C2- (cid:1) -Lactylthiamin Diphosphate Is an Intermediate on the Pathway of Thiamin Diphosphate-dependent Pyruvate Decarboxylation EVIDENCE ON ENZYMES AND MODELS*

Thiamin diphosphate (ThDP)-dependent decarboxylations are usually assumed to proceed by a series of covalent intermediates, the first one being the C2-thiazo-lium adduct with pyruvate, C2- (cid:1) -lactylthiamin diphosphate (LThDP). Herein is addressed whether such an intermediate is kinetically competent with the enzymatic turnover numbers. In model studies it is shown that the first-order rate constant for decarboxylation can indeed exceed 50 s (cid:2) 1 in tetrahydrofuran as solvent, (cid:1) 10 3 times faster than achieved in previous model systems. When racemic LThDP was exposed to the E91D yeast pyruvate decarboxylase variant, or to the E1 subunit of the pyruvate dehydrogenase complex (PDHc-E1) from Escherichia coli , it was partitioned between reversion to pyruvate and decarboxylation. Under steady-state conditions, the rate of these reactions is severely limited by the release of ThDP from the enzyme. Under pre-steady-state conditions, the rate constant for decarboxylation on exposure of LThDP to the E1 subunit of the pyruvate dehydrogenase complex was 0.4 s (cid:2) 1 , still more than a 100-fold slower than the turnover number. Because these experiments include binding, decarboxylation, catalyzed ) al. identified of

exemplified with the reaction catalyzed by the enzyme pyruvate decarboxylase in Scheme 1 (1)(2)(3)(4)(5). The existence of LThDP on enzymes has received support from two sources recently. (a) Tittmann and coworkers (6) showed that LThDP can be trapped with acid quench after rapid mixing of pyruvate decarboxylase with substrate, and (b) the analogue of LThDP with a non-cleaveable C2-␣-P bond phosphonolactyl-ThDP (PLThDP) was found to bind to both yeast pyruvate decarboxylase (YPDC, Ref. 7) and to the pyruvate dehydrogenase-E1 subunit (PDHc-E1, Ref. 8). The analogue PLThDP was also shown to be a reasonably strong inhibitor of the PDHc-E1 and could be cocrystallized with PDHc-E1, and displays behavior expected of an intermediate analogue. 2 Yet, an earlier study by Kluger et al. (10) reported no detectable decarboxylation of LThDP when exposed to pyruvate decarboxylase.
In a pioneering chemical model reported by Lienhard and co-workers (11,12), it was demonstrated that C2-␣-lactylthiazolium (LThz) salts undergo decarboxylation rather fast, and the rate of decarboxylation is dramatically accelerated by solvents of lower dielectric constant, such as ethanol. That C-C bond cleavage is rate-limiting in this reaction was confirmed by 13 C/ 12 C kinetic isotope effect studies (13). Kluger (2) subsequently described in detail the pH dependence of the rate of decarboxylation of C2-␣-lactylthiamin (LTh), showing that the rate is faster in the presence of the 4Ј-aminopyrimidine ring of the coenzyme. The unimolecular decarboxylation rate constants for these model reactions were considerably smaller than the value of the turnover numbers for such enzymes; the latter have typical values of ϳ60 s Ϫ1 /active site.
In this paper we address two issues related to the existence of this putative LThDP covalent intermediate. (a) We extended the model studies to solvents of lower dielectric constants and could achieve first-order decarboxylation rate constants as large as the enzymatic turnover numbers, and (b) we synthesized racemic LThDP and reconstituted both YPDC and PDHc-E1 with this putative intermediate. Although slower than the turnover numbers found when starting with pyruvate and enzyme-bound ThDP, the results clearly indicated that both enzymes can indeed utilize the synthetic material as substrate and, with a knowledge of the structures now available, suggested reasons why kinetic competence may be difficult to demonstrate. The results presented are especially timely in view of an unusual and unexpected ThDP-derived structure reported recently on a related enzyme, acetolactate synthase (14), on the basis of which the mechanism as drawn in Scheme 1 was challenged. Our results suggested that there is no need to invoke alternatives to the accepted mechanism; nucleophilic addition at the carbonyl carbon atom leading to formation of the tetrahedral intermediate LThDP, which then undergoes decarboxylation, adequately accounts for the observations in a vast literature on the subject.

Instrumentation
Stopped-flow experiments were carried out on an Applied Photophysics SM.18MV instrument. UV-visible spectra and kinetics were carried out on a Varian (Cary 300) or PerkinElmer (Lambda 2S) double-beam spectrophotometer.

C2-␣-Lactyl-4,5-dimethylthiazole-t-butyl ester
Under N 2 protection, n-BuLi (2.5 M solution, 8.5 ml, 21.3 mmol) was added dropwise to a solution of 4,5-dimethylthiazole (2 g, 17.7 mmol) in 20 ml of dry THF at Ϫ78°C. The mixture was stirred for 30 min at Ϫ78°C and t-butyl pyruvate (3.1g, 21.5 mmol) was added rapidly. The mixture was stirred for another 30 min at Ϫ78°C. The acetone-dry ice bath was then removed, and the reaction mixture was allowed to warm to room temperature. Concentrated HCl (2 ml) was added to quench the reaction, followed by water (50 ml). The product was extracted with CH 2 Cl 2 (25 ml, 3ϫ). The combined organic layers were dried over Na 2 SO 4 . After evaporation of the solvent in vacuo, the crude product was chromatographed on silica gel (100 g) and eluted with ethyl acetate-hexane (1:4), yielding the pure product (1.8 g, 40%). 1

Kinetic Studies
Decarboxylation of Model Compounds-The reaction was monitored by stopped-flow by coupling the production of the enamine (decarboxylation) to its oxidation by 2,6-dichlorophenoindophenol (DCPIP, max is 600 nm). The compounds were stored as the conjugate acid form [COOH], then converted to the conjugate base for decarboxylation, k 1 , as shown, where AcThz and AcTh are the 2-acetyl derivatives resulting from oxidation of the enamine. Background was obtained by scanning water. In a typical stopped-flow experiment, one syringe contained LThz (final concentration 0.2 mM) and 30 mM DCl in CH 3 CN (contains 1% D 2 O). The other syringe contained DCPIP (final concentration 0.5 mM or as specified) and "proton sponge" (1,8-bis(dimethylamino)-naphthalene) at a final concentration of 100 mM in CH 3 CN (contains 1% Me 2 SO). The temperature was controlled at 32°C. To study the decarboxylation reaction with THF as solvent, the experimental conditions were the same except for use of THF as solvent and Et 3 N (5% v:v) as base. The decarboxylation of LTh in CH 3 CN and THF was also studied under the same conditions. Assay for PDHc-E1-catalyzed Decarboxylation of LThDP as Substrate-The reduction of DCPIP at 600 nm was monitored at 30°C as in equations 2 and 3 (16,17): Here, C2-␣-hydroxyethylideneThDP is the enamine in Scheme 1, and the enzyme is either YPDC or PDHc-E1. A solution (1-ml total volume) of LThDP (0.2 mM), MgCl 2 (2 mM), and DCPIP (0.1 mM) in potassium P i buffer (20 mM, pH 7.0) was placed in a 1-ml quartz cell. Potassium P i buffer served as the reference. The PDHc-E1 (at a final concentration 42.5 g/ml, or as specified) was added to initiate the reaction.
Overall PDHc Assay with LThDP as Substrate-The formation of NADH was monitored at 340 nm at 30°C. The E1 (2 mg) and E2-E3 subcomplex (5 mg) were pre-incubated in 0.2 ml Tris-HCl (0.1 M, pH 8.0) for 5 min for reconstitution. The PDHc (from reconstitution) was then added to a solution (1-ml total volume) of LThDP (0.2 mM), NAD ϩ (2.5 mM), Mg 2ϩ (1 mM), and dithiothreitol (2.6 mM) in Tris-HCl buffer (0.1 M, pH 8.0), and the reaction buffer was placed in a 1-ml quartz cell. The Tris-HCl buffer served as the reference. The reaction was initiated by the addition of coenzyme A to a final concentration of 0.1 mM.
Monitoring the Release of Pyruvate from PDHc-E1-bound LThDP-The depletion of NADH was monitored at 340 nm and 30°C, coupling the conversion of pyruvate to lactate via lactate dehydrogenase and NADH. A solution (1-ml total volume) of 0.2 mM NADH, 10 units/ml lactate dehydrogenase, 2 mM Mg 2ϩ , 200 M LThDP in 20 mM potassium P i buffer (pH 7.0) was placed in a 1-ml quartz cell with potassium P i buffer serving as the reference. The PDHc-E1 (final concentration 42.5 g/ml, or as specified) was added to initiate the reaction.
Pyruvate Release from LThDP Catalyzed by the E91D Apo-YPDC-A solution of apo-E91D YPDC (0.5 mg/ml), 5 mM LThDP, 100 mM potassium P i , 2 mM MgCl 2 (pH 6.1) was incubated at 25°C. At various time intervals, 20-l aliquots were removed, and the pyruvate released was measured using the NADH/lactate dehydrogenase coupled assay in the absence of pyruvate. The control contained every component except E91D apo-YPDC.
Acetaldehyde Formed from LThDP by the Apo-enzymes of YPDC-A solution of apo-enzyme (0.5 mg/ml), 5 mM LThDP, 100 mM potassium P i , 2 mM MgCl 2 (pH 6.1) was incubated at 25°C. At various time intervals, 20-l aliquots were removed, and the acetaldehyde produced was measured using the NADH/ADH-coupled assay in the absence of pyruvate. The control contained every component except the apo-enzyme.

Enamine Formation from LThDP by the Apo-E91D Variants of YPDC Monitored by Oxidative Carbanion Trapping Using DCPIP Reduction-
Apo-enzyme (0.5 mg/ml) was incubated with 2.5 mM LThDP containing 100 mM potassium P i , 2 mM MgCl 2 (pH 6.1) at 10°C, and the reaction was initiated by the addition of 15 M DCPIP. The wild-type YPDC was incubated with 2.5 mM LThDP solution as a control, because this form is purified fully active with tightly bound ThDP.

Synthesis of LThDP
The racemic LThDP was first synthesized by Kluger and Smyth (10) in 1981 by condensing ThDP with pyruvic acid t-butyl ester, followed by removal of the t-butyl group with TFA. They reported that the t-butyl ester of LThDP underwent decomposition during column chromatography, even on a reversed-phase HPLC column. Therefore, the final LThDP product they used for the enzymatic activity measurements contained a significant background of ThDP. Contamination of LThDP with ThDP interferes with the interaction of LThDP and the apo-enzyme, because the ThDP would compete for the active center. We developed methods to separate the t-butyl ester of LThDP from ThDP on a C18 reverse-phase HPLC column. The NMR spectrum of both, the t-butyl ester of LThDP and of LThDP itself, displayed no detectable ThDP.

Interactions of LThDP with Yeast Pyruvate Decarboxylase
Reconstitution of the Apo-E91D Variant of YPDC with LThDP-Previous research from our laboratory demonstrated that the E91D substitution of YPDC endows the enzyme with a very useful property, the ability to readily exchange its ThDP at the optimum pH of 6.0 (20). Experiments were designed to monitor the release of both acetaldehyde and pyruvate from LThDP, i.e. catalysis of both the forward and reverse reactions in Scheme 1. As shown in Fig. 1, the rate of reduction of acetaldehyde by NADH/ADH increased with time when the apo-E91D enzyme was incubated with LThDP. Because the rate is proportional to acetaldehyde concentration, this suggests that the E91D variant catalyzes the release of acetaldehyde from LThDP, and the concentration of acetaldehyde increases over time. The rate of reduction of pyruvate by NADH/ lactate dehydrogenase also increased with time when apo-E91D enzyme was incubated with LThDP ( Fig. 1), leading to similar conclusions as with acetaldehyde release.

Reconstitution of Other Apo-enzymes of YPDC with LThDP-
The production of acetaldehyde was next monitored from several active center variants with two substitutions, the E91D to assure that ThDP could be replaced (20), and a second active center substitution already known to have a dramatic effect on the steady-state rate constants (18,22). As seen in Fig. 1, although the substitution E477Q/E91D still allows formation of acetaldehyde, with the E51D/E91D and D28N/E91D double substitutions, essentially no rate can be seen for acetaldehyde formation. We conclude that the residues Glu-51 and Asp-28 have important roles somewhere between the binding of LThDP and acetaldehyde release (i.e. one of the steps depicted as k 3 , k 4 , and k 5 in Scheme 1), whereas Glu-477 has a more modest role in these steps. At the same time, evidence is also clear for release of pyruvate from LThDP ( Fig. 1, open circles).

Interactions of LThDP with PDHc or the PDHc-E1 Subunit
Based on the proposed mechanism, the fate of the LThDP after binding to the apo-enzyme could be either decarboxylation in the forward direction (equivalent to formation of enamine), or decomposition in the reverse direction, leading to pyruvate release. The enamine could be trapped by DCPIP (in an E1-specific assay) or oxidized by E2 and E3, eventually producing NADH in the overall PDHc assay. The release of pyruvate could be detected by coupling the reaction via the lactate dehydrogenase/NADH assay. The PDHc-E1 activities (expressed as turnover numbers) as deduced from reaction progress curves for the E1-specific assay, the PDHc overall assay, and for pyruvate release from LThDP, were 0.115, 0.185, and 0.075 s Ϫ1 , respectively (23), all very low and rather similar. Apparently, under the steady-state conditions employed in these experiments, the observed activity (the rate-limiting step) is the release of ThDP to regenerate the apo-enzyme.
A single turnover experiment was next carried out to study the rate of decarboxylation of LThDP by oxidative trapping of the putative enamine with DCPIP. The reaction progress curve of DCPIP reduction (Fig. 4) consists of two phases, the presteady-state phase (the first 5 s) and the steady-state phase (linear part between 10 and 50 s). The total change in absorbance during the initial burst is about 0.11 A units, corresponding to 7.1 M DCPIP reduction and 7.1 M of PDHc-E1 active center concentration. This value is in good agreement with the SCHEME 1. Mechanism of pyruvate decarboxylase, YPDC.

C2-␣-Lactylthiamin Diphosphate Reactivity in Enzymes, Models
enzyme concentration used in the assay, 0.75 mg/ml corresponding to 7.5 M concentration of active centers (molecular weight of PDHc-E1 dimer is 200,000 with two active centers). This result suggests that both active centers on the PDHc-E1 bind LThDP and catalyze the reaction. The data were fitted to a first-order decay followed by a steady-state rate. The rate constant of the first-order decay is 0.4 s Ϫ1 (for each active center, 0.8 s Ϫ1 /E1 dimer). This rate constant is an apparent rate constant encompassing three consecutive steps, binding of LThDP to apo-E1 (including a conformational change of the LThDP from the "S" to the "V" conformation), decarboxylation of LThDP, and oxidation of the enamine by DCPIP. Thus, it is a lower limit of the decarboxylation rate constant for LThDP on PDHc-E1. The actual rate constant of this step could be much faster.

Decarboxylation of Model Compounds in a Non-polar Environment
It had been reported in 1970 (11,12) that the decarboxylation of LThz is accelerated by transferring the reaction to a non-polar solvent. Previous studies on the decarboxylation of model compounds were performed in water and EtOH, and it was concluded that the rate in EtOH is 10 4 times faster than in water. But, all of the rate constants obtained were quite small (the largest first-order decarboxylation rate constant previously reported is 3.63 ϫ 10 Ϫ2 s Ϫ1 , Ref. 2) and far below the enzyme turnover number (for YPDC, the k cat value is ϳ60 s Ϫ1 /subunit).
We studied the decarboxylation of the LThz and LTh (Fig. 5)
In Equation 1, the first step is proton transfer, and it is presumably much faster than the remaining steps when a suitable base is chosen. The rate constant k 1 is the unimolecular decarboxylation rate constant, and k 2 [DCPIP] is the secondorder rate constant for oxidation. The reactions were carried out at 0.5 mM DCPIP and 0.2 mM LThz or LTh. In other words, had the reaction proceeded to completion, the DCPIP concentration would be no lower than 0.3 mM (0.5-0.2). Because k 2 is 1.  Table I for decarboxylation of LThz and LTh in CH 3 CN and in THF is ϳ50 s Ϫ1 , seven times smaller than the quantity k 2 [DCPIP], assuring first-order conditions and that the decarboxylation step is rate-limiting overall. Apparently, for LTh in THF, a decarboxylation rate constant comparable in magnitude to the enzymatic turnover numbers could be achieved. DISCUSSION We had two goals in carrying out the reported experiments, both related to the issue of whether or not the LThDP is an intermediate on ThDP-catalyzed decarboxylation pathways. The model studies have identified conditions under which the first-order rate constant for the decarboxylation step could be as large a as 53 s Ϫ1 , several orders of magnitude faster than those reported in the pioneering studies of Lienhard and coworkers (11,12) and Kluger (2). This was achieved by the detection of the enamine via its oxidation by DCPIP, a reaction that in the model studies has been shown to be very fast, hence the reduction of DCPIP indeed monitors the rate of decarboxylation. By selecting the appropriate base to initiate the reaction (to convert the carboxylic acid form of LThDP to the conjugate base, because the zwitterionic intermediate is the substrate undergoing decarboxylation) and solvents, we could push the rates to the stopped-flow range with solvents of low dielectric constant. The results affirm that the decarboxylation step per se requires little catalysis on the enzyme aside from a low effective dielectric constant, consistent with predictions from Lienhard and co-workers (11,12). Of course, the rate constant for the unimolecular step on the enzyme could indeed exceed the turnover number, i.e. the decarboxylation step need not be rate-limiting on the enzyme. In fact, Tittmann et al. (6) reported that the rate constant for the decarboxylation step could be considerably larger than the turnover number on pyruvate decarboxylases from both Zymomonas mobilis and Saccharomyces cerevisiae.
The results with the enzyme are more complex, but also quite informative. When starting with pyruvate, Tittmann et al. (6) did observe formation of some LThDP, yet in an early reconstitution experiment using LThDP and apo-pyruvate decarboxylase no activity could be detected (10). Armed with the E91D YPDC variant, which can readily exchange its ThDP (20), and with the E. coli PDHc-E1, which can also be generated as an apo-enzyme, we decided to reexamine the issue with the enzymes. The results with YPDC gave the first indication that LThDP was indeed a substrate for the enzyme, albeit at a considerably slower rate than when starting with ThDP and pyruvate. Almost certainly, under steady-state conditions this is because of the rate-limiting loss of ThDP from the enzyme. Nevertheless, the evidence was clear that LThDP was being partitioned on YPDC, reverting to pyruvate on the one hand, and undergoing decarboxylation in the forward direction.
With the PDHc-E1, under steady-state conditions both the reversal to pyruvate and decarboxylation were in evidence. In this case, we also undertook pre-steady-state experiments, in which we used DCPIP to monitor the rate of decarboxylation. The rate constant in this case is of the order of 0.4 s Ϫ1 , still two orders of magnitude slow compared with the turnover number. There are several possible reasons for this, because the rate constant measures a composite of three steps, binding of LThDP, decarboxylation, and oxidation. The value determined in this experiment is indeed a lower limit. In the E1-specific assay using DCPIP to trap the enamine, a PDHc-E1 activity of 0.385 units/mg, translating to a k cat of 1.28 s Ϫ1 /E1 dimer or 0.64 s Ϫ1 /active site was detected (obtained by varying the DCPIP concentration leading to this plateau value). Because the E1 activity is much higher according to the overall NADH assay, the rate-limiting step could be the oxidation of the enamine by DCPIP, and the rate constant 0.64 s Ϫ1 could be the upper limit for detection of enzyme-bound enamine with the DCPIP assay. (Parenthetically, our results point to this important limitation of the E1-specific assay relying on DCPIP when surveying variants of the E1 subunits, where the true k cat and K m values are often not reflected accurately.) The first-order rate constant of 0.4 s Ϫ1 reports on the slowest step of the three steps, of which the rate-limiting step is most likely the binding of LThDP to E1. The rate constant for binding of ThDP to the E1 is ϳ10 4 s Ϫ1 M Ϫ1 , ϳ10-times faster than of HEThDP, 3 so the binding of LThDP could indeed be 100 times slower than of ThDP. For an LThDP concentration of ϳ1 mm, the pseudofirst-order rate constant for binding could be as slow as 0.4 s Ϫ1 , the value determined by our experiments. In turn, the true rate constant for decarboxylation of LThDP on the PDHc-E1 could be much faster, as also found by Tittmann et al. (6) when reacting holo-pyruvate decarboxylases with pyruvate.
The results suggest that the early suggestion of Kluger and Smyth (10) of a "closed transition state" as an explanation of the inability to observe any activity with LThDP is better ascribed to a major conformational change that the LThDP must undergo once bound to the active center. Off the enzyme, LThDP most likely exists as an S conformer where the letter S describes the disposition of the two aromatic rings (thiazolium and 4Ј-aminopyrimidine) connected by a methylene bridge with respect to each other, when there is a substituent present at the thiazolium C-2 position (25). Off the enzyme, in such an S conformation the substituent at C-2 is on the side opposite to that of the 4Ј-amino group to minimize steric hindrance (26). As an analogue of LThDP, Sax and co-workers (27) solved the structure of phosphonolactyl thiamin, where the -COOH of the lactyl group is replaced by -P(OCH 3 )O 2 H (replacing the scissile C2-␣-COOH by an unreactive C2-␣-P bond), a structure that clearly displays the S conformation characteristics. In contrast, in the crystal structure of PLThDP complexed to PDHc-E1, not only is the V coenzyme conformation found, but the C-2 atom and the N-4Ј atom are brought to within 3.5 Å or less of each other. 2 That the substituent at the C-2 atom induces steric hindrance is evidenced by the finding that both PLThDP, a stable LThDP analogue, and HEThDP, exist as their 1Ј,4Јiminotautomeric forms (7,8). Hence, the bound conformation of LThDP must also be quite different from the conformation found in the free compound, and, the rate constant for this conformational change could be limiting the rate here detected. At the same time, as mentioned above, the rate of oxidation of the enamine by DCPIP on enzymes is not a true measure of the rates of decarboxylation, because DCPIP may have difficulty being recognized by (i.e. gain access to) the active center, and this barrier could vary from enzyme to enzyme.
The dramatic solvent effect on the rate of decarboxylation in the two models reported here is explained by concentration of charges in going from the zwitterionic LThDP analogue ground state to the enamine-like transition state, with a much reduced charge separation. This observation suggests, but does not prove, that a similar "environmentally induced" rate acceleration also exists on the ThDP enzymes. In our laboratory, we have searched for such an effect on both YPDC and PDHc-E1. We exploited the reaction depicted by the rate constants k 4 /k -4 in Scheme 1, with the hypothesis that should the enzyme push the equilibrium from HEThDP to the enamine, this would be consistent with such an enzymatic "solvent effect," because a lower "effective dielectric constant" of the enzyme should favor the enamine over the HEThDP from the electrostatic view. In model studies, we showed that the pK a of HEThDP at the C-2␣ position is ϳ15 in Me 2 SO-water mixtures and extrapolates to 18 in pure water (28), whereas PDHc-E1 can ionize it at pH 7 with a rate constant Ͼ1 s Ϫ1 accounting for a rate acceleration of the k -4 step of at least 10 million-fold by the enzyme. 3 We had also shown that the C2-␣-hydroxybenzyl analogue can be converted to the enamine on YPDC (29) by ionization at the C-2␣ position (pK a for this ionization in models is near 15; Ref. 30 and 31), and more recently we established this C-2␣ ionization of HEThDP on YPDC as well (32), with its even higher pK a quoted above. All of these results on the enzymes could be explained by an environmentally induced stabilization of the enamine intermediates. Our sole result regarding the effective dielectric constant on ThDP enzymes is derived from the fluorescence emission maximum of thiochrome diphosphate (a fluorescent ThDP analogue that is a competitive inhibitor of several ThDP enzymes) on YPDC, and its relationship to the same C2-␣-Lactylthiamin Diphosphate Reactivity in Enzymes, Models quantity measured in a series of 1-alkanols and water by Li and co-workers (29). The value recorded on YPDC fell between the values recorded in 1-pentanol and 1-hexanol interpolating to an effective dielectric constant near 13. Hence, in the enzyme reactions, once the LThDP is formed with the requisite V conformation, the barrier for the decarboxylation step can be significantly reduced by simply providing an apolar (non-aqueous) environment. On the basis of our results here reported it is appropriate to ask whether the notion of a tricyclic thiamin, in which the amino nitrogen adds to C-2 of the thiazolium moiety and where LThDP is not on the reaction pathway of these enzymes, as suggested by Pang et al. (14) for the enzyme acetolactate synthase, is relevant to chemistry taking place on the enzymes. We wish to raise several points not raised in that paper that tend to argue against their hypothesis. (a) Regarding the reactivity of the C2-carbanion/ylide/carbene (all simply resonance contributions to the resonance hybrid), its structure had been established by Arduengo and colleagues (33); perhaps the most striking feature of their finding is that the 13 C chemical shift of the thiazolium C-2 carbon is deshielded from 157 to 253 ppm on formation of the C2-carbanion/ylide/carbene. Subsequently, Brown (34) and Ikemoto 4 at Rutgers confirmed those results and showed that the species generated could lead to the expected products from either pyruvate or aldehydes. Chen et al. (35,36) at Rutgers also showed that whether the C-2 conjugate base reacts as a carbanion or as a carbene, nucleophilic addition accounts for the reactivity, and insertion can be ruled out.
(b) The model here reported for the decarboxylation step leads to first-order rate constants for LThDP analogues approaching the value of the turnover numbers of the relevant enzymes, and this holds even in the absence of the pyrimidine ring. We are aware of no similarly satisfactory models for decarboxylation catalyzed by tricyclic thiamin. (c) Tittmann et al. (6) identified LThDP on the pathway of pyruvate decarboxylase from Z. mobilis by acid quench of the reaction, whereas Joseph et al. 5 have also identified LThDP on YPDC variants using the same method. Although we do not question the results reported by Pang et al. (14), we believe that in view of the evidence reported here and discussed there is no need to invoke the alternative mechanism proposed by Pang et al. (14). It is also unlikely that with the same ThDP cofactor some enzymes would use the tricyclic thiamin mechanism, while YPDC and PDHc-E1 would proceed via the accepted mechanism in Scheme 1.
As a matter of fact, a case can be made that ThDP enzymes must avoid formation of a dead end tricyclic thiamin intermediate, because the same mechanism that enables the enzymes to activate the amino group by its conversion to the 1Ј,4Ј-imino ThDP (7,8,37) would also create the reactivity to readily carry out nucleophilic addition of the N-4Ј-imino nitrogen to the C-2 atom, were the C-2 atom within bonding distance of the N-4Ј atom. It is the enforced V conformer that ensures this separation of these two centers, and any relaxation in this distance, for example by reducing the size of the hydrophobic residue supporting this conformer (Ile-415 in YPDC), could lead to deleterious effects on reactivity (24). That ThDP enzymes, because of the high chemical reactivity of both ThDP and its C2-substituted reaction intermediates, must avoid multiple potential side reactions has also been well demonstrated by Hu and Kluger (9) on benzoylformate decarboxylase.