Investigation of metal–dithiolate fold angle effects:Implications for molybdenum and tungsten enzymesHemant K. Joshi, J. Jon A. Cooney, Frank E. Inscore, Nadine E. Gruhn, Dennis L. Lichtenberger†, and John H. Enemark†

Department of Chemistry, University of Arizona, Tucson, AZ 85721

Edited by Jack Halpern, University of Chicago, Chicago, IL, and approved January 31, 2003 (received for review November 8, 2002)

Gas-phase photoelectron spectroscopy and density functionaltheory have been used to investigate the interactions betweenthe sulfur �-orbitals of arene dithiolates and high-valent transi-tion metals as minimum molecular models of the active site fea-tures of pyranopterin Mo�W enzymes. The compounds(Tp*)MoO(bdt) (compound 1), Cp2Mo(bdt) (compound 2), andCp2Ti(bdt) (compound 3) [where Tp* is hydrotris(3,5-dimethyl-1-pyrazolyl)borate, bdt is 1,2-benzenedithiolate, and Cp is �5-cyclopentadienyl] provide access to three different electronicconfigurations of the metal, formally d1, d2, and d0, respectively.The gas-phase photoelectron spectra show that ionizations fromoccupied metal and sulfur based valence orbitals are moreclearly observed in compounds 2 and 3 than in compound 1.The observed ionization energies and characters compare verywell with those calculated by density functional theory. A‘‘dithiolate-folding-effect’’ involving an interaction of the metalin-plane and sulfur-� orbitals is proposed to be a factor in theelectron transfer reactions that regenerate the active sites ofmolybdenum and tungsten enzymes.

Coordination by the sulfur atoms of one or two ene-1,2-dithiolate (dithiolene) ligands of the novel substituted

pyranopterin-dithiolate (‘‘molybdopterin’’; ref. 1) is a commonstructural feature of mononuclear molybdenum-containingenzymes (2–4). These enzymes catalyze a wide range of oxi-dation�reduction reactions in carbon, sulfur, and nitrogen me-tabolism. Fig. 1 shows the structure of the active site of sulfiteoxidase, a representative example (5, 6) of the coordinationof the pyranopterin-dithiolate (hereafter abbreviated S2pdt;ref. 7). These structural results raise fundamental questionsabout the role of the S2pdt coordination in the overall cata-lytic cycle of molybdenum enzymes (8). The unusual ability ofene-1,2-dithiolate ligands to stabilize metals in multiple oxida-tion states has been recognized since the compounds werefirst investigated (9). Proposed roles for the S2pdt ligand in-clude functioning as an electron transfer conduit from themetal to other prosthetic groups (10) and as a modulator ofthe oxidation�reduction potential of the metal site (10). Dur-ing catalysis, the metal center is proposed to pass through theM(VI�V�IV) oxidation states, i.e., the Mo d electron countchanges from d0 to d1 to d2. Thus, studies of discrete metaldithiolate complexes encompassing these and related electronconfigurations may provide insight concerning metal thiolatebonding and reactivity in enzymes.

Previous structural studies of model molybdenum com-plexes of the type (Tp*)MoE(1,2-dithiolate) [where E is O orNO, Tp* is hydrotris(3,5-dimethyl-1-pyrazolyl)borate, and the1,2-dithiolates are bdt (1,2-benzenedithiolate), bdtCl2 (3,6-dichlorobenzenedithiolate), and qdt (2,3-quinoxalinedithio-late)] have shown that the fold angle of the dithiolate metal-lacycle along the S���S vector (Fig. 2) varies in a way thatdepends on the occupation of a d orbital that is in the equa-torial plane (11). For (Tp*)MoO(bdt), which has a formal 4d1

electron configuration, the fold angle is 21.3° (12, 13). How-ever, for (Tp*)Mo(NO)(bdt), the fold angle is 41.1° (11). Al-though the metal center of (Tp*)Mo(NO)(bdt) is formally d4,the strong �-acceptor character of the NO ligand results in

the d orbital in the equatorial plane being empty, whereas for(Tp*)MoO(bdt) this d orbital is half-filled (11). The largerfold angle for (Tp*)Mo(NO)(bdt) has been ascribed to a sta-bilizing interaction of the filled p� orbitals on sulfur with theempty metal orbital in the equatorial plane on bending (11),which is illustrated schematically in Fig. 2. For other reported(Tp*)Mo(NO)(dithiolate) compounds, the fold angle is always40–45° and is essentially independent of the electron-donatingor electron-withdrawing nature of substituents on the dithio-late ligand (11). In contrast, for (Tp*)MoO(dithiolate) com-plexes the fold angle ranges from 6.9° in (Tp*)MoO(bdtCl2)(14) to 29.5° in (Tp*)MoO(qdt) (11), suggesting a much shal-lower potential surface for the interaction of the filled sul-fur-p� (S�) orbitals with the half-filled equatorial metal or-bital (metal in-plane or Mip). For comparison, the calculatedfold angles for dithiolates in protein structures (Table 1)range from 7–30° (5, 15–17).

The (Tp*)MoO(dithiolate) compounds have also been in-vestigated by a variety of spectroscopic techniques, includingEPR (12, 13), electronic absorption, resonance Raman, andmagnetic CD spectroscopies (10, 18, 19). The general conclu-sion from these studies is that the singly occupied molecularorbital in these formally d1 centers has substantial sulfur char-acter (10, 18, 19). From these studies it was proposed thatpseudo-� interactions between the in-plane metal orbital andsulfur in-plane lone pairs play an important role in electrontransfer reactions that regenerate the active sites of enzymes(10). However, Fig. 2 suggests that sulfur out-of-plane orbitalscould also have a major effect on these processes. To date,the role of fold angle on metal–sulfur interactions has notbeen quantitatively assessed.

A technique capable of assessing the metal and ligand char-acter of orbitals is gas-phase photoelectron spectroscopyusing ionization sources with different photon energies (20).Previously we have reported photoelectron spectra for(Tp*)MoE(tdt) (where E � O, S, or NO, and tdt is 3,4-toluenedithiolate) by using HeI, HeII and NeI photonsources that indicate strong mixing between Mo and S or-bitals (21). Interestingly, the first ionization energies of(Tp*)Mo(NO)(tdt), a formally d4 metal center, and(Tp*)MoO(tdt), a formally d1 metal center, differ by only0.05 eV, despite the variation in axial ligand and total numberof metal d electrons. This similarity in ionization energies wasascribed to the ‘‘electronic buffer effect’’ of the dithiolate li-gand (21). Subsequent photoelectron spectroscopic studies ofother pairs of (Tp*)MoE(dithiolate) complexes (E � O, NO)have shown that the first ionization energies of the membersof each pair are within �0.2 eV of one another. For a seriesof (Tp*)MoO(dithiolate) complexes with varying peripheral

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: HOMO, highest occupied molecular orbital; SCF, self-consistent field.

Data deposition: The atomic coordinates for 2 have been deposited in the CambridgeStructural Database, Cambridge Crystallographic Data Centre, Cambridge CB2 1EZ, UnitedKingdom (CSD reference no. 201539).

†To whom correspondence may be addressed. E-mail: dlichten@u.arizona.edu orjenemark@u.arizona.edu.

www.pnas.org�cgi�doi�10.1073�pnas.0636832100 PNAS � April 1, 2003 � vol. 100 � no. 7 � 3719–3724

CHEM

ISTR

YSP

ECIA

LFE

ATU

RE

substitutions on the dithiolate ligand, the first ionization ener-gies show a range of �0.7 eV and correlate with the lowestenergy oxidation potential in solution (11, 22). Gas-phasephotoelectron spectroscopy avoids possible complications ofsolvent and solid state effects, but a major drawback is thatonly neutral species are easily studied by this technique, andthe (Tp*)MoO(dithiolate) system does not allow access to theimportant d0 and d2 electron configurations of the catalyticcycle. Additional analysis of the molybdenum-dithiolate inter-action in the photoelectron spectra of (Tp*)MoO(dithiolate)is also difficult because the broad area of ionizations from theTp* ligand is not well separated from the region containingMo 4d and S 3p ionizations (11, 14, 21, 23). A third potentialcomplication of interpretation of photoelectron spectra of(Tp*)MoO(dithiolate) complexes is the formation of both sin-glet and triplet states from ionization of such a d1 system,though we have seen no indication that this is a problem (24).Thus, we have further sought a simple system that can pro-vide experimental gas-phase photoelectron data for formallyd0 and d2 metal dithiolate complexes and which is amenableto density functional theory calculations with a minimum ofsimplifying assumptions.

The known bent-metallocene dithiolate compounds[Cp2M(dithiolate), M � Ti, Mo and Cp is �5-cyclopentadi-enyl] provide access to d0 (Ti) and d2 (Mo) electron configu-rations (25, 26). A general bonding description of Cp2MX2compounds is well understood (27–29), and Lauher and Hoff-mann (27) first explained that the variation in fold angle forCp2M(dithiolate) compounds is caused by the occupancy ofthe metal d orbital in the equatorial plane (Mip) with respectto the dithiolate ligand. This orbital is empty for the folded d0

(Ti) case and filled for the more nearly planar d2 (Mo) case.

For the (Tp*)MoO(dithiolate) molecules that we have previ-ously studied there is also a metal-d orbital in the equatorialplane with respect to the dithiolate ligand that is, in this case,half-occupied for the d1 [Mo(V)] electron configuration.Fourmigue and coworkers (26, 30, 31) have studiedCp2M(dithiolate) and [CpM(dithiolate)2]�1,0 in the context ofnovel molecular materials. They have found that the d0 com-pounds, Cp2Ti(dithiolate), have large fold angles (e.g., 46.0°for Cp2Ti(bdt); ref. 26) in the solid state. The barrier to inter-conversion between the positive and negative extremes of foldangle in solution has been determined by NMR to be 14kcal�mol�1 (25, 30), close to the computed value of 15kcal�mol�1 (31). The observed folding for the Ti d0 systemscould facilitate interaction of the filled S� orbitals with theempty Mip orbital, similar to the diagram shown in Fig. 2. Incontrast, for the d2 metal system, Cp2Mo(bdt), the fold anglein the solid state is 9.0° (32). Fourmigue et al. (33) havefound that the oxidation potentials of Cp2M(dithiolate) com-pounds are not significantly different for M � Mo and W,suggesting that the highest occupied molecular orbital(HOMO) has substantial sulfur character. Pilato et al. (34)have prepared Cp2Mo(dithiolate) compounds in which thedithiolate contains an appended pterin group as mimics forpyranopterin-dithiolate centers. Molecular orbital calculationsand EPR data for [Cp2Mo(dithiolate)]� cations indicate thatthe HOMO is significantly ligand based (33–38). A study of(�5-tBuC5H4)2Zr(Se2C6H4) (39) demonstrated that these sys-tems are amenable to gas-phase photoelectron spectroscopy.To our knowledge, however, there has been no detailedinvestigation of the gas-phase photoelectron spectra ofCp2Mo(dithiolate) compounds to experimentally probe themolybdenum and sulfur character in the occupied valenceorbitals of these systems to assess whether the valencemolecular orbitals are primarily metal, primarily sulfur, orstrongly mixed. Here we present gas-phase PES data for(Tp*)MoO(bdt) (1) and Cp2M(bdt) [M � Mo (2), Ti (3)].The analysis of the experimental data are aided by densityfunctional theory calculations on the neutral ground stateof the molecules 1-3, and the ion states formed by photo-ionization.

Materials and MethodsSynthesis of Compounds. The compounds (Tp*)MoO(bdt) (12),Cp2Mo(bdt) (32), and Cp2Ti(bdt) (40, 41) were synthesizedaccording to published procedures and under anaerobic con-ditions by using either an inert atmosphere glove bag or stan-dard Schlenk line techniques. The modified synthesis ofCp2Mo(bdt) involved a 1:3 mixture of water and organic sol-vent to give better yield of the final product. Electronic ab-sorption (1,2-dichloroethane solutions on a modified Cary 14with OLIS interface, 250–900 nm), EPR (frozen glasses at 77K, in dry degassed toluene, at �9.1 GHz with a Bruker ESP300) and infrared (KBr disks on a Nicolet Avatar ESP Fou-rier transform infrared, 4,000–400 cm�1) spectroscopies and

Fig. 1. The active site of chicken liver sulfite oxidase illustrating a pyranop-terin dithiolate appended to the Mo center (5).

Fig. 2. The bonding interaction of the symmetric combination (S��) of sulfur

out-of-plane orbitals with the metal in-plane d orbital on folding of thedithiolate along the S���S vector indicated by the line.

Table 1. Fold angles measured for the dithiolate unit instructurally characterized representative pyranopterinMo enzymes

Enzyme Resolution, Å Fold angle, ° Ref.

Sulfite oxidase 1.9 6.6 and 7.0 5Aldehyde oxidoreductase

(MOP) (oxidized)1.28 16.6 15

Xanthine dehydrogenase 2.1 20.1 and 14.3 16Dimethyl sulfoxide reductase

(oxidized)1.3 18.2 and 33.1 17

3720 � www.pnas.org�cgi�doi�10.1073�pnas.0636832100 Joshi et al.

mass spectrometry (FAB ionization in a 3-nitrobenzyl alcoholmatrix on a JEOL HX110) were used to identify thecompounds.

Photoelectron Spectroscopy. Photoelectron spectra were re-corded by using an instrument, procedures and calibrationthat have been described in more detail (21). During datacollection, the instrument resolution (measured by using full-width half maximum of the argon 2P3/2 peak) was 0.020–0.030eV. The sublimation temperatures were (10�4 Torr, moni-tored by using a ‘‘K’’ type thermocouple passed through avacuum feedthrough and attached directly to the sample cell)195–205°C for compound 1, 190–200°C for compound 2, and175–185°C for compound 3.

In the figures the vertical length of each data mark repre-sents the experimental variance of that point. The valenceionization bands are represented analytically with the best fitof asymmetric Gaussian peaks (42). The number of peaksused in a fit was based solely on the features of a given bandprofile. The peak positions are reproducible to about �0.02eV (�3�). The parameters describing an individual Gaussianpeak are less certain when two or more peaks are close inenergy and overlap.

Theoretical Methods. The Amsterdam density functional theorypackage (ADF 2000.01) was used to study the electronicstructures of the compounds 1–3 (43–47). The optimized ge-ometry of 1 (Table 3, which is published as supporting infor-mation on the PNAS web site, www.pnas.org) was obtainedbeginning from the crystal structure geometry (Table 4, whichis published as supporting information on the PNAS web site)(13), and optimized geometries of 2 and 3 (Tables 5 and 6,which are published as supporting information on the PNASweb site) were obtained beginning from geometric coordi-nates of a biscyclopentadienyl metal dithiolate system with aligand fold angle of 1.2° and in C1 symmetry. The 3,5-di-methyl groups on Tp* were replaced by hydrogen atoms (Tpligand) to simplify the computations. A generalized gradientapproximation, with the exchange correction of Becke (48)and the correlation correction of Lee et al. (49), was used forall density functional calculations. Core levels (up to 3d forMo, up to 1s for C, N, O, and S) were treated as frozen orbit-als. No core levels for Ti were treated as frozen. The calcula-tions used triple-� basis sets with Slater type orbitals and a

polarization function for all elements besides Mo. Calcula-tions on the ground-state molecules were performed in thespin-restricted mode. Spin-unrestricted calculations were per-formed on the relevant ion states at the fixed geometry of theneutral molecule (Table 7, which is published as supportinginformation on the PNAS web site). The difference betweenthe total self-consistent field (SCF) energy of the ion and thetotal SCF energy of the neutral ground state is the �SCF es-timate of the ionization energy. A linear correction was ap-plied for comparison of the calculated and the observed ener-gies, i.e., calculated �SCF energies were shifted by thedifference between the experimentally obtained and the cal-culated first ionization energies. Similarly, orbital energy esti-mates of the ionization energies were shifted by the differ-ence between the observed first ionization energy and thenegative of the calculated eigenvalue of the HOMO.

Results and DiscussionPhotoelectron Spectra. The low energy valence regions of thegas-phase photoelectron spectra of 1-3 collected with bothHeI and HeII photon sources are presented in Fig. 3. Thisenergy region contains the ionizations that correspond to re-moval of electrons from the first two or three occupied mo-lecular orbitals of these molecules. For the Ti-containing mol-ecule (3), the spectra contain two ionizations that correspondto removal of electrons from the symmetric and antisymmet-ric (bands S�

� and S��, respectively) combinations of sulfur-�

orbital (S�) orbitals. For the Mo-containing molecules (1 and2), an additional ionization is observed in this region that cor-responds to removal of an electron from the half (1) or fully(2) occupied equatorial metal d orbital. The specific assign-ment of the ionizations can be made by comparison of thespectra to those of related molecules and by comparison ofthe relative ionization intensities as the source energy is var-ied (21). From previous experimental studies (20, 50) and cal-culations of atomic photoionization cross-sections (51), it isexpected that ionizations from orbitals with significant Ti 3dor Mo 4d contributions will increase in intensity comparedwith ionizations of primarily S 3p character when data col-lected with a HeII photon source is compared with data col-lected with a HeI photon source. Mixing of metal and sulfurcharacter in orbitals will result in smaller changes in relativeintensities of ionizations being observed.

For molecule 1, the lowest energy ionization band (labeled

Fig. 3. Gas-phase photoelectron spectra of (Tp*)MoO(bdt) (1), Cp2Mo(bdt) (2), and Cp2Ti(bdt) (3) with HeI and HeII excitation.

Joshi et al. PNAS � April 1, 2003 � vol. 100 � no. 7 � 3721

CHEM

ISTR

YSP

ECIA

LFE

ATU

RE

Mip in Fig. 3) at 7.04 eV is similar in appearance to the lowenergy band of related complexes having alkoxide ligands (23,52) that has been assigned to ionization arising primarily fromthe half-filled metal-based orbital (21, 23, 52). The two ion-izations between 7.25 eV to 8.00 eV can then be assigned toionizations associated with symmetric (S�

�) and antisymmetric(S�

�) combinations of S� orbitals based on previous assign-ment of analogous complexes (21, 22). More detailed assign-ment of the S�

� and S�� ionizations is not possible because of

the significant overlap of these low-energy features and thebeginning of the Tp* ligand based ionizations. The ratio ofthe areas of the Mip ionization and the sum of the overlap-ping S�

� and S�� ionizations changes very little with change in

photon source from HeI to HeII; the Mip�(S�� � S�

�) ratio inHeI is 0.66:1.00, that in HeII is 0.63:1.00. The observation ofonly small changes in the relative intensities of the first threebands with ionization source energy was suggested to be evi-dence of substantial mixing of metal and sulfur character in(Tp*)MoO(tdt) (21).

The photoelectron spectra of the Ti molecule (3) show twooverlapping ionization bands at 7.18 eV and 7.43 eV thatmust contain contributions from the S�

� and S�� ionizations.

The ratio of the areas of the low-energy band to the high-energy band increases significantly with change in photonsource from HeI to HeII. The ratio in HeI is 1.00:1.02; thatin HeII is 1.00:0.37. This behavior is consistent with the lowerenergy peak containing contributions from Ti. Therefore, thelower energy peak is assigned to S�

�, which has the propersymmetry to interact with the empty metal orbital on folding.The higher energy peak in 3 is assigned to the S�

� ionization,which does not mix significantly with the empty Ti d acceptororbital (discussed later in the calculations section). The sym-metric combination of the sulfur orbitals (S�

�) is more desta-bilized than the antisymmetric combination (S�

�) because ofthe greater interaction with the arene ring � orbital of appro-priate symmetry (39). This analysis compares well with that of(�5-tBuC5H4)2Zr(Se2C6H4) (39), for which the first band wasassigned as a combination of Se�

� and Se�� ionizations.

Compound 2 shows three low-energy ionizations, one at6.29 eV and an overlapping pair of peaks at 7.04 eV and 7.24eV. The HeII data for 2 clearly show an increase in the rela-tive intensity of the middle band, and perhaps a slight de-crease in the relative intensity of the highest energy band.These results imply that the middle band (7.04 eV) is mainlymetal-based, the lowest energy band (6.29 eV) is primarily S�

�

with some additional Mo 4d or C 2p character, and the high-est energy band (7.24 eV) is S�

�. The ratio of the areas of the

S���Mip�S�

� ionizations in HeI photon source is 1.00:1.03:0.82;that in HeII is 1.00:1.39:0.67. The energy of the metal basedionization (Mip) in 1 and 2 stays constant (at 7.04 eV) despitesignificant perturbations in the first coordination sphere andformal oxidation state of the metal.

Computational Results. The electronic structures of 1–3 werecalculated by using C1 symmetry and geometry optimized mo-lecular structures. The geometry optimizations are carried outon isolated molecules (45), corresponding to the conditions ofthe PES experiment. These molecular structures compare wellwith the reported structures determined from x-ray crystallog-raphy (Tables 8 and 9, which are published as supporting in-formation on the PNAS web site). As a result of geometryoptimization, the dithiolate fold angle changes from the crys-tallographic fold angle of 21.3° (12) to 31.0° in 1, from 9.0°(32) to 4.8° in 2, and from 46.0° (53) to 41.6° in 3.

Density functional theory calculations provide additionalinsight into the metal–dithiolate interactions and indicatethat, on dithiolate folding, the mixing of metal d and sulfurp� orbitals can be favored by their energy and symmetrymatch. The energies and general characters of the ionizationsobserved for molecules 1-3 by photoelectron spectroscopymatch those calculated by the �SCF method and by compari-son of orbital energies (Table 2 and Table 10, which is pub-lished as supporting information on the PNAS web site). Thecalculated orbital energies for 2 are low in comparison to theexperimental values. This appears to be a function of the foldangle. Fig. 4 shows a Walsh diagram for the variation of theorbital eigenvalues for 2 with fold angle. The S�

� eigenvaluebecomes less negative and Mip becomes more negative withincreasing fold angle. The S�

� eigenvalue and the onset of fur-ther ionizations remain relatively constant. The angle atwhich the orbital eigenvalues most closely resemble that de-termined by PES is �15°. Contour plots of the orbitals for1-3 that correspond to the ionizations evaluated by photoelec-tron spectroscopy are shown in Fig. 5. The HOMO in 1 isprimarily a metal in-plane orbital, whereas it is primarily asulfur out-of-plane orbital in 2 and 3. The HOMO-1 orbitalfor 2 is a metal in-plane orbital. As mentioned earlier, themetal-based ionization has similar energy for 1 and 2. Thus,dithiolate appears to buffer the electron density at the metalcenter by affecting the overlap of S� orbitals with the metalcenter, thereby overcoming the striking differences of � do-nor atoms (N and O) in 1 and � donor Cp ligands in 2. Fold-ing of the ene-dithiolate can influence this overlap of S� (out-of-plane) and metal in-plane orbitals. The HOMO-2 orbital

Table 2. Experimental and calculated �SCF and orbital ionization energies (eV) for the ionizations from sulfur antisymmetric (S��),

sulfur symmetric (S��), and metal in-plane (Mip) orbitals of 1–3

Compound method

(Tp*)MoO(bdt) (1) Cp2Mo(bdt) (2) Cp2Ti(bdt) (3)

PES(assignment) �SCF†

Orbitalenergy†

PES(assignment) �SCF

Orbitalenergy

PES(assignment) �SCF

Orbitalenergy

First ionization 7.04 (Mip) 7.04 (6.51) 7.04 (4.35) 6.29 (S��) 6.29 (6.46) 6.29 (3.96) 7.18 (S�

�) 7.18 (6.67) 7.18 (4.58)Second ionization 7.54 (S�

�) S 7.49T 7.20

7.60 7.04 (Mip) 6.75 6.58 7.43 (S��) 7.37 7.36

Third ionization 7.71 (S��) S 7.53

T 7.47‡

7.85 7.24 (S��) 6.94 6.93 — — —

Onset of further ionizations 8.23 Did not converge 8.52 8.32 8.07 7.99 8.19 8.12 8.16

PES, experimental vertical ionization; orbital energy, negative of the orbital eigenvalue; S, singlet; T, triplet; SOMO, smallest occupied molecular orbit. �SCFionization energies and orbital energies have been shifted so that first �SCF ionization energy and highest occupied orbital energy agree with the experimentalfirst ionization energy. Values in parentheses are unshifted values of the HOMO�SOMO for �SCF, and the negative of the orbital eigenvalue of the HOMO�SOMOfor orbital energy.†The methyl groups of the Tp* ligand were replaced with H atoms to simplify calculations.‡SCF converged to three decimal places.

3722 � www.pnas.org�cgi�doi�10.1073�pnas.0636832100 Joshi et al.

for 1 and 2 and the HOMO-1 orbital for 3 are primarily theantisymmetric sulfur (S�

�) orbitals with little contributionfrom the metal. The symmetric S�

� orbital has the right sym-metry and energy to match the metal in-plane orbital on fold-ing of the dithiolate unit, and the substantial mixing of theout-of-plane S�

� orbitals and metal in-plane orbital on foldingis shown experimentally by the intensity changes observed inthe HeI and HeII spectra of 2 and 3 (Fig. 3).

ConclusionsThe gas-phase photoelectron spectra of Cp2M(bdt), M � Ti(3) and Mo (2) clearly show that, under favorable conditions,

the ionizations from primarily metal-based orbitals and pri-marily S�-based orbitals can be experimentally distinguishedfrom one another. In addition, we have unambiguously dem-onstrated that the symmetric S�

� orbital, which can interactwith the in-plane metal orbital on bending, is more easily ion-ized than the antisymmetric S�

� orbital. These results experi-mentally verify the bonding model originally put forth byLauher and Hoffmann (27) for bent-metallocene dithiolatecompounds such as 2 and 3. Our analysis and assignment ofthe low-energy ionizations of compounds 2 and 3 also providea framework for investigation and direct assignment of the S�

and metal-based ionizations in related metal dithiolate sys-tems that may be more strongly mixed than 2 and 3 or thatmay have less favorable separation of their metal-sulfur re-gion from the ionizations of other ligands.

The variation in the fold angle of the dithiolate ligand re-flects the electron occupation in the equatorial in-plane metalorbital, as has been noted (11, 27). Folding of the dithiolateligand enables the S� electron density to modulate the elec-tron density in the equatorial plane of the metal, and thusfolding can play a very significant role in the metal–sulfuranisotropic bonding interaction. In the case of 3, which hasformally a Ti(IV) d0 metal center, the dithiolate ligand can bethought of as a six-electron donor. Each of the thiolate �-orbitals provides two electrons, and two additional electronscome from the S�

� orbital. Hence, folding the dithiolate ligandeffectively stabilizes 3 as an 18-electron complex. This ‘‘di-thiolate folding effect’’ is in contrast to the nitrosyl groupwhere the transformation from linear to bent coordination ofthe ligand transfers a pair of electrons from the metal to alocalized orbital on the ligand (54).

Implications for Enzymes. For molybdenum and tungsten en-zymes the ‘‘dithiolate-folding-effect’’ of the S2pdt ligand may

Fig. 4. Graph of orbital eigenvalues against fold angle for 2. The geometryoptimized coordinates of the Cp2MoS2 core of 2 were kept fixed, and thebenzene group was rotated relative to the MS2 plane while keeping the bondlengths constant.

Fig. 5. Calculated frontier orbitals (HOMO, HOMO-1, and HOMO-2) of 1–3. All contours have the same cutoff level (0.05).

Joshi et al. PNAS � April 1, 2003 � vol. 100 � no. 7 � 3723

CHEM

ISTR

YSP

ECIA

LFE

ATU

RE

be important in stabilizing the multiple oxidation states thatresult from electron transfer or oxygen atom transfer (OAT).For example, in sulfite oxidase (Fig. 1), reduction of themetal in-plane orbital due to the release of an equatorial oxy-gen atom during OAT should make the dithiolene unit ofS2pdt more planar compared with the fully oxidized enzyme.In this regard, we note that the calculated dithiolate fold an-gles for the two chemically equivalent, but crystallographicallydistinct, molybdenum centers in sulfite oxidase of 6.6° and7.0° (Table 1) are smaller than that for the oxidized center ofaldehyde oxidoreductase (16.6°), whose oxidized structure hasbeen determined to high resolution (1.28 Å) (15). The exactoxidation state of sulfite oxidase in the crystal structure is notknown, but there is evidence for photoreduction of the mo-lybdenum center in the synchrotron beam during data collec-tion (5, 55). The fold angle studies on model systems supportthe view that the reported structure of sulfite oxidase (5) is ofa reduced form of the enzyme. Dithiolate folding can alsomodify the electropositive nature of the metal center by vary-ing the overlap of sulfur lone pairs with the metal in-planeorbital. The S� orbitals of dithiolate can therefore serve as anessential instrument for the buffering of electron density atthe metal center by either involving themselves in strong mix-

ing with an empty metal orbital, as shown here for the d0 Ticomplex (3), or by localization of electron density on the S�

orbitals in the presence of a filled metal orbital, as shownhere for the d2 Mo complex (2). Dithiolate folding may alsoprovide a mechanism for the sulfur-� orbitals of the S2pdtligand to act as an effective electron transfer pathway fromthe in-plane orbitals on the metal center to other redox part-ners via the pyranopterin. Finally, we note that for molybde-num and tungsten enzymes the energies involved in substratebinding, docking with electron transfer partners, and dynamicmotions of the protein skeleton, may modulate the dithiolatefold angle and, consequently, the electron distribution, overallreduction potential and reactivity of the metal center.

Supporting InformationGeometry optimized and crystal structure coordinates for1–3, selected computational input, and results are available inTables 3–11, which are published as supporting informationon the PNAS web site.

Support by National Institutes of Health Grant GM-37773 (to J.H.E.)and National Science Foundation Grant CHE-9618900 (to D.L.L.) isgratefully acknowledged.

1. Rajagopalan, K. & Johnson, J. L. (1992) J. Biol. Chem. 267, 10199–10202.2. Hille, R. (1996) Chem. Rev. 96, 2757–2816.3. Sigel, A. & Sigel, H. (2002) Molybdenum and Tungsten: Their Roles in Biological

Processes, Metals Ions in Biological Systems (Dekker, New York), Vol. 39.4. Garner, D. C., Banham, R., Cooper, S. J., Davies, E. S. & Stewart, L. J. (2001)

in Handbook on Metalloproteins, eds. Bertini, I., Sigel, A. & Sigel, H. (Dekker,New York), pp. 1023–1090.

5. Kisker, C., Schindelin, H., Pacheco, A., Wehbi, W., Garrett, R. M., Rajago-palan, K. V., Enemark, J. H. & Rees, D. C. (1997) Cell 91, 973–983.

6. Kisker, C. (2001) in Handbook of Metalloproteins, eds. Messerschmidt, A.,Huber, R., Weighardt, K, & Poulos, T. (Wiley, New York), Vol. 2, pp.1121–1135.

7. Enemark, J. H. & Cosper, M. M. (2002) in Molybdenum and Tungsten: TheirRoles in Biological Processes, Metal Ions in Biological Systems, eds. Sigel, A. &Sigel, H. (Dekker, New York), Vol. 39, pp. 621–654.

8. Schindelin, H., Kisker, C. & Rees, D. (1997) J. Biol. Inorg. Chem. 2, 773–781.9. Eisenberg, R. (1970) Prog. Inorg. Chem. 12, 295–369.

10. Inscore, F. E., McNaughton, R., Westcott, B., Helton, M. E., Jones, R.,Dhawan, I. K., Enemark, J. H. & Kirk, M. L. (1999) Inorg. Chem. 38, 1401–1410.

11. Joshi, H. K., Inscore, F. E., Schirlin, J. T., Dhawan, I. K., Carducci, M. D., Bill,T. G. & Enemark, J. H. (2002) Inorg. Chim. Acta 337, 275–286.

12. Dhawan, I. K., Pacheco, A. & Enemark, J. H. (1994) J. Am. Chem. Soc. 116,7911–7912.

13. Dhawan, I. K. & Enemark, J. H. (1996) Inorg. Chem. 35, 4873–4882.14. Inscore, F. E., Joshi, H. K., McElhaney, A. E. & Enemark, J. H. (2002) Inorg.

Chim. Acta 331, 246–256.15. Rebelo, J., Dias, J., Huber, R., Moura, J. & Romao, M. (2001) J. Biol. Inorg.

Chem. 6, 791–800.16. Enroth, C., Eger, B. T., Okamoto, K., Nishino, T., Nishino, T. & Pai, E. F.

(2000) Proc. Natl. Acad. Sci. USA 97, 10723–10728.17. Li, H. K., Temple, C., Rajagopalan, K. & Schindelin, H. (2000) J. Am. Chem.

Soc. 122, 7673–7680.18. Carducci, M. D., Brown, C., Solomon, E. I. & Enemark, J. H. (1994) J. Am.

Chem. Soc. 116, 11856–11868.19. McMaster, J., Carducci, M. D., Yang, Y.-S., Solomon, E. I. & Enemark, J. H.

(2001) Inorg. Chem. 40, 687–702.20. Green, J. C. (1994) Acc. Chem. Res. 27, 131–137.21. Westcott, B. L., Gruhn, N. E. & Enemark, J. H. (1998) J. Am. Chem. Soc. 120,

3382–3386.22. Helton, M. E., Gruhn, N. E., McNaughton, R. L. & Kirk, M. L. (2000) Inorg.

Chem. 39, 2273–2278.23. Westcott, B. L. & Enemark, J. H. (1997) Inorg. Chem. 36, 5404–5405.24. Wang, X.-B., Inscore, F. E., Yang, X., Cooney, J. J. A., Enemark, J. H. & Wang,

L.-S. (2002) J. Am. Chem. Soc. 124, 10182–10191.25. Guyon, F., Lenoir, C., Fourmigue, M., Larsen, J. & Amaudrut, J. (1994) Bull.

Soc. Chim. Fr. 131, 217–226.

26. Fourmigue, M. (1998) Coord. Chem. Rev. 178–180, 823–864.27. Lauher, J. W. & Hoffmann, R. (1976) J. Am. Chem. Soc. 98, 1729–1742.28. Green, J. C. (1981) Struct. Bonding (Berlin) 43, 37–112.29. Green, J. C. (1998) Chem. Soc. Rev. 27, 263–272.30. Guyon, F., Fourmigue, M., Audebert, P. & Amaudrut, J. (1995) Inorg. Chim.

Acta 239, 117–124.31. Domercq, B., Coulon, C. & Fourmigue, M. (2001) Inorg. Chem. 40, 371–378.32. Kutoglu, A. & Kopf, H. (1970) J. Organomet. Chem. 25, 455–460.33. Fourmigue, M., Domercq, B., Jourdain, I. V., Molinie, P., Guyon, F. &

Amaudrut, J. (1998) Chem. Eur. J. 4, 1714–1723.34. Pilato, R. S., Eriksen, K., Greaney, M., Gea, Y., Taylor, E., Goswami, S.,

Kilpatrick, L., Spiro, T., Rheingold, A. & Stiefel, E. I. (1993) Am. Chem. Soc.Symp. Ser. 535, 83–97.

35. Fourmigue, M., Lenoir, C., Coulon, C., Guyon, F. & Amaudrut, J. (1995) Inorg.Chem. 34, 4979–4985.

36. Kaiwar, S. P., Hsu, J. K., Vodacek, A., Yap, G., Liable-Sands, L. M., Rheingold,A. L. & Pilato, R. S. (1997) Inorg. Chem. 36, 2406–2412.

37. Fomitchev, D. V., Lim, B. S. & Holm, R. (2001) Inorg. Chem. 40, 645–654.38. Domercq, B., Coulon, C., Feneyrou, P., Dentan, V., Robin, P. & Fourmigue,

M. (2002) Adv. Funct. Mater. 12, 359–366.39. Guimon, C., Pfister-Guillouzo, G., Meunier, P., Gautheron, B., Tainturier, G.

& Pouly, S. (1985) J. Organomet. Chem. 284, 299–312.40. Klapotke, T. & Kopf, H. (1988) Z. Anorg. Allg. Chem. 558, 217–222.41. Lowe, N. D. & Garner, C. D. (1993) J. Chem. Soc. Dalton Trans. 14, 2197–2207.42. Lichtenberger, D. L. & Copenhaver, A. S. (1990) J. Electron Spectrosc. Rel.

Phenom. 50, 335–352.43. Baerends, E. J., Ellis, D. E. & Ros, P. (1973) Chem. Phys. 2, 41–51.44. Fonseca Guerra, C., Snijders, J. G., te Velde, G. & Baerends, E. J. (1998) Theor.

Chem. Acc. 99, 391–399.45. Te Velde, G., Bickelhaupt, F. M., Baerends, E. J., Fonseca Guerra, C., Van

Gisbergen, S. J. A., Snijders, J. G. & Ziegler, T. (2001) J. Comput. Chem. 22,931–967.

46. te Velde, G. & Baerends, E. J. (1992) J. Comput. Phys. 99, 84–98.47. Versluis, L. & Ziegler, T. (1988) J. Chem. Phys. 88, 322–329.48. Becke, A. D. (1988) Phys. Rev. A 38, 3098–3100.49. Lee, C., Yang, W. & Parr, R. G. (1988) Phys. Rev. B 37, 785–789.50. Glass, R. S., Gruhn, N. E., Lichtenberger, D. L., Lorance, E., Pollard, J. R.,

Birringer, M., Block, E., DeOrazio, R., He, C., Shan, Z. & Zhang, X. (2000)J. Am. Chem. Soc. 122, 5065–5074.

51. Yeh, J. & Lindau, I. (1985) At. Data Nucl. Data Tables 32, 1–155.52. Chang, C. S. J., Rai-Chaudhuri, A., Lichtenberger, D. L. & Enemark, J. H.

(1990) Polyhedron 9, 1965–1973.53. Kutoglu, A. (1972) Z. Anorg. Allg. Chem. 390, 195–209.54. Enemark, J. H. & Feltham, R. D. (1974) Coord. Chem. Rev. 13, 339–406.55. George, G. N., Pickering, I. J. & Kisker, C. (1999) Inorg. Chem. 38,

2539–2540.

3724 � www.pnas.org�cgi�doi�10.1073�pnas.0636832100 Joshi et al.