Nitrogenase: standing at the crossroads

  • Published on

  • View

  • Download


  • 559

    Nitrogenase catalyzes the ATP-dependent reduction ofdinitrogen to ammonia, which is central to the process ofbiological nitrogen fixation. Recent progress towardsestablishing the mechanism of action of this complexmetalloenzyme reflects the contributions of a combination ofstructural, biochemical, spectroscopic, synthetic andtheoretical approaches to a challenging problem withimplications for a range of biochemical and chemical systems.

    Addresses*Howard Hughes Medical Institute, Division of Chemistry andChemical Engineering, 147-75CH, California Institute of Technology,Pasadena, CA 91125, USA; e-mail: dcrees@caltech.eduDepartment of Biochemistry, University of Minnesota, Minneapolis,MN 55455, USA; e-mail:

    Current Opinion in Chemical Biology 2000, 4:559566

    1367-5931/00/$ see front matter 2000 Elsevier Science Ltd. All rights reserved.

    AbbreviationEPR electron paramagnetic resonance

    IntroductionRationality notwithstanding, anyone studying nitrogenaseshould be excused for occasionally wondering whether aFaustian bargain might be required to establish the mecha-nism of dinitrogen reduction by this enzyme. The obstaclesto studying nitrogenase are multiple and varied: theyinclude the extreme oxygen lability of the proteins; the sizeand complexity of the ironsulfur containing metallocen-ters; the similar and nondescript visible spectroscopicproperties of multiple clusters, precluding facile monitoringof changes in individual groups; the absence of model com-pounds to evaluate potential mechanistically relevant statesand intermediates; and the kinetic complexities of a multi-substrate, multi-component enzyme system with differentsimultaneously populated states. The contrast betweenstudies on nitrogenase and metalloenzymes that react withdioxygen is informative; for enzymes reacting with dioxy-gen, the interplay between enzymology and chemicalstudies has led to the identification of key intermediatesand a bountiful stream of model compounds that provide asecure chemical foundation for a detailed mechanisticunderstanding. Nevertheless, there are grounds for opti-mism that nitrogenase may enjoy similar levels of insightbased upon the increasingly higher resolution structures ofindividual proteins and their varied complexes, augmentedby genetic manipulation of the proteins and by spectro-scopic studies. Together with developments in thechemistry of relevant metalloclusters and in the theoreticalmodeling of these clusters, we may yet be able to peerinside the nitrogenase black box and achieve a detailedmechanistic description. Following a brief overview, thisreview highlights outstanding issues in our understanding

    of nitrogenase. Due to space limitations, only recent orselected references can be cited.

    Introduction to nitrogenaseNitrogenase is a two protein component system that cat-alyzes the reduction of dinitrogen to ammonia coupled tothe hydrolysis of ATP (reviewed in [15]). Rather remark-ably, even after 35 years of study, the overall reactionstoichiometry is still not unambiguously determined. Theuncertainties are expressed in the following equation forthe overall enzyme reaction:

    N2 + (6+2n)H+ + (6+2n)e + p(6+2n)ATP 2NH3 + nH2 + p(6+2n)ADP + p(6+2n)Pi

    In the standard model, the evolution of one molecule ofdihydrogen is coupled to the reduction of one molecule ofdinitrogen, and two molecules of ATP are hydrolyzed perelectron transferred, so that n = 1 and p = 2. However,there has not been a compelling demonstration of anobligatory mechanistic coupling of dihydrogen evolutionand dinitrogen reduction, although dihydrogen is alwaysobserved as a product with ammonia, and a limiting ratioof 2 ammonia to 1 dihydrogen is obtained at high dinitro-gen gas pressure [6]. Because nitrogenase can also reduceprotons to dihydrogen in the absence of dinitrogen, thestoichiometry of the observed product distribution couldrepresent the outcome of two parallel paths from a com-mon branch in the electron transfer process. Additionally,while p = 2 represents the apparent limiting stoichiometryof ATP hydrolyzed per electron transferred, under manyconditions the coupling is much less efficient and ratios >2are observed. There are also reports of ratios approaching1 when an all-ferrous form of Fe-protein is used as theelectron source (see below).

    The most extensively studied form of nitrogenase is themolybdenum-containing system that consists of two com-ponent metalloproteins, the molybdenumiron (MoFe-)protein and the iron (Fe-) protein, where the Fe-protein isthe nucleotide-binding and electron-donating component,and the MoFe-protein contains the substrate-reducingsite. Alternate nitrogenases exist that are homologous tothis system, but with the molybdenum apparently substi-tuted by vanadium or iron [7]. We now have availableX-ray crystallographic structures for the Fe-protein [810]and MoFe-protein [1113,14] as well as for two complexesbetween the two proteins [15,16]. The Fe-protein is adimer of two identical subunits that symmetrically coordi-nates a single [4Fe:4S] cluster. The isolated Fe-protein canbind MgADP or MgATP at a stoichiometry of twonucleotides per dimer, and participates intimately in thecoupling between ATP hydrolysis and electron transfer tothe MoFe-protein. Structurally, Fe-protein adopts a

    Nitrogenase: standing at the crossroadsDouglas C Rees* and James B Howard

  • polypeptide fold characteristic of P-loop-containingnucleotide-binding proteins such as observed for ras,G-proteins and related proteins [17,18]. The MoFe-protein is an 22 heterotetramer, where the and sub-units exhibit similar polypeptide folds consisting of threedomains of the /-type, with some extra helices. Thisprotein contains two copies each of two different types ofunusual metallocenters (Figure 1): the FeMo-cofactor, thelikely substrate reduction site; and the P-cluster, which is

    believed to participate in electron transfer from theFe-protein to the FeMo-cofactor. Each cluster containseight metals and associated sulfurs that are distinctivelyarranged in ways that have been neither observed in anyother enzymes nor modeled synthetically. These metallo-clusters are coordinated by ligands contributed bydifferent domains of the MoFe-protein that are present ina common core composed of a four-stranded, parallel-sheet flanked by -helices (Figure 2); a similar structural

    560 Mechanisms

    Figure 1

    Structural models for the nitrogenasemetalloclusters and coordinating ligands.(a) FeMo-cofactor. (b) The dithionite reducedPN form of the P-cluster. (c) The oxidized POXstate of the P-cluster. Protein Data Bank[57,58] coordinate sets 3MIN and 2MIN wereused for (a), (b), and (c), respectively. Allfigures in this review were prepared withMOLSCRIPT [59].

    Cys b 95 Cys b 95

    Cys a 62 Cys a 62

    Cys a 88Cys a 88


    Cys b 70Cys b 70



    Fe3 Fe1 Fe1

    Cys a 275









    His a 442








    Ser b 188 Ser b 188

    Cys a 154Cys a 154

    Cys b 153 Cys b 153

    (a) (b) (c)

    Current Opinion in Chemical Biology

    Figure 2

    Ribbon diagrams of a common domain coreobserved to coordinate complexmetalloclusters in the nitrogenaseMoFe-protein and iron-only hydrogenases.This core consists of a four-stranded, parallel-sheet surrounded by -helices, with strandorder 2134, going from top to bottom asdepicted in this figure. The metalloclustersinteract with the carboxy-terminal end of thesheet, near the crossover position [60]between strands 1 and 3. Cluster ligandsare positioned in the loop between strands 3and 4, and immediately after loop 4. (a) TheFeMo-cofactor and MoFe-protein -subunitdomain 3 (residues 350 to 442). (b) TheFeMo-cofactor and MoFe-protein -subunitdomain 2 (residues 222 to 297). (c) TheP-cluster and MoFe-protein -subunitdomain 1 (residues 85 to 188). Similarinteractions are also observed between theP-cluster and -subunit domain 2. (d) TheH-cluster and residues 224 to 381 of theiron-only hydrogenase. Residues 269 to334 loop away from this region and havebeen excluded. Protein Data Bank coordinatesets 3MIN and 1FEH were used for (ac),and (d), respectively.





    Current Opinion in Chemical Biology

  • organization is also observed in the coordination of theH-cluster at the active site of iron-only hydrogenases [19].

    Kinetics and mechanism of substrate reductionAt the protein level, the basic mechanism of nitrogenaseinvolves four steps: first, formation of a complex betweenthe reduced Fe-protein with two bound ATP moleculesand the MoFe-protein; second, electron transfer betweenthe two proteins coupled to the hydrolysis of ATP; third,dissociation of the Fe-protein accompanied by re-reductionand exchange of ATP for ADP; and finally, repetition of thiscycle until sufficient numbers of electrons (and protons)have been accumulated so that available substrates can bereduced. The relative positions of the metalloclustersobserved in the structures of complexes of the nitrogenaseproteins [15] indicate that electron transfer from theFe-protein to the FeMo-cofactor proceeds through theP-clusters (Figure 3); the edge-to-edge distance of ~14 iscompatible with inter-protein electron transfer rates morerapid than the observed turnover time [20,21].

    Thorneley and Lowe [22] have attempted to quantify thekinetically identifiable steps of the nitrogenase mechanism.Although modifications are necessary [5,23], especially withregard to the generality of dithionite as a reductant, signifi-cant features of the nitrogenase mechanism embodied inthis scheme can be stated: under saturating conditions ofFe-protein, dissociation of the Fe-proteinMoFe-proteincomplex represents the overall rate limiting step, with aturnover time of ~5 s1; ATP hydrolysis only occurs in thecomplex of the two nitrogenase proteins; substrates onlybind to the reduced forms of MoFe-protein that are exclu-sively generated by Fe-protein and ATP; patterns ofcompetition between various substrates can be understoodto reflect binding to different oxidation states of MoFe-protein; and CO inhibits the reduction of all substratesexcept for proton reduction. It should be noted that thestates of MoFe-protein leading to substrate reduction havenot yet been produced either by coupling to other bio-chemical enzyme reducing systems or by electrochemicalreduction. Hence, the role of Fe-protein with ATP hydrol-ysis appears to be more than a simple electron donor and islikely to provide some larger conformational or structuralcontribution to the overall process.

    Beyond this kinetic analysis, a more detailed moleculardescription of the electron transfer processes has notbeen obtained. Missing from our picture of nitrogenaseduring turnover are the mechanistic details of such fun-damental processes as the relationship between ATPhydrolysis and inter-protein electron transfer; experimen-tal evidence concerning the path of electron transferbetween Fe-protein and MoFe-protein; and everyonesfavorite question where and how do substrates interactwith the FeMo-cofactor? Progress towards these goals isbeing realized through multiple approaches, includingthe use of mutant nitrogenases, alternate substrates,inhibitors, spectroscopic studies, model chemistry and

    theoretical approaches. While space limitations precludea detailed discussion, a few selected examples serve toillustrate the potentials of these approaches as they arenow being applied to nitrogen fixation.

    Mutants have been instrumental in many systems for dis-secting contributions of specific residues to the catalyticmechanism of enzymes. In the case of nitrogenase, site-directed mutagenesis studies have identified substitutionsof residues around the FeMo-cofactor (Figure 4) that havealtered substrate reduction properties. Among the bestcharacterized are replacements of residue His 195 [24,25],which donates a hydrogen bond to one of the bridging sul-furs of the cofactor. Replacement of this histidine with

    Nitrogenase: standing at the crossroads Rees and Howard 561

    Figure 3

    A slice through the ADPAlF4-stabilized nitrogenase complex [15] thatincludes the ADPAlF4, [4Fe:4S] cluster, P-cluster, and FeMo-cofactor,illustrating the relationship between the nucleotide-binding and electrontransfer sites. Relative to the isolated Fe-protein, the change inconformation of the switch II region serves to reposition the Asp129sidechains for nucleotide hydrolysis and also moves the [4Fe:4S] clustercloser to the P-cluster. The linkage of these conformational changesthrough the switch II region provides the structural basis for the couplingof ATP hydrolysis and electron transfer by nitrogenase [55]. This figurewas prepared from Protein Data Bank coordinate set 1N2C.



    [4Fe:4S] cluster10


    Switch II



    b -subunit a -subunit



    Current Opinion in Chemical Biology

  • glutamine results in an enzyme that can still reduce acety-lene to ethylene yet cannot reduce dinitrogen. However,dinitrogen can bind to the enzyme and is an inhibitor ofacetylene reduction. These observations begin to separatethe components of the enzyme necessary for substratebinding from those necessary for electron and proton addi-tion to the substrate. This clearly shows that dinitrogenbinding to the FeMo-cofactor is necessary, but not suffi-cient, for catalysis. Other substitutions at this position resultin enzymes that catalyze a four-electron reduction of acety-lene to ethane, an activity that is not exhibited by thewild-type enzyme [26]. An exciting recent development isthe application of microbial selection methods [27] toidentify mutated forms of nitrogenase with altered sub-strate reduction properties. Through this approach, theGly 69Ser variant of the MoFe-protein was identified; itcan fix dinitrogen, but has significantly lower acetylenereduction efficiency.

    Although it is commonly assumed that substrates andinhibitors are binding directly to a metallocluster, mostlikely the FeMo-cofactor, direct demonstration of this hasbeen notoriously difficult, and has still not been achievedfor N2. Hoffman, Hales and co-workers [28] have beenable to demonstrate that CO can bind to the FeMo-cofac-tor, a process that has also been monitored by stopped-flowIR spectroscopy [29]. More recently, the Seefeldt and

    Hoffman groups [30] have identified electron paramagneticresonance (EPR) signals from bound CS2. Although thespecies represented by these signals account for only a fewpercent of the total FeMo-cofactor present, raising ques-tions about the mechanistic relevance of these states, theseobservations give hope that more detailed characterizationof intermediates will be possible using spectroscopicapproaches such as ENDOR and other EPR techniques,IR, resonance Raman and Mssbauer methods.

    Model compounds provide the opportunity to evaluate thestructural and reactivity properties of relevant systems ingreater detail, and are becoming increasingly realistic in cap-turing aspects of the metalloclusters of the MoFe-protein, as de...