Spectroscopic and Kinetic Studies of Y114F and W116F Mutants of Me2SO Reductase from Rhodobacter capsulatus*

Mutants of the active site residues Trp-116 and Tyr-114 of the molybdenum-containing Me2SO reductase from Rhodobacter capsulatus have been examined spectroscopically and kinetically. The Y114F mutant has an increased rate constant for oxygen atom transfer from Me2SO to reduced enzyme, the result of lower stability of the Ered·Me2SO complex. The absorption spectrum of this species (but not that of either oxidized or reduced enzyme) is significantly perturbed in the mutant relative to wild-type enzyme, consistent with Tyr-114 interacting with bound Me2SO. The as-isolated W116F mutant is only five-coordinate, with one of the two equivalents of the pyranopterin cofactor found in the enzyme dissociated from the molybdenum and replaced by a second MoO group. Reduction of the mutant with sodium dithionite and reoxidation with Me2SO, however, regenerates the long-wavelength absorbance of functional enzyme, although the wavelength maximum is shifted to 670 nm from the 720 nm of wild-type enzyme. This “redox-cycled” mutant exhibits a Me2SO reducing activity and overall reaction mechanism similar to that of wild-type enzyme but rapidly reverts to the inactive five-coordinate form in the course of turnover.

Dimethyl sulfoxide reductase catalyzes the reduction of dimethyl sulfoxide (Me 2 SO) to dimethyl sulfide (DMS) 3 and is a member of a large class of mononuclear molybdenum-containing enzymes (1)(2)(3) having an L 2 MoO(X) core (where L represents a pyranopterin cofactor coordinated to the molybdenum via an enedithiolate side chain, and X represents a ligand in most cases provided by the polypeptide, a serinate in the case of Me 2 SO reductase). Two forms of Me 2 SO reductase exist, as exemplified by the enzymes from Escherichia coli (a heterotrimer consisting of the DmsABC gene products) and Rhodobacter species (the monomeric DorA protein). Both are components of anaerobic respiratory chains and catalyze the reductive abstraction of oxygen from Me 2 SO according to the following stoichiometry using electrons derived from the respiratory pool.
The Rhodobacter sphaeroides DorA enzyme is a soluble periplasmic protein of 85 kDa possessing the molybdenum cofactor as the sole redox-active center. Because of the distinctive absorption features of its molybdenum center, the DorA enzyme has become a paradigm for understanding the electronic structure of molybdenum centers of the L 2 MoO(X) variety (4 -12). In addition to steady-state kinetics under a variety of conditions, rapid reaction studies with both the Rhodobacter capsulatus (10) and R. sphaeroides enzyme (13) have also been reported. With the R. sphaeroides enzyme, the reaction of reduced enzyme with Me 2 SO at low pH is biphasic. A fast [Me 2 SO]dependent phase (k lim ϳ 1000 s Ϫ1 ) yields a spectrally distinct E red ⅐Me 2 SO complex, with absorption maxima at 490 and 550 nm, followed by a slower oxygen abstraction step (35 s Ϫ1 ) that yields DMS and E ox (13). The hyperbolic dependence of the rapid phase on [Me 2 SO] gives a K d Me 2SO of ϳ155 M and indicates that the observed E red ⅐Me 2 SO species is not the initial Michaelis complex but instead is the product of its breakdown. Oxidized enzyme also reacts with excess DMS to give this same E red ⅐Me 2 SO species (5,14), demonstrating the reversibility of the oxygen atom transfer reaction.
The crystal structure of Me 2 SO reductase (14 -18) suggests that two active site residues, Trp-116 and Tyr-114, are of particular interest. Trp-116 hydrogen-bonds to the labile MoAO group of oxidized enzyme and, thus, may be directly involved in catalysis. The role of Tyr-114 is less clear, although it also is close to the molybdenum center and has been suggested to hydrogen-bond to the oxygen atom of bound Me 2 SO in the E red ⅐Me 2 SO complex (24,25). With development of recombinant expression systems (in E. coli in the case of the R. sphaeroides enzyme (21,22) or homologously in R. capsulatus (23)), both Tyr-114 (24,25) and Trp-116 (26) have been targeted for mutation. The Y114F mutant turns over substantially more rapidly with both Me 2 SO and trimethylamine-N-oxide (TMAO, an alternate oxidizing substrate) relative to the wildtype enzyme in steady-state assays, although this increase in rate is offset by a decrease in substrate affinity (24,25). Steadystate analysis (25) has yielded a k cat of 81.4 s Ϫ1 and K m of 185.5 M for the Y114F mutant of the R. capsulatus enzyme, which compare with values of 42.9 s Ϫ1 and K m of 9.7 M for wild-type enzyme. It has thus been suggested that Tyr-114 stabilizes the substrate-bound intermediate E red ⅐Me 2 SO (24,25). Initial characterization of the W116F mutant from R. capsulatus showed that its molybdenum center is five-coordinate, with one of the pterin ligands of the LMoO 2 coordination sphere (that designated crystallographically as "Q") replaced by a second MoAO (26). The mutant enzyme exhibited ϳ10% of wild-type catalytic activity. The present work reports a detailed spectroscopic and rapid reaction kinetic study of wild-type R. capsulatus Me 2 SO reductase and the Y114F and W116F mutants and provides new insights into the structural and functional roles of Tyr-114 and Trp-116 in the wild-type enzyme.

EXPERIMENTAL PROCEDURES
Protein Purification; Purification of Recombinant Me 2 SO Reductase-R. capsulatus strain 37b4 carrying the recombinant dorA gene and its mutants, generated as described previously (23,25), was grown photoheterotrophically using malate as the carbon source and in the presence of ϳ60 mM Me 2 SO. All purification steps for the recombinant enzyme forms were performed as described previously (13), including the redox cycling used to generate the functional form of the enzyme. Purified enzyme was concentrated and stored in liquid nitrogen until needed.
UV-visible Spectroscopy-UV-visible spectra were obtained using a Hewlett-Packard 8452 diode-array spectrophotometer. Steady-state and rapid reaction experiments were performed using an Applied Photophysics Inc. SX.18MV stopped-flow apparatus equipped with a photodiode array detector; data analysis was performed using software provided by the manufacturer. Solutions of Me 2 SO reductase were made anaerobic by placing enzyme in a tonometer and repeatedly evacuating and flushing with O 2 -free argon. Reduced enzyme was prepared by titration using sodium dithionite and following the reaction spectrophotometrically. Calibration of sodium dithionite was done by titration of dithionite stock solutions against 2,6-dichlorophenolindophenol (DCIP) as described previously (13).
Enzyme-monitored Turnover Experiments-Enzyme-monitored turnover experiments were performed by mixing ϳ50 M enzyme with 2 mM Me 2 SO and 25 mM dithionite in a stoppedflow apparatus at 5°C. For deconvolution of the spectra seen in the course of the reaction, parent spectra for oxidized enzyme and the E red ⅐Me 2 SO complex seen in the course of the oxidative half-reaction were obtained by mixing enzyme with buffer alone and with 100 mM DMS, respectively. The parent spectrum for the Mo(V) form of wild-type and Y114F Me 2 SO reductase was taken as the stable resting line-shape observed in the course of turnover with TMAO, and the reduced parent spectra were obtained by taking the final resting spectrum from enzyme-monitored turnover experiments performed with excess dithionite as reductant. Spectral deconvolution was performed using the Moore-Penrose pseudoinverse of a matrix of parent spectra (M) using the equation C ϭ (MЈM) Ϫ1⅐MЈ , which for each experimental spectrum obtained in the course of the reaction yields the relative contributions of each of the parent spectra. Further manipulations were performed using SigmaPlot Version 8.0. Because sodium dithionite (which absorbs strongly in the near-UV) was used as reductant, spectral analysis was restricted to wavelengths above 400 nm. Simulations of these reactions were performed using Applied Photophysics software. In some cases (e.g. with the Y114F mutant and Me 2 SO as substrate) the generally similar component spectra for oxidized, Mo(V), and E red ⅐Me 2 SO enzyme forms required fitting of the experimental data by hand. All these experiments were performed in 50 mM KH 2 PO 4 , 0.6 mM EDTA, pH 6.0.
Electron Paramagnetic Resonance Spectroscopy-Samples for EPR characterization of R. capsulatus enzyme forms were prepared by mixing oxidized enzyme and made anaerobic in a glass tonometer with a concentrated dithionite solution (either with or without TMAO) followed by rapid freezing in a liquid nitrogen cooled acetone bath. EPR measurements were carried out using a Brüker ESP 300 spectrometer with the following instrument settings: modulation frequency of 100 kHz, modulation amplitude of 5 gauss, sweep width of 300 gauss (from 3250 -3550 gauss), and 2 milliwatts of microwave power.
Resonance Raman-Samples for resonance Raman studies were prepared in 10 mM KH 2 PO 4 , pH 6.0; EDTA was removed, and a low buffer concentration was used to minimize the possibility of any interference due to phosphate or EDTA vibrational modes. Samples were concentrated to 1-3 mM by Microcon ultrafiltration at 5°C. Cryogenic samples (25-35 l) were frozen by placing the protein solution in the sample well of a nickel-plated sample holder affixed to the cold finger of an APD Cryogenics Inc. liquid helium cryostat. Resonance Raman spectra were obtained and calibrated using previously described instrumentation and methods (26). The PEAKFIT program from Jandel Scientific was used to determine Raman band positions with a maximal error of Ϯ2 cm Ϫ1 . Base-line correction was performed by polynomial fits using an in-house MATLAB Version 6.1.0.450 macro (The MathWorks, Inc.) running on a SGI Octane2 work station; corrected spectra were plotted using SigmaPlot Version 8.0. Fig.  1 shows the UV-visible spectra of the (redox-cycled) oxidized, reduced, and E red ⅐Me 2 SO forms of wild-type, Y114F, and W116F Me 2 SO reductase at pH 6.0. The spectra for wild-type enzyme (Fig. 1A) are well characterized (21)(22)(23)(24)(25). Oxidized enzyme has an absorption maximum at 720 nm (indicative of bis(enedithiolate) coordination) and shoulders at ϳ560, 470, and 360 nm; the reduced enzyme has absorption bands at 640 and 380 nm; the E red ⅐Me 2 SO complex has absorption maxima at 560 and 485 nm. For the Tyr-114 mutant (Fig. 1B), as seen previously, the spectrum of the oxidized enzyme is very similar to that seen for the wild-type enzyme (24,25). Such differences as do exist include a modest bleaching in the near UV (best evidenced by loss of the band centered at 350 nm) and a slight red-shift of the long-wavelength absorbance from 720 nm in the wild-type enzyme to 730 nm. The spectrum of the reduced Y114F mutant is also similar to that of the wild-type enzyme but with some extinction loss observed in the 375-nm absorption maximum and a red-shift in the long-wavelength absorption from 640 to 660 nm. On the other hand, as previously observed by Ridge et al. (25), the spectrum of the E red ⅐Me 2 SO complex is significantly perturbed upon mutation of Tyr-114 to phenylalanine. Compared with the spectrum of the wild-type complex, which has strong absorbance maxima at 485 and 550 nm, the complex seen with the Y114F mutant has significantly weaker absorption bands at 470 and 565 nm. That the E red ⅐Me 2 SO spectrum is so perturbed upon mutation is consistent with Tyr-114 being involved in hydrogen-bonding to the oxygen of the complexed Me 2 SO (24,25).

UV-visible Spectroscopy of Me 2 SO Reductase Mutants-
The as-isolated W116F mutant lacks appreciable long-wavelength absorbance (Fig. 1C). On the basis of the similarity of the spectrum to that for the pentacoordinate "HEPES-modified" form of the wild-type enzyme, in which the Q pterin has dissociated from the molybdenum (20), the mutant enzyme is also thought also to be predominantly five-coordinate (26). Redoxcycling oxidized W116F mutant by reduction with MV⅐ and reoxidation with Me 2 SO results in a spectrum with the characteristic long-wavelength absorption of bis(enedithiolate) coordination, albeit blue-shifted to 680 nm relative to the 740 nm of wild-type enzyme (Fig. 1C). We conclude that the oxidized W116F mutant regains bis(enedithiolate) coordination upon redox-cycling. The spectrum of the reduced W116F mutant is also perturbed relative to that of the wild-type enzyme and resembles that seen for the E red ⅐Me 2 SO complex of wild-type enzyme (rather than the reduced form), with absorption maxima at 470 and 550 nm (as compared with 470 and 565 nm for wild-type enzyme). Surprisingly, the addition of DMS (to a concentration of 10 mM) to the oxidized W116F enzyme results in a uniform bleaching over the entire visible spectrum, with little if any indication of the characteristic spectrum of the E red ⅐Me 2 SO complex of wild-type enzyme formed under these conditions. The spectrum seen with the mutant in fact resembles that of the reduced "as-isolated" pentacoordinate form of the enzyme (20), suggesting that the Q pterin has dissociated upon treatment with DMS ( Fig. 1C, inset).
Oxidative and Reductive Half-reaction Kinetics of Wild-type and Y114F Me 2 SO Reductase-To better understand the catalytic behavior of R. capsulatus Me 2 SO reductase, the rapid reaction kinetics of reduced wild-type enzyme and mutants was examined. Our previous work was done with the wild-type R. sphaeroides enzyme, and the work here with the wild-type R. capsulatus enzyme was done to ensure proper comparison with the R. capsulatus mutants that were the primary focus. Measurements were performed at pH 6.0, 25°C and are summarized in Table 1. In the reaction of reduced enzyme with Me 2 SO, the E red ⅐Me 2 SO species formed in the 2-3-ms dead time of the stopped-flow apparatus with both the wild-type and Y114F mutant R. capsulatus enzyme (consistent with the limiting rate of reaction seen with the R. sphaeroides enzyme of ϳ1000 s Ϫ1 ; Ref. 13). With wild-type enzyme this intermediate  DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 broke down to oxidized enzyme with a rate constant of 43 s Ϫ1 , comparable to the 38 s Ϫ1 seen with the R. sphaeroides enzyme and in excellent agreement with the previously reported steadystate k cat of 42.9 s Ϫ1 (25). With the Y114F mutant, the E red ⅐Me 2 SO complex decayed to E ox with a rate 103 s Ϫ1 , approximately twice as fast as seen with wild-type enzyme and in satisfactory agreement with the reported steady-state k cat for the mutant of 81.4 s Ϫ1 (25).

Spectroscopic and Kinetic Studies of Me 2 SO Reductase Mutants
Reoxidation of reduced wild-type Me 2 SO reductase by Me 2 SO is known to be incomplete, with a significant amount of the E red ⅐Me 2 SO complex remaining at the end of the reaction (20). Fig. 2 shows the initial and final spectra observed during the reaction of reduced enzyme with Me 2 SO for the wild-type enzyme (2A) and Y114F mutant (2B) along with the independently determined spectra for the oxidized enzyme forms. Fits of the final resting spectrum for wild-type enzyme using the known spectra for oxidized and E red ⅐Me 2 SO forms suggests that only ϳ60% of the enzyme becomes fully oxidized at the completion of the reaction, consistent with earlier work (20). On the other hand, the spectrum seen at the end of reaction with the Y114F mutant is virtually identical to that for the oxidized enzyme, implying that the E red ⅐Me 2 SO^E ox ϩ DMS equilibrium lies further to the right for the mutant than is the case with the wild-type enzyme. This is consistent with the conclusion that Tyr-114 interacts with bound Me 2 SO; its mutation to Phe both increases the rate of decay of the E red ⅐Me 2 SO intermediate and decreases its thermodynamic stability.
We next examined the reaction of oxidized enzyme with the non-physiological reductant sodium dithionite as a surrogate for the physiological DorC cytochrome. As expected on the basis of previous work with the R. sphaeroides enzyme (13), reduction of the recombinant wild-type R. capsulatus enzyme (Fig. 3A) was biphasic, with an absorption spectrum very similar to for the "high-g split" Mo(V) intermediate, with wild-type R. sphaeroides Me 2 SO reductase first appearing, then decaying to that of the fully reduced enzyme. As with the R. sphaeroides enzyme, decay of the Mo(V) species exhibited non-exponential behavior for reasons that are not understood. Nevertheless, at pH 6.0 and 5°C, fits to the data (in the reaction of 100 M wild-type enzyme with ϳ25 mM dithionite) yielded rate constants of 5 and 0.8 s Ϫ1 for the Mo(VI)/Mo(V) and Mo(V)/Mo(IV) transitions, respectively. The Mo(V) species also accumulated to a significant degree in the reaction of the Y114F mutant with dithionite, with rate constants for formation and decay of 1 and ϳ0.2 s Ϫ1 , respectively. The Y114F mutant, thus, reacts somewhat more sluggishly with dithionite than does wild-type enzyme.
To confirm that the intermediate seen with both wild-type and Y114F enzyme indeed represented the absorption spectrum of a Mo(V) species, freeze-quench EPR experiments were performed. The insets of Figs. 3, A and B, show the EPR spectra seen in the course of the reduction with dithionite, with sample being frozen within 5 s of the addition of dithionite. The observed signals are indistinguishable from the catalytically relevant high-g split Mo(V) signal described previously (9,13) with wild-type enzyme, indicating that despite its proximity Tyr-114 does not interact to any significant degree with the molybdenum center in the Mo(V) state.

SO reductase
The oxidative half-reaction refers to the reaction of 100 M reduced Me 2 SO reductase with excess Me 2 SO, and the reductive half-reaction refers to reaction of 100 M oxidized enzyme with ϳ25 mM sodium dithionite at 5°C. All reactions were performed in 50 mM KH 2 PO 4 , 0.6 mM EDTA, pH 6.0. Enzyme-monitored Turnover with Wild-type and Y114F Me 2 SO Reductase-We have recently demonstrated the utility of enzyme-monitored turnover studies with the R. sphaeroides Me 2 SO reductase (13), and with the spectra of the several catalytically relevant species characterized above we have extended this approach to the mutant forms. In such experiments enzyme (at a sufficiently high concentration that its absorption spectrum can be accurately monitored) is mixed with excess concentrations of both Me 2 SO (or TMAO) as oxidizing substrate and dithionite as reducing substrate. There is a rapid approach to the steady state which is subsequently maintained until the limiting substrate is exhausted. Spectra observed in the course of turnover may be treated as the weighted sum of the component spectra of all catalytically important enzyme species (e.g. E ox , E red , and the high-g split species) and deconvoluted to give quantitative time courses for each species (see "Experimental Procedures"). Again, studies with the wild-type R. capsulatus enzyme recapitulated our earlier work with the wild-type R. sphaeroides enzyme but were necessary for proper comparison with the R. capsulatus mutants.

Oxidative half-reaction
With wild-type R. capsulatus enzyme and TMAO as substrate (mixing 100 M oxidized enzyme with 6 mM TMAO and ϳ25 mM dithionite at pH 6.0 and 5°C), very similar results to those seen previously for the wild-type R. sphaeroides enzyme were obtained (data not shown). In particular, the spectra seen throughout the steady-state reflected a high degree of accumulation of the high-g split Mo(V) intermediate. With Me 2 SO rather than TMAO as oxidizing substrate, the additional spectrum of the E red ⅐Me 2 SO complex (obtained by mixing oxidized enzyme with 100 mM DMS) was needed in the spectral deconvolution. The time courses for each of the four catalytically relevant species (oxidized, reduced, high-g split, and E red ⅐Me 2 SO) are shown in Fig. 4A. The results of the analysis were similar, albeit not identical, to those seen with the R. sphaeroides enzyme. In particular, the E red ⅐Me 2 SO complex accumulates more rapidly and to a greater extent with the R. capsulatus enzyme than with the R. sphaeroides enzyme, and the high-g split Mo(V) species, thus, does not predominate to the same degree. Fits of the time course yielded an effective rate constant for oxygen atom transfer of 0.08 s Ϫ1 in the steadystate, reflecting a very low level of turnover. This could be attributed to the large accumulation of the E red ⅐Me 2 SO intermediate (ϳ90% of the total enzyme concentration) as the product DMS accumulates and rebinds E ox to give E red ⅐Me 2 SO to an increasing degree. This indicates that the R. capsulatus enzyme has an even higher affinity for DMS than does the R. sphaeroides enzyme and is even more susceptible to product inhibition.
With the Y114F mutant of the R. capsulatus enzyme and TMAO as oxidizing substrate, the enzyme-monitored turnover analysis yielded comparable results to those seen with wild-type enzyme. Again, the spectrum seen in the steady state reflected a high accumulation of the high-g split Mo(V) species (data not shown). On the other hand, with Me 2 SO as oxidizing substrate (the concentration of Me 2 SO used in the reaction, 2.2 mM, was scaled from 2 mM to account for the somewhat higher K m Me 2SO seen with the Y114F enzyme; Ref. 25) the E red ⅐Me 2 SO species did not predominate in the steady state to nearly the same degree as was seen with wild-type enzyme, increasing to only 55% of the total enzyme by the time Me 2 SO was depleted. Fits to the kinetic data yielded an effective rate constant for oxygen For the W116F mutant, the Split EPR signal was obtained by steady-state turnover of the enzyme with TMAO and sodium dithionite, whereas the Unsplit signal was formed by the addition of substoichiometric amounts of sodium dithionite to oxidized enzyme. atom transfer of ϳ0.18 s Ϫ1 versus the 0.08 s Ϫ1 seen with wildtype enzyme. This indicates significantly less inhibition by dimethyl sulfide with the mutant as compared with wild-type enzyme, consistent with the faster breakdown of the E red ⅐Me 2 SO intermediate seen in the oxidative half-reaction studies above. Again, this reflects the lower affinity of the oxidized Y114F mutant for DMS as compared with wild-type enzyme due to loss of the hydrogen bond between bound Me 2 SO and Tyr-114 (24,25).
The Kinetic Behavior of W116F Me 2 SO Reductase-Appreciating that the W116F mutant regains bis(enedithiolate) coordination on redox-cycling, the kinetic behavior of the mutant was next examined. Despite modest differences in its absorption spectrum (and presumably active-site coordination) (Fig. 1C), the reduced W116F mutant reacts very rapidly with Me 2 SO to give the E red ⅐Me 2 SO intermediate, whose spectrum is comparable with that seen with wild-type enzyme. As with wild-type enzyme, this species subsequently decays rapidly to the oxi-dized form with the W116F mutant (k obs ϭ 61 s Ϫ1 ; Table 1) considerably faster than the 7 s Ϫ1 seen for k cat in the steadystate assay with the mutant (25). The redox-cycled W116F mutant is, thus, fully catalytically competent. We attribute the relatively slow steady-state rate by Ridge and coworkers (25) to Q pterin dissociation (with concomitant loss of activity) in the course of repeated turnovers. As with the Y114F mutant, the end point spectrum seen in the reaction of the reduced W116F mutant with Me 2 SO indicates no residual E red ⅐Me 2 SO complex formation (Fig. 2C). Again, the E red ⅐Me 2 SO^E ox ϩ DMS equilibrium lies further to the right for the W116F mutant than is seen with wild-type enzyme.
Although the reaction of reduced W116F Me 2 SO reductase with Me 2 SO is straightforward, the reductive half-reaction and enzyme-monitored turnover experiments are more complex. With the W116F mutant, significantly less Mo(V) intermediate accumulates in the course of the reaction with dithionite. As followed at 474 nm, an initial absorbance decrease is followed by an increase with rate constants of 0.74 and 0.41 s Ϫ1 ( Table 1). Simulation of the reaction using these rate constants gives 15, 45, and 40% of the enzyme as oxidized, Mo(V), and reduced forms, respectively, at maximum accumulation of the Mo(V) species (Fig. 2C). The absorption spectrum obtained for the Mo(V) species resembles that observed during reductive titrations of the non-functional pentacoordinate "as-isolated" enzyme ( Fig.  1C), suggesting that the Q pterin has dissociated from the molybdenum center in the course of reaction with dithionite.
Enzyme-monitored turnover experiments with the W116F mutant using either TMAO or Me 2 SO as substrate are also complex, and in the case of the TMAO reaction, it is not possible to fit the observed spectra using only the established parent spectra (oxidized, reduced, Mo(V), and as-isolated). Given the difficulties in obtaining unique solutions using a larger number of parent spectra (including, e.g. that for the inactive five-coordinate species in both oxidized and reduced states), these data were not analyzed in greater detail. Furthermore, although the procedure used to obtain the high-g split EPR signal with the other enzyme forms (turnover with TMAO and sodium dithionite) did yield a strong EPR signal (Fig. 3C, inset), this signal is fundamentally different from that observed for any other form of Me 2 SO reductase. It is in fact similar to the low pH signal of Arabidopsis thaliana sulfite oxidase, with g 1,2,3 ϭ 2.0070, 1.9760, 1.9654 (26,29). Sulfite oxidase has an LMoO 2 center with single pyranopterin (as well as a cysteinyl ligand contributed by the protein), suggesting that the Q pterin has dissociated in the signal-giving species for the W116F mutant.
DMS Reduction of W116F Me 2 SO Reductase-The propensity of even wild-type Me 2 SO reductase to form catalytically inert forms has led to the suggestion (28) that a reverse assay involving the PES-dependent oxidation of DMS (rather than the conventional forward assay monitoring Me 2 SO reduction) is a better indicator of the catalytic competence of the enzyme. In the presence of the mediator dyes PES (or phenazine methosulfate) and the oxidant DCIP, oxidation of DMS to Me 2 SO can be monitored by the loss of absorbance of DCIP at 600 nm as the dye becomes reduced. Previous studies by Ridge et al. (27) failed to observe any reverse activity using as-isolated W116F Me 2 SO reductase, consistent with the hypothesis that the as- isolated mutant enzyme has a non-functional five-coordinate molybdenum center. As indicated above, however, redox-cycling the W116F mutant restores the functional bis(enedithiolate) coordination. We have, therefore, examined the pH dependence of activity in the reverse DMS:DCIP/PES assay. Fig. 5 shows that although the activity observed was small, the characteristic bell-shaped pH profile is clearly observed and is consistent with DMS:DCIP/PES activity observed for both wild-type and Y114F recombinant enzymes (25). The pH optimum for wild-type enzyme is pH 8.3, with two pK a values of 7.5 and 9.1 (25), whereas maximal activity was seen at pH 9.2 with corresponding pK a values of 8.8 and 9.6 (25) with the Y114F mutant. The large errors observed for the reverse assay with the W116F mutant were the result of both low activity as well as the fact that the activity observed was only about twice that of the base-line activity measured in the absence of enzyme. Although the activity observed for the W116F mutant was ϳ100-fold lower than that of either the wild-type or Y114F mutant, the pH optimum could reasonably be assigned to pH 8.8 yielding pK a values of 8.0 and 9.3. We attribute such activity as is observed to enzyme that is (transiently) bis(enedithiolate) coordinated in the steady state. The relatively low activity observed reflects the low level of this functional form of the enzyme. Nevertheless, that the same pH profile is seen with wild-type and both Y114F and W116F mutants indicates that neither Tyr-114 nor Trp-116 is responsible for the ionizations giving rise to the pH dependence of the wild-type enzyme.
Resonance Raman of Wild-type and Mutant Me 2 SO Reductases-As a final probe of the molybdenum center in Me 2 SO reductase, we have examined the wild-type enzyme and mutants by resonance Raman spectroscopy. In general, we find that the resonance Raman spectra of wild-type enzyme for the three principal catalytic intermediates, oxidized, reduced, and E red ⅐Me 2 SO, are in very good agreement with the earlier work of Garton et al. using the R. sphaeroides enzyme (5). Key vibrational modes are given in Table 2. Fig. 6 shows the resonance Raman spectrum of the oxidized W116F and Y114F mutants compared with that for the recombinant wild-type enzyme (in all cases, redox-cycled to ensure bis(enedithiolate) coordination) using an excitation wavelength of 647.1 nm. Vibrational modes in the MoOS stretching region (330 -400 cm Ϫ1 ) for the W116F mutant in general downshift slightly (no more than 5 cm Ϫ1 ) as compared with the wild type (indicating slightly longer MoOS bond distances in this mutant), but the overall form of the spectrum is not significantly changed. The MoOS stretching region for the Y114F mutant is identical to that of the wild-type enzyme (with an error of Ϯ2 cm Ϫ1 ). For wild-type enzyme, the MoAO stretching frequency is 871 cm Ϫ1 with maximal enhancement with 568.2 nm excitation. The MoAO stretching mode frequency for the Y114F mutant is also 870 cm Ϫ1 with very low intensity at 647.1-nm excitation. For the W116F mutant, the MoAO stretching frequency is 870 cm Ϫ1 , and the intensity of this mode relative to the neighboring mode at 856 cm Ϫ1 is essentially unchanged with excitation wavelengths ranging from 488 to 647.1 nm. Of greater significance is the "sharpening" of the high energy CAC stretch at 1569 cm Ϫ1 accompanied by an ϳ20-cm Ϫ1 shift to lower energy in the W116F mutant. This indicates a weaker CAC bond and greater -delocalization within the dithiolene unit, reflecting a more symmetrical disposition of the two enedithiolate units in the molybdenum coordination sphere for W116F relative to the  wild-type enzyme. The resonance Raman spectrum of the Y114F mutant is nearly identical to that of wild-type enzyme above 1000 cm Ϫ1 with only a slight (5 cm Ϫ1 ) downshift of the highest energy CAC stretch. Maintenance of the general MoOS and CAC vibrational profile of the oxidized wild-type enzyme in the W116F and Y114F mutants clearly indicates retention of the full bis(enedithiolate) character of the molyb-denum center in the (redox-cycled) mutants, consistent with the results discussed above.
The absorption spectrum of the dithionite-reduced W116F mutant resembles that of the wild-type E red ⅐Me 2 SO complex rather than free reduced enzyme, as reflected in absorption bands at 470 and 560 nm. Fig. 7 shows the resonance Raman spectra for the dithionite-reduced W116F as compared with the wild-type enzyme. Despite the difference in absorption spectra, it is evident that the dithionite-reduced W116F Me 2 SO reductase is nearly vibrationally equivalent to the reduced wild-type enzyme. The only significant difference is in the high energy CAC stretch which shifts to slightly lower energy (ϳ10 cm Ϫ1 ) in the W116F mutant. As expected, the dithionite-reduced Y114F mutant is very similar to the wild-type enzyme in the dithiolene stretching region with modes at 1002, 1014, 1045, 1074, and 1128 cm Ϫ1 identical to wild-type within the error of the instrumentation. As seen with the W116F mutant, the high energy CAC stretch shifts to lower energy by 6 cm Ϫ1 . The increased intensity of the modes at 1002, 1155, and 1524 cm Ϫ1 in the spectrum for the Y114F mutant indicates an increased contribution from the pentacoordinate form of the enzyme. The MoOS stretching region of the Y114F mutant is similar to that of the wild-type enzyme with modes at 346, 387, and 401 cm Ϫ1 (349, 384, and 401 cm Ϫ1 in wild-type), but where there is only a single intense mode at 365 cm Ϫ1 in wild type (and the W116F mutant), there are two modes in the Y114F mutant spectrum at 357 and 371 cm Ϫ1 . The additional mode may be due to enhancement of a MoOS stretch from a small portion of the pentacoordinate form of the enzyme upon Q pterin dissociation.
The resonance Raman spectra of the E red ⅐Me 2 SO complex with wild-type enzyme and Y114F mutant are shown in Fig. 8. The spectrum of wild-type enzyme exhibits modes at 495 and 863 cm Ϫ1 that are only seen when Me 2 SO is present. These modes have been assigned to the MoOO stretch and the SAO stretch of bound Me 2 SO, respectively, based on the isotopic  shifts of these two modes upon 18 O substitution (5). The similar MoOS stretching regions of the wild-type reduced and E red ⅐Me 2 SO spectra indicate that the molybdenum center is reduced in the latter species. With the Y114F mutant resonance Raman spectrum, no Me 2 SO modes at ϳ495 and 863 cm Ϫ1 were observed (Fig. 8), indicating a significant weakening of the sulfur-oxygen bond. That the complex had indeed formed was reflected in the obvious color change to deep pink on the addition of DMS to the oxidized enzyme, an indicator of E red ⅐Me 2 SO complex formation; this color change persisted after the sample was frozen at 30 K. The MoOS stretching region does indicate reduction of the molybdenum center, but there is some evidence of Q pterin dissociation; in the CAC stretch region the mode at 1523 cm Ϫ1 (P pterin CAC stretch from the pentacoordinate form) is 5-6-fold more intense than the 1568 cm Ϫ1 mode (the Q pterin CAC stretch).

DISCUSSION
Here we report a comprehensive rapid kinetic study of wildtype R. capsulatus Me 2 SO reductase as well as Y114F and W116F mutants, in conjunction with a detailed spectroscopic analysis of the several catalytically relevant forms. As expected, the kinetic and spectroscopic properties of the wild-type R. capsulatus enzyme are very similar to those of the closely related enzyme from R. sphaeroides, although it is more prone to product inhibition by DMS. Both Y114F and W116F mutants destabilize the E red ⅐Me 2 SO intermediate seen in the catalytic sequence and largely relieve the enzyme from such inhibition. The Y114F mutant reacts somewhat more rapidly with both Me 2 SO and TMAO than does wild-type enzyme (but at the expense of reduced substrate specificity); it reacts with dithionite somewhat more slowly than does wild type but forms the same high-g split Mo(V) intermediate. In enzyme-monitored turnover experiments with the Y114F mutant, this Mo(V) species accumulates essentially quantitatively during turnover with TMAO. That the EPR spectrum is unchanged on mutation indicates that Tyr-114 does not interact significantly with the Mo(V) form of the molybdenum center in the wildtype enzyme. With the W116F mutant, the observed EPR signal is similar to that seen with A. thaliana sulfite oxidase at low pH (26,29), most likely reflecting dissociation of the Q pterin from the molybdenum in the course of the reaction.
The W116F mutant is isolated in a non-functional five-coordinate state but is converted to the functional six-coordinate bis(enedithiolate) form by reduction and reoxidation with Me 2 SO (so-called redox cycling). It is this form of the enzyme that likely accounts for the steady-state activity of the mutant reported previously (25), and indeed the reconstituted form of the W116F, once reduced, reacts somewhat more rapidly with Me 2 SO than does the wildtype enzyme. In the context of earlier controversies regarding molybdenum coordination in the functional form of the enzyme, our results underscore that it is the bis(enedithiolate) form that is active, even in the W116F mutant.
The absorption spectra for the several catalytically relevant species (including oxidized, high-g split, E red ⅐Me 2 SO, and reduced forms) for wild-type enzyme and the Y114F and W116F mutants have also been determined. Mutation of Tyr-114 to Phe results in little change in the absorption spectrum of oxidized or reduced enzyme. The spectrum of the E red ⅐Me 2 SO complex of the Y114F mutant is significantly perturbed as compared with wild-type enzyme, consistent with the proposed hydrogen bond to bound Me 2 SO in the E red ⅐Me 2 SO complex and the observed kinetic and thermodynamic destabilization of the intermediate. Mutation of Trp-116 to Phe, on the other hand, causes a blue shift of the long-wavelength absorption of (redox-cycled) oxidized mutant to 680 nm (from 720 nm for wildtype enzyme). Interestingly, the reduced W116F mutant resembles the E red ⅐Me 2 SO species of wild-type enzyme, a point discussed further below. The E red ⅐Me 2 SO species is destabilized to an even greater degree in the W116F mutant, to the point that the addition of DMS to the oxidized mutant (the most convenient way to form this species with wild-type or Y114F enzyme, albeit only incompletely in the latter case) does not yield the E red ⅐Me 2 SO species but, rather, the inactive, five-coordinate form of the mutant. The intermediate is observed transiently, however, in the reaction of the reduced W116F mutant with Me 2 SO.
The spectrum of the reduced W116F mutant has absorption maxima at 470 and 550 nm and is reminiscent of that of the  DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 JOURNAL OF BIOLOGICAL CHEMISTRY 35527 E red ⅐Me 2 SO complex seen with wild-type enzyme. Although the absorption bands of the reduced mutant (in the absence of Me 2 SO) are considerably weaker than those of the wildtype enzyme, they still presumably arise from metal to (substrate) oxygen charge-transfer transitions (5). In the low temperature x-ray absorption spectroscopy of wild-type reduced Me 2 SO reductase, there is evidence for a bound water molecule not seen in the crystal structure of the reduced enzyme (4). The reduced W116F enzyme is pink in color to the eye, very similar to that of the Me 2 SO-bound complex of wild-type enzyme. We suggest that this is due to binding of the water molecule to the reduced molybdenum center of W116F, the oxygen atom of solvent mimicking bound substrate in the E red ⅐Me 2 SO complex to some degree. It has been previously noted that the reduced wild-type enzyme, which is lime green in color at room temperature, turns pink upon freezing (12), a transition that presumably also reflects the temperature-dependent binding of solvent to the molybdenum center. Such thermochromic behavior is a well documented phenomenon and is of considerable interest in the material sciences. Precedence in systems relevant to Me 2 SO reductase has recently been shown in the [MoO(qdt) 2 ] Ϫ model compound, where on lowering the temperature below 250 K, the green-colored Mo(V) complex abstracts an electron from one of the dithiolene moieties, resulting in an orange Mo(IV) complex with a delocalized qdt⅐ radical (19). Such a formal electron transfer process is unlikely in the case of the enzyme, but it is possible that ambient thermal energy may be sufficient to disrupt binding of solvent to the wild-type reduced center at room temperature. The implication is that Trp-116 serves to prevent formal binding of water, which may represent a first step in the process of Q pterin dissociation.

Spectroscopic and Kinetic Studies of Me 2 SO Reductase Mutants
In any case, the data presented here suggest that turnover of the W116F mutant with Me 2 SO involves several steps, as shown in Scheme 1. The pentacoordinate asisolated species is first reduced to the solvent-bound bis(enedithiolate) Mo(IV) species, passing through a Mo(V) species in which one of the sulfurs of the Q pterin enedithiolates has possibly re-coordinated to the metal (this species also appears during one-electron reduction of the bisdithiolene-oxidized enzyme). Water coordinates weakly to the reduced enzyme. The addition of Me 2 SO to the reduced mutant displaces this water, leading to an oxidative half-reaction closely resembling that of the wild-type enzyme; that is, rapid formation of the E red ⅐Me 2 SO species (via a preceding Michaelis complex) followed by oxygen atom abstraction to give the bis(enedithiolate)-coordinated oxidized enzyme. Unlike the wild-type enzyme, however, formation of the E red ⅐Me 2 SO complex by rebinding of dissociated DMS to the oxidized mutant enzyme does not occur (consistent with the low activity observed for the non-physiological reverse assay). In fact, the addition of DMS to oxidized enzyme results in formation of the pentacoordinate as-isolated enzyme. A predisposition for Q pterin dissociation is likely the basis for the relatively low steady-state catalytic activity previously observed (26) for the W116F mutant (as distinct from the larger rate constants for single turnover as determined here). Our results indicate that W116F functions catalytically with full coordination to both pyranopterin cofactors, but that mutation of Trp-116 to Phe results in a greater propensity for the Q pterin to dissociate in the course of turnover.
Resonance Raman analysis of the Y114F and W116F mutants has also been undertaken. Importantly, once redox-cycled, our results indicate that the molybdenum center of the latter mutant possesses full bis(enedithiolate) coordination as seen in the wild-type enzyme, although MoOS bond lengths are somewhat longer than in the wild-type enzyme (as evidenced by downshifting of a majority of the distinctive MoOS stretching modes Ͻ 400 cm Ϫ1 ). There is also less double bond character and greater -delocalization within the dithiolene moiety of the pyranopterin cofactor. The modest structural changes giving rise to these vibrational differences appear to have relatively little effect on catalysis as the reaction of reduced enzyme with Me 2 SO appears to be largely unaffected by the mutation.
The present work demonstrates the role of Tyr-114 in stabilizing the E red ⅐Me 2 SO complex encountered in the course of catalysis by hydrogen-bonding to the oxygen of bound substrate. Although not essential for enzyme turnover, it plays a role in substrate binding, as reflected in the effect of its mutation on K m as well as on the E ox ⅐DMS^E red ⅐Me 2 SO equilibrium in the active site. This may be of major significance in the context of the use of Me 2 SO as an electron acceptor by bacteria in the environment since Me 2 SO concentrations are often in the submicromolar range (3). Trp-116, on the other hand, appears to be important less from a strictly catalytic standpoint than in helping to maintain the structural integrity of the molybdenum active center. By inhibiting the binding of water to the reduced form of the enzyme, displacement of the Q pterin from the molybdenum coordination sphere by a second MoAO group is retarded.