Involvement of the P cluster in intramolecular electron transfer within the nitrogenase MoFe protein.

Nitrogenase is the catalytic component of biological nitrogen fixation, and it is comprised of two component proteins called the Fe protein and MoFe protein. The Fe protein contains a single Fe4S4 cluster, and the MoFe protein contains two metallocluster types called the P cluster (Fe8S8) and FeMo-cofactor (Fe7S9Mo-homocitrate). During turnover, electrons are delivered one at a time from the Fe protein to the MoFe protein in a reaction coupled to component-protein association-dissociation and MgATP hydrolysis. Under conditions of optimum activity, the rate of component-protein dissociation is rate-limiting. The Fe protein's Fe4S4 cluster is the redox entity responsible for intermolecular electron delivery to the MoFe protein, and FeMo-cofactor provides the substrate reduction site. In contrast, the role of the P cluster in catalysis is not well understood although it is believed to be involved in accumulating electrons delivered from the Fe protein and brokering their intramolecular delivery to the substrate reduction site. A nitrogenase component-protein docking model, which is based on the crystallographic structures of the component proteins and which pairs the 2-fold symmetric surface of the Fe protein with the exposed surface of the MoFe protein's pseudosymmetric αβ interface, is now available. During component-protein interaction, this model places the P cluster between the Fe protein's Fe4S4 cluster and FeMo-cofactor, which implies that the P cluster is involved in mediating intramolecular electron transfer between the clusters. In the present study, evidence supporting this idea was obtained by demonstrating that it is possible to alter the rate of substrate reduction by perturbing the polypeptide environment between the P cluster and FeMo-cofactor without necessarily disrupting the metallocluster polypeptide environments or altering component-protein interaction.

Nitrogenase is the catalytic component of biological nitrogen fixation, and it is comprised of two component proteins called the Fe protein and MoFe protein. The Fe protein contains a single Fe 4 S 4 cluster, and the MoFe protein contains two metallocluster types called the P cluster (Fe 8 S 8 ) and FeMo-cofactor (Fe 7 S 9 Mo-homocitrate). During turnover, electrons are delivered one at a time from the Fe protein to the MoFe protein in a reaction coupled to component-protein association-dissociation and MgATP hydrolysis. Under conditions of optimum activity, the rate of component-protein dissociation is rate-limiting. The Fe protein's Fe 4 S 4 cluster is the redox entity responsible for intermolecular electron delivery to the MoFe protein, and FeMo-cofactor provides the substrate reduction site. In contrast, the role of the P cluster in catalysis is not well understood although it is believed to be involved in accumulating electrons delivered from the Fe protein and brokering their intramolecular delivery to the substrate reduction site. A nitrogenase component-protein docking model, which is based on the crystallographic structures of the component proteins and which pairs the 2-fold symmetric surface of the Fe protein with the exposed surface of the MoFe protein's pseudosymmetric ␣␤ interface, is now available. During component-protein interaction, this model places the P cluster between the Fe protein's Fe 4 S 4 cluster and FeMo-cofactor, which implies that the P cluster is involved in mediating intramolecular electron transfer between the clusters. In the present study, evidence supporting this idea was obtained by demonstrating that it is possible to alter the rate of substrate reduction by perturbing the polypeptide environment between the P cluster and FeMo-cofactor without necessarily disrupting the metallocluster polypeptide environments or altering component-protein interaction.
The MgATP-dependent reduction of nitrogen gas to yield ammonia has a minimal stoichiometry usually indicated as follows.
The reaction is catalyzed by nitrogenase, which is comprised of two component proteins called the Fe protein, a homodimer, and the MoFe protein, an ␣ 2 ␤ 2 heterotetramer (reviewed by Dean et al. (1993), Howard and Rees (1994), and Kim and Rees (1994); see Fig. 1 for structural models). Turnover requires the sequential delivery of single electrons from the Fe protein to the MoFe protein and involves the association and dissociation of the protein partners in a process where MgATP hydrolysis is coupled to electron transfer. Three different metal clusters are believed to be involved in electron transfer and substrate reduction. These include an Fe 4 S 4 cluster, which is bridged between the identical subunits of the Fe protein, and two pairs of unusual metal clusters, called the P cluster and FeMo-cofactor, both of which are contained within the MoFe protein. There is one P cluster and one FeMo-cofactor within each independently operating MoFe protein ␣␤-unit. Each P cluster is constructed from two Fe 4 S 4 subcluster fragments linked by a corner-tocorner disulfide bond and is bridged between the ␣and ␤-subunits at an interface exhibiting pseudo-2-fold symmetry. The P cluster is coordinated to the protein by residues Cys ␣62 , Cys ␣88 , Cys ␣154 , Cys ␤70 , Cys ␤95 , Cys ␤153 , and Ser ␤188 (numbers refer to the primary sequences of the component proteins from Azotobacter vinelandii) with Cys ␣88 and Cys ␤95 bridging the Fe 4 S 4 subcluster fragments. FeMo-cofactor contains a metal-sulfide core (Fe 7 S 9 Mo) and one molecule of (R)-homocitrate. The metal-sulfide core is constructed from MoFe 3 S 3 and Fe 4 S 3 subcluster fragments, joined by a ring of three sulfide bridges connecting pairs of iron atoms. FeMo-cofactor is contained entirely within the ␣-subunit and is covalently attached to the protein through a thiolate ligand provided by Cys ␣275 to an iron atom at one end of the prosthetic group and by the imidazole ␦-nitrogen atom of His ␣442 to the molybdenum atom at the opposite end.
There is compelling evidence that the Fe protein's Fe 4 S 4 cluster is the obligate electron donor to the MoFe protein and that it cycles between the 1 ϩ and 2 ϩ redox state during the sequential single-electron deliveries (Smith and Lang, 1974;Stephens, 1985;Lindahl et al., 1985). It is also known that FeMo-cofactor provides the substrate reduction site (Shah and Brill, 1977;Hawkes et al., 1984;Scott et al., 1990;Scott et al., 1992;Kim et al., 1995). In contrast, the specific role of the P cluster in catalysis is much less certain, but it is believed to be involved in accumulating electrons delivered from the Fe protein and brokering their intramolecular delivery to the substrate reduction site. Thus, questions concerning the role of the P cluster in catalysis include: (i) whether or not the P cluster is involved in mediating the delivery of electrons from the Fe protein's Fe 4 S 4 cluster to the FeMo-cofactor; (ii) if so, how many electrons can be accumulated by the P cluster; and (iii) what the path is for intramolecular electron delivery between the P cluster and FeMo-cofactor. Kim and Rees (1992) have previously recognized four helices, which are oriented in parallel and located between the P clus-ter and FeMo-cofactor, that could participate in electron transfer between the two clusters (Fig. 2). The Tyr ␤98 residue is located on one of these helices and is situated in a direct line between the two clusters. This residue is also one of a group of hydrophobic residues that provide the polypeptide environment of the P cluster without being in contact with it. Furthermore, Tyr ␤98 approaches the terminal carboxyl of the homocitrate moiety of FeMo-cofactor and may indirectly interact with it by hydrogen bonding through water. 1 The homologous residue within the ␣-subunit, ␣-Tyr 91 , is also found on a helix located between the clusters but its side chain is directed away rather than toward the FeMo-cofactor (Fig. 2). To assess the role of the P cluster and the possible participation of Tyr ␤98 in intramolecular electron transfer, this residue was substituted by Phe, Leu, and His and the catalytic, kinetic, and spectroscopic properties of the resulting altered MoFe proteins were examined. Thus, the experimental rationale was to ask whether or not it is possible to perturb intramolecular electron transfer without disrupting the polypeptide environments of the metalloclusters or altering component protein interaction.
In a parallel set of experiments, the ␣-subunit residue, Tyr ␣91 , was also substituted by Phe, Leu, and His, and the catalytic and kinetic properties of the altered His ␣91 MoFe protein were examined as well. These latter experiments were intended to serve as a control, with the substitutions at this position considered less likely to have an effect on intramolecular electron transfer.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-Methods for site-directed mutagenesis, gene replacement, and the isolation of mutant strains were performed as described or cited previously (Brigle et al., 1987;Jacobson et al., 1989). Oligonucleotides used for mutagenesis, hybrid plasmids used for strain constructions, and strain designations are shown in Table I. For site-directed mutagenesis of the A. vinelandii nifD gene, encoding the MoFe protein ␣-subunit (Brigle et al., 1985), an approximately 1.4kilobase DNA fragment containing the nifD gene was cloned into EcoRI-digested pUC119 (Vieira and Messing, 1987). This plasmid was designated pDB697 and is the one from which plasmids pDB880, pDB809, pDB814 were derived (Table I). For site-directed mutagenesis of the nifK gene (Brigle et al., 1985), encoding the MoFe protein ␤-subunit, an approximately 1-kilobase HindIII-KpnI DNA fragment containing the 5Ј half of the A. vinelandii nifK gene was cloned into HindIII-KpnI-digested pUC118 (Vieira and Messing, 1987). This plasmid was designated pDB698 and is the one from which plasmids pDB796, pDB797, and pDB860 were derived (Table I).
Cell Growth, Protein Purification, and Assays-Wild-type and mutant strains of A. vinelandii were cultured at 30°C in a modified, liquid Burk medium (Strandberg and Wilson, 1968) containing 0.01 mM Na 2 MoO 4 to repress the alternative nitrogenase systems. Nitrogenase was purified and assayed according to the methods described by Burgess et al. (1980) with the modifications described by Peters et al. (1994) and references cited therein. The purity of protein samples was determined by denaturing polyacrylamide (12%) gel electrophoresis using modifications of the procedure described by Laemmli (1970) as discussed by Scott et al. (1992). Proteins subjected to polyacrylamide gel electrophoresis were stained using Coomassie Brilliant Blue. Flavodoxin used in stopped-flow spectrophotometric experiments was isolated from A. vinelandii cells as described by Klugkist et al. (1986). The molybdenum content of the isolated wild-type and His ␤98 MoFe protein was determined to be 1.9 Ϯ 0.1 and 1.8 Ϯ 0.1 g atoms of molybdenum/ mol, respectively, by inductively coupled plasma atomic emission spectrometry using a simultaneous spectrometer (Jarrell-Ash ICAP 9000) and a sequential scanning spectrometer (Jarrell-Ash Atomscan 2400). The iron content of the purified proteins was also determined and found 1 J. T. Bolin, personal communication. The Fe 4 S 4 cluster contained within the Fe protein, and the P cluster and FeMo-cofactor contained within the MoFe protein, are included as space-filling models. In the docking model, the P cluster is located between the Fe protein's Fe 4 S 4 cluster and FeMo-cofactor.

FIG. 2. Coil diagram of a section of the pseudosymmetric in-
terface between the MoFe protein ␣and ␤-subunits. The view is approximately the same as shown in Fig. 1. The spatial relationships of the Tyr ␤98 residue and the Tyr ␣91 residue with respect to the P cluster and FeMo-cofactor are indicated in panel A. The pseudosymmetric nature of the two helices, which, respectively, contain the Tyr ␤98 residue and the Tyr ␣91 residue, is also recognized in the conservation of primary sequences between the ␣and ␤-subunits, as shown in panel B. Conserved residues are boxed, and the tyrosine residues targeted for substitution and the proximal P cluster-coordinating cysteine residues are indicated by numbers above or below the respective sequences.
to be 26 Ϯ 2 g atoms of iron/mol of purified protein. Acetylene K m values for both the wild-type and the His ␤98 MoFe proteins were obtained from Lineweaver-Burk plots as described by Kim et al. (1995). Maximum specific activity determinations (Table II) and other kinetic experiments were performed using saturating levels of Fe protein (at least a 40:1 Fe protein:MoFe protein molar ratio). Except for the stopped-flow experiments described below, all nitrogenase assays contained a total of 0.5 mg of the nitrogenase component proteins in a final assay volume of 1 ml. Protein concentrations were determined by the method of Lowry et al. (1951). Experimental results were reproducible, and all data shown are representative of at least three separate experiments.
Stopped Flow Spectrophotometry-Methods used to determine the rates of primary and secondary electron transfer are described in detail elsewhere (Thorneley, 1975;Fisher et al., 1991). Primary electron transfer refers to the transfer of a single electron from Fe protein in the 1 ϩ redox state to the semireduced as-isolated MoFe protein, whereas secondary electron transfer refers to the transfer of a second electron to the one-electron-reduced MoFe protein. The underlying principle involved is that oxidation of Fe protein's Fe 4 S 4 cluster results in an increased absorbance at 430 nm that may be monitored by visible spectroscopy. Stopped-flow spectrophotometry was performed using a commercially available SF-61 instrument equipped with a kinetic data acquisition analysis and curve-fitting system (Hi-Tech, Salisbury, Wiltshire, United Kingdom). The SHU-61 sample handling unit was installed inside an anaerobic chamber (Vacuum Atmospheres Corp., Rosedale, CA) operating at less than 1 ppm O 2 . Sample flow components were thermostatted by closed circulation of water by a Techne C-85D circulator (Techne Ltd., Duxford, Cambridge, UK) attached to a FC-200 Techne flow cooler situated outside the anaerobic chamber. All stoppedflow reactions were studied at 23°C in 25 mM Hepes buffer, pH 7.4, containing 10 mM MgCl 2 and 50 mM NaCl. For primary electron-transfer determinations, syringe A contained 10 M MoFe protein, 40 M Fe protein, and 10 mM Na 2 S 2 O 4 , and syringe B contained 20 mM MgATP and 10 mM Na 2 S 2 O 4 . For determination of the rate of secondary electron transfer, conditions of low electron flux were initiated by the addition of 18 mM MgATP, 18 mM creatine phosphate, and 60 g of creatine phosphokinase to a mixture of 40 M MoFe protein and 0.4 M Fe protein in a stopped-flow drive syringe via a mixing block as described by Fisher et al. (1991). The second drive syringe contained 80 M Fe protein. The first stopped-flow trace was recorded as quickly as possible after the premixing procedure, and successive stopped-flow traces were obtained at time intervals up to 60 min.
The turnover rate was also measured by stopped-flow spectrophotometry. In these experiments, flavodoxin (product of the nifF gene), the physiological electron donor to the Fe protein was used as reductant. Reduction of oxidized Fe protein by the hydroquinone form of flavodoxin to yield the semiquinone form is accompanied by an increase in absorbance at 580 nm. Thus, under turnover conditions, the reduced hydroquinone form of flavodoxin is oxidized as it rapidly transfers its electron to the oxidized Fe protein, which becomes available as the MoFe protein-Fe protein complex slowly dissociates. In this way, the consumption of reducing equivalents, which occurs during nitrogenase turnover, can be continuously monitored by following the increased absorbance at 580 nm. Under normal conditions, the turnover rate estimated by the flavodoxin-oxidation rate provides an indirect measure of the rate of component-protein dissociation because this dissociation is rate-limiting in catalysis (Thorneley and Lowe, 1983). In the analysis of the altered His ␤98 MoFe protein, however, where the overall turnover rate might ultimately be limited by intramolecular electron transfer, the observed change in optical density was used to calculate an "apparent" dissociation rate because here component-protein dissociation may not be rate-limiting. For these experiments, flavodoxin was reduced to the hydroquinone state by using an excess of Na 2 S 2 O 4 at pH 8 in an anaerobic chamber. After incubation for approximately 20 min to ensure complete reduction, unreacted dithionite and its oxidation products were removed on a gel filtration column (0.5 ϫ 5 cm) packed with P-6DG (Bio-Rad, Melville, NY) that had been pre-equilibrated with anaerobic 25 mM Hepes, pH 7.4, 10 mM MgCl 2 .
Electron Paramagnetic Resonance (EPR) 2 Spectroscopy and Magnetic Circular Dichroism (MCD) Spectrophotometry-MoFe protein samples for EPR spectroscopy were exchanged into 25 mM Tris-HCl, pH 7.4, 100 mM NaCl, and 5 mM Na 2 S 2 O 4 using Biogel P-6DG gel filtration resin (Bio-Rad, Melville, NY). Purified protein samples were concentrated to at least 10 mg/ml and loaded into EPR tubes in an anaerobic chamber and then frozen in liquid nitrogen. EPR spectra were recorded at 9.22 GHz and 20 milliwatts on a Varian Associates E-line spectrometer with a modulation amplitude of 2.5 milliteslas at 100 kHz. The temperature was maintained at 12 K by liquid helium boil-off. The setup and procedure for performing MCD spectrophotometry was as described previously (Johnson, 1988).

Diazotrophic Growth of Mutant Strains and Activities of the
Purified, Altered MoFe Proteins-All six mutant strains constructed by substituting either Tyr ␣91 or Tyr ␤98 by Phe, Leu, and His, were capable of diazotrophic growth (Table I). The mutant strains, which produce altered MoFe proteins having either the ␤-Phe 98 or the ␤-Leu 98 substitutions, grew diazotrophically almost as well as wild type (4-versus 3-h doubling time), while the strain with the His ␤98 substitution grew slowest with a 6-h doubling time. All three mutant strains with substitutions at the Tyr ␣91 position exhibited, at most, only a small increase in diazotrophic doubling time.
MoFe proteins were isolated and purified in parallel from all three strains with substitutions at Tyr ␤98 , from the His ␣91 strain, and from wild type. The isolated His ␣91 MoFe protein exhibited kinetic and catalytic properties nearly identical to the wild-type MoFe protein, confirming that this substitution has no discernible effect on electron transfer from Fe protein through to substrate. As expected from the good growth rates of their parent strains, the altered Phe ␤98 and Leu ␤98 MoFe proteins exhibited catalytic activities for H 2 evolution, C 2 H 2 reduction, N 2 fixation, and concomitant MgATP hydrolysis comparable with those of the wild type. In contrast to the other altered MoFe proteins, the His ␤98 MoFe protein showed a significant decrease in the maximum specific activity for substrate reduction, while the overall rate of MgATP hydrolysis was maintained near the wild type rate (Table II). In other words, the reaction catalyzed by the altered His ␤98 MoFe protein exhibits MgATP hydrolysis that is partially uncoupled from electron transfer. For all of the altered MoFe proteins, the effect of 10% CO was to divert all electron flux to H ϩ reduction, which was insensitive to the presence of CO, as seen for wild type. No ethane was observed from catalyzed C 2 H 2 reduction.
A number of experiments were performed on the His ␤98 2 The abbreviations used are: EPR, electron paramagnetic resonance; MCD, magnetic circular dichroism. MoFe protein to determine whether or not its lower specific activity is due to an alteration in intramolecular electron delivery. The effect of varying electron flux was investigated to probe if its ability to deliver electrons to the substrate had been compromised. Component-protein ratio titrations performed in parallel for both wild-type and His ␤98 MoFe proteins revealed that their specific activities maximized at different Fe protein: MoFe protein ratios (Fig. 3). For the wild-type MoFe protein, a maximum specific activity of 2250 nmol of H 2 formed/min/mg of MoFe protein was achieved at an Fe protein:MoFe protein molar ratio of greater than 10:1 under conditions of proton reduction. Under the same conditions, a maximum specific activity of 1100 nmol of H 2 formed/min/mg of MoFe protein was observed for the His ␤98 MoFe protein, and this value was achieved at the lower Fe protein:MoFe protein molar ratio of approximately 5:1 (Fig. 3). This effect was also apparent in analogous titrations performed under conditions of acetylene reduction, where the activity of the His ␤98 MoFe protein maximizes at only about 600 nmol of C 2 H 4 produced/min/mg of MoFe protein at the same 5:1 molar ratio, while the wild type reaches the expected maximum activity of 2100 nmol C 2 H 4 formed/min/mg of MoFe protein at a molar ratio of greater than 20:1. The K m values for C 2 H 2 reduction were also determined for the wild-type and His ␤98 MoFe proteins and found to be comparable at 0.0055 and 0.0040 atm, respectively. These values are in the range previously reported for wild type (Dilworth, 1966;Schollhorn and Burris, 1967;Kim et al., 1995).
The lowered maximum C 2 H 2 reduction activity for the His ␤98 MoFe protein, when compared with its maximum H ϩ reduction activity, is not compensated for by increased H ϩ reduction under 10% C 2 H 2 to maintain a constant electron flux (Table II). Thus, the His ␤98 MoFe protein appears to suffer greater inhibition of electron flux under a 10% C 2 H 2 atmosphere than under either 100% N 2 or 100% argon atmospheres. The inhibition of electron flux during C 2 H 2 reduction catalyzed by the His ␤98 MoFe protein was partially relieved by carbon monoxide, bringing it to approximately the same level observed under 100% N 2 or 100% argon atmospheres .
Stopped-flow Spectrophotometry-The possibility that the reduced activity and uncoupled MgATP hydrolysis exhibited by the His ␤98 MoFe protein is due to the altered MoFe protein being unable to easily transfer an electron from the P cluster to FeMo-cofactor, such that intramolecular electron transfer is rate-limiting, was investigated using two different types of continuous stopped-flow spectrophotometric experiments. In the first series of experiments, the turnover rate for catalysis, i.e. the dissociation rate of the Fe protein-MoFe protein complex involving either the wild-type MoFe protein or the His ␤98 MoFe protein, was determined indirectly by measurement of the rate of nitrogenase-dependent oxidation of flavodoxin from its hydroquinone form to the semiquinone (Fig. 4). Wild-type MoFe protein-dependent flavodoxin oxidation was found to exhibit a single linear function having a calculated turnover rate of 6.3 s Ϫ1 . This value, which is calculated from the rate of absorbance change, the ⌬⑀ 580 , is 5.7 mM Ϫ1 cm Ϫ1 for the hydroquinone form of flavodoxin (Klugkist et al., 1986), and the Mo content of the isolated MoFe protein is in excellent agreement with the value of 6.4 s Ϫ1 previously reported for the dissociation rate of the Klebsiella pneumoniae nitrogenase component  a Specific activity is expressed in nmol of product formed/min/mg of purified MoFe protein under the atmosphere indicated and the conditions specified under "Experimental Procedures." b Total 2e Ϫ represents the electron pairs derived from all products in a particular experiment to allow direct comparisons among substrates. 1 nmol of C 2 H 4 or H 2 is equivalent to 1 nmol of 2e Ϫ and 1 nmol of NH 3 is equivalent to 1.5 nmol of 2e Ϫ . c ATP/2e Ϫ represents nmol of MgATP hydrolyzed per electron pair transferred to substrate. d Approximately 10% carbon monoxide (CO) was added, which is sufficient for complete inhibition of C 2 H 2 or N 2 reduction activities.
proteins determined from steady-state data (Thorneley and Lowe, 1983). For the His ␤98 MoFe protein, the apparent rate of component-protein dissociation was calculated as 6.0 s Ϫ1 but only for approximately the first 300 ms. After this brief initial time period, the apparent rate of component protein dissociation in the His ␤98 MoFe protein-catalyzed reaction was slowed approximately 2-fold to a linear rate of 2.5 s Ϫ1 . From the known amounts of all reactants in the continuous stopped-flow experiments, the number of electrons delivered to the MoFe protein prior to the observed lowering in turnover rate was calculated as 2.4 indicating that the P cluster is able to accumulate at least two electrons prior to the intramolecular electron delivery event.
In the second set of stopped-flow experiments, the rates of both primary and secondary intermolecular electron transfers to both the wild-type and His ␤98 MoFe proteins were measured, as were the absorbance changes that occur at longer times after the initial electron transfers (after 150 ms). The primary and secondary electron transfer rates to both the wild-type and His ␤98 MoFe proteins were found to be identical at 158 s Ϫ1 (Fig.  5; data for determination of the secondary electron transfer rate are not shown) and within the range previously reported (Thorneley, 1975;Fisher et al., 1991). In contrast to the wildtype MoFe protein, the reaction involving the His ␤98 MoFe protein exhibits a gradual decrease in optical absorbance after about 150 ms in the pre-steady state experiment (Fig. 5). Further, this decrease in absorbance occurs exponentially with a rate constant of 2.7 s Ϫ1 , which is reminiscent of the proteinprotein complex dissociation rate of 2.5 s Ϫ1 that is measured for the latter part of this biphasic process. Thus, the absorbance decrease may reflect reduction of the oxidized Fe protein as it dissociates from the complex.
Electron Paramagnetic Resonance and Magnetic Circular Dichroism Spectroscopic Studies-The EPR spectra of all three dithionite-reduced MoFe proteins with substitutions at the Tyr ␤98 residue were identical to wild type in terms of g value, line shape, and intensity (data not shown). Similarly, the MCD spectra of the His ␤98 MoFe protein and wild-type MoFe protein were identical in both the dithionite-reduced and thionineoxidized states.

DISCUSSION
Evidence that the P cluster is the primary acceptor of electrons from the Fe protein and that it subsequently brokers the intramolecular delivery of electrons to the FeMo-cofactor can be considered in the context of the proposed structural models for the nitrogenase component proteins from A. vinelandii (Georgiadis et al., 1992;Kim and Rees, 1992). A docking model that is based on the structures of the individual component proteins (Kim and Rees, 1992;Howard, 1993) and that takes into account amino acid substitution studies (Wolle et al., 1992) and chemical cross-linking experiments has been proposed (Willing et al., 1989;Willing and Howard, 1990). This model (Fig. 1) pairs the 2-fold symmetric surface of the Fe-protein homodimer with the exposed surface of a MoFe-protein pseudosymmetric ␣␤-unit interface. In this arrangement, the Fe protein's Fe 4 S 4 cluster is positioned in the closest possible proximity to the MoFe protein's P cluster, which then lies between the Fe protein's Fe 4 S 4 cluster and FeMo-cofactor. Evidence supporting the docking model has come from biochemical and kinetic analyses of altered component proteins having one or more amino acid substitutions located within the respective The lower panel is an expanded trace (0.8 s) and shows the absorbance changes that occur after primary electron transfer. Trace c was obtained with wild-type MoFe protein and is typical of that reported previously , whereas trace d with the His ␤98 MoFe protein-dependent reaction shows a single exponential absorbance decrease (k obs ϭ 2.5 s Ϫ1 ) after primary electron transfer. docking sites (Wolle et al., 1992;Kim et al., 1993;Peters et al., 1994;Seefeldt, 1994).
In the present work, the possibility that substitutions for Tyr ␤98 might alter intramolecular electron transfer was investigated because this residue is located on a helix, which spans the P cluster and FeMo-cofactor, and yet does not directly contact either the P cluster or FeMo-cofactor (Fig. 2). Thus, the primary objective was to determine whether or not it is possible to alter intramolecular electron transfer between these prosthetic groups without disrupting either of their respective polypeptide environments. Studies of the effects of substitutions at Tyr ␤98 were carried out in parallel with identical amino acid substitutions at the corresponding residue in the ␣-subunit, Tyr ␣91 , which served as an internal control because the side chain of this residue is directed away from rather than across the direct line from the P cluster to the FeMo-cofactor. Substitutions at Tyr ␣91 were, therefore, considered much less likely to affect intramolecular electron transfer. Of the six mutant strains resulting from the substitution of either Tyr ␣91 or Tyr ␤98 by Phe, Leu, or His, only the His ␤98 -substituted strain showed a significant increase in diazotrophic growth-doubling time, which correlated with the His ␤98 MoFe protein being the only one to exhibit significantly reduced maximal specific activity for N 2 fixation, H 2 evolution, and C 2 H 2 reduction. These results were consistent with our hypothesis that the Tyr ␣91 residue would have no role in intramolecular electron transfer.
The decreased steady-state maximum activity observed for the His ␤98 MoFe protein is best explained as arising from an alteration in electron transfer capability that occurs after intermolecular electron transfer between the Fe protein and the MoFe protein because the primary and secondary rates of intermolecular electron transfer were found to be identical for both the altered His ␤98 MoFe protein and the wild-type MoFe protein (Fig. 5). Moreover, the component-protein titration experiments can be explained by a model where, under high flux conditions (i.e. high Fe protein:MoFe protein ratios), substrate reduction catalyzed by the His ␤98 MoFe protein becomes limited by intramolecular electron transfer rather than complex dissociation. In other words, the amount of Fe protein required to achieve maximum His ␤98 specific activity is lowered in the titration experiments shown in Fig. 3 because the maximum flux through the system is limited as a consequence of a defect in intramolecular electron transfer per se.
This conclusion is supported by MCD spectrophotometric analysis of the thionine-oxidized state of the altered His ␤98 MoFe protein, which was unchanged when compared with the wild type, indicating no apparent changes in the P cluster structure or its electronic environment. 3 Similarly, no perturbation of FeMo-cofactor's S ϭ 3/2 EPR spectra was observed for any of the altered dithionite-reduced MoFe proteins when compared to the wild-type spectrum. Also, none of the altered MoFe proteins having substitutions at the Tyr ␤98 position exhibited any of the characteristic substrate reduction changes associated with perturbation of the FeMo-cofactor's polypeptide environment (Scott et al., 1990(Scott et al., , 1992Kim et al., 1995; Table II). These results, together with comparable K m values for acetylene reduction for both the wild-type and His ␤98 MoFe protein, indicate that the lowered maximum specific activity for the His ␤98 MoFe protein under conditions of high flux is unlikely to occur as a result of an alteration in the substrate reduction site. However, the unusually low electron flux, plus the increased uncoupling of MgATP hydrolysis (see below) observed only under 10% C 2 H 2 with the His ␤98 MoFe protein, both of which are relieved by CO, suggests that different substrates may be served by different or multiple electron-transfer pathways.
Effective nitrogenase catalysis requires the coupled hydrolysis of about four MgATP for each pair of electrons transferred to substrate. However, MgATP hydrolysis can become partially uncoupled from electron transfer under certain conditions, such as extremely low flux (Ljones and Burris, 1972;Hageman and Burris, 1978), high or low pH (Jeng et al., 1970;Imam and Eady, 1980), and high or low temperature (Watt et al., 1975;Watt and Burns, 1977). Certain amino acid substitutions that alter either component-protein interaction (Wolle et al., 1992;Seefeldt, 1994) or the substrate reduction site (Kim et al., 1995) have also been shown to uncouple MgATP hydrolysis from electron transfer. Such uncoupling of MgATP hydrolysis from substrate reduction can be explained either by the back donation of an electron from the MoFe protein to an oxidized Fe protein, called futile cycling (Orme- Johnson and Davis, 1977) or by the occurrence of MgATP hydrolysis upon componentprotein interaction without electron transfer . A reasonable explanation for the uncoupled MgATP hydrolysis in reactions catalyzed by the His ␤98 MoFe protein (Table II) is that, as a consequence of disturbing the intramolecular electron transfer pathway, the capacity for the P cluster to accept electrons becomes saturated (after one or more rounds of component-protein interaction and intermolecular electron transfer), and any further component-protein interactions result in MgATP hydrolysis but not necessarily a net electron transfer from the Fe protein to the MoFe protein.
This explanation is consistent with the biphasic nature of the apparent component-protein dissociation rate. Here electron transfer to the P cluster of the His ␤98 MoFe protein occurs at an initial rate comparable with the wild type only until its capacity for storing electrons is reached, at which time the apparent component-protein dissociation rate becomes substantially lowered due to a defect in intramolecular electron transfer. Moreover, the gradual absorbance decrease observed after 150 ms for the His ␤98 MoFe protein contrasts dramatically with the absorbance increase observed with the wild-type MoFe protein.
Thus, according to the model of Lowe et al. (1993), it appears that P cluster oxidation within the His ␤98 MoFe protein occurs only at a relatively slow rate and is masked by the reduction of the oxidized Fe protein as it dissociates from the complex, consistent with the hypothesis that the altered MoFe protein is unable to achieve normal intramolecular electron transfer. Further, the apparent ability of the P cluster to accumulate at least two electrons is consistent with the proposed role of the P cluster as an electron storage unit and the proposed corner-tocorner disulfide link between the P cluster subfragments (Rees et al., 1993). Although the present results provide no insight as to whether the P cluster donates single electrons or electron pairs (or both) during substrate reduction, they do provide some credence to the possibility that a two-electron transfer from the P cluster to the substrate reduction site could occur during turnover.
Finally, the helix containing Tyr ␤98 has been suggested as one possible electron transfer pathway from the P cluster to FeMo-cofactor (Kim and Rees, 1992), in particular the portion from the P cluster-ligating Cys ␤95 to Tyr ␤98 and then through a hydrogen bond to homocitrate, which ligates the molybdenum atom of FeMo-cofactor. However, both the Phe ␤98 and Leu ␤98 MoFe proteins, each of which would be incapable of hydrogen bonding to homocitrate, exhibit rates of substrate reduction similar to wild type. Thus, this pathway could be viewed as unattractive and the simple interpretation invoked that neither the hydroxyl group nor the aromatic feature of the Tyr ␤98 residue is critical for productive intramolecular electron transfer. In this context, however, it should be noted that intramo-lecular electron transfer rates measured for certain other proteins (reviewed by Farid et al. (1993)) are orders of magnitude faster than the rate reported for nitrogenase component-protein dissociation (Thorneley and Lowe, 1983), the rate-limiting step in nitrogenase catalysis. Consequently, a dramatic decrease in the rate of intramolecular electron transfer might be necessary to become manifested as a lower rate of enzyme turnover. Thus, because the rate of intramolecular electron transfer within the MoFe protein cannot be directly measured, it remains premature to conclude that neither the hydroxyl group nor the aromatic nature of the Tyr ␤98 residue is involved in intramolecular electron transfer. Moreover, the significant changes in the rates of MgATP hydrolysis and catalyzed substrate reduction exhibited by the His ␤98 MoFe protein, likely resulting from an introduced structural perturbation, suggest that this helix could provide a significant electron transfer pathway. Similar studies to those described above on the residues constituting the other prosthetic group-spanning helices should provide insight into how electrons are both accommodated within and intramolecularly transferred among the redox-active moieties of the MoFe protein. Such information will be necessary to describe the complete mechanism of biological nitrogen fixation and could impact generally on our understanding of electron transfer processes in complex biological systems.