Kinetics and Thermodynamics of Oxygen, CO, and Azide Binding by the Subcomponents of Soybean Leghemoglobin*

Leghemoglobin shows extreme high affinity behavior in the binding of both oxygen and CO. We have determined the temperature dependence of the rate constants for ligation of oxygen and CO and from these data the thermodynamics (AC’, Ap, AS”) of ligation for the purified components of soybean leghemoglobin. X-ray crystallography has shown that the heme cavity can easily accommodate ligands the size of nicotinate, and analysis of extended x-ray absorption fine structure data has that the Fe is the mean of the heme in the leghemoglobin-CO are in accord


Leghemoglobin
shows extreme high affinity behavior in the binding of both oxygen and CO. We have determined the temperature dependence of the rate constants for ligation of oxygen and CO and from these data the thermodynamics (AC', Ap, AS") of ligation for the purified components of soybean leghemoglobin. X-ray crystallography has shown that the heme cavity can easily accommodate ligands the size of nicotinate, and analysis of extended x-ray absorption fine structure data has shown that the Fe atom is in the mean plane of the heme in the leghemoglobin-CO complex. Ligation of oxygen and CO are in accord with this picture in that the Ea for oxygen binding is that expected for a diffusion controlled reaction and A.!?' for the ligation of both CO and oxygen is consistent with the simple immobilization of the ligand at the Fe, with no evidence for significant conformational changes in the protein or changes in solvation.
At 20 'C the rate constants for oxygen and CO binding vary by 26-44% among the eight leghemoglobin components. For azide binding the variation is a factor of 2. These variations appear to arise from amino acid substitutions outside either the heme cavity or the two major paths for ligand entry to the heme. The distribution of leghemoglobin components varies with the age of the soybean nodule during the growing season. The changes in composition alone, however, would only allow the concentration of free oxygen to vary by about 3%. This finding calls into question models that ascribe a significant functional role to changes in the distribution of leghemoglobin components in regulating oxygen concentration in the nodule.
Leghemoglobin (Lb)l is a monomeric hemeprotein found in the root nodules of legumes. The functions of Lb are at least 2-folk Lb facilitates the diffusion of oxygen to the respiring bacteroids and serves as an oxygen reservoir (1). Its high affinity for oxygen also results in a very low concentration of free oxygen in the bacteroid, protecting the nitrogenase from inactivation. Lb that is isolated from soybean nodules is a family of eight proteins.
The major Lb components, Lba, Lbcl, LbcZ, and Lbcs, are probably post-translationally modified to yield the respective minor components: Lbb, Lbdl, Lbd*, and Lbda (2,3). The concentration ratios of Lbcl + LbcZ and LbcZ + Lbcs to Lba vary with nodule age (4,5) and functional differences have been proposed for the different Lb components. These changes with age apparently arise from differences in the biosynthetic rates for the components (6). Ligand binding kinetic and equilibrium experiments were carried out to investigate functional differences between the Lb components and to establish precise values for kinetic and thermodynamic parameters for this model high affinity hemoglobin. Small variations between soybean Lba and unseparated Lbc were observed for oxygen equilibria (7) and for oxygen and CO binding (8,9). However, the separated Lbc and Lbd components and Lbb were not studied with respect to oxygen and CO kinetics. We have separated soybean Lb into the eight individual components. We report the oxygen and CO association and dissociation rate constants and equilibrium constants, the activation energies, and the standard state changes in free energy, enthalpy and entropy of ligation for the separated components.
The rate of azide binding has been shown to be sensitive to differences between the c~ and p hemes in many hemoglobins (lo), and both the kinetics and equilibria for the binding of this ligand were measured for the eight components of Lb, MATERIALS AND METHODS

Preparation of Soybean Lb
The planting and growth of soybean plants and harvesting of soybean nodules were performed as described previously (11). Lb was prepared using the method of Appleby et ul. (12) with modifications to the procedure described elsewhere (11)  The separate Lbs were obtained from a single large preparation of Lba and thus are averaged over at least 20,000 nodules. The errors in the rate constants for the gaseous ligands were not obtained from the variance matrix (which, owing to correlations through the time constant, underestimates errors) but from errors (standard deviations) from a regression fit to a van't Hoff plot using all the data for a given component.
For several of the minor Lbs there was only sufficient material for a single set of experiments, which involved at least five determinations at each temperature. The errors given in the table do not include errors from uncertainties in gas concentrations. Since the oxygen solutions for measurement of k' were 90% air-equilibrated, the error in oxygen must be small. Extensive tests with CO have shown that solutions can be reproduced to within 5%. Since the van't Hoff analysis errors and the gas concentration errors are independent, one can estimate that the errors in /z', l', K, and L, respectively, should be no more than 5, 5, 6, and 6%. Errors in k and 1 do not depend on errors in gas composition, nor do the errors in activation energies and enthalpy changes. Errors in Gibbs free energies should be increased at most to 0.04 kcal/mol (Table III), but errors in the entropy changes would remain as given, since the major contribution is from the error in the enthalpy change. The errors for the azide association rate constants (Table IV) were from the variance matrix and are probably too small by about a factor of 2 owing to correlations of the residuals. The same azide solution was used for all Lbs, thus the differences do not arise from errors in azide concentration.
The fractions of the various components are known to vary with the age of the nodules (5), and that information can be reduced to three variables, A, B, and C. Thus, the concentration ratios are as follows: Lbb/Lba = 0.1; Lbcl/Lbdl = LbcJ LbdZ = LbcZ/Lbds = 5; Lba/Lbcs = A; Lba/Lbcl = A/B; Lba/ LbcZ = A/C, where the variables A, B, and C can be calculated from data in the paper by Fuchsman and Appleby (5) (Tables I and III) for both Ki and AZ?',, Y can also be calculated as a function of temperature.
Both the age of the nodules and the soil temperature vary throughout the growing season. In Table V are  listed oxygen concentration values for three selected fractional saturations, for two representative soil temperatures, and as a function of time in terms of nodule age. It is apparent that the entire binding curve moves to the right, i.e. toward higher oxygen concentrations and lower fractional saturation as the growing season progresses and as the nodules age.
It has recently been shown (19) that there is an acid Bohr effect for LHb below pH 7 for oxygen, but not for CO binding. This effect derives entirely from changes in k, such that k decreases with decreasing pH. Using data from that reference, it can be calculated that the number of Bohr protons (rz~) is 0.016 at pH 7 (see Ref. 20, where X-is used for no). Let Kl and Kz be, respectively, oxygen dissociation equilibrium con-      Wyman (20), the true AP values for oxygen binding at pH 7 for phosphate buffer and the dilute protein solutions employed should be no more than 0.12 kcal/mol more negative than those values reported in Table I. Since this is well within the error limits, this minor, although systematic, correction was not made.

DISCUSSION
The rate constants reported here for the unfractionated Lb and for Lba are in good to excellent agreement with previously reportedvalues. Thus, for P, in units of FM-' s-l, the following values for Lba have been reported at 20 "C: 118 (reported here); 118, pH 6.8 (8); 116, pH 7.6 (19). A value of 150, pH 6.5, was reported (9) at 25 "C, which, with our activation energy, adjusts to 134 at 20 "C. For oxygen dissociation, we report here 5.72 se1 by relaxation and 5.1 se1 by direct stoppedflow determination using dithionite. Elsewhere, the following values have been reported, for Lba, in units of s-l: 5.55 (19); 4.6 (quoted in Ref. 19 from previous work (8) at pH 7), and 9.4 (9), which was recognized as being of low precision owing to the presence of Met-Lb which would affect the stoppedflow results (9). For CO association, in units of pM-' s-l, we report 11.6 for Lba. Others have reported: 12.7 (19), 12.7 (8), and 13.5 (9) at 25 'C, which adjusts to 11,9 with our activation energy. For the CO dissociation rate constant, 1, we report here 0.0068 se1 for Lba in good agreement with literature values: 0.0078 (19), and with a value of 0.0066 calculated from our activation energy and the value 0.012 reported at 25 "C (9). As reported under "Results," we find excellent agreement between 1 determined by flash relaxation and by NO replacement. Furthermore, /J determined by flash relaxation is in reasonable agreement with that determined directly using dithionite, although M and AZ?',+, are in better agreement with the values calculated using k from the flash relaxation than with !z determined directly in the stopped-flow apparatus. We conclude, because of the redundancies in the determinations of k and 1 and the agreements of M and AH"M with calculated values, as well as the excellent agreement of the values reported here for Lba with those in the literature, that the values for 1, l', k, and k' form a consistent set of rate constants.
In Fig. 1 are plotted on a log-log scale values of k versus K for the eight components. The slope is 1, taken from a similar plot for diverse non-cooperative hemoglobins and myoglobins (21). In obtaining Fig. 1, a one-parameter least-squares fit was obtained where only the y intercept varied. This plot confirms, on a smaller scale, the conclusions reported elsewhere (21), that changes in oxygen aftinity (K) derive almost exclusively from changes in k, the rate constant for oxygen dissociation. In a single-step model of ligation, the interpretation would be that the transition state closely resembles the reactants (22). The results of Rohlfs et ul. (23), however, show that at least four steps are required to describe oxygen binding in Lb, which would require a more detailed linear free energy analysis. Nevertheless, it appears that the small structurefunction variations for the Lb components are in accord with the much greater differences previously analyzed for diverse hemeproteins (21).
The time course of biosynthesis of the major soybean Lb components varies with age, suggesting a difference in function for the components (4, 5). The ratio of Lba to Lbcs increases greatly in young developing nodules, whereas LbcJ Lbcs increases slightly during initial nodule growth and then levels off. During the lifetime of the nodule, the ratio of LbcJ Lbc, increases very slowly. A functional difference was suggested for the different Lb components (4,5), and the oxygen binding of Lba and unseparated Lbc toward oxygen were shown to differ (7,8). From values of K and AIP for all Lb components we can calculate an oxygen binding curve for the unfractionated Lb as a function of changing composition. From the data in Table V, it can be seen that there is a progressive decrease in oxygen affinity for the total Lb of the nodule as a function of age. If one adopts a simple model for facilitated oxygen transport, such as the steady-state treatment of Wyman (24) the facilitation of oxygen transport by about 3% over the life of the nodule at constant temperature. This is a small effect and raises serious questions about the developmental importance of Lb heterogeneity in oxygen transport (25). These small changes in free oxygen resulting solely from changes in hemoglobin composition would also seem to have but a small effect on the functioning of the nitrogenase system. Soybean Lbb differs from Lba only in the final two Nterminal residues (2). It is possible that Lbb arises posttranslationally from Lba by cleavage of the N-terminal valine of Lba, followed by acetylation of alanine. For oxygen binding, Lba and Lbb appear to be identical, although they clearly differ for azide association, in that Lba binds azide about 30% more rapidly than does Lbb. The Lbd series appears to be derived from the Lbc series by post-translational modification, in which the N termini are acetylated (3). Here, except for the Lbcl/Lbdl pair, acetylation does not affect oxygen binding. Lbcl and LbcZ differed by a factor of 1.5 in K and Lbcl and LbcZ differed by 6 kcal/mol in AP. For CO association, differences between Lbcl and Lbdl, although small, also appear to be significant. There is also a significant difference between Lbc* and Lbc* for 1. For oxygen and CO binding, LbcZ and Lbd* had the most aberrant behavior, having the smallest values for k', the largest for k (and hence the highest values for K), and the lowest values for l', and the highest values for L. The oxy complex of LbcZ differs significantly in its visible spectra and titration behavior from oxy-Lba and Lbcl (26). For azide binding, Lba and Lbcs differed by more than a factor of 2 in association rate constants. For azide affinity, Lba differed from Lbcl, LbcZ, and Lbdl by more than a factor of two and differences between Lbc's and corresponding Lbd's were evident.
Sequences have been published for Lba (27,28), Lbcl (28), and LbcZ and LbcZ (29). The sites where there are differences among the four proteins are listed in Table VI together with locations based on the structural homologies of Lesk and Chothia (30) and the x-ray structure of soybean Lb (31). The main ligand path to the heme site is considered to comprise residues 61-65, for which there are no differences among the various Lbs. A second ligand path has been identified in Lb (32) for which the entry is at Lys-36. With the possible exception of the subsitution of Val for Ala in Lbcs at location 39, there are no differences among the Lbs for these two  (26). The equilibrium constants for azide binding differ by a factor of 2 for LbcZ and Lbc+ Lba and Lbc, show significant differences in K and in the azide kinetics and equilibria.
Based on proton NMR measurements and a comparison of the sequences of Lba and Lbc, and the structure of Lupin Lb, it was concluded (32) that the heme environments for Lba and Lbcl differed. If these differences are attributable solely to heme contact alterations at heme contact sites, then the substitution at site 97 (Val + Ile) can affect absorption spectra, and the same substitution at site 91 can affect NMR spectra, K, and azide kinetics and equilibria.
Clearly, however, many of the functional differences between Lbc* and LbcZ must derive ultimately from differences outside the heme cavity and outside the two channels for ligand approach in Lb. The kinetic differences found among the Lbs are similar in magnitude to those found for diverse mammalian Mbs (33), for which no substitutions occurred in the heme cavity or along the Case and Karplus (34) ligation path.
Leghemoglobin shows pronounced R-state behavior (21) in both oxygen and CO ligation. For both ligands, a value for AS" can be calculated for transferring the ligand from the gas phase to the Fe-binding site. From simple statistical thermodynamic considerations, changes in translational, rotational, vibrational, and electronic entropy can be evaluated, with uncertainty arising from low frequency vibrations (35). For latm standard states, AS" should be about -38 e.u. for both ligands. For model heme compounds, binding of both CO and oxygen give very similar values for AS", typically -34 e.u., and W for CO binding is found to be about 3.5 kcal/mol more negative than for oxygen binding (36). This latter value is in excellent agreement with that reported here from direct measurement of M. Values in the literature for tiM (37) often show large errors when calculated from the appropriate rate constants, or when differences between fl values are taken for K and L. A recent extended x-ray absorption fine structure determination (38) shows that the Fe is essentially in the mean plane of the heme in LbCO. The x-ray structure for Lb shows a large heme cavity that easily accommodates nicotinate as a ligand. These structural findings are in accord with our thermodynamic and kinetic results suggesting that