Reduced nicotinamide adenine dinucleotide phosphate-sulfite reductase of enterobacteria. 3. The Escherichia coli hemoflavoprotein: catalytic parameters and the sequence of electron flow.

Abstract NADPH-sulfite reductase (EC 1.8.1.2) from Escherichia coli is a complex hemoflavoprotein, molecular weight 670,000, containing 4 FMN, 4 FAD, 20 to 21 atoms of iron, 14 to 15 labile sulfides, and 3 to 4 molecules of a novel type of heme per enzyme molecule. This heme has been identified as an octacarboxylic iron-tetrahydroporphyrin (Murphy, M. J., Siegel, L. M., Kamin, H., and Rosenthal, D. (1973) J. Biol. Chem. 248, 2801). The enzyme catalyzes the stoichiometric conversion of sulfite to sulfide at the expense of 3 NADPH. The Km values for sulfite and NADPH are both 4 to 5 µm. Reduced methyl or benzyl viologens can serve as electron donors for sulfite reduction, but NADH cannot. In addition to sulfite reduction, the enzyme catalyzes the NADPH-dependent reduction of a variety of "diaphorase" acceptors (cytochrome c, ferricyanide, 2,6-dichlorophenolindophenol, menadione, FMN, FAD) as well as NADPH oxidase, NADPH-3-acetylpyridine adenine dinucleotide phosphate transhydrogenase, NADPH-nitrite and -hydroxylamine reductase and reduced methyl viologen-NADP+ reductase activities. All NADPH-dependent activities examined were competitively inhibited by NADP+. Agents which react with the heme prosthetic group, i.e. CO, cyanide, and arsenite, inhibit the reductions of sulfite, nitrite, and hydroxylamine (with either NADPH or reduced methyl viologen as electron donor), while all other activities are unaffected. Cyanide and CO binding to and CO dissociation from the enzyme (determined spectrophotometrically) parallel the respective development and relief of inhibition of NADPH-sulfite reductase activity. Development of inhibition requires the presence of reductant (NADPH) as well as inhibitor, in accord with the observation that CO, cyanide, or arsenite can react with reduced, but not oxidized enzyme. Treatment of enzyme with 1 µm p-chloromercuriphenylsulfonate causes the dissociation of virtually all of the FMN while permitting retention of FAD and heme. This treatment inhibits all pyridine nucleotide-dependent reactions of the enzyme except transhydrogenase and FMN reductase. The methyl viologen-sulfite reductase is unaffected. The development of fluorescence due to FMN release parallels the development of the observed inhibitions. The FAD of the FMN-free enzyme is reducible by NADPH, but the heme is not. If exogenous FMN is added, the heme becomes reducible and all NADPH-dependent activities are restored. We have concluded that electron flow from NADPH to sulfite follows the minimum linear sequence: NADPH → FAD → FMN → heme → sulfite In this scheme, FAD serves as the "entry port" for electrons from NADPH. It can transfer electrons directly to FMN (internal or external) or to pyridine nucleotides and their analogues. The heme is required for electron transfer to sulfite (and nitrite and hydroxylamine). The FMN is required for electron transfer from the reduced FAD to the heme (and hence to acceptors dependent on the heme) or, more directly, to diaphorase-type acceptors and O2. Reduced methyl viologen can donate electrons to both the FMN and heme. The patterns of inhibition by a variety of salts of the NADPH-cytochrome c and reduced methyl viologen-sulfite reductase reactions are consistent with the hypothesis that these two reactions involve independent portions of the enzyme molecule.

L. M., KAMIN, H., AND ROSENTHAL, D. (1973) J. Bid. Chem. 248, 2801). The enzyme catalyzes the stoichiometric conversion of sulfite to sulfide at the expense of 3 NADPH. The K,,, values for sulfite and NADPH are both 4 to 5 pM.
Reduced methyl or benzyl viologens can serve as electron donors for sulfite reduction, but NADH cannot. In addition to sulfite reduction, the enzyme catalyzes the NADPH-dependent reduction of a variety of "diaphorase" acceptors (cytochrome c, ferricyanide, 2,6dichlorophenolindophenol, menadione, FMN, FAD) as well as NADPH oxidase, NADPH-3-acetylpyridine adenine dinucleotide phosphate transhydrogenase, NADPH-nitrite and -hydroxylamine reductase and reduced methyl viologen-NADPf reductase activities. All NADPH-dependent activities examined were competitively inhibited by NADP+.
Agents which react with the heme prosthetic group, i.e. CO, cyanide, and arsenite, inhibit the reductions of sulfite, nitrite, and hydroxylamine (with either NADPH or reduced methyl viologen as electron donor), while all other activities are unaffected. Cyanide and CO binding to and C-0 dissociation from the enzyme (determined spectrophotometrically) parallel the respective development and relief of inhibition of NADPH-sulfite reductase activity. Development of inhibition requires the presence of reductant (NADPH) as well as inhibitor, in accord with the observation that CO, cyanide, or arsenite can react with reduced, but not oxidized enzyme.
Treatment of enzyme with 1 PM p-chloromercuriphenylsulfonate causes the dissociation of virtually all of the FMN * These studies were supported in part by Research Grants AM-13460 and AM-040663 from the National Institutes of Health, and GB-7905 from the National Science Foundation, Veterans Administration Project No. 7875-01. $ To whom inquiries should be addressed.
while permitting retention of FAD and heme. This treatment inhibits all pyridine nucleotide-dependent reactions of the enzyme except transhydrogenase and FMN reductase. The methyl viologen-sulfite reductase is unaffected. The development of fluorescence due to FMN release parallels the development of the observed inhibitions. The FAD of the FMN-free enzyme is reducible by NADPH, but the heme is not. If exogenous FMN is added, the heme becomes reducible and all NADPH-dependent activities are restored.
We have concluded that electron flow from NADPH to sulfite follows the minimum linear sequence: NADPH + FAD + FMN ---f heme -+ sulfite In this scheme, FAD serves as the "entry port" for electrons from NADPH. It can transfer electrons directly to FMN (internal or external) or to pyridine nucleotides and their analogues. The heme is required for electron transfer to sulfite (and nitrite and hydroxylamine). The FMN is required for electron transfer from the reduced FAD to the heme (and hence to acceptors dependent on the heme) or, more directly, to diaphorase-type acceptors and OZ. Reduced methyl viologen can donate electrons to both the FMN and heme. The patterns of inhibition by a variety of salts of the NADPH-cytochrome c and reduced methyl viologen-sulfite reductase reactions are consistent with the hypothesis that these two reactions involve independent portions of the enzyme molecule.
He also noted that an NADPH-hydroxylamine reductase activity copurified with the sulfite reductase.
Lazzarini and Atkinson (4) further purified this enzyme as a NADPHnitrite reductase, and Kemp et al. (5) in Atkinson's laboratory subsequently showed that the NADPH-sulfite, nitrite, and hydroxylamine reductions were catalyzed by a single enzyme. These workers showed that all three activities were inhibitable by arsenite and the mercurical p-chloromercuribenzoate as well as by cyanide; furthermore, the enzyme preparation contained a NADPH-cytochrome c reductase activity which copurified with the other activities cited, and which was repressed in cysteine-grown organisms.
We have purified to homogeneity the NADPH-sulfite reductase of E. coli (6) and shown it to be a complex hemoflavoprotein of molecular weight 670,000. The enzyme contains, per mole, the following prosthetic groups: 4 FAD, 4 FMN, 20 to 21 atoms of iron, 14 to 15 labile sulfides, and 3 to 4 moles of a novel type heme. This heme has been identified (7) as an octacarboxylic iron-tetrahydroporphyrin of the isobacteriochlorin type (adjacent pyrrole rings reduced), and has now been observed to serve as prosthetic group of several sulfite reductases, both assimilatory and respiratory (8,9). It has been termed "siroheme" (8).
It is our object to describe the catalytic mechanism whereby a 6-electron reduction is accomplished by this hemoflavoprotein, one of the most complex arrays of electron-transport prosthetic groups yet observed in a single enzyme. To this end, we have investigated the interaction of sulfite reductase with a variety of electron donors, acceptors, and inhibitors, and have studied the effect of these agents both upon catalysis and upon optical properties of the enzyme. The results reported in this paper, some of which have been presented previously in preliminary form (lo), support the following conclusions: (a) The site of entry of pyridine nucleotide electrons is probably FAD.
(5) The site of interaction of sulfite with enzyme appears to be the heme. (c) The FMN prosthetic group is required for electron transfer between the reduced FAD and the heme. These studies have not as yet assigned a specific role for the non-heme iron-labile sulfide groupings.
NaHSOs, KNO*, NH%OH.HCl, and K,Fe(CN)s were Baker "Analyzed" reagents. CO, Hz, and Nz were purchased from Matheson; the latter two gases were freed of residual oxygen before use by passage through a column of hot copper. PAPS was prepared by the method of Kredich (11). E. coli sulfite reductase was purified by the procedure of Siegel et al. (6) ; all enzyme samples used in this study had a specific activity of at least 2.8 units 1 The abbreviations used are: p-CMPS, p-chloromercuriphenylsulfonate, monosodium salt; AcPyADP+, 3-acetylpyridine adenine dinucleotide phosphate; AcPyADPH, reduced 3-acetylpyridine adenine dinucleotide phosphate; DCIP, 2,6-dichloroindophenol; MVH, reduced methyl viologen; PAPS, adenylyl sulfate-3-phosphate. per mg. Naz a5S03, specific activity 15 Ci per mole, was purchased from New England Nuclear.
Enzyme Assays-NADPH-dependent reduction reactions were measured in l.O-ml reaction volumes containing 0.1 M potassium phosphate buffer (pH 7.7), 0.2 mM NADPH, acceptor. and an appropriate amount of enzyme. Acceptors were present at the following concentrations: sulfite, 0.5 mM; nitrite or hydroxylamine, 10 mM; oxygen, 0.25 mM; ferricyanide, menadione, FMN, FAD, DCIP, or cytochrome c, 0.1 mM; AcPyADP+, 0.2 mM. Rates were measured spectrophotometrically using a Cary model 14 spectrophotometer, with a control solution which for most reactions contained buffer in place of electron acceptor in the reference cuvette; for the NADPH-cytochrome c, DCIP, and AcPyADP+ reductase reactions, in which reduction of acceptor rather than oxidation of NADPH was measured, the control solution contained buffer in place of enzyme. NADPHferricyanide reductase and NADPH oxidase activities were also corrected for the nonenzymatic reaction. Absorbance changes were followed at 340 nm for all acceptors except the following: cytochrome c, (550 nm) ; DCIP, (600 nm); AcPyADP+, (363 nm).
MVH-dependent reduction reactions were measured under anaerobic conditions in Thunberg cuvettes fitted with serum caps. Reaction mixtures contained, in 2.5 ml total volume, 0.1 1~ potassium phosphate buffer (pH 7.7), 0.1 mM MVH, acceptor (0.2 mM sulfite or NADP+), and an appropriate amount of enzyme. Buffer and acceptor, in a 2.3.ml volume, were added to the main compartment of the Thunberg cuvette, and 0.1 ml of enzyme was added to the side arm. The system was bubbled with On-free Nz for 15 min. The enzyme was then tipped in and 0.1 ml of MVH (reduced with H2/Pt asbestos) was added with a gas-tight Hamilton syringe to start the reaction.
Control mixtures contained buffer in place of electron acceptor.
Rates of MVH oxidation were measured spectrophotometrically at 604 nm using a Cary model 14 spectrophotometer.
Other Assays-Concentration of sulfite reductase was determined spectrophotometrically, using an extinction coefficient for the enzyme of 3.1 x lo5 M-I cm-r at 386 nm (6). Protein was measured by the Zamenhof (17) adaptation of the microbiuret method described previously (6). Sulfide and sulfite were measured by the methods of Siegel (18) and Grant (19), respectively; concentrations of standard solutions were determined by iodometric titration.
FMN and FAD were measured fluorometrically by the procedure of Faeder and Siegel (20) ; concentrations of standard solutions were determined spectrophotometrically by means of their absorbances at 450 nm, utilizing reported extinction coefficients (1.22 X lo4 Me' cm-l for FMN and 1.13 x lo4 Me1 cm-' for FAD (21,22)).
Spectroscopic illeusurements-Absorption spectra were measured, tiersus appropriate solvent blanks, with a Cary model 14 spectrophotometer equipped with 0 to 0.1 A and 0 to 1.0 A slide wires. Fluorescence spectra were measured in a Turner model 210 spectrophotofluorometer, equipped with constant energy attachment.
For determination of flavin concentrations, an excitation wavelength of 450 nm (band width 10 nm) and an emission wavelength of 535 nm (band width 25 nm) were utilized.
Fluorescence polarization measurements were made with a Farrand Mark II spectrophotofluorometer. All spectroscopic measurements were performed at 23-25", utilizing l-cm light paths unless otherwise indicated.
Radioactivity Measurements-Radioactivity of a5S-containing solutions was determined on appropriately diluted aliquots (4 ml of aqueous sample plus 16 ml of the xylene-Triton X-114 mixture of Greene (23), with the naphthalene omitted) with a Packard model 3375 Tri-Carb liquid scintillation spectrometer. For all samples of standards and unknowns, measurement of radioactivity was continued until the statistical counting error was less than 1 To.
Concentration and Gel FiZtration-Ultrafiltration of enzyme solutions was performed with an Amicon concentrator equipped with a Diaflo PM-30 membrane.
For removal of low molecular weight solutes from enzyme in ligand-binding experiments, l.Oml samples were applied to a column (1.5 X 15 cm) of Sephadex G-25 (coarse) and 1.2.ml fractions were collected. Following either concentration or gel filtration, enzyme content was determined in appropriate fractions by measurement of protein concentrat.ion.
In all such experiments, recovery of enzyme protein was at least 85%.

Sulj2e Reduction
Stoichiometry-As shown in Table I, E. coli sulfite reductase catalyzes the stoichiometric reduction of sulfite to sulfide at the expense of 3 NADPH, this stoichiometry being maintained throughout the course of the reaction.
For these measurements, a reaction mixture containing NADPH, sulfite, and enzyme was incubated for varying periods during which t.he amount of NlZDPH oxidized was followed by the absorbance change at 340 nm. The reaction was stopped by addition of the colorforming reagents for determination of either sulfite or sulfide. Anaerobiosis was maintained to prevent oxygen-dependent consumption of NADPH (due to the NADPH oxidase activity of the enzyme, vide infra) and thereby avoid a spuriously high NADPH-sulfide stiochiometry.
The results demonstrate that purified E. coli sulfite reductase can catalyze the complete 6electron reduction of sulfite to sulfide without the accumulation of significant quantities of sulfur-containing compounds of intermediate oxidation states. This behavior is in marked contrast to that reported for the dissimilatory sulfite reductases of Desulfovibrio (24,25) and Desuljotomaculum (26), which appear to catalyze an incomplete reduction of sulfite to sulfide, with an observed stoichiometry of 10 to 12 electrons consumed per sulfide produced.
With the Desulfovibrio enzyme, sulfurcontaining intermediates such as trithionate and thiosulfate have been reported to accumulate in the reaction mixture during the course of sulfite reduction (24-30).
Kinetic Parameters-A series of Lineweaver-Burk plots of the initial velocities of NADPH oxidation at varying sulfite and NADPH concentrations is shown in Fig. 1. From these data, the V,,, of sulfite reduction, at "infinite" concentrations of both reactants, is 1850 NADPH per enzyme per min in 0.1 M potassium phosphate buffer, pH 7.7, at 23". The K, for sulfite, at infinite concentration of NADPH, is 4.3 PM. The K, for NADPH, at infinite concentration of sulfite, is 4.5 PM (Table  11). These values are somewhat lower than those reported previously (sulfite K, = 7 to 9 /JM, NADPH K, = 18 to 60 ).kM (5, 31)), but are considered more reliable, since the present Stoichiometry of NADPH-dependent sul$te reduction Reaction mixtures contained in a 3.0-ml total volume: 0.1 M potassium phosphate, pH 7.7; 200 GM NADPH; 80 PM NaHS03; 25 nM sulfite reductase; 10 units per ml of glucose oxidase; and 10 mM glucose. The mixtures were present in anaerobic cuvettes (stoppered with tight-fitting serum caps) of l-cm light path and reactions were initiated by injecting Nz-bubbled solutions containing all substrates with 5 ~1 each of glucose oxidase and sulfite reductase, in succession, with a period of 60 s between injections. NADPH oxidation was followed by the decrease in absorbance at 340 nm with a Cary model 14 spectrophotometer.
Absorbance readings were initiated approximately 10 s after injection of sulfite reductase, and the AA340 extrapolated back to time of injection. At each of the indicated times, the reaction was stopped by addition of the color-forming reagents used in determination of sulfide (Experiment 1) or sulfite (Experiment 2). The amount of sulfite in each reaction mixture was subtracted from the amount present in a control reaction mixture from which sulfite reductase was omitted.
The amount of sulfide in each reaction mixture was determined with reference to a control in which sulfite reductase was omitted.
There was negligible nonenzymatic disappearance of sulfite or production of sulfide during the time period of the assay. NADPH oxidation was also negligible in a control sample from which sulfite had been omitted. ----measurements were obtained with 5-and IO-cm light paths and a spectrophotometer with a 0 to 0.1 A slide wire, where necessary, to facilitate measurement with substrate concentrations in the 1 to 10 PM range. All previous data were obtained using l-cm light paths.
The fact that the reciprocal plots yield parallel lines is compatible with (but does not require) a catalytic mechanism in which the first reactant, presumably NADPH, converts enzyme to a reduced form which subsequently interacts with an oxidizing substrate to yield original enzyme and final product (32). This is compatible with the previously noted (6, 10) reduction of enzyme by NADPH in the absence of acceptor, as deduced from optical and EPR spectroscopy.
pH Optimum-When the velocity of NADPH oxidation was studied as a function of pH, using the standard assay concentrations of NADPH and sulfite, the optimum pH was 7.9 ( Fig.  2). Activities were identical in 0.1 M potassium phosphate and Tris-HCl buffers. Since the pK, for HS03 is 7.2 (33), the predominant sulfite species in solution at the optimal pH is soag-.
Electron 1. Lineweaver-Burk plot of NADPH-sulfite reductase activity as a function of sulfite concentration.
Reaction mixtures contained 0.1 M potassium phosphate (pH 7.7), 0.3 to 1.2 nM enzyme, and the indicated concentrations of NADPH and sulfite. Absorbance change was followed at 340 nm in a Cary model 14 spectrophotometer equipped with 0 to 0.1 A and 0 to 1.0 A slide wires, at 23" in cells of either 5-or lo-cm path length. The reference cuvette contained buffer in place of sulfite. Initial velocities (v) are expressed as moles of NADPH oxidized per mole of enzyme per min. The points at infinite concentration of NADPH were obtained from the intercepts of a l/v sers'sus l/(NADPH) plot, at several sulfite concentrations, of the same data plotted in the figure. Such plots also yielded a series of parallel lines. viologen at the same concentration of substrates. NADH, FMNH2, GSH, and reduced cytochrome c, all at 0.2 mM concentration, did not promote conversion of sulfite to sulfide with enzyme sufficient to allow detection of a reduction rate 1% of that found with NADPH as electron donor.

Othm Reactions Catalyzed
In addition to sulfite reduction, E. coli sulfite reductase is capable of catalyzing a number of other pyridine nucleotidedependent reduction reactions.

NADPH-Nitrite
and Hydroxylamine Redmtase-As reported previously (4), E. coli sulfite reductase catalyzes NADPH-dependent reduction of hydroxylamine and nitrite to ammonia. MVH can also serve as electron donor for reduction of these substrates, but we have not studied this reaction quantitatively. Lineweaver-Burk plots of the initial velocities of NADPH oxidation at varying nitrite and NADPH concentrations yield a series of parallel lines, as was observed with sulfite as acceptor. With hydroxylamine as electron acceptor, on the other hand, a series of converging lines is obtained; we have no ready explanation for this observation.
Kinetic parameters for the NADPH-nitrite and -hydroxylamine reduction reactions are presented in Table II. The Ir,,, values for both nitrite (3100 NADPH per enzyme per min) and hydroxylamine (13,700 NADPH per enzyme per min) reduction are greater than that observed with sulfite, but the K, values for these substrates (0.8 mM for nitrite, and 10 mM for hydroxylamine) are much higher than for sulfite (4.5 PM).
The pH optima for nitrite and hydroxylamine reduction, 8.6 and 9.5, respectively, are more alkaline than that for sulfite reduction (7, 9) (Fig. 2). As shown by Kemp et al. (5), it is unlikely that NADPH-sulfite reductase functions physiologically as a nitrite or hydroxylamine reductase.

NADPH-Diaphmase and MVH-NADP+
Reductase Actitities-Sulfite reductase catalyzes the transfer of electrons from NADPH to a wide variety of acceptors, including cytochrome c, ferricyanide, DCIP, menadione, and FMN. As shown in Table III, the rates of these diaphorase-type reactions, under standard assay conditions (0.2 mM NADPH and 0.1 mM acceptor), varied from 10,000 to 28,000 NADPH per enzyme per min. FAD TA~LIC III Reactions catalyzed by Escherichia coli sul$te reductase: e$ect of inhibitors Reactions were measured as described under "Materials and aerobic solution of enzyme plus NADPH was incubated with the Methods." Rates are expressed as a-electron equivalents trans-inhibitor and the reaction initiated by addition of electron acferred per enzyme per min. With NADP', p-CMPS, and fluoride ceptor. Activities are expressed relative to a control treated in &s inhibitors, enzyme was incubated in 0.1 M potassium phosphate parallel in which buffer replaced the inhibitor solutions. Incubabuffer (pH 7.7) containing the inhibitor for the period of time in-tion times: NADP+, cyanide, arsenite, fluoride, and 0.2 rnM dicated below, and the reaction initiated by addition of electron p-CMPS, 5 min; 1 pM p-CMPS, 60 min; CO, 30 min. acceptor and NADPH.
With cyanide, CO, and arsenite, an an-  a Incubation of enzyme with MVH for 30 min causes 90% inhibition of MVH-NADPH+ reductase activity (but not MVH-sulfite reductase activity).
Therefore, inhibition of this activity by CO, which requires prolonged incubation with CO in the presence of reductant, was not examined.
b Not examined because of high rate of MVH-NADP+ reductase activity catalyzed by sulfite reductase.
(0.1 mM) also served as an acceptor for the electrons of NADPH, with a velocity of 5,600 NADPH per min per enzyme. Kinetic studies of four of these diaphorase-type reactions are summarized in Table II. Each of the reactions studied, i.e. t,he NADPHdependent reductions of cytochrome c, ferricyanide, DCIP, and menadione, yielded a series of parallel lines in Lineweaver-Burk p1ot.s. Although K, values for NADPH and acceptor varied with the reaction studied, the V,,, values, at infinite concentrations of both NADPH and acceptor, were identical within experimental error for each of the four reactions (38,000 f 2,000 NADPH per enzyme per min).
These turnover numbers represent the highest observed for any of the reactions catalyzed by sulfite reductase, and are over 20 times as great as the V,,, for sulfite reduction with NADPH as electron donor. The results suggest the presence of a common rate-limiting step in the reductions of cytochrome c, ferricyanide, DCIP, and menadione. An additional rate-limiting step, considerably slower than that for reduction of diaphorase-type acceptors, must become operative in the reduct)ion of sulfite.
The enzyme also catalyzes another diaphorase-type reaction, the reduction of methyl viologen by NADPH.
Since the potential of MVH is considerably more negative than that of NADPH, we have followed the reverse reaction, i.e. the reduction of NADP+ by MVK.
The observed velocity of the latter reaction, 36,000 NADPH per enzyme per min (Table  III), is comparable to that of the other diaphorase activities of sulfite reductase.
NADPH Oxidase-Sulfite reductase can catalyze the oxygendependent oxidation of NADPH (Table III). At 0.2 mM NADPH and 0.25 mM 02, this reaction proceeds with a velocity of 75 NADPH per min per enzyme, i.e. 4% of the rate of the NADPH-sulfite reductase activity in the standard assay. No detailed studies of the NADPH oxidase activity have been performed.
At 0.2 mM concentration of each nucleotide, the reaction velocity is 9500 NADPH per min per enzyme. When velocity is plotted versus concent.ration of either substrate, at a fixed concentration of the other, the curve exhibits a maximum, indicating inhibition by excess substrate. Detailed kinetic analyses of the transhydrogenase reaction (as well as competitive inhibition by NADP+ (vi& infra)) are compatible with a common binding site for both oxidized and reduced pyridine nucleotides.

Inhibitors
By studying the effect of inhibitors on the various reactions catalyzed by E. co& sulfit,e reductase, we hoped to define more clearly those segments of the enzyme molecule with which different electron donors and acceptors can interact, and thereby tentatively deduce a sequence of electron flow within the sulfite reductase hemoflavoprotein molecule. CO and Cyanide-We first examined the catalytic effects of inhibitors which can be expected to react with the heme moiety, i.e. CO and KCN.
These compounds have been demonstrated (6) to react with both free and enzyme-bound sulfide reductase heme to form spectrally distinct complexes.
As reported (6), CO can complex only to reduced heme, while cyanide can be a ligand to either reduced or oxidized heme. However, since only the reduced enzyme is "accessible" to cyanide, the oxidized enzyme-cyanide complex can only be observed by first preparing the reduced enzyme-cyanide complex, and then permitting it to oxidize.
The enzyme-cyanide complex forms rapidly but apparently irreversibly (6) ; the enzyme-CO complex forms reversibly, but its rate of formation and dissociation is slow. The subsequent section will describe the correlation between the spectrophotometrically observed processes of enzyme hemeligand complex formation, and the catalytic events which are presumed to involve the heme.
When sulfite reductase was incubated with CO or cyanide in the presence of reductant, as described in Table III NADPHsulfite reductase reaction was initiated by addition of 0.1 ml of 5 mM NaHSOa (when NADPH was present in the preincubation mixture) or 0.1 ml of a solution containing 5 mM NaHSOa plus 2 mM NADPH (when NADPH was not nresene in the nreincubation mixture).
All'solutions were in 0. : a solution containing 20 nm enzyme, 0.2 mM NADPH, 0.5 mM CO, and 0.1 M potassium phosphate (pH 7.7), in a total volume of 0.9 ml, was incubated, in a cuvette of l-cm path length, under anaerobic conditions at 23'for the time indicated.
The NADPH-sulfite reductase reaction was then initiated by addition of 0.1 ml of 5 mM NaHS03 to the cuvette containing the enzyme-NADPH-CO solution.
Absorbance changes were followed at 340 nm with a Cary model 14 spectrophotometer at 23".
Inset, dependence of pseudo-first order rate constant for inhibition of NADPH-sulfite reductase activity upon CO concentration.
The kinetics of develonment of inhibition of sulfite reductase activity was measured as described above with each of the CO concentrations indicated.
hydrogenase reactions was significantly inhibited by either CO or cyanide.
The relationship between spectrophotometrically observable CO and cyanide binding to the heme prosthetic group and the inhibition of sulfite reductase activity was examined in detail as described below.
The following correlations were obtained.
1. CO and cyanide bind only to reduced heme, and inhibition of activity by these agents occurs only if enzyme is reduced prior to addition of sulfite. As shown in Fig. 3, when CO or cyanide was incubated with enzyme (either with or without sulfite) in the absence of a reducing agent. and the remaining reactant(s) were added to start the sulfite reductase reaction, no inhibition of sulfite reductase activity, as compared to controls, was detected.
However, when enzyme was incubated with NADPH plus either CO or cyanide, a progressive inhibition of sulfite reductase activity was observed.
2. The rate of CO and cyanide binding to the heme equals the rate of development of inhibition of sulfite reductase activity. The rate of CO binding to reduced enzyme was followed by recording at successive time intervals the absorption spectra of the enzyme plus NADPH plus CO solution.
As described previously (6), the AAs,,o-aso of the enzyme solution is a good measure of the amount of enzyme-CO complex formed, since the AA between these two wavelengths is negligible in both oxidized and reduced enzyme, while the formation of the CO complex is accompanied by greatly increased absorbance at 600 nm with little change at 560 nm. Aliquots of the reaction mixture were measured periodically for NADPH-sulfite reductase activity, while AA600-560 was monitored.
As shown in Fig. 4, the rates of formation of enzyme-CO complex, determined spectrophotometrically, and of loss of sulfite reductase activity, exhibited identical pseudo-first order kinetic patterns, with identical rate constants of 1.56 X 1O-3 s-l at 0.5 mM CO. This corresponds to a second-order rate constant of 3.1 M-I s-l for the reaction E + CO -+ E-CO, in agreement with that reported previously (6) for the formation of the E-CO complex. The rate of inhibition of sulfite reductase activity by CO was followed as a function of CO concentration.
As shown in the inset to Fig. 4, the pseudo-first order rate constants for this process were proportional to CO concentration, and a second order rate constant of 3.2 M-l s? could be obtained from the slope of this line. This value is in excellent agreement with that measured for formation of the enzyme-CO complex.* The rate of cyanide binding to reduced enzyme has not been studied previously.
A difference spectrum between reduced enzyme plus cyanide and reduced enzyme is shown in Fig. 5. A prominent maximum is observed at 411 nm. The timedependence for the development of this absorbance change was compared to that for development of cyanide inhibition. These data are shown in Fig. 6. The AAdu between the two solutions increased according to pseudo-first order kinetics.
If one assumes that the rate of the reactions E + cyanide -+ E-cyanide, determined spectrophotometrically at the cyanide concentration indicated in Fig. 6  Absorbance changes were followed at 340 nm with a Cary model 14 spectrophotometer at 23'. Inset, dependence of pseudo-first order rate constant for inhibition of NADPH-sulfite reductase activity upon KCN concentration.
The kinetics of development of inhibition of sulfite reductase activity was measured as described above with each of the KCN concentrations indicated.
(E) . (cyanide), then a second-order rate constant of 210 .M+ s-l can be calculated. The rate of development of inhibition of sulfite reductase activity by cyanide in the presence of NADPH was then compared to the rate of formation of the E-cyanide complex.
At each of the cyanide concentrations tested, the loss in activity followed pseudo-first order kinetics (Fig. 6). As shown in the inset to Fig. 6, the pseudo-first order rate constants for cyanide inhibition of sulfite reductase activity were proportional to cyanide concentration, yielding a value for the second order rate constant for the cyanide inhibition of sulfite reductase activity of 201 M-r 0, in good agreement with the value obtained from spectrophotometric measurements for formation of the E-cyanide complex.
3. The rate of dissociation of the enzyme-CO complex equals the rate of reappearance of sulfite reductase activity.
A solution of enzyme-CO complex was prepared as described in Fig. 7 and maintained at 4". Aliquots were examined at intervals over a 5-day period for content of enzyme-CO complex (AAc~~-w,) and for sulfite reductase activity.
The results are shown in Fig. 7. As CO dissociated from the complex to yield free, oxidized enzyme (6), sulfite reductase activity reappeared.
Both reactions followed first-order kinetics with the same rate constant: 1. formed by incubating 15 FM sulfite reductase anaerobically with 0.2 rnM NADPH and 0.5 mM CO at 23" for 1 hour. The l-ml solution was then passed through a column of Sephadex G-25 as described under "Materials and Methods," and the resulting enzyme, 5.2 PM by protein determination, was incubated at 4'. At the times indicated, absorption spectra of the solution were recorded (at 23") and aliquots assayed for NADPH-sulfite reductase activity.
A control sample of enzyme was treat,ed with 0.2 mM NADPH anaerobically, passed through the Sephadex G-25 column, and incubated at 4" in parallel with the enzyme-CO complex.
Aliquots of the latter solution were assayed for sulfite reductase activity each time the enzyme-CO solution was so assayed. The activity of the control enzyme solution decayed by only 10% during the entire period of incubation at 4". Fraction of enzyme-CO The activity after 105 hours of incubation at 4" (7Oyc of control enzyme) was taken as the t = 00 value. 10e5 s-l at 4". The total sulfite reductase activity recovered was 70% of that observed with a parallel solution of enzyme treated with NADPH alone, passed through an identical column of Sephadex G-25, and incubated along with the enzyme-CO complex at 4".
When a similar experiment was attempted with the enzymecyanide complex, which reoxidizes when NADPH is removed (6), there was no detectable dissociation of cyanide, as determined spectrophotometrically, nor return of sulfite reductase activity after 1 week of incubation at 4".
The results with CO and cyanide as inhibitors, then, strongly indicate involvement of the heme prosthetic group of sulfite reductase in the passage of electrons from NADPH or MVH to sulfite, nitrite, and hydroxylamine.
Since (Table III) CO and cyanide have little effect on reactions of enzyme with pyridine nucleotides, diaphorase acceptors, or oxygen, it may be concluded that the heme is probably not involved in the latter processes.
Arsenite-Arsenite has previously been reported to inhibit the NADPH-nitrite reductase activity of E. coli sulfite reductase (4). We have found that arsenite, like CO and cyanide, forms a spectrally detectable complex with the heme of sulfite reductase. As shown in Fig. 8 When the reduced complex was mixed with air and allowed to reoxidize, the spectrum returned to that of native, oxidized enzyme. Similarly, if the reduced enzyme-arsenite complex was passed through a column of Sephadex G-25 (aerobically) to remove NADP (H) and excess arsenite, the recovered enzyme was spectrally indistinguishable from free oxidized enzyme.
This result, together with data on inhibition of enzyme activity to be presented below, suggests that arsenite can form a stable complex only with reduced enzyme; the complex dissociates rapidly when the components required for its formation are removed.
With the concentrations of arsenite tested (1 and 10 mM), complex formation with NADPH-reduced enzyme, as measured spectrophotometrically, was complete within 10 s, the minimum time required to initiate measurement in the Cary model 14 spectrophotometer.
In the experiments to be described, preincubations of enzyme with arsenite or other components, or both, were routinely conducted at 23" for 1 min, but identical results were obtained when preincubations with arsenite were as short as 10 s or as long as 20 min. When sulfite was added to preformed reduced enzyme-arsenite complex (enzyme preincubated with arsenite plus NADPH), the initial rate of sulfite reduction was strongly inhibited ( Fig. 9, Curve A). However, the reaction rate progressively increased until a steady state constant rate of about TIME AFTER ADDITION OF LAST COMPONENT Irec) FIG. 8 (left). Spectra of sulfite reductase in presence of arsenite. The following additions were made to an anaerobic solution of 2.7 I.~M enzyme in 0.1 M potassium phosphate buffer (pH 7.7), and absorption spectra were recorded as soon as possible after addition of all components with a Cary model 14 spectrophotometer at 23" in cells of l-cm path length: A, no addition; B, 10 mM NaAsO, (superimposable upon A), p; C, 0.5 mM NADPH, ..*a ; D, 10 mM NaAsOz plus 0.5 rnM NADPH, -----. FIG. 9 (center). Effect of order of addition of components upon arsenite inhibition of NADPH-sulfite reductase activity. Reaction mixtures contained, in 1.0 ml total volume, 8 nM enzyme, 0.2 mM NADPH, 0.5 mM NaHS03, and 5 mnn NaAsOz where indicated. The indicated components were preincubated for 1 min in a volume of 0.9 ml. The final component(s) was then added in a volume of 0.1 ml and the absorbance change at 340 nm followed in a Cary model 14 spectrophotometer with respect to a reference solution which contained all components except sulfite. All solutions were in 0.1 M potassium phosphate (pH 7.7), cells were 1 cm in path length, and all operations were performed at 23". Each curve represents the average of three independent measurements. There was no significant difference in the control curves for A through E. Reaction mixtures contained 0.1 M potassium phosphate (pH 7.7), 9 nM enzyme, 0.2 mM NADPH, and the indicated concentrations of NaAsOz and sulfite. Reactions were initiated by the addition of sulfite as the final component.
Absorbance change was followed at 340 nm in a Cary model 14 spectrophotometer at 23" in cells of l-cm path length. Rates (v) were determined from the linear portion of the progress curve, following a short, (<l min in all crtses) initial lag period (see Fig. 9A).
Data are plotted as l/v (min per Asa) versus l/(sulfite).
The Ki determined from the data is 0.17 mM.
40% of the uninhibited rate was achieved. In contrast, if the arsenite complex was not preformed (i.e. the following combinations: (a) arsenite plus enzyme preincubated, followed by addition of NADPH + sulfite; (b) arsenite plus sulfite plus enzyme preincubated, followed by addition of NADPH; (c) NADPH plus sulfite plus enzyme preincubated, followed by addition of arsenite; or (d) arsenite plus sulfite plus NADPH preincubated, followed by addition of enzyme), then the initial velocity was not inhibited.
However, over a period of approximately 1 min, the reaction velocities progressively decreased, reaching a steady state constant rate of, again, about 40 to 45% of the uninhibited control rate. Thus, the steady state level of arsenite inhibition is independent of the order of addition of reagents, even though the initial velocities are strongly dependent upon the operation sequence.
When the steady state rate of sulfite reduction was examined as a function of arsenite and sulfite concentration (Fig. lo), arsenite was found to behave as a competitive inhibitor (with respect to sulfite) of the NADPH-sulfite reductase reaction. The Ki for arsenite is 0.17 mM. This result indicates that the enzymearsenite complex responsible for inhibition of enzyme activity must be a reversible one. In keeping with this conclusion, when the arsenite-NADPH-enzyme solution, the spectrum of which is shown in Fig. 8, was passed through a Sephadex G-25 column to remove pyridine nucleotide and excess arsenite, the resulting enzyme was 95 y0 as active as untreated enzyme.
The results of the order of addition experiments (Fig. 9) indicate that the "reduced state" of enzyme is required for formation of an inhibitory complex with arsenite. This correlates with the requirement for NADPH for formation of a spectrophotometrically detectable enzyme-arsenite complex (Fig. 8). The progressive relief of inhibition observed under catalytic conditions (Fig. 9, Curve A) is consistent with the previously-noted reversibility of the spectrophotometrically observable complex.
The steady state level of arsenite inhibition during catalysis, as modified by arsenite and sulfite concentration (Fig. lo), could be expected to represent a complex function of: (a) the relative rates of binding of arsenite and sulfite to reduced enzyme; (5) the relative rates of "release" of enzyme from its complexes via dissociation (arsenite and sulfite) or turnover (sulfite), or both; (c) the steady state oxidation-reduction level of the enzyme heme.
The pattern of arsenite inhibition of the various reactions catalyzed by sulfite reductase is shown in Table III. The pattern is identical to that shown by CO and cyanide as inhibitors, in that sulfite, nitrite, and hydroxylamine reduction are inhibited, while the other NADPH-and MVH-dependent reduction reactions are not. The data again strongly suggest reaction of arsenite with the heme of sulfite reductase, and strengthen the conclusion that the heme prosthetic group is required for electron transfer to sulfite, nitrite, and hydroxylamine.
Mercurial: p-CMPX-Treatment of the enzyme with 1 pM p-CMPS causes release of the FMN prosthetic group while the heme and FAD moieties remain enzyme-bound and apparently functional.
This conclusion is based upon the following. 1. p-CMPS treatment causes the enzyme solution to become markedly fluorescent.
This fluorescence is unpolarized, and is due to flavin, since it exhibits activation maxima at 268, 376, and 448 nm, and a single emission maximum at 532 nm. The intensity of this fluorescence was 90% of that observed upon boiling an equivalent amount of enzyme, a procedure which releases both the FMN and FAD prosthetic groups (6). Since Fluorescence measurements were made with a Turner model 210 spectrophotofluorometer, with an excitation wavelength of 450 nm (bandwidth 10 nm) and an emission wavelength of 535 nm (band width 25 nm), using a chart recorder.
Enzyme assays were performed as described under "Materials and Methods." free FMN has an intrinsic fluorescence approximately 10 times that of FAD under the experimental conditions used (20), it is apparent that mercurial treatment must have caused the release of at least 90% of the enzymic FMN.
The appearance of flavin fluorescence with 1 ).LM p-CMPS is first order, with a half-time at 23" of 2 to 3 min (Fig. 11). Titration of 20 nm enzyme with p-CMPS (overnight incubation at 4") showed that appearance of maximal flavin fluorescence was achieved with 0.4 PM p-CMPS, i.e. 20 moles p-CMPS per mole enzyme (Fig. 12).
2. Four hundred milliliters of 20 nM enzyme were treated with 1 pM p-CMPS, and then concentrated loo-fold by ultrafiltration. Filtrate and concentrate were analyzed for FMN and FAD by the procedure of Faeder and Siegel (20). The filtrate contained 78 nM FMN (3.9 moles of FMN per mole of original enzyme) and 4 nM FAD (0.2 mole of FAD per mole of enzyme).
The concentrated enzyme, 1.5 PM on the basis of protein content, contained 0.3 PM FMN (0.2 FMN per enzyme) and 5.7 PM FAD (3.8 FAD per enzyme).
Thus, mercurial treatment results in release of at least 95% of the enzyme FMN, while permitting retention of about 95% of the enzyme FAD.
3. The absorption spectrum of the p-CMPS-treated enzyme after ultrafiltration, when compared to that of native enzyme (Fig. 13), has diminished absorbance in the region 340 to 540 nm, as would be expected from loss of half of its flavin: the AAds between native and p-CMPS-treated enzyme corresponds to 3.5 FMN per mole, assuming the ~4~~ of 12.2 X lo3 M-' cm-l (21) of free FMN.
There is no change in the spectrum in the 540 to 750 nm regions, indicating no effect of mercurial treatment of the heme prosthetic group (Fig. 13).
4. The FAD bound to the p-CMPS-treated, FMN-free enzyme, remains reducible by NADPH, as shown in the data of Fig. 13  the A~450 of 10.3 X lo3 M-' cm-l for oxidized minus reduced FAD (21). However, NADPH, despite its ability to reduce the FAD, can no longer reduce the heme of mercurial-treated enzyme (Figs. 13 and 14) under conditions which permit reduction of a substantial portion of the native enzyme's heme (6). The heme remains reducible by dithionite (Fig. 13). NADPHreducibility of the heme of mercurial-treated enzyme can be restored by addition of 10 pM FMN (5 FMN per enzyme in the experiment of Fig. 14). Thus, we can conclude that FMN is required for internal electron flow, since mercurial treatment of sulfite reductase, which removes the FMN prosthetic groups, leaves both the FAD and heme moieties functionally intact, but interrupts electron flow between them.
The catalytic consequences of this p-CMPS-induced interruption of electron flow between FAD and heme are shown in Table III.
The following reactions are relatively unaffected by p-CMPS: (a) reduction of sulfite by MVH, a result consistent with the lack of apparent mercurial effect on the heme prosthetic group; (b) transfer of electrons from NADPH to AcPyADP+, suggesting that the primary site of pyridine nucleotide interaction with enzyme is unaffected (and is therefore probably FAD); and (c) reduction of FMN by NADPH, a result consistent with the ability of FMN to interact with mercurial-treated enzyme and reverse the p-CMPS effecL3 8 It should be noted that p-CMPS at much higher concentration (0.2 mM) does inhibit the NADPH-AcPyADP+ transhydrogenase and NADPH-FMN reductase reactions. The inhibitions are not reversible by added FMN. However, even at this high concentration of mercurial, the MVH-sulfite reductase activity remains unaffected (Table III) FIG. 13. Absorption spectra of p-CMPS-treated enzyme. To 200 ml of 40 nM sulfite reductase were added 200 ml of 2 pM p-CMPS. All solutions were in 0.1 M potassium phosphate (pH 7.7). The mixture was incubated for 30 min at 23", then concentrated at 4" to a final volume of 4 ml with an Amicon ultrafiltration apparatus. Flavin analysis of the filtrate showed 78 nM FMN (3.9 FMN per enzyme) and 4 nM FAD (0.2 FAD per enzyme) released from the enzyme. The concentrated enzyme solution was centrifuged for 60 min at 40,000 X 8. A 0.5-ml aliquot was assayed for protein concentration.
Flavin analysis of the concentrated enzyme, 1.5 PM on the basis of the protein determinat.ion, revealed 5.7 I.~M FAD (3.8 FAD per enzyme) and 0.3 PM FMN (0.2 FMN per enzyme). A l.O-ml aliquot of the concentrated enzyme solution was placed in a Thunberg cuvette (with 10 ~1 of 50 mM NADPH in the sidearm) and the solution rendered anaerobic by repeated evacuation and flushing with Ns. An absorption spectrum of the enzyme following anaerobiosis was recorded (B). NADPH (final concentration, 0.5 mM) was then tipped in, and the spectrum of the reduced p-CMPS-treated enzyme recorded (C). Following this, the solution was opened to air, a few crystals of sodium dithionite added, and the spectrum quickly recorded (D). An absorption spectrum of native sulfite reductase, at the same protein concentration, is shown for comparison (AA).
In contrast, the following reactions were strongly inhibited by p-CMPS: (a) NADPH-diaphorase-type reactions, including cytochrome c reductase; (b) NADPH oxidase; (c) the NADPH-(but not MVH-) dependent reductions of sulfite, nitrite, and hydroxylamine; and (d) the reduction of NADP+ by MVH. As shown in Figs. 11 and 12, the inhibition of these activities by p-CMPS shows a dependence on both time and mercurial concentration which parallels the development of flavin fluorescence on these parameters.
The mercurial-induced inhibition of these pyridine nucleotide-dependent reactions can be reversed by addition of 50 PM FMN, as shown in Table IV.4 FAD at this concentration has but little effect.
The concentration-dependence of the FMN effect in reversing mercurial inhibition of two NADPH-dependent reactions is shown in Fig. 15. FMN, 3 to 5 PM, gave 50% of maximal stimulation in both cases. This number, 'contrasted to the dissociation constant of 0.01 PM observed by Siegel et al. (34) for the dissociation of FMN from native sulfite reductase, indicates that the action of 1 PM p-CMPS may be considered as report of Asada et al. (39) that the mercurial p-chloromercuribenzoate at such concentrations can inhibit the MVH-sulfite reductase of spinach.
4 The inhibition of certain NADPH-dependent reduction reactions observed at high concentrations of FMN ( Fig. 15 and Table  IV) could be accounted for by competition between acceptor and FMN for the electrons of NADPH.
Thus, the absolute rate of absorbance change at 340 nm is unaffected or increased by FMN, but the apparent rate of reduction of acceptor (measured as the difference in the AAM per min between a complete reaction mixture and one from which acceptor, but not FMN, has been omitted) is inhibited.
It should be noted that the K, for FMN in the NADPH-FMN reductase reaction (at 0.2 rnM NADPH) is 20~~.

WAVELENGTH (nm)
FIG. 14. Reducibility of p-CMPS-treated enzyme by NADPH: effect of exogenous FMN. p-CMPS-treated concentrated enzyme was prepared by a procedure similar to that described in Fig. 13. The resulting concentrated enzyme, in this experiment, was 2.1 pM and contained 3.9 FAD and 0.3 FMN per enzyme by fluorimetric analysis. One-milliliter aliquots of the p-CMPS-treated enzyme were placed in two modified Thunberg cuvettes fitted with rubber serum caps and the solutions made anaerobic by repeated evacuation and flushing with Nz. Ten microliters of 50 mM NADPH were then added to the cuvette in the reference compartment of the Cary model 14 spectrophotometer, and 10 ~1 of anaerobic 0.1 M potassium phosphate (PH 7.7) buffer to the sample cuvette. The difference spectrum between the two cuvettes was recorded (-----), using the 0 to 0.1 and 0.1 to 0.2 A slide wires of the Cary spectrophotometer. Following this, 10 ~1 of an anaerobic solution of 1 rnM FMN were added to both cuvettes and the difference spectrum again recorded (-). effectively decreasing the affinity of the enzyme for its FMN prosthetic group by two to three orders of magnitude.
We may conclude, then, that the FMN moiety of sulfite reductase is required for electron transfer from NADPH and reduced FAD to the heme prosthetic group (and thence to those electron acceptors dependent upon the heme for reduction, i.e. sulfite, nitrite, and hydroxylamine), as well as to diaphorasetype acceptors and 02. The FMN moiety is not required for interaction of either pyridine nucleotides or sulfite with the enzyme. NADP+-As seen in Table III, NADP+ inhibits all NADPHdependent reactions catalyzed by sulfite reductase. (Since the enzyme catalyzes a very rapid MVH-NADP+ reductase reaction, the effect of NADPf on the MVH-sulfite reductase reaction could not be tested.) When the steady state kinetics of the NADPH-sulfite and cytochrome c reductase reactions were studied as a function of NADP+ and NADPH concentration, NADP+ was found to be a competitive inhibitor with respect to NADPH in both reactions, with a K; for NADP+ equal to the K, for NADPH in each reaction (Table II). Inhibition by NADPf, on the other hand, was noncompetitive with respect to sulfite or cytochrome c. Similar kinetic studies have been performed with NADP+ as an inhibitor of the NADPH-AcPyADP+ transhydrogenase reaction. In this case, NADP+ inhibition was competitive with respect to both substrates. Sulfite reductase, 20 nM, was incubated with 1 J.LM p-CMPS in 0.1 M potassium phosphate (pH 7.7) for 20 hours at 4' ("p-CMPStreated enzyme").
A parallel sample of enzyme was incubated with buffer alone ("untreated enzyme"). Aliquots of each enzyme solution were then assayed for the activities indicated below in reaction mixtures containing either no added flavin, 50 PM FMN, or 50 PM FAD.
NADPH-sulfite, nitrite, hydroxylamine, ferricyanide, menadione, and MVH-NADP+ reductase activities were corrected for a blank containing enzyme, NADPH or MVH, and the indicated flavin, but no other electron acceptor.
NADPH-ferricyanide reductase activity was also corrected for nonenzymatic oxidation of NADPH.
NADPH-AcPyADP+, cytochrome c, and DCIP reductase activities were corrected for a blank containing all components except enzyme. Details of the aesay procedures are given under "Materials and Methods." Since both FAD and FMN, at 50 FM can serve as electron acceptors for the oxidation of NADPH catalyzed by the enzyme, these flavins, by their competition for electrons, can function m apparent "inhibitors" of the reduction of other acceptors by NADPH.
With untreated enzyme, the following percentage activities for enzyme plus 50 fbM, FMN relative to enzyme without added flavin were noted: NADPH-sulfite, 44%; -nitrite, 55%; -hydroxylamine, 75%; -ferricyanide, 88%; -menadione, 86%; -AcPyADP+, 94%; -DCIP, 106%; -cytochrome c, 112%; MVH-NADP+, 98%. Similar effects were noted when 50 PM FAD was substituted for FMN. In the data below, all activities are presented relative to that of the untreated enzyme assay in the presence of the indicated flavin. &&s--In addition to the relatively specific inhibitors cited above, a number of anions (as sodium salts) were found to inhibit the NADPH-sulfite reductase activity (Table V). In all cases, inhibition occurred rapidly, without requiring prereduction of enzyme, since it made no difference in the final activity observed whether enzyme was: (a) preincubated with NADPH or sulfite, together with salt, for 1 min prior to addition of the second reactant; or (b) preincubated with salt for 1 min prior to addition of an NADPH-sulfite mixture; or (c) added to a salt, plus NADPH plus sulfite mixture with no preincubation. The kinetics of inhibition of the NADPH-sulfite reductase activity were noncompetitive with respect to sulfite for each of the salts. can be arranged oppositely in the two reactions.
The inhibition of NADPH-cytochrome c reductase activity by the same salts follows the series: SCN-> I-> NO, > Tfr-> Cl-> SOd2-> F-. The markedly different effect of salts on the two types of reactions emphasizes again the fundamental difference between the diaphorase class and the "sulfite reductase" class of reacbions catalyzed by this enzyme.
Complex with Sulfite The experiments previously described combine to suggest strongly that the heme is involved in electron transfer to sulfite. This suggestion is supported by the spectral data of Fig. 16. This figure demonst,rates the appearance of a new species, with a visible wavelength maximum at 585 nm, and a broad Soret band in the 395 to 410 nm region, when excess sulfite is added to NADPH-reduced enzyme.5 This species persists after all reductant has been consumed and after enzyme flavins have become reoxidized.
However, as in the case of cyanide, CO, and arsenite, sulfite has no effect on the spectrum of oxidized sulfite reductase.
We have no evidence which designates the oxidation state of the added sulfite when it is in the complex shown in Fig. 16, but, for convenience, we refer to this bound form as the enzyme-sulfite complex.
This complex is quite stable in the absence of enzymatic turnover.
No alteration in the spectrum of the sulfite complex is observed following chromatography of the reoxidized enzyme-NADP(H)-sulfite solution of Fig. 16 on a column of Sephadex G-25. The g = 6 EPR signal of the heme of native oxidized enzyme (6, 10) disappears following NADPH-sulfite treatment, and is not regenerated by Sephadex chromatography. The EPR spectrum of oxidized enzyme (like the absorption spectrum) is unaffected by sulfite unless NADPH is added. To ascertain whether the spectrophotometrically observed enzyme-sulfite complex is associated with actual physical binding of sulfite sulfur to the enzyme, several preliminary experiments with [a5S]sulfite have been performed.
Enzyme (1 pM) was incubated for 5 min at room temperature with 0.1 mM NADPH and 0.1 mM Naza5S03. The incubation mixture was then passed over an aerobic column of Sephadex G-25 and the eluted fractions assayed for protein and radioactivity.
The radioactivity associated with the chromatographed protein amounted to 1.2 and 1.5 a5S per enzyme molecule in replicate experiments.
When NADPH concentration was increased to 0.5 mM, or decreased to 2 PM, the a5S bound to the enzyme decreased 3-to 4-fold.
When NADPH was totally omitted in an analogous experiment, less than 0.1 mole of 35S was retained per mole of enzyme. We do not consider that the amount of sulfite bound in these experiments necessarily represent saturation of sulfite binding sites, and experiments designed to establish the number of sites are currently in progress.

DISCUSSION
The results reported in this paper suggest the minimum linear scheme of electron flow within the sulfite reductase molecule shown in Scheme 1 (NADP+, p-CMPS, CO, CN-, and AsOzact as inhibitors).
The dotted arrow between FMN and heme indicates that the mechanism of electron flow from flavin to heme is not clear, since no role for the non-heme iron prosthetic groups of the enzyme has as yet been identified.
In this scheme, FAD and FMN prosthetic groups serve distinctly different roles in the process of electron transfer. Thus, the FAD serves as the entry port for electrons of NADPH, while the FMN serves as a transmitter of electrons between the reduced FAD and either the heme prosthetic group or artificial electron carriers such as the diaphorase-type acceptors (in this case, cytochrome c, ferricyanide, DCIP, menadione, or externally added flavins), 02, or methyl viologen. The sequence of the FAD and FMN prosthetic groups in the electron transfer process is suggested by the results with p-CMPS-treated enzyme. Such treatment apparently leads to an effective decrease in the affinity of the enzyme for FMN by at least two orders of magnitude (34) and therefore results in dissociation of this prosthetic group when enzyme is maintained in the 10e8 M concentration range. The catalytic effects of the mercurial treatment can be reversed by addition of external FMN in the 10Fe to 10e6 M concentration range. With p-CMPS-treated enzyme, freed of its FMN by ultrafiltration, the FAD moiety is reducible by NADPH and the transfer of electrons between pyridine nucleotides is unaffected. However, addition of FMN is required for transfer of electrons from NADPH to the diaphorase acceptor9 and 02, and for 6 The maintenance of a residual activity of 10 to 20% that of native enzyme activity for the NADPH-dependent reductions of transfer, via the heme, to sulfite, nitrite, and hydroxylamine. The immediate physiological acceptor for the electrons of reduced FMN has not been established.
We have not demonstrated that electron flow from reduced FMN to heme is direct, and the previously mentioned possibility that the non-heme iron-labile sulfide groups may be involved must be considered?
With MVH as donor, electron transfer to NADP+ is inhibited in the p-CMPS-treated enzyme and can be restored by addition of FMN.
However, reduction of sulfite by MVH is unaffected. Thus, while FMN is required for "reverse" electron flow from MVH to pyridine nucleotides, it is clearly not required when MVH serves as electron source for reduction of sulfit,e-type acceptors.
The ability of mercurial-treated enzyme to catalyze MVH-sulfite (but not MVH-NADP+) reduction suggests that MVH can interact with the enzyme at a site "after" FMN. However, the inability of heme-binding agents to inhibit the MVH-NADPf reductase activity demonstrates that MVH can interact with a group other than the heme. The data are compatible with an ability of MVH to reduce both the heme and FMN groups (as indicated in Scheme 1). They are also compatible with a carrier between FMN and heme (perhaps non-heme iron?) as a site of MVH reduction.
When CO, cyanide, or arsenite was added to the enzyme, a striking parallelism was observed between spectrophotometric evidence of heme complex formation with these ligands, and catalytic evidence of inhibition of reduction of sulfite-type acceptors. This parallelism, both in extent and in rate of complex formation and inhibition, compels the conclusion that the heme prosthetic group is required for electron transfer from either MVH or NADPH to sulfite, nitrite, and hydroxylamine. Furthermore, sulfite perturbs the spectrum of the reduced enzyme in a manner characteristic of formation of a complex with the heme; under such conditions, binding of [asS]sulfite to the enzyme, not reversible by dialysis, occurs. With CO, cyanide, and arsenite, binding to enzyme requires the presence of a reductant.
Since the cyanide ligand (and possibly the sulfitederived sulfur as well) remains bound to the enzyme following its reoxidation, and indeed cyanide binds to the oxidized free heme (6), the requirement for reductant may reflect conditions needed for accessibility of enzyme heme to ligands, rather than DCIP, ferricyanide, and menadione in the p-CMPS-treated en- Thus, it may become possible in the future to define more clearly the role of these groups in the catalytic mechanism.
a necessarily higher affinity of ferro-versus ferriheme for these agents.
The ability of sulfite to form a nondissociable complex with the reduced enzyme is consistent with the observed ability of sulfite to halt rapidly the development of inhibition by agents such as cyanide and CO, although sulfite cannot reverse such inhibition once an enzyme-cyanide or -CO complex has formed.
The nature of the enzyme-sulfite complex is of particular interest.
Since we have detected formation of the complex only in the presence of reductant and excess sulfite, the oxidation-reduction state of the heme iron and the sulfite cannot be precisely defined. The complex may contain sulfite itself or a reduced sulfur moiety "trapped" by exhaustion of reducing power at an oxidation state intermediate between sulfite and sulfide. Barring unexpected complexities, one might anticipate that the amount of enzyme-bound 35S diminishes when a stoichiometric excess (>3 NADPH per sulfite) of reductant is added, since sulfide would then be released.
Preliminary experiments suggest that this is indeed the case, and this may provide a clue as to why the enzyme is not usually isolated as the sulfite complex.
The sequence of electron flow indicated in Scheme 1 is supported not only by the evidence cited herein, but by additional bodies of data. Thus, the scheme is similar to that proposed by  for the yeast sulfite reductase, an enzyme which, like the E. coli enzyme, is a high molecular weight hemoflavoprotein.
The suggested role of the heme chromophore in the process of sulfite reduction per se is further strengthened by its presence in all sulfite reductases examined to date, including many which contain no flavin, such as those from higher plants (38, 39), yeast and Salmonella mutants (36,40), and the sulfate-reducing bacteria (24-26).
Genetic studies by  with Salmonella typhimurium mutants have led to the conclusion that a single gene product (that of the cys J gene in Salmonella) which can be isolated as an iron-free flavoprotein, is responsible for all of the NADPH-diaphorase reactions of enterobacterial sulfite reductase, while the products of two different genes (termed cys G and I) are required for MVH-sulfite reductase activity. Recombination of appropriate mutant extracts (e.g. cys J extract with either cys G or cys I extract) leads to reconstitution of NADPH-sulfite reductase activity in U&O. The reduction of sulfite to sulfide in E. coli also requires the products of three genes (termed qs G, P, and Q) which map in positions analogous to those of the cys G, I, and J genes on the Salmonella chromosome (45). Thus, we may conclude that NADPH and sulfite interact with the sulfite reduct,ase hemoflavoprotein at entirely different sites on the enzyme molecule, and in fact probably 'at sites located on different polypeptide chains. The physical distinction (Scheme 1) between the processes of NADPH-diaphorase activity and MVH-sulfite reductase activity is supported by the widely different effects of salts on the two types of catalytic activity.
It is of interest to note that the effect of the salts tested on the over-all enzyme activity, NADPH-sulfite reduction, can be explained (as a first approximation) as the resultant of the combined effects (Table V) of each salt on the NADPHdiaphorase and MVH-sulfite reductase activities. Thus, these two processes can be considered to occur more or less independently of one another on the intact enzyme.
Why is E. coli sulfite reductase so complex in structure? A priori, one might expect that this complexity may reflect the catalytic requirements of a complex reaction, the 6-electron reduction of sulfite to sulfide. The electron transport chain could thus serve as a storage device for electrons.
And yet, the actual reduction of sulfite to sulfide, in ot,her enzymes, can be done with smaller hemoprotcin molecules (36, 3%40), using external electron sources, either artificial (MVH) or natural (as yet incompletely described, but quite possibly involving ferredoxins (46)). Certainly, the minimum catalytic requirements for sulfite reduction in E. coli include two: the highly specific sulfite reductase heme, siroheme; and, for the pyridine nucleotide-mediated reduction (as distinct from MVH), a device for "stepping-down" a a-electron donor such as NADPH to a presumed l-electron acceptor, the heme. It is clear that much of the observed complexity of the B. cola' enzyme structure is due to the latter requirement rather than the former.
Oxidation-reduction enzymes containing flavin prosthetic groups are widespread in nature. The step-down catalytic function of such flavoproteins can be achieved in a variety of ways: (a) the flavin may act independently (e.g. NADPHcytochrome bg reductase (47), NADl'+-ferredoxin oxidoreductase (48)) ; (b) flavins may act in concert and be functionally indistinguishable (as in the mechanism proposed for microsomal NADPH-cytochrome c reductasc by Kamin et al. (49)) ; or (c) fiavins may serve functionally distinct roles but act cooperatively. Enzymes which cont,ain both FMN and FAD could logically be expected to fall into the latter class, but the class need not by any means be restricted to these (e.g. the nonequivalence of the two FAD moieties of xanthine oxidase suggested by the results of Kanda and Rajagopa1a.n (50)).
Existing st,udies with FADFMN enzymes do indeed suggest the possibility of different roles. Iyanagi and Mason (51) have isolated a form of liver NADPH-cytochromc c reductase which appears to contain both FAD and FMN rather than just FAD, as described in other laboratories (52-57).* These workers suggest that one of these groups (but not yet identifiable as FAD or FMN) may serve uniquely as the initial electron acceptor. Rajagopalan and his colleagues have informed us9 that the FMN of dihydroorotic dehydrogenase appears to be required for electron transfers involving the pyrimidine, while the FAD is required for NAD+-dependent reactions. Heterogeneity in the flavin functions in dihydroorotic dehydrogenase has previously been suggested by t'he EPR studies of Aleman et al. (58).
The sulfite reductase described in bhis study has propert.ies which make it unusually suitable for elucidation of the specific roles of individual flavin species in an enzyme which contains multiple flavins. This is possible because its FMN prosthet,ic group dissociates more readily than bhe FAD prosthetic group, and because the FMN can be specifically removed by treatment of the enzyme with the mercurial p-CMPS.
Thus, FMN-free enzyme, containing a full array of its other prost,hetic groups (including FAD), and apparently fully competent catalytically upon readdition of FMN, can be prepared. The data in this paper indicate that FAD serves specifically as the entry port for NADPH electrons in this enzyme and the FMN serves to transmit these further along the elect,ron transport chain. Thus, the flavins may operate "in series." Additional studies by Siegel et al. (34,43) have led to a proposed mechanism in 8 B. S. S. Masters and H. Kamin, using the technique of Faeder and Siegel (20) have recent,ly re-examined the flavk content of both pig and rat liver microsomal NADPH-cvtochrome c reductase. -Their results confirm those of Iyanagi and Mason (51) and show that these enzymes contain approximately equimolar quantities of FAD and FMN. 9 M. Kanda and K. V. Rajagopalan, Department of Biochemistry, Duke University, Durham, N.C., personal communication.
which the oxidation-reduction cycles of the FAD and FMN cooperate in such a fashion so as to convert "input" electron pairs from NADPH into "output" single electrons at constant potential.
Aclcnowledgments-The authors are indebted to Drs. E. Phares and G. D. Novelli of Oak Ridge National Laboratory for kindly providing the E. coli B cells used for purification of sulfite reductase.