Use of a Novel Fluorinated Organosulfur Compound To Isolate Bacteria Capable of Carbon-Sulfur Bond Cleavage

ABSTRACT The vacuum residue fraction of heavy crudes contributes to the viscosity of these oils. Specific microbial cleavage of C—S bonds in alkylsulfide bridges that form linkages in this fraction may result in dramatic viscosity reduction. To date, no bacterial strains have been shown conclusively to cleave C—S bonds within alkyl chains. Screening for microbes that can perform this activity was greatly facilitated by the use of a newly synthesized compound, bis-(3-pentafluorophenylpropyl)-sulfide (PFPS), as a novel sulfur source. The terminal pentafluorinated aromatic rings of PFPS preclude growth of aromatic ring-degrading bacteria but allow for selective enrichment of strains capable of cleaving C—S bonds. A unique bacterial strain, Rhodococcus sp. strain JVH1, that used PFPS as a sole sulfur source was isolated from an oil-contaminated environment. Gas chromatography-mass spectrometry analysis revealed that JVH1 oxidized PFPS to a sulfoxide and then a sulfone prior to cleaving the C—S bond to form an alcohol and, presumably, a sulfinate from which sulfur could be extracted for growth. Four known dibenzothiophene-desulfurizing strains, including Rhodococcus sp. strain IGTS8, were all unable to cleave the C—S bond in PFPS but could oxidize PFPS to the sulfone via the sulfoxide. Conversely, JVH1 was unable to oxidize dibenzothiophene but was able to use a variety of alkyl sulfides, in addition to PFPS, as sole sulfur sources. Overall, PFPS is an excellent tool for isolating bacteria capable of cleaving subterminal C—S bonds within alkyl chains. The type of desulfurization displayed by JVH1 differs significantly from previously described reaction results.

Microbial methods of removing sulfur from organosulfur compounds are of interest to the petroleum industry for reducing sulfur emissions and, more recently, for reducing heavy oil viscosity. As conventional crude oils are consumed throughout the world, heavier oils are being exploited which, due to their high viscosity, cannot be transported from remote field sites to refineries without adding diluents. The vacuum residue fraction of crude oils (boiling point Ն 524°C [975°F]) contributes to viscosity, and recent models indicate that alkyl sulfides compose important bridges in the network of high-molecularweight molecules in this fraction (34). Up to 40% of the sulfur in these fractions is in the form of alkyl sulfides; if these alkyl COS bonds can be selectively cleaved using a biological catalyst, reductions in molecular size and viscosity could occur.
The first requirement for developing a biological process for heavy oil viscosity reduction is obtaining a microorganism capable of alkyl COS bond cleavage without reducing the carbon value of the substrate. Precedence for this type of reaction with aromatic heterocycles can be found in the well-characterized 4S pathway that selectively removes sulfur from dibenzothiophene (DBT) (35). The dsz or sox operon (genes responsible for DBT desulfurization) (7,38) in Rhodococcus sp. strain IGTS8 has been observed in a variety of genera that are apparently widespread in petroleum-contaminated environments (6,8). Other genera capable of selective sulfur removal from DBT include Gordonia sp. strain CYKS1 (40), Arthrobacter sp. strain ECRD-1 (28), the thermophilic Paenibacillus sp. strain A11-2 (23,24), and Bacillus subtilis WU-S2B (22). Strains that desulfurize alkylated DBTs (12,13,25,28,32,33) and benzothiophene (14) have also been isolated. Recently, a metabolic pathway has been described for Rhodococcus sp. strain WU-K2R that can desulfurize naphthothiophene and benzothiophene (21).
However, there are no reports that conclusively illustrate the bacterial cleavage of alkyl COS bonds by bacteria. Van Hamme et al. (45) have shown that a variety of white-rot fungi oxidize dibenzyl sulfide to dibenzyl sulfoxide and dibenzyl sulfone prior to further degradation. However, other examples of the microbial degradation of compounds containing this type of bond (e.g., 2-chloroethyl sulfide and thiodiglycol) give no evidence for direct sulfur oxidation followed by COS bond cleavage without degradation of the alkyl or aromatic moieties (20,29,36,40). For example, metabolism by Nocardioides simplex of 1-(phytanylsulfanyl)-octadecane (used as a model compound for sulfide bridges in high-molecular-weight fractions of sulfur-rich petroleum) has been described previously (18); although the sulfur was oxidized, no COS bond cleavage was observed. This bacterium oxidized the terminal alkyl groups in the model compound, which was subsequently degraded by ␤-oxidations.
The challenge in isolating a bacterial strain capable of cleaving COS bonds within alkyl chains lies in finding an appropriate substrate. Commercially available sulfur-containing substrates typically have terminal alkyl or aromatic moieties that are susceptible to microbial attack. Several studies have evaluated and used fluorobenzoic acids, including pentafluorobenzoic acid, as conservative tracers in soil and ground water (2,5,17,37) because of their chemical stability. Thus, we hypothesized that a short-chain alkyl sulfide that contained a stable pentafluorophenyl group at each terminus would be ideal for selecting microorganisms that cleave COS bonds.
This report outlines the synthesis and use of bis-(3-pentafluorophenylpropyl)-sulfide (PFPS) as a novel compound to select, screen, and characterize isolates capable of cleaving COS bonds within alkyl chains that may occur in the residual fractions of heavy crude oils and bitumens. An isolate (Rhodococcus sp. strain JVH1) which is distinct from the known DBTdesulfurizing strains in that it cannot use DBT as a sulfur source is described. In addition, metabolites produced by JVH1 from PFPS are shown and a metabolic pathway involving specific sulfur oxidation and COS bond cleavage is presented. titatively extracted from roll tubes with three 1-ml aliquots of pentane. Prior to extraction, 1-phenylnaphthalene was added as an internal standard. After each addition of pentane, tubes were vortexed at high speed for 10 s. PFPP-acid was subsequently extracted by adjusting the pH to Ͻ2.0 and extracting with 3 ϫ 1 ml of ethyl acetate.
For routine analysis, a Hewlett-Packard 5930 gas chromatograph equipped with an HP-1 column (length, 25 m; inside diameter, 0.32 mm; film thickness, 0.17 m) and a flame-ionization detector were used. Peak areas were measured with an HP 3390A integrator. The injector was set to 300°C, and the detector was set to 320°C. The oven program was as follows: 90°C for 5 min followed by an increase in temperature of 15°C/min to 280°C, with a final hold time of 5 min.
Metabolite derivatization and identification. Prior to gas chromatographymass spectrometry (GC-MS) analysis, samples were dried with anhydrous sodium sulfate. PFPS, PFPSO, PFPSO 2 , PFPP-OH, and PFPP-acid were all detected by GC-MS without derivatization. PFPP-OH and PFPP-acid were also derivatized in 800 l of acetonitrile with 200 l of N-O-bis-(trimethylsilyl)acetamide (Pierce, Rockford, Ill.) at 70°C in a water bath for 15 min prior to GC-MS. PFPP-acid was also derivatized with diazomethane in ether.
Samples were analyzed on a Hewlett-Packard 5890 series II gas chromatograph with a 5970 series mass-selective detector and a 30-m DB-5 capillary column (J&W Scientific, Folsom, Calif.). The GC temperature program used for analysis was 90°C for 5 min followed by an increase of 10°C/min to 300°C, with a final hold time of 5 min.
Fluoride and sulfate analysis. A fluoride ion-selective electrode (Thermo Orion, Beverly, Mass.) was used for fluoride analysis after samples were mixed with an equal volume of Ionall buffer solution (Anachemia Canada Inc., Lachine, Quebec, Canada).
Sulfate was analyzed using a DX600 ion chromatography system (Dionex Corp., Sunnyvale, Calif.) equipped with an AS9-HC anion exchange column and a CD25 conductivity detector. The mobile phase was 9 mM sodium carbonate set to 1.2 ml/min, and the suppressor current was set to 50 mA.
Synthesis of PFPP-OH. All synthetic reactions and purifications were monitored by thin-layer chromatography on silica plates (Merck, Darmstadt, Germany) (250 m of silica gel 60 F 254 ) with ethyl acetate in hexane for development. Spots were visualized under UV light and by oxidation with potassium permanganate. All reaction products were purified by silica gel chromatography, and purity was evaluated using 1 H-nuclear magnetic resonance (NMR) analysis. All reaction solvents were distilled prior to use.
PFPP-OH was synthesized from allylpentafluorobenzene (Aldrich) by hydroboration with argon by treatment with 1 M BH 3 . THF (triborane) complex in tetrahydrofuran (Aldrich) followed by the addition of 30% H 2 O 2 under alkaline conditions. The product (64% yield) was extracted at pH 2 with ethyl ether and purified to give PFPP-OH, a viscous yellow liquid (  Synthesis of PFPS and PFPSS. These were synthesized by first converting PFPP-OH to 1-iodo-3-pentafluorophenyl propane. This reaction was carried out in DCM with argon at room temperature according to the method of Lange and Gottardo (27). Triphenyl phosphine (Aldrich) was dissolved followed by the addition of imidazole (Aldrich), and the ingredients were allowed to mix for 30 min prior to the addition of iodine. The reaction mixture was extracted with DCM, and filtered. The extract was concentrated and the purified iodide, a dark pink oil Hz, CH 2 ], and 3.17 [2H, t, J ϭ 9 Hz, CH 2 ]), was used to make PFPS and PFPSS with the method of Landini and Rollo (26) and using the phase transfer catalyst hexadecyltributylphosphonium bromide (Aldrich) and Na 2 S ⅐ 9H 2 O. The reaction proceeded for 2 h with argon (in water purged with argon to remove residual oxygen). The two-phase reaction mixture was extracted with ether, dried with MgSO 4, and concentrated using a vacuum to yield a viscous, cloudy, orange mixed product. Purified PFPS (62%   weeks to remove solids and any associated sulfur sources. Turbid cultures were diluted and spread on agar to obtain isolated colonies. A total of 70 pure isolates, including 3 filamentous fungal isolates and 1 yeast isolate, were screened for the ability to produce biomass and remove PFPS when PFPS was provided as the sole sulfur source at a limiting concentration (40 M). Control cultures with no added sulfur were prepared for each isolate, as some strains were able to produce biomass on the trace sulfur found in medium components. Gas chromatograph-flame-ionization detector analysis showed that 4 of the 70 isolates oxidized PFPS to PFPSO 2 , and 2 of the 70 were able to produce significant amounts of biomass with PFPS as the sole sulfur source. One strain, JVH1, was chosen for further study on the basis of the extent of growth and PFPS removal. This organism is a gram-positive, branching rod-shaped bacterium 1 m in width, with branches up to 5 m in length. Strain JVH1 was sent to MIDI Labs (Newark, Del.) for full-length 16S rRNA gene sequence analysis; the strain was found to cluster with Rhodococcus, Corynebacterium, and Tsukamurella species. It had a 99% match to Rhodococcus opacus in GenBank (accession number Y11893); this isolate has been named Rhodococcus sp. strain JVH1.

Isolation of the PFPS-desulfurizing bacterium
Strain JVH1 cleaves COS bonds. To find PFPS metabolites, strain JVH1 was grown in SFGA medium with PFPS (325 M) as the sole sulfur source. Cultures were extracted with pentane at neutral pH and then ethyl acetate at pH Ͻ2 at various times during the growth and stationary phases. Concentrated extracts, and extracts derivatized with either N-O-bis-(trimethylsilyl)acetamide or diazomethane, were analyzed by GC-MS. Table 1 shows the mass spectra of PFPS and metabolites detected in culture extracts. In each case, spectra of the metabolites and derivatized metabolites were identical to those of authentic standards synthesized for this work. PFPSO and PFPSO 2 were detected in pentane extracts along with PFPP-OH. Ethyl acetate extracts yielded PFPP-acid. Both PFPP-OH and PFPP-acid were derivatized to produce trimethyl silyl derivatives, while PFPP-acid was converted to a methyl ester. The spectra of these novel metabolites and the sulfur source PFPS have not been reported previously.
Samples taken daily were analyzed for fluoride ion to check for possible aromatic ring attack. At no time did fluoride concentrations exceed the detection limit of 0.2 mg/liter. Strain JVH1 differs from known DBT-desulfurizing strains. The detection of the sulfoxide and sulfone of PFPS indicated that strain JVH1 used a pathway similar to that described for the DBT-desulfurizing strain R. erythropolis IGTS8 and related organisms. To test the versatility of strain JVH1 regarding cleavage of COS bonds, it was incubated with various organosulfur compounds. Table 2 shows that JVH1 was able to use linear alkyl sulfides ranging from DMSO to didodecyl sulfide as sole sulfur sources. However, dioctadecyl sulfide was not used by JVH1 as a sulfur source, presumably due to low solubility and to its crystalline state in aqueous medium. In addition, dibenzyl sulfide and its sulfur oxidation products, PFPSS, 1,4-dithiane, and tetrahydrothiophene sulfone, were used as sulfur sources. In contrast, phenyl sulfide, benzothiophene, and DBT and its sulfone were not used by JVH1 to produce biomass.
Because strain JVH1 could not utilize DBT as a sulfur source, further comparisons were made with four well-known DBT-desulfurizing bacteria. The dry weights of biomass produced during a 7-day incubation with various sulfur sources were determined. Figure 1 shows that JVH1 is distinct from four known DBT-desulfurizing strains in that DBT does not support its growth whereas PFPS cannot support growth of the DBT-desulfurizing strains. However, GC-MS analysis showed that strain IGTS8 was able to stoichiometrically convert PFPS to PFPSO 2 via PFPSO. Conversely, JVH1 was unable to oxidize DBT to DBT-sulfoxide or DBT-sulfone or to extract sul-   fur from DBT-sulfone. All strains were able to grow well using DMSO and sulfate as sole sulfur sources. Strain JVH1 degrades PFPS in batch culture. Growth of strain JVH1 corresponded to the disappearance of PFPS from the culture medium, as shown in Fig. 2. Growth caused an increase in pH, presumably from the consumption of acetate. After 14 days of incubation, JVH1 had removed all of the PFPS from the medium, with no accumulation of PFPSO or PFPSO 2 . Sulfate concentrations in the culture supernatant did not exceed that of the cell-free control (ϳ10 M sulfate) at any time during the incubation.
Gray et al. (15) reported that the COS bond cleavage of DBT sulfone yielded 2Ј-hydroxybiphenyl-2-sulfinate. In an analogous manner, we hypothesized that cleavage of the COS bond in PFPSO 2 would yield PFPP-OH and a sulfinate (Fig. 3). Subsequent hydrolysis of the COS bond in the sulfinate would also yield PFPP-OH. The sulfinate was not detected by our analytical methods, but up to 18 M of PFPP-OH accumulated in the culture medium by day 4 and then disappeared by day 6 (Fig. 2). Trace amounts of PFPP-acid were also detected. Further degradation of the acid via beta-oxidation would form pentafluorobenzoic acid, but none of this proposed metabolite was detected.

DISCUSSION
PFPS allows for the efficient selection of strains capable of cleaving COS bonds within alkyl chains. The relative ease with which isolates were obtained may indicate that this type of activity is not uncommon in the environment. However, to our knowledge this activity has not been described previously.
Based on the metabolites formed, the metabolic pathway in Fig. 3 is proposed. Here, sulfur is oxidized to produce first a sulfoxide and then a sulfone in a manner analogous to that of the 4S pathway observed with DBT-desulfurizing bacteria (35). Once activated in this manner, COS bond cleavage occurs to produce an alcohol and, presumably, a sulfinate. In the 4S pathway for DBT desulfurization, DBT-sulfone (DBTO 2 ) is oxidized to 2-(2Ј-hydroxybiphenyl)-benzene sulfinate via the action of DszA, which is a DBTO 2 -monooxygenase (15). If a sulfinate were formed from PFPSO 2 , it would likely be cleaved to yield sulfite, which would then spontaneously oxidize to sulfate prior to being reduced by the microorganism for growth. Gray et al. (15) showed that the desulfinase in the 4S pathway (DszB) produces 2-hydroxybiphenyl and sulfite from 2-(2Ј-hydroxybiphenyl)-benzene sulfinate. Similarly, both sulfite and sulfate were detected by Greene et al. (16) in cultures of a Variovorax sp. strain growing on tetrahydrothiophene sulfone and by Bressler et al. (3) in cultures of a Pseudonocardia sp. strain with the sulfones of benzothiophene and 3-and 5-methylbenzothiophene as sole carbon and sulfur sources. In the current report, culture sulfate concentrations did not increase above those of the cell-free controls, indicating that sulfate was rapidly consumed by the growing cells.
While this would complete the sulfur extraction process, JVH1 further oxidized PFPP-OH to PFPP-acid. PFPP-acid would be metabolized to pentafluorobenzoic acid, presumably, but this metabolite has not been detected and the fate of the PFPP-acid is unknown. We have not pursued this question, because the primary goal of the study was COS bond cleavage. Defluorination and aromatic ring opening prior to sulfate release are not included in the proposed metabolic pathway, because fluoride ion was not detected in the culture medium. Even if only 1 in 10 fluoride ions per molecule of PFPS had been removed, the fluoride ion concentration would have risen to 6.2 mg/liter (30 times higher than the detection limit). Defluorination of oxidized aromatic rings has been previously reported among members of the Actinomycetales (9, 44) and Pseudomonadaceae (10,41,42), but (in general) fluorinated compounds are considered to be more recalcitrant than other halogenated organics (4,11). The novelty of the desulfurization activity exhibited by strain JVH1 is best illustrated by comparison to known DBT-desulfurizing bacteria unable to utilize alkyl sulfides other than DMSO as sulfur sources (19). Interestingly, Arensdorf et al. (1) reported that a Dsz-negative mutant of strain IGTS8 can transform DMSO, indicating that the DBT-desulfurization pathway is not required for this activity. In the results seen with this study, the detection of PFPSO in trace amounts (and the near stoichiometric recovery of PFPSO 2 from the four DBTdesulfurizing cultures) supports the hypothesis that these strains are unable to cleave COS bonds in alkyl chains. Assuming that these strains employ the dsz operon for DBT desulfurization (6,7,31), it would appear that DszC, a sulfoxide-sulfone monooxygenase (30) that catalyzes the conversion of DBT to DBT-sulfoxide to DBT-sulfone (7,38), has sufficiently broad substrate specificity to catalyze the first two steps of PFPS oxidation. This observation agrees with the work by Lei and Tu (30), who found that purified DszC from IGTS8 is able to oxidize benzyl sulfide and benzyl sulfoxide to benzyl sulfone.
In contrast, it would appear that DszA is not active against PFPSO 2 . These results are in agreement with recent work by Arensdorf et al. (1), who used a chemostat to isolate mutants of strain IGTS8 that could utilize octyl sulfide, 5-methylbenzothiophene, and benzothiophene as sulfur sources. In their work, specific single-nucleotide changes in DszA were responsible for expanding the normally narrow substrate range of this enzyme. Once the enzymes involved in desulfurization by strain JVH1 are isolated, comparative studies can be undertaken.
In conclusion, the novel organosulfur compound PFPS was effective in selecting microorganisms capable of cleaving COS bonds in alkyl chains. This activity will be valuable in desulfurization and viscosity reduction studies (provided that preferential use of more-bioavailable sulfur compounds can be controlled). To this end, reaction mechanisms involved in PFPS oxidation and cleavage are being investigated, and current work is focused on locating the novel desulfurization genes and enzymes in strain JVH1.