Oxygen equilibrium properties of chromium (III)-iron (II) hybrid hemoglobins.

Cr(III)-Fe(II) hybrid hemoglobins, α(Cr)β(Fe) and α(Fe)β(Cr), in which hemes in either the α- or β-subunits were substituted with chromium(III) protoporphyrin IX (Cr(III)PPIX), were prepared and characterized by oxygen equilibrium measurements. Because Cr(III)PPIX binds neither oxygen molecules nor carbon monoxide, the oxygen equilibrium properties of Fe(II) subunits within these hybrids can be analyzed by a two-step oxygen equilibrium scheme. The oxygen equilibrium constants for both hybrids at the second oxygenation step agree with those for human adult hemoglobin at the last oxygenation step (at pH 6.5-8.4 with and without inositol hexaphosphate at 25°C). The similarity between the effects of the Cr(III)PPIX and each subunits' oxyheme on the oxygen equilibrium properties of the counterpart Fe(II) subunits within hemoglobin indicate the utility of Cr(III)PPIX as a model for a permanently oxygenated heme within the hemoglobin molecule. We found that Cr(III)-Fe(II) hybrid hemoglobins have several advantages over cyanomet valency hybrid hemoglobins, which have been frequently used as a model system for partially oxygenated hemoglobins. In contrast to cyanomet heme, Cr(III)PPIX within hemoglobin is not subject to reduction with dithionite or enzymatic reduction systems. Therefore, we could obtain more accurate and reasonable oxygen equilibrium curves of Cr(III)-Fe(II) hybrids in the presence of an enzymatic reduction system, and we could obtain single crystals of deoxy-α(Cr)β(Fe) when grown in low salt solution in the presence of polyethylene glycol 1000 and 50 mM dithionite.

Human adult hemoglobin (Hb A) 1 cooperatively binds four oxygen molecules via a complex sequence of intermediate oxygenated states. Information about the intermediate species is required to understand the cooperative mechanism of Hb A, yet little is known about such intermediates because the equilib-rium concentrations of the intermediates under any conditions are markedly reduced by the cooperativity of Hb.
During recent years, we have investigated the properties of metal-substituted hybrid Hbs, ␣ 2 (M)␤ 2 (Fe) and ␣ 2 (Fe)␤ 2 (M), using the first transition metal ions (M). This metal substitution method is the most suitable modification of Hb A for studying the relationships between the functional states of Hb A and globin-metalloporphyrins interaction. In our systematic investigations, we have observed a wide variation of oxygen affinities in these hybrids, as a result of the variation in the configuration of 3d electrons of the porphyrin metal. With respect to the oxygen affinity of metal-Fe(II) hybrids, we can classify these hybrids into following four groups: (i) hybrids showing oxygen affinities as high as oxy-Hb A, (ii) hybrids showing intermediate oxygen affinities, between oxy-Hb A and deoxy-Hb A, (iii) hybrids showing affinities as low as deoxy-Hb A, (iv) hybrids showing lower oxygen affinities than deoxy-Hb A. To represent the oxygenation intermediates of Hb A, the hybrids in group (i) and (iii) are particularly important. In our recent series of studies, Ni(II)-Fe(II) hybrid Hbs in group (iii) have been used successfully to investigate the structures and functions of the intermediates appearing in the first half oxygenation of Hb A (21)(22)(23)(24)(25)(26)(27)(28). Although the Ni(II)-Fe(II) hybrid system has brought much structural and functional information about the initial-half oxygenated intermediates, this approach could not be extended to studies on the latter-half oxygenation of Hb A. Stable hybrids in group (i) were required to model the intermediates appearing in the oxygenation of the last two sites of Hb A. This paper reports the preparation and oxygen equilibrium properties of Cr(III)-Fe(II) hybrid Hbs. It also shows that these Cr(III)-Fe(II) hybrids are an excellent model for the intermediates appearing in the oxygenation of the last two sites of Hb A. The influence of Cr(III)PPIX on the oxygen equilibrium properties of ferrous subunits within Cr(III)-Fe(II) hybrid Hbs results in quantitative similarities with oxygenated heme under various conditions. This view is reinforced by other structural results, namely (i) similar porphyrin geometry and metalhistidine bonds between Cr(III)PPIX and natural oxyheme, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed. 1 The abbreviations used are: Hb A, human adult hemoglobin; Hb, hemoglobin; ␣ 2 (M)␤ 2 (Fe), hybrid Hb containing metalloporphyrin IX in the ␣ subunits and ferrous protoporphyrin IX in the ␤ subunits; ␣ 2 (Fe)␤ 2 (M), hybrid Hb complementary to the preceding one; Cr(III)P-PIX, chromium(III) protoporphyrin IX; Mn(III)PPIX, manganese(III) protoporphyrin IX; CrHb, hemoglobin containing chromium protoporphyrin IX in both ␣ and ␤ subunits; semihemoglobin ␣, hybrid Hb in which the ␤ subunits do not contain a heme; semihemoglobin ␤, hybrid hemoglobin in which the ␣ subunits do not contain a heme; IHP, inositol hexaphosphate; bis-Tris, 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)-propane-1,3-diol; oxy, oxygenated; met, Hb molecule or chain in which the iron has been oxidized to the trivalent state. and (ii) the occupation of the sixth coordination position of Cr(III)PPIX by H 2 O (or OH Ϫ ) so that the substitution of oxyheme with H 2 O bound Cr(III)PPIX may not significantly alter the heme-pocket structure. With respect to the ease of experiments under anaerobic conditions, Cr(III)-Fe(II) hybrid Hbs have several advantages over cyanomet valency hybrid Hbs as stable and useful models for the oxygenation intermediates.

EXPERIMENTAL PROCEDURES
Preparation of Cr(III)-Fe(II) Hybrid Hbs and CrHb-Cr(III)PPIX was prepared as described by Hori et al. (29). Hb A and its isolated chains, semihemoglobin ␣ and semihemoglobin ␤, were prepared in carbon monoxide forms as described by Fujii et al. (30). The preparation method for Cr(III)-Fe(II) hybrid Hbs was essentially similar to that for Ni(II)-Fe(II) hybrid Hbs (22,31) and that for Porphyrin-Fe(II) hybrid Hbs (30). The preparation of ␣ 2 (Cr)␤ 2 (Fe-O 2 ) was carried out as follows. Bound CO was removed from semihemoglobin ␤ by the method of Kilmartin and Rossi-Bernardi (32). The spectrophotometric titration of semihemoglobin ␤ with Cr(III)PPIX at 440 nm gave a well defined inflection point, from which the molecular stoichiometry of 1:1 was estimated. The solution of semihemoglobin ␤ in oxy form (500 mg) was mixed with an equimolar amount of Cr(III)PPIX, which was dissolved in a minimal amount of N,NЈ-dimethylformamide. The mixture was stirred at 4°C for 2 h and then concentrated by ultrafiltration and passed through a Sephadex G-25 (Pharmacia Biotech Inc.) column equilibrated with 10 mM phosphate buffer, pH 6.85. The sample was applied on a column of CM52 cellulose (Whatman) equilibrated with the same buffer. The column was eluted by a linear gradient of 10 mM phosphate buffer, pH 7.10, and 15 mM phosphate buffer, pH 7.45. A main peak corresponding to ␣ 2 (Cr)␤ 2 (Fe-O 2 ) was collected and concentrated by ultrafiltration. The concentrated sample was passed through a column of Sephadex G-25 equilibrated with 20 mM Tris-HCl buffer, pH 8.2 and stored in liquid nitrogen (yield, 150 mg). Preparation of ␣ 2 (Fe-O 2 )␤ 2 (Cr) was carried out by the same procedure described above, using corresponding constituents, Cr(III)PPIX, and semihemo-globin ␣. For the preparation of CrHb, apo-Hb was used. The apo-Hb was prepared from native Hb A as described by Shibayama et al. (22). The apo-Hb was combined with an equimolar amount of Cr(III)PPIX, and the reconstituted CrHb was purified by the purification procedure described above. All of the preparative works and subsequent manipulations were carried out at 4°C. All of the samples containing Cr(III)P-PIX were treated in the dark, because of their light sensitivity. 2 The purities of all the Hbs were checked by isoelectric-focusing electrophoresis (Pharmacia).
Determination of Oxygen Equilibrium Properties-Oxygen equilibrium curves were measured as described by Imai et al. (33,34). Our automatic oxygenation apparatus was interfaced to a microcomputer (Nippon Electric Co., Tokyo) for on-line data acquisition, storage, and analysis. Light-path length of the cell was variable in this apparatus and was set at 15 mm in the case of 60 M and 12 M (on a metal basis) or at 8 mm in the case of 240 M (on a metal basis). The concentration 2 We found that Cr(III)-Fe(II) hybrid Hbs that had been irradiated with a 100-watt incandescent bulb for 30 min for removal of bound CO showed very high affinity for oxygen with little Bohr effect or little IHP effect. In addition, it was found that when Cr(III)-Fe(II) hybrid Hbs and CrHb were exposed to strong light during isoelectric focusing, these samples gave broad bands. We have not found out the exact cause of such light-induced damage to Cr(III)-Fe(II) hybrid Hbs, but it is likely that some globin moiety of Cr(III)-Fe(II) hybrid Hbs were damaged by O 2 radical produced by interaction between strong light and Cr(III)P-PIX. Since we found this problem, we have prepared Cr(III)-Fe(II) hybrid Hbs in O 2 form, and we have taken care that Cr(III)-Fe(II) hybrid Hbs are not exposed to strong light.
Note that Cr(III)-Fe(II) hybrid Hbs are not damaged by the spectrometer beam for measuring oxygen equilibrium curves. To confirm the stabilities of Cr(III)-Fe(II) hybrid Hbs against monochromatic light in spectrophotometers, we measured the oxygen equilibrium curves of the hybrids twice in succession in the presence of met-heme reduction systems, there was no difference between the two measurements in the range of standard error.

TABLE I Oxygen equilibrium parameters of Cr(III)-Fe(II) hybrid Hbs
Experimental conditions are as follows: temperature, 25°C; buffer conditions, 50 mM bis-Tris or Tris with 100 mM chloride; protein concentration, the data sets are based on 60 M, and numbers in parentheses are based on 240 M, and numbers in square brackets are based on 12 M (on a metal basis); wave length of detection light, 560 nm in the case of 240 M and 60 M (on a metal basis), or 430 nm in the case of 12 M (on a metal basis).  Fig. 5.
of Hb samples and the pH conditions used in this study are listed (see Table I). Measurements were carried out at 25°C in 50 mM bis-Tris or Tris buffer with 100 mM chloride. The optical absorbance change during the successive deoxygenation and reoxygenation was monitored by a Shimadzu UV-2200 spectrophotometer at 560 nm in the case of 240 and 60 M (on a metal basis) or at 430 nm in the case of 12 M (on a metal basis). In order to maintain the met-Hb formation at minimal levels, catalase and superoxide dismutase (35,36), or an enzymatic reduction system that was composed of an NADPH generating system and a ferredoxin electron mediating system (37), were added into the Hb samples. The met-heme contents after the measurement were estimated from the absorbance change at 630 nm by the addition of a small amount of a potassium cyanide solution into the measured samples (30). The deoxygenation data were used for two-step Adair analysis. The reaction between Cr(III)-Fe(II) hybrid Hbs and oxygen molecules is expressed by the following two-step equilibria: The first and second intrinsic Adair constants, K i (i ϭ 1 and 2; in torr Ϫ1 ), is written as: p is the partial pressure of oxygen. i/(2 Ϫ (i Ϫ 1)) is a statistical factor, because the binding of O 2 to Hb(O 2 ) i-1 is statistically enhanced by a factor of the number of empty sites, 2 Ϫ (i Ϫ 1), and the release of O 2 from Hb(O 2 ) i is statistically enhanced by a factor depending on the number of filled sites, i. Fractional saturation with oxygen is expressed as Y: is the Adair equation used to analyze the data. The best fit values of the first and second intrinsic Adair constants, K i , were obtained by fitting a two-step Adair equation to each deoxygenation curve through a least-squares procedure (38).
Absorption Spectra and Extinction Coefficients-Absorption spectra ware recorded on a Shimadzu UV-2200A spectrophotometer (Shimadzu, Kyoto). Millimolar extinction coefficients were calculated on a basis of the protein concentration determined by the Lowry method (39).
Measurement of Autooxidation Rates-Autooxidation rates were measured using the techniques described by Brantley et al. (40). Oxidation of the sample was followed by recording the increase in absorbance at 410 nm. Measurements were carried out at 37°C in air equil- ␣ 2 (Mn 3ϩ )␤ 2 (Fe-CO) (42) and ␣ chain are shown as controls.
Each hybrid Hb appears as a nearly single band. CrHb and the hybrid Hbs migrate toward the higher pH region compared with Hb A, due to the presence of trivalent Cr(III)PPIX.
The pH dependence of the absorption spectrum of CrHb is presented in Fig. 2A. Upon raising pH from 6.5 to 8.4, the Soret peak shifted from 445 to 439 nm, and a visible peak at 764 nm shifted to 752 nm. In the range of pH 6.5-8.4, the isosbestic points were reproducibly observed at 748, 720, 441, and 407 nm.
The absorption spectrum of CrHb was not affected by the presence of either 50 mM dithionite under anaerobic condition or the enzymatic reduction system of Hayashi et al. (37), indicating that Cr(III)PPIX in CrHb was not reduced to Cr(II)PPIX by these reductants. Thus, the oxygen equilibrium curves of Cr(III)-Fe(II) hybrid Hbs could be measured with enzymatic reduction system in order to reduce the met-heme contents to a minimal level. Moreover, absorption spectra of deoxygenated Cr(III)-Fe(II) hybrid Hbs under anaerobic condition in the presence of 50 mM dithionite did not change remarkably for 96 h at 20°C except decreasing of dithionite absorption (Fig. 3). After these measurements, CO could reasonably bind to the ferrous subunits of Cr(III)-Fe(II) hybrid Hbs (Fig. 3). Autooxidation rate of ferrous subunits of ␣ 2 (Cr)␤ 2 (Fe-O 2 ) and ␣ 2 (Fe-O 2 )␤ 2 (Cr) were measured by detecting 410-nm absorbance change in air equilibrated buffer, 37°C condition (Fig. 4). Precipitate did not appear. The autooxidation rates of Cr(III)-Fe(II) hybrid Hbs were comparable with that of Hb A under the same condition (Fig. 4), meaning that ferrous subunits of Cr(III)-Fe(II) hybrid Hbs were as stable as those of Hb A against autooxidation. Thus, we conclude that Cr(III)-Fe(II) hybrid Hbs have enough stability for our oxygen equilibrium and crystallization experiments. The sum of the absorption spectra of ␣ 2 (Cr)␤ 2 (Fe-O 2 ) and ␣ 2 (Fe-O 2 )␤ 2 (Cr) was almost identical to that of oxy-Hb A and CrHb. (Fig. 2, B and C). Oxy-deoxy difference spectrum of each Cr(III)-Fe(II) hybrid Hb agrees closely with that of isolated ␣ or ␤ chain. These findings mean that Cr(III) subunits do not bind oxygen molecules and that the absorption spectra of Cr(III) subunits are not affected by the ligation state of the corresponding ferrous subunits within Cr(III)-Fe(II) hybrid Hbs.
Oxygen Equilibrium Parameters-K 1 and K 2 values (the equilibrium constants for the first and second oxygen molecule to bind to hybrid Hbs, respectively), P 50 values (the oxygen pressure at half-saturation), n max values (the maximal slopes of the Hill plots), and met-heme contents after measurements (the percent of the met-hemes in the total hemes) are listed in Table I. The met-heme contents are markedly reduced by adding the enzymatic reduction system, but the oxygen equilibrium parameters are little affected by the same. We also found that the oxygen affinities of the hybrids are not significantly dependent on the protein concentrations in the range from 60 to 240 M (on a metal basis) at pH 7.4, and in the range from 12 to 60 M (on a metal basis) at pH 6.5.
In the absence of IHP, Cr(III)-Fe(II) hybrid Hbs showed very high affinity for oxygen molecules. At pH 8.4, both hybrids bound oxygen noncooperatively (n max ϭ 1.0 -1.1) with very high affinity comparable with that of isolated ␣ or ␤ chain, while they exhibited n max values significantly higher than unity (n max ϭ 1.2-1.3) at pH 6.5. In both hybrids, Hill coefficients became larger as pH decreased. There were slight differences between the cooperativity of ␣ 2 (Cr)␤ 2 (Fe) and that of ␣ 2 (Fe)␤ 2 (Cr). The n max values of ␣ 2 (Cr)␤ 2 (Fe) (n max ϭ 1.1-1.2) were slightly smaller than those of ␣ 2 (Fe)␤ 2 (Cr) (n max ϭ 1.1-1.3) at all pH values examined, and the latter hybrid exhibited a slightly larger Bohr effect than the former. The oxygen equi- librium properties of both hybrids were significantly affected by the addition of IHP. The oxygen affinities were reduced, and the cooperativity and the number of released Bohr protons was increased (Fig. 5). It is important to note that the extent of the IHP effect on the K 2 values of both hybrids was very similar to those on the K 4 value of native Hb A (43) 3 (see Fig. 5). These results suggest that Cr(III)PPIX behaves like a permanent oxy-heme.

DISCUSSION
The pH-dependent spectra of CrHb show well defined isosbestic points ( Fig. 2A), indicating that CrHb exists in a pH-dependent equilibrium between two alternate states. Since Cr(III) complexes are almost universally hexacoodinated (44), and Cr(III)PPIX prefers counter anion due to an extra positive charge, it is reasonable to consider that the proximal histidine coordinates to the Cr(III) ion and that the remaining coordina-tion position is occupied by a water molecule or a hydroxyl ion, as in the case of aquomet-Hb. Thus, observed pH-dependent spectral changes may result from a coordination equilibrium between H 2 O and OH Ϫ in CrHb. Previously, Fiechtner reported that hemichrome structure is formed in CrHb (45); however, their published absorption spectrum is quite different from ours. Their OD Soret /OD 280 ratio is 1.3, whereas ours is 3.6 at pH 7.4.
The dimer-tetramer association equilibrium constants of ␣ 2 (Cr)␤ 2 (Fe-CO) and ␣ 2 (Fe-CO)␤ 2 (Cr), which were obtained by gel-filtration experiments, are about 5 ϫ 10 5 M Ϫ1 and 1 ϫ 10 6 M Ϫ1 , respectively. 4 Although part of the Cr(III)-Fe(II) hybrids dissociated into a␤ dimers under our experimental conditions used for oxygen equilibrium experiments, the Hill plots of both Cr(III)-Fe(II) hybrids exhibited little protein concentration dependence (Table I). We calculated the protein concentration dependence of the Hill plots of Cr(III)-Fe(II) hybrids using estimates of dimerization from dimer-tetramer association equilibrium constants and the assumption that a hemoglobin dimer containing Cr(III)PPIX and Fe(II)PPIX exhibits high oxygen affinity, comparable with that of isolated chains. The calculations revealed that the Hill plots were only slightly influenced by dimerization in the concentration range of above 10 M (on a metal basis). Because the experimental Hb concentrations which we measured oxygen equilibrium curves were 12, 60, and 240 M (on a metal basis), the theoretical consideration was consistent with the experimental results.
K 2 values of both Cr(III)-Fe(II) hybrid Hbs agreed with K 4 values of Hb A, including both pH and IHP dependence (Fig. 5). In recent years, both association and dissociation rate constants for the last step (␣and ␤-subunit, respectively) in oxygen binding to Hb A were determined kinetically, and the Adair constants (K 4 for ␣ subunit and K 4 for ␤ subunit, respectively) were calculated from these rate constants (46 -49). In Table II,    position of the proximal histidine relative to the hemeplane is a key determinant of oxygen affinity of Hb. In this regard, Cr(III)PPIX can be an adequate model for an oxyheme for several reasons. (i) The Cr(III) ion has an ionic radius of 0.62 Å (51), which is almost equal to that of low spin Fe(II). (ii) The Cr(III) ion is also expected to lie in the mean porphyrin plane (52). (iii) Cr(III) complexes are almost universally hexacoordinated (44), and (iv) Cr(III) porphyrin binds ligands so tightly (52) that the Cr(III)-histidine bond is expected to be as short as Fe(II)-histidine bond in oxy-Hb. Thus, both the proximal and the distal environments of Cr(III)PPIX in Hb may be similar to those of oxy-heme. Cyanomet valency hybrid Hbs have been widely used as models for understanding the nature of the intermediate species formed during the cooperative oxygenation process. There are several structural and functional reasons for using cyanidebound ferric heme as an oxyheme model. (i) The crystal structure of cyanomet-Hb (10) is closely similar to that of oxy-Hb A (53). (ii) Like an oxyheme, cyanide-bound ferric heme is low spin, with the ferric ion firmly anchored in the heme plane (4,12,54). (iii) Cyanomet subunits in cyanomet valency hybrids give NMR spectra that are very similar to those observed in cyanomet Hb in the absence of 2,3-diphosphoglycerate or IHP (5,7). (iv) The ferrous subunits of cyanomet valency hybrids show fast ligand binding kinetics (3,6) and high affinity for oxygen (8,11). On the basis of these findings, cyanomet valency hybrid Hbs are generally assumed to be a good model for the oxygenated intermediates (13)(14)(15)(16)(17)(18)(19)(20). Because the K 2 value of both ␣ 2 (Fe 3ϩ CN Ϫ )␤ 2 (Fe 2ϩ ) and ␣ 2 (Fe 2ϩ )␤ 2 (Fe 3ϩ CN Ϫ ) agreed well with the K 4 value of Hb A including pH effect (11), 5 cyanide-bound ferric heme is a good model for an oxygenated heme as Cr(III)PPIX. Actually, the oxygen equilibrium properties of both Cr(III)-Fe(II) hybrid Hbs almost agreed with those of the corresponding cyanomet valency hybrid Hbs. Yet, it should be pointed out that cyanomet valency hybrid Hbs have at least one serious disadvantage. The ferrous subunits of cyanomet valency hybrid Hbs are unstable against autooxidation. The inevitable autoxidation of ferrous subunits in these cyanomet hybrids results in asymmetric oxygen equilibrium curves (11). Since the addition of the met-Hb reducing reagent or enzymatic reduction system reduces not only the oxidized heme but also the cyanomet heme, these standard techniques for native Hb are not applicable to the cyanomet hybrids.
In contrast to cyanomet heme, Cr(III)PPIX in Hb is not reduced to Cr(II)PPIX with 50 mM sodium dithionite or with the enzymatic reduction system. This advantage allows us to measure more accurately and reasonably oxygen equilibrium curves of Cr(III)-Fe(II) hybrid Hbs in the presence of the enzymatic reduction system. Moreover, complete deoxygenation of Cr(III)-Fe(II) hybrids can be easily attained by the addition of sodium dithionite, so that we can safely carry out the experiments under anaerobic condition. For example, we actually succeeded in making the single crystals of deoxy-␣ 2 (Cr)␤ 2 (Fe) grow in a low salt solution with the presence of both polyethylene glycol and 50 mM dithionite and under anaerobic conditions (Fig. 6).