The contribution of the asymmetric α1β1 half-oxygenated intermediate to human hemoglobin cooperativity

Considerable controversy remains as to the functional and structural properties of the asymmetric α1β1 half-oxygenated intermediate of human hemoglobin, consisting of a deoxygenated and an oxygenated dimer. A recent dimer-tetramer equilibrium study using [Zn(II)/Fe(II)-O2] hybrid hemoglobins, in which Zn-protoporphyrin IX mimics a deoxyheme, showed that the key intermediate, [α(Fe-O2)β(Fe-O2)][α(Zn)β(Zn)], exhibited an enhanced tetramer stability relative to the other doubly oxygenated species. This is one of the strongest findings in support of distinctly favorable intra-dimer cooperativity within the tetramer. However, we present here a different conclusion drawn from direct O2 binding experiments for the same asymmetric hybrid, [α(Fe)β(Fe)][α(Zn)β(Zn)], and those for [α(Fe)β(Zn)]2 and [α(Zn)β(Fe)]2. In this study, the O2 equilibrium curves for [α(Fe)β(Fe)][α(Zn)β(Zn)] were determined by an O2-jump stopped-flow technique to circumvent the problem of dimer rearrangement, and those for [α(Fe)β(Zn)] 2 and [α(Zn)β(Fe)] 2 were measured by using an Imai apparatus. It was shown that the first and second O2 equilibrium constants for [α(Fe)β(Fe)][α(Zn)β(Zn)] are 0.0209 mmHg−1and 0.0276 mmHg−1, respectively, that are almost identical to those for [α(Fe)β(Zn)]2 or [α(Zn)β(Fe)]2. Therefore, we did not observe large difference among the asymmetric and symmetric hybrids. The discrepancy between the present and previous studies is mainly due to previously observed negative cooperativity for [α(Fe)β(Zn)]2 and [α(Zn)β(Fe)]2, which is not the case in our direct O2 binding study.

Although cooperative oxygenation of human adult hemoglobin (HbA) 1 has been studied extensively as a paradigm for regulatory actions of allosteric proteins, the functional and structural properties of its eight partially oxygenated intermediates remain elusive. This is mostly due to the strong cooperativity of Hb, which suppresses relative abundance of the intermediates, precluding direct study. One of the most specific methods for studying the oxygenation intermediates has been to substitute the heme in one or more of four subunits with another metalloprotoporphyrin IX, which does not bind O 2 while mimicking either normal deoxyheme or oxyheme (1)(2)(3)(4)(5)(6). However, there has been a limitation to this approach: Six asymmetric forms cannot be studied in isolation, because the asymmetric hybrid tetramers dissociate into dimers, which then reassociate to form not only the former asymmetric tetramers but also symmetric ones (i.e. dimer rearrangement).
Ackers and his colleagues (7) have partly circumvented this difficulty by measuring the dimer-tetramer equilibrium constants for asymmetric hybrid species in the presence of the symmetric parental species. In 1985, this methodology was first applied to resolution of the tetramer stabilities of all ten ligation intermediates of the deoxy-cyanomet [Fe(II)/ Fe(III)-CN Ϫ ] hybrid system, in which cyanometheme mimics a fixed oxyheme (8). The most striking finding was that the dimer-tetramer association equilibrium constant for (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤) was about 170 times that of the other three doubly liganded species at pH 7.4, which implies a hypercooperativity in specific ligation steps in the ␣1␤1 dimer within the tetramer (i.e. a 170-fold affinity change). Subsequently, the effects of pH, temperature, and single-site mutations on the dimer-tetramer equilibrium of (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤) were studied (7,9,10). It was suggested that this key intermediate assumes a form of the deoxy-Tquaternary structure as judged from the nature of the dimerdimer interface.
Based on the hyperthermodynamic stability and T quaternary structure of (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤), Ackers' group proposed a new framework for Hb cooperativity, called a symmetry rule (SR), in 1991 (9). The key features of the SR model are: (i) The two ligation steps leading to the asymmetric ␣1␤1 half-liganded intermediate show distinctly favorable cooperativity while those leading to the other three doubly liganded intermediates exhibit no favorable cooperativity; (ii) The asymmetric ␣1␤1 half-liganded intermediate assumes a deoxy-T-quaternary structure while the other three doubly liganded intermediates exhibit an oxy-R-quaternary structure.
Recently, we showed that both the published hyperstability and T structure assignment for (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤) were artifacts arising from valency exchange between the deoxy and cyanometheme sites during the long deoxy incubation that had been used routinely in the laboratory of Ackers (11,12). Ackers et al. (13) have now accepted the valency exchange artifacts in their previous studies using deoxy-cyanomet hybrids. Nevertheless, they still advocate the SR model based on their dimertetramer study using [Zn(II)/Fe(II)-O 2 ] hybrid system in which Zn-protoporphyrin IX mimics a fixed deoxyheme (13,14).  2 were carried out at 470 nm where the spectral changes of the Zn-containing ␤ subunits were negligible (6). The O 2 equilibrium curves of both symmetric Zn(II)/ Fe(II) hybrid Hbs were analyzed by a two-step Adair equation as described previously (4). All the measurements were carried out at 25°C in 50 mM Tris plus 50 mM bis-Tris buffer with 100 mM Cl Ϫ , pH 7.4. Catalase and superoxide dismutase were added to Hb samples to minimize metHb levels (21,22).

Preparation of [␣(Fe)␤(Fe)][␣(Zn)␤(Zn)
]-A 15-fold molar amount of ZnHbA was mixed with oxyFeHbA to give the Zn and Fe concentrations of 13 and 0.87 mM, respectively. The buffer used was 50 mM Tris plus 50 mM bis-Tris with 100 mM Cl Ϫ at pH 7.4 (at 25°C). The resulting hybrid mixture was deoxygenated using pure N 2 gas, followed by deoxy incubation for ϳ140 h at 25°C to reach equilibrium. According to Ackers et al. (13), equilibration time for this hybrid system was ϳ120 h at pH 7. O 2 -jump Experiments-Mixing experiments were carried out at 25°C using Orii's (23) stopped-flow apparatus. Transient absorption spectra after mixing were recorded on a UNISOKU rapid-scan spectrophotometer (Model RSP-601). At least three shots were performed and averaged. The volume ratio used in the mixer was 1:1, and the mixing dead time was 18 ms. The buffer used was 50 mM Tris plus 50 mM bis-Tris with 100 mM Cl Ϫ , pH 7.4. Before the measurements, the buffer solutions (without Hb) in both reservoirs were bubbled with pure N 2 gas for ϳ30 min at a flow rate of 30 ml/min to remove dissolved O 2 . The bubbling was conducted by tubing inserted through the stopper, which also has a gas outlet, allowing the escape of gas and at the same time preventing any back diffusion of O 2 . 250 l of the equilibrated deoxy [␣(Fe)␤(Fe)][␣(Zn)␤(Zn)] hybrid mixture (as described above) was transferred to one of the reservoirs by a gas-tight syringe, and the bubbling with pure N 2 gas was continued for a few seconds to make a uniform Hb solution in the reservoir. Then, the tip of the tube was lifted above the solution surface, and the gas flushing was continued throughout the experiment. The kinetics of O 2 binding to the asymmetric hybrid were investigated in a set of mixing experiments as follows: (i) For determination of the absorbance of the deoxy hybrid mixture at 470 nm, Abs Hb , a deoxy hybrid solution was mixed with deoxy buffer.

Y͑t͒ ϭ
Abs͑t͒ Ϫ Abs Hb Abs HbO2 Ϫ Abs Hb (Eq. 1) Also, the concentration of free O 2 in the mixed solution at a time t after mixing, x(t), is calculated by the following equation, where x 0 denotes the O 2 concentration of the reservoir buffer to mix with Hb, and [heme] is the total concentration of the hemes after mixing.
, the Y(t) values of the hybrid mixture after T-R conformational transition and prior to dimer rearrangement are essential. Such Y(t) values could be obtained directly, if the time frame of dimer rearrangement was well separated from that of the T-R conformational transition (see "Results and Discussion"). Otherwise, computer simulations were carried out using a numerical integration program combined with a nonlinear least-squares fitting routine. The dimer rearrangement kinetics were characterized by the following set of differential equations, where  (24)); the k ZZ value was fixed at 5.75 ϫ 10 Ϫ5 s Ϫ1 (which was determined using a haptoglobin kinetics method under the conditions of the present study); the k FF was fixed at 1.1 s Ϫ1 (which was assumed to be identical to the value of 1.1 s Ϫ1 at 21.5°C because of very small enthalpy for this reaction (24)). At the end of each numerical integration cycle, the O 2 redistribution in all possible ligation forms of the hybrid and HbA was taken into account. For this calculation the K 1 value of the hybrid was fixed at 0.0209 mm Hg Ϫ1 (which could be determined by the O 2 -jump experiments at low concentrations of O 2 ), whereas the K 2 value was treated as variable, and the four Adair constants of HbA determined by Imai apparatus under the same experimental conditions were used.  (open squares, Fig. 3B). The asymmetric hybrid exhibits a low affinity for O 2 and slight cooperativity (n 50 ϭ 1.1). The O 2 association equilibrium constants K 1 and K 2 for [␣(Fe)␤(Fe)][␣(Zn)␤(Zn)] were determined to be 0.0209 and 0.0276 mm Hg Ϫ1 , respectively, by fitting all of the data points in Fig. 3B (solid line is the best-fit curve). Fig. 3B also presents the previously reported O 2 equilibrium curve for this hybrid at pH 7.4 and 21.5°C, which was calcu-  Fig. 3B) (13, 14). The previously observed K 2 value is approximately twice as large as the present observation (see Table I). This discrepancy is probably due to insufficient O 2 saturation of the hybrid in the previous dimer-tetramer experiments (13). According to Ackers et al. (13), ZnHbA and FeHbC were hybridized under 1 atm of O 2 , and then the fractional population of the asymmetric AC hybrid formed was deter-mined. However, because the asymmetric hybrid cannot be fully saturated with O 2 under 1 atm of O 2 (see Fig. 3B), the hybridized sample would be contaminated by deoxy-and monoliganded hybrids, leading to overestimation of the fractional population Fig. 4 shows the Hill plots for our O 2 equilibrium curves of [␣(Fe)␤(Zn)] 2 and [␣(Zn)␤(Fe)] 2 determined by Imai apparatus (15) (circles and squares, Fig. 4), along with the previous data by the dimer-tetramer experiments (dotted and dashed lines, Fig. 4) (13, 14). Striking discrepancy is seen in the present and previous results. Our results indicate that both hybrids exhibit slight cooperativity (n 50 ϭ 1.1-1.2) whereas significant negative cooperativity was found in the previous study (n 50 Ͻ 1).

of [␣(Fe-O 2 )␤(Fe-O 2 )][␣(Zn)␤(Zn)]. O 2 Equilibrium Properties of [␣(Fe)␤(Zn)] 2 and [␣(Zn)␤ (Fe)] 2 -
To eliminate the possibility of negative cooperativity, we carried out an O 2 -jump experiment in which fully deoxygenated [␣(Fe)␤(Zn)] 2 was mixed with 131 M O 2 . If the hybrid displays positive cooperativity, O 2 binding to the original low affinity conformation (which completes within the mixing dead time) should be followed by transition from the low affinity to high affinity conformation, which increases the fractional saturation of O 2 . If negative cooperativity were the case, on the contrary, dissociation of once bound O 2 from doubly liganded hybrid should take place. As expected, a significant amount of O 2 binding was observed within 500 ms (inset, Fig. 4 Table I). Therefore, we did not observe a large difference among the asymmetric and symmetric hybrids, in contrast to the previous conclusion drawn from the dimertetramer experiments (13,14). The discrepancy between the present and previous studies is mainly due to the previously observed negative cooperativity for [␣(Fe)␤(Zn)] 2 and [␣(Zn)␤(Fe)] 2 , which is not the case in our direct O 2 binding study (Table I and Fig. 4; see also inset, Fig. 4). We do not know the exact reason why Ackers' group (13,14) observed the negative cooperativity. However, it should be pointed out here that the determination of the dimer-tetramer equilibrium constants for [␣(Fe-O 2 )␤(Zn)] 2 and [␣(Zn)␤(Fe-O 2 )] 2 by gel chromatography (14) seems to be practically impossible, because these low affinity hybrids cannot be saturated with O 2 even under 1 atm of O 2 (see Fig. 4).
Perrella et al. (26 -28) have used a cryogenic technique for trapping the reaction intermediates of Hb and CO, which in many aspects is a close approximation to O 2 . In this method the intermediate compounds between Hb and CO at various CO saturation levels are trapped by rapidly quenching an aqueous solution of Hb into a cryosolvent containing ferricyanide. The valency hybrids formed by oxidization of the unliganded hemes of the intermediates are then separated and quantified by isoelectric focusing at low temperature. According to their analysis of the distribution of all the intermediates at pH 7.0 and 20°C (27), no preferable cooperative interaction was found in the ␣1␤1 ligation. Rather, the strongest cooperativity was found in the ␣1␣2 ligation pathway.
Shibayama et al. (29) have determined the first two-step microscopic O 2 equilibrium constants of Hb using cross-linked Ni(II)/Fe(II) hybrid Hbs, carrying Ni(II)-protoporphyrin IX, which binds neither O 2 nor CO, in two subunits and normal hemes in the other two subunits. The cross-linked [Ni(II)/ , because Ni(II)-protoporphyrin IX mimics a fixed deoxyheme with respect to its effect on the oxygenation properties of the counterpart Fe(II) subunits within the same tetramer (3,30,31) and because the cross-link used has little effect on the oxygenation properties of HbA (18). At pH 7.4 and 25°C, the cooperativity represented by the Hill coefficient increased in the order of ␤1␤2 (n 50 ϭ 1.36), ␣1␤1 (n 50 ϭ 1.41), ␣1␤2 (n 50 ϭ 1.64), and ␣1␣2 (n 50 ϭ 1.72), indicating no favorable cooperative interaction in the ␣1␤1 ligation pathway. Interestingly, the observation that the ␣1␣2 ligation pathway exhibits the greatest cooperativity is consistent with the results of the CO ligation intermediates (27) and with the present data of Zn(II)/ Fe(II) hybrids (see Table I).
Is there any positive evidence for distinctly favorable intradimer cooperativity? The original evidence for this preferred ligation pathway was the hyper (170-fold) stability of (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤) relative to the other three doubly liganded species (7)(8)(9). However, this evidence has been proven to be an artifact arising from valency exchange between the deoxy and cyanometheme sites during the long deoxy incubation that had been used routinely in the laboratory of Ackers (11,12). Recently, Ackers et al. (13) attempted to minimize the valency exchange artifact by using a relatively short deoxy incubation of cyanometHbA and deoxyHbS. Based on the fractional population of the asymmetric AS hybrid formed at 2.5 h (i.e. 14% of the total), they claimed that the intra-dimer cooperativity in cyanomet ligation was still significant (i.e. a 30-fold affinity change). However, because oxidization of unliganded heme was not rigorously controlled in that study (13), the integrity of the asymmetric hybrid was uncertain. In fact, the fractional populations of the AS hybrid formed and cyanometHbA parent at 2.5 h were 14 and 38%, respectively, whereas the relative amounts of cyanomet-hemes in them (estimated by the distribution of radioactive 14 CN) were 22 and 75%, respectively (see Table I of Ref. 13). If the AS hybrid formed were pure (␣ ϩCNϪ ␤ ϩCNϪ ) A (␣␤) S , the ratio of the amounts of cyanomet-hemes in the AS hybrid and in cyanometHbA parent would be 14/2:38 (ϭ 0.18:1). However, the observed ratio was 22:75 (ϭ 0.29:1), suggesting the presence of excess cyanomet-hemes in the asymmetric AS hybrid probably due to oxidization of unliganded heme. It is therefore likely that the asymmetric hybrid formed at 2.5 h was contaminated by (␣ ϩCNϪ ␤ ϩCNϪ ) A (␣ ϩCNϪ ␤) S , (␣ ϩCNϪ ␤ ϩCNϪ ) A (␣␤ ϩCNϪ ) S , and (␣ ϩCNϪ ␤ ϩCNϪ ) A (␣ ϩCNϪ ␤ ϩCNϪ ) S , leading to overestimation of the tetramer stability of (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤). This speculation is reinforced by the fact that too much asymmetric hybrid was formed at 25 min (13% of the total; see Table I of Ref. 13). Note that, because the rate-determining step of the formation of (␣ ϩCNϪ ␤ ϩCNϪ )(␣␤) is slow, i.e. dimerization of deoxyHb with a rate constant of 2 ϫ 10 Ϫ5 s Ϫ1 (24), the fractional population of the hybrid at 25 min should not exceed 3% of the total, if pure.
A second line of evidence that supports distinctly favorable intra-dimer cooperativity has come from the dimer-tetramer studies on the Fe(II)/Mn(III), Co(II)/Fe(II)CO, Co(II)/ Fe(III)CN Ϫ , and Fe(II)/Fe(II)CO hybrid systems (32)(33)(34). It was concluded that in all these systems the ␣1␤1 halfliganded intermediate exhibits an enhanced tetramer stability relative to the other doubly liganded intermediates, although the magnitudes of such enhancement are systemdependent. However, this conclusion appears to be invalid, because there were serious flows and uncertainties in the previous work. The following points should be noted: (i) Valency exchange occurs between deoxyheme and Mn(III) protoporphyrin IX sites during the hybridization of Fe(II)Hb and Mn(III)Hb (see footnote of Ref. 35). The valency exchange results in the formation of Fe(III) and Mn(II) in the original deoxyFe(II)Hb and Mn(III)Hb, respectively, and thus plays a role in randomizing the location of the divalent and trivalent metals in the samples. Thus, one may encounter the same difficulty as previously reported for deoxy-cyanomet hybrids (12).   Ref. 33; see also Ref. 34). In both cases, the equilibrium populations of the asymmetric hybrids were evaluated using the plateau populations at the end of the slow phases (at 80 min for [␣(Fe-CO)␤(Fe-CO)][␣(Co)␤(Co)] and at 3 weeks for [␣(Fe ϩCNϪ )␤(Fe ϩCNϪ )][␣(Co)␤(Co)]). However, the simulations using the published dimer-tetramer rate constants for these hybrids showed that both hybridizations should rapidly reach equilibrium within a few minutes (data not shown). Thus, it is likely that the published tetramer stabilities of these hybrids were considerably overestimated, although the origins of the previously observed slow phases are not clear. Conclusion-This work presents direct measurements of O 2 binding to asymmetric hybrid Hb that has significant implication for the nature of Hb cooperativity. The data presented here show that, in the case of [Zn(II)/Fe(II)-O 2 ] hybrid system, the first two-step O 2 binding to Hb molecule is roughly pathwayindependent and shows only slight cooperativity regardless of ligation pathway (see Table I). These findings refute directly the claims by Ackers' group about the same hybrids (13,14) and resolve the long standing controversy of whether or not the asymmetric ␣1␤1 half-oxygenated intermediate makes a unique contribution to human Hb cooperativity. Furthermore, our results, eliminating the negative cooperativity previously observed (13,14), are more consistent with the two-state allosteric model of Monod et al. (36). The present observations are also qualitatively consistent with previous data on the [Fe(II)/ Fe(II)-CO] system by Perrella and Di Cera (27) and those for the cross-linked [Ni(II)/Fe(II)-O 2 ] system by Shibayama et al. (29), in that the greatest cooperative interaction is found in the ␣1␣2 ligation pathway, although the magnitudes of cooperativity are system-dependent. One plausible interpretation for the observed small cooperativity in the [Zn(II)/Fe(II)-O 2 ] system is that the substitution of Zn(II) protoporphyrin IX for deoxyheme leads to further stabilization of the lowest affinity state of Hb (6).