Sulfite: Cytochrome c Oxidoreductase from Thiobacillus novellus

Direct oxidation of sulfite to sulfate occurs in various photo- and chemotrophic sulfur oxidizing microorgan-isms as the final step in the oxidation of reduced sulfur compounds and is catalyzed by sulfite:cytochrome c oxidoreductase (EC 1.8.2.1). Here we show that the enzyme from Thiobacillus novellus is a periplasmically located ab heterodimer, consisting of a 40.6-kDa subunit containing a molybdenum cofactor and an 8.8-kDa mono-heme cytochrome c 552 subunit (midpoint redox poten- tial, E m 8.0 5 1 280 mV). The organic component of the molybdenum cofactor was identified as molybdopterin The standard reaction 0.05 m M cytochrome heart (Sigma) and a known amount of SOR in 20 m M 8.0, in a final volume of 1 ml. The reaction was carried out at 23 6 1 °C and was started by the addition of sulfite (final concentration, 2.5 m M ; stock solution, 200 m M in 50 m M Tris-HCl, pH 8.8, 5 m M EDTA). The rate of reaction was quantified by monitoring the increase in absorb-ance at 550 nm; 1 unit was defined as the amount of enzyme oxidizing 1 m mol of sulfite/min. Extinction coefficients for cytochrome c were as given by the manufacturer (cytochrome c from horse heart) or as deter- mined by the method of Berry and Trumpower (18) ( T. novellus cytochrome c 550 ). All activity determinations were taken as the mean of three measurements and were corrected for non-enzymatic sulfite oxidation. Measurements in which cytochrome c was supplied in non- saturating concentrations (determination of reaction mechanism) were carried out in the presence of 20 m M EDTA, i.e. under partly inhibiting conditions as described in Ref. 19 to overcome detection limits of measurements with a light path of 1 cm. The presence of EDTA in the reaction mix raises the affinity of the enzyme for cytochrome c by ; 1 order of magnitude, while there is no obvious effect on the affinity for sulfite. Assays with non-saturating substrate concentrations and partial inhibition of the enzyme hence contained 20–100 m M cytochrome c and 2–16 m M sulfite; assays with only one substrate varied, and the second substrate at saturating concentrations contained 1–12 m M cyto- chrome c and 4 -100 m M sulfite. Catalytic constants were determined both from Hanes-Plots and by direct nonlinear fitting (least squares method) using the Scientist software (Micromath), both methods yield-ing similar results. Data for mixed type inhibition were fitted to the following equation (20).

Sulfite:acceptor oxidoreductases (SOR) 1 (often termed sulfite oxidases although the physiological electron acceptor appears to be a c-type cytochrome in all cases studied so far) catalyze the final step in the degradation of sulfur-containing amino acids and have been extensively studied in mammalian and avian systems (1). SORs from these sources are located in the intermembrane space of the mitochondria (2). They form a molybdoenzyme family with assimilatory nitrate reductases and are homodimers (molecular mass, 100 -112 kDa) containing one heme b and one molybdopterin (MPT)-type pterin molybdenum cofactor/subunit (1). The recently published crystal structure of the chicken liver enzyme revealed the presence of three functional domains: a heme b-coordinating domain, an MPT binding domain and a so-called "dimerization domain," which has been assigned a major role in the formation of the homodimeric structure of these enzymes. SORs have also been studied by electron paramagnetic resonance (EPR), resonance Raman, and electron spin echo-envelope modulation spectroscopies (3)(4)(5) and exhibit spectroscopic properties distinct from those of other families of molybdenum-containing enzymes. One particularly distinguishing feature in the spectroscopy of the eukaryotic SORs studied to date is the appearance of I ϭ 1/2 superhyperfine splitting of the Mo(V) EPR signal at low pH.
In addition to the eukaryotic sulfite:acceptor oxidoreductases, enzymes catalyzing the same reaction have been found in photo-and chemotrophic sulfur-oxidizing bacteria (6 -8). During oxidation of reduced sulfur compounds such as thiosulfate or sulfide by obligately or facultatively chemolithotrophic sulfur oxidizers (most of them formerly grouped as Thiobacilli), sulfite is formed as a free or bound intermediate of the complex oxidation processes (9,10). In these organisms sulfite can be oxidized to sulfate either directly by SOR (EC 1.8.2.1), which can occur either as free enzymes or as part of thiosulfateoxidizing multienzyme complexes (10,11), or via the energy conserving adenosine-5Ј-phosphosulfate reductase pathway. The presence of one or both of these enzymes has been established for various Thiobacilli, and multienzyme complex-independent sulfite:acceptor oxidoreductases have been enriched or partially purified from various chemotrophic sulfur oxidizers (6). None of these SORs, however, have been purified to homogeneity. Analysis of the available information on these enzymes shows that they form a very heterogeneous group; membrane-bound and soluble SORs can be found as well as enzymes that are inhibited by AMP or stimulated by its addition.
SOR from Thiobacillus novellus, a member of the ␣-proteobacteria, has been described previously as a monomeric, heme, and molybdenum cofactor-containing enzyme with an apparent molecular mass of ϳ40 kDa (12) that reacts with a cytochrome c 550 from the same organism (13). However, more extensive characterization of the type of pterin molybdenum cofactor contained in the enzyme, the exact nature of the heme group, and further properties of this or any other of the Thiobacillus SORs remained outstanding. On a molecular biological level, even less data are available: to date the only confirmed gene sequence for a bacterial SOR is the soxC gene (14) from Paracoccus pantotrophus GB17 (15) (formerly Thiosphaera pantotropha (16)), which encodes an enzyme that is part of a multienzyme complex.
The present work describes the first spectroscopic and enzymological characterization of a bacterial member of the sulfite oxidase family, the SOR from T. novellus. The relationship of T. novellus SOR with other SOR enzymes is discussed based on analysis of its properties and gene sequences.

EXPERIMENTAL PROCEDURES
Materials-Chemicals were purchased from Merck, Sigma, and Fluka in per analisi or corresponding quality.
Bacteria, Media, and Plasmids-Bacteria and plasmids used are listed in Table I. T. novellus was cultivated in modified DSMZ medium 69 containing 4.2 g/l Na 2 HPO 4 , 1.5 g/l KH 2 PO 4 , 0.3 g/l yeast extract, 0.3 g/l NH 4 Cl 2 , 0.1 g/l MgSO 4 , 10 g/l Na 2 SO 3 ⅐5 H 2 O, 5 ml/l trace element solution, final pH 8.5. The trace element solution was prepared as specified for DSMZ medium 69. MgSO 4 and thiosulfate were autoclaved separately. For solid medium, 1.5% agar were added. 200-ml cultures were aerated by vigorous shaking and incubated for 3-4 days at 30°C; 10-liter cultures were grown in carboys with forced aeration over 6 -7 days. Sterile 1 M Na 2 CO 3 was used to maintain a pH of about 8.5. After 2-3 days, 500 ml of a thiosulfate feeding solution (200 g/l) were added to 10-liter cultures every second day. Cells were harvested by centrifugation, and the pellet was stored at Ϫ20°C. Escherichia coli strains were grown in liquid or solidified Luria-Bertani, 2 ϫ YT medium or, as required, NZY medium (17).
SOR Purification-Except where otherwise stated, all purification steps were performed at 4°C. 20 g of frozen T. novellus cell material were homogenized in 3 volumes of 10 mM Tris-HCl, pH 7.8, and cells were broken by sonication (1 min/ml) in a Branson sonifier. Cell debris was removed by centrifugation (Sorvall RC-5B, 18 000 ϫ g, 30 min), and the resultant crude extract was subjected to ultracentrifugation (Beckman L5-50; 2 h, 145,000 ϫ g). This was followed by an ammonium sulfate precipitation (45% saturation), centrifugation, and heat treatment of the supernatant (50°C, 30 min in the presence of 20 mM sodium sulfite). Following quick chilling of the extract and the removal of precipitate by centrifugation, ammonium sulfate was added to 70% saturation (cf. Ref. 12), and the resultant precipitate was then dissolved in 10 mM Tris-HCl, pH 7.8, and dialyzed against 2 ϫ 5 liters of the same buffer for ϳ16 h. The dialyzed extract was applied to a DEAE-Sephacel column (1.6 ϫ 15 cm), equilibrated in the same buffer, and eluted using a linear sodium chloride gradient (400 ml, 0 -500 mM NaCl). The sulfiteoxidizing activity eluted between 120 and 180 mM NaCl. Active fractions were pooled, dialyzed extensively against 2 ϫ 5 liters of 1 mM potassium phosphate buffer, pH 7.8, and subsequently applied to a hydroxyapatite column (HTP Biogel, Bio-Rad, 2.6 ϫ 20 cm) buffered in 1 mM potassium phosphate buffer, pH 7.8. A 400-ml linear gradient of 1-50 mM phosphate was applied, and SOR activity eluted between 5 and 20 mM potassium phosphate. Enzyme-containing fractions were selected for pooling so that a maximum separation from a contaminating cytochrome c 551 was achieved. Fractions were concentrated by ultrafiltration (Centriprep-10, Amicon). The retentate was gel-filtered on a Superdex 75 16/60 column (Amersham Pharmacia Biotech) equilibrated with 10 mM Tris-HCl, pH 7.8, containing 150 mM NaCl, and fractions exhibiting SOR activity were desalted and subjected to strong anion exchange chromatography on a MonoQ 5/5 fast protein liquid chromatography column (10 mM Tris-HCl, pH 7.8; gradient, 0 -250 mM NaCl in 50 ml). Active fractions, which eluted at NaCl concentrations between 130 -200 mM, were checked for purity using SDS-PAGE.
SOR Activity Measurement-SOR activity was determined spectrophotometrically using a Perkin-Elmer Lambda 11 spectrophotometer.  The standard reaction mixture contained 0.05 mM cytochrome c from horse heart (Sigma) and a known amount of SOR in 20 mM Tris-HCl, pH 8.0, in a final volume of 1 ml. The reaction was carried out at 23 Ϯ 1°C and was started by the addition of sulfite (final concentration, 2.5 mM; stock solution, 200 mM in 50 mM Tris-HCl, pH 8.8, 5 mM EDTA). The rate of reaction was quantified by monitoring the increase in absorbance at 550 nm; 1 unit was defined as the amount of enzyme oxidizing 1 mol of sulfite/min. Extinction coefficients for cytochrome c were as given by the manufacturer (cytochrome c from horse heart) or as determined by the method of Berry and Trumpower (18) (T. novellus cytochrome c 550 ). All activity determinations were taken as the mean of three measurements and were corrected for non-enzymatic sulfite oxidation. Measurements in which cytochrome c was supplied in nonsaturating concentrations (determination of reaction mechanism) were carried out in the presence of 20 mM EDTA, i.e. under partly inhibiting conditions as described in Ref. 19 to overcome detection limits of measurements with a light path of 1 cm. The presence of EDTA in the reaction mix raises the affinity of the enzyme for cytochrome c by ϳ1 order of magnitude, while there is no obvious effect on the affinity for sulfite. Assays with non-saturating substrate concentrations and partial inhibition of the enzyme hence contained 20 -100 M cytochrome c and 2-16 M sulfite; assays with only one substrate varied, and the second substrate at saturating concentrations contained 1-12 M cytochrome c and 4 -100 M sulfite. Catalytic constants were determined both from Hanes-Plots and by direct nonlinear fitting (least squares method) using the Scientist software (Micromath), both methods yielding similar results. Data for mixed type inhibition were fitted to the following equation (20).
The constants K ic and K iu refer to the competitive and the non-competitive component, respectively, of the inhibition pattern observed. Characterization of the mechanisms of reaction and inhibition was performed according to Cleland (21,22). Statistical errors are given as the standard deviation of the mean.
Denaturing PAGE-SDS-PAGE was carried out according to the method of Laemmli (23). Native PAGE was carried out using essentially the same method but without the inclusion of SDS and denaturing agents. Gels were stained for protein using either Coomassie Brilliant Blue or silver (24). The method of Cohen and Fridovich (25) was employed for SOR activity staining; heme-dependent peroxidase activity was identified on denaturing gels using 3Ј-3Ј-diaminobenzidine (26) and on native gels by enhanced chemiluminescence (27).
Purification of c-Type Cytochromes from T. novellus-Three c-type cytochromes were enriched or purified from T. novellus as a byproduct of the SOR purification procedure. They were separated from the SOR preparation either after DEAE anion exchange or after hydroxyapatite chromatography (cytochrome c 551 ). Whereas cytochrome c 550 was purified by subsequent cation exchange chromatography (28), cytochrome c 552 and c 551 were further enriched by strong anion exchange chromatography (MonoQ 5/5, Amersham Pharmacia Biotech) and gel filtration on a Superdex 75 16/60 column (Amersham Pharmacia Biotech).
Amino acid sequencing was performed using an Applied Biosystems gas phase sequenator after Western blotting on a polyvinylidene difluoride membrane (Millipore) or tryptic digest and HPLC separation of samples.
Determination of Heme and Mo Content-Heme type and content were determined spectrophotometrically in alkaline pyridine solutions according to the method of Berry and Trumpower (18). Molybdenum content was quantified colorimetrically as described previously (29,30). Midpoint potentials of heme groups were measured as in Refs. 31 and 32.
Identification of Pterin Molybdenum Cofactor-The molybdenum cofactor was characterized by analytical reverse-phase (2 xHypersil C18, 126 ϫ 4.6 mm; particle size, 5 m) HPLC chromatography of oxidation products of the cofactor following liberation from SOR upon denaturation of the enzyme (33)(34)(35). Xanthine oxidase (Sigma) and Me 2 SO reductase from Rhodobacter capsulatus (36) were used as reference sources of MPT and molybdopterin-guanine dinucleotide forms of molybdenum cofactor.
Spectroscopic Measurements-UV-visible absorption spectrophotometric measurements were carried out using either a Shimadzu UV-3000 double beam spectrophotometer or a Perkin-Elmer Lambda 11 spectrophotometer. EPR spectroscopy was carried out at the Center for Magnetic Resonance at the University of Queensland, using a Bruker ESP300E spectrometer equipped with a TE 102 rectangular resonant cavity and a cold nitrogen flow cryostat. Temperature control was effected using a Eurotherm B-VT2000 variable temperature controller. A microwave power of 10 milliwatts, a modulation frequency of 100 kHz, and a modulation amplitude of 0.25 millitesla were employed; the temperature was kept constant at 130 K. Other signal acquisition parameters were selected so that the spectral resolution was limited by the modulation amplitude. Microwave frequencies between 9.460 and 9.465 GHz were monitored using an EIP 548B microwave counter, and field positions were recorded using a Bruker ER 035M NMR Gaussmeter. Samples for EPR contained 0.4 mg/200 l of purified enzyme in either 50 mM Tricine, pH 8.8, or Hepes, pH 7.0. To generate the Mo(V) species of SOR, 20 mM sulfite were added. Computer simulations of Mo(V) EPR spectra were carried out as described in earlier work (37).
Immunological Methods-SOR antiserum was prepared in rabbit by injection of 2 ϫ 750 g of purified SOR. Western blotting and immunodetection were carried out using standard procedures (17). SOR antiserum was diluted 1:2000.
Molecular Biological Methods-Standard molecular biological techniques were used throughout (17,38). Restriction enzymes were purchased from Life Technologies, Inc. Taq DNA polymerase was from Promega. Genomic DNA from T. novellus was isolated as described by Nano and Kaplan (39). A representative genomic library of partially Sau3AI-digested T. novellus DNA was constructed from fragments Ͼ6 kilobases in the lambda ZAP express vector (Stratagene) according to the manufacturer's instructions. DNA sequencing was carried out by Sequiserve (Vaterstetten, Germany) using an automated sequencer and the chain-termination method (40).

RESULTS
Sulfite:cytochrome c oxidoreductase from T. novellus was purified using a modified and expanded protocol based on the procedure in Ref. 12; the results of the purification are summarized in Table II. The enzyme eluted from the final strong anion exchange column in two overlapping peaks containing the reduced and the oxidized form of SOR, as determined by the UV-visible absorption of the cytochrome component. Native PAGE of the purified SOR revealed the presence of only one protein band upon Coomassie staining (Fig. 1A). This band also stained for SOR activity and cross-reacted with the T. novellus SOR antiserum (Fig. 1A). SDS-PAGE of the purified enzyme yielded two Coomassie-staining bands with apparent molecular masses of ϳ40 and ϳ8 kDa, respectively, identifying the enzyme as a heterodimer (Fig. 1B). The overall molecular mass of the enzyme as determined after SDS-PAGE is in good agreement with a value of ϳ46 kDa obtained after gel filtration on Superdex 75.
Characterization and Content of Cofactors-Optical spectra of the purified enzyme clearly reveal the presence of a sulfitereducible c-type heme with the following absorption maxima in the reduced state (values in parentheses refer to maxima in the oxidized spectrum): ␣, 552 nm (Ϫ); ␤, 523 nm (Ϫ); ␥, 416 nm (411 nm); ␦, 317 nm (363 nm) (Fig. 2). Heme staining after native and denaturing PAGE identified the heme as an integral part of the enzyme (Fig. 1A) with the heme group being covalently bound by the 8-kDa component of the holoenzyme (Fig.  1B). Absorption spectra of purified SOR in alkaline pyridine confirmed that the heme group contained in T. novellus SOR is a c-type heme (18). SOR contains 0.8 Ϯ 0.15 heme groups/ holoenzyme as estimated from the absorption spectrum. Redox titration at pH 8.0, the pH at which SOR exhibited maximum activity, suggested a midpoint potential for this heme group of E m8.0 ϭ ϩ280 mV.
Colorimetric determination of Mo content gave an average value of 1.2 Ϯ 0.25 Mo atoms/holoenzyme. The analytical data then suggest a content of one molybdenum atom and one heme group/␣␤-holoenzyme. The type of pterin molybdenum cofactor contained in SOR was elucidated by liberation and subsequent oxidation of the cofactor followed by reverse phase HPLC chromatographic analysis. As there are several known forms of the pterin molybdenum cofactor, xanthine oxidase (Sigma) and dimethyl-sulfoxide reductase from R. capsulatus (36) were used as reference sources of MPT and molybdopterin-guanine dinucleotide-type cofactors, respectively. The dinucleotide form of the pterin molybdenum cofactor is often found in bacterial enzymes (3). The elution profile of the cofactor oxidation products from the reverse phase HPLC column clearly shows that SOR from T. novellus contains an MPT-type cofactor (Fig. 3), as the oxidized cofactor extracted from this enzyme shows the same elution profile as that from xanthine oxidase but is distinct from that exhibited by Me 2 SO reductase.
EPR Spectroscopy-Although samples of the resting enzyme in Hepes, pH 7.0 (Fig. 4A), and in Tricine, pH 8.8 (Fig. 4A), did not give rise to a Mo(V) EPR signal, upon the addition of 20 mM sulfite to either sample a well defined Mo(V) EPR signal was observed (Fig. 4A). The g-values, anisotropy, narrow line widths, and particularly the weak features at 350 and 355 millitesla justify the assignment of the signal to Mo(V); the spacings and intensities of these latter features are consistent with their being due to the ϳ25% natural abundance of I ϭ 5/2 95/97 Mo nuclei. The spectra were simulated (Fig. 4A) assuming g 1 , 1.9914; g 2 , 1.9661; and g 3 , 1.9541. A high resolution spectrum of the g 1 -signal recorded at a lower modulation am-plitude (0.1 millitesla; Fig. 4B) clearly showed that the "wings" present on this feature are asymmetrically positioned about the main I ϭ 0 g 1 -feature and are thus due to underlying 95/97 Mo hyperfine lines rather than to either unresolved superhyperfine splitting or to ⌬m I ϭ 1 1 H "spin-flip" transitions. Studies at up to 200 milliwatt microwave power (not shown) also provided no evidence for 1 H spin-flip transitions, though at 130 K these may be expected to be difficult to detect (cf. Ref.

42).
At both pH values the signals obtained were very similar to the so-called "high pH" EPR signal that can be readily observed with avian and mammalian sulfite oxidoreductases (43)(44)(45). However, another characteristic EPR signal of these enzymes, the "low pH" form, was not observed with T. novellus SOR samples at pH 7.0 and could not be induced by the addition of 200 mM NaCl (Fig. 4A), as reported for the eukaryotic enzymes (43,44). Within the limits of the noise level of the data, the signal from T. novellus SOR at pH 7.0 and containing 200 mM Cl Ϫ (Fig. 4A) is indistinguishable from that seen at high pH (Fig. 4A). The EPR spectrum (Fig. 4A) of a sulfite-reduced SOR sample in 0.5 M phosphate buffer, pH 7.0, containing 200 mM NaCl was also similar to the one seen at high pH. Thus, the difference in behavior of the T. novellus enzyme and the eukaryotic enzyme is not merely an effect of Hepes buffer anion binding. The signal in phosphate buffer is, perhaps, slightly  Fig. 4B (high resolution spectrum of sulfite-reduced enzyme in Hepes, pH 7.0). Spectra were recorded at 130 K, 10 milliwatt microwave power, 0.25 or 0.1 millitesla modulation amplitude, 100 kHz modulation frequency. Spectra are presented normalized in intensity for enzyme concentration; between 10 and 80 spectra were averaged in each case. different in that the g 2 -feature appears to be asymmetrically broadened, the g 3 -feature appears to have a lower line width, and a shoulder is apparent on the high field side of this feature. Nevertheless, the best simulation of this signal as a single species was still that of Fig. 4A. The samples investigated failed to elicit the characteristic phosphate-inhibited signal (45,46) observed in eukaryotic sulfite oxidoreductases under similar conditions. It would appear that the ability of the molybdenum center to adopt a low pH ligand environment is an obligatory prerequisite for subsequent formation of the phosphateinhibited species and hence this latter species is also not observed in the T. novellus enzyme.
Enzymological Characterization of SOR-The pH optimum of the reaction catalyzed by SOR from T. novellus was determined in 20 mM Tris-HCl and 20 mM bis-trispropane-HCl (BTP) buffer in steps of 0.5 pH units and was found to be pH 8.0 and pH 8.5, respectively. Apart from the slightly higher optimum, the reaction showed a much greater pH dependence in BTP buffer than in Tris, with the activity falling to 65% of the maximum within 1 pH unit either side of pH 8.5 in BTP, whereas the corresponding values in Tris-HCl were 80 and 95% of the maximum (data not shown). The optimum temperature of the SOR-catalyzed reaction was determined to be 65°C, about 35°C above the optimum growth temperature of T. novellus. Preincubation of oxidized and sulfite-reduced SOR solutions at various temperatures with subsequent determination of the remaining activity confirmed the great heat stability of this enzyme; although samples of oxidized SOR lost 50% activity after incubation at 60°C for 2 min, 50% inactivation of the reduced enzyme was only observed after 15 min at 70°C (data not shown).
An isoelectric point of pH 5.5 was determined by preparative isoelectric focusing on a MonoP column (Amersham Pharmacia Biotech). Controlled tryptic digestion of T. novellus SOR had no influence on enzyme activity as had already been reported (12) and as would be expected for a heterodimeric protein. N-terminal sequencing of both subunits and tryptic digestion of the larger, Mo-containing subunit yielded the amino acid sequences indicated in Fig. 6.
Catalytic Properties-Apparent K m values for sulfite and cytochrome c (horse heart) at saturating concentrations of the second substrate were determined to be 27 and 4 M, respectively. To investigate the reaction mechanism, measurements of enzyme activity with both substrates supplied in nonsaturating concentrations were carried out. A double reciprocal plot of the data (Fig. 5) yielded a set of parallel lines, suggesting the presence of a ping-pong mechanism (47). This kind of mechanism is also found in the eukaryotic SORs and in this case involves two active centers. V max as determined from a secondary plot was 2500 units/mg. The calculated V max is only valid under the experimental conditions used.
Inhibition Studies-SOR from T. novellus is strongly inhibited by anions and buffer substances such as phosphate and bis-trispropane as well as by the substrate analogue nitrate (Table III). The enzyme is subject to product inhibition by sulfate, which shows a mixed-type non-competitive pattern with respect to sulfite. K ic and K iu values were determined to be 24 and 16 mM for sulfite. A similar product inhibition pattern with a K ic value of 8 mM for sulfite (determined from a Dixon-Plot) has been reported for the chicken liver SOR (19).
Reactivity of SOR with Purified c-Type Cytochromes-Three c-type cytochromes were purified or enriched from T. novellus extracts. These heme-containing proteins have apparent molecular masses of 12.6 kDa (c 550 ), 75 kDa (c 551 ), and 125 kDa (c 552 ), with purity coefficients (E soret red /E 280 ) of 5.65, 2.79, and 0.996. Purified preparations of the c-type cytochromes were used as electron acceptors in SOR activity assays. It was found that SOR reacts exclusively with cytochrome c 550, as previously proposed (13). This finding is consistent with the midpoint  potentials at pH 8.0 determined for cytochromes c 550 , c 551 , and c 552 , which were ϩ302, ϩ253, and ϩ242 mV. The apparent K m value of SOR for the native cytochrome c 550 under substratesaturating conditions was 2.5 M, a value almost identical to the one determined for the artificial acceptor cytochrome from horse heart. Furthermore, this value is sufficiently low as to be possibly physiologically relevant.
Cloning and Analysis of the sor Region from T. novellus-Two-hybrid phages, TNSOR32 and TNSOR34, containing the sor gene region were isolated from a representative genomic library in lambda ZAP express using a partial sorA gene probe. The probe was amplified from genomic DNA using degenerate oligonucleotides deduced from the N-terminal and tryptic amino acid sequences of SorA, the large SOR subunit. After excision of the phagemid vectors pTNSOR32 and pTNSOR34 from the hybrid phages, 2378 bp of the insert of pTNSOR32 (Fig. 6) were sequenced on both strands by primer walking.
Analysis of the sorAB Gene Sequence-The sequenced region ( Fig. 6) contains two open reading frames encoding a 40.15-kDa (43.4 kDa unprocessed) polypeptide with strong similarities to known SORs (error probabilities: eukaryotic SORs, between 5 ϫ 10 Ϫ39 and 8 ϫ 10 Ϫ35 , bacterial SoxC protein 1 ϫ 10 Ϫ25 ) and a novel 8.8 kDa (11.55 kDa unprocessed) mono-heme c-type cytochrome. The two open reading frames, designated sorA and sorB, are separated by only 16 bp and encode the amino acid sequences determined for the subunits of the purified SOR (Fig. 6). Putative ribosome binding sites were identified for both open reading frames (Fig. 6). A putative promoter for ECF-type sigma factor-dependent expression (48) is located upstream of sorA and a possible hairpin loop terminator structure with a calculated free energy of formation of Ϫ109.2 kJ/ mol is found 122 bp downstream of sorAB, suggesting that sorAB form an operon. The proposed promoter region, which contains a Ϫ35 region "GAATAT" followed by Ϫ10 region "TCTTC" at a distance of 17 bp, retains 64% of the suggested consensus sequence for ECF-dependent promoters (48), which consists of a Ϫ35 region "GAACTT" and a Ϫ10 region "TCTRA" The identification of a putative promoter for ECF-type sigma factor-dependent expression upstream of sorA is particularly interesting as ECF-type sigma factors are proposed to regulate genes encoding proteins with extracytoplasmic functions (48). Also, SorA and SorB are preceded by signal sequences. Both observations are indicative of a periplasmic location of the SOR holoenzyme. While SorA is preceded by a 32-amino acid leader of the double arginine type (49), which is likely to target the protein for export via the tat system (50,51), the SorB signal sequence (27 amino acids) is characteristic of Sec-dependent export (52). The similarity of the sorA gene product to eukaryotic SORs is restricted to two core domains, the molybdopterin binding and the dimerization domain. No similarities between either SorA or SorB and the heme b binding domain of the eukaryotic enzymes were detected. A conserved cysteine, which coordinates the MPT cofactor in the eukaryotic enzymes (53), is also found in SorA (Fig. 7). This finding is perhaps to be expected from the similarity of the EPR signal of T. novellus SOR to the high pH signal of the eukaryotic enzymes.
The sorA and sorB gene products are soluble proteins; there is no suggestion that SOR may be a transmembrane or membrane-associated protein. The calculated pI values of 6.52 and 4.67 for the processed SorA and SorB polypeptides, respectively, are consistent with the experimentally determined value pI ϭ 5.5 for the holoenzyme. DISCUSSION The characterization described herein of the sulfite:cytochrome c oxidoreductase from T. novellus has revealed a novel, heterodimeric type of SOR. The previously described 40-kDa protein (12) corresponds to the pterin molybdenum cofactor binding subunit of the enzyme. The unidentified heme moiety described in the earlier work on the enzyme (12) has been found to reside on a separate, 8.8-kDa subunit, which has been shown by sequence analyses, alkaline pyridine denaturation and elec-tronic absorption spectroscopy to be a c-type cytochrome with a novel polypeptide primary sequence. Hence, the structure of this enzyme is clearly distinct from that of the well studied, eukaryotic SORs, which are homodimeric proteins with a heme b group and a pterin molybdenum cofactor residing on each of the identical subunits.
Similar to the eukaryotic SORs, the T. novellus SOR contains an MPT-type cofactor. This is somewhat unusual, because most of the bacterial molybdoenzymes (3,54) characterized to date contain a dinucleotide form of the pterin molybdenum cofactor. At the same time, most of these enzymes are members of the Me 2 SO reductase family, a notable exception being the molybdopterin-cytosine dinucleotide-containing aldehyde oxidase from Desulfovibrio gigas (55). The discovery of an MPT-type cofactor in a bacterial enzyme belonging to the sulfite oxidase family and in a bacterial member of the third family of molybdoenzymes, the xanthine oxidase family (56), does, however, suggest some specificity of distribution of the forms of the pterin molybdenum cofactor among the molybdoenzyme families.
T. novellus SOR appears to be located in the bacterial periplasm as is indicated by the presence of sequences encoding signal peptides in the gene sequences of both subunits. This observation is supported by the fact that the subunits of the mature protein are both N-terminally processed. Given that eukaryotic SORs are found in the mitochondrial intermembrane space (2), all SORs studied so far seem to reside in cell compartments outside the cytoplasm and have a similar topological position with respect to the respiratory chain. Inspection of the leader sequences encoded by sorA and sorB implies that translocation of the two subunits to the periplasm likely occurs by two distinct pathways. The MPT binding SorA subunit is exported via the tat system (50, 51), known to export proteins that are at least partially folded and with the cofactors inserted, whereas SorB is targeted for Sec-dependent export. This pathway transports unfolded proteins across the cell membrane, and the cofactors necessary for the function of the protein are subsequently inserted in the periplasm (52). A similar situation has been described recently for a periplasmic nitrate reductase (57). How proteins with subunits that are exported by two different translocation systems are assembled in the periplasm is as yet an interesting and unanswered question. Furthermore, the evolutionary background to such a system will prove interesting.
The natural electron acceptor for SOR from T. novellus appears to be another periplasmic protein, a previously identified cytochrome c 550 (13). This reaction exhibits a very low K mvalue of 2. The presence of a cytochrome c 551 group within the SOR preparation described by (13) was shown to be due to a contaminating protein that can be removed from the preparation and is unable to accept electrons from SOR. These findings are consistent with the observation in Ref. 13 that cytochrome c 551 is also found in fractions not associated with enzyme activity and that the addition of the cytochrome to SOR preparations did not enhance the overall sulfite oxidizing activity.
The mechanism of sulfite oxidation as proposed by Rajagopalan (1) on the basis of data collected on the mammalian and avian SORs involves oxidation of sulfite at the molybdenum site, sequential transfer of the abstracted electrons to the heme group followed by a reaction of this group with an external electron acceptor. The mechanism has been termed a "hybrid ping-pong" mechanism and has been compared with a similar type of mechanism found in a biotin transcarboxylase by Northrop (59). In SOR, such a mechanism is almost inevitable:   (4); dark gray box, areas of higher conservation; bold print, conserved cysteine involved in cofactor binding; italics, leader peptides of bacterial SORs; black line, underlines heme b coordinating domains of eukaryotic enzymes; circles, amino acids involved in cofactor or substrate binding (4). oxidation of SO 3 2Ϫ to SO 4 2Ϫ is a two-electron process, whereas the electron storage capability of the enzyme is three electrons (two on the Mo ion as Mo(VI) 3 Mo(IV) and one on the heme group as Fe(III) 3 Fe(II)). As electron egress occurs via the heme group, which can only store one electron, two reactions with a one-electron acceptor are required to restore the enzyme to the Mo(VI)-containing state whereby another sulfite anion can be oxidized. We were able to show that SOR from T. novellus does indeed catalyze the oxidation of sulfite by a ping-pong type mechanism, with the stable intermediate form characteristic of ping-pong mechanisms being the sulfite-reduced enzyme (t 1/2 Ͼ 48 h). In conjunction with the physical data obtained for the enzyme our results suggest that, unlike the hybrid ping-pong mechanism discovered in Ref. 59, which is a combination of a two site ping-pong mechanism with a random ternary mechanism, SORs combine a two site ping-pong mechanism with a (probably random) sequential mechanism at the second, the heme site. The sulfite oxidation reaction catalyzed by SOR is also sensitive to high concentrations of various anions and buffer substances (19,60, and this study), and the pH optimum of the reaction, as well as the overall pattern of activity, are influenced by the type of buffer employed. The strong inhibition of T. novellus SOR by nitrate may be caused by formation of a catalytically inactive enzyme-nitrate complex.
In view of the different domain structures of the eukaryotic and the bacterial enzymes it is interesting that both types show enhanced heat stability when reduced with sulfite. In eukaryotic enzymes this property has been attributed to a change in position of the heme b group relative to the main enzyme body (4,61), since this domain is connected to the rest of the enzyme via an exposed hinge region of amino acids. This adjustment of the position of the heme group is thought to facilitate intramolecular electron transfer processes (61). The observation made suggests that despite the fact that the catalytic groups of the bacterial SOR reside on different, noncovalently connected subunits, a reduction of this enzyme by sulfite also results in intramolecular changes reflected in the altered physical properties reported.
Despite the overall similarity of the two types of sulfite: cytochrome c oxidoreductase the results of EPR spectroscopy suggest that the active site at the molybdenum center of T. novellus SOR might be different from that of the eukaryotic enzymes. Samples investigated failed to elicit a pH-dependent change in EPR signal, which was hitherto thought to be characteristic of this type of enzyme. These findings are in contrast to those of Toghrol and Southerland (12) who reported a pHdependent change of EPR signal conformation in Tris-buffered samples. The low pH form obtained by these authors, however, appears to be a mixed species of high and low pH signal. It seems likely that the coordination sphere of the molybdenum atom in the bacterial enzyme does not allow for the change in position of the Mo(V)-OH species thought to be responsible for the change between the high and low pH form of the chicken liver enzyme (5). Hepes buffer is known to bind strongly to the molybdenum centers of some enzymes and thus we attempted to generate the low pH signal in phosphate buffer. However, not only did the low pH signal remain elusive, we were also unable to generate the phosphate-inhibited EPR signal observable in the eukaryotic enzyme. Again, these findings imply a lack of coordination flexibility in T. novellus SOR compared with the eukaryotic enzymes. Activity measurements suggest FIG. 7-continued that this apparent lack of conformational flexibility of at least the Mo(V) state does not preclude catalytic activity at a level comparable to the eukaryotic enzymes.
T. novellus is one of the facultatively autotrophic members of the formerly so-called "Thiobacilli" and exhibits a great versatility as far as utilization of substrates is concerned. In 1997, Kelly et al. (10) suggested that the oxidation of sulfur compounds in this organism might proceed via the so-called "Paracoccus pathway," which involves a multienzyme-complex and is found in Paracoccus versutus (formerly T. versutus (62)) and P. pantotrophus (15). One reason for this assumption may have been the close phylogenetic relation between these members of the ␣-proteobacteria. Analysis of the sor gene region, however, suggests that oxidation of sulfur compounds via SOR does not involve a multienzyme complex. The SOR structural genes sorAB appear to form an independent operon with a putative promoter for ECF-type sigma factor-dependent expression upstream of sorA and a terminator structure downstream of sorB. This structure does not resemble the environment of the soxC gene from P. pantotrophus, which is located in a genetic region containing the other genes relevant for the function of the thiosulfate-oxidizing multienzyme complex (11,14). SOR expression in T. novellus is induced by the presence of thiosulfate in the growth medium (63), and SOR can amount to about 1.6% of the total cell protein in cells grown lithoautotrophically with thiosulfate, suggesting a major role for the thiosulfate oxidation pathway in which SOR catalyzes the final step, the conversion of the oxidation intermediate sulfite to sulfate.