Gene Overexpression, Purification, and Identification of a ...

3 downloads 2297 Views 404KB Size Report
The oxidation of dibenzothiophene to dibenzothiophene sulfone has been linked to the enzyme encoded by the sox/dszC gene from Rhodococcus sp. strain ...
JOURNAL OF BACTERIOLOGY, Oct. 1996, p. 5699–5705 0021-9193/96/$04.0010 Copyright q 1996, American Society for Microbiology

Vol. 178, No. 19

Gene Overexpression, Purification, and Identification of a Desulfurization Enzyme from Rhodococcus sp. Strain IGTS8 as a Sulfide/Sulfoxide Monooxygenase BENFANG LEI1

AND

SHIAO-CHUN TU1,2*

Departments of Biochemical and Biophysical Sciences1 and Chemistry,2 University of Houston, Houston, Texas 77204-5934 Received 7 May 1996/Accepted 2 August 1996

The oxidation of dibenzothiophene to dibenzothiophene sulfone has been linked to the enzyme encoded by the sox/dszC gene from Rhodococcus sp. strain IGTS8 (S. A. Denome, C. Oldfield, L. J. Nash, and K. D. Young, J. Bacteriol. 176:6707–6717, 1994; C. S. Piddington, B. R. Kovacevich, and J. Rambosek, Appl. Environ. Microbiol. 61:468–475, 1995). However, this enzyme has not been characterized, and the type of its catalytic activity remains unclassified. In this work, the sox/dszC gene was overexpressed in Escherichia coli, a procedure for the purification of the expressed enzyme was developed, and the properties of and the reactions catalyzed by the purified enzyme were characterized. This enzyme binds one flavin mononucleotide (Kd, 7 mM) or reduced flavin mononucleotide (FMNH2) (Kd < 1028 M) per 90,200-Da homodimer, and FMNH2 is an essential cosubstrate for its activity. Patterns of product formation were examined under different FMNH2 availabilities, and results indicate that this enzyme catalyzes a stepwise conversion of dibenzothiophene to the corresponding sulfoxide and subsequently to the sulfone. On the basis of isotope labeling patterns with H218O and 18O2, dibenzothiophene sulfoxide and sulfone obtained their oxygen atom(s) from molecular oxygen rather than water in their formation from dibenzothiophene. The enzyme also utilizes benzyl sulfide and benzyl sulfoxide as substrates. Hence, it is identified as a sulfide/sulfoxide monooxygenase. This monooxygenase is similar to the microsomal flavin-containing monooxygenase but is unique among microbial flavomonooxygenases in its ability to catalyze two consecutive monooxygenation reactions. Organic sulfur in fossil fuels has long been a major cause of environmental pollution. Substantial efforts by industrial and academic research communities have been directed toward the development of biotechnology for desulfurization of petroleum and coal. Dibenzothiophene (DBT) has been widely used as a model compound to screen microorganisms which might be used in desulfurization of fossil fuels. A number of microorganisms (1, 3, 13, 14, 22) have been found to metabolize DBT. More recently, Rhodococcus sp. strain IGTS8 (5, 12), Rhodococcus erythropolis D-1 (9), and Rhodococcus sp. strain SY-1 (25, 26) have been isolated and shown to convert the sulfur of DBT to water-soluble sulfate while keeping the organic moiety mostly unmodified. The sulfur-specific reactions make these strains most promising contenders in industrial applications. Two groups (4, 27) have cloned the three genes, sox/dszABC, encoding the Rhodococcus sp. strain IGTS8 desulfurization enzymes (EA,B,C). As shown in Fig. 1, the enzyme (EC) encoded by sox/dszC catalyzes the oxidation of DBT to DBT sulfone (DBTO2), which is subsequently converted into 2-hydroxy-biphenyl and sulfate by the enzymes EA and EB encoded by sox/dszAB. Thus far, as shown by the dashed arrows in Fig. 1, it is not clear whether DBT is directly oxidized to DBTO2 or converted stepwise to DBTO2 with DBT sulfoxide (DBTO) as an intermediate (4). The question of whether EC is an oxidase or a mono- or dioxygenase also remains unresolved. Although the genes encoding EA,B,C have been cloned, very little is known about the properties or cofactor-effector requirements of these enzymes. The extract of either IGTS8 or Escherichia coli expressing EA,B,C shows only low-level activi-

ties in vitro. Apparently, some critical factor(s) in addition to EA,B,C is required for the overall desulfurization processes. In fact, the in vitro desulfurization activities have been found to be enhanced by a reduced flavin generation system (24, 32). However, neither the enzyme(s) affected by this treatment nor the action mechanism of reduced flavin has been elucidated. These findings highlight the critical need for further enzymological characterization of EA,B,C. In this report, we describe the overexpression and purification of the cloned EC, now named sulfide/sulfoxide monooxygenase. Furthermore, the purified enzyme was characterized with respect to subunit structure, flavin binding, substrates and products, and mode of oxygenation. MATERIALS AND METHODS Materials. Flavin mononucleotide (FMN), flavin adenine dinucleotide (FAD), riboflavin, NAD(P)H, and catalase were from Sigma. Benzyl sulfide, benzyl sulfoxide, benzyl sulfone, DBT, DBTO, and DBTO2 were purchased from Aldrich. E. coli BL21 (DE3) and expression vector pET-21b were from Novagen. Rhodococcus sp. strain IGTS8 was from the American Type Culture Collection (ATCC 53968). Oxygen-18-labeled water (10% labeling) and molecular oxygen (95 to 98% labeling) were obtained from Cambridge Isotope Laboratories. Glucose oxidase was from Fluka. Procedures in the literature were followed to synthesize cuprous perchlorate, CuClO4 (7), and to purify FAD (20). All phosphate buffer (Pi; pH 7.0) consisted of molar fractions of 39% sodium monobase and 61% potassium dibase. Enzyme purification. EC was purified from E. coli BL21 harboring pDS2. The sox/dszABC genes (4, 27) were first cloned into pUC18 at the EcoRI and SmaI sites to generate pDS1 via PCR amplification (2) using the IGTS8 high-molecular-weight DNA and primers 59-TCAAGGCCTGAATTCGGCTGGTGG-39 and 59-CGATCAGCGCCTCGGATCCTCAGG-39. The underlined base was that mutated to create an EcoRI site. pDS2 was obtained by subcloning the ApaI-HindIII fragment of pDS1 containing the sox/dszC gene into pET-21b at the XbaI and HindIII sites. The ApaI and XbaI ends were blunted for ligation. pDS2-transformed BL21 was grown in 5 liters of medium containing, per liter, 20 g of tryptone, 10 g of peptone, 5 g of NaCl, 1 ml of 1 M NaOH, and 100 mg of ampicillin at 378C to an optical density at 600 nm of about 1.5 in a BioFlo IIc

* Corresponding author. Mailing address: Department of Biochemical and Biophysical Sciences, University of Houston, Houston, TX 77204-5934. Phone: (713) 743-8359. Fax: (713) 743-8351. 5699

5700

LEI AND TU

J. BACTERIOL.

FIG. 1. Scheme 1.

fermentor (New Brunswick Scientific). The expression of EC was induced at 208C by adding 0.15 g of isopropyl-b-D-thiogalactopyranoside (IPTG). Cells were harvested by centrifugation after overnight induction to an optical density at 600 nm of ;6. Fifty grams of wet cell paste so obtained was sonicated for 10 min in 150 ml of 10 mM Pi on ice. Lysed cells were centrifuged at 19,000 3 g for 20 min to remove cell debris. DEAE-cellulose (40 ml) was added to the lysate, and the mixture was stirred for 20 min. The protein-bound resin was first washed with 300 ml of cool water followed by 50 ml 0.1 M Pi. Proteins were then recovered with 350 ml of 0.5 M Pi. The sample so obtained was applied to a phenyl-Sepharose 6 fast flow (high sub; Pharmacia) column (1.5 by 30 cm) preequilibrated and washed with 0.5 M Pi. When the A280 decreased to near the baseline, the elution buffer was changed to 5 mM Pi to recover bound proteins, including EC. The enzyme sample (85 ml; 640 mg of protein) was dialyzed against 50 mM Pi and loaded for further purification on a DEAE-Sepharose (Pharmacia) column (2.5 by 40 cm) preequilibrated with 50 mM Pi. The column was eluted with 900 ml of 10 mM Pi containing a linear gradient of NaCl from 0.1 to 0.3 M. Vibrio harveyi NADPH-specific flavin reductase (FRP) was purified as described previously (18). Photobacterium (formerly Vibrio) fischeri major flavin reductase (FRG) was purified from E. coli JM109 containing pFRG. pFRG was obtained by cloning the frg gene (33) encoding FRG into pUC18 at the PstI and EcoRI sites via PCR amplification (2) using P. fischeri genomic DNA and primers 59-CATCATAAGTTCTGCAGACAAGAA-39 and 59-CTTTAAATAGAATTC TACCGTAG-39. The underlined bases were mutated to create the PstI and EcoRI restriction enzyme sites. E. coli JM109 harboring pFRG was first cultured in 5 liters of medium (the same as that for the pSD2-containing BL21) at 378C to an optical density at 600 nm of about 1.5 and then at 258C for 2 more days in a BioFlo IIc fermentor. Forty grams of wet cell paste so obtained was sonicated for 10 min on ice in 120 ml of 50 mM Pi containing 0.5 mM dithiothreitol. Lysed cells were centrifuged, and the supernatant was directly loaded on a DEAESepharose column (2.5 by 40 cm) preequilibrated with 50 mM Pi and isocratically eluted with 0.3 M Pi. The FRG pool so obtained was adjusted to 0.8 M (NH4)2SO4 and loaded on a phenyl-Sepharose 6 fast flow (high sub; Pharmacia) column (1 by 20 cm) preequilibrated with 0.8 M (NH4)2SO4 in 50 mM Pi. The column was eluted first with 50 ml of the same equilibration solution and then with 0.5 M (NH4)2SO4 in 50 mM Pi to recover FRG. FRG was concentrated by 70% ammonium sulfate precipitation and dialyzed overnight against 50 mM Pi containing 0.5 mM dithiothreitol. Protein concentrations were determined by the method of Lowry et al. (19) using bovine serum albumin as a standard. Purities of protein samples were tested by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDSPAGE) (15). Enzyme assays. Activities of EC were determined by high-performance liquid chromatographic (HPLC) analysis of the conversion of DBT or DBTO to DBTO2 in reactions carried out in two different procedures: flavin reductasecoupled assay and Cu(I) assay. Most of the coupled assays utilized the flavin reductase FRP. The EC-FRP coupled reaction was carried out by incubating 10 mM EC, 50 mM DBT or DBTO, 0.01 mM FRP, 50 mM FMN, and 0.3 mM NADPH in 1 ml of 5 mM Pi until all NADPH was oxidized. Similar reactions were performed with benzyl sulfide or benzyl sulfoxide replacing DBT or DBTO2 under otherwise identical conditions. For the measurement of EC specific activity, the coupled assay was also modified to utilize 0.02 mM P. fischeri FRG, 50 mM FMN, 0.4 mM NADPH, 0.15 mM DBT, and various levels of EC. Procedures developed for bacterial luciferase assay (16, 17) were modified for the Cu(I) assay. Ten microliters of saturated CuClO4 solution in acetonitrile was added to 1 ml of 5 mM Pi containing flavin at a designated concentration and 10 mM EDTA to reduce the flavin. The resulting reduced flavin was withdrawn into a syringe and vigorously injected into 1 ml of air-saturated 5 mM Pi containing 25 mM EC and 50 mM DBT. For all assays, products and the unreacted reactants were extracted with 0.5 ml hexane. After 2 min of vigorous vortexing and 1 min of centrifugation in a microcentrifuge, the hexane phase was collected and analyzed with a Waters HPLC system equipped with a reverse-phase C18 mBondapak column. The column was eluted isocratically with 60% acetonitrile

aqueous solution at a flow rate of 1.0 ml/min. DBT, DBTO, and DBTO2 were monitored by A285 and had retention times of 13.31, 4.33, and 5.45 min, respectively. Relative molar amounts of DBT, DBTO, and DBTO2 were based on areas under the corresponding peaks calibrated with millimolar extinction coefficients at 285 nm of 9.3, 5.4, and 5.9 mM21 z cm21, respectively. The A210 was monitored to analyze benzyl sulfide, benzyl sulfoxide, and benzyl sulfone, which had retention times of 15.20, 4.97, and 5.86 min, respectively. FPLC. The molecular weight of EC was determined by molecular sieve chromatography using a computer-controlled Pharmacia fast-performance liquid chromatography (FPLC) system and an FPLC Sephadex 75 column (1 by 30 cm). The column was preequilibrated and eluted with 50 mM Pi at a flow rate of 1.0 ml/min and was calibrated with blue dextran, FMN, and protein standards bovine serum albumin (Mr, 68,000), ovalbumin (Mr, 43,000), and cytochrome c (Mr, 12,380). The respective retention times of these substances were found to be 7.25, 18.35, 8.13, 9.55, and 12.47 min. Flavin binding. The binding of reduced FMN (FMNH2) and FMN by EC was measured by the method of Hummel and Dreyer (8). For one series of FMNH2 binding experiments, a Sephadex G-25 column (1 by 30 cm) was equilibrated with 50 mM Pi containing 10 mM glucose, 10 mM EDTA, a designated level of FMNH2, and 3 mg of each of glucose oxidase and catalase per ml. An EC sample containing the same concentration of FMNH2 was loaded on the column and eluted with 50 mM Pi containing the same level of FMNH2. FMNH2 in the buffer and column was produced by photochemical reduction of FMN with EDTA using long-wavelength UV after deoxygenation through the activities of glucose oxidase and catalase. FMNH2 in the sample was obtained by adding solid Na2S2O4. FMNH2 was measured by monitoring the A445 after its oxidation. For another series of measurements of FMNH2 binding, essentially the same procedures were followed except that FMNH2 was included only in the enzyme sample and not in the column or the elution buffer. In addition, the elution buffer was not made anaerobic with glucose and the deoxygenating enzymes. FMN binding was similarly carried out by both procedures except that neither the column nor the buffer contained EDTA, glucose, or the deoxygenating enzymes. 18 O labeling. 18O labeling of DBT and DBTO by EC was carried out by the FRP coupled reaction using 10% 18O-labeled H2O or 95 to 98% 18O2. For the reaction using 18O2, 50 ml of 5 mM Pi containing 10 mM EC, 50 mM FMN, 600 mM NADPH, and 2 mg of DBT or DBTO in 2 ml of ethanol was added to a 125-ml flask with a tee-joint stopper on top and a side arm for sample addition. 16 O2 was first removed by three cycles of vacuum application and refilling with N2 purified over a BASF R3-11 catalyst (Ace Glass Co.). Fifty microliters of 40 mM FRP in a microsyringe was added through the side arm into the N2-saturated solution, and the solution was denitrogenated by vacuum application. The reaction was initiated by introduction of 18O2 and continued for 2 h with stirring. For another 18O labeling experiment, 45 ml of 10% 18O-labeled H2O was mixed with 2 ml of enzyme, 0.5 ml of FMN, 0.5 ml of NADPH, 0.05 ml of FRP, and lastly 2 ml of DBT in ethanol, all at the same final concentrations as for the 18O2 labeling reaction. The reaction solution was stirred under aerobic conditions for 2 h. For both the 18O2 and H218O labeling experiments, the final reaction solutions were centrifuged to remove DBT precipitates. The supernatants were each extracted with 100 ml of hexane, and the extract was concentrated to about 0.5 ml by boiling and transferred to a 1-ml glass vial. Crystals of DBTO2 product were obtained at 2208C overnight. High-resolution mass spectra were obtained by using a Bell and Howell 21-110B instrument.

RESULTS Overexpression of EC. Overexpression of the cloned sox/ dszC gene, under the control of the T7 promoter, was achieved by IPTG induction of the T7 RNA polymerase gene integrated into the host cell’s genome (28). Cells grown at 25, 30, and 378C all generated soluble EC to about 2% of the level of total cellular proteins on the basis of image analysis of SDS-PAGE gels (Fig. 2, lane 2). However, the expression of sox/dszC could be improved by growing the cells at 208C, yielding EC at about 8% of the total cellular protein level (Fig. 2, lane 3). The enhancement of sox/dszC expression by cell growth at 208C was found to be quite reproducible. Purification of EC. EC was purified, stepwise, by batch adsorption and phenyl- and DEAE-Sepharose chromatography. Since EC was a major cellular protein and its activity assay was time-consuming, EC was routinely detected by SDS-PAGE during purification and the purified EC was then confirmed by activity assay. The results for a typical preparative run are shown in Table 1. EC bound tightly to phenyl-Sepharose in 0.5 M Pi but was effectively eluted with 5 mM Pi. This property allowed direct loading of a large volume of sample in 0.5 M Pi from batch adsorption. Moreover, a substantial amount of

VOL. 178, 1996

SULFIDE/SULFOXIDE MONOOXYGENASE

FIG. 2. Effect of growth temperature on expression of sox/dszC in E. coli. Expression of the gene in pDS2 was induced in host E. coli BL21 at 258C (lane 2) and 208C (lane 3) as described in the text. Lanes 1 and 4 are lysate of the host cells and the purified EC, respectively. Lysates (;30 mg) and the purified protein (2 mg) were resolved on an SDS–14% polyacrylamide gel. Lane 5 was for protein standards consisting of, from top to bottom, the myosin H chain (200 kDa), phosphorylase b (97.4 kDa), bovine serum albumin (68 kDa), ovalbumin (43 kDa), carbonic anhydrase (29 kDa), b-lactoglobulin (18.4 kDa), and lysozyme (14.3 kDa).

impurity proteins, about a quarter of the total proteins, directly flowed through the column during the loading. The final treatment by DEAE-Sepharose chromatography generated apparently homogeneous enzyme on the basis of SDS-PAGE (Fig. 2, lane 4). Characterization of purified enzyme. Molecular weight (Mr) determination by gel filtration chromatography was carried out as described in Materials and Methods. The retention time for EC was found to be 7.66 min, corresponding to an Mr of 86,000. In comparison with a monomeric Mr of 45,100 on the basis of the deduced amino acid sequence (4, 27), the gel filtration results indicate a homodimer structure for EC. The purified enzyme has an absorption peak at 281 nm but no peaks in the range of 300 to 700 nm. Apparently, the enzyme does not contain any tightly bound flavin or heme cofactor, since both have absorption peaks in the 300 to 700 nm region. One milliliter of the enzyme with an A281 of 1 in a 1-cm light path was found to contain 0.71 mg. The molar extinction coefficient at 281 nm was then calculated to be 1.3 3 105 M21 z cm21 by using an Mr of 90,200 for the homodimeric enzyme. TABLE 1. Purification of the EC expressed by E. colia Purification stepb

Vol (ml)

A280

Total protein (mg)

% 45-kDa bandc

Yield (%)

Lysate Batch adsorption Phenyl-Sepharose DEAE-Sepharose

180 350 80 170

157 4.6 10.4 2.8

8,100 1,230 640 340

8 47 75 .95

100 89 74 52

a

From 50 g of cells (wet weight). Samples obtained at each step of the purification treatment were analyzed by SDS-PAGE. Patterns for the lysate and the purified enzyme from DEAE-Sepharose chromatography are shown as lanes 3 and 4, respectively, in Fig. 2. c The relative percentage of the 45-kDa band was obtained by image analysis of the SDS-PAGE gels. b

5701

Enzyme activity and substrate specificity. Enzyme activity was determined by the conversion of DBT to DBTO2. It has been shown that flavin reductase can enhance the overall desulfurization process catalyzed by the IGTS8 lysate (32). Therefore, a series of experiments were carried out to test the possible requirement of reduced flavin for the activity of EC. V. harveyi FRP produces FMNH2 at the expense of NADPH and was used to provide FMNH2 for the EC reaction. The purified EC was incubated with a solution containing DBT, FMN, NADPH, and FRP in air-saturated 5 mM Pi until NADPH was all oxidized. The hexane extract of the reaction mixture showed two HPLC peaks, with the first and second peaks corresponding to DBTO2 and the unreacted DBT, respectively. About 79% of the DBT was converted into DBTO2. No product formation from DBT was detected in the absence of FRP under otherwise identical conditions. Flavin specificity was subsequently examined. The purified FRP has a bound FMN cofactor and possibly can provide FMNH2 for the EC reaction even without exogenously added flavin. Consequently, the Cu(I) assay was used for the determination of EC flavin specificity. FMN, FAD, and riboflavin, each at 50 mM, were reduced by Cu(I)-EDTA. The resulting reduced flavins were withdrawn into syringes and injected into air-saturated buffer containing 25 mM EC and 50 mM DBT. According to HPLC analysis, 33, 7, and 0% of the DBT reacted when FMNH2, reduced riboflavin, and reduced FAD, respectively, were used. The EC apparently uses FMNH2 as a preferred cosubstrate under our experimental conditions. It should be pointed out that the enzyme has to compete with autooxidation for the utilization of reduced flavin in the Cu(I) assay. Differences in autooxidation rate among the three reduced flavins tested, if any exist, may affect the observed flavin substrate activities. The FRP-coupled assay and the subsequent sample extraction and analyses were also carried out with benzyl sulfide or benzyl sulfoxide replacing DBT. In the reaction using benzyl sulfide, 71% benzyl sulfoxide, 6% benzyl sulfone, and 23% unreacted benzyl sulfide were detected. When benzyl sulfoxide was used as a substrate under otherwise identical conditions, 11% of the sulfoxide was converted into benzyl sulfone. EC activities were also determined at 238C by the FRGcoupled assay as a function of the amount of the purified enzyme as described in Materials and Methods. With enzyme concentrations up to 0.23 mg z ml21, the amounts of DBTO2 formation were linearly proportional to time up to 20 min and the rates of DBTO2 production were linearly proportional to the amount of enzyme used. A specific activity of the purified EC was thus calculated to be 4.4 nmol of DBTO2 formation per min per mg of enzyme under our experimental conditions. Flavin binding. Gel filtration of a sample containing the enzyme and excess FMNH2 on a Sephadex G-25 column preequilibrated and eluted with Pi, which was deoxygenated but contained no FMNH2, generated two flavin peaks (Fig. 3a). The first and second peaks had the same retention volumes as did those for EC alone and free FMN alone, respectively. Only the second peak was detected in another run under identical conditions but without EC. Therefore, the first peak is the FMNH2 associated with the enzyme and the second peak is the unbound flavin. On the basis of protein and flavin contents of the first peak, 10.7 nmol of FMNH2 was bound to 10 nmol of the dimeric enzyme. In another experiment, 1 ml of 10 mM EC with 8.3 mM FMNH2 was loaded on a G-25 column preequilibrated and eluted with a buffer containing the same level of FMNH2. As shown in Fig. 3b, a flavin peak and a trough were detected, corresponding to binding of 11.0 nmol of FMNH2 by 10 nmol of dimeric enzyme. This latter experiment was re-

5702

LEI AND TU

FIG. 3. Demonstration of FMNH2 binding by sulfide/sulfoxide monooxygenase by gel filtration. (a) One milliliter of 10 mM dimeric enzyme with 40 mM FMNH2 reduced with Na2S2O4 was applied to a Sephadex G-25 column (1 by 30 cm) deoxygenated with 3 mg each of glucose oxidase and catalase per ml in 50 mM Pi–10 mM glucose–10 mM EDTA and eluted with air-saturated 50 mM Pi. (b) One milliliter of 10 mM dimeric enzyme with 8.3 mM FMNH2 was loaded on the same column preequilibrated and eluted with 50 mM Pi containing 10 mM glucose, 10 mM EDTA, 3 mg each of glucose oxidase and catalase per ml, and the same level of FMNH2. FMNH2 was photochemically reduced as described in the text. A445 was monitored after the oxidation of FMNH2 in each fraction by air. The fraction volume was 0.58 ml.

peated under identical conditions but with 20 mM FMNH2 in the enzyme sample and the buffer used for column equilibration and elution. Results showed that 10.8 nmol of FMNH2 was bound by 10 nmol of dimeric enzyme. These results indicate that one molecule of homodimeric enzyme binds one FMNH2 and the binding is tight. Assuming that 99% of the enzyme molecules bound FMNH2 in the presence of 8 mM free FMNH2, the Kd for the enzyme-FMNH2 complex was estimated to be 8.1 3 1028 M. Since the enzyme bound about the same amount of FMNH2 in the absence of exogenously added FMNH2 in the column equilibration or the elution buffer, the real Kd must be less than 1028 M. Elution of an enzyme sample containing excess FMN on the same column preequilibrated with Pi buffer showed an A280 peak of protein (Fig. 4a, solid circles) with little A445 followed by an A445 peak with a shoulder near the side of the protein peak (Fig. 4a, open circles). The A445 peak came out much earlier than did FMN alone on the same column. These results indicate that EC significantly interacts with FMN but that the interaction is not as tight as that for FMNH2 binding. When the experiment was carried out in the column preequilibrated and eluted with 27.7 mM FMN, a typical equilibrium binding of FMN to EC was observed (Fig. 4b). Similar results were also obtained with 8.7, 18.7, and 45.2 mM FMN, and the molar ratio of bound FMN to total dimeric enzyme (FMNb/Et) is shown as a function of free FMN concentration (FMNf) in Fig. 4c. Assuming that EC has n independent binding sites for FMN with the same affinity, a double reciprocal plot of FMNb/Et versus FMNf would give 1/n as the intercept and Kd/n as the slope. Figure 4d shows such an analysis of the data from Fig. 4c. The intercept and the slope were 0.9 and 8.0 mM, respectively, corresponding to 1.1 FMN molecules bound per dimeric enzyme and a Kd of 7.0 mM. Reaction steps. When DBTO (50 mM) was tested as a substrate for EC in the FRP-coupled assay, HPLC analysis revealed that all the DBTO was converted to DBTO2. In comparison, 79% of the DBT was oxidized to DBTO2 with no DBTO detected under identical conditions. Apparently, EC can catalyze the oxidation of both DBT and, more efficiently, DBTO to DBTO2. These results suggest that DBTO is an

J. BACTERIOL.

FIG. 4. Binding of FMN by sulfide/sulfoxide monooxygenase. (a) One milliliter of 10 mM dimeric enzyme with 40 mM FMN was loaded on a Sephadex G-25 column (1 by 30 cm) preequilibrated and eluted with 50 mM Pi. A280 for the protein peak and A445 in each fraction were measured. (b) One milliliter of 10 mM dimeric enzyme with 27.7 mM FMN was applied to the same G-25 column preequilibrated and eluted with 50 mM Pi containing the same level of FMN. (c) FMNb/Et was obtained from the flavin peak associated with the enzyme in panels a and b as well as three similar runs at different FMNfs. FMNb/Et is shown as a function of FMNf. (d) Double reciprocal plotting of FMNb/Et and FMNf using data from panel c for determining the FMN binding stoichiometry and dissociation constant (see the text). The fraction volume was 0.58 ml.

intermediate in the oxidation of DBT to DBTO2. Furthermore, in a continuous reaction system like the coupled assay, the enzyme converts DBT to DBTO less efficiently than it subsequently converts DBTO to DBTO2. To test this two-step reaction scheme, we have devised a Cu(I) assay for a pulse addition of FMNH2. In the Cu(I) assay, FMNH2 reduced by Cu(I)-EDTA was injected into air-saturated solution containing DBT and EC. In this system, free FMNH2 is rapidly oxidized within seconds at 208C (6) and thus is not available for sustained reaction. Unlike the FRP coupled assay, in which FRP continuously generates FMNH2 and EC turns over again and again, the Cu(I) assay allows EC to turn over once or only a limited number of times depending on the relative rates of the enzyme reaction and the autooxidation of free FMNH2. Therefore, the DBTO so generated has a better chance to escape from the subsequent conversion to DBTO2. The results of such a reaction showed 14, 34, and 52% DBTO, DBTO2, and DBT, respectively, in the reaction solution. As expected for the two-step reaction, DBTO was indeed detected along with DBTO2 as a product. A series of Cu(I) assays using different initial FMNH2 concentrations were carried out. The amount of DBT that was oxidized increased gradually from 7 to 40%, and the molar ratio of DBTO2 to DBTO also increased from 0.5 to 3.0 as the initial FMNH2 concentration changed from 5 to 40 mM (Fig. 5a). In another series of assays, EC was first incubated under anaerobic conditions with DBT and various levels of FMNH2 reduced photochemically with EDTA and the solution was then injected into an air-saturated buffer to initiate the reaction. Again, larger molar ratios of DBTO2 to DBTO were detected when higher initial FMNH2 concentrations were used. The EC-FRP coupled assay was also employed to examine the pattern of product formation as a function of initial NADPH concentration. DBTO was not observed in the conversion of DBT to DBTO2 in the coupled assay using 50 mM DBT and 300 mM NADPH but was detected at lower NADPH concentrations (Fig. 5b). Importantly, at increasing NADPH

VOL. 178, 1996

FIG. 5. Effects of NADPH, DBT, and FMNH2 concentrations on DBTO and DBTO2 levels in the FRP-coupled and Cu(I) assays of sulfide/sulfoxide monooxygenase. In the Cu(I) assay (a), 10 ml of saturated CuClO4 solution in acetonitrile was added to 1 ml of 5 mM Pi containing flavin at a designated concentration and 10 mM EDTA to reduce the flavin. The resulting reduced flavin was withdrawn into a syringe and vigorously injected into 1 ml of air-saturated 5 mM Pi containing 25 mM EC and 50 mM DBT. The initial FMNH2 concentrations immediately after mixing are indicated. The coupled reactions were carried out in 1 ml of 5 mM Pi containing 50 mM (b) or 100 mM (c) DBT, 10 mM enzyme, 50 mM FMN, 0.01 mM FRP, and designated levels of NADPH. The reaction solution in both assays was extracted with 0.5 ml of hexane 5 min after the initiation of the reaction. The extracts were subjected to HPLC analysis. Only the DBTO (shaded peaks) and DBTO2 (open peaks) product peaks are presented.

concentrations, the amount of DBTO2 formed increased in parallel whereas the amount of DBTO showed an initial increase followed by a gradual decrease (Fig. 5b). The same experiment was repeated with 100 mM DBT as the starting substrate. Again, larger amounts of DBTO2 were obtained at higher NADPH concentrations. An even more pronounced pattern of an initial increase and a subsequent decrease in the amount of DBTO obtained was observed (Fig. 5c). 18 O labeling. In order to determine the source of oxygen atoms attached to sulfur in the products, 18O labeling experiments were performed with combinations of 10% 18O-labeled H2O–normal O2 (H2O labeling) and normal H2O–95 to 98% 18 O2 (O2 labeling) as described in Materials and Methods. Hexane extract of the reaction solutions contained about 70% DBTO2 product and 30% unreacted DBT. Each 100-ml extract was concentrated to ;0.5 ml, resulting in the formation of DBTO2 crystals upon standing with DBT left in a soluble form. The crystalline product samples were subjected to high-resolution mass spectrometric analyses using reagent ions generated from CH4. The chemical ionization generates m/e [M11]1 as a major ion (21). DBTO2 without 18O labeling has a molecular weight of 216.0245. Therefore, major m/e peaks

SULFIDE/SULFOXIDE MONOOXYGENASE

5703

FIG. 6. Mass spectra of the product in the 18O labeling reactions catalyzed by sulfide/sulfoxide monooxygenase. As described in Materials and Methods, the reactions were carried out in 50 ml of 5 mM Pi containing 2 mg of DBT or DBTO, 10 mM enzyme, 50 mM FMN, 600 mM NADPH, and 0.4 mM FRP by using pairs of 18O2-normal H2O and normal O2–10% 18O-labeled H2O. The crystalline DBTO2 product samples were subjected to high-resolution mass spectrometric analyses using reagent ions generated from CH4. (a to c) Mass spectra of the DBTO2 products from DBT in the H218O labeling reaction, from DBT in the 18O2 labeling reaction, and from DBTO in the 18O2 labeling reaction, respectively.

217.0323, 219.0366, and 221.0408 are expected for the DBTO2 product containing 0, 1, and 2 18O atoms, respectively. The DBTO2 obtained with the H218O labeling sample showed peaks at m/e’s 217.0323, 218.0355, and 219.0286 with relative intensities of 100, 15, and 4.8, respectively (Fig. 6a). These peaks match the calculated m/e’s of 217.0323, 218.0357, and 219.0282 for the protonated DBTO2 isotopic formulas of 12 C121H916O232S1, 13C12C111H916O232S1, and 12C121H916O2 34 1 S , respectively. The peak m/e 219.0286 does not match the 18 O-labeled product 12C121H918O16O32S1, which has a calculated m/e of 219.0366. Furthermore, the intensity ratio of m/e 217.0323 to m/e 219.0286 was observed to be 20.8:1, in good accord with the theoretical ratio 22.5:1 calculated on the basis of 34S incorporation, at its natural abundance (4.22%), into DBTO2. In contrast, the expected ratio should be 9:1 if only one of the two oxygen atoms in DBTO2 is derived from water or 4.5:1 if both oxygen atoms are from water in a stepwise conversion of DBT to DBTO and then to DBTO2. Thus, H2O does not contribute any oxygen atom to the product DBTO2. The mass spectrum of the 18O2 labeling sample showed three peaks with an intensity distribution similar to that of the H218O labeling sample, but all the peaks shifted in m/e up by 4 (Fig. 6b). The predominant peak at m/e 221.0408 matches the calculated m/e, 221.0408, of 18O-labeled protonated DBTO2, 12 C121H918O232S1. These results indicate that the two oxygen

5704

LEI AND TU

J. BACTERIOL.

atoms inserted into DBT by EC are both from O2 instead of H2O. DBTO2 obtained in the EC-FRP coupled reaction using DBTO and 18O2 has a major peak at m/e 219.0364 (Fig. 6c) matching well the calculated m/e, 219.0366, for 12C121H9 16 32 1 18O O S . Therefore, the oxygen atom added to DBTO by EC to form DBTO2 is also from O2. DISCUSSION In this work, we have succeeded in the overexpression of the sox/dszC gene (from Rhodococcus sp. strain IGTS8) in E. coli and devised a simple procedure for the purification of the expressed enzyme EC (Table 1). We found that the purified EC has no bound prosthetic group that is associated with any absorption in the visible-wavelength range. However, the DBT oxidation activity of EC requires exogenously added FMNH2. Therefore, the availability of purified EC has enabled us to identify FMNH2 as a preferred cosubstrate for this enzyme. Our finding is consistent with and provides a molecular basis for the interpretation of the earlier reports that the inclusion of an FRP-dependent FMNH2-generating system in the lysate of Rhodococcus sp. strain IGTS8 (32) and the addition of NADH and FMN to a partially purified enzyme from R. erythropolis D-1 (24) enhance the DBT oxidation activity. However, we did not detect any significant activity of reduced FAD for EC, in contrast to earlier reports that the DBT oxidation activity can be enhanced by the inclusion of FAD in the lysate of E. coli cells that express EC (4) or by the addition of FAD and NADH to a partially purified enzyme from R. erythropolis D-1 (24). While it is difficult to identify the exact cause for the apparent discrepancy because of the undefined composition of the crude enzyme samples used in these earlier studies, it is possible that the apparent activity of FAD observed earlier was derived from FMN either present as a contaminant in the FAD sample or generated from FAD by a hydrolytic enzyme(s) in the crude enzyme samples. Kilbane (11) previously proposed a theoretical overall pathway for the desulfurization of DBT which includes a stepwise conversion of DBT to DBTO and subsequently to DBTO2. However, the intermediacy of DBTO in such a process has never been experimentally established. In fact, the exact type of EC catalytic activity has, thus far, remained unspecified. By several lines of investigation we found that EC is capable of catalyzing both the oxidation of DBT to DBTO and the subsequent conversion of DBTO to DBTO2. Furthermore, as shown below, EC functions as a monooxygenase in both reactions: DBT 1 FMNH2 1 O2 3 DBTO 1 FMN 1 H2O

(1)

DBTO 1 FMNH2 1 O2 3 DBTO2 1 FMN 1 H2O

(2)

The formation of both DBTO and DBTO2 from DBT in reactions catalyzed by the purified EC was shown by either the Cu(I) assay or the EC-FRP coupled assay (Fig. 5). The direct conversion of DBTO to DBTO2 by EC was further demonstrated. Figure 5 also shows that as increasing levels of FMNH2 were made available, the amount of DBTO increased first and then decreased whereas DBTO2 formation showed a continuous increase. These results are consistent with a two-step conversion of DBT to DBTO2 involving DBTO as an intermediate. It should be noted that just recently Mrachko et al. (23) also reported in an abstract the purification of EC to homogeneity and concluded that EC functions as a monooxygenase in con-

verting DBT to DBTO2 in two steps. Although a detailed comparison of our and their studies cannot be made at this time, the report by Mrachko et al. (23) and our results are apparently mutually confirmative. The source of the inserted oxygen in DBTO and DBTO2 was examined by isotope labeling using H218O and 18O2. Results of labeling with H218O (Fig. 6a) show that none of the oxygen in DBTO or DBTO2 was derived from water. On the other hand, the 18O2 labeling results show that both oxygen atoms in DBTO2 originated from molecular oxygen when the DBTO2 was formed from DBT (Fig. 6b) and that a single oxygen atom from O2 was inserted into DBTO to generate the DBTO2 (Fig. 6c). These findings, taken together, allow us to deduce that the single oxygen in DBTO is also derived from O2 in its formation from DBT. Therefore, EC can be identified as a monooxygenase capable of forming DBTO from DBT and, subsequently, generating DBTO2 from DBTO. We also tested benzyl sulfide and benzyl sulfoxide with the purified EC and found that both were active substrates for the formation of the sulfone product. Therefore, the monooxygenase activity of EC is not limited to thiophene and thiophene sulfoxide as the only substrates. Consequently, we propose sulfide/sulfoxide monooxygenase as a general name for EC. The vast majority of known flavin-dependent external monooxygenases are associated with a tightly bound FAD cofactor which can be reduced by NADH or NADPH by their own catalytic activities (31). The sulfide/sulfoxide monooxygenase now joins bacterial luciferase (reference 30 and references therein) and 2,5-diketocamphane 1,2-monooxygenase (29) as an exception. First, these enzymes do not exhibit any significant flavin reduction activity and they rely on a separate enzyme species to reduce the flavin. Second, the preferred flavin is FMN for all three enzymes. Sulfide/sulfoxide monooxygenase, bacterial luciferase, and 2,5-diketocamphane 1,2-monooxygenase also show some interesting similarities and differences among themselves. One flavin site per dimer has been detected for each of these enzymes. However, the native sulfide/sulfoxide monooxygenase is a homodimer and the 2,5-diketocamphane 1,2-monooxygenase consists of two subunits with equal molecular weights, but bacterial luciferase is a heterodimer. Both sulfide/sulfoxide monooxygenase and bacterial luciferase bind FMNH2 substantially more tightly than they do FMN, but the former enzyme has affinities for oxidized and reduced FMN higher than those of the latter enzyme. The reduced flavin binding affinity for 2,5-diketocamphane 1,2-monooxygenase has not been determined, but this enzyme apparently has a very high affinity for FMN binding. Last but not least, sulfide/sulfoxide monooxygenase is highly unusual in its activity of catalyzing two consecutive steps of monooxygenation reaction. Most known flavin-dependent monooxygenases are from microbial sources. None of these microbial enzymes other than the sulfide/sulfoxide monooxygenase has been shown to exhibit such a two-step monooxygenation activity. However, the liver microsomal flavin-containing monooxygenase catalyzes the monooxygenation of, among other substrates, organic sulfides (such as dimethyl sulfoxide) to the corresponding sulfoxides and subsequently to sulfone (10). In this respect, the sulfide/sulfoxide monooxygenase provides the first example of a microbial enzyme resembling the microsomal flavin-containing monooxygenase. Interestingly, the latter monooxygenase is distinct from the sulfide/sulfoxide monooxygenase in being a monomer, having a bound FAD cofactor, and being able to reduce the bound flavin by its own catalysis.

VOL. 178, 1996

SULFIDE/SULFOXIDE MONOOXYGENASE ACKNOWLEDGMENTS

This work was supported by grants GM25953 from the National Institutes of Health and E-1030 from the Robert A. Welch Foundation. REFERENCES 1. Afferden, M., S. Schacht, J. Klein, and H. G. Tru ¨per. 1990. Degradation of dibenzothiophene by Brevibacterium sp. DO. Arch. Microbiol. 153:324–328. 2. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. A. Smith, J. G. Seidman, and K. Struhl (ed.). 1994. Current protocols in molecular biology, p. 15.1.1–15.1.7. Wiley & Sons, Inc., New York. 3. Crawford, D. L., and R. K. Gupta. 1990. Oxidation of dibenzothiophene by Cunninghamella elegans. Curr. Microbiol. 21:229–231. 4. Denome, S. A., C. Oldfield, L. J. Nash, and K. D. Young. 1994. Characterization of the desulfurization genes from Rhodococcus sp. strain IGTS8. J. Bacteriol. 176:6707–6717. 5. Gallagher, J. R., E. S. Olson, and D. C. Stanley. 1993. Microbial desulfurization of dibenzothiophene: a sulfur specific pathway. FEMS Microbiol. Lett. 107:31–36. 6. Gibson, Q. H., and J. W. Hastings. 1962. The oxidation of reduced flavin mononucleotide by molecular oxygen. Biochem. J. 83:368–377. 7. Hemmerich, P., and C. Sigwart. 1963. Cu(CH3CN)21, ein Mittel zum Studium homogener Reaktionen des einwertigen Kupfers in wa¨ssriger Lo ¨sung. Experientia 19:488–489. 8. Hummel, J. P., and W. J. Dreyer. 1962. Measurement of protein-binding phenomena by gel filtration. Biochim. Biophys. Acta 63:532–534. 9. Izumi, Y., T. Ohshiro, H. Ogino, Y. Hine, and M. Shimao. 1994. Selective desulfurization of dibenzothiophene by Rhodococcus erythropolis D-1. Appl. Environ. Microbiol. 60:223–226. 10. Jakoby, W. B., and D. M. Ziegler. 1990. The enzymes of detoxication. J. Biol. Chem. 265:20715–20718. 11. Kilbane, J. J. 1989. Desulfurization of coal: the microbial solution. Trends Biotechnol. 7:97–101. 12. Kilbane, J. J., and K. Jackowski. 1992. Biodesulfurization of water-soluble coal-derived material by Rhodococcus rhodochrous IGTS8. Biotechnol. Bioeng. 40:1107–1114. 13. Kodama, K., K. Umehara, K. Shimizu, S. Nakatani, Y. Minoda, and K. Yamada. 1973. Identification of microbial products from dibenzothiophene and its proposed oxidation pathway. Agric. Biol. Chem. 37:45–50. 14. Laborde, A. L., and D. T. Gibson. 1977. Metabolism of dibenzothiophene by a Beijerinckia species. Appl. Environ. Microbiol. 34:783–790. 15. Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London) 227:680–685. 16. Lei, B., and J. E. Becvar. 1991. A new reducing agent of flavins and its application to the assay of bacterial luciferase. Photochem. Photobiol. 54: 473–476. 17. Lei, B., K. W. Cho, and S.-C. Tu. 1994. Mechanism of aldehyde inhibition of Vibrio harveyi luciferase: identification of two aldehyde sites and relationship between aldehyde and flavin binding. J. Biol. Chem. 269:5612–5618.

5705

18. Lei, B., M. Liu, S. Huang, and S.-C. Tu. 1994. Vibrio harveyi NADPH-flavin oxidoreductase: cloning, sequencing and overexpression of the gene and purification and characterization of the cloned enzyme. J. Bacteriol. 176: 3552–3558. 19. Lowry, O. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall. 1951. Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193:265–275. 20. Massey, V., and B. E. P. Swoboda. 1963. The flavin composition of pig heart muscle preparations. Biochem. Z. 338:474–484. 21. Millard, B. J. 1978. Quantitative mass spectrometry, p. 5–6. Heyden & Son, Ltd., Philadelphia. 22. Monticello, D. J., D. Bakker, and W. R. Finnerty. 1985. Plasmid-mediated degradation of dibenzothiophene by Pseudomonas species. Appl. Environ. Microbiol. 49:756–760. 23. Mrachko, G. T., O. S. Pogrebinsky, C. H. Squires, D. J. Monticello, and K. A. Gray. 1996. Biochemical characterization of the biodesulfurization of dibenzothiophene by Rhodococcus sp. strain IGTS8, abstr. O-17, p. 356. In Abstracts of the 96th General Meeting of the American Society for Microbiology 1996. American Society for Microbiology, Washington, D.C. 24. Ohshiro, T., Y. Kanbayashi, Y. Hine, and Y. Izumi. 1995. Involvement of flavin coenzyme in dibenzothiophene degrading enzyme system from Rhodococcus erythropolis D-1. Biosci. Biotechnol. Biochem. 59:1349–1351. 25. Omori, T., L. Monna, Y. Saiki, and T. Kodama. 1992. Desulfurization of dibenzothiophene by Corynebacterium sp. strain SY1. Appl. Environ. Microbiol. 58:911–915. 26. Omori, T., Y. Saiki, K. Kasuga, and T. Kodama. 1995. Desulfurization of alkyl and aromatic sulfides and sulfonates by dibenzothiophene-desulfurizing Rhodococcus sp. strain SY1. Biosci. Biotechnol. Biochem. 59:1195–1198. 27. Piddington, C. S., B. R. Kovacevich, and J. Rambosek. 1995. Sequence and molecular characterization of a DNA region encoding the dibenzothiophene desulfurization operon of Rhodococcus sp. strain IGTS8. Appl. Environ. Microbiol. 61:468–475. 28. Studier, F. W., and B. A. Moffatt. 1986. Use of bacteriophage T7 RNA polymerase to direct selective high-level expression of cloned genes. J. Mol. Biol. 189:113–130. 29. Taylor, D. G., and P. W. Trudgill. 1986. Camphore revisited: studies of 2,5-diketocamphane 1,2-monooxygenase from Pseudomonas putida ATCC 17453. J. Bacteriol. 165:489–497. 30. Tu, S.-C., and H. I. X. Mager. 1995. Biochemistry of bacterial bioluminescence. Photochem. Photobiol. 62:615–624. 31. van Berkel, W. J. H., and F. Mu ¨ller. 1991. Flavin-dependent monooxygenases with special reference to p-hydroxybenzoate hydroxylase, p. 1–29. In F. Mu ¨ller (ed.), Chemistry and biochemistry of flavoenzymes, vol. II. CRC Press, Boca Raton, Fla. 32. Xi, L., J. D. Child, and C. H. Squires. 1995. FMN reductase is involved in the desulfurization of DBT by Rhodococcus erythropolis IGTS8, poster 17T. 14th Enzyme Mechanisms Conference, Scottsdale, Ariz., 4 to 8 January 1995. 33. Zenno, S., K. Saigo, H. Kanoh, and S. Inouye. 1994. Identification of the gene encoding the major NAD(P)H-flavin oxidoreductase of the bioluminescent bacterium Vibrio fischeri ATCC 7744. J. Bacteriol. 176:3536–3543.