Kinetics of MgATP-dependent Iron Chelation from the Fe-Protein of the Azotobacter uinelundii Nitrogenase Complex

Chelation of Fe from the Fe-protein component (Av2) of Azotobacter vinelandii nitrogenase has been investigated. The chelation, which requires MgATP binding by Av2, is best described as a two-exponential process. The rates for the two phases differed by -10-fold and increased as the concentration of MgATP was increased. The rates for both phases were 50% of maximum at approximately 1.5 mM MgATP. At MgATP concentrations >lo0 PM, the more rapid phase represented -25% of the total Fe chelated from Av2. However, below 100 WM MgATP, the proportion of the faster phase decreased until at 20 I.~M MgATP, only a single phase could be detected. The properties of Av2 were studied at various stages of Fe chelation. The partially chelated protein was isolated from the reaction by gel filtration and was subjected to a second MgATP-dependent Fe chelation. Material isolated after the completion of the first phase regained biphasic kinetics in subsequent chelation reactions. However, if MgATP was present during the isolation of Av2, then only a single phase was observed in the subsequent chelation studies. In addition, the enzymatic activity of Av2 decreased concomitantly with total Fe chelation. To account for these observations, a model is presented in which Av2 exists in two conformers. Fe chelation is proposed to occur from either conformer but only when two MgATP are bound. Both conformers bind MgATP with the same affinity but are distinguished by a 10-fold difference in chelation rate. The two conformers are in equilibrium and can interconvert only in the absence of MgATP. That is, MgATP binding prevents the conversion of the two conformational states.

sented -25% of the total Fe chelated from Av2. However, below 100 WM MgATP, the proportion of the faster phase decreased until at 20 I.~M MgATP, only a single phase could be detected.
The properties of Av2 were studied at various stages of Fe chelation. The partially chelated protein was isolated from the reaction by gel filtration and was subjected to a second MgATP-dependent Fe chelation. Material isolated after the completion of the first phase regained biphasic kinetics in subsequent chelation reactions. However, if MgATP was present during the isolation of Av2, then only a single phase was observed in the subsequent chelation studies. In addition, the enzymatic activity of Av2 decreased concomitantly with total Fe chelation.
To account for these observations, a model is presented in which Av2 exists in two conformers. Fe chelation is proposed to occur from either conformer but only when two MgATP are bound. Both conformers bind MgATP with the same affinity but are distinguished by a 10-fold difference in chelation rate. The two conformers are in equilibrium and can interconvert only in the absence of MgATP. That is, MgATP binding prevents the conversion of the two conformational states.
Biological reduction of dinitrogen to ammonia is catalyzed by the nitrogenase complex which is composed of the Feprotein and the MoFe-protein. The MoFe-protein contains the substrate binding and reduction site(s) while the Feprotein serves as the unique electron donor for the reaction.
For every electron transferred to the MoFe-protein, two ATP are bound and hydrolyzed by the Fe-protein (1,2). Because the Fe-protein is a single electron donor, multiple cycles of * This work was supported, in part, by the Science and Education Administration, United States Department of Agriculture Grant 82-CRCR-1-1119, and National Institutes of Health, Grant GM 34321. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Present address: Dept. of Biochemistry, Michigan State University, East Lansing, MI 48824. electron transfer and ATP hydrolysis are needed for the reduction of nitrogenase substrates (3). Each electron transfer appears to involve a cycle of association/dissociation of the protein complex (4, 5).
The Fe-protein has a single 4Fe:4S cluster which bridges the two identical subunits of the protein. The four cysteinyl ligands for the Fe:S center are residues 97 and 132 from each subunit (residue numbering is based upon the protein sequence of Av2)' (6). Although the exact mechanistic role of ATP hydrolysis in substrate reduction is not known, MgATP .binding to the Fe-protein induces a conformational change that alters the properties of the Fe center. For example, the midpoint potential, E ; , decreases -100 mV (7, 8), the EPR spectrum becomes more axial (9), and the magnetic circular dichroism spectrum changes (10).
One of the most striking effects of MgATP binding by the Fe-protein is the change in the reactivity of the Fe:S cluster with chelators. Walker and Mortenson (11,12) and Ljones and Burris (13) found that the Fe center was rapidly removed by chelators in the presence of MgATP but not in the presence of MgADP or in buffer alone. Furthermore, MgADP inhibited the MgATP-induced Fe chelation. The specificity and MgATP dependence of the reaction have been exploited to investigate other properties of the enzyme. For example, the amount of MgATP-dependent Fe chelation is proportional to the enzyme activity and can be used to estimate the fraction of active Fe:S center present (13). Also, using iodoacetic acid to label cysteinyl residues exposed during the Fe chelation reaction, the putative cluster ligands were identified (6). Finally, the nucleotide-binding constants have been estimated from the initial rate of chelation as a function of MgATP concentration (12)(13)(14)(15).
The time course for Fe chelation from reduced Fe-protein was described by previous workers as kinetically complex yet was not further characterized (11,12). Furthermore, Fe chelation from oxidized Av2 occurs in two sequential steps with the formation of a discrete 2Fe:2S intermediate (16). Thus, a detailed analysis of the kinetics of Fe chelation from reduced Av2 also might reveal new. properties of the protein and its Fe:S center. The time course of Fe chelation might provide a basis for comparing the reduced and oxidized states of Av2. Furthermore, in order to use the rate of Fe chelation as a measure of protein conformational change induced by nucleotide binding or by other agents, the relationship of initial rates to the full time course of Fe removal should be established. Finally, Fe chelation may prove a useful tool to probe subtle conformational differences between mutant forms of the protein.

6619
In this paper, we describe the kinetics for the full time course of Fe chelation from reduced Av2. As we observed for oxidized Av2, the removal of Fe is biphasic. However, unlike the sequential chelation of Fe from oxidized Av2, the Fe appeared to be removed, without intermediates, from two different forms of reduced Av2. The two forms, which interconvert in the absence of nucleotide, had a 10-fold difference in rate of Fe chelation.

EXPERIMENTAL PROCEDURES
Materials-ATP, creatine phosphate, creatine kinase, and 2,2'bipyridyl were obtained from Sigma. MgATP solutions were prepared by dissolving MgClz and ATP in 50 mM Tris-HCI, pH 8.0 followed by titrating the solution to pH 8.0 with NaOH. The concentration of ATP solutions was determined spectrophotometrically using an ~(260 nm) = 1.53 X lo' M" cm". ATP, creatine phosphate, and creatine kinase solutions were prepared fresh daily. MgATP concentrations were determined from the known concentrations of Mg and ATP and the stability constant, 5.01 X lo' M" (17). For the concentrations of Mg and ATP used in these experiments, Mg2ATP need not be considered (17).
To facilitate the preparation of 40 mM stock solutions of 2,2'bipyridyl, the suspension of the chelator in 50 mM Tris-HC1, pH 8.0, was warmed until the crystals melted and then was mixed vigorously. Because 2,2'-bipyridyl solutions are unstable for periods greater than a few weeks, new stocks were prepared weekly. The chelator concentration was determined using an e(300 nm) = 1.51 X 10' M" cm-I in 50 mM HC1 (18). The c(520 nm) for the Fe(2,2'-bip~ridyl)~ was 8.4 X lo3 M" cm" (11). For the o-bathophenanthroline disulfonate Fe complex, 4535 nm) = 2.21 X lo4 M" cm" was used (13). The manipulations of enzymes and reagents were carried out under an Ar atmosphere on a Schlenck manifold or in a glove box. Hamilton gas tight syringes and stainless steel cannulae were used for solution transfers. Ar was purified by passing the gas over heated BASF catalyst.
Enzyme Purification and Characterization-Avl and Av2 were purified by a modification of the method of Burgess et al. (19). The average specific activity of eight preparations of Av2 used in our experiments was 2770 f 240 (mean f S.D., n = 8) nmol Hz formed min" mg". Protein concentrations were determined by amino acid analyses. Upon treatment of Av2 with o-bathophenanthroline disulfonate (6 mM) under anaerobic conditions, 0.2 f 0.1 (mean f S.D., n = 8) mol of Fe/mol of Av2 was removed. When 5 mM MgATP was included in the chelation mixture, an additional 3.7 f 0.2 (mean f S.D. n = 8) mol of Fe/mol of Av2 was obtained. The Fe content also was determined from the protein hydrolyzate used for amino acid analysis. These measurements yielded an average of 3.6 f 0.1 mol of Fe/mol of Av2. Comparison of the Fe determined by both methods indicates that ATP induced Fe chelation is quantitative and that our preparations contain no Fe resistant to chelation.
Prior to the addition of Na2SZO4, the vials were deaerated by nine alternating cycles of evacuation and filling with Ar. The assay was initiated by addition of Av2 to vials containing Avl at 30 "C. After an 8-min incubation in a shaking water bath at 30.0 "C, the reaction was quenched with 0.25 ml of glacial acetic acid. Gas analysis was carried out on a Shimadzu gas chromatograph equipped with a 5 A molecular sieve column and a thermal conductivity detector. The Avl concentration was varied to determine optimal activity of Av2.
To detect traces of Avl in Av2 preparations, the enzyme assay was performed using high concentrations of Av2 (-20 PM) without added Avl, and the incubation time was increased to 2-5 h. We estimate that our preparations of Av2 contain 0.01 molar % Avl which is less than the amount we were able to detect by acrylamide gel electrophoresis.
Fe ChelationfromAu2 by I,P'-Bipyridyl-Fe chelation studies were performed in 1-cm quartz cuvettes equipped with 10-mm quartz tubing fused to their tops. The cuvettes were sealed with white rubber septa (Aldrich). To provide an inert gas barrier above the septum, a second septum was inserted into the cup formed by the sleeve of the lower septum. The cuvette and the space between the septa were deaerated separately using nine cycles of vacuum and Ar. The integrity of the cuvettes was demonstrated by -4 % HZ loss in 24 h. added in the following order; buffer, Av2, 2,2'-bipyridyl, creatine In a typical Fe chelation experiment, reaction components were phosphate and creatine phosphokinase, and MgATP. Unless otherwise indicated, the buffer used was 50 mM Tris-C1, pH 8.0, containing 100 mM NaCl. MgClz was added in amounts equimolar to ATP for ATP concentrations greater than 1.0 mM; at ATP concentrations below 1.0 mM, MgC12 was maintained at 1.0 mM. The final concentration of creatine phosphate was 2.0 mM, and of creatine phosphokinase, 0.1 mg/mI. MgATP and 2,2'-bipyridyl concentrations were varied as required, and the final concentrations of these reagents are reported in appropriate figures. With each addition, the absorbance at 520 nm was determined. For the 2,2'-bipyridyl addition, a second absorbance reading was taken after a 20-min incubation period to measure the amount of ATP-independent Fe chelation. The difference between the first and second absorbance readings in the presence of 2,2'-bipyridyl was less than 1% in all experiments. Comparable data were collected in which buffer was substituted for Av2. Buffers and substrates contributed less than 1% of the final absorbance. Mixing of reagents required 25-30 s and is included in the kinetic analysis. All experiments were performed at 30.0 "C.
Kinetic Analysis of Fe Chelation-Absorbance data were collected in a Beckman model DU-8B spectrophotometer equipped with a thermostat-controlled, six-position cell changer. The data were stored in a North Star computer via the RS232 port on the spectrophotometer. Because analysis of observations taken at a constant time interval weights too heavily later stages of a reaction, a subset of 100-150 data points was selected using a constant absorbance change interval for the full reaction time course. A data weighting scheme based on a particular model was considered inappropriate.
The selected data were analyzed by nonlinear regression programs described by Bevington for the Marquardt algorithm (20). (The programs have been adapted for North Star FPBASIC by Dr. Eric Eccleston of our laboratory.) All data sets were fit to equations consisting of 1,2, and 3 exponential terms with the absorbance value at infinite time (Absi,f) permitted to vary as an adjustable parameter. The calculated Abs,,f value deviated by less than 3% from the experimentally determined value, but optimum convergence of the experimental results with the exponential equations required this additional degree of freedom.
The statistical parameter, chi-squared ( x ) , was calculated for the fit of the data to equations 1-3 (see below). To determine which >20 was taken to indicate a preferential fit of the data to the equation equation best fit the data, the xs were compared as ratios. A ratio with the smaller x.

Time Course of Fe Chelation-A number of investigators have shown that the chelation of Fe from Av2 requires
MgATP; with chelator alone or with MgADP, only small quantities of Fe not associated with catalytically active Fe:S centers are removed (6,(11)(12)(13)(14)(15). At saturating concentrations of MgATP and high concentrations of chelator, the reaction is complete in less than 10 min. However, if the reaction is slowed by reducing the chelator concentration, the full time course of the Fe chelation can be measured and is found to be kinetically complex.
In Fig. lA is shown the absorbance change with time for one concentration of reactants. When the absorbance is plotted as a logarithmic function versus time, the complexity of the kinetics becomes apparent; namely, there is a significant deviation from linear between the early and late time points (see Fig. 1B).
In order to obtain a quantitative measure of the curvature, t h e data were analyzed by a nonlinear least squares procedure (see "Experimental Procedures"). The data were fit t o Equations 1,2, and 3 having one, two and three exponential terms, respectively.* 'Nomenclature used for rate constants: k., kb, kc refer to the observed, experimentally measured, rate constants for Fe chelation obtained by curve fitting the data to Equations 1-3; k values with numerical subscripts refer to specific rate constants defined by Schemes 1-3. For the curve fitting routine, the parameters on the righthand side of Equations 1-3, including the Absinf value, were allowed to vary. The results of a typical analysis are presented in the legend to Fig. 1. When a single exponential equation was used to 'fit the data (see Fig. lB), there was a large systematic deviation of the experimental values (see Fig. 1D for a plot of the statistical residuals). In general, the data were best represented by the two exponential Equation 2 (see  1L)) with little systematic deviation between the calculated and observed values. Attempts to fit the data to an equation containing three exponential terms failed to further reduce the residuals. Fits of the data to Equations 1 and 2 converged rapidly (four to eight iterations) and were relatively insensitive to the starting values of the parameters while calculations using Equation 3 rarely converged to a unique final set of parameters. We conclude that a two-exponential fit is necessary and sufficient to account for the full time course of Fe chelation from Av2.
When the data in Fig. lA were fit by Equation 2, k, and kb were found to differ by approximately an order of magnitude. This large difference in apparent rate constants facilitated detection of both exponential terms. The more rapid rate constant, k,, has a pre-exponential term (Pl) that represents -25% of the total absorbance change. The remaining 75% of the absorbance change is associated with the pre-exponential term (P2) of the slower phase, kb (Equation 2). Thus, the Fe chelation occurs as a "burst" in absorbance followed by a slower development of the remaining absorbance. In subsequent figures, the contribution of the first phase to the chelation kinetics will be expressed as %P1, where %P1 = series of control experiments were undertaken to demonstrate that the observed kinetic complexity of Fe chelation is a property of Av2 and not a consequence of the experimental conditions or protein purification methods. A potential source of the kinetic complexity could be the generation of an inhibitor during the chelation reaction. For example, MgADP, the product of nitrogenase turnover, is a potent inhibitor of the Fe chelation reaction (13). Because even highly purified Av2 contains traces of Avl (in our preparations, -0.01 molar % (see "Experimental Procedures")), slow nitrogenase turnover can occur during the Fe chelation reaction. To prevent the accumulation of MgADP, a M~TP-regenerating system was included in all our experiments. The capacity of the regenerating system was verified by adding Avl (up to 1 molar %) to Fe chelation reaction mixtures. Even this relatively large amount of Avl had no effect on the kinetics of the Fe chelation reaction as long as the complete MgATP regeneration system was present (Fig. 2).
Other potential inhibitors, such as the products of the Fe  chelation reaction (apoAv2, Fe-2,2'-bipyridyl complex, or sulfide), had no effect on the reaction time course (data not shown). Likewise, successive Fe chelation experiments performed without removing the products of the previous reaction had biphasic kinetics identical to the initial reaction. In addition, Av2 purified by an alternate method omitting the heat step exhibited the same chelation kinetics. Finally, a biphasic reaction time course with a %P1 of -25% was also observed with the chelator o-bathophenanthroline disulfonate (data not shown). Thus, the two-exponential character of the Fe chelation time course appears to be an intrinsic property of Av2.
Competence of MgAdPPNP to Induce Fe Chelation-The results above do not exclude the possibility that Fe chelation may be a result of MgATP hydrolysis and enzymic turnover of Av2. To investigate this possibility, the nonhydrolyzable MgATP analog, MgAdPPNP, was used in place of MgATP. Not only was MgAdPPNP competent to induce complete Fe chelation, but also the kinetics of the reaction were best described by Equation 2 (Table I). The slower rates observed with MgAdPPNP probably are due to Av2 not being saturated with the analog. Thus, hydrolysis of MgATP is not a necessary condition for Fe release nor does it produce the observed biphasic kinetics.
Both Kinetic Phases Represent Fully Chelated Fe-The work of Cowart et al. (21) indicates that Fe is chelated from transferrin in a biphasic process. For transferrin, the first phase was shown to be a spectral change due to the formation of a mixed ligand complex of chelator and protein; the removal of Fe occurred only in the second phase. We examined the possibility that the early kinetic phase in our experiments also simply represented a spectral change. Fe chelation experiments were performed in which the amount of proteinbound Fe and of chelated Fe was determined by gel filtration. The results are shown in Table 11. Even at one min the Fe chelated from Av2 was that expected if both kinetic phases involved the chelation and release of the Fe from the protein. Reaction was initiated with ATP, and quenched at the indicated time by addition of 10 mM ADP. In control experiments, it was shown that MgADP immediately stopped the absorption change due to chelation of Fe from Av2. The reaction mixture was transferred anaerobically to a 1.0 X 10-cm Sephadex G-25 column equilibrated with deaerated 25 mM Tris-C1, pH 8.0, containing 3 mM Na2S204. 0.5-ml fractions were collected. The Fe content of the protein and salt peaks were determined. The observed data represents the percentage of the Fe removed from the protein at the indicated times. An identical reaction mixture was monitored spectrophotometrically, and the reaction time course was fit to Equation 2.
* Expected percentage of Fe removed from Av2 calculated from the kinetic parameters in footnote a assuming two steps of Fe chelation. e Expected percentage of Fe removed from Av2 assuming that no Fe was chelated in the initial rapid phase of a sequential process. Based upon this assumption, a differential equation was derived to calculate the theoretical values. Thus, our results with Av2 are different from those with transferrin.
Dependence of Fe Chelation on the Concentration of Av2-The interaction between Av2 molecules was considered as a potential mechanism for generating the observed complex time course of Fe chelation. The effect of varying the concentration of Av2 was studied in a series of Fe chelation experiments, and the results are presented in Fig. 3. While there are changes in the apparent rate constants, %P1 and the ratio of ka to k,, remain unaltered. That is, the biphasic nature of Fe chelation is not a consequence o f protein concentration.
Loss of Enzymatic Activity with Fe Chelation-The Fe:S cluster of Av2 is essential to its catalytic activity. We examined the relationship between Fe chelation and the catalytic activity of Av2 by removing samples of Av2 from an Fe chelation reaction mixture and assaying for Av2 catalytic activity? The results are shown in Fig. 4. Despite the higher level of uncertainty inherent in the methods for activity ~~~ -~ ~ Control experiments showed no effect of 2,2'-bipyridyl in the nitrogenase assay under these conditions.  ..  measurements, the time course of activity loss also appears to be biphasic. Furthermore, the kinetic parameters for activity loss and Fe chelation are in reasonable agreement. Most importantly, partial enzyme activity persisted after the first phase of Fe chelation and was not completely lost until all Fe was chelated.
Effect of 2,2'-Bipyridyl Co~entrution on Fe Chelation-In part, the rate of Fe chelation from Av2 is dependent upon the concentration of chelator (Fig. 5). For all concentrations of 2,2'-bipyridyl studied, the reaction was biphasic. Both rate constants ( k , and kb) were second order in chelator concentration above -1 mM chelator (slopes of lines in Fig. 5A are   2.0). Further, %P1 and the ratio of kG to kb were the same over a 7-fold change in chelator concentration which results in an approximately 50-fold change in reaction velocity (Fig.   5B). For the studies presented here, 1.2 mM or higher chelator was used to ensure that pseudo-1st order conditions with respect to chelator were maintained. These results support the idea that biphasic Fe chelation is a property of the protein and not a consequence of the experimental protocol.
ATP C o~e n t r a t~o n -~e~~e~e of Fe Chelation-The dependence of the initial rate of Fe chelation on MgATP concentration has been exploited previously to estimate the stoichiometry and binding constants for MgATP with the Feprotein (12)(13)(14)(15). Because Fe chelation is biphasic, the initial rates used in earlier studies may be inadequate to accurately describe the MgATP dependence of the reaction. Thus, we have reinvestigated the role of MgATP concentration in chelation. In order to obtain measurable chelation rates over the range of MgATP concentrations to be studied, the concentration of 2,2'-bipyridyl was varied (See legend to Fig. 6). For those MgATP concentrations where the 2,2'-bipyridyl concentration was changed, a reaction was run at both the higher and lower chelator concentrations. Once k, and k b were cor- rected for the second order dependence upon 2,2'-bipyridyl concentration (see Fig. 5), the two reactions were found to have identical kinetic parameters. The results were normalized to a single chelator concentration 2.4 mM and are presented in Fig. 6.
The rate constants ka and k b had a similar sigmoidal dependence on the MgATP concentration and reached maximum values above 2-3 mM MgATP. The most striking effect of MgATP concentration was the change in %P1 (see Fig.  6C). Even though the individual rate constants increased 4-5-fold above -75 PM MgATP, %P1 remained constant. As the concentration of MgATP was decreased below 75 PM, %P1 decreased concomitantly. This shift to a first order process at low MgATP is evident in the semilogarithmic plot of Fig. 7 where the second phase is no longer discernible, c.f. the results in Fig. 1 for a higher MgATP concentration. Curve fitting analysis confirmed the presence of only one exponential process. Note that >95% of the Fe was chelated in the single, slow phase observed at low MgATP concentration. Thus, all Fe is accessible to chelation at all MgATP concentrations by either the apparent biphasic or monophasic process.

Regulation of the Kinetic Heterogeneity of Fe Chelation by
MgATP-The results presented in the section above indicate that the apparent number of kinetic phases as well as the rate of Fe chelation are controlled by the MgATP concentration. This raises the question as to whether the observed two phases at higher ATP concentrations indeed represent two subpopulations of Fe. Equation 2 predicts that at late time points the remaining Fe should be composed of a single, kinetically homogeneous subpopulation. To test this hypothesis, an Fe chelation reaction was conducted at a MgATP concentration (1 mM) sufficient to exhibit maximal biphasic kinetics. After -4 half-lives of the first phase, the chelation rate was substantially accelerated by the inclusion of additional MgATP, yet the progress curve remained a single exponential process. As predicted from the results of previous experiments (Fig.  6), the rate of chelation obtained a maximal value at -2-3 mM MgATP. Thus, we conclude that when the first phase is complete, the second behaves as a homogeneous population of molecules. Furthermore, the addition of MgATP is not, in itself, sufficient to induce biphasic reaction kinetics. Comparable results are obtained when the reaction is accelerated at late time points with 2,2'-bipyridyl (data not shown).
The results shown in Fig. 8 suggest that a kinetically homogeneous form of Av2 exists in the second phase of Fe chelation. Isolation of the protein at this point should provide material for further study to determine the properties of at least one of the Fe subpopulations. The Av2 remaining at late time points in the chelation reaction was isolated by ion exchange or gel filtration chromatography. When the reisolated protein was subjected to a further Fe chelation experiment, the time course of the reaction was again biphasic, with kinetic properties of native Av2. Thus, the isolation process apparently was sufficient to restore kinetic heterogeneity to Av2. That is, removal of a reaction component might have permitted regeneration of the two Fe subpopulations. To identify which, if any, component was involved, the protein from the Fe chelation reaction was isolated by gel filtration on columns equilibrated with either 2,2'-bipyridyl or MgATP. The reisolated Av2 was subjected to a second Fe chelation experiment and the kinetic pattern determined. For protein isolated in the presence of 2,2'-bipyridyl, a biphasic time course indistinguishable from native Av2 was observed. In contrast, Av2 isolated in the presence of MgATP had essentially a single exponential time course (Table 111). Thus, not only is MgATP required for Fe chelation from Av2 but it appears to affect the conversion of different species of Fe which, in turn, are responsible for the biphasic kinetics of chelation.
In the experiments above, the initial MgATP concentration was sufficiently high that the Fe chelation was biphasic and the reaction behaved as though it contained two Fe populations. In contrast, when the concentration of MgATP is below -75 pM, the reaction appears to be a single exponential process throughout (see Fig. 8). This raises new, corollary questions: at low MgATP, does the reaction contain a single population of Fe and can the reaction be made biphasic if it was initially monophasic?
To investigate these questions, the Fe chelation reaction was performed using the appropriate low MgATP concentration to produce a single exponential time course. After -60% of the Fe was chelated, the MgATP concentration was increased to 0.2 mM. As shown in Fig. 9, the reaction was accelerated and immediately became biphasic. Although the two new phases accounted for only the last 40% of the Fe to be chelated, they had the same %P1 and ratio of k, to kt, as observed for reactions which were biphasic throughout. Furthermore, the rates for the two phases were identical to those obtained for a reaction initially at the higher MgATP concentration. These experiments demonstrate the unique ability of MgATP to induce the expression of kinetic heterogeneity in Av2. In addition, we conclude that heterogeneity among Fe  TABLE I11 Effect of isolation conditions on the chelation kinetics of Au2 Data for k J k b and %P1 are presented as the numerical mean of experiments with the standard deviation in parentheses. Reactions were initiated with [ A v~] = 40 WM, 2,2'-bipyridyl = 3.0 mM, and [ATP] = 0.5 mM. After 3-4 half-lives of the rapid reaction phase, as determined in duplicate spectrophotometric studies, the reaction mixture was chromatographed. For gel filtration experiments, the reaction mixture was applied anaerobically to a 1. X 10-cm Sephadex G-25 column equilibrated with 25 mM Tris-C1, pH 8.0, containing 3.0 mM Na&04, with the indicated additional compounds. For DEAE-chromatography, the reaction mixture was applied to a 0.7 X 4-cm DEAE-Sepharose column equilibrated with 25 mM Tris-C1, pH 8.0, containing 3.0 mM Na2S204. The protein peak was eluted with 0.5 mM NaCl in the same buffer. In both gel filtration and DEAE procedures, the protein peak was collected directly into an anaerobic spectrophotometer cuvette and subjected to a second round of Fe chelation.  atoms in Av2 is latent under these conditions despite the apparent single exponential nature of the Fe chelation process.

DISCUSSION
Model for Fe Chelation-Two limiting models described by Equation 2 can be considered for the biphasic Fe chelation from Av2. Fe could be chelated sequentially in two partial reactions (Scheme 11, or the Fe could be removed by parallel reactions from two different sub-populations of Av2 (Scheme 2). An additional possibility, that the first phase represents formation of a mixed ligand complex of chelator and protein, has been excluded by the gel filtration experiments (Table  11). Although Schemes 1 and 2 are kinetically equivalent, their specific rate laws are considerably different and can be distinguished by experimental conditions.
Av2' (4Fe) SCHEME 2 However, neither of these mechanisms 2s sufficient to account for the experimental results. For example, the sequential model (Scheme 1) posits the formation of an intermediate with fewer than the original four Fe. Yet, the enzymatic activity decreased simultaneously with removal of all four Fe, not with just the first phase (Fig. 4). This would require the unlikely condition that Av2 with fewer than four Fe be catalytically active. In addition, Av2, reisolated after the first phase of Fe chelation, exhibited the same biphasic Fe chelation behavior as the original protein (Table 111). For Scheme 1 to account for either of these results, the original Fe center would have to be reconstituted during the subsequent manipulation of the protein. Although other Fe:S proteins are known to reform active Fe centers from fragmented clusters, such reconstitution has not been found in the presence of chelator as is our condition (22,23).
Likewise, the parallel model of Scheme 2 is not consistent with all the experimental results. Because both rate constants for Fe chelation have the same dependence on MgATP (see Fig. 6), Scheme 2 is inadequate to explain the change in number of phases as the concentration of MgATP is altered. As for Scheme 1, it is difficult to rationalize by Scheme 2 how the two phases of Fe chelation could be re-established after exhaustion of the first phase and isolation of the protein (Table 111).
To overcome the limitations in Schemes 1 and 2, we have developed Scheme 3 as a minimal model consistent with our chelation data. Scheme 3 is a modification of Scheme 2 with Av2b is more rapid than chelation. Thus, the reaction is from a pool of components in rapid equilibrium and the observed rate is a weighted average of the two pathways. The result is that the reaction appears monophasic and slow (see Fig. 7).
As the concentration of MgATP is increased, the rate of conversion between the nucleotide-free forms is less than kp which allows the two pathways to be expressed as biphasic kinetics. Over these intermediate MgATP concentrations the fast phase (%P1) appears to increase as the MgATP increases (see Fig. 6). That is, the fraction of Av2a appears to be increasing with increasing MgATP concentrations. This is an illusion explained by Scheme 3 and Equation 4. Namely, the pre-exponential values (observed %P) represent the true percentage of the Av2 forms only at saturating MgATP where the conversion is prevented. At lower concentrations, the observed rates and percentage of the phases are influenced by the fraction of the protein which is MgATP free, k3, and k d . Thus, the change from monophasic to biphasic kinetics can be effected by the shift from low to higher MgATP concentrations at any point in the reaction (see Fig. 9, A and B ) .
That is, the kinetic heterogeneity remains latent at low MgATP by the conversion of the nucleotide-free Av2 forms.
Finally, when conversion of Av2 forms is blocked by the higher MgATP concentrations, the more reactive Av2a will be depleted in the rapid initial phase leaving a kinetically homogeneous material, Av2b (See Fig. 8). It is this latter material which, if isolated in the presence of MgATP, retains a single kinetic phase in subsequent chelation reactions yet can re-equilibrate to the two forms in the absence of nucleotides (see Table 111).
In order to investigate Scheme 3 in quantitative detail, we have solved the relevant differential equations, assuming that where These equations have a large number of variables which, we anticipate, might allow for multiple solutions. To fit the data, we have limited ourselves to those we consider potentially relevant to the nitrogenase enzymology. We have constrained the two rate law constants, kl and kz, to the limiting, observed rates at saturating MgATP. K, and the co-operativity were considered to be the same for both conformers and limited to approximately 2 X M-' that has been estimated by Watt from equilibrium binding studies (8). One set of values clearly best fit the experimental results (see Fig. 6). The striking feature of the model is the requirement for the very slow rate of conversion between Av2a and Av2b. However, other slow conformational changes have been observed in macromolecules such as the cis-trans isomerization, the binding of slow inhibitors, and in the complement protein family of C-3, C-4, and a-2-macroglobulin (24)(25)(26). In order for a faster rate of conversion between Av2a and Av2b to be accommodated by Scheme 3, a large negative cooperativity for MgATP binding and PM dissociation constants would be required. We considered these parameters outside the acceptable limits of the previously reported binding constants (1,2,8).
Most importantly, Equation 4 appears adequate to describe the full time course of Fe chelation from Av2 in the presence of nucleotides and chelators. Our results appear to require two active conformers of Av2. It should be noted that there are other reports of multiple forms of nitrogenase Fe-protein.
For example, recent low temperature Mossbauer, magnetic susceptibility, and EPR results on reduced Av2 have been interpreted as indicating the presence of Av2 molecules in two spin states, S = 1 h and S = 3% (27)(28)(29). Because the Fe:S center is affected by the protein isomerization in all of these phenomena, there is intrinsic interest in the properties of the conformations. Aithough a synthesis of the origin of the various multiple forms is not yet possible, it seems reasonable that there is some underlying correlation which is related to the unique property of the Fe-protein as the specific electron donor in nitrogenase.
Regardless of the physical interpretation of the differences in Fe:S cluster reactivity, the present study makes clear that the interaction of Av2 with MgATP is more complex than previously realized. That is, factors other than MgATP binding per se can dramatically affect the reactivity of the Fe:S cluster to chelators; these factors establish the reactivity difference between Av2a and Av2b. In addition, the ability of MgATP to inhibit equilibration between conformers of Av2 reveals a previously unnoticed allosteric effect in Av2. This effect should be borne in mind when correlation is made between structure and function. For example, MgATP may restrict protein flexibility and may bias the enzyme toward catal~ically active conformations. Further, care must be taken that all Av2 molecules in a reaction comprise a single ensemble, or that the rate of protein isomerization is taken into consideration. Finally, the isomerization properties of Fe-protein may be altered by selective mutation. This study provides a basis to assess the effects of the mutations.