Autoxidation and Oxygen Binding Properties of Recombinant Hemoglobins with Substitutions at the αVal-62 or βVal-67 Position of the Distal Heme Pocket*

Background: Tertiary structure of the ligand binding pocket influences oxygen binding and autoxidation of hemoglobin. Results: E11 mutants have increased autoxidation rate. βE11Phe increases, whereas βE11Ile decreases the oxygen binding affinity of hemoglobin. Conclusion: Bulky residues at βE11 affect ligand binding and cause noticeable tertiary structural changes at the heme pockets. Significance: Hemoglobin distal heme pocket mutations alter oxygen binding properties without changing the quaternary structure. The E11 valine in the distal heme pocket of either the α- or β-subunit of human adult hemoglobin (Hb A) was replaced by leucine, isoleucine, or phenylalanine. Recombinant proteins were expressed in Escherichia coli and purified for structural and functional studies. 1H NMR spectra were obtained for the CO and deoxy forms of Hb A and the mutants. The mutations did not disturb the α1β2 interface in either form, whereas the H-bond between αHis-103 and βGln-131 in the α1β1 interfaces of the deoxy α-subunit mutants was weakened. Localized structural changes in the mutated heme pocket were detected for the CO form of recombinant Hb (rHb) (αV62F), rHb (βV67I), and rHb (βV67F) compared with Hb A. In the deoxy form the proximal histidyl residue in the β-subunit of rHb (βV67F) has been altered. Furthermore, the interactions between the porphyrin ring and heme pocket residues have been perturbed in rHb (αV62I), rHb (αV62F), and rHb (βV67F). Functionally, the oxygen binding affinity (P50), cooperativity (n50), and the alkaline Bohr Effect of the three α-subunit mutants and rHb (βV67L) are similar to those of Hb A. rHb (βV67I) and rHb (βV67F) exhibit low and high oxygen affinity, respectively. rHb (βV67F) has P50 values lower that those reported for rHb (αL29F), a B10 mutant studied previously in our laboratory (Wiltrout, M. E., Giovannelli, J. L., Simplaceanu, V., Lukin, J. A., Ho, N. T., and Ho, C. (2005) Biochemistry 44, 7207–7217). These E11 mutations do not slow down the autoxidation and azide-induced oxidation rates of the recombinant proteins. Results from this study provide new insights into the roles of E11 mutants in the structure-function relationship in hemoglobin.

Autoxidation occurs when the heme-iron atom is oxidized from the ferrous (Fe ϩ2 ) to the ferric (Fe ϩ3 ) state with release of superoxide (O 2 . ) or perhydroxy (HO 2 . ) radical (16). The reaction proceeds with a combination of two mechanisms depending on the concentration of oxygen (O 2 ) (4). The biological importance of autoxidation is that H 2 O 2 can be produced by dismutation of superoxide and leads to both oxidative stress and reactive oxygen species generation. In addition, the resulting Fe ϩ3 on hemoglobin cannot bind O 2 . Therefore, Hb A is physiologically inactivated when all four iron atoms of the tetramer are oxidized to form methemoglobin (met-Hb). This reaction carries on without catalysis and occurs naturally in normal red blood cells. This process can be promoted by anions (17) and nitric oxide (NO) (6). In erythrocytes, an enzymatic reducing system reverses oxidation of human hemoglobin (Hb A) to met-Hb (18). However, in cell-free solutions the autoxidation reaction leads to more rapid rates of hemin loss resulting in denaturation of the apoglobin (19). It has long been recognized that Hb can potentially formulate into a substitute for red blood cell transfusion termed hemoglobin-based oxygen carriers (20). However, acellular hemoglobin * This work was supported, in whole or in part, by National Institutes of Health Grants HL-024525 and GM-0846140. This work was also supported by The Arnold and Mabel Beckman Foundation and the Howard Hughes Medical Institute. 1  diluted into blood dissociates into dimers with subsequent loss in the cooperative oxygenation properties (21) and reacts with endothelium-derived NO with consequent smooth muscle dystonia and clotting disorders (22). Hemoglobins have been cross-linked (23) and polymerized (24) into larger molecules to counteract these effects and achieved mixed results in clinical trials (25,26). Further improvements can potentially be made by producing "designer" hemoglobin with unique oxygen binding and autoxidation properties.
With the advent of protein engineering, recombinant hemoglobin (rHb) (27)(28)(29)(30) with unique properties can be prepared in sufficient quantity for hemoglobin-based oxygen carrier studies. We and others have previously generated B10Leu mutants and shown that replacement of the ␣B10Leu residue (␣Leu-29) with phenylalanine increases the oxygen affinity and decreases the autoxidation rate of the macromolecule (6,13,15,31,32). We have also produced an octameric hemoglobin with wild type biological activities by substituting the ␣-subunit Asn-78 residue with cysteine (33). Combination of these mutations produced octameric macromolecules with unique properties that were utilized as resuscitation solutions in mice suffering traumatic brain injury combined with hemorrhagic shock. We observed that the high oxygen affinity rHb (␣N78C/L29F) conferred a neuroprotective effect in the selectively vulnerable region of the hippocampus (34). In light of our success in utilizing rHb as a resuscitation fluid, we have renewed our effort in identifying rHb mutants possessing unique properties for biomedical applications. We report here the characterization of six E11Val mutants. By replacing the ␤Val-67 with phenylalanine, we have generated a mutant with oxygen affinity even higher than that of rHb (␣L29F). Our results affirm the notion that heme pocket mutations can produce macromolecules with unique oxygen binding properties without changing the quaternary structure of hemoglobin.

EXPERIMENTAL PROCEDURES
Materials-Human normal blood samples were obtained from the local blood bank, and Hb A was isolated according to Russu et al. (35). Restriction enzymes and related enzymes needed for DNA work were products of New England Biolabs.
The QuikChange site-directed mutagenesis kit was purchased from Stratagene. Reagent grade chemicals were obtained from Sigma.
E. coli JM109 was used as the host for protein expression. The cell culturing and induction conditions and the purification procedure under CO atmosphere have been described by Shen et al. (29,30). The purified proteins were subjected to Edman degradation to estimate the amount of N-terminal methionine cleavage, and the molecular weights of the rHb subunits were confirmed by mass spectrometry in an ion trap instrument equipped with an electrospray ionization source (29). All rHb samples in this study had the correct molecular weights and Ͻ5% of N-terminal methionine.
Oxygen Binding Properties-Oxygen dissociation curves were measured using a Hemox Analyzer (TCS Medical Products) as previously described (29). Samples were prepared with 0.1 mM Hb (in terms of heme) in 0.1 M sodium phosphate buffer. Catalase and superoxide dismutase were added to the sample to prevent and/or slow down the formation of met-Hb (36). Experiments were conducted at 29°C as a function of pH. Oxygen affinity and cooperativity were determined for each sample from the resulting oxygen equilibrium curves. To analyze the results, the equilibrium binding curves were fit to the Adair equation using a nonlinear least-squares procedure. The oxygen affinity of the sample was determined from the P 50 value (in millimeters of Hg). To measure cooperativity, values for n 50 (the Hill coefficient) were calculated from the slope of the Hill plot at 50% saturation. The values had an accuracy of Ϯ10%. Experimental results have S.D. Ϯ 4% between runs.
Structural Study with 1 H NMR Spectroscopy-Proton spectra of Hb A and rHbs in CO or the deoxy form were collected on Bruker Avance DRX-300 or DRX-600 spectrometers. Hb samples were exchanged into 0.1 M sodium phosphate, pH 7.0, and concentrated to an ϳ5% (3.1 mM in terms of heme) solution. Deuterium oxide was then added to a final concentration of 5%, and measurements were made at 29°C. A jump-and-return pulse sequence was used to suppress the water signal (37). The methyl proton resonance (4.76 ppm upfield of the water signal at 29°C) of 2,2-dimethyl-2-silapentene-5-sulfonate (DSS) was used as internal reference for the proton chemical shifts.
Autoxidation of Hb A and rHbs-Autoxidation of hemoglobin samples was measured according to Wiltrout et al. (15). Briefly, concentrated rHbs stored in CO form were oxygenated under a constant stream of 100% O 2 in a rotary evaporator for a 1 h minimum under lamp light and in an ice bath. The rHbO 2 was then diluted immediately to 60 M in terms of heme (A 577 ϳ 1.0) with 0.1 M sodium phosphate, pH 7.0, and 1 mM EDTA. The samples were then transferred without delay into capped cuvettes and a temperature-controlled holder at 25°C in a Cary 50 UV-visible spectrophotometer (Varian). Absorbances at 560, 577, 630, and 700 nm were recorded every 15 min for 7 h. The optical density readings at 700 nm were normally less than 0.02 at the end of data acquisition and used to assure the lack of sample precipitation during the time of data acquisition. They were treated as background readings and subtracted from the readings of the other three wavelengths. The major components in the reaction mixtures were oxy-Hb, met-Hb, and hemichrome, and their respective concentrations can be calculated from these absorbances and their extinction coefficients determined at the corresponding wavelengths (38). The millimolar extinction coefficients (mM Ϫ1 cm Ϫ1 ) for oxy-Hb and met-Hb at pH 7. Samples were prepared similarly for azide-induced oxidation experiments except that the reaction mixture had an additional 0.1 M sodium azide. Absorbances were measured at 577, 630, and 700 nm every 15 min for 7 h. Readings at 700 nm were used for base-line subtractions and indicators for protein precipitation. The millimolar extinctions coefficients (mM Ϫ1 cm Ϫ1 ) for azidomet were those reported by Jeong et al. (13), and they are ⑀ 577 ϭ 8.37 and ⑀ 630 ϭ 1.98 after converting into per heme base. Calculations were treated as above, and the final equation for determining the concentration of oxy-Hb has the form of 0.063A 577 Ϫ 0.2664A 630 . For some rHb mutants, precipitation occurred 5 h after adding azide. Therefore, only data collected for the first 4 h were used in calculations.
The Hill coefficients (n 50 ), an indicator of cooperativity in the oxygenation process of Hb, are plotted as a function of pH in Fig. 2B. All mutants investigated cooperatively bind oxygen. rHb (␣V62I) and rHb (␣V62L) exhibited a high level of cooperativity, comparable to Hb A, whereas other distal heme pocket mutants in this study exhibited noticeable decreases in the Hill coefficient. As summarized in Table 1, the Hill coefficients at pH 7.4 for Hb A, rHb (␣V62L), rHb (␣V62I), rHb (␣V62F), rHb (␤V67L), rHb (␤V67I), and rHb (␤V67F), are 2.8, 2.8, 3.0, 2.4, 2.7, 2.5 and 2.0. In particular, the rHb (␤V67F) mutant has Hill coefficients between 1.5 and 2.3 within the pH range of 5.7-8.2. These values are significantly lower than those of Hb A but indicate that the mutant still maintains a low level of cooperativity.
The affinity between oxygen and Hb decreases upon lowering the pH of the solution. This alkaline Bohr Effect can be estimated from the slope of the log P 50 versus pH plot, and the values for Hb A and the six E11 mutants are listed in Table 1. With the exception of rHb (␤V67F), all other proteins on the list have a calculated value of 0.45-0.48. The lower Bohr Effect (0.36) observed for rHb (␤V67F) implies that the mutant has a smaller change in oxygen affinity upon lowering pH. Consequently, less oxygen could be released in muscle capillaries with high levels of lactic acid.
Autoxidation and Azide-induced Oxidation-The autoxidation rates of Hb A and mutants were measured in 0.1 M sodium phosphate at pH 7.0 and 25°C. Data were acquired for 7 h and treated as a mono-exponential process according to Wiltrout et al. (15). The results represent the initial autoxidation rates, and the rate constants, k auto , as summarized in Table 2, are the average rate constants for the first 7 h of the reactions. Under the experimental conditions employed, HbO 2 A has an autoxidation rate constant of 7.8 ϫ 10 Ϫ4 h Ϫ1 . All E11 mutants in this study have higher k auto values than that of Hb A. The values obtained range from 3.2-fold higher for rHb (␣V62I) to 7-fold higher for rHb (␤V67F).
Structural Properties-1 H NMR has been used routinely to probe tertiary and quaternary structural changes in hemoglobin (41). The E11 mutants generated for this study in CO (Fig.  3) and deoxy (Fig. 4) forms were analyzed with this technique in 0.1 M sodium phosphate buffer at pH 7.0 and 29°C.
NMR Spectra of Hemoglobins in CO Form- Fig. 3 shows the 1 H NMR spectra for Hb A and rHbs in the CO form. Resonance signals between 9 and 14 ppm have been assigned to the exchangeable protons in the intersubunit interfaces. The resonances at 12.9 and 12.1 ppm have been assigned to the NH ⑀1 of the side chains of ␣His-122 and ␣His-103, respectively (42)(43)(44). These two histidines are located within the ␣ 1 ␤ 1 interface. The resonance at 10.6 ppm has been assigned to the NH ⑀1 of ␤Trp-37 (44,45), which is an ␣ 1 ␤ 2 interface residue. No change in the chemical shift for these resonances was observed for all six E11 rHbs in relation to Hb A (Fig. 3A), indicating no perturbation to their quaternary structure at the ␣ 1 ␤ 1 and ␣ 1 ␤ 2 interfaces.
We observed marked changes in the resonances in the region between 9.5 and 10.5 ppm. In particular, an additional peak at ϳ9.8 ppm can be observed for the 2 mutants with aromatic amino acid substitution. However, we do not have resonance assignments for these signals, and we cannot elaborate on the origins of these disturbances.
Resonances from 0 to Ϫ3 ppm belong to the non-exchangeable ring current-shifted protons and provide information about the tertiary structure of the heme pockets. The signals at Ϫ1.75 and Ϫ1.82 ppm relative to DSS have been assigned to the ␥ 2 -CH 3 group of ␣Val-62 and ␤Val-67, respectively (46,47). For the ␣-subunit E11 mutations, the resonance corresponding to ␣Val-62 is absent from the spectra. Likewise, the resonance for the ␤Val-67 is absent from the spectra for the ␤-subunit mutations. For the ␤V67L and ␤V67F mutants, the resonances between Ϫ0.5 to Ϫ1.1 ppm differ significantly from those of HbCO A, but we have not yet assigned the resonances for these signals.
Combining these observations, we conclude that the mutants in the liganded form maintain their quaternary structure. Localized structural changes in the mutated heme pocket have been detected for rHb (␣V62F), rHb (␤V67I), and rHb (␤V67F).
NMR Spectra of Hemoglobins in Deoxy Form- Fig. 4 shows the 1 H NMR spectra for Hb A and rHbs in the deoxy form. The hyperfine-shifted proton resonances of the N ␦ H exchangeable proton of proximal histidines are between 60 and 80 ppm from DSS. The resonances at 63 and 76 ppm have been assigned to ␣His-87 and ␤His-92 of the deoxy Hb A (48,49), respectively. For the three ␣-subunit E11 mutants, the resonance at 76 ppm remains unchanged. However, the signal at 63 ppm has shifted 1.5-2.5 ppm downfield (Fig. 4A). The results suggest that these ␣-subunit mutations slightly disturb the tertiary structure of the proximal heme pockets in the mutated subunits. Among the three ␤-subunit mutants, only rHb (␤V67F) displays a spectrum that differs from that of Hb A in this region. The resonance at 63 ppm remains unchanged, whereas the signal at 76 ppm broadened and shifted 3.5 ppm downfield (Fig.  4A). The observation is indicative of a structural change in the proximal heme pocket of the ␤-subunit of rHb (␤V67F), and the change is more severe than those observed for the ␣E11 mutants.
The spectral region between 16 -25 ppm downfield from DSS is presented in Fig. 4B. The resonances at 17.0 and 22.6 ppm represent signals generated by the porphyrin ring of the ␣and ␤-subunits, respectively (41). The rHb (␣V62L), rHb (␤V67L), and rHb (␤V67I) have a similar NMR pattern in this region as Hb A (Fig. 4B). For the rHb (␣V62I) and rHb (␣V62F) mutants, the signal at 17.0 ppm has shifted 1.5 ppm downfield and upfield, respectively, from its original position. A much larger perturbation in the spectrum has been observed for the rHb (␤V67F) mutant. The resonance at 22.6 ppm has shifted ϳ3 ppm upfield and broadened. All these results are indicative of the local structural disturbances in the heme pocket where the mutation has taken place.
Resonances at 14.1 and 11.2 ppm have been assigned to H-bonds between residues in the ␣ 1 ␤ 2 interface of deoxy Hb A. The resonance at 14.1 ppm represents the H-bond between ␣Tyr-42 and ␤Asp-99 (50). The resonance at 11.2 ppm denotes the exchangeable NH indole proton of ␤Trp-37 that H-bonds to ␣Asp-94. Both of these resonances are important T-state markers of the deoxy-Hb A (44,50,51). rHb (␣V62I) exhibits a slight chemical shift of 0.3 ppm upfield for the ␣ 1 ␤ 2 interface peaks. Overall, the E11 mutations exhibit little to no perturbation of the ␣ 1 ␤ 2 interface.
As in the CO form, ␣His-122 and ␣His-103 in the ␣ 1 ␤ 1 interface of deoxy-Hb A give resonance signals at 13.0 and 12.2 ppm from DSS, respectively. Mutants that have substitution at the ␤E11Val position do not affect the resonances at these positions. However, the ␣E11 mutants clearly suppress the resonance signal at 12.2 ppm. Apparently, these ␣-subunit mutations interfere with the H-bond between ␣His-103 and ␤Gln-131 (42).
Overall, in the deoxy form the substitutions that we have made at the E11 position have not disturbed the ␣ 1 ␤ 2 interfaces. The ␣-subunit mutants have a weaker H-bond between ␣His-103 and ␤Gln-131, which are ␣ 1 ␤ 1 interface residues. The ␣-subunit mutations slightly affect the proximal histidyl residue, whereas the ␤E11Val to phenylalanine replacement creates a more pronounced disturbance on the proximal histidyl residue. In addition, interactions between the porphyrin ring  and heme pocket residues have been perturbed in rHb (␣V62I), rHb (␣V62F), and rHb (␤V67F).

DISCUSSION
Hemoglobin functions as an O 2 carrier and the heme pocket is the ligand binding/active center of the protein. In the center of the cavity are the proximal F8His and the distal E7His. F8His covalently binds the heme onto the protein for O 2 binding, whereas E7His stabilizes the bound O 2 with hydrogen bonds (52,53) and acts as a gate for ligand entry (52,54).
Two other important residues within the reaction center are B10Leu and E11Val (Fig. 1). Hemoglobin mutants carrying an aromatic replacement at the B10 position have been characterized by several groups (6,13,15,31,52,55). Wiltrout et al. (15) substituted B10Leu with either phenylalanine or tryptophan. Among the four mutants, only rHb (␣L29F) has a similar autox-  idation rate and a higher O 2 affinity than Hb A (Fig. 2). The other three mutants bind O 2 with low affinity and have higher autoxidation rates than that of the wild type protein (15). In general, the two ␣-subunit mutants have lower azide-and NOinduced oxidation rates, whereas the two ␤-subunit mutants have a lower NO-induced but higher azide-induced oxidation rates than Hb A (6,15). Presumably, an aromatic residue at the B10 position decreased the volume of the distal pocket and creates a steric hindrance that hinders the accessibility of the iron atom to solvent water molecules, NO, and azide anions, resulting in diminished autoxidation and induced oxidation rates. This interpretation is consistent with O 2 and CO binding and CO geminate rebinding studies in rHb(␣V1M/L29F, ␤V1M) and rHb(␣V1M/L29W, ␤V1M) triple mutants (31). In that work placement of the aromatic residues at position B10 was shown to dramatically reduce (i) the fraction of geminate recombination and (ii) association rate constants for ligand binding, demonstrating that the mutations interfere with ligand entry and bond formation in the ␣-subunit. The phenylalanine substitution further stabilizes the bound oxygen with the positive edge of the phenyl ring multipole and consequently increases the oxygen binding affinity (5). The distal heme pocket of the ␤-subunit assumes a comparatively more open structure in the R-state (9), and similar B10 substitutions are less effective in lowering the autoxidation rates.
E11Val occupies a unique position in the distal heme pockets. Its methyl ␥ 2 is 5.2 and 4.2 Å from the iron atom of the porphyrin group of the deoxy Hb A ␣and ␤-subunits, respectively (56). According to the 1.25 Å resolution crystal structural model (PDB ID 2DN1 (57)), the distance between the methyl ␥ 2 of E11Val and the bound O 2 atom is 4.07 and 3.52 Å in the ␣and the ␤-chains of HbO 2 A, respectively (Fig. 1). Apparently, ␤⌭11Val can restrict or interfere with ligand binding, and this restriction is exerted to a lesser extent by ␣⌭11Val in the ␣-subunits.
We have replaced the ␣Val (E11) with leucine, isoleucine, and phenylalanine in this study. These mutants have similar cooperativity (n 50 ) and alkaline Bohr Effect ( Fig. 2 and Table 1) as Hb A. The O 2 binding affinities (P 50 ) of these mutants are also similar to that of Hb A at neutral or basic pH. Under acidic conditions, rHb (␣V62L) and rHb (␣V62F) have identical O 2 binding curves, and the affinities are slightly higher than that of Hb A. As for the rHb (␣V62I) mutant, it exhibits a slightly lower O 2 binding affinity than that of Hb A at acidic pH (Fig. 2). Similar observations have been reported by Mathews et al. (10) for the rHb (␣V62I) mutant and by Maillett (31) for the rHb(␣V1M/L29F, ␤V1M) mutant.
Leucine has a flexible aliphatic side chain. Replacing ␣E11Val with leucine causes minimal disturbance to the structure of the macromolecule in the R-state (Fig. 3). In the T-state (Fig. 4B), we have detected a weakening of an H-bond (between ␣His-103 and ␤Gln-131) located in the ␣ 1 ␤ 1 interfaces, and this phenomenon is common for all three ␣-subunit mutants. According to the x-ray crystallographic model (57), ␣Val-62 is Ͼ16 Å from ␣His-103. The replacements at ␣E11 probably affect ␣His-103 via ␣His-87, the proximal histidine. Our hyperfine-shifted protein resonance spectra (Fig. 4A) indicate a shift in the ␣His-87 signal for these three ␣-subunit mutants. The main-chain oxy-gen of ␣His-87 forms H-bonds with ␣Arg-92 and ␣Val-93. These two residues are part of a loop that connects and forms H-bonding network with the ␣G helix, where ␣His-103 is located. It is of interest to note that a ␣His-87 to glycine substitution can shift the signal from the NH ⑀1 of ␣His-122 upfield (58). ␣His-122 is another residue located in the ␣ 1 ␤ 1 interface and over 22 Å from ␣His-87. Further experiments are needed to test our speculation.
Phenylalanine has a bulkier side chain than that of leucine. Consequently, slight changes in the tertiary structure of the heme pocket in the R-state (Fig. 3B) and interactions among the porphyrin ring and surrounding amino acids (Fig. 4B) have been noticed for rHb (␣V62F). However, these structural alterations have not translated into functional changes, and the O 2 binding affinities of rHb (␣V62F), rHb (␣V62L), and Hb A are similar. These observations are consistent with measurements of O 2 binding to a rHb (␣V1M/V62F, ␤V1M) mutant, which showed that ligand association was promoted by removal of the ␥-CH 3 group but countered by steric hindrance to ligand capture, resulting in little impact on O 2 affinity (31). The increase in side-chain volume from leucine to phenylalanine at the ␣E11 position does not significantly change O 2 binding ( Fig. 2A) or the azide-induced oxidation (Table 2) rates of the macromolecule.
Tame et al. (59) have determined the crystal structures of rHb (␣V62L) and rHb (␣V62I) in the deoxy state. The extra methyl group of the isoleucine side chain has replaced a water molecule found inside the heme pocket of the ␣-subunits of hemoglobins. Our 1 H NMR analyses indicate that the ␣E11Valto-isoleucine substitution causes a change in the porphyrin ring interacting in the T-state of the mutant (Fig. 4B). The nonexchangeable ring current-shifted proton spectrum between Ϫ0.5 to Ϫ1.0 ppm also suggests a slight change in the tertiary structure of the heme pocket of rHb (␣V62I) in the R-state. However, none of these structural changes significantly affect the O 2 binding affinity of rHb (␣V62I) (Fig. 2A).
The presence of the ␥-CH 3 group in the ␣-subunit appears to slightly inhibit azide-induced oxidation. Rate constants measured for Hb A and rHb (␣V62I) are similar and lower than those observed for rHb (␣V62L) and rHb (␣V62F) ( Table 2). This result implies that removal of the ␥-CH 3 group at position 62 facilitates azide anion access to the iron atom. In contrast, rates of autoxidation were similar in rHb (␣V62L), rHb (␣V62I), and rHb (␣V62F) but still 3.2-5.5-fold higher than that of Hb A ( Table 2).
Due to the close proximity of ␤E11Val to the ligand binding site, a selected mutation at this position can affect the activity of hemoglobin. The crystal structure of deoxy rHb (␤V67L) has been resolved at 2.18 Å. The ␤E11Leu side chain is disordered, and the mutation has caused a small movement in the E-helix (11). Our hyperfine-shifted proton resonance spectrum for the T-state of rHb (␤V67L) also indicates a slight shift in the signal that normally appears at 22.6 ppm (Fig. 4B). With the disappearance of the signal at Ϫ1.0 ppm, the ring-current shift proton resonance spectrum (Fig. 3B) suggests a slight structural change in the heme pocket of this mutant at the R-state. These changes do not significantly affect the O 2 binding affinity of rHb (␤V67L) (Fig. 2 and Ref. 11). Presumably, due to the flexibility of the leucine side chain, it is easier for the oxygen and azide anions to access the heme irons. rHb (␤V67L) has a higher autoxidation rate and azide-induced oxidation rate than Hb A ( Table 2). The situation is analogous to that of Hb Sydney, which has a ␤E11Val-to-alanine replacement. With a smaller side chain at the ␤E11 position, Hb Sydney is easily autoxidized and unstable (60).
Like the valine side chain, a branched isoleucine side chain on the E-helix cannot rotate about the C ␣ -C ␤ bond, and its ␦-methyl group shields the binding site and hinders oxygen binding (11). The conformations of rHb (␤V67I) in both T-and R-states are similar to those of Hb A. Among the 1 H NMR spectra generated for this mutant, we have detected a missing peak at Ϫ1.82 ppm (assigned to the ␥ 2 -CH 3 group of ␤Val-67) from the ring current shift proton resonance spectrum for the R-state of the mutant. Compared with valine, the bulkier side chain of isoleucine effectively hinders the binding of oxygen and azide anions. Consequently, rHb (␤V67I) has a lower binding affinity for O 2 (Fig. 2) and similar azide-induced oxidation rate ( Table 2). The autoxidation rate of rHb (␤V67I) is only 3.8 times higher than that of Hb A.
Unlike the corresponding ␣-subunit mutation, placement of a phenylalanine residue at the ␤E11 position significantly changes the heme pocket of the mutant. Noticeably, in the T-state, the signal at 76 ppm and assigned to the N ␦ H of ␤His92 shifted downfield (Fig. 4A), whereas the signal at 22.5 ppm and assigned to the ␤-subunit porphyrin ring shifted upfield. The spectra for rHb (␤V67F) at the R-state are similar to those of Hb A and rHb (␤V67L) (Fig. 3). Therefore, the ␤E11Val to phenylalanine substitution affects the proximal ␤His-92 and the interactions between the porphyrin ring and surrounding amino acid residues when the macromolecule assumes the T-state conformation. This substitution does not cause meaningful structural changes to the protein in the R-state conformation.
The structure of rHb (␤V67F) has not been elucidated. However, an x-ray crystallographic model for the corresponding myoglobin mutant (V68F) is available (61). The benzyl side chain of the myoglobin mutant points away from the ligand binding site and packs at the back of the distal pocket. Consequently, the heme iron is more readily accessible for ligand binding than that of the myoglobin ␤V67L and ␤V67I mutants. Presumably, the ␤Phe-67 side chain on rHb (␤V67F) can assume a similar conformation and its preference to stay in the R-state, resulting in a drastic increase in oxygen binding affinity (Fig. 2).
Studies of geminate recombination, O 2 equilibrium binding, and ligand association and dissociation reactions are consistent with this interpretation (31). In that work, rHb (␣V1M, ␤V1M/ V62F) was shown to have decreased rates of O 2 association due to steric hindrance of the bulky aromatic ring, but this effect was more than counteracted by a marked decrease in O 2 dissociation rates attributed to removal of the ␥CH 3 group, resulting in a substantial increase in O 2 affinity. Fitting of the equilibrium binding curve to a two-state model indicated a decrease in the R 7 T equilibrium constant (L), suggesting a possible structural mechanism consistent with the observations made in this work.
Even though phenylalanine has a bulky aromatic side chain and is expected to effectively decrease the distal pocket volume, it is not as effective as the methyl ␥ 2 of the native valine side chain in protecting the iron atom of the heme. The k az of rHb (␤V67F) is 3.6-fold higher than that of Hb A. The autoxidation rate of rHb (␤V67F) is 7-fold higher than that of Hb A and comparable to that of rHb (␣L29W) (15). Eich et al. (6) determined the NO-induced oxidation rate of rHb (␤V67F) in the R-state, and it is 4.5-fold slower than that of Hb A. The higher azide-induced (this work) and lower NO-induced (6) oxidation rates of rHb (␤V67F) compared with those of Hb A illustrate differences in the mechanisms of these oxidation reactions. NO-induced oxidation depends upon the rate of NO entry into the pocket, whereas azide-induced oxidation occurs on a slower time frame and is dependent upon the affinity for azide. In rHb (␤V67F), the presence of the aromatic group in the ligand binding pocket interferes with ligand entry, slowing NO-induced oxidation, but removal of the ␥CH 3 group may increase azide affinity, explaining the increase in azide-induced oxidation based on the findings from the myoglobin study (62). This interpretation is consistent with CO bimolecular binding, and geminate rebinding results in Hb(␣V1M, ␤V1M/V62F) mutant that found k entry Ј was reduced by the aromatic substitution (31). A more detailed structural model of rHb (␤V67F) is needed to delineate whether and how additional residues located in the heme pockets participate in the mechanism. It is reasonable to postulate that the tertiary structure of the distal heme-pocket is of primary importance in determining rates of autoxidation and ligand binding. However, the differential stability of the Hb mutants in their R-and T-states contributes to the oxygen affinity of the proteins.
We have recently employed recombinant octameric hemoglobins carrying ␣B10Leu to phenylalanine or tryptophan replacements as resuscitation solutions in mice after experimental traumatic brain injury plus hemorrhagic shock (34). Both mutants have reduced rates of NO-induced oxidation. However, the ␣L29F mutant has high, whereas the ␣L29W mutant has low O 2 binding affinity (15). We have found that the high O 2 affinity ␣L29F mutant has the lowest numerical brain tissue oxygen concentration and unexpectedly a better neuronal survival (34). Interestingly, Eaton et al. (63) have reported that Hb with higher O 2 affinity gave better protection to animals placed at greatly reduced environmental oxygen pressures. Deer mice adapted to high altitude also carry Hb isoforms with higher intrinsic O 2 affinities (64). Furthermore, Cole et al. (65) proposed a mathematical model for O 2 transport and postulated that cell-free oxygen carriers should have P 50 in the range of 5-15 mm Hg, which is higher or equivalent to that of Hb A. The rHb (␤V67F) in this study has an O 2 binding affinity higher than that of the ␣L29F mutant (Fig. 2), and it has reduced NO-induced oxidation rate (6). Thus, our work here has further advanced the understanding of the structure-function relationship of hemoglobin. Incidentally, it also provides a candidate (rHb (␤V67F)) to be used in resuscitation solutions and can be tested in a mouse model representing traumatic brain injury combined with hemorrhagic shock.