Revisitation of the βCl-Elimination Reaction of d-Amino Acid Oxidase

d-Amino acid oxidase (DAAO) from pig has been reported to catalyze the β-elimination of Cl− from βCl-d-alanine via abstraction of the substrate α-H as H+ (“carbanion mechanism”) (Walsh, C. T., Schonbrunn, A., and Abeles, R. H. (1971) J. Biol. Chem. 246, 6855–6866). In view of the fundamental mechanistic importance of this reaction and of the recent reinterpretation of the DAAO dehydrogenation step as occurring via a hydride mechanism, we reinvestigated the elimination reaction using yeast DAAO. That enzyme catalyzes the same reactions as the pig enzyme but with a much higher efficiency and a substantially different kinetic behavior. The reaction is initiated by a very rapid and fully reversible dehydrogenation step. This leads to an equilibrium (kon ≈ kreverse) between the complexes of oxidized enzyme-βCl-d-alanine and reduced enzyme-βCl-iminopyruvate. In the presence of O2 the latter complex can partition between an oxidative half-reaction and elimination of Cl−, which proceeds at a rate of ≈50 s−1. This step forms a complex between oxidized enzyme and enamine that is characterized by a charge transfer absorption (which describes its rates of formation and decay). A minimal scheme that lists relevant steps of the reductive and oxidative half-reactions and elimination pathways along with the estimate of the corresponding rate constants is presented. β-Elimination of Cl− is proposed to originate at the locus of the enzyme-βCl-iminopyruvate complex. A chemical mechanism that can account for elimination is discussed in detail.

In their seminal paper that appeared in this journal in 1971, Walsh et al. (1) reported that upon incubation with D-amino acid oxidase (DAAO) 2 (EC 1.4.3.3) ␤Cl-D-alanine (␤Cl-D-Ala) eliminates Cl Ϫ , which was later confirmed by others (2)(3)(4). Based on this, the mechanism depicted in Scheme 1 for the elimination reaction was put forward (1). In this reaction the substrate ␣H as H ϩ is abstracted to form a carbanion intermediate, which then releases Cl Ϫ to yield the final products pyruvate and NH 4 ϩ . In this reaction the redox state of the flavin cofactor remains unaltered.
"Normal" dehydrogenation of ␤Cl-D-Ala would yield ␤Clpyruvate (␤Cl-Py) as the final product and lead to reduction of the flavin cofactor (5). The initial experiment (1) provided the first clues for formulating a general concept for the mechanism underlying flavin dehydrogenation and represents the birth of the so-called carbanion mechanism. This concept then gained wide acceptance for decades to come. However, some doubts soon emerged (2, 6 -8); experiments by Hersh and Jorns (9) demonstrated that for a DAAO in which the flavin cofactor was replaced by 5-deaza-flavin, Cl Ϫ was not eliminated. This puzzle remained unanswered for a long time, as the flavin ought not to be involved directly in elimination. Later experiments based on the concept of the linear free energy relationship were not in favor of the carbanion mechanism either (10,11). A way out of the conundrum emerged when the three-dimensional structure of two related DAAOs was identified (12,13); these studies showed unambiguously that there is no functional group at the active center of the enzyme that could act as a base in abstracting the substrate ␣H as H ϩ . Furthermore, the structures of complexes of Rhodotorula gracilis D-amino acid oxidase (RgDAAO) with D-alanine or D-CF 3 -alanine show that the substrate ␣C-H points directly toward (the Lowest Unoccupied Molecular Orbital, of) the flavin N(5) and is thus poised for hydride transfer (13). The various arguments in favor of a hydride transfer mechanism for flavin-mediated dehydrogenation reactions have been discussed extensively elsewhere (14,15). In the meantime, experimental evidence from a variety of flavoproteins has converged at the "hydride transfer" mechanism and shows it to be generally valid (14). Despite this, the carbanion mechanism is still presented to a general audience as the mechanism of choice, e.g. in modern biochemistry textbooks.
Although the mechanistic issue (carbanion versus hydride transfer mechanisms) might be regarded as being solved, the basic question as to how Cl Ϫ is eliminated is still open. This is of cardinal importance for two reasons; (i) because the hydride mechanism as such cannot induce elimination, it assumes that elimination proceeds via a different mechanism for which the flavoprotein must possess intrinsic prerequisites, and (ii) strictly speaking, a mechanism cannot be generally valid if it fails to provide a rationale for a connected experimental observation.
Therefore, we set out to provide an answer to the aforementioned unresolved questions. We investigated the ␤-elimination reaction using some modern methods and, in particular, using RgDAAO. The kinetic behavior of the latter DAAO differs from that of the enzyme from pig (pkDAAO) (16); the substrate dehydrogenation step is substantially faster, and the rate-limiting step is different. This has turned out to be the key for the mechanistic interpretations presented here.

EXPERIMENTAL PROCEDURES
Enzymes and Buffers-D-Amino acid oxidase from R. gracilis was produced and purified from recombinant BL21(DE3)pLysS E. coli cells carrying the pT7-DAAO expression plasmid as reported by Molla et al. (17). Composite buffer was 15 mM boric acid, 15 mM phosphoric acid, 15 mM sodium carbonate anhydrous, 1% glycerol, adjusted to the desired pH with NaOH, with a working pH range of 6.0 -9.0.
Absorption Measurements and Pyruvate Molar Extinction Coefficient-Enzyme concentration is indicated in terms of flavin content using an ⑀ 455 nm ϭ 12,600 M Ϫ1 cm Ϫ1 (16,17). The pyruvate molar extinction coefficient at 320 nm was determined as 20 Ϯ 1 M Ϫ1 cm Ϫ1 by linear fitting of the absorbance values of samples containing increasing, known concentrations of pyruvate in composite buffer, pH 7.0, at 25°C.
NMR Spectra-A JEOL (GX 400) 400-MHz instrument was used. Samples were Ϸ0.5 ml in 5-mm tubes, and measurements were performed at 28°C; 8 -16 pulses were recorded and averaged for Fourier transformation. Spectra evaluation and transformation was performed with MestReC.
Rapid Reaction (Stopped Flow) Measurements-Rapid reaction measurements were carried out in composite buffer at 25°C using a stopped-flow spectrophotometer equipped with a thermostat and a diode array detector (J&M Analytische Messund Regeltechnik GmbH) as detailed (18 -20). All concentrations mentioned in these experiments refer to those after mixing. See the supplemental material "Additions to Experimental Procedures" section for further conditions. The rate of chloride release was assessed by measuring the change in pH as detailed in the legend to Fig. 10 HPLC Analysis-HPLC was performed as described earlier (21).

The ␤-Elimination from ␤Cl-D-Ala Produces Pyruvate and Cl-pyruvate in the Presence of O 2
We studied the reaction with RgDAAO in the pH range from 6 to 10, whereas Walsh et al. (1) conducted the reaction at pH 8.5 using pkDAAO. With respect to product formation, qualitatively similar pictures emerge for both DAAOs, but important and substantial differences are also apparent. In view of the shortcomings of the method used by Walsh et al. (Ref. 1; derivatization using 2,4-dinitrophenylhydrazine and thiosemicarbazide), we followed the disappearance of ␤Cl-D-Ala and the formation of pyruvate and that of other products by HPLC, as described in Gibson et al. (22). In this way the amounts of involved species can be estimated as a function of time. A representative time course of formation of pyruvate and disappearance of ␤Cl-D-Ala is depicted in Fig. 1 for the reaction conducted at pH 7.0. At pH 7.0 and in the presence of Ϸ1 mM O 2 (100% saturation), formation of pyruvate (Py) and the disappearance of ␤Cl-D-Ala proceed at identical rates, suggesting that the two processes are directly kinetically linked (Fig. 1). Under these specific conditions (␤Cl-D-Ala ϭ 20 mM) an estimated 85% of the ␤Cl-D-Ala is converted to pyruvate via Cl Ϫ elimination. This contrasts sharply with the case of pkDAAO in which close to 100% ␤Cl-Py is formed at 100% O 2 saturation (i.e. no elimination occurs) (1). It should be noted here that the formation of HCl continuously lowers the pH of the reaction mixture, which affects the course of the reaction and, at pH Ͻ 7, induces enzyme denaturation. The relative amount of Py formed (corresponding to the elimination reaction) is fairly constant from pH 6 to 7; however, it decreases substantially at pH Ն 8 (not shown, see also Fig. 4). A second, primary product formed in the presence of O 2 is ␤Cl-Py (1). ␤Cl-Py in aqueous solution is  ␤Cl Elimination by Yeast DAAO assumed to exist mainly in its hydrated form and so far has not been possible to attribute to a specific peak or to estimate the quantity in HPLC elution chromatograms (supplemental Fig. S1, with peaks eluting at Ͼ6 min being attributed to products resulting from secondary reactions of ␤Cl-Py). The absence of substantial amounts of ␤Cl-Py as the product of ␤Cl-D-Ala dehydrogenation with RgDAAO has been confirmed by 1 H NMR spectra (Fig. 2, see also supplemental Fig. S2); at pH 9 a signal was observed at Ϸ4.05-4.1 ppm, a position that is compatible with the presence of hydrated ␤Cl-Py. However, far lower amounts, as estimated from proton integration, were detected than would be expected if ␤Cl-Py was a main primary product; Ͻ20% CH/D pyruvate or CH/D acetate is formed in the same incubation. Details on the analysis of products from the reaction of ␤Cl-D-Ala are presented in the supplemental material (see specifically supplemental Figs. S1 and S2). From the present results we conclude that Walsh et al. (1) did not detect primarily formed ␤Cl-Py but likely products resulting from secondary reactions.

Estimation of the Ratio of Cl-Pyruvate Formation versus Cl Ϫ at pH 8
From the aforementioned description (see also supplemental material, "Comments to NMR Experiments" section) it is evident that amounts of ␤Cl-Py and pyruvate cannot be determined by using direct methods. We thus attempted to estimate the quantity of primarily formed ␤Cl-Py by correlating it to the consumption of ␤Cl-D-Ala and dioxygen. This is based on the stoichiometry of the two competing reactions as shown in Equations 1a and 1b) (see below and the legend to Fig. 4 for details), is the normal dehydrogenation reaction in which 1 eq O 2 is consumed/molecule of ␤Cl-D-Ala to form 1 eq of ␤Cl-Py and H 2 O 2 , whereas Equation 1b is the ␤-elimination reaction that consumes ␤Cl-D-Ala but no O 2 . For this experiment we adapted the method of Gibson et al. (22) in which the time dependence of the oxidation state of a flavoprotein (which changes with time according to the reactions in Equation 1a) is monitored during turnover by using its absorption at 450 nm; see also Refs 17,20,and 23). In the specific experiment (Fig. 3) Abs 450 nm decreases very rapidly immediately upon mixing the reactants, corresponding to an apparent, partial reduction of the enzyme. As a result and for up to Ϸ40 -60 s, the system enters a stationary (turnover) phase in which E ox and E red (oxidized and reduced enzyme forms) are present in a ratio of Ϸ3-5:1. This is deduced based on the value of "start" Ϸ100% E ox and the end absorbance obtained with ␤Cl-D-Ala ϭ 2 mM, which corresponds to that of E red . After the stationary phase, the residual O 2 concentration Ͼ ␤Cl-D-Ala concentration, E ox is (re)formed. This is the case when the starting ␤Cl-D-Ala is Յ1.0 mM. When ␤Cl-D-Ala is present in sufficiently large excess over O 2 , the latter becomes exhausted, and RgDAAO is eventually fully converted to the E red form. This occurs when the starting concentration ␤Cl-D-Ala ϭ 2 mM (Fig. 3). From this it can be estimated (see inset of Fig. 3) that at ␤Cl-D-Ala Ϸ1.5 mM the system would end up in an intermediate situation where E ox and E red remain unaltered over time. Under these conditions, Ϸ0.25 mM (ϭinitial O 2 ) out of 1.5 mM In this section some residual ␤Cl-D-Ala is present (Ϸ4.0 ppm) in addition to several unidentified signals that arise during the measurement (see also supplemental Figs. S1 and S2). X denotes a singlet that is attributed to the hydrated form of Cl-pyruvate. The identity of the signal denoted by ? is discussed in the supplemental material, " Comments to NMR Experiments" section. ␤Cl-D-Ala would have been converted via normal, oxidative turnover, whereas Ϸ1.25 mM would have undergone Cl Ϫ elimination. The ratio of pathways 1b/1a can thus be estimated as Ϸ5, and k elim Ϸ5 k [E red -␤Cl-Py][O 2 ] under the specific conditions of Fig. 3 (k is the apparent rate constant for the dehydrogenation reaction as reported in Eq. 1a), where ␤Cl-D-Ala concentration is in the same range as O 2 concentration. The latter cautionary statement is necessary because the reaction with dioxygen contains a second-order term whose rate constant can only be estimated as Ϸ10 Ϫ6 M Ϫ1 s Ϫ1 (see below) (16). This experiment demonstrates the competitive behavior of oxidative and elimination pathways and their dependence on the ratio of the reagent concentrations.

Incorporation of the ␣*H of ␤Cl-D-Ala into Pyruvate
An intriguing feature of the elimination reaction is the (partial) retention of labeled ␣*H of ␤Cl-␣*H-D-Ala in ␤-position of the ketoacid product, an issue that has to be taken into account for formulating alternate mechanisms (see Equation 2) (1, 4). With ␤Cl-␣ 3 H-Ala and pkDAAO at pH 8.5, a 20 -40% label retention was found (1,4). With RgDAAO we carried out the elimination reaction in a NMR tube using ␤Cl-␣ 2 H-D,L-Ala or ␤Cl-␣ 1 H-D-Ala in D 2 O (Fig. 4). It should be pointed out that, as Walsh et al. (1) also noticed, the rate of the reaction progressively slows down with time, whereas repeated equilibration with O 2 (shaking with air) restores elimination activity.
The conversion was followed until Ն90% of the ␤Cl-D-Ala was consumed. In the NMR spectra CH 3 -and CH 2 2 H-pyruvate are easily discerned (Fig. 4). CH 3 -pyruvate shows a narrow singlet at 2.360 ppm, whereas CH 2 D-pyruvate exhibits a (1/1/1) triplet centered at Ϸ2.344 ppm. The fine structure of this latter signal results from the 1 H 2 -2 H coupling in the CH 2 2 H group. Integration of the signals from the reaction of ␤Cl-␤[ 1 H]Ala in D 2 O at pD 7.3 indicates that out of the total hydrogens ( 1 H), Ϸ50% are in CH 3 -, and Ϸ50% are in CH 2 D-pyruvate, i.e. 3 H versus 2 H, respectively, thus yielding a ratio of 50/3 versus 50/2 for CH 3 /CH 2 D ϭ 2/3 in the product pyruvate. When the reaction is started from ␤Cl-D,L-␣ 2 H-Ala, essentially only CH 2 Dpyruvate is produced. From this it follows that at pD 7.3 the extent of ␣ 2 H label "loss" is Ϸ60%, with the remaining Ϸ40% corresponding to retention. Here we want to stress that possible KIEs were not being considered in these estimates. The reaction at pD 8.3 and 9.3 is more complex as the "transient inactivation phenomenon" (1) is more pronounced and requires repeated equilibration with air to attain Ϸ90% conversion. At pD 8.3 and 9.3 a similar amount of retention is found (see Fig. 4). An analogous experiment was carried out using ␤Cl-D,L-2 H-Ala in H 2 O at pH 8.0. The integration of the signals corresponding to CH 3 -and CH 2 D-pyruvate was Ϸ55 and 45% (supplemental Fig. S2B), suggesting that solvent-borne 2 H was incorporated in the product to a somewhat larger extent than for incubations of ␤Cl-D-  gest they belong to unidentified products of secondary reactions of ␤Cl-Py, probably with residual ␤Cl-D-Ala.

Kinetic Studies
Two Charge Transfer (CT) Intermediates Are Detected during the Course of the Reaction-A detailed study of the spectral course of the reaction of RgDAAO with ␤Cl-D-Ala was carried out at pH 6 -9 using the stopped-flow instrument (see "Experimental Procedures"). At pH 8 (supplemental Fig. S3A) and 9 ( Fig. 5) the courses are very similar, although at pH 9 the various intermediates are distinguished best. This is shown in Fig. 5 where the first spectrum was recorded at 0.8 ms upon mixing the reactants. This spectrum is characterized by an Ϸ50% decrease in the original absorption of the oxidized flavin in the 450-nm region. Importantly, the phase leading to this spectrum occurs almost completely during the "dead time" of the instrument and is just about completed at 2-3 ms (see the traces in Fig. 5 and supplemental Fig. S3). Concomitantly, a long wavelength absorbance forms at Ͼ520 nm that is attributed to a CT complex. This species (named intermediate 1 charge transfer (I1-CT)) presumably consists of a mixture of chromophores derived from oxidized and reduced flavin in a ratio Ϸ1:1. This deduction is based on the observed ratio of the 440-nm absorbance of oxidized (spectrum recorded at 0 ms) and reduced enzyme (spectrum recorded at 12 s). I1-CT then converts to a second charge transfer species (I2-CT) with higher absorbance both in the 440-and 550 -650-nm region. This species attains maximal absorbance at Ϸ100 ms (Fig. 5), where the system enters a short stationary phase. At pH 8, 7, and 6, a qualitatively similar behavior is observed (see the panels of supplemental Fig.  S3), although the intensities of the corresponding species are significantly different. The absorption spectrum of I2-CT (Fig.  5) suggests that its main component is oxidized enzyme flavin in complex with an electron donor, which gives rise to the CT absorption observed at Ͼ520 nm. The time dependence of the absorbances was also analyzed using the application SpecFit with which the spectra of intermediates can be identified in sequential processes (see the supplemental material "Additions to Experimental Procedures" section for details); the spectra obtained by this deconvolution procedure for I1-CT and I2-CT can be superimposed on those recorded directly (see Fig. 5). A decrease in the absorbance of I2-CT then ensues. It leads at Ϸ12 s to a final species with a spectrum that is closely similar to that of free reduced RgDAAO (16), this occurring concomitantly with oxygen consumption in the system. Notably, the extent of absorbance increase in the 320-nm region at pH 9 (Ϸ0.05 absorbance units) is small compared with that of the same experiment at pH 6 (supplemental Fig. S3C). This is consistent with formation of small amounts of pyruvate at pH 9 (Ϸ2.5 mM from 100 mM ␤Cl-D-Ala in the experiment of Fig. 5) in contrast to substantial amounts at pH 6 (supplemental Fig.  S3C).
The ␤-Elimination Reaction at pH 6 in the Presence of O 2 -Although basically the same products are formed as at pH 7, 8, or 9, the spectral course of the reaction at pH 6 (supplemental Fig. S3C) under similar conditions of reactant concentrations is substantially different from the former one (see Fig. 5 for pH 9) as demonstrated by the following observations; (i) the decrease in absorbance in the 450-nm region that occurs during the dead time of the instrument and corresponds to the formation of I1-CT is Ϸ10% compared with Ϸ50% at pH 9 ( Fig. 5) or 8 (supplemental Fig. S3A), (ii) I1-CT is formed at a slower rate, its maximal formation is at Ϸ2-3 ms (supplemental Fig. S3C), and it is also converted at a much slower rate into I2-CT, (iii) the reaction takes Ͼ10 times longer to reach completion, (iv) at that point the enzyme exists largely in the oxidized state, (v) the amount of pyruvate formed is much larger, as reflected by the absorbance increase at 320 nm (inset of supplemental Fig. S3C), and (vi) the steady-state phase encompasses Ն100 s compared with 3-4 s at pH 9.
The ␤-Elimination Reaction at pH 8 in the Absence of O 2 -The course of the anaerobic reaction is similar at pH 9 and 8 (supplemental Fig. S4), the latter conditions corresponding to those used by Walsh et al. (1) under which pyruvate was reported to be formed exclusively. In the present case only Յ20% of the possible amount of pyruvate is formed (see the time course of absorbance at 320 nm in the inset of supplemental Fig. S4 and compare with the inset to Fig. 5). Drastic differences can also be observed when comparing the spectral courses of the pkDAAO (1) and RgDAAO incubations. Fig. 2C in Walsh et al. (1) shows that during the "steady-state" phase, a species is present that contains predominantly oxidized pkDAAO and exhibits a CT absorption. At the end of the incubation, essentially all pkDAAO was in the oxidized state (1). In the present case with RgDAAO, the spectral course of the corresponding reaction is depicted in supplemental Fig. S4 and shows that the enzyme is in the reduced state at the end of the incubation. ␤Cl Elimination by Yeast DAAO NOVEMBER 25, 2011 • VOLUME 286 • NUMBER 47

JOURNAL OF BIOLOGICAL CHEMISTRY 40991
The initial events observed under anaerobic conditions are similar to what is observed in the presence of oxygen. Thus, the very first spectrum (supplemental Fig. S4) obtained at 0.8 ms upon mixing the reactants in the stopped-flow instrument reflects an Ϸ25% decrease in the 440-nm band of the oxidized enzyme that occurs during the dead time of the instrument. This initial decrease is smaller than that found in the presence of O 2 (Ϸ40%, Fig. 5), which is a counterintuitive observation. A concomitant increase occurs in the 530 -600-nm region that reflects the formation of a charge transfer complex (I1-CT) similar to that observed under aerobic conditions (Fig. 5). From Ϸ70 ms and up to 300 -400 ms the system enters a steady-state situation (see supplemental Fig. S4, inset) that is much shorter than that observed under aerobic conditions (Fig. 5) and that gradually leads to the final species within Ϸ30 s. There are two relevant differences between the behavior of RgDAAO and that of pkDAAO (1); with RgDAAO the spectral features of the final species are consistent with formation of fully reduced enzyme flavin. By comparison, at the end of the reactions and after approximately the same incubation time under similar conditions, pkDAAO ends up in the fully oxidized state (4). A second difference pertains to the quantity of pyruvate formed. This is Ϸ0.1 eq of the available ␤Cl-D-Ala with RgDAAO (estimated by the increase in absorbance at 320 nm, see supplemental Fig. S4, inset), whereas with pkDAAO up to 0.5 eq are formed (1).
As reported for pkDAAO (1), a progressive loss of activity with incubation time was also observed with RgDAAO; this is apparent from the 320-nm trace in the inset of supplemental Fig. S4. Elimination activity can be restored by admitting oxygen. The exact reason for this behavior is still elusive, but we suggest that during the anaerobic incubation a less reactive, reduced flavin form accumulates that is (re)converted to oxidized enzyme (active in ␤-elimination) upon reaction with dioxygen.

Kinetic Course of the Cl Ϫ Elimination Reaction
Evidence for Full Reversibility of the Initial Redox Step-In rapid mixing experiments such as those shown in Fig. 5, I1-CT, the very first observable species is formed almost completely during the dead time of the instrument. Fig. 6 shows that the decrease in absorbance recorded at 0.8 ms depends on the concentration of ␤Cl-D-Ala. In Fig. 6, the trace 5 is representative and was obtained at 50 mM ␤Cl-D-Ala. The inset to Fig. 6 correlates the extent of the absorbance decrease at 450 nm occurring in the dead time and the extent of formation of I1-CT with the concentration of ␤Cl-D-Ala. This is a typical situation in which the reactants very rapidly form a complex that reacts reversibly to reach an equilibrium (24), as represented by Equation 3), where K d is the dissociation constant for the rapid equilibrium of ␤Cl-D-Ala binding, k f and k r are "forward" and "reverse" steps for the ensuing redox reaction, and E red -␤Cl-IPy is the complex of reduced enzyme with the product ␤Cl-IPy. In a situation such as that in Fig. 6, an apparent dissociation constant K d,app ϭ K d ϫ (k r /k f ) Ϸ 5 mM can be estimated from the plots in the inset of Fig. 6. The concentration ratio of the species E ox -␤Cl-D-Ala/E red -␤Cl-IPy and that of the steps k r /k f that link these species can be deduced from the spectra shown in Fig. 6 as follows; the spectrum of E ox -␤Cl-D-Ala is taken as being not relevantly different from that of uncomplexed E ox (16), and the spectrum of E red -␤Cl-IPy (trace A in Fig. 6) is obtained by deconvolution with the application SpecFit. From this it is estimated that the ratios E ox -␤Cl-D-Ala/E red -␤Cl-IPy and k r /k f are approximately 2/3 under substrate saturation conditions. Consequently the reaction up to I1-CT constitutes a (fast) approach to the equilibrium depicted in Equation 3. Because k r /k f Ϸ 2/3, the dissociation constant K d for the formation of the encounter complex E ox -␤Cl-D-Ala can be estimated as Ϸ3 mM. At pH 9.0 the situation is similar, the equilibrium being also slightly in favor of the reduced species (not shown). At pH 7 the same equilibrium is Ϸ1:1, whereas at pH 6 and under saturating conditions E ox -␤Cl-D-Ala/E red -␤Cl-IPy Ϸ4/1 (not shown, compare with supplemental Fig. S3C).
Comparison of Reaction Courses under Aerobic and Anaerobic Conditions-The shapes of the spectra of I1-CT are not significantly affected by the presence of O 2 , although their intensities and specifically the ratio of the 450/550 nm absorbance differ (supplemental Figs. S3A and S4). I2-CT is formed at similar rates from I1-CT and also attains a maximum at Ϸ100 ms both in the presence and absence of O 2 (supplemental Fig.  S5). At longer time scales the absorption versus time profiles are The spectra intersect at the isosbestic points indicated by the arrows. Trace B is the spectrum of the reduced RgDAAO-pyruvate complex obtained in a parallel experiment using D-Ala as substrate. Trace A is a spectrum obtained by deconvolution using the application SpecFit. It corresponds to subtracting 36% of spectrum 0 from spectrum 5 and normalization to 100%; it is depicted to demonstrate its similarity to the spectrum of the complex of E red with iminopyruvate (E red -IPy). The inset shows the plots of the absorbances at 458 (Ⅺ) and 550 (E) nm registered at 0.8 ms and obtained using the indicated ␤Cl-D-Ala; -is the fit of the data points based on a saturation equation (apparent K d Ϸ 5 mM). EQUATION 3 ␤Cl Elimination by Yeast DAAO very different because in the absence of O 2 the enzyme is rapidly converted to the reduced form (supplemental Fig. S5), whereas under aerobic conditions the system enters a steadystate phase and takes longer to convert to the reduced form, which occurs upon exhaustion of O 2 at 15-20 s. Fig. S5), rates of I1-CT conversions into I2-CT were estimated. The results are depicted in Fig. 7 as a function of the ␤Cl-D-Ala concentration for pH 8. At all pH values saturation behavior was found to be consistent with the substrate binding step being linked to the specific kinetic process(es) observed. In all cases fitting the data with an equation that includes a reverse step k r (see Equation 3) yielded better results (expressed as R 2 in the legend of Fig. 7) than for fits without it. Furthermore, these fits show a positive intercept on the ordinate, also indicating a reversible process (24), and allow an estimation of the rate of the apparent reverse step k r . Interestingly, the apparent rate of I1-CT conversions into I2-CT (k obs ) is faster in the presence of O 2 , which is in agreement with the two specific steps being linked via common intermediate(s). The conversion of I1-CT into I2-CT does not show relevant pH dependence, the rates (extrapolation to saturating ␤Cl-D-Ala) varying between Ϸ40 (pH 6) and Ϸ80 s Ϫ1 (pH 8 and 9).

Estimation of Rate Constants of I1-CT Conversions into I2-CT-Based on the aforementioned approach (supplemental
Estimation of the Rate of the Product Enamine Dissociation from Oxidized RgDAAO-Because elimination should proceed via formation of an enamine from ␤Cl-D-Ala, it was reasoned that this enamine is the species that gives rise to a CT band in a complex with oxidized flavin enzyme. In this complex the enamine should exist in its anionic (unprotonated amine) state and serve as the donor as a protonated ␣N would most likely not have such a capacity. As also discussed below, intermediate I2-CT is thus proposed to be this E ox -enamine complex (see Scheme 2). The rate of dissociation of this complex to yield free E ox was estimated in a double stopped-flow experiment as shown in Fig. 8. The rationale behind the experiment is that the good ligand benzoate (a competitive inhibitor) (23) effectively traps E ox upon its formation (see Scheme 2) and thus impedes any further cycling of the enzymatic elimination reaction. The disappearance of the CT band with a rate Ϸ26 s Ϫ1 thus likely corresponds to enamine release, I2-CT thus being the E oxenamine complex (see Scheme 2).  (16), a direct measurement not being feasible due to the reactivity of Cl-pyruvate. Steps k 4 , k 6 , and k 7 involve dissociation of a product as well as hydrolysis by H 2 O, the two processes not being differentiated.

JOURNAL OF BIOLOGICAL CHEMISTRY 40993
Elimination Starting from Reduced Enzyme Is Associated with Flavin Reoxidation-To define the redox state at which Cl Ϫ elimination occurs, the competence of reduced RgDAAO was investigated at pH 7.0 as at this pH substantially more pyruvate forms than at higher pH values (see above). Walsh et al. (1) addressed the same question using reduced pkDAAO; they reported that ␤-elimination did not occur. In the present case RgDAAO was first made anaerobic and subsequently converted to the reduced state with a small amount of D-Ala (Ϸ1.5 molar excess, Fig. 9), whereby special care was taken to avoid the presence of any oxidized enzyme (see the comments in the supplemental material, "Comments on the Cl Ϫ Elimination from ␤ Cl-D-Ala Starting from Reduced Enzyme" section) Fig.  9 shows that reduced RgDAAO does indeed interact with ␤Cl-D-Ala. The first part of the reaction, i.e. from 1 ms to 30 s, was examined with the stopped-flow instrument (supplemental Fig.  S6). Up to 2 s nothing happens. However, subsequently and up to 30 s a progressive increase in absorbance in the 450and 550-nm region ensues that corresponds to (re)oxidation and leads to formation of Ϸ50% of the possible quantity of oxidized enzyme (see Fig. 9) and some absorption due to a CT interaction. Concomitantly, production of pyruvate (see 320-nm trace) sets in and continues progressively. Importantly, there is a 1-2-s lag phase preceding the absorbance increases at 450 and 540 nm (supplemental Fig. S6), whereas the absorbance increase at 320 nm reflecting pyruvate formation sets in at 10 -20 s (Fig. 9). The maximal rate of this process corresponds to a quasi-steady-state in which the relative concentration of oxidized RgDAAO remains maximal. Presumably there is no real discrepancy between the present results and the negative ones reported by Walsh et al. (1). In the present case the concentration of DAAO is Ϸ10-fold higher than that used in Walsh et al. (1). Furthermore, Walsh et al. (1) used an Ϸ20-fold excess of reductant (substrate); consequently, any net reoxidation of their enzyme was prevented. We conclude from these experiments that reduced DAAO is not able to carry out elimination catalytically at an appreciable rate. However, a component present in the solution, possibly ␤Cl-D-Ala, very slowly reoxidizes the reduced enzyme, thus generating the oxidized form that is competent in catalytic ␤-elimination.
Isotope Effects-To correlate the steps of dehydrogenation and elimination, the courses of the reactions of ␣ 1 H-␤Cl-D-Ala and ␣ 2 H-␤Cl-D-Ala were compared. The rates of conversion of I1-CT into I2-CT at pH 8 are k obs ϭ 97 and 73 s Ϫ1 for ␣ 1 H-and ␣ 2 H-␤Cl-D-Ala, corresponding to a KIEs of Ϸ1.3 (see the legend to supplemental Fig. S7, A and B for details of the conditions). The rate of the (re)oxidative half-reaction(s) with ␣ 2 H-␤Cl-Ala is also slowed down as the ratio of E ox /E red is lower, and the system requires a longer time during turnover to exhaust O 2 (supplemental Fig. S7B). These effects reflect a KIE Ϸ 1.5. Solvent KIE (supplemental Fig. S7, C and D) and a corresponding proton inventory were carried out to study the involvement of solvent-borne hydrogens in the elimination process. At pH 8 and when followed by the initial rate of pyruvate production, elimination proceeds with a solvent deuterium KIE Ϸ1.7. Pyruvate formation/elimination is best followed at pH Ͻ 7 where it is the major process. At pH 6.5, the proton inventory of pyruvate formation shows a fairly linear profile with a KIE Ϸ 1.7 (supplemental Fig. S7E), this being compatible with the effect  arising from a single site and a single hydrogen being involved (25). Similarly, at pH 6.5 the conversion of I1-CT into I2-CT in H 2 O versus D 2 O shows a solvent KIE Ϸ 1.5, whereas the decay of I2-CT exhibits a larger solvent KIE Ϸ 5 (supplemental Fig.  S7D).

Identification of the ␤Cl Elimination
Step-Walsh et al.
(1) estimated the rate of Cl Ϫ elimination in a rapid-quench experiment using pkDAAO. We addressed the same question by following the formation of the coproduct H ϩ . To this end, the reaction was conducted in lightly buffered solution and in the presence of the pH indicator bromthymol blue. The neutral and anionic forms of the indicator have spectra with isosbestic points at 502 and 322 nm (not shown). At these wavelengths the spectral changes going along with the conversion of intermediates I1-CT into I2-CT in the initial phase of the elimination reaction are sufficiently large to be followed in the stopped-flow instrument. Conversely, the spectra for the conversion of intermediate I1-CT into I2-CT have isosbestic points at 462-464 and 515 nm at pH 7 (not shown); thus, changes in pH can be monitored by following the absorbance changes of the indicator at these wavelengths. Inspection of Fig. 10 shows that when followed at 502 nm (isosbestic point of the indicator bromthymol blue), conversion of intermediate I1-CT into I2-CT proceeds at Ϸ35 s Ϫ1 (curves A and B, first fast phase). Curve C shows that there are no relevant absorbance changes at Ϸ463 nm (isosbestic point for the conversion of intermediate I1-CT into I2-CT) in the absence of indicator during the first reaction phase. However, in the presence of indicator an absorbance increase was observed that reflects a lowering in pH and that proceeds at Ϸ55 s Ϫ1 (curve D). Because the rates of the preceding and of the subsequent steps differ by up to 2 orders of magnitude, it is reasonable to assume that the conversion of intermediate I1-CT into I2-CT and that of H ϩ release all reflect the same chemical event, namely, Cl Ϫ elimination.

DISCUSSION
Although pkDAAO and RgDAAO share the ability to catalyze the elimination of Cl Ϫ from ␤Cl-D-Ala, substantial differences can also be seen in the respective kinetic behaviors and in particular with respect to the dependence of the process from dioxygen. Thus, in a key statement Walsh et al. (1) wrote: "… anaerobic incubations of pkDAAO with ␤-chloroalanine yield pyruvate exclusively. When similar incubations were conducted with 100% O 2 as the gas phase, the expected keto acid product, chloropyruvate, was formed almost exclusively." The yeast RgDAAO behaves substantially different in this respect, and we have exploited this to attribute specific kinetic steps to chemical events occurring during elimination and to identify the observed intermediates.
Identification and Attribution of Kinetic Steps-Because kinetic analyses were conducted at various pH values, for the sake of clarity we focus the discussion on the most representative case of pH 8 when differences to other conditions are not relevant. Fig. 3 shows that elimination, a reaction formally not involving changes in redox states, and normal dehydrogenation (as defined by substrate dehydrogenation coupled to oxygen consumption; see also Scheme 2) are concurrent events. This is in line with previous deductions (1,26) and is depicted in more detail in Scheme 2. In that scheme, the two processes share the species E ox and E red -␤Cl-Py. Catalysis starts with fast and fully reversible binding of ␤Cl-D-Ala to form the E ox -␤Cl-D-Ala complex via steps k 1 /k -1 and with a K d (k Ϫ1 /k 1 Ϸ 3 mM) that is similar to that for D-Ala (16,19). This conclusion is derived from experiments such as those of Figs. 6 and 7 that show saturation behavior of the observed initial steps (k obs ) on ␤Cl-D-Ala. Binding is followed by a very rapid reduction of E ox -␤Cl-Ala (Scheme 2) via step k 2 that is essentially completed within Ϸ1 ms and thus cannot be observed in the stopped-flow instrument (Figs. 5 and 6 and supplemental Fig. S3). The rate of k 2 is thus Ն1000 s Ϫ1 . Remarkably, in sharp contrast to other DAAO substrates but in analogy to phenylglycines (19), this reduction step k 2 is fully reversible (step k Ϫ2 ), the single steps having approximately the same value. The very first species observed spectroscopically at Ϸ1 ms (I1-CT, Fig. 5 and supplemental Fig.  S3) thus consists of an equilibrium mixture of E ox (as free E ox and E ox -␤Cl-D-Ala) and E red (as E red -␤Cl-Py, Scheme 2).
The ensuing steps depend on the presence or absence of O 2 (Scheme 2, compare Figs. 5 and 6 and supplemental Fig. S4). The initially formed I1-CT is converted to a second intermediate (I2-CT) via k 3 at a rate Ϸ50 -80 s Ϫ1 at pH 8 (Figs. 5 and 7). ␤Cl Elimination by Yeast DAAO NOVEMBER 25, 2011 • VOLUME 286 • NUMBER 47 As shown in Fig. 7, the rate of this process is dependent on O 2 to a minor extent. The rate of the reaction of I1-CT with dioxygen, k 5 , can be assumed to be similar to that of the complex of reduced enzyme with iminopyruvate E red -IPy, i.e. around 1.2 ϫ 10 5 M Ϫ1 s Ϫ1 (16). The rates of I1-CT conversion into I2-CT depend on ␤Cl-D-Ala (Fig. 7) and show finite intercepts on the ordinate. This deserves comment. As discussed in Strickland et al. (24), such an intercept corresponds to the reverse (or an equivalent combination) of the measured forward step, the latter representing an approach to equilibrium. However, it is chemically most unlikely that the microscopic reverse of step k 3 , the elimination of Cl Ϫ , would play any role kinetically. We thus interpret the apparent "reversibility" as a combination of steps that lead to reformation of E red -␤Cl-IPy. This is assumed to be achieved via steps k 4 , k 1 , and k 2 (Scheme 2), possibly also including steps k 5 and k 6 . The values found for k r ϭ 4 -8 s Ϫ1 (Fig. 7), therefore, likely reflect a combination of the aforementioned single steps.
An estimation of the value of k 7 , the dissociation of the E red -␤Cl-IPy complex, can be obtained from the rate of disappearance of the long wavelength absorbance of the species in experiments in which O 2 is either absent or has been exhausted (see Fig. 5 and supplemental Fig. S4). The value Յ2-3 s Ϫ1 is typical for the dissociation of imino acids from the corresponding complexes with reduced enzyme (16,19). The rate of the reaction of free reduced enzyme with O 2 via k 8 is taken from the literature (16,19). The last step in Scheme 2, the dissociation of I2-CT to form E ox via k 4 , was determined directly from the experiments shown in Fig. 8.
It is of crucial importance to identify the step that corresponds to the chemical event in which Cl Ϫ elimination occurs. This can be deduced from the experiments of Fig. 10, according to which elimination is concomitant with step k 3 , the transformation of I1-CT into I2-CT. This establishes that Cl Ϫ elimination is preceded by enzyme reduction/substrate dehydrogenation, a topic that was debated in previous studies (4,6,27). In kinetic terms elimination reactions that would branch off at either intermediates E ox -␤Cl-D-Ala or E red -␤Cl-IPy (Scheme 2) would be equivalent. The overall behavior of the system can thus be described by the minimal set-up of Scheme 2 where two concurring cycles share two intermediates, E ox and E red -␤Cl-IPy. The latter is of great importance as it constitutes the branching point for the oxidative and the elimination pathways. In this scheme the limiting step(s) for the normal, oxidative turnover cycle (Scheme 2, right side) is k 5 (or k 6 ), whereas for the elimination pathway this is k 4 , the release of the enamine product. This yields a rationale for the differences in spectral courses observed in the presence or absence of O 2 as shown in Figs. 3 and 5 and supplemental Fig. S4.
Chemical Identity of Intermediates-Attributing chemical entities to species I1-CT and I2-CT is of particular importance in the context of Scheme 2. The absorbance values at wavelengths Ͼ530 nm (Figs. 5 and 6 and supplemental Fig. S3) of both species are compatible with the presence of charge transfer interactions (28). I1-CT is reasonably attributed to the complex of reduced enzyme flavin with the imino acid product (E red -␤Cl-IPy, Scheme 2) in analogy to its occurrence in the reaction with normal substrates (28). I2-CT, on the other hand, exhibits the two-banded absorption of the oxidized enzyme in addition to the CT absorption at Ͼ530 nm. From this it is reasonable to assume that the oxidized flavin behaves as the acceptor in the complex. Because I2-CT is formed concomitantly with Cl Ϫ elimination, the donor in the same complex would be the resulting enamine, which also ought to be in its NH 2 -neutral form. This interpretation is in agreement with previous, general proposals (4,28).
Chemical Mechanism of Cl Ϫ Elimination-A carbanion mechanism has been excluded for the normal DAAO dehydrogenation reaction (10,13,29) mainly because of the absence of a functional group (base) that might abstract a H ϩ to form the mentioned carbanion. The validity of this argument must also apply for the Cl Ϫ elimination reaction and thus speaks against a mechanism starting from an intermediate such as E ox -␤Cl-D-Ala (Scheme 2). The kinetic data discussed above are in support of the elimination occurring "directly" from the E red -␤Cl-IPy complex. This reaction can be defined as a "reductive elimination" as it involves net transfer of 2 e Ϫ equivalents to the leaving group and (re)oxidation of the flavin. Proposals for the chemistry of such a step have been discussed earlier, e.g. in Ref. 6; some of these involve the formation of covalent adducts between the reduced flavin and ␤Cl-IPy. This concept has been reworked and expanded in Scheme 3, taking into account newer insights and the present results. The key point is that in the -complex between reduced flavin and iminopyruvate (E red -␤Cl-IPy, Scheme 2), the flavin N(5) sp 3 orbital that interacts with the -orbital at C(2) of the acceptor (imino group) can be either a free pair or an N-H bond. Depending on whether the hydrogen is in either one of these orbitals, the overlap with the imino acid -orbital can induce either (reverse) hydride transfer as in the normal reaction (step k Ϫ2 , Scheme 2) or formation of the covalent adduct (C in Scheme 3). That formation of such an adduct cannot be observed by spectroscopic means might be due simply to an unfavorable equilibrium concentration of C and/or to the kinetics of the involved steps. The formation of covalent adducts between reduced flavin and carbonyls or imines has precedents both in the chemical system (30,31) and in flavoenzymes (32,33). Cl Ϫ elimination from adduct (C) then occurs via concerted transfer of 2 e Ϫ from the flavin to the leaving group, this being classic fragmentation as described by Grob and Schiess (34). For such fragmentations precise steric orientations of involved orbitals are necessary (in general, an antiparallel one) (34); the absence of this might be a possible reason for the absence or occurrence of ␤-elimination in related enzymes (4,26,35). In turn, the occurrence or absence of ␤-elimination in various enzymes might be dictated by the set-up of the active site, which determines the steric orientation of the ligand ␣and ␤-substituents.
Of particular interest from a mechanistic point of view is the fate of the ␣*H of ␤Cl-D-Ala, which is found in the product pyruvate at C(3) after Cl Ϫ elimination. The experimental results of the present work (Fig. 4) agree in essence with those of Walsh et al. (4,26,35) in that the retention of label is 25-40%. Walsh et al. (4) interpreted this as resulting from the involvement of a triprotonic base/acid such as a lysine at the active center of DAAO, this in turn resulting in an Ϸ3-fold dilution of the label. Scheme 3 depicts a viable and attractive alternative ␤Cl Elimination by Yeast DAAO that is based on the same concept and works in the absence of such a triprotonic functional group derived from the amino acid backbone (12,13). It also involves a triprotonic base/acid, namely, the amino group of adduct C (Scheme 3). Accordingly, the label is first transferred from the ␤Cl-D-Ala ␣C to the flavin N(5) via hydride transfer to form B. Concomitant with formation of the covalent adduct C, the N(5) label is transferred to the amino group of the adduct (Scheme 3). It remains on the same nitrogen in the enamine upon Cl Ϫ elimination to form D. From this position the label is tautomerized to the C(3) position in the product iminopyruvate/pyruvate (E and F). The involvement of the triprotonic amino group in the intermediate (C) thus gives a rationale for the percentage of label incorporation (Fig. 4) and requires that there is little or no exchange of label with solvent during the elimination turnover cycle either at the reduced flavin N(5) position or at the adduct amino group. The involvement of this amino group (C) in label transfer is also in agreement with the absence of a relevant pH effect on the degree of incorporation (Fig. 4) and on the kinetics of the elimination. In the case of a general acid-base catalysis, such an effect should be manifest. Such a mechanism is also in line with the finding of KIEs of rather small magnitude. Thus, when using ␣ 2 H-␤Cl-Ala, the intrinsic KIE would be "diluted" to 1 ⁄ 3, whereas for elimination in D 2 O as solvent the dilution factor would be 2 ⁄ 3. This corresponds to an experimentally found KIE Ϸ 1.3 in the case of ␣ 2 H-␤Cl-Ala as substrate (supplemental Fig. S7, A and B) and of a solvent KIE Ϸ 1.7 in pyruvate formation (supplemental Fig.  S7E). The various differences observed on rate constants, positions of equilibria, and spectral properties between pH 6 and 9 reflect an (apparent) pK of around 7.5. A plot of the rates of pyruvate formation versus pH yields a pK Ϸ 7.6 (supplemental Fig. S8). Likely candidates for this ionization are the amino group of ␤Cl-D-Ala or that of the enamine complexed with E ox (see Scheme 2). An alternative mechanism in which a hydride from the reduced flavin releases Cl Ϫ by direct attack at ␤-C of ␤Cl-IPy in a substitution reaction is unlikely as a major process as it would lead directly to a complex E ox -IP bypassing the observed E ox -enamine complex (I2-CT).
Elimination Starting from Reduced RgDAAO-The finding of Cl Ϫ elimination starting from reduced RgDAAO is very surprising from a mechanistic point of view. From the data of Fig.  10 and supplemental Fig. S6, it can be deduced that the intrinsic activity of reduced RgDAAO in Cl Ϫ elimination must be very low. Great care was taken to ensure that RgDAAO was present exclusively in the reduced state when the reaction was started by adding ␤Cl-D-Ala. Furthermore, supplemental Fig. S6 shows that overall elimination activity increases gradually with time concomitantly with an increase in absorbance in the 450-nm area, which reflects (re)formation of oxidized enzyme. From this it appears that a component in the system, likely ␤Cl-D-Ala, promotes "reoxidation." A conceivable mechanism for this would envisage a direct interaction of the reduced flavin with ␤Cl-D-Ala in a manner comparable with that shown in Scheme 3 and mentioned above for the E red -␤Cl-Py complex. Specifically, it would involve a direct attack of the N(5)-H at the ␤Cl-D-Ala ␤-carbon in which the hydride releases Cl Ϫ in a substitution reaction (see supplemental Scheme S1). This would generate E ox that then enters elimination catalysis as shown in Scheme 2. It should be noted that such an elimination would constitute a slow side reaction (see Fig. 9 and supplemental  ␤Cl Elimination by Yeast DAAO NOVEMBER 25, 2011 • VOLUME 286 • NUMBER 47 such as the dependence of Cl Ϫ elimination from the presence of oxygen and in particular the retention of substrate ␣C*H label into the product pyruvate. These mechanisms also highlight the inherent capacity of enzymes to catalyze reactions that differ from those that take place during normal catalysis (promiscuity).