8/13/2019 Cramer 008

1/10

3398 Journal of the Amer ican Chem ica l Soc ie ty / OO: l I / M a y 2 4 , 1 9 7 8The M olybdenum Si te of Nitrogenase. PreliminaryStru ctur al Evidence from X-R ay Absorption SpectroscopyStephen P. Cramer,la Keith 0 Hodgson,*la William 0 Gillum,la andLeonard E. MortensonlbContribution rom the Department of Chemistry. Stanford liniuersitjt,Stan ford , California 94305, and th e Department ofBiologica1 Sciences. Purdue Uniuersity,Lafapette, Indiana 47907. Receiced Septemb er 13, 1977

Abstract: On e of the two protein components of the nitrogenase enzyme system (the so-called MoF e protein) contains two Moatoms and 24 -32 Fe atom s per 220-270 000 molecular weight. Despite m any hypotheses a bout th e M o site and its involvementin dinitrogen reduction, there has been no spectroscopic means of unambiguously probing the state of M o. I n this paper, theresults of x-ray. absorption spectroscopy studies are desc ribed an d th e analysis of these results provides the first direct evidenceof the Mo coordination environment in the MoFe protein from Clostridium pasteurianum nitrogenase. The Mo is found tohave primarily S ligation and in the resting sta te (semireduced form ) of the protein there appears to be no Mo=O speciespresent. On air oxidation, however, clear evidence is found for the presence of Mo=O. The interaction of the Mo with oneother m etal, most probably Fe and most certainly not M a, is also seen and it is suggested tha t the M o is present in a M o-Fe-Scluster not yet adequately modeled by any synthetic inorganic system.

IntroductionTh e molybdenum-iron (M oF e) protein of Clos t r id iumpasteurianurn ni t rogenase conta ins two molybdenum a tomsand approximately 24 ato ms each of iron and acid-labile sulfideper molecular weight of 220 000.2aDespite many hypothesesthat molybdenum is a t the si te of dinitrogen reduction,2buntilrecently there has been no unambiguous means of spect ro-scopically monitoring th e stat e of molybdenum in this protein.Th e UV-visible absorp tion spectrum of the Mo Fe protein isa featureless, monotonic rise in absorbance from 70 0 nm to t he~ l t r a v i o l e t , ~nd t he M C D and C D spect ra in t h is r eg ion a r ea lso ~ n i n f o r m a t i v e . ~ ~ ~he l ow- t empe ra tu re E PR spec t rumof the clostridial MoFe protein as isolated (the so-calledsemireduced form) has resonances at g = 4.27, 3.78, and2.01, and these have been attribu ted to a spin 3/2 system wi thr h om b ic an d a xi al d i ~ t o r t i o n . ~he E PR of t he MoFe p rot ein

of Klebs iel la pn eumoniae grown on 87% 57Feshows about a5-G broaden ing of the sharpes t resonance,6 but no hyperf ineinteraction is observed upon incorporation of 95Mo nto theMo Fe pr ot eh 6 Similar results were obtained w ith isotopicallylabeled Azotobacter vinelandii Mo Fe protein. Althoug h theseexperiments are consistent with the presence of F e a t t h eEPR-active site, the absence of a measurable effect with 95Modoes not e l iminate the presence of an EPR-si lent M o a t th issite.8 Th e paucity of information ab out th e M o in nitrogenaseis such th at, unti l recently, the only evidence for i ts presencecame from e lementa l analysis of the MoF e protein an d thenu t ri t iona l r equ iremen t fo r M o in n it rogen f i ~ a t i o n . ~he reis obviously a gre at need for a spectroscopic method which c andefine the state of M o in nitrogenase.X-r ay absorption spectroscopy is one technique which canunambiguously examine a particular element in a complexsample in a ny physical state. Th e availabil ity o f a high-inten-sity, broad-band synchrotron radiation source at Stanf ord hasmad e da ta collection on dilute samples such as metalloproteinsfeasible for the first t ime, and reports have been published onthe spectra of iron-IO n d copper-containing proteins. In thispaper, M o K-edge x-ray absorption spectra of the ClostridialMo Fe protein a re presented and a nalyzed. Thes e results willexpand upon previously reported edge da ta l2 and present, forthe f i rs t t ime, conclusive informat ion about the M o environ-ment . The EX AF S data provide c lear evidence for pr imari lysulfur ligation to molybdenum , and together with th e edge datathey conclusively ru le out the presence of doubly boundoxygen(^).'^ I t i s suggested tha t the molybdenum in nitroge-

0002-7863/78/1500-3398$01 OO/O

nase is present in a Mo-F e-S cluster of a type not yet ade-quately modeled by an y synthetic inorganic system.

Experimental SectionProtein Purification and Lyophilization. The M oFe and Fe proteinsof Clostridium pasteurianum were prepared by the method of Zum fta n d M ~ r t e n s o n . ~echniques used t o analyze for protein, Mo, Fe , S2-,and activity ar e also cited in ref 14 . I n order to obtain the best possibleEX AFS d ata , a sam ple with the concentration of M o as high as pos-sible was required. Since the highest MoFe protein concentration thatcan be obtained in solution is ca. 200 mg/m L or about 2 mM Ma, andsince data with the best possible S / N ratio were desired, lyophilizedsamples were prepa red for these preliminary studies. For lyophiliza-tion, 3 mL of a 70 mg/m L sam ple of the Mo Fe protein with activityof 1.5 pmol acetylene reduced min-l mg-I protein was pipetted an-

aerobically under argon in to a 5 0-mL lyophilizing flask. Th e prepa-ration was shell frozen on the inside of the flask and the flask imme-diately attached t o the lyophilizer and evac uated. To further protectagainst oxidation during lyophilization, 0.1 m M sodium dithionitewas added t o the preparation before freezing. Onc e dried, the lyoph-ilized powder (ca. 200 mg of protein conta ining 1.7 pmol of M a) waspacked into a Lucite block designed for the x-ra y absorption analysis.All transfers were made in a moisture-free box under argon. Aftertransfer t he open end of the block was sealed with a piece of Lucitefixed to the block by epoxy glue.After the sam ple was analyzed, the block was opened under mois-ture-free conditions and the powder transferred to a dry tube. Th e tubewas sealed with a serum stopper, the gas phase in the tube was changedto argon, and 5 mL of 02-fr ee deionized water w as added anaerobi-cally by syringe. This solution of M oFe protein when assayed had anactivity of 1.3 pmol acetylene reduced min- mg-l protein. Thiscompa res favorably with the initial activity of 1.5 (87% recovery).Preparation of Dye-Oxidized, Full Reduced, and Air-OxidizedStates. It has been previously shown that the MoFe protein could existin three states-the dye-oxidized, the semireduced, and the full re-duced-all of which can be fully activ ated in the nitrogenas ea s ~ a y . ~ ~ ~ ~ ~ ~he lyophilized preparation described above was in thesemireduced state. In a prior publication, edge da ta were reported ona lyophilized sample which had been partially air oxidized.I2 For theexperiments to be described in this paper, the fully reduced and dye-oxidized forms of the MoFe protein were also prepared in conjunctionwith semireduced solution and lyophilized samples.Th e dye-oxidized form was prepared by ad ding anaerobically anamoun t of a co ncentra ted, 02-fr ee solution of oxidized dye, thionine(Lauths violet), to the sem ireduced form. The am ount of dye addedwas equivalent in molar concentration to the MoFe protein plusenough dye to oxidize the traces of dithionite in the pr e pa ra ti ~n . ~or1978 American Chemical Society

8/13/2019 Cramer 008

2/10

Hodgson et al. / Molybdenum Site of Nitrogenaseeach milliliter of a prepara tion of dithionite-free MoF e protein of 88m g /m L , 0.8 pmol of thionine was added in as sma ll a volume as pos-sible so that the Mo concentration would not be diluted significantly.The res ulting oxidized MoFe sam ple, which exhibited less than 10%of the original EPR signal, was exam ined as a solution frozen at dryice temperature.The preparation of the ful ly reduced form was more complicatedsince semireduced M oFe protein can only be re duced effectively byelectrons supplied by the Fe protein.I5 In addition, the electron transferrequires and utilizes adenosine triphosphate. For this experiment thefollowing solutions were prepare d and made anaerobic: Mo Fe protein,0.44 pmol/rnL containing 1.7 Mo atoms/pmol; Fe protein, 0 .4pm ol/m L containing 3.7 Fe atoms/pmol; an ATP-gene rating systemcontaining (per 0.1 mL) 105 units of creatine kinase, 8 pmol of AT P,20 pmol of M g2+ ,and I O pmol of creatine phosphate added to 1 MTris-HCI at pH 8.0 (final pH 7.8); and Na2S204, 100 pmol/mL in0.5 M Tris-HCI (pH 8.0). For preparation of fully reduced MoFeprotein 1.2 mL of MoF e protein was mixed anaer obic ally with 0.1 mLof Na2S204 in a 5-mL vial sealed with a se rum cap. Using a syringe,0.13 mL of the ATP-generating system was added to the MoFe pro-tein/dithioni te solution. After m ixing and checking that dithionitewas in excess and tha t the pH was 7 .8, 0.1 3 mL of Fe protein solutionwas added. This reaction mixture (under argon) was mixed for 2 mina t 25 OC after which time I mL was transferred rapidly and anaer o-bically into an anaerobic E XA FS cell and 0.3 mL into an anaerobicEPR tube. For comparison purposes, 0.23 mL of MoFe protein and0.07 mL of 0.05 M Tris-HC1 (pH 8.0) were added to a second EPRtube. All samples were immediately frozen in either d ry ice or liquidN2. On examination, the control tube with the semireduced MoFeprotein had the expected EPR signal with gva lues of 4.27,3 .78, and2.01. The fully reduced sample ha d, as expected, less than 5 of thisEPR ~ i g n a I . ~ ~ ~ ~ *Air-oxidized samp les were prepare d by redissolving the lyophilizedmaterial in a minimal amount of aerated 50 mM pH 7.4 Tris buffer.Bubbling began almost imm ediately, and the odor of H2S was quiteappa rent. The first spectrum of this material was recorded about 2h later. After 6 h of room tempe rature d ata collection, the sample wasreturned to 4 OC for storage. After 2 weeks of storage, the samplesspectrum was again recorded.X-Ray Absorption Measurements. The x-ray absorption spectraof all samples were obtained at the Stanford Synchrotron RadiationLaboratory (SSRL) using a Si[2,2,0] channel cut crystal mono-c h r ~ m a t o r ~nd a rgon-filled ionization detectors. The instrumentalresolution was better than 6 eV (estimated from the fwhm of thesharpest feature in the M 00 4~ - dge). The photon flux available aftermonochro matization was a str ong function of the electron storage ringoperating conditions; for example, at 3.48 GeV and 31 mA, theavailable flux was about IO 9 photons/s in a 1 X 20 mm are a at 20 keV,while at 2.71 GeV and 14.2 mA the flux was about 3 X I O 7 photons/s.All of the protein spectra were recorded when the ring was at 3.2 GeVor higher.The absorption spectra were calibrated with respect to the ab-sorption edge of M o foil, using the fi rst inflection point of the foil edgeas 20 003.9 eV.20 The first and second derivatives of all edges werecalculated using se quential applications of a cubic finite differenceapproximation over a 7-eV interval. Lon g-term reproducibility ofinflection point measurements on stan dard samples was better thanf0 .7 5 eV. The data processing required for the analysis of the EXA FSdata has been previously described in detaiL2The sample cells were Lucite blocks in which a channel was ma-chined to hold the protein. The path length was 20 mm and channeldimensions perpendicular to the beam were 2 X 25 mm. The cellwindows were m ade of 1.6 mm thick Lucite. Solid samples werepacked into the cell in an anaerobic glove box, while solutions wereinjected anaerobically with a gas-tight syringe through a rubberseptum in a glass tube at the top of the cell. The solution samples werekept frozen in a S tyrofoam box packed with dry ice while the spectrawere being recorded.The spectrum of the lyophilized material was recorded on twoseparate occasions using two different protein samples and averaged.The E XA FS presented represents about 15 h of actual da ta collectionover a 1000-eV region; it is actually the sum of about 30 individual30-min scans. The solution absorption edges each required a bout 2h of data collection over a 100-eV region. The actual hands-on in-strumental time required to obtain useful spectra was generally afactor of 2 or 3 greater than the data collection time.

3 3 9 9Results

T h e M o K - e d g e x - r a y a b s or p t i on s p e c t r u m of ClostridialM o F e p ro t e in ca n b e d iv id ed in to tw o d i s t in c t r eg ion s : ( I ) t h eab so rp t io n ed g e-w h ose sh ap e an d po s i t io n co n ta in in fo r -mat ion abou t t he type(s) o f l igands, the coord inat ion geometr y( si t e s ym m et ry ), an d possibly th e m et al ox id at io n ~ t a t e , ~ ~ .a n d ( 2 ) t h e e x t e n d e d x - r a y a b s o r p t i o n f i n e s t r u c t u r e(E X A F S ) -w h ich can rev ea l t h e ty p e , n u m b e r , an d di s t ancesof neighboring atom s.* I t was relat ively easy to o b ta in u se fu lM o a b s o r pt i on e d g e d a t a on sev e ra l d i f f e ren t s t a t e s o f t h eM o F e p ro t e in in so lu tio n . On t h e o th e r h a n d , co l lec t ion an danalysis of E X A F S d a t a r e q ui r ed a muc h bet ter s ignal to no isera t io , s in ce o sc il l a t io n s w i th am p l i tu d es t y p ica l ly f r o m I O t o0.1 t h e s i ze of t h e e d g e j u m p w e r e b e in g m e a s u r e d . T h e r e -f o r e , t h e E X A F S o f t h e M o F e p r o t e in h a s b e e n i n i t i al l y r e -corde d using a lyoph il ized powder . Th is was done to min imizes o lv e nt b a c k g r o u n d a b s o r p t i o n a n d to o b ta in t h e h ig h es t p o s -sible fraction of absorp t ion by Mo. For th e lyoph i lized p ro tein ,t h e M o c o n t r i b u t e s a b o u t 10 of t h e t o t a l ab so rp t io n , w h i l ein so lu t ions wi th 100 m g / m L o f p r o t e in ( a b o u t 1 m M i n M o)t h e M o a b s o r p t io n i s on l y a b o u t 1% o f t h e t o t a l . T h e e d g e sr e c or d e d f r o m s o l ut io n s a m p l es a n d t h e E X A F S d a t a f r o m aly o p hi l i zed sam p le w i ll b e p resen ted se p a ra t e ly , an d th eco m b in ed ev id en ce w il l b e u sed to d i scu ss t h e s t ru c tu re o f t h eM o s i t e .The Mo K Absorption Edge Region. I n t h e m e t a l K a b -sorp t ion edge reg ion o f coord inat ion complexes, one typ ical lyo b se rves a se r i e s of d i sc re t e sp ec t r a l p eak s (d u e to t r an s i t i on sof t h e 1s e l ec t ro n to u n o ccu p ied m o lecu la r o rb i t a l s o f m e ta lchara cter) super imposed upon a s teep ly r is ing absorp tion t rend(d u e to t r an s i t i on s of t h e 1s elect ron in t o con t in uum levels) .23T h e shape of t h e ed g e , w h ich i s d e t e rm in ed b y th e r e l a t i v ein t en s i ti e s an d w id th s o f t h ese lo w - ly in g b o u n d -s t a t e t r an -s i t io n s , co n ta in s i n fo rm at io n ab o u t t h e g eo m et ry o f t h e m e ta lcomplex and the na tu re o f the l igands p resen t . Thus, octa hedra lco m p lex es o f f i r s t -ro w t r an s i ti o n m e ta l s w i th r e l a t iv e ly h a rdl ig an d s ( am in es , w a te r s , h a l id es , e t c . ) h av e s t ro n g 1 -4 pt r an s i t io n s , b u t v e ry w eak 1 s -3 d a n d 1s-4s transition^.^^ Ont h e o t h e r h a n d , c o m p l e x es w it h t e t r a h e d r a l g e o m e t r y h a v es t ro n g ly en h a n ced 1 s -n d an d 1 s - (n 1) t r an s i t i on s , b ecau seof a mix ing o f d , s, a n d p c h a r a c t e r in t h e ex c i t ed s t a t e o rb i -t a l . a ,24 F in a l ly , l ig an d s w h ich h av e m u l t ip l e -b o n d in g in t e r -a c t i o ns w i t h t h e m e t a l c a n i n t e ns i f y e i t h e r t h e I s - n d o r t h eIs- n 1 )s t r ans i t i o n s o r b o th . 2 5T h e p o s i t i o ns of t h e ab so rp t io n ed g e f ea tu res a re sen s i ti v eto t h e m e t a l o x i d a t i on s t a t e a n d e n v i r o n m e n t . A s t h e e l e c t r o -n eg a t iv i ty o f t h e l i g an d s in c reases , t h e ed g e f ea tu res m o v e toh ig h er en e rg y . I2 F u r th e rm o re , fo r a g iv en se t o f l i g an d s, t h eed g e m o v es to h ig h e r en e rg y as t h e m e ta l o x id a t io n s t a t e i n -c reases . T h i s sh i f t can b e a s l a rg e a s 5 eV p e r u n i t o x id a t io ns t a t e ch an g e ( a s w i th h igh ly io n ic fluoride^).^^ I t i s typ ical lya b o u t 2 e V f o r o xi d es a n d c h l o r id e s I 2 a n d a b o u t 1 eV or lessper ox idat ion s ta te chan ge with h igh ly po lar izab le l igand s (aswith cyan ides24and su lfu r-con tain ing ligand^^^.^^). S y s tem at i ct rends in the absorp t ion edges of molybde num co mpounds haveb een d iscussed in d et a il e l s e ~ h e r e . ~ ~ , ~ ~S ev era l d i s t i n c t s t a t e s o f t h e M o F e p ro t e in h av e b eenc h a r a c t e r i z e d b y EPR s p e c t r os c o p y , a n d t h e M o a b s o r p t i o ned g es o f fo u r o f t h ese s t a t e s h av e b een in v es t ig a t ed to seew h e th e r ch an g e s in t h e o v e ra ll p ro t e in o x id a t io n l eve l can b er e l a t e d to c h a n g e s i n t h e M o o x i d at i o n s t a t e . T h e ClostridialM oF e pro tein , as i so lated in the p resence o f d i th ion i te , exh ib i tsa n EPR with g v a lu es o f 4 . 2 7 , 3 . 7 8 , an d 2.01.5T h is s t a t e i st e r m e d s e m i r e d u c e d ( o r C p l ) , s i n c e t h e E P R c a n b e e li m -in a t ed b y fu r th e r r ed u c t io n w i th d i th io n i t e , F e p ro t e in , an dM g A T P ( t o yi el d t h e f u ll y r e d u c e d , C p l f r , s t a t e ) or by oxi-d a t io n w i th th io n in e o r m e th y len e b lu e ( to y i e ld t h e d y e-oxidized, Cpldo,s ta te) . F inally , th e p ro tein c an be i r reversib ly

8/13/2019 Cramer 008

3/10

3400 J a u r n a l of t h e A m e r i c a n C h e m i c a l S o c i e t y / 100.1 I / M a y 24, 1978A I R - O X I D I ZE D - , , , ~ i\A IR -

D Y E - O X I D I ZE D 7 O X I D I ZE D

c2aam4xI

-4

zm>

OXIDIZED 81 wz

a

19990 20015 20040 19990 2 0 1 5 20040ENERGY ev)

Figure 1. Molybdenum K x-ray absorption edges of Closrridial MoFeprotein i n four different states. For these solution spectra, the total a b-sorption is dominated by water absorption (90 ),protein absorption (6%),and Fe-S absorption (3%). These components decrease monotonically asthe photon energy increases, giving rise to a sloping background on whichthe Mo K-edge is superimposed (se e Figure 2 of ref 12 ). Th e Mo compo-nent was removed from the background absorption by fitting a straightline to the total abs orption on the range of 19 930 to 19 990 eV, and sub-tracting this line from the total absorption. Since the extracted Mo ab-sorption peak positions are sensitive to the slope of the s ubtrac ted pree dge,small differenc es in peak positions on the order of 1 eV, as measured bychanges in the inflection points, are not necessarily significant (see Ex-perimental Section). The derivative spectra (right) are useful because (a)they are much less sensitive to background absorption and (b) they helpsharpen the rather broad feat ures common to most Mo edges. The ab-sorption derivatives were calculated using a cubic finite difference a p-proximation over a 7-eV (26 data point) interval.

Table I. Absorption Edge InformationSta te Conditions Inflection point(s) Peak

Fully reduced Solution (a) 200 11.1 20 030.9Solution (b) 200 11. 6 20 032.1Solution (c) 200 11.7 20 030.7Semir educe d Lyophilized (a) 20 01 1.5 20 024.5Solution (b) 20 012.1 20 033.5Dye oxidized Solution 20 012.5 20 03 1.8Air oxidized Solution (a ) 20 004.9, 20 014.9 20 042.1Solution (b) 20 004.9. 20 016.6 20 041.8

oxidized by exposure to air or 0 2 to yield th e air-oxidized ,Cplao, state( s)), in which case a com plicated EPR spec t rumresult^.^ T h e M o K abso rption edges for these four state s ar epresented in Figure 1, and the relevant inflection point andpeak energies are listed in Table I .Th e semireduced sta te of Clostr idial Mo Fe protein, Cpl ',has a smooth, featureless absorption edge, whether the proteini s lyophilized o r in solution (see Figure 2 ), and t he derivativespectrum reveals single peaks a t 20 01 1.5 and 2 0 012.1 eV,respectively. Altho ugh the slopes of the edges ap pea r to differin the region beyond 20 020 eV , this is mor e likely an artif actof the procedure for removing the background absorption thana real difference in the spectra. The similarity in the sh ape andpeak positions of th e derivative spectra leads to th e conclusionthat no dras t ic changes in the M o coordinat ion sphere occurupon lyophilization.In order to mak e some structural conclusions from th e shapeand position of th e C p P edges, i t has been assumed that thebiochemically reasonable ligand possibilities for C p P M o i n -volve nitrogen, oxygen, or sulfur coordination, and that thereasonable M o oxidation sta tes are MolI1-MoV1. Comp arisonof the C p P edge with those of known compounds (see Table

I I I19990 20015 20040 19990 20015 20040ENERG Y eV)Figure 2. A comparison of lyophilized and solution C plsr with thre e mo-lybdenum compounds with all-sulfur coordination. The two edge inflectionpoints for (MoS4)*- are 20 000.1 and 20 010.0 eV and the single edgeinflection point for the other edges (top to bottom) a re 20 010. 0,20 010.5,20 01 1 . 5 , and 20 012.2 eV. Data processing was similar to that of F igureI except that the model compounds did not require a preedge backgroundsubtraction. Note that the Is-4d transition is extremely weak except i nthe case of tetrahedral coordination for Mo i n ( M o S ~ ) ~ - .

Table 11. Relevant M o Complex Absorption Edge InformationCompd oxidn sta te energy, eVForm al M o Inflection point

M o S ~ IV 20 0 10.420 010.5M o C I ~ 111 20 010.7M o C I ~ I V 20 012.7MoC ls V 20 014.3MoOz 1v 20 015.4M o O ( O H ) ~ V 2 0 0 17 .720 018.9OO? VI

M ~ ( & C N E ~ ~ ) ( S Z C ~ H ~ ) Z( M O S ~ ) ~ - VI 20 01 1.1

11) then reveals t ha t a n inflection point of 20 01 1.5 eV is afairly low value, most ch aracteristic of sulfur ligation.26 Ithas been observed that the presence of a single oxygen inMo1 O(S2CNEt2)2 yields an edge inflection point of 20 012.1eV25and t hat increases in the num ber of oxygens for a givenoxidation state shift the edge to even higher energies (theMoIV02 edge i s a t 20 015.4 eV) . Thus , f rom the edge da taa lone , it appears unlike ly tha t C p P would have mo re than asingle oxygen ligand. The lack of suitable Mo-N complex dat aprecludes estimation of the possible nu mb er of nitrogens fro mthe current edge data .When sulfur ligation is predominant, the absorption edgeposition i s relatively insensitive to changes in th e M o oxidationstate. Although m olybd enum chlorides and oxides show edgeshifts of as much as 2 eV per unit oxidation stat e chang e, mo-lybdenum-sulfur compounds exhibit changes of 0.5 eV or lessper uni t change.25These small shifts are within the experi-mental error of the edge posi tion measurements . Thus , a l -t ho ug h t he C p P edge is in a region indicative of sulfur ligation(and clearly too low for prim arily oxygen ligation), l it t le canbe said about the formal M o oxidat ion s tate from these dataalone.T h e sh a pe of t h e C p P edge, a sm ooth rise with a single in-flection point (see Figu re 2), can b e used to el iminate both atet rahedral environment and doubly bound oxygen coordina-tion. Mixing of d, s, and p c hara cter in the orbitals produced

8/13/2019 Cramer 008

4/10

Hodgson e t al. / M o l y b d e n u m S i t e of Ni t rogenase 3401by covalent bonding and /or te t rahedra l symm etry makes hy-brid orbitals accessible and gives intensity to extra boun d sta tetransition^.^' Th e enhancement of the 1s-4d t ransi t ion hasbeen clearly observed in molybda tes and thio molybdate^.^^This particular transition generally produces an extra shoulderor peak in the M o a bs o rp ti on ed ge s, as in t h e ( M o S ~ ) ~ -dgeof Figure 2. The absence of such a f ea tu re in t he C p P edgepermits elimination of strictly tetra hedr al geometr y for Mo.Similar enhanc ement of the 1s-4d transit ion occurs formolyb denum complexes with doubly bound oxygen. All mo-lybdenum complexes with Mo=O examined to date, whetherMo (IV) , Mo(V ), or Mo (VI ), exhibit this characterist ic boundstate transition.25The intensity of this transition varies directlywith the number of Mo=O bonds. Complexes with one or twoMc=O bonds exhibit a low-energy shoulder on the absorptionedge, while a distinct peak appe ars when thr ee or four Mo=Oare present. The low-energy feature un der discussion is clearlyvisible in the air-oxidized protein and model compo und dataof Figure 3. T he C p P edge exhibits no such feature, an d doublybound oxygen can, therefore, be excluded as a possible l i-gand .Th e edges of the fully reduced and dye-oxidized states(Figure 1) are extremely similar to the CpP edge. Since Cplfrand Cpldoalso exhibit smoo th, low-energy, single inflectionpoint edges, the argu men ts presented above for sulfur ligationand against te t rahedra l geometry and doubly bound oxygenhold for the se state s as well. The positions of all three inflectionpoints, al though not identical , are within experime ntal errorof being the sam e.28Given th e present (and following) evidence for sulfur l iga-tion, i t is conceivable that oxidation sta te changes of the mo-lybdenum would not produce edge changes la rger than ex-perimental erro r, and it is thus not possible from these data tosay wh ether or not the mo lybden um is involved in th e overallchan ge of protein oxidation level. At least i t can be said thatt he re a re no drastic differences in the Mo coordination sphere(as far as l igand or geometry changes a re concerned) betweenthese three states.Th e state which did exhibit significant differences in bothedge position and sh ape was the air-oxidized Clostridial Mo Feprotein. T he C plao edge ha s a distinct low-energy inflectionpoint at 20 004.9 eV, as well as a higher inflection point at20 016.6 eV29 (see Figu re 3). Both the s hape and position ofthe Cplaoedge are charac ter is t ic of Mo(V ) or Mo(V1) coor-dinated by tw o or more oxygen ligands. Fro m the intensity ofthe bound state transition shoulder i t appears l ikely tha t themolyb denum is l igated by one or two doubly bound oxygens.The presence of other less t ightly bound oxygens cannot beexcluded becaus e they would have little effect on the sha pe ofthe edge. However, three or four Mo= O would produce a re-solved low-energy peak .25Although the air-oxidized spectrumprovides no information ab out the physiologically active stat esof nitrogenase molyb denum , i t is a useful referenc e spectru mfor confi rming tha t the other curre nt Cpl spect ra do not cor-respond to 02- inac t iva ted m ater ia l .The EXAFS Region. The extended x-ray absorption finest ruc ture x ( k ) s defined as the mo dulation of the absorptioncoefficient p on th e high-energy side of the ab sorptio n edge:

In this expression po is the absorption coefficient that wouldbe observed for the absorber in the absence of neighboringsca t te r ing a toms, p is the smoothly varying part of the ob-served absorption, and k is the photoelectron wave vector.By making a number of approximat ions, one can arr ive a ta general expression for the EXAFS x ( k ) , n terms of calcu-latable physical quantit ies :

It lIzPI-n

xm

I

I I 1 I19990 20015 20040 19990 20315 20040ENERGY ( eV )Figure 3. A comparison of Cplao after both 4 h and 2 weeksof air expo-sure) with the two molybdenum compounds containing both oxygen andsulfur ligands. The pairs of inflection points (from top to bottom) were20 006.2 and 20 016.2, 20 005.6 and 20 016.2, 20 004.9 and 20 016.6, and20 004.9 and 20 014.9 eV. The Mo compound spectra are presented mainlyto show the effects of one or two MFO on the shape of the M o absorptionedge. The Is-4d transition is now a cle ar shoulder on the edge, producingthe low-energy inflection point in the absorption spectrum which appear sas a se parate peak in the derivative spectrum-a distinct difference fromthe edge of C pP .

X sin ( 2 k R a , aas(k ) )e -Lk2 ( 2 )In this expression, N s s the number of sca t te rers (s) at dis-tance R,, f rom the absorber (a) ; f , ( r , k ) is the scattererselectron backscattering amplitude; a,, is th e total phase shift,and CT; is the mean squ are deviation of R,, . Th e derivation ofthis expressi0n3~nvolves a numb er of app roxima tions whichbreak down a t low values of k (close to the absorption ed ge).Also, for scattering atom s beyond th e first coordination sphereone must account for additional effects which diminish thephotoelec t ron wave ampl i tude a nd/or change i t s phase as i t

propagates f rom the absorber .For the purpose of understanding the EX AF S of ni t roge-nase, i t is important to consider the physical quantit ies whichaffect the fine structur e. First of all , the oscil latory part of thefine structu re is related to th e argu men t of the sine functionin eq 2: 2kR, , a , , ( k ) . From this we see tha t thefrequencyof the E X A FS oscil lations is proportional to the dista nce be-tween the absorber and sca t te rer ( in the present case M o andthe surrounding l igands). Furtherm ore, the phas e shift a a , ( k )(due to electron-atom backsc attering and propagation o ut ofand back throug h the Co ulomb hole) will affect both the fre-quency and the absolute phas e of the EX AF S. Fina l ly, theEX AF S am plitude will be directly proportional to the numberof scatterers N , , weighted by their electron-atom backsca t-te r ing ampl i tudes f s ( r , k ) and damped by both an inversesquare distance dependence (1 / R ) nd a Debye-Waller-likeexponential depend ence on fluctuation s of R a s : -o:sk2. T h esumm ation index of eq 2 is over all scattering atom s in the vi-c inity of the absorber ; thus, the EX AF S of any absorber wi tha complicated coordin ation s phe re will be a superposition ofthe individual components. The reader unfamiliar withEX A FS is urged to consult ref 21 and references cited thereinfor fur ther de ta i l s on the physics of EXAFS and currentanalysis methods.Th e EX AF S of lyophil ized C p P is presented in Figure 4 .An important fea ture of the k space EXA FS data i s the pres-ence of a beat pattern w ith maxima nea r 6 and 12.5 A- anda minimum near 9.5 A - * . This modula t ion of the EXAFSamp litude envelope indicates th e presence of at least two fre-

8/13/2019 Cramer 008

5/10

3402 Journal of the Amer ican C hemica l Soc ie ty / OO : l l / M a y 24, 1978Table 111. Fourier Transform Calculations C

Peak Postulated scattererPosition Magnitude N 0 S Fe M o

A 1.43 0.055 1.86 I .84 1.89 1.83 1.821.86 0.137 2.29 2.27 2.32 2.26 2.252.32 0.08 1 2.75 2.73 2.78 2.720 2.7 1

(2.5) (1.5) (0.9) (0.5) (0.2)(9.6) (5.7) (3.5) (2.0) (0.9)(8.1) (4.9) (2.9) (1.7) (0.7)

1.31 0.035 1.7 8 1.75 1.80 1.75 1.70(1 .7 ) 1 O (0.6) (0.4) (0.2)1.84 0.109 2.3 1 2.28 2.33 2.28 2.23(9.0) (5.4) (3.2) (1.9) (1.0)2.29 0 .06 2.76 2.73 2.78 2.73b 2.68(7.0) (4.5) (2.5) (1.5) (0.8)

B

Using approximate CYFT f 0.4 A and M F Tof 0.35. Using approxim ate OFT of 0.44 A an d M F Tof 0.30 . A transform range 4-14 A-l;B, transform range 4-12 A - I . All distances are in A. Th e values in parentheses a re the calculated numbers of a toms.

I .o:

0 45

N.YX

0 0QXW

- 0.45

- 1.05J

5.0 9.0 13.0PHOTOELECTRON WAVE VECTOR k ( I - )

Figure 4. The original (light) an d Fourier-filtered (dar k) EX AFS of ly-ophilized Cplsr. The Fourier filter ing involved transforming the k-sp acedat a (from 3 to 14.5.&-I) to R space. then back-transforming to k spaceonly those components between 1 and 3 A . Note the beat pattern in theE X A F S ampli tude envelope, indicating the presence of a t least two dif-ferent absorber-scatterer distances.quency components-that is, at least two different absorb er-scatterer distances. The frequency of this beat patte rn, givenby Vbeat 2 ~ 1 6 . 5 -I 1 A , indica tes tha t th e two majorcomponents of the EX AF S have a f requency di fference ofabout 1 A . Since the frequency of the oscil latory par t of eachcomponent (neglecting the phase shift) depends upon twice theabsorber-sca tterer distance, the difference in distances wouldbe about 0.5 i f the phase shif t aa , (k) had approximate lyequal s lope and cu rvature .Th e above discussion indicates th at even a relatively simpleexaminat ion of the Cplsr EX AF S shows tha t a t least two di f-

ferent kinds of neighboring ato ms have to be accou nted for inan analysis of the M o environment . In order to ext rac t themaximum amount of informat ion avai lable f rom the f inestructure, the phase, frequency, and m agnitude of the EX AF Scomponents have been analyzed to determine the types ofatom s close to Mo, the ir distances to Mo, and the num ber ineach coordination shell.Fourier Transform Analysis. Use of the Fourier t ransformto analyze EXAFS was int roduced by Sayers, Lyt le , andStern.30 They have shown th at th e major peaks in th e trans-form correspond to the important absorber-scatterer distances,but shifted to lower R by a few tent hs of an Angstrom. Fou riert ransformat ion of k-space E X A FS d ata i s a useful means ofobserving th e m ajor f requency components. The t ransformpeak height is related to the magnitude of a particular fre-quency comp onent, while th e peak position is related to /2 theaverage frequency of t hat comp onent.31 Onc e the effectivephase shift C~FT the difference between the observed peakposition and t rue distance ) and the effective per atom mag ni-t ude M F T the observed peak he ight X R2/N ave been ob-tained from analysis of model compounds, the transform peakpositions a nd peak heights observed for an unknow n compoundcan be used to predict scatterer distances and numbe rs. Theapplication of this procedure to a variety of molybdenumstruc tures has been discussed in deta il in ref 2 1.Th e Fourier transform spectrum of lyophilized CpP' (Figure5) in the range of 4-14 A-1 shows a major peak at 1.86 A, apeak of 60% relative intensity at 2.32 A, and a m inor peak of40% relative intensity at 1.43 A (th e low R peak a t 1 .43 Astrongly resembles the should ers caused by pha se shift non-linearity, as typified in Figu re 6 of ref 21). For t he preliminaryanalysis i t has been presumed th at each of these features cor-responds to a real Mo-scatterer distance, and a further as-sumption is that th e biochemically reasonable assignm ents forthe peak will be either N , 0 S, Fe, or Mo. Using previouslycalculated effective phase shifts and per a tom magnitu des, aset (Table 111) of possible s catterer distances and num bers canbe constructed. O ne can now comp are these calculated valueswith typical bond lengths (Table IV) in order to decide whichassignments a re chemically reasonable.Th e Fourier trans form calculations reveal that the primarypeak would correspond to abo ut nine or ten nitrogens a t 2.29A, five or six oxygens at 2.21 A , or three or four sulfurs at 2.32A. Com parison of these values with the typical Mo-X bondlengths of Tab le IV shows that th e most chemically reasonableassignment i s tha t the pr imary EX AF S component is due toabout th ree or four sulfurs. The higher R peak could be either

8/13/2019 Cramer 008

6/10

H o d g s o n e t a l . / M o l y b d e n u m S i t e of N i t r o g e n a s eTable IV. Typical Mo-X Bond Lengths'

3403

Bondlength, 8 Structure Ref2.2 12.222.232.24I .661.6951.711.741.761.851.921.931.942.1 12.1952.212.242.301.942.182.292.312.332.332.342.412.382.3852.492.762.782.552.57

2.672.802.85

abaddaegahiabk

mbPPn

C

C

j

C

m0C

PP

aC

4mb

L. T. J . Delbaere and C . K. Prout, Chem. Commun., 162 (1971).B. Spivack and Z. Dori, J . Chem. Soc., Dalton Trans. 1077 (1975).l J . R. Knox and C. K . Prout, Acta Crystallogr., Sect. B , 25, 1857(1969). L. Ricard, J . Estienne, P. Karagiannidis, P. Toledano, J.Fisher, A . Mitschler, and R . Weiss, J . Coord. Chem., 3,277 (1974).e F. A. Cotton and R. C . Elder, lnorg. Chem., 3,397 (1964). B. M .Gatehouse and P. Leverett, J . Chem. SO C.A , 849 (1969). g A. B.Blake, F. A. Cotton, and J . S. Wood, J . Am . Chem. Soc., 86 3024(1964). L. Ricard, C . Martin, R . West. and R . Weiss, lnorg. Chem.,14, 2300 (1975 ). B. Spivack and 2 Dori, J . Chem. S oc., DaltonTrans., 1 173 (1973). j J . J . Park, M . D. Glick, and J . L. Hoard, J . A m .Chem. Soc., 91 301 (1969). B. Spivack, Z . Dori, and E. 1 . Stiefel,lnorg. Nucl. Chem. Left., 11 501 (1975). W . P. Binnie, M . J .Redman, and W. J . Mallo, Inorg. Chem., 9 1449 (1970). nl M . G.B. Drew and A. Kay, d . Chem. SOC. , 1851 (1971). R. G. Dickinsonand L. Pauling, J . Am . Chem.SOC. 5,1466 (1923). M. G . B. Drewand A. Kay, J . Chem. SOC.A , 1846 (1971). p L. Dahl, private com-munication. 4 J . Dirand-C olin, L. Ricard, and R. W eiss, Inorg. Chim.Acta, 18 L21 (197 6). Abbreviations: Nimid, histidine imidazolenitrogen; x b , bridging x; Ocarb, carboxylate oxygen; Scysrysteinesulfur.sulfido-bridged iron or molybden um, while the lower R peak,if real , mus t be nitrogen or oxygen. Using t he transfo rm dis-tances an d scatterer num bers for starting values, a curve-fittinganalysis of the E XA FS h as been carried out. Fo r complicatedcoordination spheres with overlapping EXAFS Fourier

S R

p/V 1I E 3 . 4 5 6

R H IFigure 5. The Fourier transform (moduli only) of lyophilized Cpl and[Fe4S4(SCH2Ph)4I2-. f a Mo atom were substituted for a single Fe atomin the tetramer, while keeping all distances the same, the magnitude ofthe resulting Mo EXAFS Fourier transform would look v e r y similar tothe original Fe transform, but phase shifted a n extra 0.09 A to lowerR .

transform peaks, curve-fi tt ing procedures yield mor e precisedistances and better sca tterer num bers, as well as elementalidentification of the scatters involved.21Curve-Fitting Analysis. In a previous paper2' a procedurefor curve-fit t ing the extended fine struct ure by means of em-pirical phase shift and a mpl itud e functions has been describedin de ta i l. The t o t al EX AF S x ( k ) was described as the su m ofindividual absorb er-scatterer interactions, thus:x ( k ) = c Xas(k) (3)

S

where th e sum over scatterers can often neglect ato ms beyond3 8 f rom the absorber .32Th e individual in terac t ions wereparam etrized in terms of a pairwise amp litude function:A a , ( k )= coe- c lk2 /kc2 (4)

( 5 )and a pairwise phase shift :

Q a s ( k )= uo a l k a2k2Th e best values for eo, c1, c2, ao, u l and a2 were de terminedby curve-fitting the E XA FS of appropriate model compounds,and different values for Mo-C, Mo-N, Mo-0 , Mo-S, andMo-M o interactions were obtained. Since appropriate Mo-Femodels are not available, Mo-Fe parameters were obtainedthrough a combination of empirical and theoretical re-sults.33Th e tota l EXA FS of an unknown st ru c ture is f it ted wi th asum of damped sine waves whose amplitude envelopes andphase shi f ts a re predetermined by the above parameters:

In th e f i t t ing procedure, the param eters C I , c2, ao, U I , and a2are held to different fixed values depend ing on the elementalassignment of th e scatterer. Th e values of co and R,, ar e variedby the optimiza tion prog ram to give the best fit to the dat a. Ifthe e lementa l assignments for the sca t te rers a re correc t , thenR,, will be the distance from the absorber (in this case Mo) to

8/13/2019 Cramer 008

7/10

3404 Journal of the American Chemical Society / OO : l l / M a y 24, 1978I

* IX IlnXua 0 0

1.05 -0q45-I V5.0 9 0 13.0

PHOTOELECTRON WAVE V E C T O R k ( A - )Figure 6. Fourier-filtered C p P data -) fit with either a single Mo-S wave- - - - - ) or a single M o - 0 wave ....-il.v.).he phasesh ifts used wereempirical values from ref 21. The M o - 0 wave is clearly out of phase withmost of the d at a; it adjust s as well as possible by leading the Cpl EXAFSat low k and trailing it at high k . The single-shell f i t using Mo-S param-eters yielded values of 3.4 sulfurs a t 2.36 A while the fit using M o - 0param eters predicted 4.7 oxygens at 2.21 A. Fit range: 4-14 kweighting.

the scatterer, while the value of co can be used to obtain theapproxim ate number of scat terers of that type.Th e first fits on t h e fi lt er ed C p P EX AF S were des igned todetermine the source of the primary component of the f inestructure. The optimization program was run with a single sinewave, using the phase shift and amplitud e parameters fo r eitherM o - 0 or Mo-S, and varying both the numb er of atom s (viaC O ) and their distance to molybdenum (via R a J . Th e resultantfits (Figure 6) show that M o-S param eters describe the phaseof the EX AF S before the beat region substantially better th ando M o- 0 parameters . Since the s trongest EXA FS componentdominates the p hase of the f ine s t ructure in this region, thesefits support t he previous conclusions based on edge dat a abo utprimari ly sulfur l igat ion to th e molybden um.A successful curve-fitting analysis should produ ce a muchbetter fit than those of Figure 6, and it was, therefo re, impor-tant to know how many addi t ional components were requiredto adequa te ly reproduce the EXAFS. F rom the Four ie rt ransform data (Table 111), t appeared tha t th e next wave tobe added should correspond to a single mol bdenu m or severalsulfurs or i rons at a dis tance of about 2.7 Two-shell fits onthe C p P EXA FS were thus at tempted using ei ther (Mo-S andMo-Mo) , (Mo-S and Mo-S) , or (Mo-S and Mo-Fe) pa -rameters . Th e best f i ts (Figure 7 and Table V ) indicate thatthe second component is most likely a M o-F e interaction. Thisconclusion is based on two observations. First, the best (M o- SMo-F e) fit yielded a lower erro r function than an y (Mo-SMo-Mo) or (Mo -S Mo-S) fit. Second, the best fit usingMo-Mo parameters gave a negatiue M o am plitude, indicatingthat the Mo -Mo phase shift is substantially out of phase withthe high R component. Although these results do not yet provethe presence of Mo-Fe interactions, they m ake it very unlikelythat t he M o in ni trogenase is present as a bridged Mo- Modimer .Despite the improved fits obtained using two different waves,

II.051

Na

x I I a F Iv)XWLL OaO I

0.45

- 1.05II 5.0 9.0 13 0

PHOTOELECTRON W A V E VECTOR k ( I - )Figure 7. Fourier-filtered C p P data -) fit with a Mo-S wave and aMo-Fe wave - - - -). Fit range: 4-14 A - , k 6 weighting.

Table V. Two-Wave Curve-Fitting Res ultsoPostulated Calculated Calculated Minimizationscattering scatterer scatterer functionelements distances, A numbers value

S 2.359 3.3 1.030S 2.847 2.6S 2.361 3.4 1.4346M o 2.861 0.5S 2.366 4.4 0.893bMo 2.552 -0.6S 2.365 3.6 0.843Mo 2.705 -0.7S 2.360 3.3 0.753Fe 2.719 1.6

0 Fitting rang e 4-14 , -I, k6 weighting. Local minimum.

i t was only with three components that the E XA FS could beadequately reproduced. Although the Fourier transform mightbe interpreted as requiring a low-frequency oxygen or nitrogencomponent , the greates t improvement to the f i t came fromadding ano ther slightly longer Mo-S wave. Th e best three-shellfit on the ran ge of 4-1 4A-] is presented in Figu re 8, while thecalculated scatterer numbers and dis tances for this and thecorresponding 4-12 A-1 fit are presented in Table VI. T h equality of this fit is of the s ame orde r as the fits obtaine d onmodel compounds of known structure, and there app ears to beno justification for addition of new waves. No te tha t th e regionof greatest discrepancy between calc ulated an d observed (fil-tered) EX AF S is the beat region from 9 to l l A- , as seen inFigure 4. he precise shap e of this beat region is the greatestambiguity in the original spectra, and bette r data are requiredto determine whether the difference between calculated andobserved E XA FS is significant.

8/13/2019 Cramer 008

8/10

H o d g s o n e t a l . / M o l y b d e n u m S i t e of N i t r o g e n a s e 3405Table VI. Three-W ave Curve-Fitting Results

MinimizationFitting function SI SZ Ferange, A-1 value Mo-S, 8 No. Mo-S, 8 No. Mo-Fe, 8 No.4-12 0.434 2.352 3.8 2.495 1.1 2.721 1.44-14 0.501 2.353 3.8 2.494 1 . 1 2.722 1.4

DiscussionSince both x-ray absorption edges and EX AF S are new andunfam iliar tools for the stud y of metalloproteins, i t is usefulat this point to sum marize the conclus ions abo ut the Cpl mo-lybden um site, the metho ds and logic which led to these con-clusions, and, finally, the relative stringency of th e differen tfindings. The various co nclusions will be discussed separate lyand in order of decreasing certainty. At the end of this dis-cussion the curren t results will be used to propose a pair ofs t ructures compat ible with al l the data .Th e most clear-cut finding is that t he M o in nitrogenase issurrou nded primarily by sulfur ligands. The absorption edgeis in a low-energy region characteristic of all-sulfur ligation.Th e primary peak in the F ourier t ransform can be readily ac-counted for postulating three or four Mo-S distances of about2.32 A , while the al ternat ive explanat ion of f ive or s ix M o- 0

distance s of 2.27 8 ppears much less chemical ly r e a ~ o n a b le . ~ ~Th e single wave fits showed tha t th e absolute phase of th e lowk fine structu re is in very good agreemen t with empirical Mo-Sphase shifts , and s ignificant ly different from M o -0 values .Finally, the successful three-wave fit of th e E XA FS using twoMo-S waves and a M o-Fe wave makes the conclusion aboutsulfur ligation virtually certain .A second conclusion of nearly equal certainty is that noMo= O bonds exist in the fully reduc ed, semire duced , ordye-oxidized states of nitrogenase, and tha t air oxidation ofthe M oF e protein produces one or two such bon ds (and possiblysome M o -0 bonds). Th e evidence for this is the lack of alow-energy 1s-4d should er in the absorption edge of the firstthree s ta tes , and the a ppearance of such a feature upon airoxidat ion of Cpl . The C plfr ,Cplsr ,and C pldo dges all show asingle inflection point, while Cplao shows two inflection pointsat positions characteristic of Mo(V ) or Mo(V1) with doublybound oxygen. There has been speculation3 that one of theroles of AT P in nitrogenase action might be in the removal ofsuch oxygen, but the curren t edge data indicate that only theirreversibly oxidized Cp lao contain s M-0. Furthermore, theextreme similarity of the first three edges indicates that whilethe M o oxidat ion s tate m ay or may not change between thesedifferent s ta tes , no major changes of the M o coordinat ionsphere occur.Examinat ion of the E XAF S, both d irect ly in k space orFourier t ransformed in R space, yielded the third result, theexistenc e of a second set of scatterers abou t 0.4 or 0.58 urtherfrom the M o than the primary sulfur l igands. The exis tenceof this second EXAFS component is revealed by the beatpat tern in the EXA FS (between 9 and 1 1 as well as bythe Fourier transform peak a t 2.32 8 unshifted). A two-shellfitting analysis indicated tha t th e phase of the second compo-nent is significantly different from the empirical Mo -Mo phaseshift, and reasonably in phase with a calculated Mo-Fe phaseshift. Therefore, this second component is probably not due toMo, while the same analysis makes a M o-Fe assignment veryreasonable.Unfortunately, Mo-Fe compounds with isolatable Mo-FeEX AF S do not exis t and, therefore, the empirical methods ofref 21 for the extraction of pairwise phase shift and amp litudefunctions could not be appl ied. T he M o-Fe ampli tudes andphase shifts were calculated by a straightforward method,33but unlike the other parametrized functions, they have not been

1-05

0.45N

XE 0 caXW

-0.45

- 1.05I5.0 9 0 13 0

PHOTOELECTRONW A V E VECTOR k (a-1)Figure 8. Fourier-filtered C p P data -) l i t with twodiffere nt Mo-S wavesand a Mo-Fe wave - - - - - ) . Fit range: 4-14 . - I , k 6 weighting.

tested on known structures. The assignment of the secondEX AF S component as a Mo-Fe interact ion, a l though rea-sonable, is therefo re not as unamb iguou s as the first two con-clusions.Th e final conclusions, stem ming from the three-shell fit ofFigure 8 and summ arized in Table VI, are that M o in lyophi-lized C p P s surrounded by three or four sulfurs at 2.35 f .03A , one or two sulfurs a t 2.49 f .03 A , and two or three ironsat 2.72 0.05 A . Th e estimated accu racy is based on previousresults2 rath er th an on statistical errors, since it is systematicerrors which a re the greatest source of uncertainty in EX AF Sstructure determinat ion.Experience with multishell fits on model compounds ofknown structure showed that d istances were always calculatedto bet ter than 0.03 8 when the correct scat tering atoms werepostulated. Furthermore, th e calculated scat terer numberswere generally correct to within an atom. T hus th e calculatedvalues of 3.5 S at 2.35 A should be interpreted as three or foursulfurs with a n average distan ce inside the ran ge of 2.32-2.38from the m olybdenum. Since this component dominates theEX AF S, these values are the firmest number s to be extractedfrom th e fits.Assuming tha t th e longer distance secondary com ponent tothe E XA FS arises from a M o-Fe interact ion, the three-wavefit calculates I .4 Fe atoms a t 2.72 8 . ince th e Fe is most likelybridged through S to the Mo , this calculated num ber of ironsshould be considered a lower limit on the tru e value, based onthe following considerations. First, atoms beyond the prim arycoordination sphere (not directly bound to the absorber) havegreater thermal motion relative to the absorbe r than d o ligatedatoms, and increased thermal m otion causes a reduction in the

8/13/2019 Cramer 008

9/10

3406 Journal of the Amer ican C hemica l Soc ie ty / OO : I l / M a y 24, 1978EX AF S amp li tude. For example, the sulfido-bridged Fe-Fedist ance in th e 2Fe-2 S c luste r, [Fe2S2(S2-o-xy1)2I2-, is es-sentially unobservable in th e EX AF S of this compound,25whilethe Fe-Fe distance is quite distinct in the EX AF S of the morerigid 4Fe-4S cluster [ F ~ ~ S ~ ( S C H ~ C ~ H S ) ~ ] - ~see Figure 5 .Since the ampli tude calculat ions were based on a model withtightly b ound ligands [Le., M o ( S ~ C ~ H ~ ) ~ ] ,hey will un der-es t imate the number of more dis tant scat terers required toproduce a given effect on the f ine s t ructure.A second effect which reduces the EX AF S contribution ofmore d istant shells is depletion of th e outgo ing photoelectronwave by intervening scatterers. Thu s, the presence of sulfuratoms between the mo lybdenum (absorber) and i ron (scat-terer) might cause an addi t ional reduct ion in the Mo-FeEX AF S. Although it is difficult to quantitate th ermal motionand depletion effects, one should note tha t for sulfido-bridgedMo dimers , the st rength of the Mo-M o component is about50% of that expected by comparison with oxo-bridged di-m ew 2 ' Thus , the calculated value of 1.4 Fe ato ms could inreality easily represent a contribution from two or three irons.Final ly, the calculated Mo-Fe dis tance (2.72 0.05 A ) hasa greater uncertainty than the Mo-S dis tances owing to pos-sible inaccuracy in the calculated M o-Fe phase shift.Th e third compon ent of the three-wave fit may correspondto one or two sulfurs at a n averag e distance of 2.49 0.03 A .This com ponen t was necessary to produce a n acceptab le fit ,but it is conceivable tha t better Mo-F e phase shift and am -plitude functions would make a third component unnecessary.On the other hand, a fourth wave corresponding to a singleMo-N dis tance would not be detectable . Al though the cal-culated num bers a re chemical ly reasonable, the ampli tude ofthis component is only 25% that of the primary Mo-S wave,and these values are therefore subject to greater uncertain-ty.Confronted with these three pairs of scatterer num bers anddistances, and th e general information about a n all (or almostal l ) sulfur , non-Mo=O environment , a model for the Mo s i teof nitrogenase can be formulated. It is useful to begin bycomparing the calculated distances with bond lengths of knownstructures (Table IV). Th e calculated Mo-S dis tance of 2.358 is sho rter than known Mo-S,,, bond lengths, and slightlylonger than known Mo-Sb(bridging) bond lengths. An as-signme nt as MO(IV)-Sb or MO(III)- Sb is reasonable in viewof the longer M O - S b bond len gths for lower oxidation sta tes.The second Mo-S dis tance (2.49 A ) is in a region typical ofMo-S,,, bond lengths, although methionine or even persulfideS cannot be rejected. Finally, the Mo-Fe distance of 2.72 8is close to the value of 2.76 8 observed in (v5-C5H5)3M03-F ~ S ~ B I - . ~ ~Th e calculated numb ers and distances of sulfurs and ironsappear chemically reasonable, and they may be interpretedthrough comparison with known structures as representingthree or four bridging sulfides, one or two termina l cysteinesulfurs, and two or thre e sulfido-bridged irons. With this in-terpretatio n, at least two distinct models for the nitrogenaseM o site may be proposed (I and 11).

/

I1IThe essential feature of both of these models is the proximityof iron, bridged by sulfur, to molybdenum, with an extra li-gan d(s) (perhap s cysteine) on the M o. Both of these modelssatisfy all of the constraints of th e curren t absorption edge andEX AF S information, and each has its own special appeal. The

Mo-Fe-S cube 1) is a variation on well-known 4Fe-4S clus-ters, and inorgan ic clusters contain ing molybdenum an d ironhave been prepared and c rys ta l lographical ly c h a r a ~ t e r i z e d ~ ~(unfortunately these structures contain 3M o- 1Fe rather than1Mo-3Fe). A bridging s t ructure s imilar to 11 has been pro-posed for [ ( M O S ~ ) ~ F ~ ] * - , ~ ~ut the la t ter compound also hasan inverted Fe-Mo rat io.It is temp ting to envision N2 interaction s simultaneouslywith both the M o and Fe in these s tructures , or modificationsthereof. However, it should be emphasized that none of thecurrent d ata can address the problem of th e s i te of Nz reduc-tion. Furthermore, although both of these models involvefive-coordinate M o, the da ta can not rule out a s ix-coordinatespecies. A single nitrogen or oxygen ligand would pr obably bemissed in the EX AF S, given the curren t signal to noise ratio.Alternatively, a set of several cysteine ligand s with a widespread in Mo-S bond len gth s would yield a decep tively weakcontribut ion to the EX AF S. Therefore, further experimentsare necessary to test for coordinative unsaturatio n of the M osite.Summary

X-ray absorption spectroscopy has proven to be a useful toolfor study of the M o site in nitrogenase. Th e combined ab-sorption edge and E XA FS d ata a re indicat ive of primari lysulfur ligation, and curve-fitting analysis of th e fine stru ctur eindicates the existence of a Mo- Fe-S cluster. Althou gh thelimited sensitivity of this techniqu e necessitates the use of veryconcentrated samples, x-ray absorption spectroscopy is atpresent the only spectroscopic method for unambiguouslyobserving the M o under various conditions. F utu re experimentsusing CO, N2, C2H2, and Ar atmospheres may help clarify thenatu re of the small molecule interaction site(s) in this protein.Collection of extended fine structure dat a on the fully reducedand dye-oxidized states could help determine whether theMo-F e-S species is involved in the EPR -observ ed oxidationlevel changes. Furthermore, examination of the Fe edges inFe-Mo cofactor , demolybdo component I, and in tact MoF eprotein und er various conditions will yield information abo utthe types of F e present and their role in the cataly tic process.T he availability of powerful synchrotron radiation sources hasopened a new spectroscopic window on nitrogenase and me-talloproteins in general, and a wide ran ge of novel experimentsis now ac cessible.

Acknowledgments. We would l ike to thank Sebast ianDoniac h for helpful discussions concerning E XA FS analysis.W e would also like to acknowledge To m Eccles for the com-puter pro gramm ing of both da ta collection and data analysispackages and He rman Winick and the members of the Stan-ford Synchrotron Radiation Laboratory (SSRL) staff for theirtechnical assistance in suppo rt of this project. K. 0 .Hodgsonis a Fellow of the A lfred P. Sloan Foundation and S.P. Cramerwas an IBM Doctora l Fellow for 1976-1977. This work wassupported by the Nation al Science Foundation through Gra ntPCM-75-17105 and through Gr an t GB-22629 (Mortenson).Synchro tron Radiation time was provided by SS RL , supportedby the Nat ional Science Foundat ion G rant D MR-07692-A02in cooperation with the Stanf ord Linear Acceleration C enterand th e Energy Research and Development Adminis t rat ion.References and Note s

(1) (a) Stanford University: ( b )Purdue University.(2) (a)T. C. Huang. W. G. Zumfl, and L. E. Mortenson,J Bacterial., 113,884(1973); ( b )R. W. Hardy, R . C. Burns, and G. W. Parshall in Inorganic Bio-chemistry , G. L. Eichhorn, Ed., Elsevier, Amsterdam, 1973, pp 745-793.(3) R. R. Eady,B. E. Smith, K.A . Gook, and J. R. Postgate, Biochern J., 128,655 (1972).(4) (a)J. S. Chen, J. S.Multani. and L. E. Mortenson, Biochim. Biophys.Acta.310 51 (1973). ( b )The magnetic circular dichroism of the semireduced

8/13/2019 Cramer 008

10/10

T r o s t , Strege, W e b e r , Fullerton, D i e t s c h e / r - A l l y l pa l l ad i um C o m p l e x e s r o m Ole f ins 3401MoFe protein from 600 o 350 nm is slightly positive and monotonicallyincreasing-unpublished results of the authors with Dr. John Dawson.(5) G. Palmer, J. S. Multani, W. C. Cretney, W. G. Zumft, and L. E. Mortenson,Arch. Biochem. Biophys., 153 325 (1972).(6)B. E. Smith, D. J. Lowe, and R . C. Bray, Biochem. J., 135, 31 (1973).(7)W. H. Orme-Johnson and L. C. Davis in Iron-Sulfur Proteins , Vol. 3,W.Lovenberg. Ed., Academic Press, New York, N.Y., 1977, 31.8) As the authors of ref 7have noted, the Momight be present in an Em-activestructure with insufficient spin density localized on the metal for observationof the electron-nucleus interaction.(9)Even the number of Mo atoms necessary for a functional MoFe moleculeis still in some doubt (ref 7,p 25).The nutritional requirement is clearlynecessary but not sufficient evidence for the involvement of Mo in nitro-genase.(10) a)R. G.Shulman, P. Eisenberger, W. E. Blumberg, and N. A. Stombaugh,Proc. NaN. Acad. Sci. U.S.A., 72 4003 (1975);b) P. Eisenberger. R. G.Shulmann, G. S. Brown, and S.Ogawa. ibid., 73 491 (1976).11) (a)V. W. Hu, S. .Chan, and G. S. Brown, Proc. Natl Acad. Sci. U.S.A., 74,3921 (1977);b) T. K. Eccles, PhD. Thesis, Stanford University, 1977.(12) . P. Cramer, T. K . Eccles. F. Kutzler, K.0. Hodgson, and L. E. Mortenson,J. Am. Chem. Soc., 98, 1287 (1976).(13) hroughout this paper, we refer to oxo groups tightly bound to molybdenum,whose short (1.6-1.8 Mo-0 bond lengths are indicative of multiplebonding, as doubly bound oxygens.(14)W. G. umftand L. E. Mortenson, Eur. J Biochem., 35 401 (1973).(15)M. Walker and L. E. Mortenson, J Biol Chem., 249, 6356 (1973).(16)W. G. umft, L. E. Mortenson, and G. Palmer, Eur. J . Biochem., 46, 25

(17) . E. Mortenson, W. G. Zumft, and G. Palmer,Biochim Biophys. Acta, 292,18) W. G. Zumft and L. E. Mortenson, Biochim. Biophys. Acta, 416,(19) . M. Kincaid, Ph.D. Thesis, Stanford University, 1974.(20) . A. Bearden, Rev. Mod. Phys., 39, 8 1967).(21) .P. Cramer, K.0. Hodgson, E. I.Stiefel, and William E. Newton, J Am.

Chem. SOC.,100 2748 (1978).(22) .P. Cramer, K. 0. Hodgson, and E. I. Stiefel, manuscript in prepara-tion.(23)For a review of absorption edge spectroscopy prior to synchrotron radiationdevelopments, see U. C. Srivastava and H. .Nigam, Coord. Chem. Rev.,9 , 275 (1973).(24) . G.Shulman, Y. Yafet, P. Eisenberger, and W. E. Biumberg, Proc. Nati.Acad. Sci. U.S.A.. 73 1384 (1976).(25) . . Cramer, Ph.D. Thesis, Stanford University, 1977, hapter 6.(26) or lack of suitable all-N complex data, we present data on Mo chlorides.Since the electronegativity of nitrogen is between that of chlorine andoxygen, Mo-N complexes should exhibit edges intermediate to the chlorideand oxide edges, for a given metal oxidation state.(27)Appellations such as Is-3d and Is-4s are to some extent misleading, sinceit is the very fact that the final state is a mixed state with some p characterwhich makes these transitions allowed. The terms are useful in designatingthe predominant character of the final state, however.(28)Measurement of the Mo edge inflection point of a dilute sample is subjectto an extra source of error because of the protein background absorption,

(1974).422 1973).(1975).

which was almost 99 of the total absorption (A) or the solution experi-ments. The inflection point E,,, is defined as the energy at which the secondderivative d2 A/ dP = 0. If the protein background were a linear functionof energy, Aback= a bE, its second derivative would be zero and it wouldnot affect the measured inflection point. However, in practice the proteinback round is a decreasing signal with positive curvature: Aba& = a bEc 8 , b < 0, c > 0. The positive background curvature will shift the ob-served inflection poit to higher energies as the sample becomes more di-lute. It would thus be desirable to repeat these edges with protein samplesof exactly the same Mo concentration.(29) t should be noted here that the air-oxidized edges reported here are es-sentially the same as the edge previously reported for resting-state'' ni-trogenase (ref 12). t now appears that so-called resting-state'' nitrogenasewas permanently at rest, that is, irreversibly air oxidized(30) . A. Stern, D. E. Sayers, and F. W. Lytle, Phys. Rev. Sect. B, 11 48361975).(31)We note here that the EXAFS Fourier transform abscissa, labeled R , is0.5X he mathematical transform dimension; thus th frequency differenceobserved in the transform is one-half the difference observed by directanalysis of the EXAFS in k space.(32) n structures such as metalloporphyrins and pure metals, which containhighly symmetric shells of atoms around the absorbing atom, EXAFScomponents for atoms as far as 5 A are easily observed. However, for mostof the Mocompounds studied in ref 21, toms without a bonding interactionwith Mo were not easily seen.(33) he Mo-Fe phase shift, O LM ~ - F ~as calculated as the sum of (1)he the-oretical Fe scatterer phase shift and (2)he average difference betweenempirical Mo-C, Mo-0, and Mo-S and theoretical C, 0, and S phase shifts.We are, therefore, using the fact that O L M ~ - F ~aF,(calcd) [aMo-r(em-pirical) - ax(calcd)], where the quantity in brackets is the average dif-ference between empirical pairwise phase shifts and calculated scattererphase shifts. The theoretical phase shifts were taken from ref 34,whilethe procedure for using relationships between phase shifts is a modificationof that in ref 35.Since we do not know the shapeor magnitude of theMoFeamplitude function, we have assumed that it will be similar to the Mo-Samplitude, only larger by a factor of ZFe/Zsor 26/16. he approximatelylinear Zdependence of EXAFS amplitudes is discussed in ref 21.(34)P A. Lee, B. K. Teo, and A. L. Simons, J. Am. Chem. Soc., 99, 3856(1977).(35)P.H. Citrin, P. Eisenberger, and B. M. Kincaid, Phys. Rev.Lett. 36 1346(1976).(36) he long Mo-0 bond length of 2.30 in the oxo-bridged molybdenumcysteine dimer has been attributed o ligand constraints and a Mo=O transeffect.(37) a) E. it Stiefel and J. K. Gardner, Proceeding of the First InternationalConference on the Chemistry and Uses of Molybdenum , P. C. H. Mitchell,Ed.. London, 1973, p 272-277: b) R. J. P. Williams and R. A. D. Went-worth, ibid., pp 212-215: c) R W. F. Hardy and E. Knight, Jr., Prog. Phy-tochem., 1 212-215 (1968):d)G.W. Parshall, J. Am. Chem. SOC.,89,1822 (1967).(38) a) E. Stern, Phys. Rev. Sect. B, 10 3027 (1974);b) C. A. Ashley and S.Doniach, ibid., 11 1279 (1975).(39) . Dahl, private communication.(40)E. Diemann and A Mueller, Coord. Chem. Rev., 10 79 (1973).

Allylic Alkylation: Preparatio n of.;rr-Allylpalladium Complexes from OlefinsBarry M . Trost,* Paul E. Strege, Lothar Weber,Terry J. Fullerton, and Thom as J. DietscheContribution r om the Department of Chem istry, University of Wisconsin,Wisconsin 53706. Received October 20, 1977

Abstract: Allylic alkylation re quires activation of th e allylic C- H bond and formation of a C-C bond.A mild way for achievingthis activation is developed. A general proc edure to obtain .rr-allylpalladium complexes from simple and complex olefins ingood yields is described. The regio- and chemoselectivity is examined. A mechanistic rationale is put fo rth.

Th e advent of the Wack er process init iated a flood of ac-tivity in organo palladium chem istry. In studying the mech-anism of this reaction, the possibil i ty that 9-allylpalladiumcomplexes could be form ed as by-products leading to allylicoxidation was noted. Th e possibil i ty tha t orga nometa ll ic re-agents could be directly available from olefins under mildconditions attracted ou r attention a s a route to allylic alkyla-tion. Specifically, we became concerned with the problemoutlined in eq 1 , the ability to replace an allylic C-H bond with

1 XT I VAT I O N SUBSTITUTION, 1 Eq IH R

an allylic C C bond. Such a process can be envisioned to in-volve two stages, activation followed by subs ti tution. Indeed,allylic halogenation followed by coupling with an organorne-t a l k such as a cup ra te const itutes such a process. Li thia t ionfollowed by coupling with a n alkylating agen t or a carbony lcompound als o consti tute s a reasonable approac h. These ap-