3-Mercaptopropionate Dioxygenase, a Cysteine Dioxygenase Homologue, Catalyzes the Initial Step of 3-Mercaptopropionate Catabolism in the 3,3-Thiodipropionic Acid-degrading Bacterium Variovorax paradoxus*

The thioether 3,3-thiodipropionic acid can be used as precursor substrate for biotechnological synthesis of 3-mercaptopropionic acid-containing polythioesters. Therefore, the hitherto unknown catabolism of this compound was elucidated to engineer novel and improved polythioester biosynthesis pathways in the future. Bacteria capable of using 3,3-thiodipropionic acid as the sole source of carbon and energy for growth were enriched from the environment. From eleven isolates, TBEA3, TBEA6, and SFWT were morphologically and physiologically characterized. Their 16 S rDNAs and other features affiliated these isolates to the β-subgroup of the proteobacteria. Tn5::mob mutagenesis of isolate Variovorax paradoxus TBEA6 yielded ten mutants fully or partially impaired in growth on 3,3-thiodipropionic acid. Genotypic characterization of two 3,3-thiodipropionic acid-negative mutants demonstrated the involvement of a bacterial cysteine dioxygenase (EC 1.13.11.22) homologue in the further catabolism of the 3,3-thiodipropionic acid cleavage product 3-mercaptopropionic acid. Detection of 3-sulfinopropionate in the supernatant of one of these mutants during cultivation on 3,3-thiodipropionic acid as well as in vivo and in vitro enzyme assays using purified protein demonstrated oxygenation of 3-mercaptopropionic acid to 3-sulfinopropionate by this enzyme; cysteine and cysteamine were not used as substrate. Beside cysteine dioxygenase and cysteamine dioxygenase, this 3-mercaptopropionic acid dioxygenase is the third example for a thiol dioxygenase and the first report about the microbial catabolism of 3-mercaptopropionic acid. Insertion of Tn5::mob in a gene putatively coding for a family III acyl-CoA-transferase resulted in the accumulation of 3-sulfinopropionate during cultivation on 3,3-thiodipropionic acid, indicating that this compound is further metabolized to 3-sulfinopropionyl-CoA and subsequently to propionyl-CoA.

The thioether 3,3-thiodipropionic acid (TDP) 2 and its ester are effective non-toxic antioxidants (1), and they are therefore widely used as antioxidant and stabilizer in food, for food packaging, and for various technical applications. Experiment with rats showed that TDP was rapidly adsorbed after oral intake and excreted in the urine (2). In technical applications esters of TDP are important stabilizers of polyolefins (1), and polymer-bound TDP is used to replace methyl sulfide for the reductive quenching of ozonolysis reactions (3). Recently, the biotechnological production of medium-and long-chain dialkyl 3,3-thiodipropionate antioxidants by a lipase-catalyzed esterification of 3,3thiodipropionic acid in the absence of solvents was reported. In contrast to the chemical production of TDP ester, the biotechnological process does not require any materials with deleterious effects on health and environment (4). Another biotechnological process using TDP as primary product is the microbial production of polythioesters (PTEs) (5). In addition to 3-mercaptopropionic acid Ralstonia eutropha is able to use the organo sulfur compounds TDP and 3,3-dithiodipropionic acid (DTDP) as precursor substrates for production of copolymers of 3-hydroxybutyrate and 3MP (6). In contrast to 3MP the application of TDP and DTDP has numerous advantages, because they have a lower toxicity and they are odorless, inexpensive, and available on a large scale. Until today the use of TDP and DTDP as precursor substrates are limited to R. eutropha, and biotechnological production of PTE using the recombinant Escherichia coli strain JM109 pBPP1 (7) is only possible when 3MP is added to the media. Because 3MP is incorporated into the polymer if TDP or DTDP is supplied as precursor substrate, it is assumed that these compounds are enzymatically cleaved into 3MP and 3-hydroxypropionate, or two molecules 3MP, respectively (6). The corresponding TDP-and/or DTDPcleaving enzymes as well as the microbial catabolism of the intermediate 3MP are still unknown. The identification of such enzymes could help to engineer the recombinant E. coli JM109 pBPP1 toward TDP-and DTDP-based PTE production. Therefore, the corresponding genes could be used to improve the already established BPEC pathway by heterologous expression (7). For 3MP it is known that it occurs naturally as an intermediate during microbial degradation of the osmoprotectant dimethylsulfoniopropionate and during the biotransformation of the sulfur-containing amino acids methionine and homocysteine in anoxic coastal sediments (8 -11). However, to our knowledge no reports on the pathways or enzymes for further metabolism of 3MP in bacteria have been published. In contrast, the catabolism of cysteine, the structural analogue of 3MP, is well known in bacteria (12,13). In addition to the well investigated cysteine degradation pathways, a novel pathway was recently reported in eubacteria by Dominy et al. (14), which involves a cysteine dioxygenase (Cdo, EC 1. 13.11.20). This Fe 2ϩdependent enzyme catalyzes the irreversible oxidation of the sulfhydryl group of cysteine to cysteine sulfinic acid. Cdos are well known in eukaryotes and play an important role by reducing the cysteine pool and increasing the levels of important metabolites such as taurine and sulfates (15). The physiological function of this enzyme is not completely understood in bacteria. In addition to the important role in regulating the steadystate cysteine levels, also an important role of this enzyme during changes of bacterial life cycles was suggested (14), because many of the bacteria possessing a Cdo undergo a complex life cycle involving morphological changes. For example, in cells of Bacillus subtilis expression of the cdo gene is up-regulated during transition from the vegetative state to sporulation (14).
To our knowledge there are so far also no reports about bacteria that utilize the organic thioether TDP as sole source of carbon and energy. Only its use as a sulfur source was described for Mycobacterium goodii X7B (16). However, this bacterium could not grow with TDP in the absence of an additional carbon source, and degradation products of TDP were also not reported. To engineer the PTE biosynthesis in the future toward to use TDP as precursor substrates, we investigated the catabolism of TDP. The isolation and characterization of bacteria capable of using TDP as the sole source of carbon and energy are described in this study. In addition, Tn5::mob mutagenesis was carried out with one of these new isolates, Variovorax paradoxus strain TBEA6, to identify genes possibly involved in TDP catabolism.

EXPERIMENTAL PROCEDURES
Isolation of Strains Capable of Using TDP as Sole Source of Carbon and Energy-Samples from different soils, activated sludge, and freshwater tank sediment were incubated in mineral salts medium (MSM) containing 3 g/liter TDP as the sole source of carbon and energy at 30°C for 3-5 days (17). Aliquots of these cultures were then plated on the same medium solidified with 1.5% (w/v) agar, single colonies were isolated, and the best growing colonies were chosen for further studies.
Bacterial Strains and Cultivation Conditions-All bacterial strains used in this study are listed in supplemental Table S1. R. eutropha H16, strains of Variovorax paradoxus and isolate TBEA3 were cultivated at 30°C in nutrient broth or MSM supplemented with 0.1 g/liter yeast extract under aerobic conditions on a rotary shaker at an agitation of 130 rpm. Strains of E. coli were cultivated in Luria-Bertani (LB) medium or M9 medium supplemented with yeast extract (0.1 g/liter) at 37°C under the same conditions (18). Carbon sources were supplied as filter-sterilized stock solutions as indicated in the text. For maintenance of plasmids, antibiotics were prepared according to Sambrook et al. (18) and added to the media at the following concentrations (g/ml): ampicillin (75), kanamycin (50), chloramphenicol (34), and tetracycline (12.5). In E. coli heterologous expression of genes under the control of a lac-promotor was induced by addition of 1 mM IPTG to LB medium.
Chemicals-Bulk TDP was provided by Bruno Bock Chemische Fabrik GmbH & Co. KG. Organic thiochemicals of high purity grade were purchased from Acros Organics (Geel, Belgium) or Sigma-Aldrich (Steinheim, Germany) (Fig. 1). 3-Sulfinopropionate was synthesized according to Jollés-Bergéret (19); the procedure was modified by one repetition of the alkaline cleavage of the intermediate bis- (2-carboxyethyl)sulfone. Starting from 111 g of sodium formaldehyde sulfoxylate (purity, Ͼ98%) plus 108 ml of acrylic acid (99.5%), 119 g of the intermediate bis- (2-carboxyethyl)sulfone were chemically synthesized. After alkaline scission, precipitation, and washing procedures, 99 g of the disodium salt of 3SP, with a purity of ϳ90%, were finally obtained. Synthesis and purity of the substance was confirmed by HPLC and GC/MS. Strain Identification-API 20NE identification test (bio-Mérieux, Marcy-l'Etoile, France) and Bactident oxidase test stripes (Merck KgA, Darmstadt, Germany) were used according to the manufacturer's instructions. Presence of catalase was tested using 3% (v/v) H 2 O 2 .
The 16 S rRNA gene was amplified from total genomic DNA by PCR using Primers 27f and 1525r (20). The PCR product was purified using the NucleoTrap kit (Machery and Nagel, Düren, Germany) and applied as template for sequencing in which the following primers were utilized: 27f, 357f, 803f, 907r, 1114f, 1385r, and 1525r (20). The assembled sequence was compared with the GenBank TM data base and the Ribosomal Database Project using the Blast and SIMILARITY-RANK (Ribosomal Database Project) algorithms (21,22). A phylogenetic tree was constructed using ClustalX (23).
Transposon Mutagenesis and Characterization of Tn5::mobinduced Mutants-Insertional mutagenesis of V. paradoxus strain TBEA6 with transposon Tn5::mob was performed as described previously by using the suicide plasmid pSUP5011, which was delivered from E. coli S17-1 to the recipient by conjugation during spot agar mating (24,25). Transconjugants were selected on MSM plates containing 0.5% (w/v) gluconate plus kanamycin. Transconjugants impaired in growth on TDP were then identified by plating on MSM agar plates containing 0.3% (w/v) TDP plus kanamycin or 0.5% (w/v) gluconate plus kanamycin. For genotypic characterization of the Tn5::mob-induced mutants, genomic DNA was isolated (26) and digested with SalI or BamHI. The genomic DNA fragments were then ligated to pBluescriptSK Ϫ DNA, which was linearized with the same restriction endonuclease; the ligation products were then transformed into CaCl 2 -competent E. coli Top10 cells. Transformants were selected on LB medium containing ampicillin plus kanamycin, and hybrid plasmids were subsequently isolated and sequenced using the primers M13 forward, M13 reverse, and IS50L (supplemental Table S3).
Analysis of Cell-free Supernatants and Sulfur Organic Compounds by HPLC-Concentrations of TDP, 3MP, cysteine sulfinic acid, cysteamine, hypotaurine, and 3SP were analyzed by HPLC.
HPLC analysis of TDP, 3MP, cysteamine, and 3SP was carried out in a LaChrom Elite HPLC apparatus (VWR-Hitachi International GmbH, Darmstadt, Germany) consisting of a Metacarb 67H advanced C column (Varian, Palo Alto, CA, Bio-Rad Aminex equivalent) and a 22350 VWR-Hitachi column oven. The primary separation mechanism includes ligand exchange, ion exclusion, and adsorption. A VWR-Hitachi refractive index detector (Type 2490) with an active flow cell temperature control and automated reference flushing eliminating temperature effects on the refractive index baseline was used for detection. Aliquots of 20 l of cell-free supernatants, solutions of organic sulfur compounds or enzyme assay were injected and eluted with 0.005 N sulfuric acid (H 2 SO 4 ) in double distilled water at a flow rate of 0.8 ml/min. Online integration and analysis was done with EZ Chrome Elite Software (VWR International GmbH, Darmstadt, Germany). Cysteamine was detected under the same conditions using double distilled water as mobile phase. Detection of hypotaurine and cysteine sulfinic acid was carried out in a Kontron Instrument (Neu-fahrn, Germany). After derivatization with OPA reagent (27) using a Smartline Autosampler 3900 (Knauer Advanced Scientific Instruments, Berlin, Germany), 20 l of the reaction was injected onto a Novapack C18 reversed-phase column (Knauer) and monitored fluorometrically at 330/450 nm (excitation/ emission) by using a model 1046A fluorescence detector (Hewlett Packard). Substances were identified by comparison of their retention times to those of standard organic acids. The detection limit for hypotaurine is ϳ20 M and 10 M for cysteine sulfinic acid.
Quantitative Analysis of Polyhydroxyalkanoic Acid and Their Compositions by GC-Lyophilized cell material was subjected to methanolysis in the presence of methanol and sulfuric acid (MeOH: 85%, v/v; H 2 SO 4 : 15%, v/v) for 4 h at 100°C, and the resulting methylesters of the polyhydroxyalkanoic acid constituents were characterized by gas chromatography using an Agilent 6850 GC (Agilent Technologies, Waldbronn, Germany) equipped with a BP21 capillary column (50 m by 0.22 mm; film thickness, 250 nm; SGE, Darmstadt, Germany) and a flame ionization detector (Agilent Technologies). A 2-l portion of the organic phase was analyzed after split injection (split ratio, 1:5); a constant hydrogen flow of 0.6 ml/min was used as carrier gas. The temperatures of the injector and detector were 250°C and 220°C, respectively. The following temperature program was applied: 120°C for 5 min, increase of 3°C/min to 180°C, and increase of 10°C/min to 220°C and 220°C for 31 min. Substances were identified by comparison of their retention times to those of standard fatty acid methyl ester.
Analysis of 3SP Production by GC/MS-Lyophilized cells, cell-free supernatants, or aliquots of synthesized 3SP were subjected to methanolysis as described above, and the resulting methylesters of the organic acids were characterized by coupled GC/MS using an HP6890 gas chromatograph equipped with a model 5973 EI MSD mass-selective detector (Hewlett Packard). A 2-l portion of the organic phase was analyzed after splitless injection employing a BP21 capillary column (50 m by 0.22 mm; film thickness, 250 nm; SGE). Helium (0.6 ml/min) was used as carrier gas. The temperatures of the injector and detector were 250°C and 240°C, respectively. The same temperature program as described for GC analysis was applied. Data were evaluated using the NIST Mass Spectral Search program. 3 Isolation of RNA and RT-PCR-Total RNA was isolated from V. paradoxus strain TBEA6 by using the Qiagen RNeasy-Kit according to the manufacturer's instructions. RT-PCR was performed using the Qiagen "One step RT-PCR" Kit according to the manufacturer's instructions. To recognize PCR products based on DNA contaminations in isolated RNA, a control with addition of RNA after the reverse transcription step was done.
DNA Isolation and Manipulation-Chromosomal DNA of strains of V. paradoxus and R. eutropha H16 was isolated according to Marmur (26). Plasmid DNA was isolated from E. coli and V. paradoxus strains using the GeneJET TM plasmid miniprep kit from Fermentas (St. Leon-Rot, Germany) according to the manufacturer's manual. DNA was digested with restriction endonucleases under conditions described by the manufacturer or according to Sambrook et al. (18). PCR were carried out in an Omnigene HBTR3CM DNA thermal cycler (Hybaid, Heidelberg, Germany) using Platinum Taq DNA Polymerase (Invitrogen). PCR products were isolated from an agarose gel and purified using the NucleoTrap kit (Machery and Nagel, Düren, Germany) according to the manufacturer's instructions. T4-DNA-Ligase was purchased from Invitrogen. Primers were synthesized by MWG-Biotech AG (Ebersberg, Germany).
Transfer of DNA-Competent cells of E. coli strains were prepared and transformed by the CaCl 2 procedure (18).
DNA Sequencing and Sequence Data Analysis-DNA sequencing was done in a Li-Cor 4000L automatic sequencing apparatus (Li-COR Inv., Biotechnology Division, NE, USA) using the Thermo long read cycle sequencing Kit (Epicenter Technologies, WI, USA) and IRD 800-labeled oligonucleotides (MWG-Biotech, Ebersberg, Germany). BlastX was used for determination of nucleotide identity (21).
DNA-DNA Hybridization-Southern hybridization was carried out by the method described by Oelmüller et al. (29) at a temperature of 68°C.
Genome Walking-For sequencing of flanking genomic regions of known sequences a PCR-based directional genome walking method (30) was performed.
Cloning of cdo Vp , cdoA Re , and cdoB Re -The cdo Vp , cdoA Re , and cdoB Re genes were amplified from total genomic DNA of V. paradoxus strain TBEA6 or R. eutropha strain H16 by PCR using Taq DNA polymerase (Invitrogen) and the following oligonucleotides: cdo(NdeI), cdo(XhoI), cdoA(XbaI), cdoA(XhoI), cdoB(ApaI), and cdoB(HindIII) (supplemental Table S3). PCR products were isolated from agarose gels using the NucleoTrap kit (Machery and Nagel) and ligated with pGEMTeasy DNA (Promega, Madison, WI). Ligation products were transformed into CaCl 2 -competent cells, and transformants were selected on LB agar plates containing IPTG, X-Gal, plus ampicillin. For heterologous expression in the T7 promoter/polymerase-based expression vector pET23a (Novagen, Madison, WI), cdo Vp was obtained by digestion of hybrid plasmid pGEMTeasy::cdo Vp with restriction endonucleases NdeI and XhoI and purified from an agarose gel using the NucleoTrap kit (Machery and Nagel). After ligation into expression vector pET23a, which was linearized with the same restriction endonucleases, the ligation product was used for transformation of CaCl 2 -competent cells of E. coli Top10. After selection of transformants using LB media containing ampicillin, the hybrid plasmids were isolated, analyzed by sequencing, and transformed to CaCl 2 -competent cells of E. coli (DE3) strains BL21 pLysS and Rosetta pLysS (Novagen, Madison, WI).
For complementation studies and heterologous expression in the broad host vector pBBR1MCS-3 (31), cdo Vp , cdoA Re , and cdoB Re were obtained from pGEMTeasy vector, which were digested with the respective restriction endonuclease and purified from an agarose gel using the NucleoTrap kit (Machery and Nagel). The purified genes were subsequently ligated into pBBR1MCS-3, which was linearized with the same restriction endonucleases, and the ligation products were transformed to CaCl 2 -competent cells of E. coli S17-1 and E. coli Top10. Transformants were selected on LB medium containing tetracycline, IPTG, plus X-Gal. The hybrid plasmids pBBR1MCS-3::cdo Vp , pBBR1MCS-3:: cdoA Re , and pBBR1MCS-3::cdoB Re were then conjugated into the transposon-induced mutants 2/5 and 13/33 from E. coli S17-1.
Preparation of Crude Extracts-Cells from 50-to 500-ml cultures were harvested by centrifugation (20 min, 4°C, and 2,800 ϫ g), washed twice, and resuspended in 50 mM NaPO 4 buffer (pH 7.6). Cells were disrupted by sonification in a Sonopuls GM200 apparatus (Bandelin, Berlin, Germany) with an amplitude of 16 m (1 min/ml) while cooling in an NaCl/ice bath. Soluble protein fractions of crude extracts were obtained in the supernatants after 1-h centrifugation at 100,000 ϫ g and 4°C and were used for enzyme purifications.
Immobilized Metal Chelate Affinity Chromatography-To obtain purified hexahistidine-tagged fusion Cdo Vp , His Spin Trap affinity columns (GE Healthcare, Uppsala, Sweden) were used according to the instructions of the manufacturer with minor modifications. Tris-HCl (0.1 M, pH 7.6) was used as buffer component instead of sodium phosphate, and for the washing step a buffer containing 40 mM imidazole was applied. The washing step was repeated three times, and the elution step was repeated two times.
Enzyme Assay-Standard in vitro activity of cysteine dioxygenase was assayed by incubating 3 g of purified Cdo Vp for 30 min at 30°C in the presence of the following components: 10 mM cysteine, 10 mM cysteamine, or 5 mM 3MP, 400 M (NH 4 ) 2 Fe(SO 4 ) 2 ϫ 6H 2 O, 12.5 M bathocuproine disulfonate and MES buffer (62 mM, pH 6.3). The reaction was stopped by 10-min incubation at 95°C. Negative controls were done with denatured protein. The reaction products 3SP, hypotaurine, and cysteine sulfinic acid were analyzed by HPLC.
For in vivo testing of recombinant cysteine dioxygenase activity in recombinant E. coli strains, cells were cultivated in M9 medium containing 1% (v/v) glycerol at 30°C. Expression of the recombinant protein was induced by addition of 1 mM IPTG after 6 h of cultivation. Subsequently, the substrate 3MP was added to a final concentration of 0.1% (v/v), and the cells were grown for additional 24 h before they were harvested and washed twice with 0.9% (w/v) NaCl. Finally, the cell pellets were lyophilized. Analysis of the reaction product 3SP was done by GC/MS. Data Deposition-Nucleotide sequences have been deposited in the GenBank TM data base under the following Gen-Bank TM accession numbers: EF641108, 16 S rDNA gene of isolate TBEA6; EU825700, 16 S rDNA gene of isolate TBEA3; EU441166, 16 S rDNA gene of isolate SFWT; EU441167, contiguous sequence comprising bugC, fox, bugA, cdo, and ahpD; and EU449952, contiguous sequence comprising act, acd, and partial sequence of bugB.

RESULTS
Isolation of TDP-utilizing Bacteria and Taxonomic Affiliation of Isolate TBEA6-From different samples taken from soil, compost, sewage sludge, or the sediment of a fresh water tank, bacterial strains capable of using TDP as sole source of carbon and energy for growth were enriched and isolated (supplemental Table S1). The best growing isolates were designated as

3-Mercaptopropionate Dioxygenase in V. paradoxus
TBEA6, TBEA3, and SFWT; they were further characterized by methods of polyphasic taxonomy and by analysis of the 16 S rDNA sequences to unravel their phylogenetic position. All three isolates were Gram-negative, oxidase, and catalase-positive, and best growth was observed at 30°C, although isolates TBEA6 and SFWT grew also very slowly at 4°C. Colonies of TBEA6 and SFWT were yellow-pigmented, and the motile cells are short rods with a cell length of 1.5 m. Using the API 20NE identification test system, reduction of nitrate to nitrite and positive assimilation of L-arabinose, D-mannitol, gluconate, and malate were noticed for these two strains. Positive urease reaction was only identified for isolate SFWT. The API 20NE test system identified isolates TBEA6 and SWFT as strains of Ralstonia pickettii and TBEA3 as a strain of Comamonas acidovorans. The 16 S rRNA sequences of strains TBEA6 and SFWT showed high levels of sequence similarities (99%) to various V. paradoxus strains. V. paradoxus is not yet contained in the available API 20NE test system. The 16 S rDNA nucleotide sequence of strain TBEA3 exhibited the highest sequence similarity (98.9%) with those of an uncultured bacterium. Fig. 2 shows a phylogenetic tree providing an idea on the phylogenetic classifications of the three isolates.
Physiological Characterization of Tn5::mob-induced Mutants of V. paradoxus Strain TBEA6 Showing Impaired Growth on TDP-Although all three isolates were able to use TDP as sole source of carbon and energy for growth, no growth was observed with 3MP, or DTDP, or the putative cleaving product 3-hydroxypropionate. All three isolates utilized also the organic sulfur compounds taurine and 3SP, whereas homocysteine was used for growth only by isolate TBEA6. Cysteamine (0.1%; w/v) was also tested, but none of the investigated strains used this compound as carbon and energy source. Good growth was observed for strain SFWT when 0.1 (w/v) cysteamine was added to MSM agar plates containing gluconate (1%; w/v), whereas strain TBEA6 showed weak growth. All strains accumulated poly(3-hydroxybutyrate), but no accumulation of PTE such as poly(3MP) or poly(3MP-co-3-hydroxybutyrate) occurred as revealed by GC analysis of cells cultivated under conditions permissive for synthesis and accumulation of PTEs.
Both V. paradoxus strains exhibited good access to standard methods of molecular biology. They could for example be transformed by the broad host vector pBBR1MCS-3 (31).
To analyze if these mutants are still able to degrade TDP, or if they are impaired in the utilization of intermediates of TDP catabolism and to identify putative intermediates, mutants and wild type were grown in Erlenmeyer flask without baffles containing 1% (w/v) TDP plus 0.2% (w/v) sodium succinate as carbon sources in 300 ml of MSM. Samples were taken every 48 h, and aliquots of cell-free supernatants were analyzed by HPLC (Fig. 3). The initial concentration of TDP was decreasing in all cultures except in those of mutants 1/8 and 1/20. In the supernatant of mutant 1/1 accumulation of 3SP was observed by HPLC and GC/MS analysis, and the conversion of TDP to 3-sulfinopropionate was analyzed quantitatively in an additional experiment. Therefore, the wild type and this mutant were cultivated in an Erlenmeyer flask with baffles containing 0.2% (w/v) sodium succinate plus 1% (w/v) TDP in 50 ml of MSM, and the cell-free supernatants were analyzed by HPLC for contents of TDP and 3SP every 24 h (Fig. 4). The results of this experiment (Fig. 4A) clearly demonstrated that the increasing concentrations of 3SP during the time course of the experiment with the mutant correlated with decreasing concentrations of TDP. Whereas the wild type utilized TDP at a rate of 1.2 mM/h and did not form 3SP, mutant 1/1 utilized TDP at a rate of 0.32 mM/h and formed 3SP at a rate of 0.33 mM/h. For detection of other putative intermediates GC/MS analyses of the culture were performed. For this, 1 ml of the culture was taken every 24 h, and the cell-containing material was analyzed upon methanolysis. Beside 3SP, small amounts of 3-hydroxypropionate and 3MP could be detected in the culture of mutant 1/1 (Fig.  5A). Whereas 3MP and DTDP could also be detected in cultures of the wild type, identification of 3-hydroxypropionate and 3SP failed (Fig. 5B).
In contrast to the wild type and to all other mutants, mutant 1/1 was unable to use 3SP as sole source of carbon and energy for growth when provided in MSM. The mutant was grown in a baffled Erlenmeyer flask containing 0.5% (w/v) 3-sulfinopropionate, 0.5 (w/v) TDP, or 0.5% (w/v) sodium succinate as sole source of carbon and energy in 50 ml of MSM. Growth was quantified by measuring the optical density at 600 nm, and the concentrations of sodium succinate, TDP, and 3SP were determined by HPLC. The results are displayed in Fig. 4B and revealed that growth on 3SP and TDP is strongly affected in this

3-Mercaptopropionate Dioxygenase in V. paradoxus
mutant. Whereas sodium succinate was completely utilized during the first 24 h of cultivation (data not shown), the concentration of 3SP was not decreasing during the time course of the experiment, and the optical density remained at 0.3. The TDP concentration was decreasing very weakly, and in the same supernatant a weak accumulation of 3SP was detected. The optical density of this culture remained constant at ϳ0.3.
Molecular Characterization of Tn5::mob-induced Mutants and Identification of Two Chromosomal DNA Regions Harboring Genes Required for TDP Catabolism-The insertions of Tn5::mob into the genomes of these mutants were confirmed by Southern hybridization using ApaI-digested DNA isolated from the mutants and a digoxygenin-labeled K m resistance cassette derived from Tn5::mob DNA as probe. To map the insertions of Tn5::mob in these mutants, SalI-or BamHI-restricted genomic DNA fragments conferring kanamycin resistance were cloned in E. coli strain Top10. Sequencing of these DNA fragments using oligonucleotides hybridizing to the terminal region of IS50L and the multiple cloning site of the used cloning vector pBluescriptSK Ϫ revealed only four different open reading frames (ORFs) in which Tn5::mob had been inserted (Table  1). In mutants 1/8 and 2/34 Tn5::mob was mapped in a gene putatively coding for an FAD-linked oxidoreductase exhibiting 55% identical amino acids to FAD-linked oxidase domain protein of Acidovorax avenae subsp. citrulli AAC00 -1 (fox). In mutants 1/20 and 4/5 the transposon was mapped in an ORF showing high homologies (46% identical amino acids) to a putative exported protein of Burkholderia xenovorans LB400. Due to the sequence similarities to representatives of the gene family of Bordetella uptake genes (bug), coding for extracytoplasmatic solute receptor proteins, this gene (bugA) is also a putative member of this gene family, which is strongly overrepresented in several ␤-proteobacteria (32). In mutants 13/33 and 2/5 the transposon was mapped in a putative gene coding for a cysteine dioxygenase (cdo). The respective sequence exhibited 68% identical amino acids to a type I Cdo of Verminephrobacter eiseniae EF01-2. In mutant 1/1, the transposon insertion was localized in an ORF showing high sequence homology (56% identical amino acids) to a predicted family III acyl-CoA-transferase of Magnetospirillium gryphiswaldense MSR-1.
In addition to the genes in which Tn5::mob was mapped, several putative ORFs were detected when the regions upstream and downstream of these insertions were sequenced by the PCR-based directional genome walking method (30) and analyzed. Detailed analysis of these sequences revealed that the fox Vp , bugA Vp , and cdo Vp genes are contiguous, whereas there were no hints that the gene coding for the putative acyl-CoAtransferase is also adjacent to the other gene cluster. In total, the sequences of a 5.1-kbp and of a 3.2-kbp region were unraveled, respectively. Downstream of the cdo Vp gene in the first gene cluster an ORF showing 63% identical amino acids to the alkylhydroperoxidase core (ahpD) of Burkholderia xenovorans LB400 was detected. Downstream of act Vp a gene putatively coding for an acyl-CoA-dehydrogenase (acd) was identified. Furthermore, two other homologous genes of the Bordetella uptake gene family were identified: one (bugC) was identified upstream of the putative oxidoreductase gene in the first cluster; the other was detected downstream of the putative acyl CoA-dehydrogenase (bugB). The gene organization is summarized in Fig. 6.
Transcriptional Analysis of Genes Putatively Involved in TDP Catabolism-Transcriptional analysis of genes identified by transposon mutagenesis and putatively involved in TDP metabolism were performed by RT-PCR to determine if cdo Vp and act Vp are constitutively transcribed, or if the expression is induced in the presence of TDP in the wild type. Furthermore, RT-PCR was carried out with RNA isolated from the mutants 4/5 and 2/34 to analyze whether the phenotype of these mutants is caused by transposon insertion in the respective gene, or if a polar effect caused by insertion of Tn5::mob in the genes adjacent to cdo Vp is responsible for the observed phenotype. Because the strains were grown in MSM media containing 10 mM MgSO 4 , the results of the RT-PCR can also be used to determine if the expression of the genes is independent of limited concentrations of sulfate, as previously described for sulfate starvation-induced (Ssi) proteins (36). Using the same RNA, bugA Vp was amplified from RNA of mutant 2/34 and fox Vp was amplified from RNA of mutant 4/5. Successful amplification of both genes indicated independent transcription of these genes and confirms the genotype. For the same reason RT-PCR analysis of the act Vp and acd Vp genes was performed using RNA isolated from mutant 1/1. The results revealed that cdo Vp and act Vp are transcribed in the wild type irrespectively of having used succinate or TDP as the sole source of carbon and energy. This indicated that transcription of both genes is constitutive and is not induced in the presence of TDP. Transcription of cdo Vp was also detected in RNA isolated from mutants 4/5 or 2/34. In contrast, no RT-PCR product of acd Vp was detected under all used conditions. To exclude non-efficient RT-PCR, cdo Vp was analyzed as a positive control in the same reaction mixtures and could be amplified in all cases. Results are summarized in Fig. 6.
Analysis of Primary Structure of the Cdo Translational Product-Multiple sequence alignments of the putative V. paradoxus Cdo and of Cdo homologues of different species, showing high sequence similarities, revealed several highly conserved residues already described for eukaryotic Cdo proteins. This enzyme belongs to the cupin protein superfamily, and therefore the consensus sequences of two cupin motifs were identified in the amino acid sequences (37). Cupin motif 1 (GX 5 HXHX 3,4 EX 6 G) is highly conserved in the V. paradoxus TBEA6 Cdo homologue except of glutamate, which is replaced The results of transcriptional analysis of the genes are summarized in the lines below the respective genome region. RNA was isolated from the wild type of V. paradoxus strain TBEA6 cells growing in MSM containing 1% (w/v) TDP or 1% (w/v) sodium succinate or from cells of transposon-induced mutants growing in MSM containing 1% (w/v) TDP plus 0.2% (w/v) sodium succinate. ϩ, transcript detectable; Ϫ, transcript not detectable; n.t., not tested. Upper part: sequence analysis of Tn5::mob insertions of eight transposon-induced mutants revealed three different adjacent genes putatively involved in TDP metabolism. cdo, putative gene of cysteine dioxygenase; bug, putative gene for an extracytoplasmic solute receptor protein (Bordetella uptake gene); fox, putative gene for an FMN-dependent oxidoreductase. Lower part: neighborhood of the gene act putatively coding for an acyl-CoA-transferase/carnithin dehydratase in which Tn5::mob was mapped in mutant 1/1. Downstream of the act gene, a gene coding for an acyl-CoA-dehydrogenase (acd) and a gene putatively coding for an extracytoplasmic solute receptor protein (bug, Bordetella uptake gene) were located.

3-Mercaptopropionate Dioxygenase in V. paradoxus
by alanine or glycine. A common characteristic of eukaryotic Cdo proteins is the replacement of the highly conserved glutamate within motif 1 by cysteine, and the formation of a thioether bond between the thiol sulfur of cysteine and the C⑀ of a nearby tyrosine residue is proposed (38 -40). The second cupin motif (GDX 4 PXGX 2 HX 3 N) is less conserved in Cdo proteins from eukaryotes (39). However, in the sequence of the V. paradoxus TBEA6 Cdo Vp , all conserved residues of the cupin motif 2 are present. In addition to the two cupin motifs some other highly conserved residues occur in Cdo proteins. In eukaryotes the residues Tyr-58 and Arg-60 are located at the active site, and they are supposed to be directly involved in substrate coordination (40). A substitution of the equivalent of the highly conserved Arg-60 in bacterial Cdo proteins by glutamine, as previously reported (14), occurs obviously also in the V. paradoxus TBEA6 Cdo Vp and in other bacterial Cdo proteins used for this multiple sequence alignment (Fig. 7A). The highly conserved Tyr-58 was present in all sequences used for this multiple alignment. Additionally highly conserved residues of eukaryotic Cdo proteins, also found in V. paradoxus and the other Cdo sequences, are Ser-153, His-155, and Tyr-157, which probably form a catalytic triad (14,39,40).
BlastX searches using the cdo Vp nucleotide sequence of V. paradoxus strain TBEA6 as query, revealed two putative genes for Cdo in some bacteria (21). Whereas the translational product of one shows high sequence similarity (Ͼ50% identical amino acids) to Cdo of V. paradoxus (Fig. 7A), the other shows lower sequence similarity (Ͻ30% identical amino acids). The amino acid sequences of lower sequence similarities to Cdo of V. paradoxus were also aligned, and the results revealed that the highly conserved residues of Cdo proteins are also found in these sequences (Fig. 7B). As described for Cdo proteins from eukaryotes, the highly conserved glutamate (Glu) of cupin motif 1 is replaced by cysteine (Cys-93) instead of glycine as often occurring in paralogues. The cupin motif 2 is less conserved. In some cases the highly conserved serine (Ser-153) residue within the catalytic triad is replaced by histidine, and the equivalent of rat Tyr-58 is substituted by glutamine. Arg-60 is highly conserved in contrast to the paralogue where Arg-60 is substituted by glutamine.
CDO Enzyme Activity Assays in Recombinant Strains-In vivo assays were carried out using 3MP as substrate and analyzed by GC/MS. In addition to E. coli BL21(DE3)pLysS pET23a::cdo Vp and E. coli Top 10 pBBR1MCS-3::cdo Vp two other E. coli strains containing the putative cdoA and cdoB FIGURE 7. Multiple sequence alignment of Cdo homologues from various bacteria. The boxes indicate highly conserved regions known for Cdo (11). Within the cupin motif the highly conserved residues are highlighted. A, multiple sequence alignment of Cdo of V. paradoxus TBEA6 with homologues showing high sequence similarities (Ͻ50%). The highly conserved glutamic acid is replaced by alanine or glycine. B, multiple sequence alignment of cdo paralogues of B. pertussis Tohama I, R. eutropha H16, and V. eiseniae EF01-2. The glutamic acid residue of motif 1 is replaced by cysteine.

3-Mercaptopropionate Dioxygenase in V. paradoxus
genes of R. eutropha H16 were also applied. E. coli strain BL21(DE3)pLysS harboring pET23a was used as negative control. Cells were cultivated in M9 medium supplemented with 0.1 g/liter yeast extract and containing 1% (v/v) glycerol as carbon source at 30°C. After 6 h of cultivation expression of the recombinant protein was induced by addition of 1 mM IPTG, and the substrate 3MP was added to a final concentration of 0.1% (v/v). Cells were harvested after 24 h of cultivation and washed twice with saline, before the cell pellets and the supernatants were lyophilized and analyzed by GC/MS. The putative reaction product 3SP was detected in cell pellets and in the supernatants of E. coli BL21(DE3)pLysS harboring pET23a::cdo Vp (Fig. 8) and also of E. coli Top 10 harboring pBBR1MCS-3::cdoA Re . No 3SP was detected in the cell pellet and supernatants of the negative control E. coli BL21(DE3)pLysS harboring pET23a (supplemental Fig. S1) and in those of the recombinant E. coli strains Top 10 harboring pBBR1MCS-3:: cdo Vp or Top 10 pBBR1MCS-3::cdoB Re .
Enzyme activity was determined in vitro using cysteine, cysteamine, and 3MP as substrates. The putative cdo gene of V. paradoxus strain TBEA6 was heterologously expressed using the T7-promotor/ polymerase-based expression vector pET23a and E. coli BL21(DE3)-pLysS as host strain. The Cdo Vp was purified to electrophoretic homogeneity as hexahistidinetagged protein by immobilized metal chelate affinity chromatography and applied to the in vitro enzyme assay (supplemental Fig.  S2). Enzyme activity was only determined when 3MP was supplied as substrate. When cysteine or cysteamine where used as substrate neither of the expected reaction products cysteine sulfinic acid or hypotaurine could be detected, although the assays were done at different substrate concentrations (0.5-20 mM) and even with up to 10 g of purified enzyme. The oxygenation of 3MP was monitored with different concentrations of 3MP (50 -500 M) and revealed a sigmoidal dependence on substrate concentration, indicating that this reaction did not fit Michaelis-Menten kinetics. The enzyme activity saturated at 300 M 3MP, and concentrations higher than 500 M resulted in a significant decrease of the activity. The addition of cysteamine to the in vitro assays using 3MP as substrate also resulted in a clearly lower activity, and concentrations above 10 mM completely inhibited the enzyme. In contrast, the addition of cysteine gave an activating effect. The enzyme activity was also determined in a Bis-Tris/Trisbuffer system at different pH values (pH 5-9) and revealed highest activities at pH 7 (data not shown).

DISCUSSION
In total, eleven different bacterial strains capable of using TDP as sole source of carbon and energy were isolated (supplemental Table S1). Besides Schlegelella thermodepolymerans, which shows weak growth on TDP, this is the first report about bacteria capable of utilizing this thioether (41). Phylogenetic analysis revealed that the best growing isolates enriched in this study also belong to the family of Comamonadaceae (Fig. 2). High sequence similarities of the 16 S rRNA genes of ϳ99% to the 16 S rRNA gene of V. paradoxus indicated that the isolates TBEA6 and SFWT are closely related to this species. Strains of V. paradoxus are commonly known for the use of a large variety of organic compounds (42). In addition to the utilization of various sugars and amino acids, degradation of complex substrates like 2,4-dichlorophenoxyacetic acid or 3-nitotyrosine was reported (43,44). Desulfurization of toluen sulfonate and the use of a broad variety of arylsulfonates, alkylsulfonates, and sulfate esters as sulfur sources by V. paradoxus DSM 30034 was also described recently (45). Furthermore, sulfolane is used as sole source of carbon, sulfur, and energy by V. paradoxus WP1 (46). Isolate TBEA3 exhibited only low sequence similarity to cultured and characterized bacteria. This isolate probably represents a new taxon within the Comamonadaceae, 4 and studies on its taxonomic affiliation will be published separately.
Transposon mutagenesis using Tn5::mob was done to elucidate the metabolism of TDP in V. paradoxus strain TBEA6. Ten independent Tn5::mob-induced mutants affected in growth on TDP were isolated and further characterized. Although growth on TDP was highly or completely diminished, HPLC analysis of the supernatants revealed that TDP is still degraded by most of these mutants. Only mutants 1/8 and 1/20 were fully impaired in the degradation of TDP. In addition, with the exception of mutant 1/1, all other mutants show growth on 3SP.
In two mutants (2/5 and 13/33) the transposon was mapped in a gene putatively coding for a cysteine dioxygenase. The cdo Vp gene is transcribed during growth on TDP and on succinate indicating that it is not induced by TDP but presumably constitutively expressed. In addition, the presence of 1 mM MgSO 4 in the medium showed that transcription is probably not repressed by inorganic sulfate as previously described for Ssi proteins (36). Cysteine dioxygenase catalyzes the oxidation of the free thiol group of cysteine, which is thereby oxidized to cysteine sulfinic acid. Due to the structural similarity of 3MP to cysteine and because 3SP is accumulated by mutant 1/1 during growth on succinate and TDP, we suppose that 3MP is used as substrate by the Cdo of V. paradoxus strain TBEA6. This assumption is strongly supported by the detection of 3SP in in vivo and in vitro assays using a recombinant E. coli strain expressing cdo Vp and using 3MP as substrate. Advanced in vitro assays applying purified Cdo clearly demonstrated that oxygenation of 3MP depended on this enzyme. Highest enzyme activ-ities were achieved applying 50 -500 M 3MP; concentrations higher than 500 M led to a significant decrease of Cdo activity. Enzyme activity showed a sigmoidal dependence on the substrate concentration and did obviously not display Michaelis-Menten kinetics, because it was also reported for Cdo from other bacteria (14). 3MP is a highly reactive substance and inhibition of the enzyme by concentrations above 0.5 mM is explainable by the reducing properties of this thiol. Cysteamine concentration above 5 mM seems to have a similar effect as revealed by addition of different concentrations of cysteamine to the standard assay using 3MP as substrate. Because hypotaurine could not be detected when the enzyme was incubated with cysteamine, the latter was obviously not used as substrate.
The enzyme-catalyzed oxidation of 3MP is very unusual and has not been described for any characterized Cdo, yet. In contrary, all so far characterized Cdo are reported as highly specific for cysteine, and structurally related thiols were not used as substrate for oxygenation (14,47). Furthermore, 3MP was shown to be a strong inhibitor of rat Cdo and a moderate inhibitor of bacterial Cdo (14,47). Although analysis of the primary structure of the V. paradoxus Cdo revealed highly conserved residues known for various cysteine dioxygenases, there are some notable differences to described eukaryotic and prokaryotic Cdo. The highly conserved glutamate residue (Glu) is replaced by the non-polar residue alanine in cupin motif 1, whereas cupin motif 2 is almost completely conserved. Replacement of the highly conserved residue Arg-60 by glutamine was also reported for some bacterial Cdo (14); however, until now their function as cysteine dioxygenase is based on theoretical considerations only (14).
Although we observed an activating effect of cysteine when it was applied to the enzyme assay, it was not possible to detect the reaction product cysteine sulfinic acid. We have also shown that cysteamine is no substrate for this enzyme (see above). Therefore, and because the primary structure of the Cdo Vp homologue of V. paradoxus TBAE6 exhibits significant differences, this enzyme is not a cysteine dioxygenase but a thiol dioxygenase with an alternative substrate range. This assumption is also supported by a lacking pleiotropic phenotype of the two transposon-induced cdo Vp insertion mutants isolated in this study.
In eukaryotes a second thiol dioxygenase is known since 1966, however, the gene was identified only recently (48,49). The enzyme catalyzes the oxidation of cysteamine, and like Cdo it is a member of the cupin superfamily (49). Some studies characterized this enzyme as highly specific for cysteamine, others reported only low substrate specificity (49,50). Besides cysteamine, oxidation of homocysteamine, mercaptoethanol, and 3-mercaptopropionic acid by this enzyme has been shown.
Although there are no reports about further thiol dioxygenase reactions in bacteria, it is likely that bacteria also possess more than one thiol dioxygenase. Because we could also not find any hint of oxygenation of cysteamine by Cdo Vp , it is proposed that the dioxygenase described in this study is a new type of thiol dioxygenases. Due to the substrate it is referred to as 3-mercaptopropionate dioxygenase.
3MP, derived from biological and abiotic reactions, occurs widespread in coastal sediments. Kiene and Taylor (8, 51) even suggested that 3MP is a central metabolite in both catabolic and assimilatory sulfur metabolism in general. Therefore, the ability to utilize 3MP could be a beneficial trait for soil bacteria. Furthermore, degradation of this compound releases the sulfur moiety back into the sulfur cycle. In R. eutropha and in several other bacteria in silico analysis identified an additional gene putatively coding for a cysteine dioxygenase. These paralogue genes show low sequence similarity to each other, and the alignment of the primary structure revealed that one of the peculiar differences between these paralogues is the replacement of the conserved glutamate residue in motif 1. Although one paralogue cdo gene product showed replacement of glutamate by glycine or alanine, the other showed substitution by cysteine as described for eukaryotic Cdos. Different substitutions were also found for the highly conserved residues Tyr-58, Arg-60, and Ser-153 in the Cdo of rat.
Because characterization of these paralogue genes is also based on theoretical considerations, it would be interesting to characterize and compare the substrate ranges of these enzymes in the future. For R. eutropha H16 the function of one paralogue gene was analyzed in this study. The cdoA Re gene product showed high sequence similarity to the Cdo Vp of V. paradoxus TBEA6. The glutamate residue was replaced by glycine in CdoA Re . With this enzyme oxidation of 3MP was demonstrated in the in vivo assays, and heterologous expression of cdoA Re in the Tn5::mob-induced mutant 13/33 of V. paradoxus showing transposon insertion in cdo Vp gene restored growth on TDP.
As shown in this study, mutant 1/1 accumulates 3SP, and growth of the mutant in this compound is completely impaired. Therefore, it is very likely that the act Vp translational product is involved in the further degradation of 3SP. Due to the high sequence similarities of the putative gene product to family III acyl-CoA-transferases, we suppose that 3SP is activated by linkage to CoA. The resulting 3SP-CoA thioester is then probably further metabolized in one or more steps to SO 3 and propionate (Fig. 9). A homologous gene product is putatively involved in Marinomonas sp. in degradation of dimethylsulfoniopropionate. The dddD gene product of this bacterium also shows high sequence homologies to family III acyl-CoA-transferase, and it was predicted to catalyze the addition of CoA as the initial step in dimethylsulfoniopropionate degradation (52). Transcription of act Vp is also not induced by TDP or limited concentrations of inorganic sulfate in V. paradoxus strain TBEA6 as demonstrated by RT-PCR. Transcription of the gene downstream of the act Vp putatively coding for acyl-CoA-dehydrogenase was not detected in mutants 1/1 and 2/5 or in the wild type under all used conditions, indicating that this gene is probably not involved in TDP degradation and that a polar effect can be excluded.
A putative bug homologue was identified upstream of cdo Vp , and in two mutants the transposon was mapped in this gene. Although numerous orthologous bug genes were identified in genomes of several ␤-proteobacteria (32,53), the function of most of them is still unknown. The bug homologous bctC gene of Bordetella pertussis was identified as part of a tripartite tricarboxylate transporter operon. Besides its function as an extracytoplasmic solute receptor, the protein is also part of the signaling cascade leading to up-regulation of the operon in the presence of their substrates (28). Although some bug homologues were identified as part of a tripartite tricarboxylate transporter system, most of them are not linked to any transport system (53). In the R. eutropha H16 genome 154 bug homologous genes were identified, and an additionally regulatory role for the bug homologues was suggested, because the majority of these genes were located immediately adjacent or one or two coding FIGURE 9. Proposed pathway of the catabolism of TDP in V. paradoxus strain TBEA6. Initially, TDP is cleaved by a yet unidentified enzyme (1) into 3-hydroxypropionate and 3-mercaptopropionate. The latter is then oxygenated by a cysteine dioxygenase (2) thereby yielding 3-sulfinopropionate. After addition of coenzyme A by a family III acyl-CoA transferase (3), the sulfur moiety is putatively removed by a desulfinase (4) resulting in propionyl-CoA, which is then further metabolized. Whereas enzyme reactions (2) and (3) are based on experimental data and on general predictions of the respective enzymes, the proposed reactions (1) and (4) are hypothetically and are based on theoretical considerations and indirect experimental evidence.
The role of the bug homologues in TDP degradation is still unknown, although impaired growth of mutants 1/20 and 2/34 on TDP provides evidence that this gene is involved in the utilization of this thioether. Because no adjacent genes for transport proteins were identified, a sole transport function is unlikely. TDP and limited concentrations of inorganic sulfate have also no inducing effect on the transcription of bugA as revealed by RT-PCR. Identification of two additional bug homologues corresponds with the finding that the representatives of this gene family are overrepresented in several genomes of ␤-proteobacteria (32). In addition, putative signal sequences for leader peptides were identified in all three bug sequences so that the gene products are probably localized in the periplasma.
Insertion of Tn5::mob in a gene putatively coding for an FADdependent oxidoreductase impaired growth of V. paradoxus on TDP. Because polar effects can be excluded according to RT-PCR analysis, further studies have to elucidate the physiological role of this enzyme in TDP metabolism.