Mechanistic studies of p-hydroxybenzoate hydroxylase reconstituted with 2-Thio-FAD.

2-Thio-FAD (oxygen substituent at position 2 is replaced by sulfur) was used to reconstitute the apoenzyme of p-hydroxybenzoate hydroxylase. The 2-thio-FAD enzyme differs from native enzyme in several respects. While the native enzyme catalyzes the fully coupled hydroxylation of p-hydroxybenzoate, the 2-thio-FAD enzyme shows no hydroxylation of this substrate, instead reducing molecular oxygen to hydrogen peroxide. The rate of reduction of 2-thio-FAD p-hydroxybenzoate hydroxylase by NADPH in the presence of substrate was 7-fold faster than with the native enzyme. However, the oxygen reactivity of the reduced 2-thio-FAD enzyme was less than 1% that of native enzyme. This slow oxygen reaction results in the very high KmO2 observed in steady state kinetic studies of the modified enzyme. Stopped flow studies of the oxygen reaction of the reduced 2-thio-FAD enzyme in the presence of substrate confirmed the formation of a transient intermediate. The spectrum of this intermediate is very similar to those of the flavin-C(4a) adducts obtained with 2-thio-FMN lactate oxidase. This evidence suggests that reduced 2-thio-FAD p-hydroxybenzoate hydroxylase forms a flavin-C(4a)-hydroperoxide on reaction with oxygen in a reaction analogous to that with native enzyme, but that the resulting peroxyflavin is incompetent as an oxygenating species, breaking down instead to oxidized 2-thio-FAD enzyme and hydrogen peroxide.

The use of modified flavins as probes of flavoprotein structure and mechanism has been advanced in recent years with considerable success (1,2). Recent mechanistic studies in this laboratory have employed the 8-mercapto-, 8-chloro-, and 6hydroxy-FAD analogs with xanthine oxidase (3), 7,8-dichloro-FAD with amino acid oxidase (4), and iso-FMN (5) and 2thio-FMN' with lactate oxidase. Previous studies of D-amino acid oxidase and glucose oxidase had made use of the l-deaza-FAD analog, which is characterized by an oxidation-reduction potential 72 mV more negative than native FAD (6, 7 ) .  replaced by sulfur; Structure 1) had earlier been used in studies of the charge-transfer complexes of Old Yellow Enzyme (a), as a structural probe of the covalent flavin adducts of lactate oxidase (9), and with the Azotobacter uinelandii flavodoxin (10) and rabbit liver pyridoxamine 5"phosphate oxidase (11). Previous attempts to apply the 2-thio-FAD coenzyme as a mechanistic probe of adrenodoxin reductase were complicated by the presence of contaminating FAD in the flavin preparations (12). However, recent work in this laboratory has resulted in stable preparations of the 2-thio-FAD coenzyme free of FAD contamination (13). The chemical reactivity of the sulfur substituent at the flavin 2-position toward thiol reagents (methyl methanethiolsulfonate; Ref. 13) and peroxides (H202, m-chloroperoxybenzoic acid)2 to yield the 2-SSCH3 flavin disulfide and the flavin 2-S-oxide, respectively, makes this flavin a useful probe of active site structure.

C % *
The hydroxylation pathway deduced from mechanistic studies with p-hydroxybenzoate hydroxylase from Pseudomonas fluorescens involves three oxygenated flavin intermediates with distinct spectral properties (14). While the structures of intermediates I and I11 are well established as the flavin-C(4a)-hydroperoxide and the C(4a)-hydroxyflavin, respectively (15), the structure of intermediate 11 has yet to be established (14,16). Reconstitution of the enzyme with various FAD analogs might be expected to yield valuable kinetic and spectral information relevant to the structure of this flavinoxygen derivative.
However, reconstitution of the enzyme with 1-deaza-FAD resulted in a catalytically competent NADPH oxidase which failed to carry out substrate hydroxylation (17). This result was particularly significant in view of the observed formation of a transient enzyme-bound l-deazaflavin-C(4a)-hydroperoxide. This intermediate, instead of oxygenating bound phydroxybenzoate, broke down to yield oxidized enzyme and hydrogen peroxide.
We were required in all catalytic turnover experiments; for single turnover and spectral studies, residual FAD levels of 1 4 % were acceptable. 2-Thio-FAD was prepared as previously described from 2-thioriboflavin (13), and was used to reconstitute the apoenzyme following the general procedure of Entsch et al. (1 7).
Oxygen Incorporation into Substrate-Two procedures were followed in measuring the stoichiometry of substrate hydroxylation by the 2-thio-FAD enzyme. The f m t method, described in detail by Entsch et al. (17), involves high pressure liquid chromatography analyses of 3,4-dhydroxybenzoic acid formation in reaction mixtures following reoxidation of stoichiometrically reduced enzyme in the presence of substrate. The second method employed catalytic amounts of enzyme with an oxygen electrode (Yellow Springs Instrument Co., model 53). 0.15 PM 2-Thio-FAD enzyme (0.3% residual FAD) was allowed to react with 16 p~ NADPH in the presence of 1 mMp-hydroxybenzoate and 0.26 mM 0 2 at 25 "C. After exhaustion of the limiting substrate (NADPH), catalase was added. The stoichiometry of H202 production during NADPH oxidation was measured from the decomposition of peroxide by catalase (18). In the case of complete uncoupling of NADPH oxidation and substrate hydroxylation, 50% of the oxygen initially consumed was returned with the addition of catalase.
Steady State Kinetics-The general procedures of Husain and Massey (19) were followed in initial velocity measurements of 2-thio-FAD enzyme activity with varying oxygen concentrations. The stopped flow spectrophotometer previously described (20) was also convenient for steady state kinetics. The instrument was operated manually; photometer output was channeled to an x-y recorder, giving a direct plot of Az40 uersus time, to follow NADPH oxidation, and with easily attainable variation of the initial 0 2 concentration.
In the enzyme monitored turnover experiments, as in all other rapid kinetic studies, the Nova minicomputer system (21)  Enzyme Reduction-The rapid anaerobic reduction of the 2-thio-FAD enzyme by NADPH in the presence of substrate was followed in the stopped flow spectrophotometer. Using general methods previously described (19). enzyme reduction was monitored at wavelengths over the range 450-700 nm in order to determine whether transient species were involved. Kinetic measurements as a function of NADPH concentration were made at 500 nm. The slow anaerobic reduction of the enzyme in the absence of substrate was followed at 500 nm with a Cary 118 double beam spectrophotometer.
Rapid Kinetics of Enzyme Reoxidation-Reoxidation of dithionite-reduced enzyme in the absence or presence of substrate was also followed in the stopped flow spectrophotometer, as previously described (17). Anaerobic enzyme solutions were reduced with sodium dithionite; a small excess of reductant (30%) was added to remove traces of oxygen. Reduced enzyme was then rapidly mixed with oxygenated solutions; reoxidation was followed as a function of oxygen concentration at many wavelengths. The use of buffered solutions equilibrated with 100% O2 at 0 "C allowed fiial 0 2 concentrations of 1 mM in some of these experiments.
Anaerobiosis-Commercial purified nitrogen was freed of residual oxygen by passage over heated copper turnings. Anaerobic enzyme solutions were prepared by alternate cycles of evacuation and equilibration with oxygen-free nitrogen. Anaerobic solutions of 3,4-dihydroxybenzoic acid and protocatechuic acid dioxygenase (23) were used to scrub residual oxygen from the stopped flow system prior to aU anaerobic experiments. All other anaerobic techniques are described elsewhere (24).
All experiments with the 2-thio-FAD enzyme, unless otherwise noted, were performed at 4 "C in 50 m~ potassium phosphate, pH 7.0, containing 0.3 mM EDTA. Stock solutions of NADPH were kept at slightly alkaline pH to avoid nonenzymatic loss. Absorption spectra were recorded with either a Cary 118 or Cary 219 double beam spectrophotometer thermostatted at 4 "C and purged with dry air to avoid fogging. Calculation

Spectral Properties of 2-Thio-FAD p-Hydroxybenzoate
Hydroxylase-The visible absorbance spectrum of the 2-thio-FAD reconstituted enzyme at pH 7.0 is shown in a later section (Fig. 5). Previous studies (13) demonstrated a tight binding affinity of the apoenzyme for 2-thio-FAD (Kdlo-' the other substrates (19). However, variations in NADPH and p-hydroxybenzoate concentrations over the respective ranges of 14-140 PM and 16-160 ELM at 0.26 m 0 2 did not give significant changes in initial velocity. This observation implied that the K, values for NADPH and p-hydroxybenzoate were considerably lower than for the native enzyme (19).
At fiied concentrations of NADPH andp-hydroxybenzoate, an unusual initial velocity dependence was observed with varying O2 concentration. As the concentration of 0 2 was increased from 0.13-0.74 m, the rate of oxidation of NADPH was found to be directly proportional to the 0 2 concentration. This steady state kinetic behavior is consistent with a K 2 value much higher than the maximum O2 concentration (0.74 m) employed in these experiments. A second order rate constant of 830 M" s-l was determined. The highK2 of the 2-thio-FAD enzyme contrasts markedly with the&? value of 31 ,UM reported for the native enzyme at pH 6.6 (19).
In order to investigate the steady state behavior of the 2thio-FAD enzyme at lower concentrations of NADPH and phydroxybenzoate, the enzyme monitored turnover method fist described by Chance (27) was employed. This technique allows the determination of turnover numbers as the limiting substrate concentration approaches zero. The stopped flow traces shown in Fig. 1 represent the change in enzyme absorbance at 500 nm as oxidized 2-thio-FAD enzyme is mixed with 100 p~ NADPH in the presence of excessp-hydroxybenzoate and 02. The rapid reduction of the enzyme is followed by gradual reoxidation; the latter process is seen to accelerate as NADPH is exhausted. The data were analyzed according to Gibson et al. (22) to generate a series of reciprocal plots (turnover number uersus NADPH concentration) at several fmed levels of p-hydroxybenzoate.
An intersecting pattern of reciprocal plots over the p-hydroxybenzoate concentration range 10-31 ,UM was found, consistent with a ternary complex mechanism for the 2-thio-FAD enzyme involving NADPH and p-hydroxybenzoate, as found for the native enzyme (19). However, at higher concentrations of p-hydroxybenzoate, the double reciprocal plots became parallel, and secondary plots of intercept uersus reciprocal phydroxybenzoate concentration curve downward at higher concentrations (0.1-1 m), indicating a possible secondary interaction of this substrate with the enzyme. While this activating effect ofp-hydroxybenzoate at high concentrations was not further explored, it should be recalled that high substrate concentrations inhibit the native enzyme (28).
Reductive Half-reaction-The slower rate of reduction of the 1-deaza-FAD enzyme by NADPH was attributed to the 0. 15  lower oxidation-reduction potential of -280 mV measured for 1-deazariboflavin, compared to the potential of' -208 mV for riboflavin (17). The higher oxidation-reduction potential of -126 mV measured for 2-thioriboflavin (12), therefore, might predict a faster rate of NADPH reduction with the 2-thio-FAD enzyme. This prediction was confirmed in stopped flow studies of the reductive half-reaction of this enzyme.
The reduction of the 2-thio-FAD enzyme by NADPH in the presence of 1.2 mM p-hydroxybenzoate is shown in Fig.   2   3. The respective second order rate constants, taken at 4 "C, of 733 and 2100 M" s-' show that the oxygen reactivity of the 2-thio-FAD enzyme is much slower than with native enzyme. Comparable values for the native FAD enzyme, taken at pH 6.6 and 4 "C (14), are 2.6 X lo4 and 2.6 X lo5 M" s-', respectively; the reduced 2-thio-FAD enzyme reacts 120-fold more slowly with oxygen than does the native enzyme in the presence of substrate. Despite the poor oxygen reactivity, the presence of substrate does accelerate the oxidation rate to a modest extent (&fold) with the 2-thio-FAD enzyme; the size of this effect with native enzyme is 10-fold (14).
Although the 1-deaza-FAD enzyme does not hydroxylate substrate, the appearance of a l-deaza-FAD-C(4a)-hydroperoxide intermediate has been documented (17) in stopped flow studies of the oxidation of reduced 1-deaza-FAD enzyme in complex with p-hydroxybenzoate, p-aminobenzoate, and the effector 6-hydroxynicotinate. This peroxyflavin, however, is unable to carry out oxygenation of substrates, instead breaking down to yield the oxidized 1-deaza-FAD enzyme and hydrogen peroxide.
Stopped flow studies of the reduced 2-thio-FAD enzyme in the presence ofp-hydroxybenzoate, on reoxidation with 0.65-1 mM O2 at 4 "C, were made in order to establish whether this modified enzyme also formed a C(4a)-peroxyflavin intermediate. At most wavelengths, the return to the absorbance of the oxidized 2-thio-FAD enzyme-substrate complex was monophasic (kob, = 1.4 to 1.6 s-l at 0.65 mM 0 2 ) . At other wavelengths, notably in the region below 390 nm, the reoxidation was not monophasic, indicating the formation of intermediate species. The failure to observe a clear 2-thio-FAD-C(4a)-peroxyflavin intermediate under these conditions may well be a consequence of the poor oxygen reactivity of the reduced 2thio-PAD enzyme, so that the formation of the hydroperoxide is slower than its decay. This is not an unreasonable possibility, since decay rates with native and 1-deaza-FAD enzymes occur in the range 4.6-10.7 s-' at 2-5 "C (14,17).
Previous studies3 of the oxygen reactivity of the native enzyme (14) had shown that the kinetic resolution of intermediates could be greatly improved by the addition of 0.1 M ' In these studies, it was shown that 0.1 M azide selectively affected rates of formation and decay of the observed intermediates, but without effect on hydroxylation stoichiometry or enzyme stability. Furthermore, with several substrates, the same intermediates were observed in the presence and absence of azide. The effect of azide was shown to involve only specific kinetic constants, and appears to be related to a specific binding site on the enzyme. azide to reaction mixtures. For example, the oxidation of reduced native enzyme at pH 6,6 and 4 "C in the absence of substrate was found to proceed (14) via a kinetically invisible intermediate in the absence of azide, due to the rate of decay of the flavin hydroperoxide being faster than the rate of its formation: However, in the presence of 0.1 M azide, the course of reoxidation with the native enzyme at appropriate wavelengths was markedly biphasic (14)

2-Thio-FADp-Hydroxybenzoate
Hydroxylase was found to have no effect on the [oxygen]-dependent rate kl, but dramatically decreased kz to the level where the intermediate could be observed. Therefore, we undertook a stopped flow study of the oxidation of the 2-thio-FAD enzymesubstrate complex in the presence of 0.1 M potassium azide and 1 m~ 0 2 .
A clearly biphasic course of oxidation with the reduced 2thio-FAD enzyme-substrate complex was observed under these conditions. At wavelengths in the range 340-370 nm, there was a rapid increase in absorbance followed by a slower return to that of the oxidized enzyme-substrate complex. The absorbance change was greatest at 355-360 nm. By following the reaction at appropriate isosbestic wavelengths (see below), it was established that the rapid increase in absorbance was  Fig. 4. By using the methods previously described (14) and modified as under "Experimental Procedures," knowledge of the rate constants for the reaction scheme  Fig. 5. This spectrum is similar to those of the covalent C(4a) adducts of 2-thio-FMN lactate oxidase with P-bromopropionate and ar-hydroxybutynoate (9).

DISCUSSION
The 2-thio-FAD reconstituted p-hydroxybenzoate hydroxylase catalyzes the p-hydroxybenzoate-stimulated oxidation of NADPH by a mechanism similar to that observed for both the 1-deaza-FAD enzyme (17) and for the native enzyme in the presence of nonhydroxylatable effectors (14,28). This mechanism is summarized in Scheme 1. The kinetic constants determined by steady state and stopped flow methods for the 2-thio-FAD enzyme and for the native enzyme (Table I) suggest that the low oxygen reactivity observed with the modified enzyme is responsible for the unusual 0 2 dependency seen in steady state studies. This The abbreviation used is: PHB, p-hydroxybenzoate.  * A l l values for the native enzyme are from Refs. 14 and 19, and were obtained at pH 6.5-6.6,3.5-5 "C.
The value for k g is taken from Ref. 14, and was determined in the presence of the nonhydroxylatable effector 6-hydroxynicotinate.
Rate constants k'7 and k', refer to the rates of oxidation and reduction, respectively, of the enzyme in the absence of p-hydroxybenzoate.
kinetic behavior precludes the determination of accurate values for V , , , KLnB, KZADPH, as well as P m 2 .
However, we can estimate (30) that the K% value for the 2-thio-FAD enzyme must be at least 10 m~ or higher. Similar steady state oxygen dependencies were noted previously in studies of the 7,8dichloro-FAD D-amino acid oxidase (4) and the iso-FMN lactate oxidase (5). In both cases, the reactions of 0 2 with the respective reduced enzyme-imino acid or enzyme-pyruvate complexes were much slower than with the native enzymes. The low oxygen reactivity of 2-thio-FAD p-hydroxybenzoate hydroxylase is not due to an inherent poor oxygen reaction of the 2-thioflavin, since the reaction of O2 with reduced 2-thio-FMN lactate oxidase is rapid.' There does not appear to be any correlation between the oxidation-reduction potentials of the free flavins and the rates of oxidation of p-hydroxybenzoate hydroxylase (as the enzyme-substrate complex) containing 1-deaza-, native, or 2thio-FAD. The data of Table I1 show that the 1-deaza-FAD enzyme reacts with 0 2 with a second order rate constant less than 30% that of the native enzyme under identical conditions. The 2-thio-FAD enzyme at pH 7.0 and 4 "C reacts more slowly than either the 1-deaza-FAD enzyme or the native enzyme.
On the other hand, a reasonable correlation appears to exist between the limiting rate of two-electron NADPH reduction of the coenzymes bound to p-hydroxybenzoate hydroxylase and the oxidation-reduction potentials of the free flavins (Table 11). A plot of log k, versus EL (for the free riboflavin derivative) for these three enzyme forms is linear with a slope  The formation of C(4a)-peroxyflavin intermediates in the oxygen reactions of native, 1-deaza-FAD, and 2-thio-FAD phydroxybenzoate hydroxylase represents a common f i s t step in their respective oxidative mechanisms. However, neither the 2-thio-FAD nor 1-deaza-FAD enzymes carry out substrate hydroxylation. It is of interest to note that in both modified coenzymes, an electronegative element of the pyrimidine subnucleus has been replaced by a less electronegative substituent. Thus, the N(1) nitrogen is replaced by carbon in the 1deazaflavin, and the O(2a) oxygen is substituted with sulfur in the 2-thioflavin. The effect of such diminished electronegativity in the corresponding flavin-C(4a)-hydroperoxides (Structure 2) could influence the polarization of the 0-0 bond, suggested to account in part for the electrophilic nature of the terminal oxygen atom that is incorporated into substrate during hydroxylation (31,32).
The oxenoid mechanism proposed by Hamilton (32) involves a rapid equilibrium between the flavin-C(4a)-hydroperoxide and a carbonyl oxide species resulting from cleavage of the C(4a)-N(5) bond (Structure 3). According to this hypothesis, delocalization of the formal negative charge over the other electronegative atoms of the pyrimidine subnucleus would allow the terminal oxygen atom to attain substantial electrophilic character, facilitating nucleophilic attack of the substrate on the terminal oxygen in the oxygen transfer process (Structure 4). The diminished electronegativity resulting from substitutions at either N(l) or O(2a) in the 1-deaza-and 2-thioflavins would be expected to lessen the resonance stabilization of the carbonyl oxide and to diminish the electrophilic character of the terminal oxygen atom. Instead of transferring oxygen to substrate, the flavin-C(4a)-hydroperoxide would decompose to yield oxidized flavin and Hz02 (Stfucture 5). Ball and Bruice (31) have demonstrated that the N(5)ethyl-l-deazatlavin-C(4a)-hydroperoxide is capable of oxidations of iodide, thioxane, and amines, but at rates ranging from 3-17-fold slower than those with the corresponding N(1)flavin-C(4a)-hydroperoxide. It was suggested that the observed poorer reactivity of the 1-deazatlavin hydroperoxide was due to the less electronegative character of the flavin-C(4a)-position, resulting in a less polarized 0-0 bond in the hydroperoxide. While our results with the 2-thio-FAD p-hydroxybenzoate hydroxylase do not serve to distinguish between the carbonyl oxide mechanism of Hamilton (32), or the alternative ring opening suggested by Entsch et al. (14), and the model suggested by Bruice (31) involving inductive polarization of the peroxide moiety during oxygen transfer, we can suggest that slower rates of substrate oxidation should be observed for the synthetic N(5)-alkyl-2-thioflavin-C(4a)-hydroperoxide as compared to the normal peroxyflavin. Further studies of p-hydroxybenzoate hydroxylase reconstituted with suitably modified flavins should allow closer examination of the influence of electronegative character on the course of peroxytlavin decay.