Substitution of the heme binding module in hemoglobin alpha- and beta-subunits. Implication for different regulation mechanisms of the heme proximal structure between hemoglobin and myoglobin.

In our previous work, we demonstrated that the replacement of the "heme binding module," a segment from F1 to G5 site, in myoglobin with that of hemoglobin alpha-subunit converted the heme proximal structure of myoglobin into the alpha-subunit type (Inaba, K., Ishimori, K. and Morishima, I. (1998) J. Mol. Biol. 283, 311-327). To further examine the structural regulation by the heme binding module in hemoglobin, we synthesized the betaalpha(HBM)-subunit, in which the heme binding module (HBM) of hemoglobin beta-subunit was replaced by that of hemoglobin alpha-subunit. Based on the gel chromatography, the betaalpha(HBM)-subunit was preferentially associated with the alpha-subunit to form a heterotetramer, alpha(2)[betaalpha(HBM)(2)], just as is native beta-subunit. Deoxy-alpha(2)[betaalpha(HBM)(2)] tetramer exhibited the hyperfine-shifted NMR resonance from the proximal histidyl N(delta)H proton and the resonance Raman band from the Fe-His vibrational mode at the same positions as native hemoglobin. Also, NMR spectra of carbonmonoxy and cyanomet alpha(2)[betaalpha(HBM)(2)] tetramer were quite similar to those of native hemoglobin. Consequently, the heme environmental structure of the betaalpha(HBM)-subunit in tetrameric alpha(2)[betaalpha(HBM)(2)] was similar to that of the beta-subunit in native tetrameric Hb A, and the structural conversion by the module substitution was not clear in the hemoglobin subunits. The contrastive structural effects of the module substitution on myoglobin and hemoglobin subunits strongly suggest different regulation mechanisms of the heme proximal structure between these two globins. Whereas the heme proximal structure of monomeric myoglobin is simply determined by the amino acid sequence of the heme binding module, that of tetrameric hemoglobin appears to be closely coupled to the subunit interactions.

the heme cavity, and they are mainly modulated by the heme environmental structure (1)(2)(3)(4). In particular, the heme proximal structure is one of the most important determinants for the regulation of heme activities. For example, significant anionic character of the proximal His-174 in cytochrome c peroxidase arising from a strong hydrogen bond between its N ⑀ H and carboxylate of Asp-235 (5) is the primary cause for its extremely low redox potential and stabilization of higher oxidation states of the heme. Also, the proximal His-93 in vertebrate myoglobins is tightly constrained by a hydrogen-bonding network extending from His-93 to the heme 7-propionic acid group and the imidazole of His-97 (6), which is responsible for their considerably lower ligand affinity compared with that of another monomeric globin, leghemoglobin (7). Indeed, recent crystallographic studies on several hemoproteins together with the protein engineering studies report that the proximal ligand, its heme coordination structure, and its local environment virtually determine function and reactivity of hemoproteins (8 -12). Therefore, it is a fundamental problem to understand how the heme proximal structure is regulated in hemoproteins.
In our previous paper (13), we have proposed the heme binding module, a protein continuous segment regulating the heme proximal structure in the globin family. The "module" is defined as a compact structural unit on the protein structure, which consists of 10 -40 amino acid residues. Originally, the globin structure was decomposed into four modules (M1-M4), each of which corresponds to the exon on the gene structure (14). By engineering many kinds of module-substituted globin proteins and investigating their structural and functional properties, we have shown that the module can be a structural and functional unit that has advantages in producing novel artificial globin proteins, although interactions between the modules are essential for the stable structures (15)(16)(17). However, we have also succeeded in producing a novel functional globin protein by the substitution of the pseudo-module, 1 of which the boundaries are located at the middle of the modules (18). The ␤␣(PM3)-subunit, where the pseudo-module PM3, the segment from the center of the module M3 to that of the module M4 ( Fig.  1), of hemoglobin ␣-subunit was transplanted into hemoglobin ␤-subunit, exhibited the ␣-subunit-type association property, just as did the module M4-substituted ␤-subunit, ␤␣(M4)-subunit (18). Moreover, NMR analysis demonstrated that the heme coordination structure of the proximal histidine in the ␤␣(PM3)-subunit was converted into the ␣-subunit type (18).
Since the change in the heme coordination structure was not observed for the ␤␣(M4)-subunit, it is inferred that the Nterminal half of the pseudo-module PM3, which is outside the module M4 (Fig. 1), serves to modulate the heme coordination structure. These results on the ␤␣(PM3)-subunit lead us to believe that the boundaries of the pseudo-module PM3, F1 and H6 sites, have some structural and functional significance. Excitingly, a new intron has been discovered near the F1 site on the gene of the Artemia hemoglobin (19), which strongly suggests that the F1 site actually has genetic or evolutionary meanings. Accordingly, the N-terminal half of the pseudo-module PM3 is supposed to be a fundamental segment on globin structure and is referred to as the heme binding module (Fig. 1).
Structural significance of this segment was also confirmed by studies on Mb␣(HBM) 2 -globin, in which the heme binding module of myoglobin was replaced by that of hemoglobin ␣-subunit (13). On the basis of a variety of spectroscopic data including absorption, NMR, and resonance Raman spectra, heme electronic state and heme coordination structure of the proximal histidine in the Mb␣(HBM)-globin were quite similar to those in the ␣-subunit, not to those in myoglobin. This structural conversion in the Mb␣(HBM)-globin clearly demonstrated that the heme binding structure is regulated by the heme binding module.
In the present study, to further examine the structural regulation by the heme binding module in hemoglobin, we have synthesized a novel chimeric globin, ␤␣(HBM)-subunit, where the heme binding module of hemoglobin ␤-subunit was replaced by that of the ␣-subunit (Fig. 1), and investigated its heme environmental structure in detail. Unfortunately, the isolated ␤␣(HBM)-subunit was so unstable that its oxy and deoxy forms were hardly characterized to confirm the structural conversion by the module substitution. In the presence of native ␣-subunit, however, the ␤␣(HBM)-subunit was stabilized by forming ␣ 2 [␤␣(HBM) 2 ] tetramers, affording similar NMR and resonance Raman spectra to those of the ␤-subunit in tetrameric Hb A. In the results, we did not find the structural conversion by the module substitution for the ␤␣(HBM)-subunit in the complex with native ␣-subunit, which implies that the structural regulation mechanism for hemoglobin subunits in the tetramer is different from that for myoglobin. In contrast to monomeric myoglobin, it is likely that the heme binding structure of tetrameric hemoglobin is closely coupled to the subunit interactions rather than to amino acid sequence of the heme binding module.

EXPERIMENTAL PROCEDURES
Construction of Expression Vector-To construct the gene of the ␤␣(HBM)-subunit, KpnI (GGTACC) and SacI (GAGCTC) sites were introduced at the boundaries of the heme binding module in the ␣and ␤-subunits, both, by polymerase chain reaction (Fig. 1), accompanied with the mutation Arg(G6) to Glu. 3 The small KpnI-SacI fragment of the ␣-subunit amplified by polymerase chain reaction was ligated with the digested T7 expression vector encoding the ␤-subunit. Methionine was substituted for valine at the N terminus to initiate the peptide elongation for the globins. Construction of the desired expression vector was verified by the double-stranded DNA sequence analysis (373 DNA sequencer, Applied Biosystems).
Protein Preparation-The prepared gene encoding the ␤␣(HBM)subunit was transformed into an Escherichia coli strain (BL21), which was grown at 37°C overnight in 2ϫ TY (tryptone-yeast extract) culture medium containing ampicillin (100 g/ml). The expressed subunit was purified as previously reported for recombinant Hb (20,21). We confirmed the correct expression of the desired subunits by fast atom bombardment mass spectroscopy (data not shown) (22), and no additional mutations were detected. We also synthesized the wild-type ␤-subunit, 4 which has methionine at the N terminus instead of valine as the reference and confirmed that the structural properties of the wildtype ␤-subunit are virtually the same as those of native ␤-subunit isolated from human red blood cell.
The carbon monoxide derivative was prepared by adding minimal amounts of sodium dithionite to the ferric globins under CO atmosphere. The globin solution was deoxygenated by repeated evacuation and flushing with N 2 gas under gentle shaking, and complete deoxygenation was achieved by the addition of minimal amount of sodium dithionite.
Gel Chromatogram-Gel filtration measurements were performed by using a Sephacryl S-200 high resolution column (0.8-cm diameter ϫ 62-cm length) at 4°C. The buffer used for the chromatography was 50 mM Tris-HCl in the presence of 0.1 M NaCl and 1 mM Na 2 EDTA at pH 7.4, and the flow rate was 7 ml/h. The eluted fractions were monitored by absorption at the Soret band (420 nm). Tetramer-dimer dissociation constants of the samples were determined by the concentration dependence of the elution volume in the column over the range from 0.5 to 800 M (23,24). The dependence of the centroid elution volume V e versus protein concentration (C T ), Equation 1, allows us to determine the dimer-tetramer equilibrium constants for the samples, where V j is elution volumes for the individual species pertaining to the various aggregates (j-mers), and the (m j ) term represents molar concentration for the respective species.
Spectral Measurements-1 H NMR spectra at 500 MHz were recorded 2 The abbreviation used is: HBM, heme binding module. 3 According to the x-ray crystallographic study of hemoglobin, the side-chain of Arg(G6) is located on the protein surface of the ␤-subunit, and it does not directly contribute to the heme contact. The mutation at the G6 position, as found in the abnormal hemoglobin, Hb Sherwood Forest, has minor effects on the stability of the globin structure (69), although its oxygen affinity is slightly higher than that of normal hemoglobin (70). In fact, Glu(G6) is also found in other globins such as mammalian myoglobin. Therefore, it is unlikely that the mutation, Arg(G6) to Glu, in the ␤␣(HBM)-subunit is a serious problem for the evaluation of structural significance of the heme binding module in hemoglobin. 4 Wild-type subunit represents the protein expressed in E. coli; a methionine residue is located at the N terminus. "Native" subunit corresponds to the protein purified from human red blood cell. In the wild-type ␤-subunit, we confirmed that the mutation of Val to Met does not seriously perturb the globular structure and heme environmental structure of the ␤-subunit. The CD and NMR spectra for the wild-type ␤-subunit were virtually the same as those of the native ␤-subunit. on BRUKER Avance DRX 500 spectrometer equipped with the Indy workstation (Silicon Graphics). To measure proton resonances in the diamagnetic region, we used a water gate pulse sequence with 50-ms pulse and 33,000 data points over a 13-kHz spectral width (25). The hyperfine-shifted proton resonances were obtained by using a LOSAT pulse sequence with a 65,000 data transform of 150 kHz and an 8.5-s 90°K pulse. The probe temperature was controlled at 290 Ϯ 0.5 K by a temperature control unit of the spectrometer. The volume of the NMR sample was 600 l, and the protein concentration was approximately 1 mM on the heme basis. The buffer was 20 mM sodium phosphate containing 0.1 M NaCl at pH 7.4. Proton shifts were referenced with respect to the proton resonance of 2,2-dimethyl-2-silapentane-5-sulfonate.
Resonance Raman scattering was excited at 441.6 nm with a He/Ca laser (Kinmon Electronics, CDR80SG) and detected by a JEOL-400D Raman spectrometer equipped with a cooled CCD camera. The frequency of the Raman spectrometer was calibrated with indene. Sample concentration was 40 M on the heme basis. The buffer was 20 mM sodium phosphate, 0.1 M NaCl at pH 7.4.
Oxygen Equilibrium Curves and Analysis-Oxygen equilibrium curves were measured by using an improved version (26,27) of an autooxygenation apparatus (28). The wavelength of the detection light was 560 nm. The buffer was 50 mM Tris-HCl, 0.1 M NaCl at pH 7.4. The temperature of the sample in the oxygenation cell was constant at 25 Ϯ 0.05°C. Sample volume was 6 ml, and concentration was 100 M on the heme basis. The hemoglobin reductase system (29) was added to the sample before each measurement to reduce oxidized subunits. To minimize the autooxidation of the sample during the measurements, catalase and superoxide dismutase were added to the sample, and the concentration was 0.1 mM (30,31). The oxygenation data were acquired by use of a microcomputer (model PC-98XA, Nippon Electric Co., Tokyo), which was interfaced to the oxygenation apparatus (32).

RESULTS
Association Properties-Before analyzing the heme environmental structure of the ␤␣(HBM)-subunit, we investigated its association property with native hemoglobin subunits, since the heme environmental structure of globin proteins is very sensitive to the subunit assembly (33,34). Fig. 2A represents gel chromatograms of the carbonmonoxy chimeric subunit in the presence and absence of the native subunits. Under the condition employed here, the mixture of the native ␣and ␤-subunits forms a tetramer, whereas the isolated ␣-subunit remains a monomer (23). The ␤-subunit is in equilibrium between monomers and tetramers (23). As shown in Fig. 2A, the position of the elution peak for the ␤␣(HBM)-subunit corresponds to that of the ␣-subunit, indicative of its monomeric structure. In the chromatogram for the mixture of the ␤␣(HBM)and ␤-subunits, a single broad peak was observed, but this elution pattern can be simulated by simple addition of those of the isolated subunits. On the other hand, the mixture of the ␤␣(HBM)and ␣-subunits showed a single broad peak that cannot be reproduced by the simple addition ( Fig. 2A). This broad peak was significantly shifted to the lower elution volume from that of the isolated ␤␣(HBM)or ␣-subunit. These elution patterns, therefore, demonstrate that the ␤␣(HBM)subunit was preferentially associated with the ␣-subunit not with the ␤-subunit.
However, the dissociation constant between the ␤␣(HBM)and ␣-subunits is much larger than that between native ␣and ␤-subunits. Fig. 2B delineates the centroid elution volumes of the samples as a function of protein concentration, which clearly shows that the complex of the ␤␣(HBM)and ␣-subunits is in equilibrium between heterodimeric ␣[␤␣(HBM)] and heterotetrameric ␣ 2 [␤␣(HBM) 2 ]. The fitting curve of the complex shifts to the right side from that of Hb A (Fig. 2B), revealing that the dissociation into the dimers was enhanced for the complex. The tetramer-dimer dissociation constants, K D values, were estimated as 1.4 and 40 M for Hb A and the complex of the ␤␣(HBM)and ␣-subunits, respectively.
Subunit Interface Structure-To investigate the subunit interface structure of heterotetrameric ␣ 2 [␤␣(HBM) 2 ], we meas-ured the 1 H NMR spectra in the hydrogen-bonded proton region for the carbonmonoxy and deoxy forms (Figs. 3, A and B). In the downfield region from 10 to 15 ppm, three exchangeable proton signals were observed for carbonmonoxy Hb A, which have been assigned to the hydrogen bonds in the subunit interface, as described in the figure legends (35)(36)(37). The ␣ 2 [␤␣(HBM) 2 ] tetramer also exhibited these three exchangeable resonance peaks at almost the same positions as Hb A (Fig. 3A), and the spectral feature in this region was quite similar to that of Hb A. Dissociation of the ligands from the heme iron induces the structural rearrangements on the ␣ 1 ␤ 2 subunit interface in Hb A (35,36), which is clearly reflected in the downfield region of the NMR spectra (Fig. 3B). Additional exchangeable proton resonances were observed at 13.9 and 11.0 ppm for deoxy Hb A (Fig. 3B), assignable to the hydrogen bonds of ␣1 Tyr(C7)-␤2 Asp(G1) (35) and ␣1 Asp(G1)-␤2 Trp(C3), respectively (38). The deoxy-␣ 2 [␤␣(HBM) 2 ] tetramer also afforded these additional peaks, although the peak positions slightly moved from the positions for Hb A (Fig. 3B). Thus, the spectral features of the ␣ 2 [␤␣(HBM) 2 ] tetramer were almost identical with those of Hb A in the carbonmonoxy and deoxy states, implying that the structural rearrangements on the ␣ 1 -␤ 2 subunit interface induced by deoxygenation in the ␣ 2 [␤␣(HBM) 2 ] tetramer correspond to those in native Hb A.
Heme Environmental Structure in Deoxy and Carbonmonoxy Forms-By using a combination of spectroscopic methods, we examined structural effects of the module substitution on the heme vicinity in the deoxy and carbonmonoxy states. Fig. 4 represents the 1 H NMR spectra of the deoxy chimeric and native globins. In the far downfield region, the hyperfineshifted proton signals from the proximal histidine are observed for hemoglobin subunits and tetrameric Hb A (Fig. 4A) (38,39) In deoxygenated tetrameric Hb A, the proximal histidyl N ␦ H protons in the ␣and ␤-subunits are detected at 64 and 76 ppm, respectively. The ␣ 2 [␤␣(HBM) 2 ] tetramer also afforded the N ␦ H proton resonance at 64 and 76 ppm, which are almost the same positions as those of native Hb A. Such a coincidence of the N ␦ H proton resonances between the ␣ 2 [␤␣(HBM) 2 ] tetramer and Hb A demonstrates that the heme coordination structure of the proximal His in the ␣ 2 [␤␣(HBM) 2 ] tetramer corresponds to that in Hb A. In the 12-30-ppm region, the spectral pattern of the ␣ 2 [␤␣(HBM) 2 ] tetramer was quite similar to that of Hb A, although the subtle line-broadening was observed. Since the resonances in this region are originated from the protons of the heme peripheral groups (40), the similar spectral pattern indicates that heme electronic state of Hb A did not significantly change upon the substitution of the heme binding module. Thus, the transplantation of the heme binding module from the ␣-subunit to the ␤-subunit gave minor effects on the heme environmental structure of the ␤-subunit in the ␣ 2 [␤␣(HBM) 2 ] tetramer. On the other hand, the isolated deoxy-␤␣(HBM)subunit exhibited extremely rapid autooxidation and considerable aggregation during the preparation, suggestive of its seriously perturbed heme environmental structure. NMR measurement for the isolated deoxy-␤␣(HBM)-subunit was not successful due to the low stability.
To gain further insights into the heme coordination structure of the proximal histidine, we also measured the resonance Raman spectra of the globins in the deoxy form. Fig. 5 shows the low frequency region of the Raman spectra measured with a 1.00 cm Ϫ1 resolution. The stretching mode of the Fe-His bond, (Fe-His), for the ␣and ␤-subunits were observed at 220 and 222 cm Ϫ1 , respectively (41). For tetrameric Hb A, the peak position of the (Fe-His) was shifted by 7 cm Ϫ1 to the lower wave number from that of the isolated ␤-subunit, indicating that the structural strain is imposed on the Fe-His bond by association of the ␣and ␤-subunits (42). In the complex of the ␤␣(HBM)and ␣-subunits, the stretching mode of the Fe-His bond was observed at 215 cm Ϫ1 , revealing that the Fe-His bond in deoxygenated tetrameric hemoglobin is insensitive to the substitution of the heme binding module. 1 H NMR spectrum for the carbonmonoxy form is also useful for characterization of the heme environmental structure. As shown in Fig. 6, a peak from ␥ 1 -methyl proton of Val (E11) appeared at Ϫ2.0 and Ϫ2.2 ppm for the carbonmonoxy ␣and ␤-subunits, respectively, which can serve as a marker for the heme environmental structure (33, 43, 44). The corresponding signals of Hb A and the ␣ 2 [␤␣(HBM) 2 ] tetramer were detected at almost the same position (Fig. 6), suggesting that the local FIG. 3. A, NMR spectra in the hydrogen-bonded proton region for carbonmonoxy globins. Proton resonance peaks at 10.4, 11.9, and 12.8 ppm for native tetrameric Hb A have been assigned to the hydrogen-bonded protons between ␣1 Asp(G1) and ␤2 Asn(G4) (35), ␣1 His(G10) and ␤1 Asn(G10) (36), and ␣1 Asp(H9) and ␤1 Tyr(C1) (36,37), respectively. Experimental conditions were as follows: 20 mM sodium phosphate, 0.1 M NaCl, pH 7.4, at 290 K. Sample concentration was 1 mM on the heme basis. B, NMR spectra in the hydrogen-bonded proton region for deoxy globins. Proton resonance peaks at 11.0, 12.2, 12.9, and 13.9 ppm for native tetrameric Hb A have been assigned to the hydrogen-bonded protons between ␣1 Asp(G1) and tetramer is similar to that in Hb A. On the contrary, the signals between Ϫ0.5 and Ϫ1.5 ppm from other heme-surrounding residues were significantly different between Hb A and the ␣ 2 [␤␣(HBM) 2 ] tetramer, indicative of some conformational changes in the heme vicinity by the module substitution. Such a conformational change seems to be enhanced for the isolated ␤␣(HBM)-subunit. The resonance peak from ␥ 1 -methyl proton of Val (E11) was detected around Ϫ2.0 ppm, as was in the native subunits, but it is highly asymmetric and split (Fig. 6), suggesting the heterogeneity in its heme environmental structure.
Heme Environmental Structure in Cyanomet Form-Although the cyanomet form is not biologically active in globin proteins, the 1 H NMR spectra afford various structural information for liganded hemoglobin (45)(46)(47)(48). As illustrated in Fig.  7, 5-, 1-methyl, and 2-vinyl C ␣ proton signals have been assigned for the human hemoglobin ␣and ␤-subunits and tetrameric Hb A (49). Noteworthy here is that the spectral feature of the cyanomet-␣ 2 [␤␣(HBM) 2 ] tetramer was also virtually identical to that of Hb A, indicating that the heme electronic state of the ␣ 2 [␤␣(HBM) 2 ] tetramer was not significantly affected by the module substitution. For the isolated ␤␣(HBM)subunit, the resonances from the heme methyl and vinyl groups were detected at almost the same positions as those for the isolated ␤-subunit, but these resonance peaks were clearly split (Fig. 7). These split peaks were still observed after sufficient reconstitution time with the heme molecule (48 h). These results suggest that there are two kinds of heme local conformations in the isolated ␤␣(HBM)-subunit (47,48), which is consistent with the heterogeneity in the heme environmental structure found for its carbonmonoxy form (Fig. 7).
Oxygen Binding Property-To evaluate effects of the module substitution on the oxygen affinity and cooperativity for the oxygen binding, oxygen equilibrium curves were measured for the complex of the ␤␣(HBM)and ␣-subunits. Fig. 8 represents the oxygen equilibrium curves expressed by the Hill plot. The  P 50 and n max value of native Hb A were estimated to be 4.40 mm Hg and 3.13, respectively. On the other hand, the P 50 value of the complex of the ␤␣(HBM)and ␣-subunits was 3.19 mm Hg, corresponding to higher oxygen affinity compared with that of Hb A. The n max value of the complex was 1.64, and the complex still maintained the cooperative oxygen binding property.

Heme Environmental Structure of the ␤␣(HBM)-subunit-
The present study provides further insights into regulation mechanisms of the heme proximal structure in globin proteins. As shown in the 1 H NMR spectra of the cyanide form (Fig. 7), the isolated ␤␣(HBM)-subunit showed the clear peak split for the resonances from its heme peripheral groups. In addition, the asymmetric and split NMR peak from ␥ 1 -methyl proton of Val (E11) was observed for the carbonmonoxy ␤␣(HBM)-subunit (Fig. 6). These observations correspond to two kinds of heme local conformations (47,48), implying that interactions between the heme and its surrounding residues could not specify one heme local conformation in the isolated ␤␣(HBM)-subunit. The heme environmental structure in the isolated ␤␣(HBM)-subunit was, therefore, seriously perturbed upon the substitution of the heme binding module. The rapid autooxidation observed for the isolated deoxy-and oxy-␤␣(HBM)-subunit could also be attributed to this structural perturbation in the heme cavity.
Noteworthy here is that another module-substituted globin, the Mb␣(HBM)-globin, shows homogeneity of its heme local conformation and forms stable oxy and deoxy derivatives (13), although homology of amino acid sequence in the heme binding module between myoglobin and the ␣-subunit (20%) is considerably lower than that between the ␤and ␣-subunits (50%). One of the possible reasons for these different structural features between the ␤␣(HBM)-subunit and Mb␣(HBM) globin would be their distinct structures in the heme-free state. According to multi-dimensional heteronuclear NMR spectroscopy, the major part of apomyoglobin polypeptide chain adopts a well defined structure that is very similar to that of holomyoglobin. The region around the heme binding module, the F-helix, FGloop, and beginning of the G-helix, is in equilibrium between the holo-protein-like and unfolded or partially folded conformations (50). In the apo␤-subunit, however, several helices are highly disordered, and its total ␣-helical content is nearly half of that for the holo-␤-subunit (51, 52). These results mean that the globin structure of the ␤-subunit would be less stable than that of myoglobin, and the specific heme binding plays more crucial roles in construction of the stable globin structure in the ␤-subunit than in myoglobin. In other words, the globin structure of the ␤-subunit is more sensitive to engineering in the heme binding region than that of myoglobin, and the structural perturbation by the substitution of the heme binding module would, therefore, be enhanced in the ␤␣(HBM)-subunit.
Upon association with the ␣-subunit, however, the ␤␣(HBM)subunit formed stable heme environmental structure. The split resonance peaks from the heme peripheral groups disappeared in the NMR spectra of cyanomet ␣ 2 [␤␣(HBM) 2 ] tetramer (Fig.  7). Also, the deoxy and oxy derivatives were quite stable like native globins, indicative of marginal structural disorder in the heme cavity of the ␤␣(HBM)-subunit. In fact, apo-␣␤-dimer shows remarkable structural recovery from the isolated apo-␣or ␤-subunit, and its ␣-helical content is comparable with that of apomyoglobin (52). Accordingly, it is feasible that the structural perturbation by the module substitution was moderate for the complex of the ␤␣(HBM)and ␣-subunits, as the case for the Mb␣(HBM)-globin.
Of particular note here is that the complex of the ␤␣(HBM)and ␣-subunits exhibited spectral patterns quite similar to Hb A. As shown in Figs. 4 and 5, the NMR and resonance Raman peaks of the complex were observed at almost the same positions as those of Hb A. Since the hyperfine shift of the proximal histidyl N ␦ H proton and the Fe-His vibrational mode in the deoxy state depend on the bond strength between the heme iron and proximal histidine (34,41,42), such similar spectral features indicate that the heme coordination structure of the proximal histidine in the complex of the ␤␣(HBM)and ␣-subunits was identical with that of Hb A. In addition to the heme coordination structure, the ␣ 2 [␤␣(HBM) 2 ] tetramer exhibited almost the same heme electronic state as Hb A. Chemical shifts of the resonances from the heme peripheral groups were quite common between ␣ 2 [␤␣(HBM) 2 ] tetramer and Hb A both in deoxy and cyanomet derivatives (Figs. 4 and 7), suggestive of their identical heme electronic states. Thus, the substitution of the heme binding module did not significantly affect the heme environment of Hb A, and the heme environmental structure of the ␤␣(HBM)-subunit bound to the ␣-subunit was equivalent to that of the ␤-subunit bound to the ␣-subunit.
This finding is interesting in that the substitution of the heme binding module induced different structural effects on Hb A and myoglobin. The heme proximal structure of the Mb␣(HBM)-globin was quite similar to that of the ␣-subunit (13), demonstrating that the substitution of the heme binding module is sufficient to convert the heme proximal structure to the ␣-subunit-type in myoglobin. However, such a structural conversion was not encountered for the complex of the ␤␣(HBM)and ␣-subunits. These different structural effects by the module substitution strongly suggest that regulation mechanism of the heme environmental structure is different in Hb A and myoglobin. One of the key features for the different structural regulations in these two globins is that the hemoglobin subunits, ␣and ␤-subunits, form a ␣ 2 ␤ 2 heterotetramer unlike myoglobin. The heme proximal structure of monomeric myoglobin can simply be determined by the amino acid sequence of the heme binding module. In the ␤-subunit of the tetrameric Hb A, on the other hand, the subunit association with the ␣-subunit significantly affects the heme proximal structure. Comparisons of the spectroscopic properties and x-ray structures between the isolated and ␣-subunit-bound ␤-subunits clearly show some structural difference around the heme proximal structure (34,53). By binding the ␣-subunit, the subunit interface of the ␤␣(HBM)-subunit would be rearranged to that of the ␤-subunit in tetrameric Hb A, which leads to the ␤-subunit-like heme proximal structure in the ␤␣(HBM)-subunit. Based on the fact that the heme binding module includes both the residues contributing to the ␣1-␤2 subunit contact and to the heme contact ( Fig. 1), it is likely that the heme proximal structure constituted by the heme binding module is affected by the ␣1-␤2 subunit interactions. Although the limited structural information of the isolated ␤␣(HBM)-subunit prevents us from drawing the confirmative conclusion, it could be safely said that there are some interplays between the subunit interactions and the heme proximal structure in tetrameric Hb A.
Interestingly, it can be inferred that the interplays in Hb A are deeply related to its cooperative oxygen binding property. On the basis of the Perutz model (54,55), the movement of the proximal histidine induced by the ligand binding or dissociation is linked to structural rearrangement on the subunit interface to control high (R) and low (T) oxygen affinity states. Notably, deletion of the bond between the proximal histidine side chain and the polypeptide prevented the quaternary structure switching, resulting in remarkable decrease of the cooperativity (56). This result strongly supports that the heme proximal structure in native Hb A is closely coupled to the subunit interface structure, and this coupling is a key factor for its cooperative oxygen binding. Thus, structural regulation mechanism in tetrameric Hb A is more complicated than that of monomeric myoglobin, which would reflect long evolutionary history of Hb A since its heterotetramer formation (ϳ1 billion years) (57).
Oxygen Binding for the Complex of the ␤␣(HBM)and ␣-Subunits-The present study on the ␤␣(HBM)-subunit also involves some information on oxygen binding mechanism of Hb A. In the oxygen equilibrium curves (Fig. 8), the complex of the ␤␣(HBM)and ␣-subunits clearly exhibited cooperative oxygen binding, although its n max value was even lower than that of Hb A. The cooperative oxygen binding of the complex can be explained by the structural rearrangement on its subunit interface. As encountered for deoxy Hb A, two additional peaks assignable to the hydrogen-bonded protons on the functional ␣ 1 -␤ 2 interface appeared in the NMR spectra of deoxy ␣ 2 [␤␣(HBM) 2 ] tetramer (Fig. 3), strongly suggesting its T-quaternary structure. Also, since the deoxy-␣ 2 [␤␣(HBM) 2 ] tetramer showed the hyperfine shift of the proximal histidyl N ␦ H proton and the Fe-His vibrational mode at almost the same positions as Hb A (Figs. 4 and 5), the movement of the proximal histidine linked to the structural rearrangement on the ␣ 1 -␤ 2 interface is thought to be induced upon the deoxygenation, as proposed in the Perutz model (54,55). Such a structural rearrangement for the ␣ 2 [␤␣(HBM) 2 ] tetramer is probably due to high sequence homology between the ␣and ␤-subunits in the heme binding module. On the basis of the previous crystallographic and site-directed mutagenesis studies, the positions of FG3-4 and G1-4 in the heme binding module are essential for the allostericity of Hb A (55,58,59), but Leu(FG3), Asp(G1), Pro(G2), and Asn(G4) are common in the ␣and ␤-subunits. Therefore, it is plausible that the hydrogen bonds between ␣1 Tyr(C7) and ␤␣ (HBM) 2 Asp(G1) and between ␣1 Asp(G1) and ␤␣ (HBM) 2 Asn(G4) were formed on the ␣ 1 -␤␣(HBM) 2 interface of the complex, as found for the ␣ 1 -␤ 2 interface of Hb A (see Fig. 3).
However, functional roles of the other two residues at the FG4 and G3 positions are not negligible. Some Hb natural mutants at these positions showed a remarkable decrease in the cooperativity. n max values of Hb Malmo ( ␤ His(FG4) 3 Glu), Hb Alberta ( ␤ Glu(G3) 3 Gly), and Hb Potomac ( ␤ Glu(G3) 3 Gly) were 1.6, 2.0, and 1.6, respectively (60, 61), which are comparable with that of the complex of the ␤␣(HBM)and ␣-subunits. Also, it is to be noted that the complex of the ␤␣(HBM)and ␣-subunits exhibited a much higher tetramerdimer dissociation constant than Hb A. Indeed, other Hb natural mutants, Hb Georgia ( ␣ Pro(G2) 3 Leu) and Hb Hirose ( ␤ Tyr(C3) 3 Ser), whose tetramer-dimer dissociation constants are extremely large (more than 300 times greater than that of Hb A), hardly possess the cooperative oxygen binding property (62,63). Therefore, it can be inferred that the remarkable reduction in the cooperativity observed for the complex of the ␤␣(HBM)and ␣-subunits is caused not only by change in the interactions at the FG4 and G3 positions of the ␤␣(HBM)subunit with adjacent residues in the ␣-subunit but also by the enhancement of the subunit dissociation.
Summary-As described above, the substitution of the heme binding module in hemoglobin ␣and ␤-subunits significantly perturbed the heme environmental structure of the isolated ␤-subunit. The structural disorder in the ␤␣(HBM)-subunit was reduced by the association with the ␣-subunit, and its heme environmental structure was almost identical to that of the ␤-subunit bound to the ␣-subunit. In the results we could not confirm that the heme binding module corresponds to a structural unit regulating the heme proximal structure and heme electric state in globin proteins. In tetrameric Hb A, the subunit interactions appear to play a role in the regulation of the heme environmental structure. Such a structural regulation in Hb A would deeply be related to its cooperative oxygen binding mechanism, in which the structural change of the heme proximal side, especially the proximal histidine, is linked to the structural rearrangement on the subunit interface (55). At present, to characterize the tertiary and quaternary structure of ␣ 2 [␤␣(HBM) 2 ] tetramer, its x-ray crystal structure analysis is in progress. Detailed structure comparison between ␣ 2 [␤␣(HBM) 2 ] tetramer and Hb A will provide further insights into how the local folding of the heme binding module is coupled to the subunit interface structure in Hb A.