Identification and Biochemical Characterization of Arabidopsis thaliana Sulfite Oxidase A NEW PLAYER IN PLANT SULFUR METABOLISM*

In mammals and birds, sulfite oxidase (SO) is a homodimeric molybdenum enzyme consisting of an N-ter-minal heme domain and a C-terminal molybdenum domain (EC 1.8.3.1). In plants, the existence of SO has not yet been demonstrated, while sulfite reductase as part of sulfur assimilation is well characterized. Here we report the cloning of a plant sulfite oxidase gene from Arabidopsis thaliana and the biochemical characterization of the encoded protein (At-SO). At-SO is a molybdenum enzyme with molybdopterin as an organic compo-nent of the molybdenum cofactor. In contrast to homologous animal enzymes, At-SO lacks the heme domain, which is evident both from the amino acid sequence and from its enzymological and spectral properties. Thus, among eukaryotes, At-SO is the only molybdenum enzyme yet described possessing no redox-active centers other than the molybdenum. UV-visible and EPR spectra as well as apparent K m values are pre- sented and compared with the hepatic enzyme. Subcellular analysis of crude cell extracts showed that SO was mostly found in the peroxisomal fraction. In molybdenum supplemented This diluted 1:50 into 2 (cid:4) YT medium containing 1 m M Na 2 MoO 4 and 0.1 m M isopropyl- (cid:2) -thiogalactoside. Aerobic growth was continued for 24 h at 22 °C. Purification of soluble rAt-SO was performed at 4 °C by using Ni 2 (cid:2) -nitrilotriacetic acid Superflow matrix (Quiagen). Purified protein was analyzed by SDS-PAGE on 12.5% polyacrylamide gels, and pure fractions were dialyzed against 20 m M Tris/HCl, pH 8.0, 1 m M EDTA, concentrated by ultrafiltration using a YM10 membrane (Amicon), sterile filtrated through 0.2- (cid:3) m mem- branes, and stored at 4 °C or in aliquots at (cid:1) 70 °C. The protein prep-aration was stable at 4 °C for several weeks after dialyzing and concen- trating to 2–10 mg/ml. Concentration of rAt-SO was determined by UV spectroscopy using the calculated extinction coefficient of (cid:4) (cid:5) 69.820 M (cid:1) 1 cm (cid:1) 1 . Polyclonal antibodies were generated against purified rAt-SO protein (Eurogentec).

Molybdenum plays an important role as the active center in molybdenum enzymes catalyzing essential redox reactions in the global C, N, and S cycles (1). Molybdenum enzymes are important for such diverse metabolic processes as sulfur detoxification and purine catabolism in mammals (2), nitrate assimilation in autotrophs, and phytohormone synthesis in plants (3). With the exception of nitrogenase, in all molybdenum enzymes studied so far, molybdenum is part of a so-called molybdenum cofactor (Moco). 1 Sulfite oxidase (SO) from vertebrate sources is among the best studied molybdenum enzymes. Mammalian and avian SO (EC 1.8.3.1) is a homodimeric protein consisting of an N-terminal heme domain and a C-terminal Moco domain (4,5), and its crystal structure has been solved (6). SO catalyzes the reaction SO 3 2Ϫ ϩ H 2 O 3 SO 4 2Ϫ ϩ 2H ϩ ϩ 2e Ϫ , which is the important terminal step in the oxidative degradation of cysteine, methionine, and membrane components such as the sulfatides. Deficiency of this enzyme in humans leads to major neurological abnormalities and early death (7). In mammals, SO is localized in the intermembrane space of mitochondria (8), where electrons derived from sulfite are passed on to cytochrome c, the physiological electron acceptor. SO has been extensively studied by EPR (9) and resonance Raman spectroscopy (10) as well as rapid kinetic methods, and a detailed model for the reaction mechanism has been published (1). Together with eukaryotic assimilatory nitrate reductase (NR), SO forms a family of molybdenum enzymes that contain a dioxomolybdenum center (1,2). Like SO, NR is a homodimeric enzyme with a Moco domain, a heme domain, and a third FAD-binding domain that does not occur in SO (2).
In plants, during primary sulfate assimilation in the chloroplast, sulfate is reduced via sulfite to organic sulfide, which is essential for cysteine biosynthesis (11). However, there are reports that sulfite can be reoxidized to sulfate (e.g. when plants are subjected to SO 2 gas (as reviewed in Ref. 12) or when isolated chloroplasts are fed with radioactively labeled sulfite) (13). Sulfite oxidation in intact chloroplasts is enhanced by light and sensitive to inhibitors of photosynthetic electron transport (13,14) and thus was interpreted to be primarily due to nonenzymatic reactions during electron transport. Another sulfite-oxidizing activity has been found to be associated with thylakoid membranes (15), although this enriched enzyme fraction exhibited properties that were different from the hepatic enzyme. Another report describes the purification of a SO activity from the plant Malva sylvestris (16); however, the enzyme had a molecular mass of only 27 kDa.
Here we report the cloning of the plant SO gene from Arabidopsis thaliana (At-SO) and the biochemical characterization of the encoded protein. Additionally, native SO was purified from Nicotiana tabacum (Nt-SO) leaves, and its properties were compared with At-SO. At-SO is a molybdenum enzyme with molybdenum-MPT as molybdenum cofactor. In contrast to animal SO, At-SO lacks the heme domain, which is evident both from the amino acid sequence and from its enzymological and spectral properties. Thus, among eukaryotes, At-SO is the only known molybdenum enzyme lacking other redox-active centers. Another difference from animal SO is that At-SO is localized mostly in the peroxisomal fraction. Finally, using antibodies directed against At-SO, we show that a cross-react-ing protein of similar size occurs in a wide range of plant species belonging to herbacious and woody plants.

EXPERIMENTAL PROCEDURES
Screening of Genomic Library-The expressed sequence tag (Gen-Bank TM accession number AV 540461) was used to screen an A. thaliana cDNA library ( YES), resulting in the isolation and identification of a cDNA encoding for A. thaliana SO (At-sox) which was subcloned into pBluescript SKϩ (Stratagene), yielding pBS-At-sox. Hybridizations and Southern blotting were performed according to standard molecular biology techniques. The full-length At-sox cDNA (1443 base pairs) showed a single open reading frame of 1182.
Plasmids-For overexpression and purification of recombinant A. thaliana SO (rAt-SO), the QIAexpress kit (Qiagen) for metal ion chelating chromatography of His-tagged proteins was used. The His 6 tag was fused to the C-terminal end of the protein separated by two additional amino acids. At-sox was PCR-cloned into NcoI-and BglII-digested pQE-60, resulting in pQE60-At-sox. Initial experiments were performed by using pQE60-At-sox to express recombinant At-SO (rAt-SO) in Escherichia coli M15 cells. However, in order to express large amounts of protein with a high saturation of Moco, the expression cassette of pQE60-At-sox (EcoRI-HindIII fragment) was subcloned into pQE80, yielding pQE80-At-sox, allowing a pREP4-independent expression.
Protein Expression and Purification-Recombinant At-SO (rAt-SO) was expressed from pQE80-At-sox in the E. coli strain TP1000 (genotype ⌬mobAB), which was previously used for expression of recombinant human SO with a saturation of Moco (17). TP1000/pQE80-At-sox was grown aerobically at 37°C overnight in LB supplemented with antibiotics. This culture was diluted 1:50 into 2ϫ YT medium containing 1 mM Na 2 MoO 4 and 0.1 mM isopropyl-␤-thiogalactoside. Aerobic growth was continued for 24 h at 22°C. Purification of soluble rAt-SO was performed at 4°C by using Ni 2ϩ -nitrilotriacetic acid Superflow matrix (Quiagen). Purified protein was analyzed by SDS-PAGE on 12.5% polyacrylamide gels, and pure fractions were dialyzed against 20 mM Tris/HCl, pH 8.0, 1 mM EDTA, concentrated by ultrafiltration using a YM10 membrane (Amicon), sterile filtrated through 0.2-m membranes, and stored at 4°C or in aliquots at Ϫ70°C. The protein preparation was stable at 4°C for several weeks after dialyzing and concentrating to 2-10 mg/ml. Concentration of rAt-SO was determined by UV spectroscopy using the calculated extinction coefficient of ⑀ ϭ 69.820 M Ϫ1 cm Ϫ1 . Polyclonal antibodies were generated against purified rAt-SO protein (Eurogentec).
Molecular Weight Determination-Size exclusion chromatography was performed with an FPLC system (Amersham Pharmacia Biotech) using the analytical Superdex 200 column (Amersham Pharmacia Biotech). The column was equilibrated in 20 mM Tris/HCl, 200 mM NaCl, 1 mM EDTA, pH 8.0, and 100 l of purified rAt-SO was separated at a flow rate of 0.3 ml/min. The molecular weight was determined using a calibration curve obtained from the retention times of standard proteins (ovalbumin, 43 kDa; chymotrypsinogen A, 25 kDa; equine myoglobulin, 17 kDa; ribonuclease A, 14 kDa; vitamin B 12 , 1.35 kDa).
Purification of Plant SO from Leaves-Fresh N. tabacum leaves were ground in liquid nitrogen, resuspended in 3-5 ml of extraction buffer (100 mM HEPES buffer, 1 mM EDTA, 5% glycerol, 1 mM Na 2 MoO 4 , 1 mM phenylmethylsulfonyl fluoride, pH 7.3)/g of cell powder, and the homogenate was centrifuged at 21,000 ϫ g for 20 min. The supernatant designated as crude extract was directly applied to a Source Q30 anion exchange column that was previously equilibrated and washed with 5 bed volumes of extraction buffer. Subsequently, the enzyme was eluted with a gradient of 0 -1 M NaCl in extraction buffer of 15 bed volumes. Fractions were analyzed by SDS-PAGE and immunoblotting with the anti-At-SO-antibody generated in this study. Positive fractions were concentrated and subjected to size exclusion chromatography in 100-l aliquots using a Superdex 200 column (Amersham Pharmacia Biotech) equilibrated in 20 mM Tris-HCl, 200 mM NaCl, 1 mM EDTA, pH 8.0. Finally, the positive fractions were concentrated, rebuffered in 20 mM Tris/HCl, 1 mM EDTA, pH 8.0, and applied to a strong anion exchange MacroPrep Q10 m support (Bioscale Q2; Bio-Rad). The enzyme was eluted with a gradient of 0 -1 M NaCl in 20 bed volumes of extraction buffer.
SO Assays-SO was routinely assayed at 25°C in 20 mM Tris buffer, adjusted to pH 8.0 with acetic acid, containing 0.1 mM EDTA. The reaction was monitored in a Hewlett-Packard 8452A single beam diode array spectrophotometer following reduction of ferricyanide (⑀ ϭ 1020 M Ϫ1 cm Ϫ1 ) at 420 nm. One unit of SO activity was defined as the conversion of 1 mol of sulfite/min. Cytochrome c reductase activity was assayed as described (4). Protein content in crude protein extracts was determined as described (18).
Steady-state kinetic measurements were performed aerobically at 25°C using a 1.0-cm light path cuvette and a final sample volume of 1.0 ml. All experiments were performed with assay buffer (20 mM Tris acetate, pH 8.0). All experiments were performed using buffers in which the pH was adjusted with acetic acid; in each case, less than 35 mM acetic acid was used. A full steady-state kinetic study was conducted at pH 8.0 with fresh enzyme, using a saturating concentration of ferricyanide, 395 M (66-fold greater than K m ), and varying the concentration of sulfite between 2.5 and 400 M.
UV-visible and EPR Spectroscopy-The UV-visible absorption spectra of oxidized and reduced rAt-SO were recorded using a Hewlett-Packard 8452A diode array spectrophotometer and compared with the spectrum of oxidized and reduced sulfite oxidase from chicken liver (19) SO was reduced with sulfite or dithionite, and a wave scan was recorded between 300 and 550 nm. The electron paramagnetic resonance spectra were recorded with a Brü cker ER 300 spectrometer equipped with an ER 035 M gaussmeter and a Hewlett-Packard 5352 B microwave frequency counter. Temperature control at 150 K was achieved using a Brü cker ER 4111 VT continuous flow liquid nitrogen cryostat.
Cofactor Analysis-MPT was detected and quantified by converting it to the stable oxidation product, FormA-dephospho, according to Johnson and Rajagopalan (20). Oxidation, dephosphorylation, QAE chromatography, and HPLC analysis were performed as described in detail previously (21). FormA-dephospho was quantified by comparison with a standard isolated from xanthine oxidase (Sigma) for which the absorptivity was ⑀ 380 ϭ 13,200 M Ϫ1 cm Ϫ1 (20). Biologically active Moco was detected by the nit-1 assay in the absence of additional molybdate exactly as described earlier (22).
Cell Fractionation and Immunodetection of SO-Nicotiana plumbaginifolia leaves were extracted and fractionated as described (23) and separated in a 10% SDS-PAGE. For immunodetection, the polyclonal anti-rAt-SO antibody was affinity-purified against purified rAt-SO as described (24) and diluted 1:500 prior to use.

Arabidopsis SO Is a Single Domain Molybdenum Enzyme-
Using the amino acid sequence of human (XP_006727) or chicken SO (P07850), a TblastN search in GenBank TM identified an A. thaliana expressed sequence tag (GenBank TM AV 540461) that was sequenced and showed an open reading frame of 393 amino acids (43.3 kDa) with 47% identity to the primary sequence of the Moco domain of chicken SO (Fig. 1A). However, the sequence for the heme domain known from animal SO was lacking in this clone. To exclude the possibility that the identified cDNA was truncated at the 5Ј-end, the genomic structure of this gene encoding for a putative SO was analyzed (Gen-Bank TM BAC clone F28J7). The genomic sequence showed a single open reading frame with 11 introns, and it was located on chromosome III next to the RI marker mi74b at 5.8 centimorgans; however, no sequence encoding for a heme domain was found upstream or downstream of this gene. High stringency hybridization of Arabidopsis genomic DNA with the isolated cDNA clone as probe demonstrated that the gene encoding for the putative Arabidopsis SO (At-sox) was a single copy gene (data not shown). The cDNA sequence was deposited in GenBank TM (accession number AF200972) as plant SO cDNA.
The alignment of the protein sequences of Moco domains from animal SOs with sequences from plant SO and plant NR (Fig. 1A) shows that the putative At-SO has higher homologies to mammalian SO than to Arabidopsis NR. Since SO and NR belong to the same class of molybdenum enzymes, a significant conservation between the SO sequences shown and Arabidopsis NR can be seen (Fig. 1A, dark shadings). However, there are several SO-specific residues exclusively conserved between the members of the SO family, and most of them are present in the plant homologue as well ( Fig. 1A; gray shadings). For comparison, the domain structures of animal and plant SO and of NR are shown (Fig. 1B).
Purification and Properties of Recombinant SO-In order to provide evidence that the isolated cDNA encodes for a func-tional SO, the enzymatic properties of the gene product have been determined. For recombinant expression, the isolated Arabidopsis cDNA was cloned into a bacterial expression vector, allowing the expression and purification as His-tagged protein from E. coli. rAt-SO was purified to more than 98% homogeneity, and N-terminal sequencing of the purified protein revealed the correct translational start. The protein exhibited a sulfite dependent SO activity when using ferricyanide as electron acceptor. No activity was found with cytochrome c as electron acceptor (data not shown), as expected, since the heme domain is known to mediate electron transfer between the Moco domain and cytochrome c in rat hepatic SO (5). The K m value for sulfite in the ferricyanide assay ( Fig. 2A) was determined to be 33.8 Ϯ 3.2 M, which is in the same range as found for rat SO (33 M (19)). Like hepatic SO (25), rAt-SO could also be inhibited by increased ionic strength. The following buffers, salts, and the substrate analogue nitrate each caused a 50% inhibition at the given concentration: Tris/HCl (pH 8.5) at 90 mM, potassium phosphate (pH 8.5) at 30 mM, NaCl in 20 mM Tris/HCl (pH 8.5) at 70 mM, and potassium nitrate at 1 mM, respectively. rAt-SO was unable to reduce nitrate when using reduced methyl viologen as electron donor (data not shown). FPLC size exclusion chromatography of rAt-SO resulted in two distinct peaks. The major peak eluted at 14.30 ml, which corresponds to a molecular mass of 39.53 kDa, while the minor peak eluted at 12.61 ml, which corresponds to 109.37 kDa (Fig.  2B), Based on the size of At-SO predicted from the amino acid sequence (43.3 kDa), it can be concluded that the majority of the protein is monomeric.
In eukaryotic molybdenum enzymes, molybdenum is coordinated by an MPT-based cofactor. HPLC analysis of the MPT oxidation product FormA-dephospho confirmed the pterin nature of the Moco bound to rAt-SO (Fig. 2C). The ratio of MPT to rAt-SO monomer was determined to be 1:0.83. Without Moco- releasing treatment, rAt-SO was positive in the nit-1 reconstitution assay; rAt-SO donated Moco to the NR apoprotein in crude extracts of the Moco-deficient Neurospora crassa mutant nit-1, thereby restoring NR activity (data not shown), demonstrating the previous successful incorporation of the metalcontaining cofactor into rAt-SO. As will be shown below, also the spectroscopic properties of rAt-SO confirm that plant SO contains the same Moco as all other members of the SO family.
Spectral Properties of At-SO-The UV-visible absorption spectra of rAt-SO (Fig. 3A) showed the absence of a prosthetic heme group. The spectrum of the oxidized enzyme, with an absorption maximum at 360 nm and a shoulder at ϳ480 nm is very similar to the spectrum reported for the Moco domain of tryptically cleaved rat SO (5) and of the recombinant Moco domain of human SO (26). These absorption bands, attributable to cysteine-to-molybdenum and enedithiolate-to-molybdenum charge-transfer bands, respectively (27), are lost in the reduced enzyme.
The EPR spectra of rAt-SO at pH 10 and pH 6 upon partial reduction with sulfite are shown in Fig. 3B. At high pH, the EPR signal, with g 1,2,3 ϭ 1.991, 1.965, 1.958, is essentially identical to the signal exhibited by the chicken liver enzyme under the same conditions (g 1,2,3 ϭ 1.987, 1.9641, 1.953) (28). At low pH, the chicken enzyme shows a characteristic shift in g value (g 1,2,3 ϭ 2.004, 1.972, 1.966) and the appearance of hyperfine splitting due to one strongly coupled proton (a 1,2,3 ϭ 8.5, 8.0, 13.0 G) (28). With rAt-SO, however, the observed EPR spectrum was clearly a composite due to at least two EPRactive species, although the doublet to low field resembled that observed at low pH with the chicken enzyme. Repeated attempts under a variety of experimental conditions did not give a pure signal analogous to the low pH signal seen with the vertebrate forms of the enzyme, due at least in part to the instability of the plant enzyme below pH 6.0 (data not shown). Furthermore, no EPR signal comparable with the phosphateinhibited signal observed at more moderate pH with the chicken enzyme (28) is seen with the Arabidopsis protein (data not shown), even in 0.1 M phosphate buffer. On the basis of the UV-visible absorption and EPR signal at high pH, however, it is evident that the molybdenum center of the At-SO is fundamentally similar to that of the vertebrate proteins. The lack of a phosphate-inhibited or pure low pH EPR signal is most likely due to differences in substrate binding and pK a values of active site residues, respectively.

SO Protein Can Be Detected in Herbacious and Woody
Plants-Using a polyclonal antiserum raised against rAt-SO, a single band at 45 kDa was detected in immunoblots of crude protein extracts of A. thaliana leaves, the size of which is identical to the size of the protein as predicted from the primary sequence (Fig. 4A, lane 1). A similar result was obtained when analyzing crude extracts of leaves from tobacco (N. tabacum), N. plumbaginifolia, pea, spinach, barley, carrot, and poplar trees (Fig. 4A). Thus, in representative species of monocot and dicot plants and of herbacious and woody plants, a protein is expressed with a size similar to rAt-SO and with immunologically relevant epitopes that are conserved over a wide evolutionary range. Assuming that the polyclonal serum generated against rAt-SO exclusively recognizes SO proteins in all other plants analyzed, one can conclude that plant SOs contain only a single Moco domain and that they are widely expressed.
SO Is the Major Sulfite-oxidizing Activity in Tobacco-In plants, previous studies showed that SO activities were detectable using ferricyanide as electron acceptor (15) or utilizing radioactively labeled sulfur compounds (14). If SO as a novel plant molybdenum enzyme is the only sulfite-oxidizing activity in plants, no SO activity should be measurable in Moco-deficient mutants. Plant Moco mutants are characterized by the loss of activity of all molybdenum enzymes (i.e. NR, xanthine dehydrogenase, and aldehyde oxidase) (29). In A. thaliana, Moco mutants derived from five Moco-specific loci are known; however, all of these mutants are leaky and exhibit relatively high background activities of molybdenum enzymes (30,31). Therefore, we chose the N. plumbaginifolia system, where nonleaky Moco mutants are available (32,33) that were used previously for studying Moco biosynthesis in plants (22,34). Determination of SO activity in nongreen cell suspension cultures of mutants defective in the Moco loci cnxA, cnxC, and cnxE, respectively, showed that these cells still had SO activities of 39 Ϯ 6% of wild type activity in the standard assay (Fig.   FIG. 3. UV-visible and EPR spectra of rAt-SO. A, UV-visible absorption spectra of oxidized (solid line) and sulfite-reduced (dashed line) enzyme (2.9 M). Spectra were recorded in 0.02 M Tris/HCl, pH 8.5. The 360-and 480-nm absorption bands of the oxidized enzyme are due to cysteine-to-molybdenum and enedithiolate-to-molybdenum chargetransfer bands, respectively. B, EPR spectra of rAt-SO partially reduced with sulfite in 20 mM glycine, pH 10 (upper) and 20 mM bis-tris propane, pH 6.0 (lower). Instrument settings were 9.458-GHz microwave frequency, 10-mW microwave power, 100-kHz modulation frequency, 5.0-G modulation amplitude, and 150 K. 4B). No NR, xanthine dehydrogenase, and aldehyde oxidase activities were detected in these cells (Refs. 32 and 34; data not shown). Mutants in the cnxA locus are defective in the insertion of molybdenum into MPT and can be therefore partially repaired by growing them in the presence of high molybdate (34,35). SO activity in cnxA mutants grown in a medium containing 1 mM molybdate was found to be increased to 65 Ϯ 8% of wild type activity (Fig. 4B), which is a direct correlation between the availability of Moco and the SO activity in plants. We conclude, therefore, that in N. plumbaginifolia about 60% of the cellular capacity to oxidize sulfite is due to the novel molybdenum enzyme SO. The nature of the residual sulfite-oxidizing activities remains unclear.
Subcellular Localization, Purification, and Properties of Native SO from Tobacco-In higher plants, the enzymes for primary sulfate assimilation are localized in the chloroplasts, where sulfate is reduced via sulfite to organic sulfide (11). Because SO would principally counteract the function of sulfite reductase, we investigated whether SO is also localized in the chloroplasts. Crude extracts of N. plumbaginifolia leaves were separated into their subcellular fractions and SO was immunodetected after SDS-PAGE of these fractions. Fig. 5A shows that SO could be found in the peroxisomal fraction but not in chloroplasts, which coincides with the presence of the peroxisome import motif SNL at the C terminus of the enzyme (see Fig. 1A). Compared with animal SO, these three additional amino acids are an extension of the C terminus that was found to be solvent-exposed in the crystal structure of chicken SO (6). The experiments were repeated with leaves from different developmental stages and also with leaves from Arabidopsis (data not shown), and always the same results as shown in Fig.  5A were obtained. So we conclude that SO is not localized in the chloroplast and therefore probably not interfering with primary sulfate assimilation.
After showing that in plant crude extracts SO activity can be detected that is dependent on the biosynthesis of Moco, we wanted to demonstrate that this enzymatic activity is identical to the Arabidopsis SO that was identified, recombinantly expressed, and biochemically characterized above. Therefore, native Nt-SO was purified to more than 98% homogeneity by applying ion exchange chromatography followed by FPLC size exclusion chromatography (Fig. 5B), resulting in an ϳ100-fold enrichment of the protein. N-terminal sequencing of the purified protein (Fig. 5C) revealed that out of 12 amino acids, 11 were identical to the N terminus of A. thaliana SO (Fig. 5D). The K m value for sulfite was determined to be 51.4 M (Fig. 5E), which is in the same range as found for rAt-SO (33.8 M).

DISCUSSION
During our molecular studies of molybdenum metabolism in plants we isolated a cDNA from A. thaliana encoding for a protein with high homologies to animal SO. The plant sequence encodes for a SO-type Moco domain, but surprisingly the heme domain was lacking. Analysis of the genomic structure of this gene proved the absence of any sequence coding for a heme-like domain, which is consistent with 1) the absence of cytochrome c-dependent SO activity of both recombinantly expressed and native SO, 2) the absence of a heme-specific UV-visible absorption spectrum, and 3) the observed molecular mass (45 kDa) of the immunologically cross-reacting protein band in different plant species. Is it possible that such a single domain SO protein is functional in vivo? The crystal structure of chicken SO (6) suggests that the heme domain and the Moco domain need not necessarily be connected covalently. The disordered structure of the hinge region linking the molybdenum domain and the clearly separated heme domain indicates a high mobility for this part of the protein (36). Together with the long distance between the molybdenum center and the electron acceptor (32 Å), alternative arrangements of the heme domain relative to the Moco domain may be possible, suggesting a very flexible positioning of the heme domain. It is therefore possible that in plants the function of the SO electron acceptor could be taken over by a separately expressed protein. Further, depending on the nature of the still unknown native electron acceptor for plant SO, it might well be that a heme-like protein mediating electron transport between SO and the native acceptor protein is unnecessary. This suggestion gains support from the purification experiments, where native SO purified from N. tabacum contained only the single Moco domain. No evidence was found for the copurification of another heme-like protein, while in bacteria a separately expressed heme domain was copurified with the Moco domain of SO from Thiobacillus novellus (37). For functionality, however, it is very likely that plant SO needs an electron acceptor protein of a redox potential similar to a heme. Taking into account that our subcellular fractionation experiments indicated a peroxisomal localization for plant SO, the b-type cytochromes recently described for plant peroxisomes (38) are a likely candidate to fulfill the role of physiological electron acceptor for plant SO.
Among eukaryotes, plant SO is the smallest molybdenum enzyme known up to now and the only one lacking other redoxactive centers. The alignment of Moco domains of SOs from different sources with Arabidopsis SO and NR (Fig. 1) demonstrates considerable overall homology, identifying these enzymes as members of a common family, the so-called SO family (2). Arabidopsis is the first organism where within one organism both members (SO and NR) of this family are present, which makes it a unique system for studying structure-function relations of a molybdenum domain that can be part of an oxidase or of a reductase.
The enzymatic and spectral properties of At-SO are so similar to the Moco domain of the animal counterpart that it is justified to assume that this enzyme functions also in vivo as an SO. This view is supported by measuring SO activity in Mocodeficient N. plumbaginifolia cnx mutants. In those mutants, SO activity is strongly decreased as compared with the wild type, indicating that the molybdenum enzyme SO functions also under in vivo conditions as SO. This enzyme is conserved among higher plants as evidenced by the fact that antibodies raised against rAt-SO detected a dominantly cross-reacting protein of about 45 kDa in a wide range of species belonging to a variety of both herbacious and woody plants.
Although SO catalyzes more than half of the cellular sulfite oxidation, there is a residual activity detected in Moco-deficient N. plumbaginifolia cnx mutants. This activity may either be nonenzymatic as described by Dittrich et al. (13) for electron transport in chloroplasts or due to the enzymatic fraction that was described in wheat chloroplasts (14,15). However, if sulfite oxidation also takes place in chloroplasts, it has to be strictly regulated in order to avoid conflicts with the sulfate assimilation pathway that is vital for the survival of the plant cell. We found At-SO in the peroxisomal fraction rather than in chloroplasts, which suggests that the function of SO is not related to the chloroplast-based sulfur assimilation pathway. Rather it may be that plant SO has a sulfite-detoxifying function. For example, it has been shown that peroxisomal catalase is inhibited already by low concentrations of sulfite (39). Studies are presently under way to further elucidate the physiological role of SO in plants, as well as their catalytic, spectroscopic, and structural properties.