Hemox Analyzer (TCS Scientific, New Hope, PA) as described (21). The oxygen equilibrium curves were determined in 0.1 M potassium phosphate buffer (pH 7.0) at 20°C with a hemoglobin concentration of 25 M. Polymerization kinetics were determined in 1.8 M potassium phosphate buffer as described (22). Polymerization was initiated using the temperature jump method in which the temperature of deoxygenated hemoglobin solutions is rapidly changed from 0 to 30°C, and aggregation is monitored turbidimetrically at 700 nm (22,23).
Analysis of Hemoglobin Subunit Recombination-Monomeric ␣and ␤-globin subunits were prepared as described (24). The hemolysates were treated with carbon monoxide prior to separation to reduce formation of methemoglobin. Isolated ␣and ␤-globin monomers were allowed to combine at 0°C for 1 h. The amounts of HbS and HbAS3 formed were determined by HPLC using a PolyCAT A cation exchange column (PolyLC, Columbia, MD). Hemoglobins were eluted from the column with a linear gradient of Buffer A (35 mM BisTris, 1.5 mM KCN, 3 mM ammonium acetate, (pH 6.47)) and Buffer B (35 mM BisTris, 1.5 mM KCN, 16.85 mM ammonium acetate, 150 mM sodium acetate, (pH 7.0)) (25). The flow rate was 1 ml/min with detection at 415 nm. The relative amounts of each hemoglobin were calculated by integration of the area under each peak.
Analysis of Heterotetramer Formation-Equimolar amounts of purified oxygenated HbS and HbAS3 or HbA were mixed and allowed to equilibrate overnight at 0°C as described previously (17). The mixtures were analyzed by cation exchange chromatography as described above with the addition of 3 mM sodium dithionite to the elution buffers. Sodium dithionite was added to allow separation of hemoglobins under anaerobic conditions, which was necessary for detection of heterotetramers.
Production of HbAS3, Knock-out Transgenic Sickle Mice-Mice that express HbAS3 exclusively were produced by breeding the HbAS3 mice with mouse ␣and ␤-globin gene knock-out animals (26,27). These animals were subsequently bred with knock-out transgenic sickle mice (28) to obtain mice that were homozygous for mouse ␣and ␤-globin gene knock-outs and contained the human LCR ␣, LCR ␥-␤ S , and LCR ␤ AS3 transgenes. These animals expressed approximately equal amounts of HbAS3 and HbS.
Hematological Indices and Histopathology-Blood was collected from anesthetized animals into Microtainer EDTA collection tubes. The red blood cell count was measured on a HemaVet 1500 hematology analyzer. Hemoglobin concentration was determined spectrophotometrically after conversion to cyanmethemoglobin with Drabkin's reagent. Before determining the hemoglobin concentration, red cell membranes were formed into pellets at 14,000 rpm for 5 min in an Eppendorf centrifuge. Removal of the membranes inhibits artifactually high values caused by membrane-bound, denatured hemoglobin. Hematocrit was measured with a JorVet J503 microhematocrit centrifuge. Reticulocyte count was determined by flow cytometry after staining with thiazole orange. Urine osmolality was measured after food and water were withheld from the mice for 4 h. Tissues were fixed in 70% alcoholic formalin, embedded in paraffin, sectioned, and stained with hematoxylin-eosin or Gomori's iron stain by standard methods. Fig. 1 illustrates the major contacts present in the HbS polymer. We reported previously amino acid substitutions (␤Gly 22 3 Asp and ␤Thr 87 3 Gln) that inhibit HbS polymerization (12). To provide the modified ␤-globin polypeptide with a competitive advantage for interacting with ␣-globin chains, we made an additional amino acid substitution (␤Gly 16 3 Asp). Codon changes were introduced into the human ␤-globin gene by site-directed mutagenesis, and the modified sequences were inserted downstream of a 22-kb DNA fragment containing the DNase hypersensitive sites 1-5 (5Ј-HS 1-5) of the ␤-globin LCR (16). These constructs were injected into fertilized mouse eggs, and transgenic lines were established. Hemolysates obtained from several animals were analyzed by HPLC to quantitate the amounts of human, mouse, and hybrid hemoglobins (17). The purity of the human hemoglobins was assessed by denaturing reverse-phase HPLC, which separated the ␣and ␤-globin subunits (19,20). Two types of contacts occur between deoxy-HbS tetramers in the doublestranded fibers. Contacts along the long axis of the fiber are termed axial contacts, whereas contacts along the sides of tetramers are termed lateral contacts. The ␤6 valine plays a crucial role in the lateral contact by interacting with the hydrophobic ␤85 phenylalanine and ␤88 leucine on a neighboring tetramer. An important axial contact is the interaction of the ␤22 glutamic acid with an ␣20 histidine on an adjacent tetramer. The recombinant hemoglobin HbAS3 contains amino acid changes at positions 16,22, and 87 of the ␤-globin chain. The modification at ␤16, glycine to aspartic acid, is known as HbJ-Baltimore and provides HbAS3 with an enhanced ability to interact with the ␣-globin subunit to form ␣␤ AS3 dimers. The amino acid modification at ␤22, glutamic acid to alanine, disrupts the axial contact interaction with the ␣20 histidine, and the mutation at ␤87, threonine to glutamine, disrupts the lateral contact of the ␤6 with the hydrophobic pocket.

Functional Analysis of Recombinant Human Hemoglobins-
The oxygen equilibrium curve for purified HbAS3 is sigmoidal (data not shown), demonstrating that oxygen binding is cooperative. The P 50 value (partial pressure of oxygen at which hemoglobin is half-saturated) was determined for HbAS3 and compared with HbA and HbF ( Table I). The P 50 for stripped HbAS3 is 7.2 mm Hg and is increased to 10.3 mm Hg in the presence of 2 mM bisphosphoglycerate. Although stripped HbAS3 has a higher oxygen affinity than HbA or HbF, its affinity in the presence of bisphosphoglycerate is similar to that of HbF. In a physiologic environment, HbAS3 should be indistinguishable from HbF, which is a protein that is known to function in vivo to provide protection against sickling crises.

Inhibition of HbS Polymerization by Recombinant Human
Hemoglobins-The ability of HbAS3 to inhibit deoxy-HbS polymerization was analyzed by delay time measurements (22). Briefly, HbS (100%) or mixtures of HbS (75%) and HbF or HbAS3 (25%) were deoxygenated, and polymerization as a function of time was measured spectrophotometrically (22,23). Fig. 2A demonstrates that HbS polymerizes rapidly but that HbF and HbAS3 markedly delay HbS polymerization. HbAS3 inhibits HbS polymerization at approximately the same level as HbF, which is known to inhibit sickling in vivo at a 3:1 ratio (6,7,29). This finding strongly suggests that HbAS3 would inhibit HbS polymerization in vivo if HbAS3 levels constituted 25% of total hemoglobin.
The delay times determined in Fig. 2A were all measured at a concentration of 60 mg/dl. Fig. 2B illustrates the results of similar experiments performed using different concentrations of total hemoglobin. The ratio of HbS to HbF or HbAS3 in these experiments was 3:1. In this figure the log of the reciprocal of the delay time and the log of hemoglobin concentration are plotted. As reported by Hofrichter et al. (30), an empirical relationship between delay time and hemoglobin concentration is described by the equation 1/t d ϭ ␥S n , where S ϭ [Hb] total / [Hb] soluble and ␥ is an experimental constant. The n value is related to the size of nuclei formed during polymerization. The n values of the data shown in Fig. 2B are between 2 and 3, which agree well with those shown previously in high phosphate buffer (22). The delay time plots for mixtures of HbS with HbF and HbS with HbAS3 overlap, indicating that HbAS3 has an ability to inhibit HbS polymerization that equals that of HbF.
Competition for Dimerization with the ␣-Globin Subunit-Previous work demonstrated that two mutations were sufficient to provide the ␤-globin subunit with a significant ability to inhibit the polymerization of sickle hemoglobin (12). The anti-sickling HbAS2 described in that work was the starting point for the improved anti-sickling hemoglobin, which we have termed AS3. HbAS3 combines the mutations at position 22 and 87 of AS2 with a third mutation at position 16, which is known as HbJ-Baltimore (31). This mutation increases the ability of a ␤-subunit to compete for dimerization with an ␣-subunit (32).
The parameters that determine how well a ␤-subunit will interact with an ␣-subunit have been defined using naturally occurring mutant hemoglobins (33). The dimerization relation-ship is straightforward and is based primarily upon the charge of the ␣and ␤-chains (34). Under physiologic conditions, the ␣-subunit is positively charged (pI 8.4), and the ␤-subunit is negatively charged (pI 6.7) (35). Mutations that increase the net negative charge on the ␤-subunit increase its ability to dimerize with the ␣-subunit. Similarly, mutations that decrease the negative charge on the ␤-subunit decrease its ability to dimerize with the ␣-subunit. The ␤ S -subunit is particularly vulnerable to competition with a more negatively charged ␤-subunit because the sickle mutation, glutamic acid to valine, reduces the negative charge on the ␤ S -chain. For this reason, erythrocytes in individuals with sickle trait do not contain 50% HbS and 50% HbA but contain ϳ42-45% HbS (36). Competition for the ␣-subunit, therefore, appears to be a potential method to increase the amount of anti-sickling hemoglobin in an erythrocyte at the expense of sickle hemoglobin.
The addition of the HbJ-Baltimore mutation causes the net charge on ␤ AS3 to equal that of ␤ A ; the addition of an aspartic acid at ␤16 compensates for the loss of a glutamic acid at ␤22. Based strictly upon charge, ␤ AS3 should have a competitive advantage over ␤ S for dimerization with the ␣-subunit. When ␣-, ␤ S -, and ␤ AS3 -globin subunits are combined at an ␣-globin: ␤-globin ratio of 1:1, HbAS3 comprises 59.5% of total hemoglobin. As the ␣-globin subunit becomes limiting, the proportion of HbAS3 increases dramatically, reaching a level of 75.1% of total hemoglobin at an ␣-globin:␤-globin ratio of 0.5:1 and 82% of total hemoglobin at a ratio of 0.1:1. These results demonstrate that ␤-subunits with the HbJ-Baltimore mutation have a significant competitive advantage over ␤ S -polypeptides for interaction with ␣-subunits. The ␤ AS3 -subunit interacts with the ␣-subunit even more efficiently than would be expected on the basis of charge alone. Apparently, the specific combination of negative residues in ␤ AS3 -subunits improves the ability of this polypeptide to interact with ␣-globin chains.
Additionally, a recombinant therapeutic hemoglobin must be able to form heterotetramers with deoxy-HbS to achieve effective inhibition of polymerization. Earlier studies have shown that the inhibitory effect of HbF on the polymerization of deoxy-HbS is dependent on the formation of heterotetramers (␣ 2 ␤ S ␥) (37). A similar mechanism was proposed to explain the polymerization inhibitory effects of the naturally occurring minor hemoglobin, HbA 2 (␣ 2 ␦ 2 ) (38). Unlike the ␣ 2 ␤ A ␤ S heterotetramer, the ␣ 2 ␤ S ␥ and ␣␤ S ␦ heterotetramers are excluded from the sickle polymer, which accounts for the increased inhibitory effects of HbF and HbA 2 relative to HbA. To determine whether recombinant HbAS3 could form heterotetramers (a 2 ␤ S ␤ AS3 ), we mixed oxygenated HbAS3 with oxygenated HbS. At equilibrium, a binomial (1:2:1) distribution of parent hemoglobins and heterotetramers is observed (Fig. 3). Similar results were obtained for oxygenated mixtures of HbA and HbS.

Correction of Abnormal Red Blood Cell Morphology and Hematological Parameters in Knock-out Transgenic Sickle/AS3
Mice-To determine whether HbAS3 would inhibit HbS polymerization in vivo, we bred mice that express HbAS3 with our knock-out transgenic sickle mice and obtained animals that were homozygous for mouse ␣and ␤-globin gene knock-outs and contained the human LCR ␣, LCR ␥-␤ S , and LCR ␤ AS3 transgenes. Blood smears from sickle control and sickle/AS3 animals are illustrated in Fig. 4. Many sickled cells are observed in the sickle control; however, no sickled cells are observed in sickle/AS3 mice (two representative animals). The upper left panel is a wild-type control. The blood smears of sickle/AS3 animals also lack the anisocytosis, poikilocytosis, and polychromasia that are characteristic of erythrocyte morphology observed in SCD mice and HbS patients (28,36).  AS3 mice. In sickle/AS3 animals, red blood cell counts are nearly doubled, hemoglobin levels are increased by 6 g/dl, hematocrit is normalized, and reticulocytosis is reduced to levels approaching those of wild-type mice. Sickled erythrocytes are not observed in blood smears or in tissues of sickle/AS3 mice. These data demonstrate that HbAS3 is a potent anti-sickling hemoglobin in vivo. Amelioration of Spleen, Liver, and Kidney Pathology and Restoration of Kidney Function in Sickle/AS3 Mice-Histologi-cal sections of wild-type, sickle, and sickle/AS3 animals are shown in Fig. 6A. The spleens of sickle mice are characterized by a massive expansion of red pulp, dramatic pooling of sinusoidal erythrocytes, vaso-occlusion, and a complete loss of lymphoid follicular structure. In sickle/AS3 mice, normal splenic red and white pulp is observed, and virtually no pools of sickle erythrocytes or infarcts are evident. In addition, splenomegaly is substantially diminished in sickle/AS3 mice (Fig.  6C); sickle/AS3 spleens are ϳ0.6% of total body weight com- , sickle (n ϭ 5), and sickle/AS3 (n ϭ 5) mice. Red blood cell (RBC) counts of sickle/AS3 mice are nearly doubled, hemoglobin levels are increased by 6 g/dl, hematocrit is normalized, and reticulocytosis is reduced to levels approaching those of wild-type mice. pared with sickle spleens that are almost 4% of body weight.
Livers of sickle animals are characterized by focal areas of necrosis and pronounced congestion of the intrahepatic vascu-lature with pooling of sickled red blood cells. Large aggregates of erythroid progenitors are evident in the sinusoids, and this extramedullary hematopoiesis is indicative of severe anemia.  6. Normalization of spleen, liver, and kidney pathology in sickle/AS3 mice. A, spleen, liver, and kidney sections were analyzed at high magnification (ϫ100), and kidney was examined at low (ϫ10, bottom three panels) magnification. In sickle/AS3 mice, normal splenic red and white pulp is observed, and no pools of sickle erythrocytes or infarcts are evident. In livers of sickle/AS3 animals, focal areas of necrosis and aggregation of sickled erythrocytes are not observed; also, extramedullary hematopoiesis and hemosiderin deposition are absent. Kidneys of sickle/AS3 mice appear normal and free of the disruptive vascular red blood cell pooling and hemosiderin deposits observed in mock treated animals. B, urine concentrating ability is restored to wild-type levels in sickle/AS3 animals. n ϭ 5 for wild-type, sickle, and sickle/AS3, respectively. C, correction of splenomegaly in sickle/AS3 mice. n ϭ 5 for wild-type, sickle, and sickle/AS3. There is also abundant hemosiderin deposition subsequent to Kupffer cell erythrophagocytosis. In sickle/AS3 animals, focal areas of necrosis and aggregation of sickled erythrocytes are not observed; also, extramedullary hematopoiesis and hemosiderin deposition are absent.
In the kidneys of mock transduced mice, engorgement and occlusion of blood vessels results in vascular, tubular, and glomerular changes. Sequestration and occlusion are most obvious at the corticomedullary junction where dilated capillaries are easily observed in this region of reduced oxygen tension. Reduced medullary blood flow in HbS patients causes extensive tubular damage that results in hyposthenuria, and this same loss of urine-concentrating ability is observed in the sickle mice. Kidneys of sickle animals also accumulate abundant hemosiderin in the cortical region, and these aggregates are easily visualized with Gomori's iron staining (data not shown). In contrast, kidneys of sickle/AS3 mice appear normal and free of the disruptive vascular red blood cell pooling and hemosiderin deposits observed in mock treated animals. Most importantly, urine-concentrating ability is completely restored in sickle/AS3 mice (Fig. 6B).
Expression of ␤ AS3 -globin should effectively lower the concentration of HbS in erythrocytes of patients with SCD, especially in the 30% of these individuals who coinherit ␣-thalassemia (39). We recently corrected our knock-out transgenic mouse model of SCD using lentiviral transduction of the ␤ AS3 anti-sickling gene into purified hematopoietic stem cells (40). These results suggest that stem cell-and genetics-based therapies using recombinant ␤ AS3 -globin may be able to be translated to human sickle patients.