Direct evidence for carbanions and covalent N 5 -flavin-carbanion adducts as catalytic intermediates in the oxidation of nitroethane by D-amino acid oxidase.

Abstract We have found that d-amino acid oxidase is rapidly and irreversibly inhibited by cyanide during oxidative turnover of the carbanion of nitroethane. The inhibited enzyme has an absorption spectrum which is characteristic of a covalent flavin-substrate adduct. The following summarizes the results and interpretation of experiments which establish the mechanism of cyanide inhibition as well as the chemical mechanism of oxidation of nitroalkanes by d-amino acid oxidase. 1. The kinetic mechanism of oxidation of the nitroethane carbanion (S-) to form acetaldehyde (P), hydrogen peroxide, and nitrite, was deduced from a combination of stopped flow and O2-monitored kinetic measurements and is the following. [see PDF for equation] Evaluation of all the rate and equilibrium constants showed that the major pathway for flavin oxidation is the oxidation of ErP by O2. 2. The rate of inhibition of the enzyme by cyanide is regulated by κ2, and O2 is not consumed during the inhibition process. Since the rate of cyanide inhibition increases as the O2 concentration is raised, cyanide does not react with ErP or with a species in rapid equilibrium with ErP. Consequently cyanide must react rapidly with an intermediate, EX, which is formed from E0S in a reaction, or reactions, controlled by κ2. This is depicted kinetically as follows: [see PDF for equation] The inhibited enzyme, EI, does not react with substrate or O2 under the conditions, and within the time scale, of routine kinetic experiments. 3. The inhibited enzyme (EI) contains, per FAD, 1 eq each of substrate and cyanide, but is lacking the nitro group. 4. Treatment of EI with hot methanol produces a free flavin-substrate adduct in good yield and with no irreversible spectral or chemical modification. At 70° and pH values greater than 8 the free flavin-substrate adduct releases cyanide and is readily converted (l10 min) under aerobic conditions to FAD, acetaldehyde, and, presumably, H2O2. Anaerobically, FADH2 is produced. These solvent-catalyzed reactions of the adduct in large part mimic the enzyme-catalyzed oxidation of the substrate. 5. The spectral and ionization properties, as well as the chemical reactivity, of the free flavin-substrate adduct closely resemble those of 5-substituted dihydroflavins in general, and those of 5-cyanomethyl-1,5-dihydroflavin in particular. For these reasons we assign the structure 5-cyanoethyl-1,5-dihydro-FAD to the free flavin-substrate adduct. The kinetic turnover mechanism, the locus of action of cyanide, and the structure of the free flavin-substrate adduct, taken together, enable us to propose a detailed chemical mechanism for the oxidation of nitroethane carbanion by d-amino acid oxidase. E0S is a noncovalent complex in which the substrate carbanion is sufficiently close to the flavin (presumably the N5 position) to perturb the electronic properties of the latter. Attack of the carbanion at N5 of the flavin is controlled by κ2 and results in the formation of 5-nitroethyl-1, 5-dihydro-FAD. Elimination of nitrite forms a highly reactive cationic imine (EX) at the N5 flavin position to which solvent adds to [see PDF for equation] form a carbinolamine. Finally, FADH2, is eliminated from the carbinolamine, leaving acetaldehyde (P) noncovalently bound to Er. Cyanide attacks the cationic imine (EX) in competition with solvent to form 5-cyanoethyl-1,5-dihydro-FAD (EI). This adduct is not reactive with O2 under conditions normally used to study the enzyme.

Evaluation of all the rate and equilibrium constants showed that the major pathway for flavin oxidation is the oxidation of E,P by OZ.
2. The rate of inhibition of the enzyme by cyanide is regulated by k, and 0, is not consumed during the inhibition process.
Since the rate of cyanide inhibition increases as the O2 concentration is raised, cyanide does not react with E,P or with a species in rapid equilibrium with E,P. At 70" and pH values greater than 8 the free flavin-substrate adduct releases cyanide and is readily converted (<lo min) under aerobic conditions to FAD, acetaldehyde, and, presumably, HzOr. Anaerobically, FADH2 is produced. These solvent-catalyzedreactions of the adduct inlarge part mimic the enzymecatalyzed oxidation of the substrate. 5. The spectral and ionization properties, as well as the chemical reactivity, of the free flavin-substrate adduct closely resemble those of S-substituted dihydroflavins in general, and those of 5-cyanomethyl-1,5-dihydroflavin in particular. For these reasons we assign the structure S-cyanoethyl-l,Sdihydro-FAD to the free flavin-substrate adduct. The kinetic turnover mechanism, the locus of action of cyanide, and the structure of the free flavin-substrate adduct, taken together, enable us to propose a detailed chemical mechanism for the oxidation of nitroethane carbanion by Damino acid oxidase.
E&i is a noncovalent complex in which the substrate carbanion is sufficiently close to the flavin (presumably the N" position) to perturb the electronic properties of the latter. Attack of the carbanion at N" of the flavin is controlled by k, and results in the formation of 5-nitroethyl-1,5-dihydro-FAD.
Elimination of nitrite forms a highly reactive cationic imine (EX) at the Ns flavin position to which solvent adds to The molecular, as opposed to the kinetic, details of the oxidation-reduction processes catalyzed by flavoenzymes have remained elusive during the 40 years since the first flavoprotein, namely Old Yellow Enzyme, was discovered by Warburg and Christian (1). Given the three oxidation-reduction states of the flavin nucleus we must ask whether these states are interconverted through an orderly series of conventional chemical intermediates in which electron transfers are achieved by rearrangements of covalent adducts formed successively with the flavin nucleus by the reducing and oxidizing substrates, or whether the oxidation-reduction processes occur without the formation of covalent adducts. We address ourselves in this paper to the question of the chemical pathway by which enzyme-bound FAD is converted to FADH2 by nitroethane at the active site of D-amill acid oxidase.
If a covalent substrate-flavin adduct, whose kinetic properties qualify it as an intermediate lying directly on the catalytic pathway, could be trapped as a stable and isolable entity, the mechanism of Equation 3 would be ruled out. Furthermore, the structure of the trapped adduct would determine the fundamental chemical features of flavoenzyme catalysis and would serve to distinguish between Equations 1 and 2. Unfortunately, neither of the postulated adducts (see Equations 1 and 2), regardless of the position on the flavin nucleus at which nucleophilic addition of the physiological substrate occurs, is likely to have chemical properties which would allow trapping by a chemical agent in a process competitive with enzymatic turnover.
i\ priori, there are three chemical mechanisms by which enzyme-bound FAD might be converted to FADHz by physiological substrates of simple flavoprotein oxidasesl (Equations 1, 2, and 3).

Consequently,
we have taken a novel approach involving the use of preformed carbanions of nitroalkanes as analogs of the postulated enzyme-bound carbanions derived from physiological substrates.
Our intention, therefore, is to test specifically the validity of Equation 1 as a description of the chemical mechanism of flavoprotein oxidase catalysis. We were led to these experiments through the observation that n-amino acid oxidase is slowly inactivated during the oxidative turnover of nitromethane (6). The spectrum of the inactivated enzyme showed that chemical modification of the flavin nucleus had occurred. We reasoned that a 2nd molecule of the nitromethane anion (which had been slowly formed by ionization of the neutral substrate originally present) had probably interacted with a flavin-substrate adduct to form a species incapable of reacting with OZ. Consequently, we searched for an inhibitory anion of simple structure and found that cyanide was an extremely effective inhibitor of nitroalkane oxidation. Equation 4 outlines our strategy and emphasizes the close relationship between the mechanism of oxidation of nitroalkanes and that of physiological substrates (Equation 1) if the latter must be converted to enzyme-bound carbanions before covalent adduct formation with the flavin nucleus. 1 BY "simple flavoprotein oxidases" we mean those flavoenzymes which contain flavin as the only known oxidation-reduction coenzyme or cofactor and which catalyze the following over-all reaction (where X denotes 0 or NH). H,X-C-FADH in the normal course of catalysis by the addition I of H?X (which in our experiments is H20). The latter adduct is formally identical with that which would result from the interaction of the enzyme-bound carbanion of a physiological substrate with the flavin nucleus (see Equation 1).
We shall show that the scheme of Equation 4 accurately describes the chemical mechanism of reduction of the flavin nucleus of n-amino acid oxidase by the carbanion of nitroethane. We shall also show, by examination of the spectral and chemical properties of the free flavin adduct resolved from EZ, that flavin reduction results from rearrangements of covalent flavin-substrate adducts involving the N5 position of enzyme-bound FAD.
EXPERIMENTAL PROCEDURE Naterials n-Amino acid oxidase was purified from hog kidney by the method described by Yagi et al. (7,8). The benzoate-holoenzyme complex was stored at 5". The holoenzyme was separated from benzoate by dissolving approximately 40 mg of enzyme in 1 ml of 0.1 M sodium pyrophosphate buffer, pH 8.3, and adding a few crystals of nn-alanine. This solution was then passed through a Sephades G-25 column yielding benzoate-free holoenzyme. The absorbance ratio -4~~~4~~~ of the holoenzyme was 10.3. Assay of the enzyme under standard conditions (airequilibrated solution at 25" containing 10 pM FAD, 0.2 M DLalanine, and 0.1 M sodium pyrophosphate, pH 8.5) gave a specific activitv based on E;?~ = 23.0 (9) of 13.8 pmoles of product mine1 mg -1 while the turnover number, based on FAD content, was 11.3 s-1 (10).
Crystalline catalase and glucose oxidase (type IV) were purchased from Sigma Chemical Co. Alcohol dehydrogenase was obtained from the Boehringer Mannheim Corp. These three enzymes were used without further purification.
Nitroethane was purchased from the Aldrich Chemical Co. Before use, it was redistilled and if not used immediately was stored in the dark at 3". The anion of nitroethane was formed by equilibration of nitroethane (1.0 M) with an equivalent of KOH. The course of neutralization was followed by the pH change. The final pH of the reaction mixture was usually between 10 and 10.5. The anion was stored at 3" for no longer than 1 day. Prolonged storage resulted in the solution gradually turning yellow.
Tetranitromethane was purchased from the Aldrich Chemical Co. FAD was obtained from Sigma. Other chemicals were reagent grade purchased from commercial sources and used without further purification. KCN and NaCN were obtained from Fisher and yielded identical experimental results. Cyanide solutions were not stored at pH 8.3 for periods longer than 1 day. NanCN (1 mCi/O.9 mg) was obtained from New England Nuclear.

Meihods
Kinetic and Spectral ~l~easurements-L411 reactions were performed, unless otherwise stated, in 0.1 ~\r sodium pyrophosphate, pH 8.3, at 25". The solution was adjusted to the desired pH with I-ICI or KOH. All pH values were determined on a Radiometer 22 pH-meter with a GK-2322-C electrode.
The enzyme activity was monitored on the Yellow Springs oxygen electrode system with a solution initially air-equilibrated at 25". The 02 concentration in this solution was 0.24 m&I (11). The enzyme reaction was usually init,iated by the addition of nitroethane anion to the Orequilibrated solution containing the enzyme. This order of addition was chosen to eliminate changes in the state of ionization of nitroethane anion as much as possible. Nitroethane anion had little effect on the 02 electrode calibration.
Varying 02 concentrations were obtained by the use of gas flow meters manufactured by Kontes Glass Co. Either pure O2 or air was used in one valve, while N, was used in the other. The resulting mixture was bubbled through the assay solution for 3 to 5 min. The concentration of O2 in the assay mixture was measured relative to an air-equilibrated solution.
All spectra and optical kinetics were measured either on the Cary 15 or Gibson-Durrum Stopped Flow Spectrophotometer. Both were temperature-regulated to 25 + 0.2". These experiments were arranged such that less than 10% of the nitroethane anion was lost. due to protonation during the course of the experiment. The pH of the reaction mixture was measured at the end of each experiment.
Data from stopped flow experiments were plotted when applicable as first order rate constants over at least three half-lives. If the total absorbance change was less than 0.1, voltage changes were plotted directly. However, if the absorbance change was greater than 0.1, the voltage changes were first converted to absorbance. The anaerobic and oxidative half-reactions, as well as turnover, were monitored at several wave lengths.
Solutions were made anaerobic by slowly bubbling V2+-deoxygenated Nz through the solution for at least 5 min. This time was sufficient for complete anaerobiosis as judged by 02 depletion measured on the 02 electrode. For static experiments, Thunberg or glass-stoppered cuvettes were used.
In stopped flow experiments, solutions were made anaerobic with lo-' M glucose oxidase and 10v3 M n-glucose.
Resolution of Enzyme-The apoenzyme of n-amino acid oxidase was prepared by dialysis against 1 M KBr as described by Massey and Curti (12). However this method for resolution of holoenzyme was not successful when the enzyme was inactivated by cyanide and nitroethane. The flavin-substrate adduct was separated instead from the inactive holoenzyme in 0.1 M sodium pyrophosphate, pH 8.3, by the addition of boiling methanol to a final concentration of 80%. The protein was removed by centrifugation at 15,000 x g for 10 min at 5". The flavin-substrate adduct remained in the supernatant solution with a yield of 80 to 100%. The methanol was removed by flash evaporat,ion (without heating) and the resulting solid was dissolved in the desired buffer.
Unless noted, the modified FAD prepared in this manner was used for further experiments on the same day. Attempts at recovering active apoenzyme from the holoenzyme which had been inactivated by cyanide in the presence of nitroethane were unsuccessful.
Product Snalysis-Acetaldehydc was assayed for as dcscribcd by Bergmeyer (13). Hydrogen peroxide was determined by the influence of catalase on the initial velocity of the reaction or by the amount of OS consumed in the presence and absence of catalase. Nitrite was assayed by the method of Griess-Ilosvay as described by Egami and Taniguchi (14). Nitroethanc anion at the highest concentrations used resulted in less than a 1O'i; int.erference in any of these assay methods.

RKSULTS
Reaclion Xtoichiomefry-We recently established the reaction stoichiometry of the oxidation of nitromethane by n-amino acid osidase (6). In this paper we deal exclusively with the much more highly reactive nitroalkane anion. Consequently, we redetermined the reaction stoichiometry for nitroethane anion and found the following (Equation 5).
1.07 CH$ZHNOz-+ 1.0 02 + Hz0 -+ (5) 1.01 CH,CHO + 0.86 NOz + 1.07 Hz02 We conclude that 1 mole each of HzOz, N01, and acetaldehyde is produced for each nitroethane anion and OS consumed. This result is identical with that found for the oxidation of nitromethane by n-amino acid oxidasc (6) and simpler than the stoichiometry given by glucose oxidase (2). We note (recognizing that the nitroalkanes are of the same oxidation state as amines and alcohols) that nitroalkane oxidation is formally analogous to the oxidative deamination of an amino acid normally catalyzed by this ctlzyme when the latter reaction is written to include nonenzymatic hydrolysis of the imine (15, 16) released by the enzyme (Equations 6 and 7 which show that the apoenzyme is incapable of osidizing nitroethane anion with greater than 0.2% of the efficiency of the holoenzyme. These results also emphasize how much more reactive the substrate becomes when proton abstraction is carried out before interaction of the substrate with the enzyme. The true difference in reactivity of the neutral and carbanionic nitroethane species with the holoenzyme (which in the cast of glucose oxidase and nitromethane is a factor of lo6 in favor of the carhanion (2)), would require a systematic kinetic analysis over a wide range of pH.
No oxidation of nitroethane by free FAD in the dark has been detected, although photochcmical oxidation does occur (6).
Most of the experiments involving the prior formation of tlrcn nitroethane anion were carried out within a few minutes of platsing the anion in a solution at pH 8.3 and 25". We found, usitlg the tetranitromethane method (17), that the half-time for protonation of the carbanion under these particular conditions is 54 min.
Consequently, very little neutral nitroethane was l)rescllt in any of the experiments described here. Steady State Turnover Kinetics-When initial rates of Q2 COUP sumption were plotted in double reciprocal form with ltitroalkatle anion and O2 as variable substrates, parallel line pattertrs resultcld (see Fig. 1). The steady state rate equation describing these results is given by Equation 8.
The values of the steady state coefficients do-', +I-', and 42-l at 25" and p1-I 8.3 (0.1 M sodium pyrophosphatc) for nitroethanc carbanion are given in Table II.
The value of 41-l is 200 timcxs I greater than the corresponding value (see Table II) found with nitromethane at pH 8.3 (6). ,4s is generally the case in fla\-oprotein oxidase reactions (18,19), the steady state coefficients arc related in a simple fashion to the kinetic constants obtained 1,~ direct measurement in the stopped flow experiments to be d(lscribrtl.
They are very similar to those for nitroethane carbanion. We should emphasize that concentrations of nitroethane anion exceeding about 5 mM inactivate the enzyme. This requires special care in the estimation of initial velocities. The inhibition, noted previously (6), is most probably due to the interaction of a second nitroethane anion with the enzyme in a process entirely analogous to the effects of cyanide to be described.
The supposition that nitroalkanes function as simple substrate analogs of n-amino acids was strongly reinforced by the finding that benzoate, a classical competitive inhibitor of the n-amino acid substrates, acts as a competitive inhibitor of nitroethane anion.
The double reciprocal plots with nitroethane anion as variable substrate and at different levels of benzoate were accurately linear and gave a Ki value of 3.0 pM. This value is in excellent agreement with the value of the dissociation constant (3 PM) determined for the binary enzyme-benzoate complex (20).

Stopped Flow Anaerobic
Half-reaction Measurements-When enzyme and nitroethane anion were mixed anaerobically in the stopped flow apparatus the reduction of flavin, monitored at 463 nm, occurred in three phases.
The last part of the second phase, together with the third phase, is shown in Fig. 2. The first phase was detected by comparison of absorbance amplitudes with those expected from static spectra and was too rapid to be resolved (tIlz < 3 ms), while the second and third phases were easily resolved independently and from each other. The spectrum corresponding to the termination of each phase is shown in Fig. 3. The species formed very rapidly in the first unresolved phase (I$$) has a spectrum slightly less intense than that of EO at wave lengths less than 500 nm.
Above 500 nm, however, EoX differs from E. in having weak absorption which extends to 800 nm.
The flavin in E&3 is clearly in the oxidized state. The spectrum of the species (EJ') produced in the second phase is indicative of a reduced form of the enzyme, but is not identical with that of E,.
The third, very slow, phase does result in the formation of E,, as is shown in Fig. 3 by comparison with E, produced by n-alanine.
The interpretation of these spectral changes in the anaerobic half-reaction is straightforward. Firstly, the initial and rapid (unresolved) burst in absorbance at 550 nm, which is due to the formation of E& and which occurs within the deadtime of the apparatus, can be used to titrate E,$. These data are given in Fig. 4 2. Example of anaerobic reductive half-reaction monitored by transmittance increase at 463 nm in the stopped flow apparatus.
Visible on this relatively long time scale is the final portion of the second kinetic phase (fs = 0.18 s and easily resolved on faster sweep) corresponding to the conversion of E,, to E,P and the third, very slow, phase corresponding to the conversion of E,P to E, + P (see Equation  9 and discussion in text). The reaction was carried out in anaerobic 0.1 M sodium pyrophosphate, pH 8.3, at 25", by mixing enzyme (39.5 pi) with an equal volume of 50 mM nitroethane carbanion. E,P, spectrum obtained from the absorbance difference between the termination of the second and third kinetic phases of the anaerobic reductive half-reaction (see Fig. 2). E, (--), fully reduced enzyme obtained with n-alanine under anaerobic conditions. E, (---), fully reduced enzyme obtained after the third phase of the anaerobic reductive half-reaction with nitroethane carbanion (see Fig. 2). All spectra were measured, or calculated from measurements, in 0.1 M sodium pyrophosphate, pH 8.3, at 25", with 41 MN enzyme.
Although not shown, the spectrum of E&' extends to 800 nm. Since the formation of E&' from Eo is too rapid to be observed, the absorbance of E&' is obtained as the difference between the absorbance at the beginning of the reductive half-reaction (due to Ed) and the absorbance at the end of the reaction (which is zero, because E, does not absorb above 530 nm).
The reactions were carried out anaerobically in 0. FIG. 5. Double reciprocal plot of anaerobic reductive half-reaction data obtained from stopped flow measurements at 463 nm (see Fig. 2). In terms of Equations 9 and 10 and Scheme I, the data of the second kinetic phase (0-O) give kt = 6.2 s-1 (which can be seen more clearly in the inset) and K1 = 14 mM. The data of the third, slow, phase (X-X) give /cl = 0.01 s-l. The experimental conditions were identical with those of Fig. 2 ). I& (E-FA1)H.J was found in stopped flow experiments to be oxidized by 02 with a bimolecular rate constant of 2.5 X lo4 M-l s-l.
However, the rate of formation of E, (k, = 0.01 s-l) is 250 times smaller than the maximum turnover number and is much too slow for this species to be anobligatory catalytic intermediate.
In this and all other respects, Equation 7 is formally analogous to the kinetic mechanism required for many physiological substrates of both the D-and L-amino acid oxidases (19,21,22).
The analogy extends even to the fact that product dissociation from oxidized enzyme partially controls the maximum turnover velocity.
Stopped Flow Monitored Turnover E'xperiments-Typical stopped flow turnover traces monitored at 463 nm are shown in Fig. 6. There is an initial rapid absorbance decrease corresponcling to formation of E,.P during the first half of the first turnover, followed by slower absorbance decrease as turnover proceeds and the 02 is depleted.
A small part of the slow absorbance decrease during turnover is again due to enzyme inactivation by a 2nd molecule of nitroethane anion. The half-time of the initial rapid absorbance decrease becomes smaller as the concentrations of the substrate and of O2 are increased at fixed concentrations of 02 and substrate, respectively.
Such experiments also show that, the amplitude of the initial rapid phase increases as the substrate concentration is raised (see Fig. 6) and decreases as the O2 concentration is raised.
We conclude that E,P reacts with O2 to form a species EoP which absorbs at 463 nm. At this stage, therefore, our results from the half-reaction and turnover measurements are explained by the following scheme (Equation 10). The inhibition is complete and, for all practicsal purposes at 25" (see later) irreversible. Moreover, the inhibited enzyme does not oxidize I)-alanine.
The dependence of the rate of inhibition on the concentration of cyanide in the case of uitroethanc carbanion is given in Fig. 7. The non-zero ordinate intercept is due t,o the relatively slow inhibition of the enzyme by a second nitroctliane anion, as mentioned previously. These &monitored experiments could not be carried out at cyanide concentrations greater than 1 mM because of the rapidity of in-Ilibition.
We shall subsequently show that the rat.e of inhibition reaches a limiting value at higher cyanide concentrations. The rate of enzyme inhibition by cyanide is also a function of t.he concentration of nitroethaue anion and of OS, as shown in Fig.  8, A (Fig. 9). The first order absorbance decay caused by cyanide is saturable, as shown in Other experimental conditions given in legend to Fig. 6.
These results are highly significant for two reasons. Firstly (noting that the experimental conditions used in Figs. 7 and 10 were almost identical), the rates of inhibition from Fig. 7 accurately fit the initial portion of the curve obtained from the stopped flow turnover experiments of Fig. 10. We conclude, therefore, that the absorbance changes caused by CN-at 455 nm under turnover conditions (Figs. 9 and 10) directly monitor the inhibition process. Secondly, the maximum rate of the 455-nm absorbance change in Fig. 10, namely 0.83 0, is in excellent agreement with the value of 0.77 s-r which is obtained for the rate of the anaerobic half-reaction under similar conditions (that is, 2 mM nitroethane anion as shown in Fig. 5). This point is further verified by the experiments of Fig. 11, which show that at high cyanide and saturating nitroethane anion the maximum rate of the 455-nm absorbance change is about 5 s-l.
This value, is the lack of effect of cyanide on the turnover number of the enzyme as measured in the standard n-alanine assay (see "Methods").
when corrected to infinite cyanide by noting that cyanide is present at a concentration of 2.8 X Km'"-, agrees almost precisely with the value of 6.2 s-1 determined for kz in anaerobic reductive half-reactions (see Fig. 5). Theseexperiments, therefore, identify kinetically the locus of cyanide action as being within the sequence from E$5' to E,P and will prove to have important predictive value when we describe the properties of the flavinsubstrate which can be separated from the enzyme after cyanide treatment.
The kinetic experiments just described clearly suggest that inhibition by cyanide does not require the prior or simultaneous interaction of O2 with the enzyme even though the experimental design thus far utilized requires 02 to be present. This important conclusion was checked by a direct method in which the amount of O2 consumed during inhibition was measured as a function of the cyanide concentration. Fig. 12 shows that the total amount of O2 consumed during inhibition, when compared to the concentration of enzyme-bound FAD, approaches zero at high cyanide concentrations. Control experiments, also presented in Fig. 12, show that turnover with n-alanine is negligibly affected by cyanide.
These measurements completely substantiate the conclusion that cyanide interacts with a reduced, rather than oxidized, flavin species which is an obligatory catalytic intermediate during enzyme turnover with nitroethane anion. Furthermore, the enzyme r-sulting from treatment with 0.2 IV cyanide during nitroethane turnover in Fig. 12 had the spectrum characteristic of inhibited enzyme (see Fig. 14). The results of Fig. 12 are additionally useful because they establish, for the purpose of experiments to be discussed subsequently, that only about 0.06 of a complete turnover can occur in the presence of 0.2 M cyanide.
From the data now at hand we can examine the reaction scheme of Equation 10, which was deduced entirely from the results of stopped flow half-reaction and turnover experiments in which cyanide was not present, to see whether this scheme satisfactorily accommodates the information from the cyanide inhibition esperiments.
Since cyanide was proved to interact with a catalytic species lying between E',S and E,P, and since E,.P reacts with 02, the question as to whether both cyanide and 02 react with L&P can be tested experimentally. Fig. 8B clearly shows that O2 increases, rather than decreases, the rate of inactivation by cy- anide.
If cyanide reacted with E,P, competition between O2 aud cyanide would result, rather than the synergistic effect actually observed in Fig. 8B.% We must conclude, therefore, that cyanide in fact reacts with an intermediate EX which is formed in a first order process, or processes (governed by kz), from E&' and which is converted in turn by a first order process to E,P. This is depicted in Equation  11. The chemical structures and spectral properties of the intermediates of the scheme of Equation 11, as well as a more complete interpretation of the kinetic data, wTil1 be treated under "Discussion."  left ordinate refers to wave lengths above 280 nm, while the right ordinate refers to wave lengths less than 280 nm. The shift in X,,, from 332 to 320 nm observed when the flavin-substrate adduct is separated from EI is exactly reversed when El is reconstituted from adduct and apoenzyme.
All spectra were recorded at 25" in 0.1 M sodium pyrophosphate, pII 8.2.
the second phase of the anaerobic half-reaction was increased in the presence of cyanide.
This must result from the fact that E,.P (the formation of which from E. is responsible for the second phase of the half-reaction in Fig. 2) has greater absorbance than EI.
This absorbance difference is maximal at 400 nm (compare Figs. 3 and 14) and Fig. 13 shows a plot of the amplitude of the second phase as a function of cyanide.
The interaction of cyanide with the enzyme under these (anaerobic) conditions gives the same apparent Km value, namely 20 m&r, as is observed in the kinetic studies of enzyme inhibition (see Fig. 10). This K, value, in terms of Equation\ 11, represents the ratio lcz,P:kBI. Furthermore, the absorbance increment at 400 nm caused by 0.09 1~ cyanide (as well as the entire absorption spectrum) is esactly that predicted for the spectrum of EI shown in Fig. 14. This experiment confirms, therefore, that EI can indeed be formed under anaerobic conditions. Studies of Cyanide-inactivated Enzynze-The absorption spcctrum of EI, the inhibited species of the enzyme resulting from cyanide treatment during turnover, is shown in Fig. 14. The spectrum is dominated at pH 8.3 by a peak at 332 nm (eSS2 = 5.4 X lo3 M-l cm'? and is totally different from the spectrum of any enzyme species detected either statically or kineticall) with nitroethane anion in the absence of cyanide (provided that, appreciable inhibition by nitroethane has not occurred) or wit,11 physiological substrates in the presence or absence of cyanide. The spectrum of EI does, however, resemble the spectrum of the inhibited enzyme recovered after exposure to high concentrations of nitromethane (6). The derivative of FAD which is trapped by cyanide can be resolved in good yield (between 80 and 100yO) from the apoenzyme by precipitation of the protein with hot 80% methanol (see "Methods").
It is interesting to note that the conventional KBr procedure developed for the resolution of E-FAD (12) fails to resolve EI, suggesting, as will be subsequently confirmed, that the flavin in EI has a reduced, rather than oxidized, conformation.
After the methanol is removed by flash evaporation, the absorption spectrum of the free flavin adduct closely resembles that of the inactivated holoenzyme when measured in the same by guest on March 24, 2020 http://www.jbc.org/ Downloaded from solvent.
As Fig. 14 shows, the 332-nm peak of the inactivated holoenzyme is blue-shifted by only 12nm after the flavin adduct is separated from the apoenzyme.
The spectrum of EI (X,,, = 332 nm) could be recovered by combining the resolved flavin adduct with apoenzyme which had been prepared from native holoenzyme by the KBr procedure. This shows that no irreversible changes (covalent or otherwise) occur in the flavin adduct during resolution of the inhibited holoenzyme by hot methanol. We next used 14CN-in order to establish the amount of cyanide contained in the flavin-substrate adduct. This was accomplished by first separating the 14C-labeled inhibited enzyme from free "CN-on Sephadex G-25 as shown in Fig. 15. On the basis of c332 = 5.4 X lo3 M-l cm-l for EI (see Fig. 14), the ratio of 14CNto flavin in the pooled fractions containing EI (the first peak in Fig. 15) was 0.81.
Samples of 14CN--labeled holoenzyme which had been separated by gel filtration were then resolved by hot 80% methanol into flavin adduct and apoenzyme (which precipitated) and the supernatant solution containing the free flavin adduct was taken to dryness (without heating) by flash evaporation.
The residue was dissolved in 70% methanol and part of this solution was cycled twice more through the flash evaporation procedure.
The ratios of 14C to flavin in these examples, namely 0.92 and 0.90, established that the specific radioactivity of the flavin adduct was reasonably constant and that the free flavin adduct contains 1 eq of cyanide, Next, a sample of the flavin adduct in 0.1 M sodium pyrophosphate, pH 7.9, which had undergone three cycles of flash evaporation, was heated at 70" for 10 min.
This procedure, as we shall show, converts the free adduct to FAD and acetaldehyde.
Part of the heated solution was analyzed for the ratio of 14C:FAD, which was found to be 15. Determination of number of cyanide equivalents bound per flavin in EI.
The figure shows the separation of '4CN--labeled EZ ($first peak) from free 14CN-(second peak) on Sephadex G-25, using 0.1 M sodium pyrophosphate, pH 8.3. Absorbance (@---a) of EZ was measured at 330 nm while 14C (X---X) was measured as disintegrations per min per ml of each 2.3-ml fraction. EZ was formed by incubating (in a total volume of 3.0 ml of 0.1 M sodium pyrophosphate, pH 8.3) 0.140 mN D-amino acid oxidase with 10 mM ['*C]cganide (2.48 X 10' dpm per ml) and 2.0 mM nitroethane carbanion.
After inhibition was complete (as judged by the absorbance of 0.95 at 332 nm and the value Aa32 = 5.4 X lo3 M-' cm-1 for El), the solution was applied to the Sephadex G-25 column (25 X 1 cm) with the results shown in the figure.
Comparison of the absorbance of the solution obtained by pooling Tubes 9, 10, 11, and 12 (A332 = 0.26) with the radioactivity (9.58 X 10' dpm per ml) showed that EZ contains 0.81 cyanide eq per flavin (see also Table III). 4409 1.02. The increase in this ratio, compared to the values obtained before heating, is probably due to incomplete conversion to FAD (see "Discussion"). The remainder of the heated solution was made 80% in methanol and subjected to two cycles of flash evaporation (with heating). Control experiments showed that between 80 and 90% of free 14CN-would be lost from the solution of adduct during flash evaporation under these conditions.
The ratio of 14C:FAD in the solutions of adduct which had been heated and then subjected to flash evaporation was found to be 0.40 and 0.37, indicating that heating the flavin adduct at 70" for 10 min removes most of the cyanide.
No int'ellsive efforts were made to see whether this ratio approached zero after very long times of flash evaporation.
These experiments do suggest, however, that a small fraction of the adduct is converted in a reaction with a relatively low activation energy to a species from which cyanide can not be released on heating (see "Discussion").
Such a species would explain why the ratio &&?:A445 in the FAD spectrum (see Fig. 18) is about 20% t'oo large (23).
The results of the experiments involving 14CN-are given in Table III.
In summary, we conclude that both the inhibited enzyme (El) and the free flavin adduct contain 1 eq of cyanide and that the free adduct, when heated to 70" for 10 min at pH 7.9, produces FAD and free cyanide in good yield.
The number of substrate equivalents bonded to the flavin in the holoenzyme intermediate (EX,see Equation 11) with which cyanide rapidly reacts was determined by measuring the catalytic activity remaining at saturating cyanide as the enzyme was titrated with nitroethane anion. Essentially no turnover occurs under these conditions (see Fig. 12). These results are graphed in Fig. 16 and show that the formation of fully inhibited enzyme requires the addition of 1 eq of nitroethane anion per FAD. Therefore, EI must contain one substrate equivalent per flavin. It was next of interest to show whether or not EI contained the nitro group. This was carried out by inactivating the enzyme with saturating cyanide in the presence of nitroethanc anion and then analyzing for the presence of NO, after gel filtration on Sephadex G-25. Fig. 17 shows that each enzyme-bound FAD equivalent forms 1 eq of free NO1 before or during inactivation.
Again, comparison of the experimental condit.ions of Fig. 17 and those of Fig. 12, with particular emphasis on the saturating cyanide concentrations employed, rules out the possibility that the NO1 detected in Fig. 17 resulted from substrate turnover.
As an additional check, however, the consumption of O2 in the experiments of Fig. 17 was measured and found to be negligible.
Studies of Isolated Flavin Adduct-The isolated flavin adduct', which we have shown contains 1 eq each of substrate and cyanide, but is lacking nitrite, is readily oxidized to FAD by heating at 70" for 10 min at pH 8.3 in the presence of 02 (see Fig. 18). Free cyanide is released during this reaction, as was shown in Table  III.
The yield of FAD in this reaction is about 70%, whether the calculation is based on the amount of 14CN-released or on the value of es20 (namely 6.3 X lo3 M-I cm-l) for the free flavin adduct at pH 8.3. When the adduct is heated at 70" at pH 8.3 in the absence of 02, the characteristic spectrum of the adduct is replaced by a spectrum which is very similar to that of FADH2 (see Fig. 18). When O2 was admitted to the solution of the adduct which had been heated anaerobically at 70" FAD was formed extremely rapidly (see Fig. 181, as would be expected if FADHz had been formed.
The FAD formed in these reactions was identified not only by its spectral properties but also by its unique ability to activate the apoenzyme of D-amino acid oxidase to the same extent as authentic FAD.
The trace of FAD evident  showing that EI and the flavin-substrate adduct each contain 1 eq of cyanide per flavin and that most of this cyanide is released upon heating the adduct aerobically at pH 8.3 and 70" for 10 min.
The starting material for these experiments was t,he WN--labeled EI which was obtained as described in the legend of Fig. 15 .22 x 10 .97 x 10 .05 x 10 .38 x 10 .08 x 10 I a Obtained by comparing absorbance of EI at 332 nm b Obtained by comparing absorbance of adduct at 320 nm (c320 = 6.3 X lo3 M-I cm-l) with specific radioactivity.
c Obtained by comparing absorbance of the FAD at 450 nm (~4~0 = 11.3 X lo3 M-I cm-l) with specific radioactivity. These experiments were based on the fact that at high cyanide concentrations the enzyme is stoichiometrically inhibited by substrate because all but 5% of turnover is prevented (see Fig. 12). Enzyme (0.064 mzJ) was incubated with nitroethane carbanion (at concentrations giving the ratio of substrate to enzyme-bound FAD indicated on the abscissa) and 0.2 M cyanide in 0.1 M sodium pyrophosphate, pH 8.3, at 25". The percentage of activity remaining was determined by the standard assay with n-alanine (see "Methods"). The observed abscissa intercept of 1.1 is higher than the expected value of 1.0 because of the small amount of turnover which occurs under these conditions (see Fig. 12 Fig. 12). After the products were separated on Sephadex G-25 (25 X 1 cm column) as shown in the figure, 0.4% rmole of El and 0.46 pmole of nitrite were detected. After correction for the small amount of nitrite produced during turnover, this result shows that 0.98 eq of nitrite was released for each cquivalent of EI produced by cyanide treatment.
in the spectrum obtained by heating under anaerobic conditioirs (Fig. 18) does not arise from the leakage of O2 because the ratio of L14.15 after heating anaerobically, to 1 1 445 after heating anacrobically and admitting 02, was the same in tno experiments in which the adduct concentrations tliffercd by a factor of 2. We find that the aerobic conversion of the cyanide-containing flavin adduct to Fhl) is extremely sensitive to pII and tempcrature.
Less than lyO reaction occurred at 25" and pI1 8. The spectrum of FAD (---) was obtained after heatine the adduct at 70" for 10 min in an air-equilibrated solution. FLD was also obtained in similar yield by-admitting air to the solution of the adduct which had been heated anaerobically.
Left ordinate refers to wave lengths greater than 280 nm, while right ordinate refers to wave lengths less than 280 nm. All reactio& were carried out in 0.1 M sod&m pyrophosphate, pII 8.3, and snectra were recorded in this solution at 25". The concentration of adduct was 0.111 mM, while the concentration of FAD formed by heating aerobically was 0.082 mM.
10 min, while only about 4Oc/, conversion took place when the adtluct was heated to 70" for 10 min at pH 5.2.
Some of the adduct solution which had been heated at 70" in t,hc presence of 02 until the formation of FAD had ceased was examined for the presence of acetaldehyde by measuring the consumption of NASH in the alcohol dehydrogenase reaction at 111-I 6.0. When 0.09 mM adduct was heated aerobically, 0.034 m&j a&aldehyde and 0.068 mM FAD were detected.
Control exl)erinjcnts show-ed that the amount of free cyanide relrased from the adduct during heating would cause the observed yield of acetaldchyde to be 37.5% too low, owing to the reaction of cyanidc with ncetaldehyde.
When this correction factor was al?plied, the ratio of acetaldehyde to FAD detected alter heating the atltluct became 0.80.
No acetaldehyde was detected before tljt, adduct was heated.
Although measurements of HZOZ formation (which would be difficult) were not attempted, we can conclutlc with some confidence that, after cyanide release, the reactions undergone by the freshly prepared adduct at 70" give rise t,o exactly the same products as are observed in the rnzymatic turnover reaction.
We have noted, however, that after storage of the adduct at 5" for 8 days in the presence of O2 and at ~1-1 8.3, it is no longer possible to obtain a good yield of F.4D upon heating at 70".
WC have investigated the structure of the substrate-flavin adduct using, as a basis for comparison, the unique spectral and ionization properties of various well characterized flavin derivatives (25-28). Fig. 19 shows the absorption spectra of the adduct at pH values of 9.75 and 2.25. The shift in X,,,,, (320 + 330 nm) should be noted.
The large absorbance change at 260 nm is used in Fig. 19 as the basis for a spectrophotometric titratiojj, resulting in a pK, value of 6.4. When the flavin-substrate adduct was subjected to 6 N HCl the solution became purple, having absorption bands at 525, 390, 355, and 308 nm. This must represent oxidation of the adduct cation to give a mixture of products (including a free radical species which was detected I  I  I  I  I  by ESR measurements). Oxidation of the cationic adduct ap pears to be very facile, since attempts to obtain the spectrum of the cationic adduct itself in 6 N HCl; through which Nz was bubbled, were unsuccessful.
Addition of nitrite to the acid solution slowly generated the pale yellow flavoquinonium chloride (FAD HCI) which, upon neutralization with KOH, was deprotonatcd to form neutral flavoquinone (FAD).

n1scuss10s
Our experiments enable us to propose a detailed chcmicnl mechanism for a flavoprotein oxidase reaction. In order to facilitate the presentation and discussion of this mechanism, w(' shall first bring together the major facts which have resulted from our spectrokinetic and chemical studies of t.he oxidation of nitloethane ajjiojj by u-amino acid oxidase.
These facts arr as follows.
1. Iso equilibrates rapidly with S-(h'l = 14 mnl) to form E,$, the spectrum of which is similar to that of Eo except for weak absorbance at 1o11g w-ave lengths.
2. EoS is converted to E,P at a maximum rate of 6.2 s-l 3. The release of acetaldehgde (P) f rom E,P is not an obligatory process in turnover.
4. E,P reacts with O2 to form BJ' with a bimolecular rate constant of about lo4 M-l s-j.
5. Acetaldehyde (P) is released from BoP at a rate of 4.5 s-l to regenerate Eo.
6. The enzyme reacts rapidly with cyanide during turnover to give a spectrophotometrically distinct species, EI, which is ijjcapable of reaction with either S-or 02. The rate of formation of EZ is identical both with the rate of inactivation of the enzyme and with the rate of conversion of E$ to its immediate product. The inactivation of the enzyme by cganidc is therefore regulated by kz (6.2 s-j), representing the first order breakdown of EOS during catalysis. 7. Inactivatioii of the enzyme by cyanide is not an Or-rrquiring process.
Increasing concentrations of either S-or 02 increase the rate of inactivation by cyanide. Cyanide must react, thcrrfore, with a species EX which lies between E&' and EJ'. absorption spectrum at pH 8.3 which is characterized in particular by a maximum at 332 nm (6332 = 5.4 X lo3 m-r cm-l). 10. The species EI can be resolved in high yield into apoenzyme and a cyanide-containing covalent flavin-substrate adduct. Resolution is accompanied by a 12-nm blue shift at pH 8.3 in the absorption maximum of the adduct (~20 = 6.3 x lo3 n1-l cm-l).
11. The absorption maximum of the free flavin-substrate adduct undergoes the shift 320 + 330 nm as the solvent is changed from pH 9.75 to 2.52.
Protonation of the anionic flavin-substrate adduct (corresponding to the change in absorption maximum from 320 to 330 nm) occurs with a pK value of 6.4.
12. The free flavin-substrate adduct generates, in the presence of 02, the normal products of the enzyme-catalyzed reaction, namely, approximately 1 mole each of FAD and acetaldehyde (and presumably HzOJ, in a highly temperature-and pH-sensitive reaction.
13. In the absence of 02, the free flavin-substrate adduct generates FADH2.
14. In the presence of 6 N HCl and nitrite, the free flavinsubstrate adduct produces FADH+.
These facts are logically accommodated by the following chemical mechanism (Scheme 1). The basic framework of Scheme 1 results from the interpretation of the stopped flow spectrokinetic measurements. The position of EX is rigorously determined from the correlation of the effects of cyanide with the kinetic results.
However, as in all kinetic studies, only those intermediates whose breakdown is wholly or partially rate-limiting in the reaction sequence under study will influence the rate measurements.
Consequently, we have added certain intermediates, which, by well known chemical processes, serve to connect logically the other species for which direct spectrokinetic and chemical evidence exists. Our discussion of chemical mechanism is predicated throughout on the evidence obtained from studies of the isolated flavin-substrate adduct which, as we shall show, implicates covalent interaction of the substrate at the N" position of the flavin nucleus. E$ is visualized as a noncovalent complex (Ki = 14 mM, see Fig. 4) which is in rapid equilibrium with Eo and S-and in which the substrate carbanion is bound at the substrate site very close to the N-5 position of the flavin nucleus.
The spectral properties of E&3, particularly the long wave length aborption (see Fig. 3)) must reflect electronic interaction between the donor carbanion and the acceptor flavin.
We have measured by flow methods a spectrum3 very similar to that of E&3 when n-amino acid oxidasc interacts with fi-chloroalanine (29). The next intermediate for which kinetic and spectral evidence exists is E,P, which is formed from E&i at a limiting rate of 6.2 s-r (see Figs. 2 and 5). Because of the resemblance of the spectrum of E,P to that of E, (see Fig. 3) we tentatively identify EJ' as a noncovalent complex in which acetaldehyde resides close enough to the enzyme-bound FADHZ to slightly perturb the electronic properties of the latter. The simplest hypothesis accounting for the initiation of covalent catalysis, considering the results of this paper and the lack of any experimental evidence to the contrary, is that of nucleophilic attack of the substrate carbanion within E&3 on N-5 of the flavin nucleus, to give 5nitroethyl-1 ,5-dihydroflavin. Although the latter intermediate is not observed spectrophotometrically, its spectrum is expected to resemble that of an enzyme-bound 5substituted 1,5-dihydroflavin (such as EI shown in Fig. 14) and to be quite different from that of E&.
We assign kz (with a value of 6.2 s-i) as rep resenting the attack of the substrate carbanion at the N-5 flavin position.
This is an interesting result, since we have been unable to detect any nonphotochemical reaction of nitroalkanes (equilibrated at pH 8.3) with free FAD.
The enzyme, therefore, catalyzes the attack of the bound substrate carbanion at the N-5 position of the adjacent FAD, presumably by assisting the orientation of the carbanion for nucleophilic attack and perhaps by assisting electron redistribution of the flavin nucleus through 3 The int.eraction of p-chloroalanine with n-amino acid oxidasc in stopped flow anaerobic reactions monitored at 550 nm occurs in four phases (29,30). The first phase is rapid and is associated with a a-fold kinetic isotope effect when cY-deuterated p-chloroalanine is used (30).
The spectrum corresponding to the termination of the first phase (which is almost identical with the spectrum observed after 5 min, as shown in Fig. 1 of Reference 29) closely resembles the spectrum of E&3 obtained with nitroethane carbanion (see Fig. 3, this paper).
It seems likely, therefore, that the species formed in the first phase of the p-chloroalanine reaction is the enzyme-bound substrate a-carbanion. Evidence presented subsequently suggests in model systems (25,26). It is of interest, however, that that the N-l position of the enzyme-bound flavin-substrate ad-anaerobic half-reaction measurements of the interaction of gluduct is unusually basic. case oxidase with nitroethane anion revealed a spectrum cor-The chemical events following the formation of 5-nitroethyl-responding to a well known tautomer of the cationic iminr, 1,5-dihydroflavin can be deduced by considering the action of namely 5-ethyl flavoquinonium ion.* It appears very likely cyanide on the enzyme during turnover.
We note that nitrite is therefore that the mechanism of Scheme I, at least for the seeliminated either prior to, or simultaneously with, the attack of quence E0 ---f EX, applies also to the glucose oxidase reaction. cyanide to form EI (see Fig. 17) and that O2 reacts with E,P.
The reductive half-reaction in the n-amino acid oxidase mecha-The location of EX, the intermediate which is rapidly attacked nism is completed by the dissociation of acetaldehyde (P) from by cyanide, can be determined as follows. Firstly, EX is not E,P to give enzyme-bound 1 ,Bdihydroflavin which is indistin-EoS, because the rate of cyanide attack on EX is controlled by guishable from that produced by n-alanine (see Fig. 3). The kz which has already been assigned as the attack of the substrate latter is oxidized rapidly by 02 with a bimolecular constant k4 = carbanion on flavin within ES. Consequently, EX must lie 2.5 x lo* M+ s-l to regenerate E,,. However, the rate of disbeyond EoS. Secondly, no 02 is consumed during rapid inac-sociation of acetaldehyde from E,P (t,/z = 69 s, see Fig. 5) is tivation at high cyanide concentrations (see Fig. 12). Therefore competitive with the oxidation of E,P by OS only at 02 concen-EX can not be an intermediate derived from E,P after the reac-trations of the order of 1 PM. In routine turnover experiments, tion of the latter with 0%. The possible location of EX is now therefore, the dissociation of E,P to give E, is not an obligatory limited to the reaction sequence spanned by 5-nitroethyl-l , 5-di-process in the catalytic mechanism. hydroflavin and E,P, and including these species. E,P is elim-The deduction that EX lies directly on the catalytic pathinated as a candidate for EX by the important finding (Fig. 8B) way between E,$' and E,P is obviously of great importance that the interactions of 02 and cyanide with the enzyme in turn-and warrants further scrutiny. Given the results of the flow over do not result in competitive behavior between these ligands. measurements of the anaerobic half-reaction and the fact that Such behavior (i.e. a monotonic decrease in kobs when plotted 02 is not required for CN-inhibition, a possible alternative according to Fig. 8Bl  ET -= (13) V It has been our general experience that the composition of t,he steady state coefficients for flavoprotein oxidase reactions can be routinely identified by the kind of correlation of steady state and rapid flow measurements described here. It is not possible, of course, to deduce the composition of the coefficients from steady state measurement,s alone. The trapping and isolation of the covalent flaviil-substrate adduct, together with the precision with which the locus of action of cyanide has been determined, are the pirotal results of this paper.
Fortunately, there exists a large body of inforrnation concerning the physical properties and chemical reactivity of a wide variety of substituted flavins (24-28). This knowledge is in turn soundly based on structural determinations by s-ray crystallographic methods (32). We shall now summarize the evidence which points to the structure of carbinolamine. The fact that this feature is entirely missing in 5-cyanoethyl-l ~ 5-dihydro-FAD probably explains why this compound is relatively inert to oxidation by O2 at 25", both on and off the enzyme.
The marked pH dependence of adduct oxidation at 70" is probably explained by the ionization of N'. The anionic adduct would be expected to expel the negative cyanide ion much more effectively than the neutral adduct. The fact that the pK, value of the N' proton may be appreciably higher when the adduct is bound to the enzyme (which is suggested by the similarity in the X,,, values for EI and for the neutral adduct as shown in Figs. 14 and 19, respectively) would also tend to stabilize the enzyme-bound cyanoethyl adduct toward oxidation. Equation 14 accounts for at least 70% of the reaction undcrgone by the freshly prepared adduct at 70" and pH 8.3. There appear to be at least two other reactions taking place at pH 8.3. Firstly, traces of FAD arc formed in the absence of 02 (see Fig. 18). This could occur by elimination of the resonancestabilizetl carbanion of CH&H&N. Secondly, storage of the adduct for 8 days at 5" and pH 8.3 in the presence of 02 yields only about 40% of the theoretical amount of FAD. Heating at 70" had little effect on the spectrum which, apart from a peak at around 459 nm corresponding to FAD, showed a broad band iu the 330 to 359 nm region.
It is possible that, the latter absorption is due to the pseudo base derived from the addition OH-to C4" of the osidized adduct (25).
It has now been clearly established that flavoproteins as well as free flavins (35)(36)(37)  physiological substrates such as glucose, and whether the lifetimes of such carbanions are sufficiently short that proton removal and adduct formation are to be regarded as synchronous processes (see dashed arrows in Equation  14) are questions that, can only be answered by further experimentation.