Identification of the naturally occurring flavin of nitroalkane oxidase from fusarium oxysporum as a 5-nitrobutyl-FAD and conversion of the enzyme to the active FAD-containing form.

Nitroalkane oxidase from Fusarium oxysporum catalyzes the oxidation of nitroalkanes to aldehydes with production of nitrite and hydrogen peroxide. The UV-visible absorbance spectrum of the purified enzyme shows a single absorption peak at 336 nm with an extinction coefficient of 7.4 mM-1 cm-1. Upon denaturation of the enzyme at pH 7.0, a stoichiometric amount of FAD is released. The spectral properties of the enzyme as isolated are consistent with an N(5) adduct of the flavin. This is not due to a covalent linkage with the protein, since the free flavin adduct can be isolated from the enzyme at pH 2.1. The free flavin adduct shows an absorbance spectrum with a lambdamax at 346 nm (10.7 mM-1 cm-1) and is not fluorescent. Under alkaline conditions the free adduct decays, yielding FAD; the rate of this process is pH-dependent with a pKa of 7.4. Adduct decay is also observed with the native enzyme; in this case, however, the rate of decay is 160-fold slower (at pH 8.0) and not dependent on pH. During this process a large increase in enzymatic activity ( approximately 26-fold at pH 7.0) is observed, the rate of which is equal to the rate of flavin adduct conversion to FAD. Thus, the native flavin adduct is not active but can be converted to FAD, the active form of the flavin. Maximal activation is pH- and FAD-dependent; two groups with pKa values of 5.65 +/- 0. 25 and 8.75 +/- 0.05 must be unprotonated and protonated, respectively. The m/z- of the free flavin adduct is 103.0645 higher than that of FAD, as determined by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. This corresponds to a molecule of nitrobutane linked to FAD. A mechanism is proposed for the formation in vivo of the nitrobutyl-FAD of nitroalkane oxidase.

Flavoprotein oxidases comprise a large group of enzymes that catalyze the removal of a hydride equivalent from a substrate, transferring the electrons initially to the flavin cofactor and then to molecular oxygen to form hydrogen peroxide and the oxidized product. The most studied examples are the ␣-hydroxy acid oxidases, such as lactate oxidase (1), glycolate oxi-dase (2), and flavocytochrome b 2 (3) and the amino acid oxidases, of which D-amino acid oxidase is the best understood (4). There is strong evidence that the first step in the formation of the respective keto or imino acid products is removal of the substrate ␣-proton to form a carbanion. This conclusion is based upon halide elimination from ␤-substituted substrates (5), mechanism-based inactivation by propargyllic substrates (6), and kinetic isotope effects (7). The substrate carbanion is proposed to form an adduct at the N(5) position of flavin; such an adduct would then break down to form the reduced flavin and keto or imino acid product (8). Evidence for such an adduct comes from: (a) the trapping of a species with the properties of an N(5) adduct when glycolate is a substrate for lactate oxidase (9) and (b) the formation of adducts when nitromethane or nitroethane carbanion is used as substrate for D-amino acid oxidase (10). Flavoprotein oxidases that do not have ␣-hydroxy or amino acids as substrates appear to utilize different mechanisms for cleavage of the substrate carbon-hydrogen bond. Monoamine oxidase is proposed to utilize a radical mechanism for substrate oxidation, either abstracting an electron from the amine to initiate catalysis (11) or removing a hydrogen atom from the ␣-carbon (12). In addition, carbanion mechanisms have not generally been invoked for the alcohol oxidase group of oxidases, which includes glucose oxidase, cholesterol oxidase, and methanol oxidase (13); rather, both radical (14) and hydride transfer (15) pathways have been proposed. This variety of mechanism is remarkable in that all of these enzymes contain the same isoalloxazine ring as a cofactor, either at the FMN or FAD level. How the protein moiety modulates the reactivity of the flavin is not yet understood.
In 1978, Kido et al. (16) reported the initial characterization of a nitroalkane oxidase from the plant fungus Fusarium oxysporum. This enzyme catalyzes the oxidation of nitroalkanes by molecular oxygen to the respective aldehydes or ketones with production of nitrite and hydrogen peroxide (Scheme 1). Almost no mechanistic or structural studies of this enzyme have been reported so far. The neutral form of the nitroalkane is required for catalysis, in contrast to the situation with all other flavoproteins (17).
Nitroalkane oxidase has been reported to contain 0.32 mol of FAD/mol of monomer and to show a colorless UV-visible absorption spectrum with a shoulder at about 340 nm. This suggests that the flavin is in a reduced form (16,18). Furthermore, the addition of exogenous FAD is reported to be required for catalysis (16,17). These two properties, an apparently airstable reduced flavin and a requirement for a second molecule of FAD for catalysis, would make nitroalkane oxidase unique in its utilization of flavin cofactors. As a requirement for further structural and mechanistic studies, we set out to study the cofactor(s) of nitroalkane oxidase. In this paper we report the isolation and characterization of the native flavin adduct of nitroalkane oxidase.
Assays-The enzyme concentration was determined by the method of Bradford (20) using bovine serum albumin as the standard. The enzyme activity was determined using a variation of the coupled assay with aldehyde dehydrogenase developed by Heasley and Fitzpatrick (17), following the increase in absorbance at 340 nm. Final concentrations of 1 mM NAD, 0.02 units of aldehyde dehydrogenase, and 0.2 M Tris-Cl, pH 8.0, were used in a volume of 1 ml at 30°C. FAD and DTT were omitted from the assay mixture. Concentrations of NAD and aldehyde dehydrogenase were chosen such that neither was limiting in the overall reaction. One unit of enzymatic activity corresponds to the conversion of 1 nmol of substrate/min.
To monitor time-dependent changes in enzymatic activity, nitroalkane oxidase (5-14 M) was incubated in 25 mM potassium phosphate, 1 mM EDTA, pH 7.0, at 36°C unless otherwise stated. At different times aliquots were withdrawn and assayed for enzymatic activity. For experiments in which the effect of FAD, DTT, or 2-mercaptoethanol on the rate of activation was measured, the enzyme was incubated with the compound for 5 min on ice before starting the incubation. When the pH was varied, the following buffers were used at a final concentration of 0.1 M: MES, pH 5.4 -6.5; MOPS pH 6.4 -7.5; HEPES, pH 7.2-8.7; glycine, pH 8. 8 -9.8. For experiments involving nitroethane as a mixture of the anionic and neutral form (1:4, respectively), nitroethane (10%, v/v) was incubated in 0.2 M Tris-Cl, pH 8.0, at room temperature for 7 h.
UV-visible spectra were recorded using a Hewlett Packard model HP 8453 spectrophotometer. Fluorescence emission spectra were recorded with an SLM model 8000 spectrofluorometer thermostated at 25°C. Circular dichroism spectra were recorded with a Jasco J-600 spectropolarimeter thermostated at 25°C. For experiments involving anaerobiosis, the sample was placed in a 1-ml cell equipped with side arms and made anaerobic by several cycles of evacuating and flushing with purified nitrogen obtained by passage through a column of BASF catalyst heated at 120°C. Reverse phase HPLC of the flavin extracted from nitroalkane oxidase was performed according to Light et al. (21), using a Waters instrument equipped with a Lambda Max model 481 detector set either at 346 or 450 nm and a -Bondapack C-18 (0.39 ϫ 150-mm) column (Waters). Reverse phase ion pair chromatography of the free flavin adduct was performed as follows: eluent A, 0.3% tetrabutylammonium bromide, 3% methanol in 25 mM ammonium acetate, pH 4.8; eluent B, 100% methanol; flow rate, 1.5 ml/min. Elution was as follows: A from 0 to 5 min and then from 0 to 30% B in 24 min.
Isolation and Characterization of Flavin in Nitroalkane Oxidase-The free flavin adduct was prepared using the following method: 40 l of 6 N HCl were added to 1 ml of nitroalkane oxidase (12-28 M) in 25 mM potassium phosphate, 1 mM EDTA, pH 7.0; the resulting pH was 2.1-2.3. After 15 min of incubation on ice, the sample was loaded onto a PD-10 column (Pharmacia Biotech Inc.) equilibrated with 25 mM potassium phosphate, pH 2.1. Fractions were collected and checked by UV-visible spectrometry for flavin content; those containing free flavin were pooled. For determination of the activation energy of the decay of the free flavin adduct, 0.12 ml of 1 M Tris base were added to 1 ml of free flavin adduct in 25 mM potassium phosphate, pH 2.1; the resulting pH was 7.1. The rate of breakdown was determined by following the increase of absorbance at 450 nm. When the rate of adduct breakdown was studied as a function of pH, the following buffers were used at a final concentrations of 0.1 M: MES, pH 5.4 -6.5; MOPS, pH 6.4 -7.5; HEPES, pH 7.2-8.7; glycine, pH 8.8 -9.8.
For FAB and MALDI-TOF mass spectrometry analyses of the free flavin adduct, the sample was prepared with the following procedure: nitroalkane oxidase (10 -30 mg) was pelleted at 70% ammonium sulfate saturation; the pellet was resuspended in 1-2 ml of 10 mM ammonium acetate and dialyzed at 4°C against three 500-ml changes of the same solution and one change of 500 ml of 1 mM ammonium acetate. Acetic acid was added to the enzyme to give a pH of 2.1-2.3. After 15 min of incubation on ice, the sample was loaded onto a PD-10 column (Pharmacia) equilibrated with 1 mM ammonium acetate, pH 2.1. The fractions containing the free flavin (total volume, 5-8 ml) were pooled and, for FAB analysis, concentrated to a final volume of 6 -20 l using a Savant DNA SpeedVac model DNA100. Samples were then analyzed in the Laboratory for Biological Mass Spectrometry (Department of Chemistry, Texas A & M University). FAB mass spectrometry experiments were performed on a VG Analytical 70S instrument. Thioglycerol was used as the matrix. To improve the ionization yield, NaI was added to the sample just before the analysis. MALDI-TOF mass spectra were acquired using a Voyager Elite XL mass spectrometer (PerSeptive Biosystems, Framingham, MA) equipped with delay extraction. MALDI mass spectra of both the commercially available FAD and the 5-nitrobutyl-FAD extracted from nitroalkane oxidase were acquired in positive and negative ion modes using ␣-cyano-4-hydroxycinnamic acid as the matrix; however, a signal corresponding to the intact 5-nitrobutyl-FAD was only observed in the negative ion mode. The samples were prepared for MALDI using the overlayer method of sample preparation previously described (22). Approximately 1 pmol of the sample was applied to the sample plate for the MALDI analysis. For the accurate mass measurements, 10 spectra were acquired in the delayed extraction reflection mode. Each spectrum consists of an average of 50 individual laser shots. The spectra were acquired with an acceleration voltage of Ϫ20 kV, a pulse voltage of Ϫ5 kV, and a delay time of 250 ns. Des-Arg-bradykinin (M r 903.4603) and the matrix dimer (2M Ϫ H) Ϫ ion were used as internal standards for mass calibration.
Data Analysis-Data were fit to Equations 1-3, using the Kaleida-Graph software from Adelbeck Software (Reading, PA). Equation 1 describes a single first-order reaction and was used for the time-dependent change in enzymatic activity. Equation 2 describes a bellshaped curve and was used for the specific activity of activated nitroalkane oxidase as a function of pH. Equation 3 describes data that decreased with unit slope at low pH values and was used for the rate of free adduct decay yielding FAD as a function of pH.

RESULTS
Spectral Properties of Nitroalkane Oxidase-The near UVvisible absorbance spectrum of freshly purified nitroalkane oxidase shows a single peak centered at 336 nm ( Fig. 1). When the enzyme was heated at 100°C for 30 min, the absorbance spectrum changed to that of an oxidized flavin (Fig. 1). Similar results were obtained by incubating the enzyme for 30 min in 7 M urea or in 80% methanol at 75°C. The released cofactor was identified as FAD by HPLC; based upon the ⑀ 450 value for FAD of 11.3 mM Ϫ1 cm Ϫ1 (23), an extinction coefficient of 7.4 mM Ϫ1 cm Ϫ1 was determined at 336 nm for the enzyme-bound flavin. A stoichiometry of 0.98 FAD/monomer could be calculated from these results. The absorbance spectrum was unaffected by the addition of nitropropane in the presence or absence of oxygen. The visible fluorescence emission spectrum of purified nitroalkane oxidase is unusual in that it shows a maximum at 475 nm, in contrast to free FAD, which has a maximum centered at 535 nm ( Fig. 1, inset). The intensity of emission was about 10% that of free FAD. Upon the addition of 1-nitropropane, the intensity decreased by about 30%; thus, the flavin microenvironment is perturbed by the presence of the substrate. The circular dichroism spectrum of purified nitroalkane oxidase showed a single negative band at 346 nm and no optical activity above 420 nm (Fig. 2). This spectrum is distinct from previously reported circular dichroic spectra of both oxidized and reduced flavoproteins (24,25).
To determine more directly if the enzyme contained reduced FAD, it was denatured at pH 7.0 with 7 M urea anaerobically. This resulted in the slow appearance of the oxidized flavin spectrum. No substantial changes in the absorbance spectrum were observed upon incubating purified nitroalkane oxidase (8 -11 M) at 15°C for 90 min with 1 mM methylmethanethiosulfonic acid either in the presence or absence of 0.2 mM 1-nitropropane. Sulfite or 5,5Ј-dithiobis-(2-nitrobenzoic acid) similarly had no effect on the absorbance spectrum.
Time-dependent Activation of Nitroalkane Oxidase-Nitroalkane oxidase as isolated has a specific activity of about 10 units/nmol at 30°C and pH 8.0 using 1-nitropropane as substrate when assayed in the absence of exogenous FAD. When the native enzyme (14 M) was incubated at 36°C in 25 mM potassium phosphate, pH 7.0, a large increase in activity was observed over several hours (Fig. 3). The rate of this process was equal to the rate of appearance of the typical oxidized flavin absorbance spectrum (data not shown). When the incubation was performed in the presence of 0.35 mM FAD, a higher specific activity at the end of the process was observed. At pH 7.0, the specific activity increased about 26-fold when FAD was present during the incubation. No subsequent change in specific activity was observed upon further incubation (52 h). The flavin present after activation was identified as FAD by HPLC. Thus, the native flavin species is not active but can be converted to FAD, yielding active enzyme. The rate of conversion into FAD was not dependent on the presence of added FAD or the pH of incubation (k obs ϭ 0.0013 min Ϫ1 , at 36°C). The same results were obtained when the incubation was carried out in the presence of 5 mM DTT or 2-mercaptoethanol.
The extent of the increase in activity was both pH-and FAD-dependent (Fig. 4). Two groups with pK a values of 5.4 -5.9 and 8.7-8.8 must be unprotonated and protonated, respectively, for maximum activity. This may be due either to an enzyme conformational change or to ionization of groups di-rectly involved in FAD binding. When the extent of activation at pH 7.1 was determined in the presence of different concentrations of FAD up to 0.3 mM, half-maximal activation occurred at 10.4 Ϯ 1.5 M (data not shown). A possible explanation for the FAD dependence was the formation of apoenzyme during the incubation. This model is supported by the observation that the specific activity of a sample incubated in the absence of FAD (ϳ140 units/nmol after 16 h at pH 7.0) increased when an aliquot was withdrawn and incubated for 2 min in the presence of 0.22 mM FAD (ϳ210 units/nmol).
Properties of Nitroalkane Oxidase Containing FAD-Activated FAD-containing nitroalkane oxidase was isolated by gel filtration after different times of incubation (16 -24 h) in the presence and absence of FAD. In these experiments, the concentration of enzyme after gel filtration was typically 4.9 -10.6 M in FAD content, and the specific activity was 81-126 units/ nmol. In all cases, the visible absorbance spectrum of the en- zyme showed the presence of bound FAD (Fig. 5). The absorption maximum in the visible region (449 nm) is well resolved with a shoulder around 475 nm, suggesting a hydrophobic microenvironment at the N(3) position of the flavin (26). Assuming that the extinction coefficient for bound FAD is similar to that of free FAD, as is the case for the majority of flavoproteins, the stoichiometry of FAD/monomer can be calculated. A value of 0.3-0.4 was estimated when the enzyme concentration was 5-7 M. In all cases, the total flavin content was 0.5-0.7 mol/mol monomer, ruling out a requirement of a second flavin for catalysis. The flavin fluorescence emission of nitroalkane oxidase containing FAD was maximal at 535 nm ( ex either 417 or 450 nm) (Fig. 5, inset). The intensity of the emission was about 25% that of free FAD. The circular dichroism spectrum of nitroalkane oxidase containing FAD is quite different from that of the enzyme as isolated, suggesting a different pattern of optically active centers in the two flavins (Fig. 2).
The FAD in the enzyme could be reduced under anaerobiosis by 1 mM nitroethane, as seen by the bleaching of the flavin peak in the visible region (Fig. 6), in contrast to the result with enzyme as isolated. Upon the admission of oxygen, the spectrum returned to that of the oxidized enzyme. FAD-containing nitroalkane oxidase formed a flavin-sulfite adduct; in the presence of 0.56 M sulfite, the bleaching of the visible peak was monophasic, with a rate of 0.093 min Ϫ1 at pH 7.0, 16°C (data not shown). The addition of nitroethane in the presence of oxygen resulted in enzymic turnover. As shown in Fig. 7, the absorbance maximum at 449 nm was bleached (70 -80%) upon the addition of nitroethane at pH 7.0; after removing the substrate by gel filtration, the spectrum was that of enzyme containing oxidized FAD.
Isolation and Characterization of the Native Flavin-In contrast to the results at pH 7, incubating nitroalkane oxidase for 15 min at pH 2.1 followed by gel filtration to remove the apoprotein prevented conversion of the flavin to FAD. The absorbance peak of the flavin isolated in this fashion was centered at 346 nm (Fig. 1), suggesting that the free flavin was still in the form of an adduct. Thus, the modified flavin spectrum of the enzyme is not due to a covalent interaction with the protein. The flavin isolated at pH 2.1 was not fluorescent. The circular dichroism spectrum of the flavin extracted at low pH showed a single band with positive polarity at 325 nm (Fig. 2); both the change in polarity and the red shift of the circular dichroic signal with respect to that of the native enzyme suggest a different conformation of the flavin bound to the enzyme and free in solution. The isolated flavin was stable in solution for a few days at 4°C and pH 2.1; when the pH was raised to 7.0 or above, a rapid and full reappearance of the absorbance spectrum of oxidized flavin was observed. A concomitant increase of fluorescence emission at 525 nm ( ex 450 nm) was also observed, with a rate equal to that of the absorbance changes.
The resulting oxidized flavin was identified as FAD by HPLC (21), positive ion FAB and MALDI-TOF mass spectrometry, and ion pair reverse phase HPLC. Reverse phase ion pair HPLC showed that more than 96% of the extracted flavin is in the form of the adduct (retention time (R t ) ϭ 2.52 min), whereas the rest is oxidized FAD (R t ϭ 18.39 min). After incubation of the free flavin at pH 7.1, a single species eluting at 18.38 min was obtained. Based upon the known ⑀ 450 value for FAD (23), an extinction coefficient of 10.7 mM Ϫ1 cm Ϫ1 was calculated at 346 nm for the flavin extracted at pH 2.1.
The rate of decay of the isolated adduct to FAD was pH-dependent, with a maximum value above pH 8.0 of 0.2 min Ϫ1 at 36°C (results not shown). A pK a value of 7.4 Ϯ 0.2 was determined for this process. A value of 0.25 kcal/mol has been determined for the activation energy for breakdown of the free flavin adduct, by incubating the free flavin at pH 7.1 at different temperatures ranging from 16 to 40°C (results not shown).
Structural Studies on the Isolated Flavin-The structure of the isolated flavin was studied by MALDI-TOF and positive ion FAB mass spectrometry. Careful handling was required during the isolation of the flavin, since under acidic conditions and high temperatures FAD is known to hydrolyze, giving a series of products (27). Using negative ion MALDI-TOF mass spectrometry, two peaks with m/z Ϫ values of 784.1415 Ϯ 0.0068 and 887.2138 Ϯ 0.0078 were observed (Fig. 8). When commercially available FAD was used as a standard, a single peak with an m/z Ϫ value of 784.1500 Ϯ 0.0078 was obtained. Thus, the peak at m/z Ϫ 784 can be identified as FAD formed during the laser desorption ionization process. The difference of 103.0645 Ϯ 0.0103 between the two species is consistent only with the addition of C 4 H 9 O 2 N to FAD. The most reasonable structure that accounts for this is nitrobutane bound to reduced flavin (Scheme 2). Indeed, the calculated molecular mass of this molecule is 887.2126; the m/z Ϫ difference (887.2138 Ϫ 887.2126) of 0.0012 mass units corresponds to a mass measurement accuracy of 1.3 ppm. 2 A single peak with an m/z ϩ value of 889 was observed when the free flavin adduct was analyzed by positive ion FAB mass spectrometry (data not shown). Three peaks with m/z ϩ values of 786, 808, and 830, corresponding to the non-, mono-, and di-sodium molecules, respectively, were obtained when FAD was used as a control.
In Vitro Formation of Nitrobutyl-FAD from FAD-containing Nitroalkane Oxidase-A nitrobutyl moiety bound to the FAD in nitroalkane oxidase could be formed by the addition of two molecules of nitroethane to the flavin with loss of nitrite. However, no adduct is obtained upon incubating the FAD-containing enzyme with the neutral form of nitroethane (Fig. 7). To determine if the anion of nitroethane is required, FAD-containing nitroalkane oxidase was incubated at pH 8.0 with nitroethane containing about 25% nitroethane anion. As shown in Fig.  9, a complete bleaching of the 449-nm peak was observed; after gel filtration the spectrum was identical to that of the native enzyme with an absorbance maximum at 336 nm. Thus, the FAD-containing form of nitroalkane oxidase can be converted to the species present in the enzyme as isolated by incubation with a mixture of nitroethane and nitroethane anion.

DISCUSSION
When nitroalkane oxidase was first described by Kido et al. (16), an absolute requirement for added FAD or FMN for activity was reported. This suggested that nitroalkane oxidase might be a flavoprotein, but no spectral data were given at that time. Recently, Kurihara et al. (18) reported that the enzyme contains a chromophore that absorbs at 340 nm. More interestingly, denaturation of the protein with urea released oxidized FAD, consistent with nitroalkane oxidase being a flavoprotein. However, the stoichiometry was quite low, only 0.32 per monomer. This low stoichiometry suggested that the flavin content might be due to a contaminant. In this work we have taken advantage of the improved purification procedure developed by Heasley (19) for isolating and characterizing the cofactor of nitroalkane oxidase.
The cofactor released when purified nitroalkane oxidase is denatured by different methods at pH 7.0 has been identified as FAD. A stoichiometry of 0.98 FAD/monomer of enzyme has been determined; the previously reported value of 0.32 was presumably due to the presence of apoprotein in the preparation (18). A different form of flavin could be isolated at pH 2.1 The spectral properties of this flavin are not those of oxidized flavin, the typical oxidation state for the flavin cofactor in a flavoprotein oxidase. Rather, the FAD appears to be in the form of an adduct; the protein moiety is clearly not involved in this adduct. Taking advantage of the stability of this molecule at low pH, we have developed a method for isolating the flavin for structural studies. Using negative ion MALDI-TOF mass spectrometry, two peaks were observed in the mass spectrum of the flavin isolated at low pH. The lower mass species is FAD, presumably formed by fragmentation of the flavin adduct during the ionization process. The second species is FAD, with a nitrobutyl moiety bound.
Several observations suggest the N(5) of the isoalloxazine ring as the most likely site for the nitrobutyl substitution. Both the absorbance and fluorescence emission maxima of the enzyme are in the typical range for N(5) adducts, i.e. 300 -365 and 440 -490 nm, respectively (28). Indeed, the absorbance maximum of nitroalkane oxidase of 336 nm compares well with the value of 332 nm observed with D-amino acid oxidase upon formation of a 5-cyanomethyl-FAD (10) and with lactate oxidase reconstituted with N(5)-ethyl-FMN (28). Moreover, the spectrum of a recombinant mutant of glycolate oxidase in which 50 -80% of the flavin is in the form of an N(5) adduct is very similar to that of nitroalkane oxidase when about half of the adduct has been converted to FAD (29). The fluorescence emission maximum is also consistent with an N(5) flavin adduct; a value of 476 nm has been reported for the flavin N(5) adduct formed upon reaction of lactate oxidase with bromoacetate (28). Further evidence comes from the spectral properties of the flavin adduct of nitroalkane oxidase free in solution. The nitrobutyl-FAD extracted from nitroalkane oxidase showed an absorbance maximum at 346 nm and no fluorescence emission. The maximum absorbance of a series of N(5)-alkyl-substituted lumiflavins has been shown to be 325-350 nm, depending on the nature of the substituent (30); with these compounds no fluorescence emission has been detected at room temperature, whereas at 77 K in rigid glasses (a condition that has been suggested to resemble the motion constraints of an enzyme active site) the emission maximum is around 476 nm. Moreover, the 5-cyanomethyl-FAD of D-amino acid oxidase (10) and the bromoacetyl flavin adduct of lactate oxidase (28) have absorbance maxima at 330 and 345 nm, respectively, when free in solution. The pK a determined for the rate of the flavin adduct decay to FAD in solution provides additional support for the involvement of the N(5) locus in the adduct formation. This value can be attributed to ionization of the N(1) position of the isoalloxazine ring, since no ionizable groups are present on the 2-nitrobutyl moiety of the molecule. The value of 7.4 Ϯ 0.2 is in the expected range for an N(5)-substituted dihydroflavin (31)(32)(33), whereas it has been reported that C(4a)-substituted flavins do not show ionizable groups in the range 2-10 (31,32,34).
As far as the position of the nitro group of the nitrobutyl moiety is concerned, the stability of the flavin adduct bound to the enzyme strongly suggests that this group is positioned on the ␤-carbon of the butyl moiety (Scheme 2). An alternative is to position the substituent on the ␣-carbon; however, this appears to be unlikely, since in this case the molecule should be broken down as a normal substrate as in the case of 1-nitropropane or 1-nitropentane (17). Moreover, the formation of the adduct from a combination of neutral and anionic nitroethane is readily explained by such a structure.
Our data show that the 5-nitrobutyl-FAD form of nitroalkane oxidase is not catalytically active. This conclusion is supported by the observation that when the adduct bound to the enzyme is converted to FAD, the enzymatic activity of the protein increases with the same rate. Moreover, only FADcontaining enzyme can be reduced by substrate. The specific activity of nitroalkane oxidase correlates well with the amount of oxidized FAD bound to the enzyme. In all cases, the stoichiometry of total flavin (FAD plus adduct) bound to the enzyme was between 0.5 and 0.7; this clearly establishes the requirement of one flavin for catalysis.
The 5-nitrobutyl-1,5-dihydroflavin in solution is stable at acid pH, while it undergoes a rapid hydrolysis, giving FAD at neutral or alkaline pH. The rate of this process is pH-depend- SCHEME 2 FIG. 9. Visible absorbance spectrum of FAD-containing nitroalkane oxidase during incubation with a mixture of neutral and anion form of nitroethane. FAD-containing nitroalkane oxidase (14.2 M in FAD content) was incubated with 2% nitroethane in 0.2 M Tris-Cl, pH 8.0, at 30°C. In order to have 25% of the anion form in solution, the substrate was pretreated as described under "Experimental Procedures." Enzyme is shown before (--) and after 10 min of incubation (--) and after gel filtration (-⅐-⅐). ent, and a pK a with a value of 7.4, assigned to the N(1) position of the substituted 1,5-dihydroflavin, has been determined. General base catalysis can be invoked for the breakdown process, with the electron flow starting from the N(1) locus of the flavin (Scheme 3). The same pattern of events is likely to happen within the enzyme active site when the temperature is raised at 36°C. In the latter case, the rate of breakdown is not dependent on pH and is about 160-fold slower (at pH 8.0). Thus, the enzyme active site contributes to the stabilization of the molecule.
Nitroalkane oxidase has been reported to require the neutral form of the substrate for catalysis, based on pH studies (17); this conclusion is further supported by the lack of reactivity of the enzyme upon mixing with nitroethane anion. 3 In contrast, in the presence of the neutral form of nitroethane, the enzyme undergoes normal turnover and oxidized FAD is observed upon removal of the substrate (Fig. 7). This requirement for the neutral nitroalkane makes the enzyme unique, since both glucose oxidase and D-amino acid oxidase use the nitroethane anion as the substrate (10,35). In both cases, the reactions are nonphysiological. With D-amino acid oxidase, a covalent nitroalkane-flavin adduct can be formed in the presence of nitroethane anion and cyanide (10).
However, as shown here, when the enzyme is reacted with a mixture of the anion and neutral forms of nitroethane, upon removal of these molecules the 5-nitrobutyl-FAD is obtained (Fig. 9). These results clearly show that the anion nitroethane has to be formed in solution, ruling out the possible involvement of the enzyme active site in this process, i.e. through the abstraction of the ␣-proton. The nitrobutyl flavin is presumably formed in vivo during turnover of the nitroethane used for induction, due to the presence of a small amount of the anion nitroethane at the pH used for growing the cells. During cell growth, nitroethane is the only nitro compound present in the medium; thus, the formation of the adduct would be due to a similar side reaction. 4 These results can be rationalized with a variation of the model proposed for the reaction of D-amino acid oxidase with nitroethane and cyanide (10) (Scheme 4). Catalysis is initiated by proton abstraction of the ␣-carbon of nitroethane. An amino acid residue in nitroalkane oxidase with a pK a of 6.7 has been recently proposed to act as the base for abstracting the ␣-proton from the substrate (17). Next, the substrate carbanion forms an adduct at the N(5) position of the flavin; such an adduct would then break down to form the reduced flavin, and the product would be released. Nitroethane anion preformed in solution can react with the highly reactive cationic imine formed after the elimination of nitrite giving the 5-nitrobutyl-FAD. This molecule would be frozen in the enzyme active site, probably as a consequence of hydrogen bonds and/or salt bridges between the nitro group of the second nitroalkane added and a positive charged group(s) in the active site of the enzyme. It is noteworthy that the nitro group is now positioned on the ␤-carbon of the nitrobutyl moiety, while for breakdown of the adduct it has to be on the ␣-carbon. 5 Although not investigated in detail, it has been proposed that with D-amino acid oxidase, a similar addition of a second nitroethane anion "can interrupt the oxidation-reduction process by forming a stabilized flavin-substrate adduct" (10).
To our knowledge, the results reported here for the first time show the structure and elucidate the mechanism of formation of an N(5) adduct produced in vivo by the natural microorganism. It appears puzzling that in F. oxysporum nitroalkane oxidase is produced in high yields with the cofactor in an inactive form, whereas the active form required for catalysis is easily released in solution. This raises the question of whether the inactive 5-nitrobutyl-FAD of nitroalkane oxidase should be considered a catalytic by-product or a sort of storage unit of FAD, in order to have the catalytically competent flavin species readily available through an activation mechanism not yet identified.