APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 2003, p. 4689–4696 Vol. 69, No. 80099-2240/03/$08.00�0 DOI: 10.1128/AEM.69.8.4689–4696.2003Copyright © 2003, American Society for Microbiology. All Rights Reserved.

Growth of Rhodosporidium toruloides Strain DBVPG 6662 onDibenzothiophene Crystals and Orimulsion

Franco Baldi,1,2* Milva Pepi,2 and Fabio Fava3

Department of Environmental Sciences, Ca Foscari University, S. Marta, Dorsoduro 2137, I-30121 Venice,1 and InteruniversityConsortium for Environmental Chemistry, Marghera, I-30175 Venice,2 and DICASM, Faculty of Engineering,

University of Bologna, I-40136 Bologna,3 Italy

Received 29 January 2003/Accepted 2 June 2003

Strains DBVPG 6662 and DBVPG 6739 of Rhodosporidium toruloides, a basidiomycete yeast, grew onthiosulfate as a sulfur source and glucose (2 g liter�1 or 10.75 mM) as a carbon source. DBVPG 6662 has adefective sulfate transport system, whereas DBVPG 6739 barely grew on sulfate. They were compared for theability to use dibenzothiophene (DBT) and related organic sulfur compounds as sulfur sources. In the presenceof glucose as a carbon source and DBT as a sulfur source, strain DBVPG 6662 grew better than DBVPG 6739.In the presence of thiosulfate as a sulfur source, the two yeast strains did not use DBT, DBT-sulfone,benzenesulfonic acid, biphenyl, and fluorene. When the two strains were grown in the presence of glucose,strain DBVPG 6662 transformed 27% of the DBT present (10 �M) at a rate of 0.023 �mol liter�1 h�1 in 36 h.Traces of 2,2�-dihydroxylated biphenyl were transiently accumulated under these conditions. When the samestrain was grown on glucose in the presence of a higher concentration of DBT (0.5 g liter�1), mainly in aninsoluble form, the whole surface of the DBT crystals was colonized by a thick mycelium. This adherentstructure was imaged by confocal microscopy with fluorescent concanavalin A, a lectin that specifically bindsglucose and mannose residues. When DBVPG 6662 was grown on glucose in the presence of a commercialemulsion of bitumen, i.e., orimulsion, 68% of the benzo- and dibenzothiophenes and DBTs was removed after15 days of incubation. The fungus adhered by hyphae to orimulsion droplets. When cultivated in the presenceof commercial emulsifier-free fuel oil containing alkylated benzothiophenes and DBTs and having a compo-sition similar to that of orimulsion, strain DBVPG 6662 removed only 11% of the total organic sulfur thatoccurs in the medium and did not adhere to the oil droplets. These results indicate that strain DBVPG 6662is able to utilize the organic sulfur of DBT and a large variety of thiophenic compounds that occur extensivelyin commercial fuel oils by physically adhering to the organic sulfur source.

Biological removal of sulfur from fossil fuels to decrease thequantity of the sulfur oxides produced during their combustionis a challenge for microbiologists and other scientists. Withprokaryotes, biocleaning of fossil fuels has only been partiallysuccessful because chemolithotrophic bacteria and archaea (8,29, 30) only remove inorganic sulfur. Various scale-up pro-cesses are available for removal of inorganic sulfur, but noindustrialized process for organic sulfur degradation has beendeveloped. The removal of organic sulfur, i.e., the sulfur co-valently bound in complex organic molecules, such as mercap-tans, thiophenols, thioethers, dithioethers, and dibenzothiophene(DBT), is being investigated by many laboratories, and differ-ently adapted microbial cultures are being tested. Many strainsare not useful because they mineralize the carbon skeleton (9,12, 13, 21, 24, 25). Bacteria that desulfurize organic sulfurcompounds without metabolizing the carbon skeleton are lesscommon and generally use a sulfur-selective oxidative pathway(12, 15, 18, 30, 35, 36). Isolation of new microorganisms capa-ble of carbon-sulfur bond-targeted desulfurization is still ofprimary interest. Bacteria belonging to the gram-positive do-main (7, 17, 19, 34), mainly from the genus Rhodococcus (22,27), are responsible for DBT desulfurization. Few gram-nega-

tive bacteria are capable of organic sulfur degradation; somerecently reported ones are Sphingomonas (26) and Paenibacil-lus (20) spp. Among eukaryotic organisms, Cunninghamellaelegans has been reported to desulfurize DBT (9). This fungusgrows on DBT by forming DBT-5-oxide and DBT-5-dioxidebut not biphenyl. The fungus Paecylomyces sp. carries out sul-fur-specific oxidation of DBT by producing 2,2�-dihydroxybi-phenyl (11). No yeasts have been reported to grow desulfuriz-ing DBT and/or related thiophenic compounds that extensivelyoccur in industrial fuel oils.

Rhodosporidium toruloides strain DBVPG 6662 was isolatedfrom industrial waste several years ago (2, 31, 32) and grows onthiosulfate as a sulfur source in the presence of glucose as acarbon source. The aim of the present study was to investigateand characterize the ability of this strain to utilize DBT andrelated organic sulfur compounds that occur in fossil and oilfuels as sulfur sources.

MATERIALS AND METHODS

Yeast cultures in organic and inorganic sulfur sources. Two strains from theindustrial yeast collection of the Dipartimento di Biologia Vegetale Universita diPerugia (DBVPG), belonging to the European Culture Collection Organization,were compared for organic sulfur utilization: DBVPG 6739, which grows to avery limited extent on sulfate as a sulfur source, and DBVPG 6662, which doesnot. The two strains were routinely grown in Bacto Sabouraud Dextrose Broth(Difco, Detroit, Mich.) and incubated overnight at 28°C in a rotary drum. Cellswere harvested by centrifugation at 6,000 � g for 20 min and washed twice with10 mM PIPES [(piperazine-N,N�-bis(2-ethanesulfonic-acid)] buffer (Sigma, Mi-

* Corresponding author. Mailing address: Department of Environ-mental Sciences, Ca Foscari University, Calle Larga, S. Marta, Dor-soduro 2137, 30121 Venice, Italy. Phone: 39-041-2578432. Fax: 39-041-2348901. E-mail: baldi@unive.it.

4689

lan, Italy) at pH 7.4. They were then inoculated into 250-ml conical flaskscontaining 50 ml of a minimal medium (MM) consisting of 2 g of yeast nitrogenbase liter�1 without amino acids (Difco) and 0.011 mM ammonium chloride(Aldrich, Milan, Italy), which replaced ammonium sulfate, to lower the sulfatecontent of the MM to 3.1 mM. The carbon source was 2 g of D-glucose (Aldrich)liter�1 (10.75 mM); sulfate (Aldrich), thiosulfate (Aldrich), or sulfite (Aldrich)was used, at concentrations ranging from 0.080 to 6.2 mM, as a sulfur source.

To demonstrate that DBT is a sulfur source for these yeast strains, inorganicsulfur was replaced with an organic sulfur source such as 500 mg of DBT (Merck,Milan, Italy) liter�1 (2.71 mM) with or without 2 g of D-glucose liter�1 in MM.For a control, fluorene (Aldrich), a sulfur moiety-free DBT analog, was alsotested at 500 mg liter�1 (3.0 mM) as a potential carbon source for both yeaststrains in MM with thiosulfate (5 mM) as a sulfur source. A further controlexperiment was also performed with MM plus glucose (2 g liter�1) but no addedsulfur source.

Yeast biomass. Yeast biomass was determined by (i) counting cells stainedwith acridine orange by epifluorescence microscopy (Axiophot; Zeiss, Ober-kochen, Germany), (ii) analyzing total protein by the Bradford method (3), (iii)measuring A600 with a UV-visible light spectrophotometer (Shimadzu UV-160),and (iv) weighing the dried biomass on a filter. Glucose concentration wasdetermined as described by Miller (23). Growth yields were determined atseveral different initial glucose concentrations in the presence of DBT. The yielddetermination was based on glucose consumption, which was measured by treat-ing 1.5-ml samples of the yeast culture with 1.5 ml of a reagent containing 1%(wt/vol) dinitrosalicylic acid, 0.2% (wt/vol) phenol, 0.05% (wt/vol) sodium sulfite,and 1% (wt/vol) sodium hydroxide (all chemicals were from Aldrich). The glu-cose concentration was then determined spectrophotometrically at 575 nm(A575).

Studies of oxygen consumption. The yeast strains were pregrown on MMcontaining 2 g of D-glucose liter�1 (10.75 mM) and 50 �M DBT. DBT-inducedcells were harvested in the exponential growth phase (by centrifugation at 6,000� g for 20 min), washed twice with phosphate buffer (pH 7.4), and resuspendedin 5 ml of fresh MM. A 100-�l aliquot of the resulting cell suspension was addedto 3 ml of fresh phosphate buffer and introduced into the measuring cell of anoxygen monitor (model 5300; Yellow Springs Instrument Co., Yellow Springs,Ohio) equipped with a Clark-type oxygen electrode. DBT, benzothiophene (Al-drich), DBT-sulfone (Aldrich), benzenesulfonic acid (Aldrich), biphenyl (Al-drich), and fluorene were dissolved in N,N�-dimethylformamide (Carlo Erba,Milan, Italy) and added to cell suspensions to a final concentration of 50 �M.Depending on the experiment, 10.75 mM glucose or 5 mM thiosulfate was alsoadded to the assay vessel as a carbon or sulfur source, respectively. Oxygenconsumption was measured at 28°C. The instrument was calibrated constantlywith air-saturated fresh MM. The measurements were corrected for endogenousrespiration and chemical substrate oxidation. Yeast concentration was expressedin terms of total cell protein determined by the Bradford method (3).

DBT biodegradation in liquid cultures. DBT biodegradation by the two R.toruloides strains was studied in cells growing in batch cultures. The strains wereinoculated into sterile conical flasks (1 liter) containing 200 ml of sterile MMamended with 10 �M DBT (by spiking each culture with 0.2 ml of a 10 �M DBTsolution in dimethylformamide) as a sulfur source and 10.75 mM glucose as acarbon source. An identical culture was developed with no inoculum as an abioticcontrol. All flasks were closed with sterile aluminum foil and incubated at 28°Cin static mode to reduce abiotic DBT loss due to volatilization. DBT and itspotential biodegradation metabolites were extracted from the cultures. Dupli-cate 10-ml samples were removed from each culture at pre-established times.Each sample was placed in a 20-ml vial equipped with a Teflon-coated butylstopper and an aluminum crimp sealer, and the samples were subjected to batchsolvent extraction with 1 ml of hexane in an ultrasonic bath. The organic phasewas then analyzed through a Beckman high-performance liquid chromatography(HPLC) system equipped with a Beckman Ultrasphere octyldecyl silane column(4.6 by 25 mm; particle diameter, 5 �m) and a Beckman Coulter 168 SystemGold diode array detector operating at 240 and 250 nm. The column temperaturewas 35°C; the injection volume was 20 �l. The solvents used were water acidifiedwith 1% (vol/vol) acetic acid (A) and methanol acidified with 1% (vol/vol) aceticacid (B). The solvent gradient method used was as follows: initial solvent com-position, 60% A and 40% B; isocratic elution for 15 min; solvent compositionchanged to 20% A and 80% B in 20 min; isocratic elution for 8 min; and back tothe initial solvent composition for 12 min. Five standard solutions ranging from5 to 100 �M DBT were used for calibration. Extraction efficiency was determinedby the standard addition method; it was typically found to be 93% � 3%. Fivereplicates of the same sample gave a coefficient of variation of 7.3%. The DBTdetection limit of the method was 0.54 �M.

The water phase resulting from DBT extraction was then acidified to pH 2.5

with trichloroacetic acid (2 M; Aldrich) and subjected to two successive batchextractions with 5 ml of ethyl acetate to recover polar metabolites of DBTdegradation. The organic phase obtained was evaporated to dryness under an N2

stream. The residue collected was redissolved in methanol to a final volume of0.5 ml. Qualitative and quantitative analyses of the main DBT desulfurizationorganic metabolites, such as DBT-sulfone, benzothiophene, benzenesulfonicacid, biphenyl, and 2,2�-dihydroxybiphenyl, that occur in the hexane fraction andin the 20-fold-concentrated ethyl acetate fraction were done by the same HPLCprocedure. One-milliliter samples were also periodically taken from each of theyeast cultures for analysis of the main inorganic metabolites of DBT desulfuriz-ation. Each sample was harvested and centrifuged at 11,000 � g for 3 min(Eppendorf 5415), and 0.2 ml of the resulting supernatant was analyzed forsulfite concentration by the iodometric method (14) and for sulfate by ionchromatography. In the latter case, the anion concentration was measured witha Dionex DX-120 IC system equipped with an IonPac AS14 column (4 by 250mm) and a conductivity detector combined with an ASRS-Ultra conductivitysuppressor system (Dionex Corporation, Sunnyvale, Calif.). The eluent was asolution of 3.5 mM Na2CO3 and 1.0 mM NaHCO3 prepared in ultra-resi-ana-lyzed water; the flow rate was 1.2 ml min�1, and the injection volume was 20 �l.A linear four-point calibration curve (range, 0.2 to 20.0 mg/liter, i.e., 0.0021 to0.21 mM) for SO4

2� was developed by using a pure standard of this compound.The limit of SO4

2� detection was 0.052 �M.Morphology changes and cell adhesion determined by CSLM. Morphology

changes in strain DBVPG 6662 from vegetative cell to hypha formation weremonitored in MM with 2 g of glucose liter�1 (10.75 mM) in the presence of DBTcrystals at a concentration (500 mg liter�1) much higher than the water solubilityof DBT. Strain 6662 was also grown in MM with 2 g of glucose liter�1 amendedwith 1 g of orimulsion 4097 or with fuel oil (MTZ 4098) containing organic sulfur(determined as described below). The culture was sampled at pre-establishedtimes throughout the experiment, and the collected sample was analyzed byconfocal scanning laser microscopy (CSLM). A confocal microscope (MRC-500;Bio-Rad Microscience Division) mounted on a Nikon Microphot microscope wasused to observe strain DBVPG 6662 growing on MM with DBT crystals (500 mgliter�1), orimulsion, or fuel oil. Yeast colonization of crystals was checked afterdifferent periods of inoculation. To determine polysaccharide production duringgrowth on the different thiophenic compounds, a 1-ml aliquot of suspension wasincubated for 1 h with 5 �l of concanavalin A (ConA; 1 mg ml�1; Sigma)conjugated with fluorescein isothiocyanate (FITC; C7642; Sigma) (5). IndividualCSLM images were horizontal high-resolution optical thin sections (10 nm) (6).A krypton-argon laser with maximum emission lines at 488 nm was used tomeasure the excitation source of the fluorescent lectin. Black and white photo-graphs were taken with T-max 100 ASA film (Kodak, Rochester, N.Y.). Imageanalysis of recorded sections of samples was carried out with Bio-Rad Comossoftware.

Determination of organic sulfur in orimulsion and fuel oil. Strain DBVPG6662 was also tested for the ability to take up sulfur from organic sulfur incommercial fuels. The yeast was transferred five times at 28°C in MM containing2 g of glucose liter�1 (10.75 mM) and 1 mM thiosulfate as a sulfur source toavoid DBT contamination. The strain was then inoculated (0.021 � 0.005 mg ofprotein ml�1) into duplicate flasks (250 ml) containing 50 ml of MM with glucose(10.75 mM) and 1 g (20 g liter�1) of orimulsion 4097 or of MTZ 4098 fuel oil(with a medium sulfur content of 1.5%, wt/wt). These materials were kindlysupplied by the ENEL Waste Treatment Research Center (Brindisi, Italy).Orimulsion is a fuel consisting of natural bitumen dispersed in fresh water (26 to30%, wt/wt), stabilized by addition of a patented surfactant (33). The qualitativeand quantitative analyses of the organic sulfur compounds that occur in thecultures amended with the two commercial products were carried out at thebeginning and end (after 15 days) of the culture incubation period. The batchesof whole cultures were extracted (in three successive steps) with dichlorometh-ane in accordance with Environmental Protection Agency SW-846 method3510B (10). Extracts were dried on a column of anhydrous sodium sulfate(Aldrich) and carefully concentrated to a final volume of 10 ml. Aliquots (1 ml)of the resulting solution were diluted to 30 ml with dichloromethane. A 2-cmcolumn with activated silica gel, pretreated at 150°C for 24 h, was used to removeparticles and very high-molecular-weight components as previously described(16) for oil spill characterization. Aliquots of 1 �l were injected in the splitlessmode in a gas chromatograph-mass spectrometer (Finnigan Magnum) equippedwith an SPB-5 fused silica column (0.32-�m film thickness; 30 by 0.32 m [insidediameter]; Supelco). The carrier gas was helium at 1,110 kPa. The injectortemperature was 200°C. The analytical conditions were injection and holding for1 min at 35°C and then a temperature increase of 10°C min�1 to 300°C. Com-pounds were detected by ion trap mass spectrometry. The ion trap operated at anominal voltage of 70 eV, scanning 80 to 300 atomic mass units s�1. DBT and

4690 BALDI ET AL. APPL. ENVIRON. MICROBIOL.

benzothiophene were identified by analysis of standard solutions and mass spec-tra. Methylated residues of these sulfur compounds were also identified by ionprofile (m/z 148, 162, and 176 for methylbenzothiophene, dimethylbenzothio-phene, and trimethylbenzothiophene, respectively; m/z 198, 212, and 226 formethyl-DBT, dimethyl-DBT, and trimethyl-DBT, respectively). Methylated res-idues were quantified with respect to the sum of the peaks of each class ofcompounds and normalized to DBT.

RESULTS

Yeast growth on different sulfur species. R. toruloides strainDBVPG 6662 cultivated in the presence of different inorganicsulfur sources grew on thiosulfate (�1 mM) but not at all onsulfate or sulfite (up to 80 �M). Strain DBVPG 6739 grew ona lower concentration of thiosulfate (�0.080 mM) and slightlyon sulfate but not on sulfite (Fig. 1). Strain DBVPG 6662 alsogrew on high concentrations of DBT at a rate of 5.67 mg ml�1

day�1 but only in the presence of glucose as a carbon andenergy source. It did not grow on DBT as a combined sourceof energy, carbon, and sulfur. Strain DBVPG 6739 grew at aslower rate (0.715 mg ml�1 day�1) under the same laboratoryconditions. Neither strain grew on fluorene, as a DBT analoglacking a sulfur moiety, in the presence of thiosulfate (Fig. 2).

The metabolism of DBT, DBT-sulfone, and benzenesulfonicacid, along with that of biphenyl, fluorene, and the solventN,N�-dimethylformamide, by the two yeast strains was alsostudied through O2 consumption measurements. None of theorganic compounds tested was oxidized by R. toruloidesDBVPG 6662 and DBVPG 6739 in the absence of glucose(Table 1). However, in the presence of glucose, DBT, DBT-sulfone, and benzenesulfonic acid significantly sustained thegrowth of DBVPG 6662 and poorly sustained that of strainDBVPG 6739. In the presence of D-glucose with no organicsulfur source, the yeast strains did not grow. Fluorene, biphe-nyl, and N,N�-dimethylformamide were not oxidized by DBT-induced cells of both strains in the presence of thiosulfate(Table 1).

To demonstrate direct DBT transformation, two batch ex-periments were performed with both strains in glucose-

amended MM with dissolved DBT (10 �M). DBT was found tobe metabolized partially by both cultures. Strains DBVPG6662 and DBVPG 6739 exhibited average DBT degradationrates (calculated by subtracting the average depletion ratemeasured in the abiotic culture from those recorded in theparallel inoculated cultures) of 0.023 and 0.0088 �mol liter�1

h�1, respectively, with average percentages of removal of theinitially occurring DBT of 24 and 17%, respectively, after 36 hof incubation (Fig. 3). The DBT depletion observed in thestrain 6739 cultures was only slightly faster and higher thanthat observed in the corresponding abiotic controls. In fact, adetectable amount of the spiked DBT was also found to dis-appear over time in the abiotic controls (Fig. 3); analysescarried out on the headspace gas of identical cultures de-scribed in parallel in hermetically closed microcosms (data notshown) indicate that the detected DBT abiotic losses were

FIG. 1. Growth of R. toruloides strains DBVPG 6662 (‚, E, �) andDBVPG 6739 (Œ, F, ■ ) in the presence of the inorganic sulfur sourcesthiosulfate (‚, Œ) sulfite (E, F), and sulfate (�, ■ ). The mineralmedium contained a background sulfate level of about 300 mg liter�1

(3.1 mM).

FIG. 2. Growth curves of R. toruloides strains DBVPG 6662 (‚, E,�) and DBVPG 6739 (Œ, F, ■ ) in the presence of DBT with glucose10.75 mM (‚), DBT without glucose (E), and fluorene with 5 mMthiosulfate (�).

TABLE 1. Substrate-dependent oxygen uptake by cells of R.toruloides DBVPG 6662 and DBVPG 6739a

Substrate(s)b

Avg rate of oxygen consumption (nmolof O2 min�1 mg of protein�1 � SD)

R. toruloides6662

R. toruloides6739

D-Glucose NDc NDDBT ND NDDBTS ND NDBSa ND NDBP ND NDFN ND NDDMFM ND ND

DBT � D-glucose 37.85 � 3.04 NDDBTS � D-glucose 39.82 � 4.29 1.31 � 0.6BSa � D-glucose 33.92 � 1.01 0.87 � 0.5

BP � thiosulfate ND NDFN � thiosulfate ND ND

a Cells were pregrown in a medium containing 50 �M DBT and then harvestedand resuspended in phosphate buffer with different substrates (500 �M).

b DBTS, DBT-sulfone; BSa, benzensulfonic acid; BP, biphenyls; FN, fluorene;DMFM, N,N�-dimethylformamide.

c ND, not detectable (�0.5 nmoles of O2 min�1 mg of protein�1).

VOL. 69, 2003 RHODOSPORIDIUM TORULOIDES GROWS ON DBT 4691

mostly due to volatilization phenomena. Traces of 2,2�-dihy-droxylated biphenyl were transiently accumulated in theDBVPG 6662 cultures, whereas none of the metabolitessought, such as DBT-sulfone, sulfate, and sulfite, was producedin either the DBVPG 6662 or the DBVPG 6739 cultures. Thedetected compound in DBVPG 6662 cultures was character-ized as 2,2�-dihydroxylated biphenyl, as it coeluted with pure2,2�-dihydroxybiphenyl under a variety of different HPLC an-alytical conditions and displayed a diode array UV-visible lightabsorption spectrum identical to that of the pure compound.Biphenyl was detected (at 0.0654 � 0.0032 �M) in both cul-tures and in the control from the beginning of the experiment.This compound was neither produced nor metabolizedthroughout the whole experiment; it was found to be an im-purity in the DBT product used in the study.

Glucose metabolism and biomass growth in the DBT-amended cultures were studied through a parallel experimentcarried out under the same conditions. DBVPG 6662 metab-olized glucose rapidly and extensively (Fig. 4), exhibiting adoubling time of 5.9 h, producing 0.139 mg of protein liter�1,consuming 1.5 mg of D-glucose, and giving a yield of 0.085. Bycontrast, strain DBVPG 6739 metabolized glucose weakly (Fig.4) and exhibited a doubling time of 20.0 h, a cell production of0.016 mg of protein liter�1, a glucose consumption of 0.13 mg,and a cell yield of 0.01.

DBVPG 6662 growth in the presence of DBT crystals andcommercial fossil fuels. Given its ability to use DBT and otherorganic sulfur compounds as a source of sulfur, strain DBVPG6662 was subjected to further physiological and metabolic in-vestigations. In particular, when DBVPG 6662 interacted withDBT crystals, it underwent a morphologic transformation fromyeast cells to hyphae. This event was first investigated by con-focal microscopy with the lectin ConA-FITC. This lectin spe-cifically binds the carbohydrate residues of D(�)-glucose andD(�)-mannose (1). The cell wall of R. toruloides is known tocontain glucose, mannose, galactose, and glucosamine, butonly mannose seems to be involved in the adhesion process (9).

DBVPG 6662 cells growing in MM with glucose (10.75 mM)and DBT (500 mg liter�1) were found to adhere to and colo-nize only DBT crystals (Fig. 5A) and not the walls of theculture vessel. At the start, the contact with DBT caused asignificant reduction in cell volume. The smallest cells (size, 1.4� 0.13 �m) showed a huge concentration of ConA fluores-cence at the cell envelope (Fig. 5B). Yeast cells (size, 5 � 0.34�m) were not, or only faintly, stained by ConA. After 6 days ofincubation, hypha formation was induced and the cells formeda thick mycelium (Fig. 5C) on the DBT crystal surface. Con-focal microscopy image analysis of a series of 20 scanned se-quential optical planes (x and y), 0.6 �m in thickness, andtherefore equivalent to 12 �m in the z plane, showed myceliumrich in exopolysaccharides (Fig. 5D), which were concentratedmainly at the poles of elongated hyphal cells (Fig. 5E). Themycelium developed and branched, forming a large epiphyticbiomass solidly attached to the hydrophobic DBT crystals.

Strain DBVPG 6662 was also investigated for the ability touse organic sulfur in commercial fossil fuels. The strain wasfound to grow significantly on glucose in the presence oforimulsion, i.e., a bitumen amended with an emulsifying agentand water. However, the same strain grew only slightly in thepresence of the MTZ 4098 fuel oil (data not shown). Bothcommercial products contain a number of sulfur compoundsheavier and more hydrophobic than DBT, such as methylated,dimethylated, and trimethylated benzo- and dibenzothio-phenes, which constituted about 96% (wt/wt) of the orimulsionand 80% (wt/wt) of the fuel oil (Table 2). The growth of thestrain throughout the 15 days of incubation resulted in theremoval of about 68% of the organic sulfur initially detected inthe orimulsion-amended culture and of 11% in that in the fueloil culture. DBVPG 6662 cells were found to adhere to andaggregate orimulsion droplets (Fig. 6A) by producing the

FIG. 3. Consumption of DBT (10 �M) in liquid cultures of R.toruloides strains DBVPG 6662 (■ ) and DBVPG 6739 (F) and in anuninoculated culture (control) (E) throughout a 36-h experiment. Re-sults are means of single determinations on duplicate cultures; errorbars not visible are smaller than the symbols and show ranges ofduplicate samples.

FIG. 4. Consumption of D-glucose (2 mg liter�1) as a carbon andenergy source by cultures of R. toruloides strains DBVPG 6662 (‚) andDBVPG 6739 (E) and in an uninoculated culture (�) during 120 h ofincubation at 28°C in the presence of dissolved DBT (10 �M) as asulfur source. Simultaneous determination of the biomass (microgramsof protein per milliliter) that occurred in the cultures of strainsDBVPG 6662 and DBVPG 6739 is also shown (Œ and F, respectively).Results are means of single determinations on duplicate cultures; errorbars not visible are smaller than the symbols and show ranges ofduplicate samples.

4692 BALDI ET AL. APPL. ENVIRON. MICROBIOL.

ConA-positive exopolysaccharides. This was followed by a sig-nificant reduction in cell size and polar adhesion of cells toproduce hyphae (Fig. 6B), extending almost to the samegrowth pattern as on DBT crystals. At the end of the incuba-tion period (day 15), the hyphae were longer than those on

DBT crystals (Fig. 6C) and they trapped orimulsion droplets.Staining of specimens with ConA-FITC showed clearly thathyphae were rich in exuded exopolysaccharides. The orimul-sion droplet surface was strongly fluorescent, being coated bythis polysaccharide biofilm (compare Fig. 6D and B).

FIG. 5. (A) Cells of R. toruloides strain DBVPG 6662, observed intransmission mode, at the beginning of their colonization of DBTcrystals (500 mg ml�1). Bar, 25 �m. (B) The same specimen observedin fluorescence mode after staining with ConA-FITC; smaller yeastcells became highly fluorescent because of exopolysaccharide produc-tion. Bar, 25 �m. (C) After 6 days of incubation, DBT crystals, ob-served in transmission mode, were intensively colonized by a myceliumproduced by the basidiomycete R. toruloides. Bar, 25 �m. (D) Thesame specimen observed in fluorescence mode after staining withConA-FITC. Hyphae are producing a large amount of exopolysaccha-ride. Bar, 25 �m. (E) Detail showing that the largest amounts ofexopolysaccharide were produced at the polar regions of elongatedcells. Bar, 5 �m.

VOL. 69, 2003 RHODOSPORIDIUM TORULOIDES GROWS ON DBT 4693

DISCUSSION

The present study originated from previous research onCr(VI) resistance in yeasts (2, 31) in which it was found that R.toruloides strain DBVPG 6662 has a variety of transport sys-tems for reduced forms of sulfur and an inefficient sulfatetransport system. This was naturally shut down to avoid uptakeof chromate, a natural competitor for sulfate uptake (2, 28).The basidiomycete yeast was isolated from chromate-rich in-dustrial wastes and was the only microorganism capable ofgrowing at Cr(VI) concentrations of up to 500 �g ml�1 (31).

Several experimental findings indicate that R. toruloidesDBVPG 6662 requires reduced forms of sulfur for growth.Indeed, the strain grew on reduced forms of inorganic sulfursources, such as thiosulfate, and of organic sulfur sources, suchas cysteine, methionine, and thioglycolic acid (32), along withDBT, DBT-sulfone, and benzenesulfonic acid (Fig. 1 and 2 andTable 1). The utilization of the organic sulfur compounds isstrictly dependent on the presence of glucose in the culturemedium. A lack of growth (Fig. 1 and 2) or oxygen consump-tion (Table 1) in media with such organic sulfur sources in theabsence of glucose and a lack of growth and an absence ofoxygen consumption in the presence of fluorene (a DBT ana-logue with no sulfur moiety) in the presence of thiosulfateindicate that the reduced S-aromatic compounds were usedexclusively as a sulfur source by the strains. The absence ofdetectable sulfate or sulfite production in these cultures fur-ther supports this hypothesis. Unfortunately, little is knownabout the nature and fate of the organic metabolites resultingfrom sulfur removal from the metabolized organic sulfur com-

pounds; indeed, only traces of 2,2�-dihydroxybiphenyl weredetected in the DBT- and glucose-amended culture of strainDBVPG 6662, which was the strain that exhibited the greatestactivity on the range of organic sulfur sources utilized by thetwo yeasts used in the study (Fig. 1, 2, and 3). Our inability todetect other metabolites of DBT desulfurization probably re-sulted from the small content of the organic sulfur compoundsconsumed by the two strains in these tests (Fig. 3). Therefore,the metabolites were produced in quantities too small for de-tection by the analytic techniques used in this study.

When grown in the presence of DBT at a concentration inexcess of its water solubility, strain DBVPG 6662 extensivelycolonized the surface of DBT crystals in the culture medium(Fig. 5). This is an expected feature of this basidiomycetousstrain, which is a common phylloplane epiphyte that typicallyadheres to plant surfaces (5). However, the fact that growth didnot occur on the inner surface of the culture vessel suggeststhat intimate contact between yeast cells and the solid sulfursource is essential for efficient uptake of the sulfur nutrient. Tobetter investigate this phenomenon, the lectin ConA, a goodspecific marker for the trimannosyl core of N-linked glycans(1) to which the lectin strongly binds, was used to monitorglycosylation during yeast cell adhesion to DBT crystals. Thelectin used in this study turned out to be useful for identifica-tion of morphological and biochemical changes in the glyco-sylation of outer cell layers in response to adaptation to DBT.On the basis of the confocal microscopy of DBVPG 6662 cellsgrowing on DBT crystals (Fig. 5), a significant decrease in cellsize and a high level of mannoprotein production seemed to bethe prevalent adaptive mechanisms through which the strainperformed DBT metabolism and sulfur assimilation. The de-velopment of a thick wall of glycoproteins probably protectedthe cells and/or contributed to DBT dissolution, thus permit-ting gradual and controlled sulfur uptake. This hypothesis issupported by findings according to which a wild-type R. toru-loides strain adheres to surfaces by means of mannoproteins(i.e., glycoproteins containing mannose), which accounted for56% of the surface glycoproteins (4, 5), mostly concentrated inthe pole regions of cells (5).

Strain DBVPG 6662 was also found to massively adhere toand colonize orimulsion droplets by extensively coating theirsurface with glycoproteins (Fig. 6). However, no cell adhesionor colonization was observed on droplets of fuel oil exposed tothe strain under identical culture conditions. The two commer-cial products used were similar in organic sulfur compoundcomposition and content (Table 2). Thus, the extensive colo-nization that was observed only on orimulsion droplets may beascribed to the fact that orimulsion incorporates an emulsifier,which, according to the literature (36), may have significantlyenhanced the apparent water solubility and/or the dissolutionrate in water of the highly hydrophobic organic sulfur com-pounds, making them more available to yeast cells. The findingthat a higher and broader removal of the organic sulfur com-pounds was observed in the culture with orimulsion (68%)than in the culture with fuel oil (11%) (Table 2) seems tosupport the hypothesis according to which intimate contactbetween cells or mycelium and the organic sulfur source isrequired for effective sulfur source transformation and assim-ilation in strain DBVPG 6662.

In conclusion, R. toruloides strain DBVPG 6662 exhibited an

TABLE 2. Concentrations of the main organic sulfur compounds oforimulsion and fuel oil detected in sterile controls and the

biologically active DBVPG 6662 cultures after 15 daysof incubation

Organic sulfur compoundAvg concn (�g g�1) � SDb

Control Inoculateda

OrimulsionBenzothiophene NDc NDMethylbenzothiophene 0.9 2.75 � 1.34Dimethylbenzothiophene 19 5 � 1.4Trimethylbenzothiophene 42 12 � 4.2DBT 15 5.5 � 2.12Methyl-DBT 97 34 � 11.3Dimethyl-DBT 242 80 � 29Trimethyl-DBT 467 144 � 35Total organic sulfur 883 283 (32)

Fuel oil (MTZ 4098)Benzothiophene 2 3 � 0.98Methylbenzothiophene 23 29 � 7.1Dimethylbenzothiophene 77 83 � 19Trimethylbenzothiophene 104 106 � 22DBT 92 81 � 17Methyl-DBT 261 232 � 59Dimethyl-DBT 474 392 � 89Trimethyl-DBT 452 399 � 92Total organic sulfur 1,485 1,325 (89)

a The data in parentheses are average percentages of the total sulfur com-pounds detected in biologically active cultures with respect to that detected in thecorresponding controls at the end of the treatment.

b Duplicate analyses were performed on two different samples.c ND, not detected.

4694 BALDI ET AL. APPL. ENVIRON. MICROBIOL.

ability to utilize a large variety of organic sulfur compounds,including those that occur in commercial sulfur-rich productssuch as orimulsion and fuel oils, as sulfur sources. Intimatecontact between yeast cells and the sulfur source, provided byyeast colonization of organic sulfur crystals or oil emulsiondroplets, appears to be essential for more efficient organicsulfur source metabolism and sulfur uptake. The yeast seemsto selectively remove sulfur from the organic molecules with-out using their carbon skeleton as a carbon and energy source.The transient accumulation of 2,2�-dihydroxybiphenyl, alongwith our inability to detect other DBT desulfurization organicmetabolites, does not, however, exclude the possibility thatorganic intermediates of the process were degraded by theyeast through cometabolism. Anyhow, the yeast selectively uti-lizes DBT and thiophenic compounds but not many of theirstructurally correlated sulfur-free hydrocarbons, and this sug-gests that it may be used successfully for fuel and fossil oilindustrial desulfurization.

The results of this study are also of interest because theydocument for the first time that (i) the yeast strain R. toruloides

DBVPG 6662 is able to utilize a large variety of organic sulfurcompounds as sulfur sources, (ii) complex mixtures of alky-lated benzo- and dibenzothiophenes can be significantly me-tabolized and used as sulfur sources by a naturally occurringeukaryotic microorganism, and (iii) complex thiophenic frac-tions that occur in the commercial product orimulsion can beefficiently removed by R. toruloides strain DBVPG 6662, whichcan combine effective biotransformation of such sulfur frac-tions with the ability to produce an abundant mycelium. Thisbiomass can be easily removed from the oil bulk.

ACKNOWLEDGMENTS

This research was financed by ENEL.We thank G. Maspero (ENEL-Milan) for analysis of the organic

sulfur in orimulsion and fossil fuel and G. Varallo (ENEL, WasteTreatment Research Center, Brindisi, Italy) for supplying commercialoil samples.

REFERENCES

1. Baenziger, J. U., and D. Fiete. 1979. Structural determinants of concanavalinA specificity for oligosaccharides. J. Biol. Chem. 254:2400–2407.

FIG. 6. (A) Cells of R. toruloides strain DBVPG 6662, observed in transmission mode, starting to colonize the hydrophobic surface oforimulsion droplets. (B) The same specimen observed in fluorescence mode after staining with ConA-FITC. Yeast cells adhering to orimulsionbecame fluorescent because of exopolysaccharide production. Cells reduced their size and formed elongated aggregates (arrows). (C) After 15 daysof incubation, development of long hyphae by R. toruloides was observed in transmission mode. Mycelium entrapped orimulsion droplets. (D) Thesame specimen observed in fluorescence mode after staining with ConA-FITC. Orimulsion droplet became coated by exopolysaccharides,suggesting that strain 6662 has an intrinsic emulsifying activity different from that of the emulsifying agent already present in orimulsion.

VOL. 69, 2003 RHODOSPORIDIUM TORULOIDES GROWS ON DBT 4695

2. Baldi, F., A. Vaughan Martini, and G. J. Olson. 1990. Chromium(VI)-resistant yeast isolated from a sewage treatment plant receiving tannerywastes. Appl. Environ. Microbiol. 56:913–918.

3. Bradford, M. M. 1976. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye bind-ing. Anal. Biochem. 72:248–254.

4. Breierova, E., and Kockova-Kratochvilova. 1992. Cryoprotective effects ofyeast extracellular polysaccharides and glycoproteins. Cryobiology 29:385–390.

5. Buck, J. W., and J. H. Andrews. 1999. Attachment of the yeast Rhodospo-ridium toruloides is mediated by adhesives localized at the sites of bud celldevelopment. Appl. Environ. Microbiol. 65:465–471.

6. Caldwell, D. E., D. R. Korber, and J. R. Lawrence. 1992. Confocal lasermicroscopy and digital image analysis in microbial ecology. Adv. Microb.Ecol. 12:1–67.

7. Chang, J. H., S. K. Rhee, Y. K. Chang, and H. N. Chang. 1998. Desulfuriz-ation of diesel oils by a newly isolated dibenzothiophene-degrading Nocardiasp. strain CYKS2. Biotechnol. Prog. 14:851–855.

8. Clark, T. R., F. Baldi, and G. J. Olson. 1993. Coal depyritization by thethermophilic archaeon, Metallosphaera sedula. Appl. Environ. Microbiol.59:2375–2379.

9. Crawford, D. L., and R. K. Gupta. 1990. Oxidation of dibenzothiophene byCunninghamella elegans. Curr. Microbiol. 21:229–231.

10. Environmental Protection Agency. 1994. Separatory funnel liquid-liquid ex-traction; SW-846; method 3510B (rev. 2). In Test method for evaluating solidwaste. Environmental Protection Agency, Washington, D.C.

11. Faison, B. D., T. M. Clark, S. N. Lewis, C. Y. Ma, D. M. Sharkey, and C. A.Woodward. 1991. Degradation of organic sulfur compounds by a coal-solu-bilizing fungus. Appl. Biochem. Biotechnol. 28–29:237–251.

12. Grossman, M. J. 1996. Microbial removal of organic sulfur from fuels: areview of past and present approach, p. 345–359. In M. L. Occelli and R.Chianelli (ed.), Hydrotreating technology for pollution control: catalysts,catalysis and process. Marcel Dekker, Inc., New York, N.Y.

13. Grossman, M. J., M. K. Lee, R. C. Prince, K. K. Garrett, G. N. George, andI. J. Pickering. 1999. Microbial desulfurization of a crude oil middle-distil-lated fraction: analysis of the extent of sulfur removal and the effect ofremoval on remaining sulfur. Appl. Environ. Microbiol. 65:181–188.

14. IRSA-CRN. 1976. Metodi fisici e chimici: i solfiti, metodi analitici per leacque, parte II, vol. II. La Pergamena, Rome, Italy.

15. Izumi, Y., T. Ohshiro, H. Ogino, Y. Hine, and M. Shimao. 1994. Selectivedesulfurization of dibenzothiophene by Rhodococcus erythropolis D-1. Appl.Environ. Microbiol. 60:223–226.

16. Karstensen, K. H., O. Ringstad, I. Rustad, K. Kalevi, K. Jorgensen, K.Nylund, T. Alsberg, K. Olafsdottir, O. Heidenstam, and H. Solberg. 1992.Methods for chemical analysis of contaminated soil samples—tests of theirreproducibility between Nordic laboratories. Talanta 46:423–437.

17. Kayser, K. J., L. Cleveland, H. S. Park, J. H. Kwak, A. Kolhatkar, and I. J.Kilbane. 2002. Isolation and characterization of a moderate thermophile,Mycobacterium phlei GTIS10, capable of dibenzothiophene desulfurization.Appl. Microbiol. Biotechnol. 59:737–746.

18. Kilbane, J. J. 1989. Desulfurization of coal: the microbial solution. TrendsBiotechnol. 7:97–101.

19. Kim, S. B., R. Brown, C. Oldfield, S. C. Gilbert, S. Iliarionov, and M.

Goodfellow. 2000. Gordonia amicalis sp. nov., a novel dibenzothiophene-desulphurizing actinomycete. Int. J. Syst. Evol. Microbiol. 50(Pt. 6):2031–2036.

20. Konishi, J., T. Onaka, Y. Ishii, and M. Suzuki. 2000. Demonstration of thecarbon-sulfur bond targeted desulfurization of benzothiophene by thermo-philic Paenibacillus sp. strain A11–2 capable of desulfurizing dibenzothio-phene. FEMS Microbiol. Lett. 187(2):151–154.

21. Lee, M. K., J. D. Senius, and M. J. Grossman. 1995. Sulfur-specific microbialdesulfurization of sterically hindered analogs of dibenzothiophene. Appl.Environ. Microbiol. 61:4362–4366.

22. McFarland, B. L. 1999. Biodesulfurization. Curr. Opin. Microbiol. 2(3):257–264.

23. Miller, G. L. 1959. Use of dinitrosalicylic acid reagent for determination ofreducing sugars. Anal. Chem. 31:426–428.

24. Monticello, D. J., and W. R. Finnerty. 1985. Microbial desulfurization offossils fuels. Annu. Rev. Microbiol. 39:371–389.

25. Monticello, D. S., D. Bakker, and W. R. Finnerty. 1985. Plasmid-mediateddegradation of dibenzothiophene by Pseudomonas species. Appl. Environ.Microbiol. 49:756–760.

26. Nadalig, T., N. Raymond, Ni’matuzahroh, M. Gilewicz, H. Budzinski, andJ. C. Bertrand. 2002. Degradation of phenanthrene, methylphenanthrenesand dibenzothiophene by a Sphingomonas strain 2mpII. Appl. Microbiol.Biotechnol. 59(1):79–85.

27. Ohshiro, T., T. Hirata, and Y. Izumi. 1996. Desulfurization of dibenzothio-phene derivatives by whole cells of Rhodococcus erythropolis H-2. FEMSMicrobiol. Lett. 142:65–70.

28. Ohtake, H. C., C. Cervantes, and S. Silver. 1987. Decreased chromateuptake in Pseudomonas fluorescens carrying a chromate resistance plasmid. J.Bacteriol. 169:3853–3856.

29. Olson, G., B.-M. Pott, L. Larsson, O. Holst, and H. T. Karlsson. 1994.Microbial desulfurization of coal by Thiobacillus ferrooxidans and thermo-philic archaea. Fuel Proc. Technol. 40:277–282.

30. Omori, T., L. Monna, Y. Saiki, and T. Kodama. 1992. Desulfurization ofdibenzothiophene by Corynebacterium sp. strain SY1. Appl. Environ. Micro-biol. 58:911–915.

31. Pepi, M., and F. Baldi. 1992. Modulation of chromium(VI) toxicity by or-ganic and inorganic sulfur species in yeasts from industrial wastes. Biometals5:179–185.

32. Pepi, M., and F. Baldi. 1995. Chromate tolerance in strains of Rhodospo-ridium toruloides modulated by thiosulfate and sulfur amino acids. Biometals8:99–104.

33. Sommerville, M. 1999. Orimulsion containment and recovery. Pure Appl.Chem. 71:193–201.

34. van Afferden, M., S. Schacht, J. Klein, and H. G. Truper. 1990. Degradationof dibenzothiophene by Brevibacterium sp. DO. Arch. Microbiol. 153:324–328.

35. Volkering, F., A. M. Breure, and W. H. Rulkens. 1998. Microbiologicalaspects of surfactant use for biological soil remediation. Biodegradation8:401–417.

36. Wang, P., and S. Krawiec. 1994. Desulfurization of dibenzothiophene to2-hydroxybiphenyls by some newly isolated bacterial strains. Arch. Micro-biol. 161:266–271.

4696 BALDI ET AL. APPL. ENVIRON. MICROBIOL.