Localization of a Catalytic Intermediate Bound to the FeMo-cofactor of Nitrogenase* □ S

Nitrogenase catalyzes the biological reduction of N 2 to ammonia (nitrogen fixation) as well as the reduction of a number of alternative substrates, including acetylene (HC (cid:1) CH) to ethylene (H 2 C (cid:1) CH 2 ). It is known that the metallocluster FeMo-cofactor located within the nitrogenase MoFe protein component provides the site of substrate reduction, but the exact site where substrates bind and are reduced on the FeMo-cofactor remains un-known. We have recently shown that the (cid:2) -70 residue of the MoFe protein plays a significant role in defining substrate access to the active site; (cid:2) -70 approaches one face of the FeMo-cofactor, and when valine is substituted by alanine at this position, the substituted nitrogenase is able to accommodate a reduction of the larger alkyne propargyl alcohol (HC (cid:1) CCH 2 OH, propargyl-OH). During this reduction, a substrate-derived intermediate can be trapped on the FeMo-cofactor resulting in an S (cid:1) 1/2 spin system with a novel electron paramagnetic resonance spectrum. In the present work, trapping of the propargyl-OH-derived or propargyl amine (HC (cid:1) CCH 2 NH 2 , propargyl-

Nitrogenase is comprised of two component proteins, called the iron protein and the MoFe protein, which together catalyze the nucleotide-dependent reduction of N 2 to ammonia (Equation 1).
During catalysis, electrons are delivered one at a time from the iron protein to the MoFe protein in a reaction coupled to the hydrolysis of 2 eq of MgATP for each equivalent of electrons transferred (1,2). The MoFe protein contains two metalloclusters called the P-cluster [8Fe-7S] and the FeMo-cofactor [7Fe-9S-Mo-X-homocitrate], where X is proposed to be nitrogen, carbon, or oxygen (3). The P-clusters are thought to mediate electron transfer from the iron protein to the FeMo-cofactor, which in turn provides the site for substrate binding and reduction. The structure of the FeMo-cofactor has been elucidated from the solution of x-ray structures of MoFe proteins (3)(4)(5)(6)(7), yet where and how substrates interact with the FeMocofactor is still unknown. Different models for where substrates bind to the FeMo-cofactor have been developed; they were built on evidence from model compounds, theoretical calculations, and kinetic and biophysical studies on the wildtype (WT) 1 and genetically altered MoFe proteins (8). Some models propose binding and reduction of substrates at the molybdenum atom, whereas others suggest binding and reduction of substrates at one or more of the six iron atoms that constitute the central portion of the FeMo-cofactor. Models have also been proposed that involve substrate binding of both molybdenum and iron at different steps during the reduction reaction (9 -11).
Recently, we have pursued genetic and biophysical approaches on nitrogenase to localize the substrate binding site on the FeMo-cofactor (12)(13)(14)(15)(16). It has been demonstrated that substitution of ␣-70 Val , a residue that approaches one Fe 4 S 4 face of the FeMo-cofactor (involving iron atoms 2, 3, 6, and 7), by amino acids with smaller side chains expands the substrate specificity to include larger alkynes (12,14,15). For example, substitution of ␣-70 Val by alanine has been shown to expand the substrate range of nitrogenase to include propyne (HCϵCCH 3 ) or propargyl alcohol (HCϵCCH 2 OH, propargyl-OH) (14). When the ␣-70 Ala MoFe protein is freeze-trapped during the reduction of propargyl-OH, a reduction intermediate bound to the FeMo-cofactor is captured (15). Using 13 C-and 1/2 H-labeled propargyl-OH and electron nuclear double resonance spectroscopic methods, we have recently deduced that the trapped intermediate has two hydrogen atoms added (i.e. allyl alcohol, H 2 CϭCHCH 2 OH) and is bound to iron such that the two terminal hydrogen atoms are spectroscopically indis-tinguishable, suggesting the bio-organometallic complex shown in Scheme 1 (17).
In the present work, a model is developed from studies of nitrogenase for binding of a propargyl-OH reduction intermediate as well as binding of a propargyl-NH 2 reduction intermediate to a specific iron atom within the FeMo-cofactor.

EXPERIMENTAL PROCEDURES
Protein Purification and Activity Assays-Azotobacter vinelandii strains DJ1310 and DJ1316 expressing the ␣-70 Ala and ␣-70 Ala /␣-195 Gln variant MoFe proteins, respectively, were constructed using site-directed mutagenesis and gene replacement techniques as described previously (12,13). The ␣-70 Ala and ␣-70 Ala /␣-195 Gln variant MoFe proteins were purified using a poly(His)-metal affinity chromatography system described earlier (18). The wild-type iron protein component of nitrogenase was purified essentially as described previously (19). All manipulations of proteins were conducted in septum-sealed serum vials under an argon atmosphere, and all anaerobic liquid and gas transfers were performed using gas-tight syringes. Acetylene reduction, H 2 evolution, and N 2 reduction activities were determined as described earlier (20,21). NH 3 was quantified using a liquid chromatographic fluorescence method with o-phthalaldehyde mercaptoethanol as described previously (22). Thirty l of an assay reaction producing NH 3 was added to 1 ml of a solution containing 19 mM phthalic dicarboxyaldehyde, 3.4 mM 2-mercaptoethanol, 5% (v/v) ethanol, and 190 mM potassium phosphate, pH 7.3, and allowed to react in the dark for 30 min. The mixture was injected and separated on a C-18 guard column and detected by fluorescence ( excitation / emission of 410/472 nm). The NH 3 quantification was standardized with NH 3 Cl. The K i for propargyl-OH was estimated from the apparent affinity for acetylene (K m(app) ) in the presence of varying propargyl-OH concentrations and then plotting K m(app) versus propargyl-OH concentration (23).
Preparation of Non-turnover and Turnover MoFe Protein Electron Paramagnetic Resonance Samples-Non-turnover (resting state) MoFe protein (100 M) samples were made in 100 mM MOPS buffer, pH 7.0, with 30 mM sodium dithionite (Na 2 S 2 O 4 ) under 1 atm of argon. Nitrogenase turnover samples were prepared by the addition of the MoFe protein (100 M) to a buffer mixture (100 mM MES, 100 mM MOPS, 100 mM Tris, and 100 mM TAPS) at defined pH values and including 30 mM sodium dithionite, 10 mM ATP, 15 mM MgCl 2 , 20 mM phosphocreatine, 2 mg/ml bovine serum albumin, and 0.3 mg/ml creatine phosphokinase. The reaction was initiated by the addition of iron protein (50 M) and was allowed to react for ϳ20 s at room temperature before being frozen in EPR tubes in liquid nitrogen. Proteins were added from a concentrated stock solution and did not significantly alter the reaction pH. Where appropriate, propargyl-OH (3 mM) or propargyl-NH 2 (15 mM) was included in the initial reaction mixture. Standardized 4-mm quartz EPR tubes were used for all samples.
EPR Spectroscopy-X-band EPR spectra were recorded on a Bruker ESP-300 E spectrometer equipped with an ER 4116 dual-mode X-band cavity including an Oxford Instruments ESR-900 helium flow cryostat. EPR spectra were recorded at a modulation frequency of 100 kHz, a modulation amplitude of 1.26 millitesla (12.6 gauss), a sweep rate of 10 millitesla/s, and a microwave frequency of ϳ9.65 GHz (with the precise value recorded for each spectrum to ensure exact g alignment). All spectra were recorded at 8000 and a microwave power of 2.0 mW with each trace being the sum of five scans. The software program IGOR Pro (WaveMetrics, Lake Oswego, OR) was used for all subsequent manipulation of spectral data.
Theoretical Calculations-Spin-unrestricted all-electron density functional calculations of the FeMo-cofactor with bound intermediates used the blyp functional with numerical basis sets as implemented in the software program DMol3 (24 -26). Calculations of the FeMo-cofactor with bound intermediates within the protein involved a large protein component composed of 1032 amino acid residues, 1332 associated water molecules, and the P-cluster together with the FeMo-cofactor. This large protein component was selected as all of chains A and B of 1M1N (3) together with residues 494 to 523 of chain D and all water molecules within 4 Å of these atoms or the FeMo-cofactor. All hydrogen atoms were added and energy-minimized by force field methods (force field cvff, program Discover) to optimize the protein and water hydrogen bonding. Structures from density functional calculations of the intermediates bound to the FeMo-cofactor were substituted into the protein and investigated by further force field optimizations. The Pcluster and the N cen Fe 7 MoS 9 O 2 atoms of the FeMo-cofactor plus bound intermediate were fixed during all force field calculations.

RESULTS AND DISCUSSION
Relevant Features of the ␣-70 Ala -substituted MoFe Protein-Substitution of the MoFe protein ␣-70 Val residue by alanine expands the substrate range for nitrogenase to include the short chain alkynes propyne (HCϵCCH 3 ) and propargyl alcohol (14). When propargyl-OH is used as a nitrogenase substrate and freeze-quenched under turnover conditions, a paramagnetic intermediate is observed that results from conversion of the resting state S ϭ 3/2 spin system ( Fig. 1, trace 1) to an S ϭ 1/2 spin system having a rhombic EPR signal with g values of 2.123, 1.998, and 1.986 ( Fig. 1, trace 4) (15). This new EPRactive state originates from an intermediate derived from propargyl-OH that is bound to the FeMo-cofactor. Electron nuclear double resonance studies coupled with the use of isotopically labeled propargyl-OH have established that allyl alcohol, CH 2 ϭCH-CH 2 OH, is the probable species bound (Scheme 1) (17).
In contrast to propargyl-OH, when propyne is used as the substrate, freeze-quenching under turnover conditions does not result in the formation of a trapped adduct that can be observed by EPR. An obvious explanation for this difference is that the -OH group of propargyl-OH stabilizes the bound intermediate through hydrogen bonding interactions with a functional group provided by an amino acid located within the vicinity of the substrate binding site. Identification of the proposed functional group thus provided an unprecedented opportunity to identify where and how an alkyne substrate might interact with the nitrogenase active site. Toward this end, the environment around ␣-70 Val and the FeMo-cofactor in the resting state x-ray structure of the MoFe protein (3) was examined and the imidazole of ␣-195 His identified as the most likely candidate to provide the hydrogen bonding interaction with propargyl-OH. This possibility suggested two predictions that could be experimentally tested. First, if a propargyl-OH reduction intermediate is stabilized by a hydrogen bond interaction with the imidazole group of ␣-195 His , such stabilization should be pHdependent having a pK a value near that expected for a histidine residue. Second, an ability to trap the EPR-active intermediate should be dependent on having a histidine residue at the ␣-195 residue position.
Appearance of the Freeze-trapped Propargyl-OH Reduction Intermediate Is pH-dependent- Fig. 1 illustrates the pH dependence of the intensity of the propargyl-OH-derived EPR signal. The maximal EPR signal intensity, which is interpreted to indicate the maximal concentration of the trapped adduct, is observed at a pH ϳ6.7. As the pH value increases from 6.7 to 8.7, the intensity of the EPR signal steadily decreases (Fig. 1,  traces 4 and 5). No significant changes in EPR lineshape are observed over this pH range, indicating a simple depopulation of the intermediate bound state as pH rises. A more complete data set for the intensity of the propargyl-OH-dependent EPR signal versus pH is presented in Fig. 2, also illustrating a depopulation of the trapped adduct as the pH value increases from 6.7 to 9.0. This result is consistent with the deprotonation of a group having a pK a ϳ7.5 and the fact that a protonated state is required for formation of a hydrogen bond necessary to elicit the EPR signal associated with the trapped species. In proteins, histidine residues can have pK a values ranging from 5 to 9, depending on the protein environment (27). The ⑀NH ϩ portion of ␣-195 His is surrounded by the hydrophobic residues ␣-65 Ala , ␣-66 Gly , ␣-70 Ala , ␣-71 Val , and ␣-381 Phe together with the negatively charged sulfur atoms of the FeMo-cofactor, generating a net negative electrostatic potential. This environment would be predicted to increase the pK a of the imidazole of ␣-195 His (28). Thus, one explanation of the results is that the protonated state of the ⑀NH ϩ of ␣-195 His acts as a hydrogen bond donor to a non-bonded electron pair of the oxygen of bound allyl-OH (Fig. 3, panel A). As the ⑀N of ␣-195 His becomes deprotonated (with an observed pK a of 7.5), the hydrogen bond with the oxygen of allyl-OH would greatly diminish.
The pH dependence for the formation of the propargyl-OH trapped intermediate can be contrasted with the previously reported pH dependence for acetylene reduction by nitrogenase. In the latter case, maximal activity is around pH 7.3 with a decline in activity at pH values higher and lower than pH 7.3 (29). The pH optimum value of 6.7 observed here for the formation of the propargyl-OH-trapped species is significantly lower than the pH optimum value of 7.3 for activity, suggesting that a specific deprotonation event is controlling the trapping of the intermediate.
Appearance of a Propargyl-NH 2 Reduction Intermediate Is Also pH-dependent-More convincing evidence that the appearance of the trapped EPR-active species is the result of specific hydrogen bonding between a reduction intermediate and an active site residue (rather than a nonspecific effect on enzyme activity) was obtained by using propargyl amine (propargyl-NH 2 ) as a substrate instead of propargyl-OH. If the appearance of the trapped species derived from propargyl-OH is dependent on hydrogen bonding provided by the protonated imidazole group of ␣-195 His , it is predicted that the appearance of a trapped species when propargyl-NH 2 is used as substrate would require a deprotonated imidazole group (Fig. 3B) because the pK a of propargyl-NH 2 is higher than that of imidazole. In control experiments, it was shown that, like propargyl-OH, propargyl-NH 2 is an inhibitor of acetylene reduction (50% inhibition observed by inclusion of 20 mM propargyl-NH 2 in an assay with 0.003 atm of acetylene at pH 8.0) and therefore interacts with the active site. When the ␣-70 Ala MoFe protein is freeze-quenched during turnover using propargyl-NH 2 as substrate, an EPR-active species is also detected with the same lineshape as the propargyl-OH-derived species with apparent g values of 2.12, 2.00, and 2.00 (Fig. 4). Given the similarity in lineshape of the EPR spectra elicited under freeze-quench conditions when either propargyl-OH or propargyl-NH 2 is used as substrate, it is presumed that an allyl-NH 2 adduct is bound to an iron atom of the FeMo-cofactor in a way similar to that proposed for propargyl-OH (17). One difference in behavior between these two substrates is the lower intensity of the propargyl-NH 2 (15 mM)-elicited EPR signal (approximately one-fourth) when compared with propargyl-OH (3 mM)-dependent EPR signal intensity. Another important difference is the pH profile for the formation of the respective intermediates (Fig. 2). As can be seen, the population of the trapped propargyl-NH 2 reduction intermediate EPR-active species increases with rising pH (maximizing at about pH 8.2) followed by a rapid decline. The important observation is that the pH required for maximizing the propargyl-NH 2 -elicited EPR signal is significantly shifted to a higher pH than required for maximizing the propargyl-OH-elicited EPR signal. This feature is in line with a requirement for a protonated imidazole group for hydrogen bonding to propargyl-OH and a deprotonated imidazole group for hydrogen bonding to propargyl-NH 2 (Fig. 3). The rapid decline in population of the trapped species when propargyl-NH 2 is used as substrate can be explained by the deprotonation of both the imidazole group of ␣-195 His and the amino group of propargyl-NH 2 (30). Taken together, the different pH dependences required to populate intermediate states when either propargyl-OH or propargyl-NH 2 is used as substrate provides a compelling case for the protonation or deprotonation of an imidazole group of ␣-195 His as controlling the stabilization of bound intermediates by hydrogen bonding interactions.
The Imidazole Group of ␣-195 His Is Required for Intermediate Stabilization-Previous work has shown that substitution of the MoFe protein ␣-195 His residue by glutamine results in an altered MoFe protein that can bind N 2 as effectively as wildtype MoFe protein but is not able to effectively reduce N 2 (Ͻ2% N 2 reduction) (31)(32)(33). In contrast, the ␣-195 Gln -substituted MoFe protein is unaffected in its ability to reduce acetylene or protons (32,33). The explanation offered for these features is that the imidazole group of ␣-195 His is specifically required to stabilize an N 2 reduction intermediate or is required as a proton donor for N 2 reduction (32,33). These observations taken together indicate that ␣-195 His is the most likely candidate for providing the ionizable group responsible for stabilizing propargyl-OH or propargyl-NH 2 intermediates. If ␣-195 His is responsible for hydrogen bonding interactions necessary to stabilize a propargyl-OH reduction intermediate, such an interaction should be lost by substitution of glutamine for the MoFe protein ␣-195 His residue. This prediction was tested by the construction and characterization of a doubly substituted MoFe protein where ␣-195 His is substituted by glutamine and ␣-70 Val is substituted by alanine. The ␣-70 Ala /␣-195 Gln doubly substituted MoFe protein retains Ͼ65% of the wild-type acetylene and proton reduction activities (Table I), indicating no major disruption in the binding or reduction of these substrates. Like the ␣-195 Gln MoFe protein, the doubly substituted MoFe protein shows very low N 2 reduction activity (Ͻ1%). Further, propargyl-OH remains an effective inhibitor of acetylene reduction for the doubly substituted MoFe protein with a K i of 8 mM compared with the published value of 4 mM for the ␣-70 Ala MoFe protein (14), clearly indicating that propargyl-OH continues to associate with the doubly substituted MoFe protein. However, no EPR-detectable adduct is observed when the ␣-70 Ala /␣-195 Gln MoFe protein is freeze-quenched during turnover when either propargyl-OH or propagyl-NH 2 is used as substrate (data are shown in Fig. 2 for propargyl-OH).
Refined Identification of the Substrate Binding Site by Density Functional and Force Field Calculations-The above results all point to the formation of a hydrogen bond between the -OH or -NH 2 group of the propargyl-OH-or propargyl-NH 2bound reduction intermediate and the imidazole of ␣-195 His , thereby localizing the position of the -OH or -NH 2 groups to be within ϳ2 Å of the ⑀N of ␣-195 His . Further, electron nuclear double resonance spectroscopic characterization (17) has indicated that the CϭC portion of allyl-OH is bound to a single iron atom in an 2 configuration (Scheme 1). With these constraints, the location of the bound intermediate is largely defined. To further define the likely binding site, theoretical calculations were used. The strategy employed density functional methods to elucidate the detailed geometry of bonding of the intermediate to the FeMo-cofactor and then to test the fit into the ␣-70 Ala protein using force field methods with all hydrogen atoms explicitly included. The density functional calculations were made on a model that includes the essential coordination features of the FeMo-cofactor, namely Fe 7 MoS 9 N cen (SCH 3 )-(OCH 2 COO)(C 3 N 2 H 4 ), with net charge Ϫ3 corresponding to the resting state (34). The resulting structures with bound intermediate were then substituted (in silico) for FeMo-cofactor in the protein with ␣-70 Ala and relaxed to assess their ability to meet two criteria, which are hydrogen bonding with ⑀N of ␣-195 His and accommodation by the ␣-70 Ala protein but not the wild-type ␣-70 Val protein.
From these calculations, the best binding modes for allyl-OH and allyl-NH 2 were deduced, where the alkene portion is bound 2 at Fe-6 in a position that is closer to exo than endo. For both allyl-OH and allyl-NH 3 ϩ , a good H-bond is formed with the ⑀N of ␣-195 His and with a -S (S2B) of FeMo-cofactor (Fig. 5). Normal van der Waals contact occurs between the methyl group of ␣-70 Ala and the bound intermediates, but as expected there is impossible conflict with the side chain of ␣-70 Val in the wild- type MoFe protein. Binding in a similar fashion to Fe-2, Fe-3, and Fe-7 was significantly less favorable compared with binding to Fe-6 either by not satisfying the need for an H-bond between the -OH or -NH 2 and the imidazole of ␣-195 or by resulting in van der Waals collisions of the bound adduct with the surrounding protein.
These results define in detail the location and geometry of the bound allyl-OH and allyl-NH 3 ϩ intermediates in the substituted ␣-70 Ala MoFe protein. Calculations using the same methods and strategy as described above were made for C 2 H 2 and C 2 H 4 as substrate and product, respectively, without the alanine substitution for the ␣-70 Val residue. These calculations reveal that the same 2 coordination geometry of propargyl-OH and propargyl-NH 2 is possible for C 2 H 2 /C 2 H 4 at Fe-6 in the wild-type MoFe protein.
Mechanistic Implications-Experiments described here have established a requirement for the imidazole group of ␣-195 His to elicit a characteristic EPR spectrum under freeze-quench conditions when either propargyl-OH or propargyl-NH 2 is used as substrates for the ␣-70 Ala -substituted MoFe protein and show that development of these respective spectra are dependent upon and differentiated by the pH of the reaction conditions. These results confine the substrate reduction site for these substrates within hydrogen bonding distance of the ⑀N of ␣-195 His . The binding site was further refined using density functional and force field calculations and by consideration of the binding geometry to iron as indicated by electron nuclear double resonance experiments. Although the binding of propargyl-OH and propargyl-NH 2 to Fe-6 within the FeMo-cofactor is strongly favored by the present work, they do not preclude the possibility that alkynes might also bind to other sites and in different ways. Indeed, there is abundant evidence indicating that there is more than one acetylene-binding site located within the MoFe protein (35). Therefore it is important to keep in mind that the trapped species recognized by freeze-quench EPR analyses reported here and elsewhere does not necessarily represent the only binding site that is possible even for propargyl-OH and propargyl-NH 2 . Rather, it is one binding site that can be observed and defined as a consequence of stabilization of a reduction intermediate. The important mechanistic implication is that it provides the first experimentally supported evidence for both the location and geometry for binding of any nitrogenase substrate.
Work described here provides no direct experimental information concerning the location and geometry of alkyne binding to FeMo-cofactor in relation to the binding of the physiological substrate N 2 . Nevertheless, it is striking that a completely independent approach was previously used to gain evidence that the MoFe protein ␣-195 His residue has an important role in controlling N 2 reduction (31,32), probably through the stabilization of a reduction intermediate or as a proton donor during the catalytic process. In our view, this observation is compatible with the current work that shows ␣-195 His is also required to trap an alkyne reduction intermediate that can be observed by EPR spectroscopy. Although this information can be interpreted in a variety of ways, one reasonable interpretation is that an N 2 reduction intermediate binds at the same place and in the same way as determined here for propargyl-OH and propargyl-NH 2 . In this context there is ample evidence in the chemical literature for metal-hydrazine complexes having the same configuration as the metallacyclopropane ring structure proposed for alkyne reduction intermediates (36). Future work will focus on testing whether or not N 2 or its reduction intermediates can be trapped and characterized using the genetic, biochemical, and biophysical approaches described for characterizing alkyne reduction intermediates. a Specific activities are reported in units of nmol of product/min/mg of MoFe protein and were determined with an iron protein:MoFe protein molar ratio of 40:1.
b % WT were calculated by a ratio of the activity of the altered MoFe protein to the activity of wild-type.