N-Methylglutamate synthetase. Substrate-flavin hydrogen transfer reactions probed with deazaflavin mononucleotide.

N-Methylglutamate synthetase, reconstituted from apo-protein with 5-deazaFMN, catalyzes the reversible formation of N-methylglutamate via the same two-step mechanism previously elucidated for native enzyme (Reactions 1 and 2).(See article). This conclusion is based on the observation that: 1. Enzyme-bound deazaFMN (gamma-max equals 410, 338, epsilon410 equals 10,400 m-minus 1 cm-minus 1) is reduced by L-glutamate, N-methyl-L-glutamate but not D-glutamate. At saturating concentrations of L-glutamate Reaction 1 proceeds at 1% of the rate observed with FMN-reconstituted enzyme. 2. Substrate-reduced deazaFMN enzyme is reoxidized by methylamine or ammonia. 3. A glutaryl enzyme intermediate, isolated by Sephadex G-25 chromatography, contains radioactivity when prepared from [U-14C]glutamate, [alpha-3H]glutamate, or N-[glutaryl U-14C]methylglutamate; however, this intermediate is not labeled from N[methyl 14C]methylglutamate. 4. The amount of radioactivity incorporated into the intermediate is stoichiometric with the amount of deazaFMN reduced during its formation. 5. Intermediate prepared with [U-14C]glutamate yields alpha-[14C]ketoglutarate when denatured with acid and N-[glutaryl-U-14C]methylglutamate when incubated with methylamine. In the absence of methylamine deazaFMN enzyme intermediate slowly decays to yield alpha-hydroxyglutarate. 6. The rate of deazaFMN glutaryl enzyme intermediate formation at a fixed glutamate concentration is equal to the rate of the over-all reaction while the rate of intermediate reaction with methylamine is approximately 50 times greater than the over-all reaction. DeazaFMN enzyme intermediate prepared with [alpha-3H]-glutamate yields [3H]deazaFMNH2 when denatured with acid or phenol and N-[3H]methylglutamate when incubated with methylamine. These results show that the alpha-hydrogen of glutamate is transferred to deazaFMNH2, presumably at the 5 position, during Reaction 1 and that the same hydrogen is utilized for the reformation of the alpha C-H bond during Reaction 2. These results provide the first direct evidence for enzymic hydrogen transfer from substrate to flavin.

FMN in which the N-5 nitrogen of FMN is replaced by CH (4,5). Flavins and deazaflavins have been shown to exhibit many similar properties: (a) Both react with sulfite to form a covalent adduct at the 5 position (6,7); (b) both are chemically reduced by pyridine nucleotides (7) or by dithionite (5) and photochemically reduced in the presence of EDTA (5); (c) reduced flavins and reduced deazaflavins can be reoxidized by oxygen (5), cytochrome c (5), earbonyl compounds (8) and disulfides (7); (d) both compounds in the oxidized state form nonfluorescent complexes with tryptophan and &resorcylic ayid (7) ; and (e) theoretical calculations indicate that the 5 position is the most electrophilic position in flavins and deazaflavins (9). Except for photochemical reduction (5), deazaflavins are more slowly reduced than flavin and the reduced species is more slowly reoxidized as compared with flavin (5, 7).
The major difference in the oxidation-reduction properties of flavins and deazaflavins thus appears to be quantitative rather than qualitative.
Substitution of deazaFMN for FMN in the N-methylglutamate synthetase could thus afford a unique opportunity to directly study the involvement of flavin in enzymic hydrogen transfer. The results of this study are the subject of this paper. A preliminary communication describing these results has previously been published (10). EXPERIMENTAL PROCEDURE

Materials
Deazariboflavin was synthesized according to the method of O'Brien et al. (4). This compound was phosphorylated according to the procedure of Flexser and Farkas (11). DeazaFMN and commercial FMN (Sigma Chemical Co.) were purified in the dark on DEAE-cellulose as described by Massey and Swoboda (12). DeazaFMN was judged pure since it showed a single blue Auorescent spot in paper (Whatman No. 3MM) chromatography (5yo NaZHPOG.7Hz0 inwater; l-butyl alcohol-6 N HCl-water (80:2:20)) and thin layer chromatography (ethyl acetate-formic acidwater (7:Z:l)). The preparation of N-[glutaryl-U-l%]-and N-[me~hyZ-l*C]methylglutamate was performed as described by Pollock and Hersh (3 respectively. This ratio varies between 11.0 and 12.5 for native enzyme preparations.
Preparations of enzyme reconstituted with either FMN or deazaFMN yield values within the range 18.0 to 24.0, indicating 50 to 70% reconstitution.
The catalytic efficiency of deazaFMN-reconstituted enzyme as compared to FMN-reconstituted enzyme is shown in Fig. 3. That the deazaFMN enzyme is catalytically active is evidenced by the fact that 100 turnovers per mol of enzyme-bound deazaFMN have occurred at the 120-min time point. The FMN-reconstituted enzyme exhibits a specific activity equal to 69% of the native enzyme while the specific activity of the deazaFMN enzyme is 3.3% of the native enzyme.
Reduction and Reoxidation of DeazaFMN Enzyme by Substrates -Flavin reduction occurs during the reaction of native enzyme or FMN-reconstituted enzyme with amino acid substrates, according to Reaction 1. A similar reaction is observed with dea.zaFMN enzyme upon addition of L-glutamate or N-methyl-L-glutamate (Fig. 4), but not n-glutamate, indicating that in Reaction 1 deazaFMN enzyme exhibits the same specificity as native enzyme.
In order to rule out the possibility that the residual FMN enzyme (see Fig. 3) was acting as a catalyst for the reduction of deazaFMN enzyme, the following experiment was performed. The concentration of FMN enzyme in the dcazaFMN enzyme preparation was estimated from Fig. 3 (3)). FMN enzyme was added to this preparation such that the concentration of residual FMN enzyme was increased 2-fold in one experiment and 5-fold in a separate experiment. The half-times for reduction of deazaFMN by 25 mM glutamate at 10" were 3.9 min for the original deazaFMN preparation, and 4.0 and 3.8 min for the 2-fold and 5-fold increase in FMN enzyme, respectively. At 410 nm, the wavelength at which deazaFMN reduction is measured, no appreciable spectral changes are observed for reduction of FMN (3). Thus, if FMN enzyme was catalyzing the reduction of deazaFMN enzyme, the measured half-times would have had to decrease 2-fold and 5-fold respectively.
We, therefore, conclude that deazaFMN enzyme reacts directly with glutamate, and thus is, in itself, enzymatically active.
Substrate reduction of deazaFMN enzyme leads to the appearance of a new absorption band at 325 nm and an isosbestic point at 334 nm. This latter observation is consistent with the direct conversion of oxidized deazaFMN to a reduced species without the participation of any spectrally discernible intermediates. The spectrum of the reduced species is consistent with the formation of 1,8dihydrodeazaFMN since a similar hypsochromic shift of the 340 nm band of the oxidized chromophore also occurs in the case of free deazariboflavin during titration with dithionite and partially reduced samples of deazariboflavin show an absorption maximum at 320 nm (Fig. 5). A similar absorption maximum was not observed by Edmondson et al. (5) during the photoreduction of deazariboflavin in the presence of EDTA. This band may have been obscured in their experiments by the atypically high absorbance of the oxidized chromophore in this region of the spectrum.
In the presence of excess amino acid substrate the reduction of native enzyme (3), FMN-reconstituted enzyme, and deazaFMN-reconstituted enzyme show multiphssic kinetics. There is an initial rapid phase of reduction, amounting to approximately 75% of the total reaction, which follows pseudo-first order kinetics when plotted according to the procedure of Guggenheim (16). The slower phase of flavin reduction is presumed to be due to the presence of partially denatured enzyme since it shows the same isosbestic point as the initial fast phase and is too slow to account for the catalysis observed in turnover experiments. The formation of a single spectrally discernible reduced species is also suggested by experiments with native enzyme which show that the pseudo-first order rate constant for the rapid phase of reduction and the per cent of the total reaction occurring in the fast phase is independent of wavelength over the range of 300 to 530 run. The rates of flavin reduction were therefore calculated from the pseudo first order rate constants obtained during the initial rapid phase of flavin reduction. Comparison of the rates of glutamate reduction of enzymebound FMN and deasaFMN was obtained by measuring the rate of flavin reduction at several glutamate concentrations under conditions described under "Methods" and extrapolating the results to infinite glutamate. With FMN-reconstituted enzyme, plots of l/&b, versus l/[glutamate] for 62, 250, 625, and 1250 PM glutamate yielded a value of 20.8 mi0 at infinite glutamate. With deazaFMN enzyme the rate was too slow to measure at low glutamate concentrations. However, a constant value of 0.17 min-i was obtained at 2.5 and 25 mM glutamate, indicating saturation at the lower concentration. Thus reduction of deazaFMN-reconstituted enzyme proceeds at approximately 1 y0 the rate of FMN-reconstituted enzyme. Free reduced deazaFMN, released from substrate-reduced enzyme by acid denaturation, is slowly reoxidized (tllz = 3 hours), similar to results obtained for the reoxidation of 1,8dihydrodeazaflavins (5). Rapid reoxidation of enzyme-bound reduced deazaFMN occurs upon addition of either ammonia or methylamine (Fig. 6). Reoxidation in the presence of saturating glutamate with 22.2 mM ammonium sulfate is too fast to measure on the Gary spectrophotometer and at the same concentration of FIN. 6. Reoxidation of substrate-reduced deazaFMN enzyme by ammonia. Curve 1 is oxidized enzyme in the same buffer as described in Fig. 4. Curve 2 was recorded 37 min after the addition of 22.2 mM L-glutamate at 8", and Curve S was obtained after the addition of 22.2 rnM ammonium sulfate to the reduced enzyme, and represents the spectrum of the enzyme in the steady state. methylamine the reaction is complete within 1 min. These results show that deazaflavin reduction is slower than reoxidation since the tl/z for reduction in the presence of saturating glutamate is 4.0 min.
The addition of a 2-fold excess of ammonia with respect to glutamate causes nearly complete flavin reoxidation, as judged by the increase in absorption at 410 nm (Fig. 6). The extent of reoxidation is difficult to evaluate precisely since the spectrum obtained after reoxidation differs from the original spectrum of the oxidized enzyme, as evidenced most notably by the shift of the absorption maximum at 410 to 415 nm (Fig. 6). Experiments in which the glutamate concentration was held constant and the ammonia concentration varied to give partial flavin reoxidation indicate that the equilibrium strongly favors flavin reoxidation. This can be contrasted with the equilibrium constant of 1 obtained for Reaction 1 with native enzyme (3).
The addition of amine substrate alone to oxidized deazaFMN enzyme does not cause perturbation of the absorption spectrum. However, a perturbation virtually identical to that shown in Curve S (Fig. 6), as judged by the position of the new absorption maxima (415,340 nm) and minimum (370 nm), is observed with nonreducing amino acid substrate analogues, such as n-glutamate, a-ketoglutarate, or L-cY-hydroxyglutarate. This suggests that the spectrum obtained after reoxidation with amine substrate is due to an L-glutamate-oxidized enzyme complex. This complex would be expected to be the predominant enzyme species under the conditions of Fig. 6 since the steady state would favor an oxidized form of the enzyme and glutamate is present at saturating levels.  (Fig. 7). No radioactivity is incorporated when apoprotein is incubated with [ U-14C]glutamate. The amount of radioactivity incorporated after reaction with [U-*%1-or [a-aH]glutamate is approximately stoichiometric (70 to 90% yields) with the amount of deazaFMN reduced during intermediate formation, indicating that deaza-FMN is the sole oxidant of the amino acid substrate. Radioactivity is also incorporated into the enzyme when N-[glutaryl-U-14C]methylglutamate is used as the reductant, but virtually no radioactivity ehromatographs with enzyme when N-[methyE-"Clmethylglutamate is used (Fig. 7). These results show that the glutaryl group, but not the amino group, of the amino acid substrate is incorporated into the intermediate, indicating that cleavage of the LY carbon-nitrogen bond of the substrate occurs during intermediate formation, similar to native enzyme (3) and consistent with the release of the amino group as the first product of the over-all reaction.

Isolation of a DeazaFMN Glutaryl Enzyme
Intermediate prepared with [ U-i4C]glutamate yields cr-['"Clketoglutarate when denatured with acid. Reaction of this intermediate with methylamine results in the disappearance of LYketoglutarate and the formation of stoichiometric amounts of N-methylglutamate (Table I). These results show that the catalytically active glutaryl residue is released as a-ketoglutarate upon acid denaturation of the intermediate and that oxidation at the (Y carbon of the amino acid substrate occurs during intermediate formation similar to native enzyme (3).
Intermediate prepared from [&H]glutamate does not contain radioactive a-ketoglutarate, indicating that intramolecular shift formed was determined by the decrease graphed on a Sephadex G-25 column, as previously described (10). in absorbance at 410 nm (edesas~~~ -edeasa~~~nZ = 9400 ~-1 cm-l). Specific activities of the labeled substrates were: [U-Wlgluta-Since substrate does not fully reduce deazaFMN enzyme a value mate, 3.2 X 10' cpm/nmol; N-[gluta@-U-Wlmethylglutamate, for edseEs~~~nl could not be determined and was assumed to be 1.2 X 10" cpm/nmol; and N-[methyl-W]methylglutamate, 1.6 X 10% of the value determined for e&a&MN. If the intermediate is denatured by acid, 30% of the tritium in deaza-FMNHz is converted to *Hz0 during air reoxidation and the remaining 70% of the tritium chromatographs with deazaFMN during paper (Fig. 8)  on a Dowex AG-1 column, as previously described (10). The column was washed with 0.5 M formic acid (a-hydroxyglutarate and amino acids elute in this fraction) and deaaaFMN was then eluted with 10.0 M formic acid. Formic acid was removed by extraction with ether, the concentrated sample was applied in a l-inch strip to a paper chromatogram and the chromatogram was developed in 5y0 Nae-HP04.7H10 in water. DeaeaFMN was located by fluorescence (A). The chromatogram was then cut into l-cm strips, deazaFMN was eluted by shaking in 1.0 ml of water for 24 hours, and the radioactive content of each fraction determined by liquid scintillation counting (B).
acid and N-[aH]methylglutamate when reacted with methylamine (Table II)  and isolated by Sephadex G-25 chromatography as described for deazaFMN enzyme in the legend to Fig. 7. The amount of tritium incorporated into native and FMN-reconstituted enzymes corresponded to 0.9 and 0.7 mol of tritium/mol of FMNH, formed, respectively. The FMNHl formed was determined by the decrease in absorbance at 450 nm (crux -~ruxn~ = 10,300 M-I em-l, calculated as described for deazaFMN enzyme in Fig. 7). The isolated intermediates were incubated for 30 min at 30" in 100 mM potassium Tricine buffer, pH 8.3, 2 mM dithioerythritol, 5 mM sodium pyrophosphate, 0.5 mM magnesium disodium EDTA, 2.5 mM 2-mercaptoethanol and, where indicated, 200 mM methylamine. The intermediates were then denatured with 50/, trichloroacetic acid, amino acid formation was determined as described under "Methods," and 3Ht0 was determined by the radioactivity lost after repeated evaporation.  Table  III with the turnover number of the enzyme in the corresponding over-all reaction measured under the same conditions except that the substrate, absent in the half-reaction, is saturating and the observed values were corrected for the blank reaction seen with apoprotein.
The concentration of enzyme active sites used to calculate turnover numbers was determined from the absorbance of the deazaFMN enzyme preparation at 410 nm using the known molar extinction coefficient (10,400) at this wavelength.
Intermediate formation, under the same conditions except that methylamine was not present, was measured by monitoring the decrease in absorbance of the enzyme at 410 nm. The pseudo-first order rate constant for the reaction was obtained by plotting the data according to the procedure of Guggenheim (16). This method is equivalent to measuring the rate of formation of cr-ketoglutarate as an estimate of intermediate formation (Fig. 11). b The rate of the over-all reaction was determined as described above except the glutamate concentration was varied at a constant [Wlmethylamine concentration (100 mM). The rate of intermediate reaction was measured under the same conditions except unlabeled methylamine was used and glutamate was not present. Enzyme intermediate, prepared with [U-*"C]glutamate as described in the legend to Fig. 7, was incubated with methylamine for 10, 20, 40, 60, and 90 s, denatured with trichloroacetic acid, and the amount of N-(glutaryl-i4C]methylglutamate formed was determined by Dowex 50 column chromatography. Sufficient points for a pseudo-first order plot were not obtained since 87 and 100% of the reaction occurred at 10 and 20 s, respectively. Thus, the half-time of this reaction was estimated to be on the order of 3 s, assuming pseudo-first order kinetics.
in the over-all reaction. is nonexchangeable with solvent and less than 50% is lost during air oxidation, indicating that the a! substrate hydrogen is probably attached to C-5 in deasaFMNHz.
A similar hydrogen transfer appears likely with native enzyme since other aspects of catalysis are similar even though the relative reaction rates are different.
Indirect evidence for hydrogen transfer from substrate to flavin in enzymic reactions has been presented by Louie (25). Direct hydrogen transfer has also been observed by Shinkai and Bruice (8) during the reduction of the carbonyl group of pyridoxal by 1,5-dihydro-5-deaza-3, lo-dimethylisoalloxazine.
Although our data provide the first direct demonstration of enzymic hydrogen transfer from substrate to flavin, the mechanism of this hydrogen transfer is not presently understood. Theoretical calculations indicate that the 5 position of flavins and deazaflavins is the most electrophilic and, therefore, most likely to receive a transferred hydride ion (9). However, the observed hydrogen transfer reaction could proceed via a proton transfer mechanism since the evidence does not require a concerted transfer of a proton and 2 electrons to the 5 position of (deaza)FMN and recent studies have provided evidence for carbanion formation in the reactions catalyzed by the enzymes n-amino acid oxidase, L-amino acid oxidase, and lactate oxidase (26-31).
The reduction of certain carbonyl compounds by reduced flavins and deazaflavins has been reported (8, 32) and our observations suggest that the formation of ar-hydroxyglutarate observed with deazaFMN enzyme also occurs via a slow reduction of enzyme-bound cY-ketoglutarate (or an intermediate at the same oxidation level as Lu-ketoglutarate) by reduced deazaFMN.