Serine 121 Is an Essential Amino Acid for Biotin Sulfoxide Reductase Functionality*

Rhodobacter sphaeroides f. sp.denitrificans biotin sulfoxide reductase (BSOR) catalyzes the reduction of d-biotin d-sulfoxide (BSO) to biotin, an important step in oxidized vitamin salvaging. In addition to BSO, the enzyme also catalyzes the reduction of a variety of other substrates, including methionine sulfoxide, with decreased efficiencies, suggesting a potential role as a general cell protector against oxidative damage. Recombinant BSOR, expressed as a glutathione S-transferase fusion protein, contains the molybdopterin guanine dinucleotide cofactor (MGD) as its sole prosthetic group, which is required for the reduction of BSO by either NADPH or reduced methyl viologen. Comparison of the amino acid sequences of BSOR and the closely related MGD-containing enzyme, dimethyl sulfoxide reductase, has indicated a number of conserved residues, including an active site serine residue, serine 121, which has been potentially identified as the fifth coordinating ligand of Mo in BSOR. Site-directed mutagenesis has been used to replace serine 121 with cysteine, threonine, or alanine residues in the BSOR sequence to asses the role of this residue in catalysis and/or Mo coordination. All three BSOR mutant proteins were expressed, purified to homogeneity, and demonstrated to contain both MGD by fluorescence spectroscopy and Mo by inductively coupled plasma mass spectrometry, similar to wild-type enzyme. However, all three mutant proteins were devoid of BSOR activity using either NADPH or reduced methyl viologen as the electron donor. These results strongly suggest that serine 121 in BSOR is essential for catalysis but is not essential for either Mo coordination or MGD binding.

In prokaryotes, such as Escherichia coli (1) and the photo-synthetic bacterium Rhodobacter sphaeroides f sp. denitrificans, 2 BSOR has been demonstrated to be expressed only at low levels, severely limiting studies of the native enzyme. BSOR has been partially purified from E. coli and has been demonstrated to require two accessory proteins for activity: a small, heat stable, thioredoxin-like moiety referred to as protein-(SH) 2 and an unidentified flavoprotein (2). The R. sphaeroides BSOR has been successfully cloned and heterologously expressed as a functional glutathione S-transferase fusion protein (3). Following cleavage of the GST affinity tag, the fusion protein has been purified to homogeneity and demonstrated to contain MGD as its sole prosthetic group. In contrast to the E. coli variant, R. sphaeroides BSOR has been shown to function efficiently in the absence of any additional redox-active or structural proteins (3). R. sphaeroides BSOR has been shown to be able to utilize a variety of alternate substrates, in addition to BSO, including nicotinamide-N-oxide, dimethyl sulfoxide, methionine sulfoxide, and trimethylamine-N-oxide, but exhibits a marked preference for BSO (3). The ability of the enzyme to utilize alternative oxidizing substrates has led to the suggestion that R. sphaeroides BSOR may play a role as a general protector of the cell from oxidative damage, similar to the role proposed for methionine sulfoxide reductase (4,5) and superoxide dismutase (6). Preliminary data on bacterial culture growth in the presence of hydrogen peroxide support this hypothesis.
BSOR is a member of the Me 2 SO reductase family of prokaryotic Mo-containing enzymes. This family comprises a number of related molybdoenzymes that share the characteristic property of requiring a complex of Mo and two molybdopterin guanine dinucleotide (MGD) (7,8) cofactors as their sole redoxactive prosthetic group and that catalyze the reduction of S and N-oxides of organic molecules. In addition to Me 2 SO reductase and BSOR, members of this class include bacterial (dissimilatory) nitrate reductase, trimethylamine N-oxide reductase, and formate dehydrogenase (9).
The structure of the Mo active site in BSOR has only recently been characterized by the application of various spectroscopic techniques, including electron paramagnetic resonance, extended x-ray absorption fine structure, and resonance Raman. The presence of the bis(MGD) molybdenum cofactor was confirmed, as was the presence of four thiolate ligands and a mono-oxo site in the oxidized form and a des-oxo site in the reduced form of this enzyme (10,11). This data has suggested that the coordination sphere of the Mo is very similar in both BSOR and Me 2 SO reductase.
Multiple sequence alignments of proteins belonging to the Me 2 SO reductase family of molybdoenzymes have indicated that one of the ligands to Mo is an amino acid side chain of either serine, cysteine, or selenocysteine (12). In Me 2 SO reduc-* This work was supported by Grant GM 32696 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. tase this residue corresponds to serine 147, whereas in BSOR the corresponding residue is serine 121. Mutation of Ser 147 to a Cys in Me 2 SO reductase has been shown to result in differential effects on substrate utilization. These include a loss of approximately 61-100% of its activity toward substrates such as Me 2 SO and methionine sulfoxide and a Ͼ400% increase in activity for adenosine N 1 -oxide reduction (13).
To investigate the role of Ser 121 in the functionality of BSOR, we have generated the corresponding alanine, cysteine, and threonine mutants and examined the proteins for both Mo and MGD incorporation and their associated catalytic activities.

Chemicals, Enzymes, and Reagents
Factor Xa protease, Pfu DNA polymerase, and DpnI restriction enzyme were purchased from Promega, and media for bacterial growth was purchased from Difco. Restriction enzymes were purchased from New England Biolabs Inc. (Beverly, MA). Aprotinin, antibiotics, biotin, dithiothreitol, phenylmethylsulfonyl fluoride, reduced glutathione, and basic buffer chemicals were purchased from Sigma. SDS, acrylamide, bis-acrylamide, protein molecular weight markers, and protein assay solution were purchased from Bio-Rad. Isopropylthio-␤-galactoside was obtained from Research Products International Corp. (Mt. Prospect, IL.). d-Biotin d-sulfoxide was prepared as described by Pollock and Barber (14).

Site-directed Mutagenesis
Site-directed mutagenesis was performed following the basic protocols from the QuickChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA). The following oligonucleotides were used for the mutagenesis of serine 121: wild-type, 5Ј C CAT GTG GAC ACC TAT TCC ATT GCA GCC GGG C 3Ј; Ser 3 Ala, 5Ј C CAT GTG GAC ACC TAT GCT ATT GCA GCC GGG C 3Ј; Ser 3 Cys, 5Ј GC CAT GTG GAC ACC TAT TGT ATT GCA GCC GGG C 3Ј; Ser 3 Thr, 5Ј C CAT GTG GAC ACC TAT ACG AT T GCA GCC GGG C 3Ј (bold characters indicate changed bases). Both sense and antisense complementary oligonucleotides were used in the same PCR reaction, and the primers were designed to introduce a second silent mutation that eliminated a BsrDI site from the wild-type sequence to facilitate mutant screening. 50 ng of template DNA (BSOR gene in the GST vector); 125 ng each of sense and antisense primer; 10% Triton X-100; 5% Me 2 SO; 1.5 units of Pfu DNA polymerase; dATP, dGTP, dCTP, and dGTP (each 200 M); and 1ϫ Pfu buffer were mixed in a 100-l reaction volume. Hot start PCR was utilized for amplification of template DNA, with the mutation introduced in both strands by the mutagenic oligonucleotide pair. 18 cycles of PCR corresponding to 95°C for 30 s, 55°C for 1 min, and 68°C for 15 min were performed following an initial denaturation step of 95°C for 5 min. Following PCR, the wild-type dam methylated template was digested with 10 units of DpnI at 37°C for 1 h. Following template digestion, competent E. coli DH5␣ cells were directly transformed with 2-and 4-l aliquots of the digested PCR reaction without further purification. Colonies that grew on the amp-resistant LB plates were grown and miniprepped for DNA isolation. Positive mutants were identified following restriction digests with BsrDI, which in the mutants cleaved the DNA at one less site than in the wild type. Following initial screening, positive mutants were sequenced using the Sequenase kit (U. S. Biochemical Corp.) and [␣-35 S]dATP.

Protein Expression and Purification
For the isolation of both wild-type and mutant recombinant R. sphaeroides BSORs, transformed E. coli JM109 cells were grown overnight, and BSOR expression was induced by addition of isopropylthio-␤-galactoside as described previously (3). For BSOR isolation, the cells were harvested by centrifugation at 3000 ϫ g for 20 min, and the cell pellets were resuspended in phosphate-buffered saline supplemented with dithiothreitol (10 mM), aprotinin (0.1 mg/ml), EDTA (1 mM), phenylmethylsulfonyl fluoride (0.1 mM), and sodium molybdate (1 mM). The cells were sonicated on ice, and the BSOR was purified as described previously using a combination of glutathione-agarose affinity chromatography, anion exchange chromatography (Mono-Q), and fast protein liquid chromatography gel filtration (Superose 12) (3).

Factor Xa Proteolysis
Fusion protein samples were cleaved by treatment with Factor Xa protease (1% w/w) in 50 mM Tris-HCl buffer containing 150 mM NaCl and 1 mM CaCl 2 (pH 8) at 16°C for 16 h.

Protein Analysis
Recombinant wild-type and mutant BSORs were examined for purity and size using SDS-PAGE. Protein samples (2-3 g of total protein) were analyzed using a 10.0% SDS-PAGE gel (15) stained with Coomassie Blue.

Enzyme Activities
Spectrophotometric assays-BSOR activities (NADPH:BSOR and NADPH:FR) were routinely determined using a Shimadzu (Columbia, MD) UV2501 spectrophotometer. Assays were performed at 25°C in 50 mM Tris-HCl buffer, pH 8.0 at 340 nm, monitoring the oxidation of NADPH (250 M) in the presence of either BSO (1.7 mM) or FeCN 6 3Ϫ (630 M) as electron acceptor and using 2.5 g of purified enzyme in a final volume of 1 ml. NADPH concentrations were calculated using an ⑀340 of 6.22 mM Ϫ1 cm Ϫ1 . Activities, measured as initial rates, were expressed as micromoles of NADPH consumed per minute per nanomole of enzyme. Kinetic parameters were derived from the experimental initial rate data by least-squares fitting to the original hyperbolic rate equation using the software Enzfitter (Elsevier Biosoft, Ferguson, MO).
Reverse Phase HPLC Analysis for Biotin Determinations-Wild-type and mutant BSOR activities were also examined using the MV . ϩ :BSOR assay and reverse phase HPLC to detect the formation of the product, biotin. Each reaction contained 250 l of 116 mM Mes, pH 6 buffer, 100 l of 100 M methyl viologen (in 50 mM Tris, pH 8), 100 l of BSO or biotin (5 mg/ml in 20 mM NaOH), and 15 g of enzyme. The mixture was degassed for 1 h to eliminate any oxygen present, and the reaction was initiated by injecting 20 l of 1 M dithionite. The reaction was allowed to proceed for 0.5 h, following which the enzyme was separated from the reactants and products by spin column filtration. 100 l of the reaction mixture was injected onto a C 18 reverse phase column, and the reactants were separated from the products using a gradient of 0.05% trifluoroacetic acid, pH 2.5 and 0.05% trifluoroacetic acid, acetonitrile (70:30) as described previously (3).
Molybdenum Cofactor Analysis-Wild-type and mutant enzyme samples were denatured with 1% SDS followed by boiling for 30 min. The SDS was precipitated with KCl (0.25 M final concentration), and the cofactor was separated from the protein utilizing ultrafiltration spin columns (ultrafree-MC 5,000 molecular weight cut-off, Millipore Corp., Bedford, MA.). The presence of cofactor in each sample was determined by fluorescence spectroscopy using a Shimadzu RF5301 spectrofluorometer. Excitation spectra were obtained using an emission wavelength of 460 nm, and emission spectra were obtained using an excitation wavelength of 370 nm. Isolated cofactor samples were treated with nucleotide pyrophosphatase as described previously (3).
Molybdenum Analysis-Mo analysis was performed utilizing inductively coupled plasma mass spectrometry at the University of Georgia, Department of Chemistry, Athens, GA using sodium molybdate as standard.

Wild-type and Mutant BSOR Expression and
Purity-To ensure that the kinetic properties of the wild-type and mutant enzymes were not influenced by contaminating proteins, recombinant R. sphaeroides wild type and the three BSOR mutants, corresponding to S121A, S121T, and S121C, were purified to homogeneity and analyzed by SDS-PAGE. As shown in Fig. 1, following anion exchange chromatography of the Factor Xa-cleaved proteins, all of the purified enzymes exhibited a single, high molecular mass band corresponding to a mass of approximately 80 kDa following SDS-PAGE, which was of the appropriate size predicted from the amino acid sequence. An additional low molecular mass band of approximately 29 kDa was also apparent for each sample and corresponded to the cleaved GST tag. All three mutant proteins were expressed as the appropriately sized proteins even though the individual expression efficiencies varied. Whereas the S121T mutant was expressed at levels comparable with wild type, the expression levels of the S121A and S121C mutants were approximately 2-5-fold lower than that of the wild-type BSOR.
Wild-type and Mutant BSOR Activities-To assess the functionality of the mutant proteins, their respective activities were examined utilizing the spectrophotometric assays for NADPH: BSOR and NADPH:FR activities and by HPLC detection of biotin produced following the MV . ϩ :BSOR assay. In addition, all proteins were examined for their reactivity with alternative substrates, such as trimethylamine-N-oxide.
As illustrated in Table I, utilizing equal molar amounts of protein and compared with the wild-type enzyme, none of the mutants exhibited any significant, detectable activity using either the NADPH:BSOR or NADPH:FR assays. The very low levels of activity observed with some of the substrates, corresponding to 0.3-3% of wild-type values, were indistinguishable from background levels. In addition, there was no change in substrate specificity for any of the mutants examined.
The results of the reverse phase HPLC analysis of the MV . ϩ :BSOR activities of the wild-type and mutant forms of BSOR are shown in Fig. 2. Under the reaction conditions utilized for the assay, biotin eluted at 17.7 min and BSO eluted at 9.1 min, partially overlapping the peak at approximately 8.7 min resulting from the control reaction utilizing only methyl viologen, dithionite, and buffer (Fig. 2, trace D) in the absence of BSOR. Biotin formation was detected only when the wild-type enzyme was utilized in the assay mixture (Fig. 2, trace C). In contrast, utilizing the S121A, S121T, or S121C mutants, no biotin production was observed in any of the assays (Fig. 2, traces E-G). Peak intensities observed following HPLC analyses of the wildtype and mutant reaction mixtures are shown in Table II, indicating the absence of any detectable conversion of BSO to biotin by the S121A, S121T, or S121C variants of BSOR.
Cofactor Analysis-The fluorescence excitation and emission spectra of the molybdenum cofactor isolated from wild-type BSOR and the S121A, S121T, and S121C mutants are shown in Fig. 3. All three mutant proteins were demonstrated to incorporate the Mo cofactor as shown by the fluorescence excitation and emission spectra, which yielded characteristic maxima at 370 and 470 nm, respectively, and were very similar to the corresponding spectra obtained from the wild-type enzyme. Small shifts toward lower wavelengths were observed for the fluorescence emission maxima, especially for the S121T and S121C mutants, which may reflect different oxidation states of the cofactor isolated from the mutant proteins. Treatment of the isolated cofactor samples with nucleotide pyrophosphatase resulted in the characteristic approximately two-fold increase in the fluorescence emission intensity, confirming the presence of the dinucleotide form of the molybdenum cofactor in all three mutants.
Molybdenum Analysis-Total Mo analyses of the wild-type, S121A, S121T, and S121C variants of BSOR were performed using inductively coupled plasma mass spectrometry. As shown in Table III, all the mutant proteins were determined to have incorporated Mo, although the mutants were observed to have slightly lower Mo stoichiometries when compared with the wild-type BSOR. The lowest level of Mo incorporation, corresponding to 72% of the wild-type BSOR Mo content, was observed for the S121T mutant, whereas Mo incorporation was higher for both the S121A and S121C mutants. DISCUSSION The preceding results represent the first application of sitedirected mutagenesis to probe the role(s) of specific active-site amino acid residues in the function of R. sphaeroides BSOR. FIG. 1. SDS-PAGE analysis of recombinant R. sphaeroides BSOR wild type and Ser 121 mutants. Purified recombinant wildtype BSOR and the three BSOR mutants (2-3 g each) were analyzed on a 10% SDS-PAGE gel. Lanes M and MЈ represent standard molecular mass and prestained molecular mass markers, respectively. Lane WT corresponds to wild-type BSOR, lane T represents the S121T mutant, lane A represents the S121A mutant, and lane C represents the S121C mutant. The masses of the protein bands in the standard molecular mass markers are indicated in kDa.

FIG. 2. HPLC analysis of the reaction products generated by wild-type and Ser 121 mutant BSORs. MV .
ϩ :BSOR reactions were performed as described under "Materials and Methods," and 100 l of each reaction mixture was injected onto a C 18 reverse phase column. The substrate, BSO, was separated from the product, biotin, using a gradient of 0.05% trifluoroacetic acid, pH 2.5 and 0.05% trifluoroacetic acid, acetonitrile (70:30) with a flow rate of 1 ml/min. Control reactions containing either BSOR or biotin in the absence of BSOR or only dithionite and methyl viologen in buffer are shown in traces A, B, and D, respectively. The reaction products catalyzed by wild-type BSOR are shown in trace C. The reaction products generated by the mutant BSORs S121C, S121A, and S121T are shown in traces E, F, and G, respectively. Replacement of Ser 121 with Ala, Thr, or Cys residues has been demonstrated to result in the production of three mutant forms of BSOR that all retain the incorporation of both MGD and Mo but that are all devoid of both NADPH:BSOR and NADPH:FR activities. Both BSOR and Me 2 SO reductase from R. sphaeroides share regions of extensive sequence similarity and exhibit the unique feature within the diverse array of known molybdoenzymes of containing MGD as their sole prosthetic group. Sequence alignments have indicated approximately 38% overall sequence conservation between the two proteins including regions of the primary structures identified as molybdenum cofactor-binding signatures (14).
X-ray crystallographic studies of Me 2 SO reductase have demonstrated the presence of a novel metal cluster comprising a single Mo and two MGD cofactors. The single Mo atom is coordinated by four thiols, two derived from each MGD cofactor, in addition to an oxygen atom (O ␥ ) that comprises part of the side chain of Ser 147 (12).
The results of multiple sequence alignments of members of the Me 2 SO reductase family of MGD-containing molybdoproteins are shown in Fig. 4. In addition to Me 2 SO reductase from organisms such as R. sphaeroides (16) and R. capsulatus (17), the family also comprises BSOR from such organisms as E. coli (18), Helicobacter pylori (19), and Haemophilus influenza (20) and other E. coli enzymes, such as trimethylamine-N-oxide reductase (21), the dissimilatory forms of nitrate reductase (22,23), and formate dehydrogenase (24,25). Within the N-terminal portion of these sequences, the residue corresponding to Ser 147 in Me 2 SO reductase is strongly conserved as a serine, cysteine, or selenocysteine (SeC) residue, indicating that in proteins belonging to the Me 2 SO reductase family of Mo-containing enzymes, the putative Mo ligand corresponds to either an oxo or sulfido group derived from an S, C, or SeC side chain.
The active site of BSOR has recently been examined using a combination of electron paramagnetic resonance, resonance Raman, and x-ray absorption spectroscopies (10,11). The results of these studies indicate that BSOR has a very similar Mo site architecture to Me 2 SO reductase, with four thiol ligands donated by the two MGD cofactors and an oxo ligand donated by a serine residue, which has been proposed from sequence alignments to correspond to Ser 121 . A proposed catalytic scheme for BSOR, derived from the resonance Raman studies (11), has indicated that Ser 121 remains coordinated to the Mo center both in the oxidized and reduced states throughout the catalytic cycle of the enzyme. The results of our mutagenesis studies are in agreement with this active site model and suggest that the oxo group of Ser 121 provides the fifth ligand to the Mo center.
Recent site-directed mutagenesis studies of Ser 147 in Me 2 SO reductase have demonstrated that replacement of Ser 147 by Cys resulted in the production of a Me 2 SO reductase mutant that retained significant functional activity with both Me 2 SO and trimethylamine-N-oxide. Whereas substantial decreases in catalytic activities with alternate substrates such as methionine sulfoxide and BSO were observed, the S147C mutant was shown to exhibit enhanced activity (400% increase compared with wild type) with adenosine N 1 -oxide as the oxidizing substrate, suggesting a change in substrate specificity (13). Recent extended x-ray absorption fine structure studies (26) have also indicated that the side chain sulfur of the S147C variant functions as a molybdenum ligand, conferring partial activity.
In contrast, replacement of Ser 121 in BSOR with either Cys, Thr, or Ala resulted in the production of three nonfunctional mutants that were catalytically inactive when assayed spectrophotometrically with all of the oxidizing substrates utilized by the wild-type enzyme, suggesting that no alteration in substrate specificity occurred. In addition, HPLC analysis of the reaction products generated by the wild-type and mutant enzymes also clearly demonstrated that biotin formation was absent in the reactions catalyzed by all three mutants. We analyzed for product formation, rather than monitoring the disappearance of reduced methyl viologen, to avoid potential problems due to incomplete anaerobiosis during the assay. These results also confirmed that a functional Mo site is also required for both NADPH:BSOR and NADPH:FR activities, the latter representing a partial activity that is unique to R. sphaeroides BSOR.
Site-directed mutagenesis has also been utilized to examine FIG. 3. MGD cofactor analysis from wild-type and Ser 121 mutant BSOR. The molybdenum cofactor was isolated from 150 g of wild-type BSOR (-), 90 g of S121A (-), 85 g of S121T (-⅐-), and 70 g of S121C (-) mutant BSOR, as described under "Materials and Methods." Fluorescence excitation spectra were recorded using an emission wavelength of 460 nm, and emission spectra were recorded using an excitation wavelength of 370 nm. the role of the putative Mo ligand in other members of the MGD-containing Me 2 SO reductase family with varied results. Generation of the S176C mutant in the catalytic (dmsA) subunit of E. coli Me 2 SO reductase resulted in the production of an inactive enzyme (27), whereas generation of the SeC141C mutant in E. coli formate dehydrogenase resulted in production of an enzyme with decreased activity (28). The variety of effects of residue substitution on the catalytic activities of members of the Me 2 SO reductase family of enzymes suggests that the nature of the fifth ligand, derived from the protein side chain in the coordination environment of the Mo center, plays a pivotal role in regulating the functionality of the metal center. Changes in the coordination chemistry of the Mo center in these enzymes could have significant impact on the oxidationreduction potentials of the Mo VI /Mo V and Mo V /Mo IV redox couples, could also alter the accessible redox states available during turnover or could change the conformation of the active site.
Our results indicate that Ser 121 is an essential amino acid for functionality in BSOR. However, whereas this residue functions as a coordinating ligand to the Mo, it is not crucial for retention of either Mo coordination or MGD binding because all the mutants retained both cofactor and Mo. Additional spectroscopic studies of the Ser 121 mutants will be required to determine whether the Cys or Thr variants fail to provide a suitable fifth ligand for molybdenum coordination, resulting in loss of activity.