J. CHEM. SOC. DALTON TRANS. 1995Dalton Perspectives503Inorganic Reaction Mechanisms:the Bioinorganic-Organometallic InterfaceRichard A. HendersonNitrogen Fixation Laboratory, University of Sussex, Brighton BN I 9RQ, UKBioinorganic and organometallic chemistry are often consid-ered to represent the extreme ends of the spectrum of inorganicchemistry. This division is a consequence primarily of thestructural aspects associated with each area, the type of ligandsinvolved, a perception of the types of oxidation states of themetals in the two systems and the methodology employed tostudy the materials. However, the apparent barrier betweenthese two disciplines start9 to disappear when we considerreactivity and mechanistic chemistry. After all, a substitutionreaction is a substitution reaction, whether it’s in a cobalt(Ir1)complex, such as [COCI(NH,),]~ + , a manganese(0) complexsuch as [Mn(q5-C,Hs)(CO)t (thf)] (thf = tetrahydrofuran) ora zinc(1r) centre surrounded by a polypeptide, as in carboxy-peptidase.The same set of basic mechanistic principles apply tothe reactivity irrespective of the system.This article will illustrate how studies designed to understandthe reactivity of metalloenzymes can lead to new chemistry, orgive insights into contemporary chemical issues which are farremoved from any biological problem.The Development of Inorganic Reaction MechanismsBy the 1950s the mechanistic basis of organic chemistry was welladvanced and thus it was a natural consequence that theprinciples established there should be applied to the study of thereactions of inorganic complexes.Prior to this period isolatedsuggestions about the mechanisms of inorganic reactions hadbeen made, but it is only from the middle of the twentiethcentury that a methodical, systematic approach was under-taken. The development of a Periodic Table-wide comprehensivepicture of reaction mechanisms is a gargantuan task because ofthe large variety of oxidation states, co-ordination numbers andgeometries associated with the elements. In the early days thedevelopment of the principles of inorganic reaction mechanismswas dominated by studies on classical co-ordination com-pounds. In particular, ‘Werner type’ octahedral cobalt(m)amine complexes and square-planar platinum(I1) complexes,with an emphasis on substitution and redox reactions. Sub-sequently, the field flourished so that the substitution andelectron-transfer reactivity patterns of many metals in a varietyof different co-ordination environments are now established.Included in this expansion came studies on organometallicsystems, which exhibited reactions not observed with classicalco-ordination compounds.Certainly the basic mechanisms of all the fundamentalelementary reactions including substitution, electron transfer,insertion, oxidative addition and reductive elimination areestablished.The purist would probably argue that there is stilla great deal to be done even in defining the mechanisms ofsubstitution or redox reactions.This is a point of view withwhich I have little quarrel. What I do advocate, though, is thatwe look beyond the approach that has been adopted so far. Ifwe assess the role of reaction mechanisms critically, it must beconcluded that understanding the mechanism of any one par-ticular reaction is of only limited value. At the very least,understanding the mechanism of a reaction must complementsynthetic and structural chemistry. At best the results of amechanistic study not only define how a reaction proceeds atthe atomic level, but can also be employed to help in thepreparation of new complexes, rationalise the previously un-explained behaviour of analogous complexes, or even open up anew, and previously unexplored, area of chemistry. At worst, amechanistic study only adds another rate constant (or severalnumbers if the activation parameters are also determined),associated with a reaction type we already understand in greatdetail! There has been an overemphasis in the past to discussinorganic reaction mechanisms in terms of a particular oxi-dation state of a metal, an electron configuration, or a particulargeometry.This approach owes much to the developmentalyears where researchers were defining the reactions of octa-hedral cobalt(rI1) complexes (say). It is too limiting: we need tolook for more general principles, those which are of suchgeneral utility that there is no hesitation in applying them tomore complex, multistep processes as are found in the action ofhomogeneous catalysts and metalloenzymes. It is just this typeof approach upon which I will elaborate.Mechanisms: Bioinorganic and OrganometallicChemistryIn the last couple of decades there has been a move in inorganicreaction mechanisms towards studying bioinorganic systems.2This is an area where we should carefully assess what theinorganic mechanist can usefully contribute.Undoubtedly someexcellent work is being done in this area but does not alwaysaddress a biologically relevant problem.There are two areas in which mechanistic studies cancontribute meaningfully to our understanding of bioinorganicsystems.(1) Studies on isolated, purified metalloproteins: defining thestructure-function relationship of the protein in terms of ourunderstanding of the structure and reactivity of simple com-plexes.For example, understanding: (i) how the polypeptidebackbone augments the reactivity of the active site and (ii) whyparticular metals (or groups of metals) with a specific set ofligands are employed by the biological system.(2) Using simple metal complexes to establish a chemicalprecedent for the proposed mechanisms of the elementaryreactions of metalloenzymes. This type of study is often referredto as involving ‘model systems’. This is misleading and leads tothe criticism that the metal site under investigation bears norelation to that in the enzyme. Undoubtedly, based on astructural comparison, this criticism is justified, but that is notwhat these studies purport to do. What is being ‘modelled’ is thereactivity (or part of the reactivity) of the enzyme.Consideration of the types of mechanisms being studied in (2)shows that we are looking at the reactions of ligands, and themanner in which they can be activated or transformed.Theactivation and transformation of co-ordinated molecules is alsoa major concern in organometallic and co-ordination chem504istry, and catalysis. It follows then that the mechanistic prin-ciples established for the reactions of ligands will pervade manyareas of chemistry. Consequently, although understanding thereactivity of substrates of metalloproteins is a meritorious goalin its own right, the application of the knowledge we gain fromthis research can result in the development of new chemistry inareas far removed from metalloenzymes: chemistry which mightotherwise have been overlooked.Defining the Mechanisms of Catalysts: FunctionalGroup ChemistryThere are three stages by which any catalyst, including metallo-enzymes, converts reactants into products: binding of the sub-strate; transformation of the substrate into product and finallythe release of the product.The complexity of a generalisedenzyme is shown in Scheme 1. Thus the catalysis consists of atleast three elementary reactions corresponding to each of thestages listed above. However, this is just the start of the com-plexities. For example, it is common for the transformationstage to consist of several elementary reactions and, addition-ally in some cases, the enzyme may consist of several proteins,all of which must interact to produce the enzyme.Thus themanner in which the proteins interact must also be defined.Considering the complexity of any catalyst system it is rarely,if ever, possible to define the intimate mechanism of the action ofthe catalyst at the atomic level by studies on the enzyme itself.There are three limitations in defining the detailed mechanismof any catalyst: (1) the large number of steps involved, and thecatalytic nature of the process, means that it is often not possibleto look at any one step in isolation from the rest; (2) the rate-limiting step of the catalysis limits the kinetic information; anysteps occurring after the rate-limiting step are kinetically hidden;( 3 ) a large number of species may be present in solution, inequilibrium with one another, but only some of these species areinvolved directly in the catalytic cycle; with all the othercomplications it is often difficult to establish which is the ‘active’species.The problems inherent in mechanistic studies on catalysts asthey turn over are well illustrated by studies on Wilkinson’scatalyst, [RhCl(PPh,),], which hydrogenates a l k e n e ~ .~ Earlyattempts to study the mechanism of this catalyst led to a varietyof rate laws, all established by simulating the rate data usingseveral variables. Given the number of steps involved and thenumber of independently adjustable parameters, it is hardlysurprising that satisfactory fits can be obtained to severaldifferent rate laws.It is possible, at least on paper, to break up any catalytic cycleinto a series of elementary reactions.Anyone can arrive at adetailed mechanism for each of these elementary steps based ongross observations such as product distributions or the resultsof isotopic labelling studies on the catalyst, as it turns over.However, the mechanism of the elementary step must fulfil twoprerequisites. First, it must be chemically reasonable, andsecondly should be demonstrable in simple chemical complexes.At its best the reactivity is being defined on a complex which isstructurally analogous to the active site in the catalyst. How-ever, this is not always feasible and really the main criterion is asite which has similar electronic characteristics to those of theactive site of the catalyst, for the elementary reaction underconsideration.The model system must fulfil certain prerequi-sites: (a) the structure of the reactant and the product(s) mustbe established; (b) the co-ordination sphere of the complexshould be robust, the only reactive site being the one pertinentto the elementary reaction under investigation; (c) the reactionbeing studied must be stoichiometric and as simple as poss-ible, in order that the maximum mechanistic information can beobtained.It is using this sort of approach that the detailed mechanismsof catalysts can be elucidated. Ultimately, by modelling eachelementary reaction of the catalyst, we can present an overallJ.PEp-productrearrange men\\’HEM. SOC.DALTON TRANS. 1995+ p2L-substrate Icoreadants , n 1 e‘. m H+, 9 OH - erc./ 4- intermediateScheme 1 General picture of the action of an enzyme, illustrating thecomplexity of the system including: the multiprotein nature of theenzyme (PI + P2); binding of the substrate; transformation of the sub-strate including intermediates and release of the product. E,, En and E,represent different states of the enzymepicture of the catalytic action. As an example, Scheme 2 showsthe catalytic cycle for the conversion of dinitrogen intoa m m ~ n i a . ~ . ~ This cycle is based on the chemistry defined onmolybdenum (the physiological metal) and tungsten complexes,such as tran~-[M(N,)~(dppe),], (dppe = Ph,PCH,CH,PPh,)and illustrates the possible reaction pathways that the enzymenitrogenase could adopt.Clearly, the very detailed picture of the reactivity that emergesis all the possible ways in which the reactants are converted intoproducts, not just the pathway that the enzyme employs.It is,of course, impossible to define the enzyme’s preferred routefrom studies on these simple chemical systems. What is definedis all the pathways by which dinitrogen can be converted intoammonia or hydrazine at a single metal site. In order to under-stand the reactivity of any catalyst at the atomic level it is justas important to understand the pathways that it does not use as itis to know the pathways that the catalyst does adopt. What areestablished are chemical precedents for the transformations ofthe substrate.Scheme 2 then illustrates what is really the func-tional group chemistry of the co-ordinated dinitrogen moleculetowards protons.In the remainder of this Perspective I will discuss selectedmechanistic studies on relatively simple complexes where theinitial goal of the research was to understand the reactivity ofvarious metalloenzymes, but which subsequently gave rise toinsights into more diverse areas of chemistry.Action of Carbox ypeptidase and IntramolecularNucleophilic AttackThe enzyme carboxypeptidase is a hydrolytic, pancreaticenzyme which specifically cleaves the C-terminus amino acid ofpolypeptides. It has a molecular weight of ca. 34 600, and itscrystal structure has been determined. The active site of theenzyme has been identified as a hydrophobic cleft at thebottom of which is a zinc atom. The zinc is in a + 2 oxidationstate, has a distorted geometry, and is co-ordinated to thepolypeptide backbone by three amino acid residues: twohistidines (His-69 and His-196) and a bidentate glutamate(Glu-72).The fourth co-ordination site is occupied by a watermolecule.The zinc centre in carboxypeptidase is a hard metal ion, andit is this characteristic which is fundamental in understandingthe action of the enzyme. It has been known for many years7that hard metal ions such as Cu2+, Co2+, Mn2+, Ca2+, Mg”,Zn2+, etc. are capable of hydrolysing a wide range of aminoacid derivatives, such as, peptides, amides, phosphate estersJ . CHEM.SOC. DALTON TRANS. 1995 505Scheme 2characterised and the mechanisms of the interconversions studied in stoichiometric reactionsPathways for the conversion of dinitrogen into ammonia or hydrazine at a single metal site. All species shown have been structurallyScheme 3 Pathways for the hydrolysis of amino acid derivatives atsimple metal sites: top, intramolecular route; bottom, intermolecularroutesulfonate esters, acetals and esters. The mechanism originallyproposed for this hydrolysis is shown in the bottom line ofScheme 3. The mechanism described here is that intuitivelyexpected of the enzyme. The metal ion is the site where thepolypeptide binds, and is activated. Being hard it withdrawselectron density from the substrate, thus rendering the carbonylcarbon more susceptible to nucleophilic attack by the freehydroxide ion.It is difficult to define the intimate details of the mechanismof hydrolysis of amino acid derivatives at the metal ions listedabove because their substitution lability makes it impossible tobe sure of the exact nature of the co-ordination sphere.How-ever, using the robust cobalt(Ir1) amine * complex, p-cis-[Co(OH,)(glycine ester)(NH2CH,CH,NHCH2CH2NHCH2-CH2NH2)I3 + in which the co-ordination sphere of the cobaltremains intact and of defined stereochemistry throughout thereaction, it is possible to show, by isotope labelling studies, thatthe major hydrolysis pathway involves intramolecular attack ofco-ordinated hydroxide on the carbonyl carbon, as shown onthe top line of Scheme 3.This mechanistic result is unexpected.Hydroxide bound to a hard metal site is intuitively expected tobe a weaker nucleophile than free hydroxide ion because of theelectron-withdrawing effect of the hard metal. The prevalenceof the intramolecular hydrolysis route is due to the correctstereochemistry and proximity of the co-ordinated hydroxideto the carbonyl carbon.It is not clear which of these two pathways operates in carb-oxypeptidase and enzyme mechanisms based on both theseroutes have been proposed for the hydrolysis of peptides asshown in Scheme 4.The general principle of intramolecular nucleophilic attackby co-ordinated hydroxide has been elaborated upon outsidethe biological sphere as illustrated in Scheme 5 .Thus a varietyof different reactions of synthetic utility can be accomplished bythis type of process including hydrolysis of amino acid deriv-atives, transesterification and hydrolysis of nitriles.Other co-ordinated nucleophiles can undergo similar re-actions. O In particular, amido-groups (generated by deproto-nation of co-ordinated amines) can attack suitably positionedcarbonyl carbon atoms in a-ketocarboxylates, a-aminocarb-onyls, phosphate esters, disulfides and nitriles, as shown inScheme 6. Finally, if co-ordinated amido-groups can act asintramolecular nucleophiles, they can also act as nucleophilesto external molecules. This allows a further framework to bebuilt on the existing ligands and thus result in the synthesisof more elaborate organic molecules including macrocyclicligands, as shown in Scheme 7.This chemistry is now a long way from its origins, based ondefining the mechanism of nucleophilic attack at co-ordinatedamino acid derivatives in order to understand the way in whichcarboxypeptidase works at the atomic level.Yet the basicmechanistic principle of intramolecular nucleophilic attackapplies to a range of different hard metal ions and has resultedin a useful synthetic method to prepare multidentate or macro-cyclic ligands, and organic molecules (after the metal hasbeen removed).Nitrogenases and the Protonation of UnsaturatedHydrocarbonsThe physiological substrate of the nitrogenases, dinitrogen, hasalready been introduced in Scheme 2. There are three nitro506 J.CHEM. SOC. DALTON TRANS. 1995Scheme 4 Possible pathways- RO-- Fro- Q Cu-N v 0Scheme 5philes; R = aryl, R' = alkylExamples of intramolecular attack of oxygen-based nucleo-0Scheme 6arylExamples of intramolecular attack of amido-groups; R =genases, distinguished by their metal content and their productspecificities. One contains iron and molybdenum, another ironand vanadium and a third which apparently contains only iron.Each enzyme consists of two metalloproteins, as shown inScheme 8.For the molybdenum nitrogenase the first protein, commonto all the enzymes, is a Fe,S, protein (molecular weight ca.60000) which transfers electrons to the larger protein. Thelarger protein (molecular weight ca. 230 000) contains twodifferent types of metal clusters: the so-called P clusters whichare believed to act as electron reservoirs before the electrons aretransferred to the other cluster; the cofactor.The cofactorcontains the molybdenum and is also believed to be thesubstrate binding site. It is an iron-sulfur-based cluster, asshown in Scheme 8, in which a molybdenum atom is located a tone end and is further bound to a homocitrate (2-hydroxy-butane-l,2,4-tricarboxylate) molecule in a bidentate fashion.O F 0 ..co<N=&- 0MeCOCHCOMeScheme 7groups on non-bound moleculesExamples of nucleophilic attack of co-ordinated amido-The cluster is bound to the polypeptide backbone only by theimidazole ring of His-442 (at molybdenum) and the sulfur ofCys-275 at the extreme tetrahedral iron atom.The cofactorfrom the vanadium enzyme is believed to be structurallyanalogous to the molybdenum cluster except for the presence ofa vanadium atom in place of molybdenum. It is not obviousfrom this structure at which metal the substrates bind:molybdenum, tetrahedral iron or three-co-ordinate iron.In terms of reactivity the nitrogenases are particularly diverse(and hence challenging to anyone trying to model their re-actions), because of their ability to transform many substratesin uitro including: nitrous oxide, azide ion, cyanide, isocyanides,cyanamide and unsaturated hydrocarbons (alkynes and cyclo-propene). Understanding the transformation of these 'alter-native' substrates at simple metal complexes can give valuable,indirect, information about the nature of the substrate bindingsite.One of these 'alternative' substrates, which is particularlyimportant, is acetylene.The molybdenum nitrogenase reducesacetylene to ethylene, and this is used as a field test for thepresence of nitrogenase in the soil. The formation of ethane ischaracteristic of the presence of the vanadium nitrogenase.There are two distinct problems associated with the reductionof acetylene which need resolution at the atomic level. First,the nitrogenases reduce acetylene stereospecifically to cis-CHDCHD in the presence of D 2 0 as shown in equation ( 1 ) .C,H, + 2D' + 2e- -- cis-CHDCHD (1)Secondly, the factors which discriminate between the evolutionof ethylene and the formation of ethane need to be defined if wJ.CHEM. SOC. DALTON TRANS. 1995 507Molybdenum-Iron protein0 P Clusters Cofador->SubstrateScheme 8 The electron-transport chain, and the structures of the metal clusters in molybdenum nitrogenaseMScheme 9 Pathways for the transformation of alkynes to alkenes or alkanes at a single metal siteare to understand how the molybdenum enzyme only producesethylene whereas the vanadium nitrogenase can also producesome ethane as described by equation (2). It is clear fromC2H2 + 4H' + 4e- - C2H6 (2)equations (1) and (2) that the mode of activation involves thesequential addition of electrons and protons. In modelling thistype of reaction it is common to use a relatively low-oxidation-state complex; that is the electrons necessary to complement theproton additions to the ligand are effectively stored in the metal.The overall picture for the transformation of alkynes at asingle metal site, at least as far as we understand it to date, isshown in Scheme 9.Clearly a picture as detailed as this is notthe result of a single study but rather from three independentstudies on the stoichiometric reactions (3)-(5). In the systems[V(T~~-C~H~)~(T~~-P~CCP~)] + 2HC1-[ V ( T ~ ~ - C ~ H ~ ) ~ C I ~ ] + cis-PhCHCHPh (85%) +trans-PhCHCHPh(l5%) (3)trans-[Mo(q2-RCCH),(dppe),] + HCl-truns-[Mo(CCH,R)Cl(dppe),l + RCCH (4)~~u~s-[Mo(T~~-C~H~)~(~PP~)J + 2HC1-trans-[MoCI,(dppe),] + CzH4 + C,H, (5)described by equations (3) and (4) the factors involved in theprotonation of co-ordinated alkynes are being probed, sites ofprotonation and stereospecificity, etc., whereas in equation (5)the factors discriminating between formation of ethane andevolution of ethylene can be defined.The study on reaction (3)13 demonstrates that, even whenbound to a highly symmetrical metal site, protonation of a sym-metrical alkyne can result in predominant formation of the cis-alkene. This stereoselectivity is due to initial, rapid protonationat the metal, followed by intramolecular migration of thehydride on to the alkyne thus producing the cis-vinyl species.This corresponds to the pathway in the inner circle of Scheme9.Provided the carbon-carbon double bond is retained, thestereochemistry of the subsequently formed alkcne is defined bythat of the vinyl species and hence by the initial site of pro-tonation.A small amount of the trans-alkene is also produced,presumably as a consequence of the direct, but slow, proton-ation of the co-ordinated alkyne at the face remote from themetal.The study on the bis(a1kyne) system l4 shown in equation (4)further confirms the initial stages in the protonation of co-ordinated alkyne described above. However, it also illustrates acommon reaction pathway for these electron-rich, low-oxid-ation-state systems, that is preferential protonation of the vinylspecies at the remote carbon atom to produce an alkylidene,trans-[Mo(CHCH,R)Cl(dppe),]+ (R = alkyl or aryl), whichultimately loses a proton to give the corresponding alkylidyne508 J .CHEM. SOC. DALTON TRANS. 1995100 IC2H4C2H60 20 40 60 200[HCl]/mmol dm-3t o o n,0 0.2 0.4 0.6 0.8[HCl]/mol dm"y .2+ H2+H+ M=N=N; - M=N-N:H 4. H HFig. I Left: hydrocarbon product distribution from the protonation of tr~ns-[Mo(q~-C,H,),(dppe)~], showing that ethane is producedpreferentially at low acid concentrations whereas ethylene is formed at high concentrations. The curves drawn are those defined by the kinetics andthe elementary rate and equilibrium constants derived therefrom. Right: the product distribution from the protonation of trans-[Mo(N,),(E~~PCH,CH,PE~~)~] showing that the hydrazido-complex is produced at high acid concentrations, but dinitrogen is evolved at lowconcentrationstrans-[Mo(CCH,R)Cl(dppe),]; the driving force for this re-action is the attainment of the closed-shell, eighteen-electronconfiguration.This pathway is shown in the bottom right-handside of Scheme 9.There are two important points which need clarification.First, protonation of the vinyl species at the remote carbonatom necessarily results in the formation of a carbon-carbonsingle bond. Rapid free rotation about this single bond is apathway for the equilibration of the cis- and trans-vinyl species,resulting in the loss of stereospecificity for the reaction. Sec-ondly, protonation of the remote carbon atom in the vinylspecies results in an increase in the formal oxidation state of themetal by two units, whereas protonation of the bound carbonatom, producing the alkene, does not change the formal oxid-ation state of the metal.Clearly, the stereospecificity exhibitedby nitrogenase towards acetylene dictates that the enzyme doesnot proceed uia an alkylidene species and it may be that acontrolling factor is the ability, or not, of the enzyme site toundergo a change in oxidation state. It is worth mentioning, inthis context, that a much poorer stereospecificity is observed inthe reaction of nitrogenase with propyne. One explanation ofthis loss of stereospecificity is that the reaction with propyneoccurs in part by this alkylidene pathway.The study on the reaction shown in equation (5) l 5 was ableto define the factors which discriminate between the formationof alkane and evolution of alkene by the simple addition ofelectrons and protons.These pathways are summarised on theleft-hand side of Scheme 9. Equation (5) is slightly misleadingsince this is the stoichiometry only at low concentrations ofacid. At higher concentrations the carbon mass balance isretained, but the amount of ethylene increases whilst theamount of ethane decreases proportionately until at very highconcentrations of acid only ethylene is produced and thelimiting stoichiometry is that shown in equation (6). Detailedtrans-[Mo(q2-C,H,),(dppe),] + 2HC1-trans- [MoH,Cl , (dppe) ,] + 2C , H, (6)mechanistic studies show that at low acid concentrations pro-tonation at the alkene results in the formation of an ethylcomplex which is the precursor to ethane, whilst at higher con-centrations the slower protonation of the metal becomes im-portant and results in labilisation of the bound ethylene.Thehydrocarbon product distribution, as determined by GLC, canbe simulated from the values of the elementary rate constantsand equilibrium constants determined from the kinetic analysisas shown in Fig. 1. The interesting observation here is thecounter-intuitive result that the alkane is produced at low acidconcentrations. That is, the hydrocarbon requiring the additionof protons to be formed is produced preferentially at low ratherthan at the expected high concentrations of acid. For com-parison, also shown in Fig. 1 is the intuitively expected be-haviour observed for the protonation of dinitrogen in trans-[Mo(N,),(Et,PCH,CH,PEt,),l by HCI.' In this system,initial protonation occurs at the metal to give [MoH(N,),-(Et,PCH,CH,PEt,),] + and at low acid concentrations dinitro-gen is released. Diprotonation of the dinitrogen ligand to formthe hydrazido(2-)-species [Mo(NNH,)Cl(Et,PCH,CH,P-Et,),] + only occurs at high concentrations of acid.It is the unexpected result of the product distribution ofthe hydrocarbons outlined above upon which we can build.Protonation of [M~H(q~-C~H~)(dppe)~], at low concen-trations of HC1, produces propene as shown in equation (7).[MoH(q3-C3H5)(dppe),] + 2HC1-[MoH,Cl,(dppe),] + MeCHCH, (7)However, at higher concentrations of HCI propyne is formedand the proportion of propene decreases whilst maintaining aconstant carbon mass balance.At high concentrations of HCIthe limiting stoichiometry is given by equation (8). The[MoH(q3-C3H5)(dppe),] + 2HC1-[MoH,Cl,(dppe),] + MeCCH + H, (8J. CHEM. SOC. DALTON TRANS. 1995 509loo\ 80 MeCCHMeCHCH2 201 /.I0 300 600[HCI]/[ Mo]High acid concentrationI Low acid concentration I11 +H /Scheme 10 General scheme showing how the protonation of alkenecomplexes can give rise to alkane, alkene or alkyne. The reversibility ofthe initial protonation steps means that the isomerisation of alkenescould also be accomplished by this type of reaction, but to date this hasnot been observedmediate, [M~(q~-MeCHCH,)(dppe)~l~ + and commits thesystem to producing propyne. This fourteen-electron inter-mediate is so unsaturated that it abstracts hydrogen atoms fromthe co-ordinated alkene thus producing [MoH,(q2-MeCCH)-(dppe),]' + , which subsequently releases propyne.This work on the protonation of co-ordinated, unsaturatedhydrocarbons is now a long way from studies aimed atunderstanding the product specificity of the various nitro-genases, but the mechanistic principles established in that studyhave been exploited to develop a complete picture of how wecan control the hydrocarbon released by protonation of analkene complex as summarised in Scheme 10.Hydrogenases, Nitrogenases and Dihydrogen LigandsThe hydrogenases are a class of metalloenzymes which performthe reaction shown in equation (9).Hydrogenases in uiuo areH 2 e 2 H + + 2e- (9)Fig. 2 Top: hydrocarbon product distribution for the protonation of[MoH(q3-C,H,)(dppe),] showing the preferential formation ofpropyne at high acid concentrations, whilst propene is the exclusiveproduct at low concentrations.Curves drawn are those defined by thekinetic analysis and the derived elementary rate and equilibriumconstants. Bottom: the mechanism of protonation of [MoH(q3-C,H,)(dPPe),lhydrocarbon product distribution for the reaction of HCl with[M~H(q~-C~H~)(dppe)~] over the range of acid is shown inFig. 2. That the more unsaturated hydrocarbon is producedat higher acid concentrations is reminiscent of the behaviourdescribed above for the ethylene complex, but now we actuallyform an alkyne by protonation of an allyl species derived from analkene complex!Detailed mechanistic studies, including low-temperaturedetection of intermediates by NMR spectroscopy, show that themechanism for the formation of the two hydrocarbons is asshown at the bottom of Fig.2. Again the unexpected productdistribution observed in the protonation of [MoH(q3-C,H,)(dppe),] has its origins in the same effects observed in thereactions of tr~ns-[Mo(q~-C,H,)~(dppe)J: competitive pro-tonation of hydrocarbon ligand and metal. At low concentra-tions of HCI rapid protonation of the allyl ligand results in theevolution of propene via [MoH(q 2-MeCHCHz)(dppe)2] + . Inaddition, competitive, rapid protonation of the metal produces[MoH,(q3-C3H5)(dppe),] +, which at higher acid concentra-tions becomes further protonated (probably on the allyl group)to produce [MoH,(q2-MeCHCH2)(dppe),]' +, and thenrapidly loses dihydrogen to produce the co-ordinativelyunsaturated, five-co-ordinate, formally fourteen-electron inter-probably unidirectional, but in uitro examples are known whichare either uni- or bi-directional.'* There are three types ofhydrogenases distinguished by the elements that they contain:[Fe]-, [Ni-Fel- and mi-Fe-Sel-hydrogenases. The functions ofmany hydrogenases are still not entirely clear but include: (i)delivery of electrons from dihydrogen to the membrane-boundelectron-transport chain and (ii) to trap dihydrogen evolved inother enzyme reactions, for instance the dihydrogen producedas a by-product of dinitrogen fixation (see below).The periplasmic [Fel-hydrogenases from sulfate-reducingbacteria from Desulfouibrio species have been isolated in apurified state. From Desulfouibrio uulgaris the hydrogenaseconsists of two subunits.The a subunit has a molecular weightof ca. 46000 and has three Fe,S, clusters bound throughcysteine amino acid side chains. The p subunit is smaller with amolecular weight of ca. 9600 and probably has a predominantlyelectron-transfer function.The pi-Fel-hydrogenase from Desulfovibrio gigas alsoconsists of two subunits. The a one has a molecular weight of ca.56000-68000 and it is this which contains the nickel. Thesmaller p subunit has a molecular weight of ca. 28 000-35 000and has only an electron-transfer function using two Fe,S,clusters. The metals in the a subunit are probably in one Fe,S,cluster and a single nickel atom, which is close to, or possiblyeven contained within, the cluster.Extended X-ray absorptionfine structure (EXAFS) indicates that the co-ordination sphereof the nickel comprises 3 f 1 light atoms (N or 0) and 2 & 1S atoms.Apart from the physiological reactions shown in equation (9),hydrogenases are also capable of catalysing H/D exchange a510 J . CHEM. SOC. DALTON TRANS. 1995C u a - H - - CU-H* + H+CU-H* + Cu2* - 2Cu+ + H*Scheme 1 1 Mechanism for the oxidation of dihydrogen to protonsL H'Scheme 12 Representation of the elementary steps in the binding ofdinitrogen at the active site of the molybdenum nitrogenase, involvingthe displacement of dihydrogenshown in equation (10) and the equilibration of ortho- andpara-dihydrogen.D, + H + e H D + D+ (10)To understand the reactivity of both uni- and bi-directionalhydrogenases we must model reaction (9) in both directionsindependently, and look for suitable chemical precedents in thereactions of simple compounds.Considering first the activationof dihydrogen, three ways in which a metal can facilitate thisprocess have been identified: heterolytic cleavage; homolyticcleavage and oxidative addition. The heterolytic cleavage ofdihydrogen is accomplished by a variety of simple metalcomplexes such as those of Cu", Ag', Hg', Hg", Pd", Rh'", Ru"and Ru"'. The mechanism of this reaction was established in the1950s for copper acetate, as shown in Scheme 1 1, and remainsessentially unchanged to this day.The homolytic cleavage ofdihydrogen by two molecules of [CO(CN),]~~ or Ag' has alsobeen identified. A mechanism which, in effect, bridges theforward and reverse reactions of equation (9) is the cleavage ofdihydrogen by an oxidative-addition reaction typified byequation (1 1). Mechanistically, the cleavage of dihydrogen intrans-[IrCl(CO)(PPh,),] + H, -+cis, trans-[IrH,Cl(CO)(PPh,),] (1 1)this type of reaction must be considered to involve the transientinteraction of an intact dihydrogen with the co-ordinativelyunsaturated metal centre prior to the cleavage step. However, itwas only with the isolation and structural identification of thefirst dihydrogen complex2' that it was appreciated thatrelatively long-lived intermediates containing dihydrogenligands may well be detected.In the last few years there havebeen many reports of dihydrogen complexes at a variety ofdifferent sites (including hydrogenases and nitrogenases), bothas isolable complexes and transient intermediates.The reactivity of the molybdenum nitrogenase implicates theinvolvement of an intermediate dihydrogen species at thesubstrate binding site.21 In the absence of a substrate,nitrogenase acts as a hydrogenase and evolves dihydrogen byreduction of protons [i.e. the reverse of equation (9)]. Thisevolution of dihydrogen cannot be suppressed entirely, even atextreme pressures of dinitrogen. The limiting stoichiometry ofthe molybdenum nitrogenase is that described by equation (1 2).N, + 8H' + 8e- - 2NH, + H, (12)This obligatory evolution of dihydrogen, together with theobservation that dihydrogen is an inhibitor of dinitrogenreduction, led to the proposal that dinitrogen binds to the activesite of nitrogenase by displacing dihydrogen.The dihydrogenoriginates from a hydrido-species, which is formed by protonsbinding to the reduced active site, as shown in Scheme 12.However, pre-steady-state kinetic analysis and the isotopiccomposition of the liberated gas obtained from labelling studieson the enzyme indicate that the evolution of the dihydrogenfrom this hydridic site is catalysed by acid and, in particular,that this protonation step involves direct attack at the hydrideligand, but with no interaction between the proton and the metal.At the time this mechanism was proposed it was withoutprecedent in simple chemical system^.^,^^ Using the tetra-hydride [MoH,(dppe),] it was shown that dinitrogen (and avariety of other small molecules such as nitriles, isocyanides,alkynes, carbon dioxide and sulfur dioxide, hydrogen sulfide,water, azide ion, halide ion and nitric oxide) could be boundrapidly, and in some cases activated, at a hydridic site in aprocess which is accelerated by acid.The unexpectedmechanistic feature in the protonation reactions of [MoH,-(dppe),] is that this tetrahydride binds two protons beforedihydrogen is evolved, as shown in Scheme 13. Theintermediate, '[MoH,(dppe),12 +', which is the precursor todihydrogen evolution, must be a dihydrogen species, sinceotherwise the maximum oxidation state of molybdenum isexceeded.The question remains, 'how is a dihydrogen ligand formedin the protonation of this and other polyhydridic sites?' Twomechanisms are possible. First, initial protonation of the metalfollowed by coupling two hydride ligands and secondly, directprotonation of a hydride ligand.Although dihydrogen ligandsare formed in the protonation reactions of [MoH,(dppe),] it isnot possible to define the intimate mechanism of their formationusing this system since there are too many steps to be ableto 'home in' on just the dihydrogen ligand formation. In prin-ciple it is possible to distinguish between the two limitingmechanisms by determining the relative proportions of HD andH, in the reaction of [MoH,(dppe),] with D'; in the directprotonation pathway HD would be the exclusive gaseousproduct.However, intramolecular exchange and rapidintermolecular proton exchange (involving both hydrides andphenyl rings of the phosphine ligands) complicates the analysis.In order to define the intimate mechanism of protonation ofhydrido-complexes it is clear that we must move to a simplersystem. Stoichiometrically, the simplest possible process that wecan study is the single protonation of a hydrido-complex, inwhich the nett result of the reaction is that a proton is added tothe metal as shown in the generalised equation (13). If themechanism of this reaction is a simple direct protonation of themetal then the associated kinetics is very simple: a first-orderdependence on both the concentration of the acid and thecomplex.This is the kinetic behaviour observed in thereaction23 of [WH,(dppe),] with HCl to form [WH,-(dppe),] + . However, studies on the analogous reaction of[WH,(PMePh,),] to form [WH,(PMePh,),] + reveal a morecomplex kinetic beha~iour,~, reflecting a more complicatedmechanism as shown in Scheme 14.The rate law for this reaction demonstrates that there are twopathways by which the metal centre is protonated. The simplerpathway (k3) involves direct protonation of the metal (asobserved for [WH,(dppe),]}, but this pathway is slow. Themajor pathway is associated with a rate law which describes amechanism involving an initial, rapidly established, equilibriumprotonation followed by a relatively slow intramolecularrearrangement step. Owing to the stoichiometric simplicity ofthe reaction, and that the only potentially protonatable sites on[WH,(PMePh,),] are the metal and the hydride ligands, we canascribe the initial protonation step to attack at a hydride ligandto form the dihydrogen species as shown in Scheme 14.Thisdihydrogen ligand then undergoes an intramolecular oxidativecleavage to generate the [WH,(PMePh,),] + product.This was the first unambiguous demonstration that hydrideligands were susceptible to direct attack by protons, despitemuch speculation that such mechanisms did operate, not only innitrogenases and hydrogenases, but also in metal complexesJ.CHEM. SOC. DALTON TRANS. 1995 51 1~~ ~~ ~Preparation of leaving group,coupling of hydride ligandsLabilisation of dihydrogen ligand Activation of bound substrateScheme 13intermediateMechanism for the rapid binding of a variety of small molecules at a polyhydridic site, involving protonation and a dihydrogen ligandHI I k I 6 = WC!+hSecond stageScheme 14metal and another pathway involving protonation of a hydride ligand. Phosphine ligands omitted for clarityMechanisms and rate equations for the protonation of [WH,(PMePh,),] to give [WH5(PMePh,),]+, showing direct attack at theThe success of the approach is due to the use of rapid reactiontechniques, in effect following the progress of the reaction as itoccurs. The kinetics observed for the direct protonation of ahydride ligand followed by an intramolecular cleavage step issufficiently diagnostic to be used as a general method forestablishing the mechanism of protonation of other hydridecomplexes provided those complexes contain no other ligandscapable of being protonated.In addition, the quantification ofthe elementary rate and equilibrium constants associated withthis pathway allows detailed analysis of the fundamentalreactions associated with this type of complex such as theacidity of the dihydrogen ligand, etc.Again we see that the establishment of chemical precedent forthe elementary reactions of metalloenzymes can have a moregeneral applicability in chemistry.ConclusionI hope to have shown in this Perspective that studying thereaction mechanisms of relatively simple transition-metalcomplexes can define what are feasible elementary reactions formetalloenzymes.In addition, more importantly, that thesestudies can lead to an understanding of reactivity beyond that ofthe biological problem, and even to the development of newareas of chemistry. I have tried to illustrate this with work onthree very different metalloenzymes where the impact of theinorganic mechanistic work spreads across the areas ofbioinorganic, co-ordination, organometallic and syntheticorganic chemistry. Clearly the inorganic reaction mechanistmust not become so polarised towards one area of chemistrythat the relevance of the results to other, more diverse areas,is overlooked.References1 H.Taube, J. Chem. Soc., Dalton Trans., 1991, 547 and refs. therein;see also Comprehensive Coordination Chemistry, eds. G. Wilkinson,R. D. Gillard and J. A. McCleverty, Pergamon, Oxford, 1987, vol. 1,2 See, for example, Advances in Inorganic and BioinorganicMechanisms, ed. A. G. Sykes, Academic Press, London, 1982- 1984,1986, vols. 14.chs. 7.1-7.5.3 J. Halpern, Inorg. Chim. Acta, 1981, 50, 11 and refs. therein.4 D. J. Evans, R. A. Henderson and B. E. Smith, Bioinorganic Catalysis,ed. J. Reedijk, Marcel Dekker, New York, 1993,89 and refs. therein.5 T. E. Glassman, M. G. Vale and R. R. Schrock, J. Am. Chem. Soc.,1992,114,8098.6 D. W. Christianson and W. N. Livscomb, Ace. Chem. Res.. 1989.22.7891011121314151617181962 and refs. therein.H. Kroll, J. Am. Chem. Soc., 1952,74, 2036.D. A. Buckingham, D. M. Foster, L. G. Marzilli and A. M. Sargeson,Inorg. Chem., 1970,9, 1 1.D. St C. Black, Comprehensive Coordination Chemistry, eds.G. Wilkinson, R. D. Gillard and J. A. McCleverty, Pergamon,Oxford, 1987, vol. 1, p. 415.R. W. Hay, Comprehensive Coordination Chemistry, eds.G. Wilkinson, R. D. Gillard and J. A. McCleverty, Pergamon,Oxford, 1987, vol. 6, p. 41 1.D. St C. Black, Comprehensive Coordination Chemistry, eds.G. Wilkinson, R. D. Gillard and J. A. McCleverty, Pergamon,Oxford, 1987, vol. 6, p. 151.M. J. Dilworth, R. R. Eady, R. L. Robson and R. W. Miller, Nature(London), 1987,327, 167 and refs. therein.R. A. Henderson, D. J. Lowe and P. Salisbury, J. Organomet. Chem.,in the press.R. A. Henderson, K. E. Oglieve and P. Salisbury, unpublished work;see also, A. Hills, D. L. Hughes, N. Kashef, A. J. L. Pombeiro andR. L. Richards, J. Chem. Soc., Dalton Trans., 1992, 1775 and refs.therein.R. A. Henderson and K. E. Oglieve, J. Chem. Soc., Dalton Trans.,1991, 3295 and refs. therein.R. A. Henderson, J. Chem. Soc., Dalton Trans., 1984,2259.R. A. Henderson and K. E. Oglieve, J. Chem. Soc., Chem. Commun.,1993,474.G. Voordouw, Adv. Inorg. Chem., 1992,38, 397 and refs. therein.J. Halpern, J. Organomet. Chem., 1980,200, 133 and refs. therein.20 G. J. Xubas, R. R. Ryan, B. I. Swanson, P. J. Vergamini andH. J. Wasserman, J. Am. Chem. Soc., 1984, 106, 451; P. G. Jessopand R. H. Morris, Coord. Chem. Rev., 1992, 121, 155 and refs.therein.21 R. N. F. Thorneley and D. J. Lowe, Molybdenum Enzymes, ed.T. G. Spiro, Wiley, New York, 1985, p. 221 and refs. therein.22 R. A. Henderson, J. Chem. Soc., Chem. Commun.. 1987, 1670 andrefs. therein.23 R. A. Henderson, unpublished work.24 R. A. Henderson and K. E. Oglieve, J. Chem. SOC., Chem. Commun.,1992, 441 and refs. therein.Received 27th September 1994; Paper 4105901