Engineering Hydrogen Sulfide Production and Cadmium Removal by Expression of the Thiosulfate Reductase Gene (phsABC) from Salmonella enterica Serovar Typhimurium in Escherichia coli

Abstract

The thiosulfate reductase gene (phsABC) from Salmonella enterica serovar Typhimurium was expressed in Escherichia coli to overproduce hydrogen sulfide from thiosulfate for heavy metal removal (or precipitation). A 5.1-kb DNA fragment containing phsABC was inserted into the pMB1-based, high-copy, isopropyl-β-d-thiogalactopyranoside-inducible expression vector pTrc99A and the RK2-based, medium-copy, m-toluate-inducible expression vector pJB866, resulting in plasmids pSB74 and pSB77. A 3.7-kb DNA fragment, excluding putative promoter and regulatory regions, was inserted into the same vectors, making plasmids pSB103 and pSB107. E. coli DH5α strains harboring the phsABC constructs showed higher thiosulfate reductase activity and produced significantly more sulfide than the control strains under both aerobic and anaerobic conditions. Among the four phsABC constructs, E. coli DH5α (pSB74) produced thiosulfate reductase at the highest level and removed the most cadmium from solution under anaerobic conditions: 98% of all concentrations up to 150 μM and 91% of 200 μM. In contrast, a negative control did not produce any measurable sulfide and removed very little cadmium from solution. Energy-dispersive X-ray spectroscopy revealed that the metal removed from solution precipitated as a complex of cadmium and sulfur, most likely cadmium sulfide.

Heavy metals are commonly found at many hazardous waste sites in industrialized countries. Many soluble metals can form insoluble complexes with hydroxides, carbonates, phosphates, and sulfides (21). One of the best-known natural metal precipitation mechanisms is due to sulfide production from sulfate by sulfate-reducing bacteria (SRB) found in anoxic sediments containing high concentrations of lead and mercury (9). A recent bioremediation technology utilizes hydrogen sulfide generated by SRB in anaerobic bioreactors to precipitate soluble metal species in aqueous waste streams as insoluble metal sulfides (25). The primary focus of this study was to develop a genetically engineered bacterium capable of producing sulfide under aerobic, microaerobic, or anaerobic conditions for heavy metal precipitation.

Among several bacterial hydrogen sulfide-generating systems, we chose the thiosulfate reductase gene (phsABC; phs represents production of hydrogen sulfide) from Salmonella enterica serovar Typhimurium to overproduce hydrogen sulfide. Thiosulfate reduction is a common but incompletely understood feature among bacteria (17). Thiosulfate reductase catalyzes the dissimilatory reduction of inorganic thiosulfate to hydrogen sulfide and sulfite (6). The enzyme has been purified from Desulfovibrio vulgaris (1), D. gigas (13), and a thermophilic iron-oxidizing bacterium, strain TI-1 (22).

Mutant and biochemical tests suggested that thiosulfate reductase activity from S. enterica serovar Typhimurium has an absolute requirement for the F₀F₁-ATP synthase (20). Sequence analyses of the chromosomal phsABC region from S. enterica serovar Typhimurium revealed a functional operon with three open reading frames (ORFs), designated phsA, phsB, and phsC (14). Amino acid sequence analyses revealed significant similarity between PhsA and the sequence of molybdoprotein oxidoreductases and between PhsB and the sequence of the iron-sulfur protein of the reductases. PhsC does not show any significant homology to any sequences in the GenBank database, but it retains characteristics similar to those of hydrophobic subunits of the reductases (14). Single-copy phs-lac translational fusions required both anaerobiosis and thiosulfate for full expression, whereas multicopy phs-lac translational fusions responded to either thiosulfate or anaerobiosis, suggesting that oxygen and thiosulfate control of the phs operon involves negative regulation (14).

There are several potential advantages of using thiosulfate and thiosulfate reductase for heavy metal remediation. First, thiosulfate is a relatively inexpensive source of sulfur for sulfide production. Second, thiosulfate is a weak metal chelator that facilitates mobilization of heavy metals in contaminated soils and is effective at reducing metal toxicity from some common metals in aquatic environments (15). As thiosulfate reductase catalyzes the stoichiometric production of hydrogen sulfide and sulfite from thiosulfate (6), the sulfite may be further reduced to sulfide by a group of bacteria, providing another equivalent of sulfide for metal precipitation. Finally, it should be possible to engineer sulfide-dependent metal removal by transferring the recombinant thiosulfate reductase system to certain environmental bacteria lacking in the dissimilatory sulfate reduction pathway.

MATERIALS AND METHODS

Strains and plasmids.

The bacterial strains and plasmids used in this study are listed in Table 1. All plasmids were transformed into Escherichia coli DH5α, and the thiosulfate reductase gene was expressed in the presence of thiosulfate. The plasmid (pEB40) containing the phsABC operon from S. enterica serovar Typhimurium was a gift from Ericka L. Barrett (University of California, Davis).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristicsa	Source or reference
E. coli DH5α	F− φ80dlacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17(rk−mk+) phoA supE44 λ−thi-1 gyrA96 relA1	BRLb
pTrc99A	Expression vector; Ptrc lac1qrrnBT1T2 Ampr	2
pJB866	Expression vector; Pm xylS839 oriT Tcr	4
pEB40	pUC19 with 5,136-bp EcoRI-SalI DNA fragment containing native phs regulatory and structural genes from S. enterica serovar Typhimurium LT2	8
pSB74	pTrc99A with 5,136-bp EcoRI-SalI DNA fragment containing native phs regulatory and structural genes from S. enterica serovar Typhimurium LT2	This study
pSB77	pJB866 with 5,136-bp EcoRI-SalI DNA fragment containing native phs regulatory and structural genes from S. enterica serovar Typhimurium LT2 (XhoI-compatible ends)	Bang et al.c
pSB103	pTrc99A with 3,653-bp PCR amplified DNA fragment containing only structural genes of phsA, phsB, and phsC from pSB74	This study
pSB107	pJB866 with 3,653-bp PCR amplified DNA fragment containing only structural genes of phsA, phsB, and phsC from pSB74	This study

^aAbbreviations: P_trc, hybrid trp-lac promoter; lac1^q, gene encoding the lac repressor protein; rrnBT1T2, transcription terminators; P_m, m-cleavage pathway promoter of the TOL plasmid; xylS839, gene encoding the repressor protein that controls P_m; oriT, origin of transfer; Amp^r, ampicillin resistant; Tc^r, tetracycline resistant; phs, gene for the production of hydrogen sulfide from thiosulfate. 

^bBRL, Gibco BRL, Inc., Gaithersburg, Md. 

^cS. W. Bang, D. S. Clark, and J. D. Keasling, Abstr. 99th Gen. Meet. Am. Soc. Microbiol. 1999, abstr. Q-302, p. 591, 1999. 

Molecular techniques.

Plasmid DNA was isolated by the alkaline-sodium dodecyl sulfate procedure of Birnboim and Doly (3) or purified using QIAGEN plasmid kits (Qiagen, Inc., Valencia, Calif.). Restriction digests and ligations of DNA samples were performed as recommended by the supplier (Roche Molecular Biochemicals, Inc., Indianapolis, Ind.).

PCR amplification was performed in accordance with the suggestions of the supplier of the Expand High Fidelity PCR System (Roche Molecular Biochemicals, Inc.) and the thermal cycler manufacturer (MJ Research, Inc., Waltham, Mass.). Oligonucleotide primers were synthesized by a commercial vendor (Genemed, Inc., San Francisco, Calif.). The primer sequences derived from the phsABC region were 5′-tcagcgaattctaataacaggagg-3′ (forward) and 5′-cattattttatggatccgctcagac-3′ (reverse) and 5′-tcagctggatccaataacaggagg-3′ (forward) and 5′-cattattttatgaattcgctcagac-3′ (reverse). Restriction sites for BamHI and EcoRI were inserted into the sequences for directional cloning and are underlined.

Restriction fragments containing the native and the PCR-amplified phsABC region were ligated into pTrc99A (2) and pJB866 (4) expression vectors. The constructed plasmids were then transformed into E. coli DH5α by the procedure of Hanahan (11), and the genes were expressed under appropriate culture conditions.

Culture conditions.

E. coli DH5α cells were preadapted to the morpholine propanesulfonic acid (MOPS) minimal medium used by several serial subcultures. E. coli DH5α cells harboring various phsABC genetic cassettes were inoculated onto a Luria agar (Miller's Luria broth agar) plate supplemented with appropriate antibiotics. The cells were incubated overnight at 37°C. A half inoculating loop of the cells from the plate was transferred into fresh MOPS medium (50 ml) supplemented with 10 mM glucose, thiamine at 1 μg/ml, and the appropriate antibiotic (ampicillin at 100 μg/ml for cells harboring pTrc99A derivatives or tetracycline at 12.5 μg/ml for cells harboring the pJB866 derivatives) and inducer (3 mM isopropyl-β-d-thiogalactopyranoside [IPTG] for cells harboring pTrc99A derivatives or 1 mM m-toluate for cells harboring the pJB866 derivatives). To prevent abiotic metal precipitation with phosphate, the K₂HPO₄ component of the original MOPS-buffered minimal medium (19) was replaced with glycerol 2-phosphate (1.32 mM). The modified MOPS medium was used in all of the experiments in this study. The cells were incubated overnight at 37°C with aeration and agitation (200 rpm). Five milliliters each of the overnight-grown cultures was transferred into 50 ml of fresh MOPS medium supplemented with the same components under the conditions described above. The cells were harvested by centrifugation at 6,000 × g for 10 min. The cell pellets were resuspended in various volumes of fresh MOPS medium to achieve the same cell density (optical density at 600 nm, 1.0) and used as an inoculum for further experiments. Aliquots (0.5 ml) of the cell suspensions were transferred to 50 ml of fresh MOPS medium supplemented with the same components plus 3 mM Na₂S₂O₃·5H₂O.

For cadmium removal experiments, various concentrations of CdCl₂ were added to the medium. The cultures were grown at 37°C without agitation to minimize hydrogen sulfide escape from the culture medium. A redox indicator, resazurin, was added into a separate series of cultures to indicate the reduction state of the culture.

Thiosulfate reductase activity of cell extracts.

E. coli DH5α strains harboring the phsABC constructs were grown in MOPS medium as described above. After several serial subcultures, the cells were inoculated into a fresh medium and grown at 37°C with shaking (for aerobic culture) or without shaking (for anaerobic culture). At an optical density at 600 nm of 0.7, the cells were harvested, washed twice with ice-cold 0.1 M Tris-acetate buffer (pH 9.0), and resuspended in the same buffer to the same cell density (approximately 3.5 × 10⁹ cells/ml). One milliliter each of the cultures was disrupted on ice by sonication (10 pulses at 10% duty cycle at a power setting of 2 on a Sonifier [model S-450; Branson Ultrasonic Co., Danbury Conn.]). After centrifugation (14,000 × g for 30 min at 4°C), the supernatants were collected and used for enzyme assay. Thiosulfate reductase activity in the cell extracts was determined by reacting the sulfite product with pararosaniline (5a). One unit of thiosulfate reductase activity is defined as the production of 1 μmol of sulfite in 1 min. Mean values from three replicate experiments are reported. The relative activity is the ratio of the thiosulfate reductase activity in the cell extracts from E. coli harboring phsABC constructs grown aerobically or anaerobically to that in the control cells grown aerobically.

Sulfide production by E. coli DH5α expressing phsABC.

Because thiosulfate interferes with the inorganic acid-labile sulfide assay (16), alternative methods to measure sulfide production were needed. For a simple sulfide detection assay, a semisolid LB agar medium (0.2% Noble agar in Luria broth) containing 2.5 mM FeCl₂·4H₂O and 3 mM Na₂S₂O₃·5H₂O with the appropriate antibiotic and inducer was used. The formation of a black precipitate (FeS) in the medium was considered to be an indication of sulfide production.

Direct measurement of sulfide was performed with a Sure-Flow Combination silver-sulfide electrode (model 9616; Orion, Inc., Beverly, Mass.). A closed capped serum vial was used in this experiment to prevent sulfide escape as hydrogen sulfide gas. Cell cultures of the same cell density (optical density at 600 nm, 0.5) were prepared in 250 ml of fresh MOPS medium supplemented with 10 mM glucose, thiamine at 1 μg/ml, the appropriate antibiotic and inducer, and 3 mM Na₂S₂O₃·5H₂O. A fraction of each culture (10 ml) was transferred into a 25-ml serum vial and capped with a rubber stopper and an aluminum seal. The vial was incubated at 37°C without shaking. At timed intervals, samples (2 ml) were withdrawn, filtered to collect cell-free supernatant, and added to an equal volume of sulfide antioxidant buffer, which was prepared in accordance with the suggestions of the electrode manufacturer (Orion, Inc.). The sulfide concentrations in the samples were measured using the sulfide probe, which was calibrated with known concentrations of Na₂S·9H₂O in the same antioxidant buffer. The samples were analyzed in triplicate, and mean values are reported. Sulfide concentrations of less than 50 μM were not detectable using the electrode.

Cadmium removal analysis.

All E. coli cells were grown and prepared as described in the culture conditions section. Cadmium was added to the medium to final concentrations between 50 and 200 μM. At various times, samples (1.0 ml) of culture medium were withdrawn. After centrifugation (10,000 × g for 15 min), 0.1 ml of the culture supernatant was filtered (Millipore MF membrane, 0.45-μm pore size) and transferred to 9.9 ml of 10% nitric acid solution to measure the concentrations of the metals remaining in the solution. The cadmium concentration in the samples was determined by inductively coupled plasma optical emission spectrometry (ICP-OES). Cadmium concentrations were determined in triplicate, and the values reported are means. The samples were prepared, and all ICP-OES operating settings were selected in accordance with the recommendations of the manufacturer (Perkin-Elmer, Norwalk, Conn.). Cadmium standards for ICP-OES analysis were prepared by dilution to obtain the desired concentrations of CdCl₂.

EDXS analysis of cadmium sulfide.

E. coli DH5α(pSB74) was grown overnight in the modified MOPS medium containing 200 μM CdCl₂. Small samples of the cells were fixed for 18 h in 0.1 M sodium cacodylate buffer (pH 7.4) containing 1% paraformaldehyde and 2% glutaraldehyde, washed twice in 0.2 M sodium cacodylate (pH 7.4) for 20 min, and then postfixed in unbuffered 2% osmium tetroxide with 2.5% potassium ferrocyanide for 2 h. The samples were dehydrated in graded ethanol, passed through a propylene oxide transition, and then infiltrated overnight in 1:1 propylene oxide and Spurr's resin. The samples were embedded, and the resin was cured at 60°C for 24 h as previously described (7, 18).

Energy-dispersive X-ray spectroscopic (EDXS) analysis was performed with a JEOL 200CX electron microscope at 200 keV using a high-angle energy-dispersive X-ray detector with a resolution of 165 eV for Mn K-alpha radiation for elements with atomic numbers greater than 11 (23). Unstained thin (60-μm) sections were placed on acid-cleaned, uncoated 300-mesh gold grids. The data was analyzed using EM:SPEC Systems software (version 3.1).

RESULTS

Construction of phsABC genetic cassettes.

A 5.1-kb EcoRI-SalI-digested DNA fragment from S. enterica serovar Typhimurium containing the native phsABC fragment encompassing the structural and putative regulatory region was inserted into the pTrc99A and pJB866 expression vectors, resulting in plasmids pSB74 and pSB77, respectively (Fig. 1; Table 1). The DNA fragment also contains a truncated ORF of unknown function downstream of phsABC.

FIG. 1.

Construction and genetic structure of the phsABC cassettes. (A) The 5,136-bp EcoRI-SalI DNA fragment containing the native phs regulatory and structural genes from S. enterica serovar Typhimurium LT2(pEB40). (B) Structure of the 3,653-bp PCR amplified DNA fragment containing only structural phsABC cloned into pTrc99A, creating pSB103 (diagram 1), and into pJB866, creating pSB107 (diagram 2). (C) Structure of the native phsABC operon cloned into pTrc99A, creating pSB74 (diagram 1), and into pJB866, creating pSB77 (diagram 2). The numbers represent the locations of nucleotide bases. An asterisk indicates putative sequences. Abbreviations: phsABC, thiosulfate reductase gene; CAP, catabolite activator protein-binding site; RBS, ribosome-binding site; P_trc, hybrid trp-lac promoter; lac1^q, gene encoding the lac repressor protein; rrnBT1T2, transcription terminators; P_m, m-cleavage pathway promoter of the TOL plasmid; xylS839, gene encoding the repressor protein that controls P_m.

A 3.7-kb PCR-amplified phsABC fragment, designed to eliminate the putative promoter, the catabolite activator protein-binding site, and the ORF of unknown function downstream of phsABC, was inserted into the pTrc99A and pJB866 expression vectors, resulting in plasmids pSB103 and pSB107, respectively.

Thiosulfate reductase activity of cell extracts.

E. coli DH5α strains harboring the native phsABC constructs had higher thiosulfate reductase activities under both aerobic and anaerobic conditions than E. coli DH5α strains harboring the engineered phsABC constructs (Table 2). In general, the high-copy-number plasmid constructs had more activity than their medium-copy-number counterparts, and cells grown anaerobically had higher activity than cells grown aerobically.

TABLE 2.

Thiosulfate reductase activity by cell extracts of E. coli DH5α strains harboring phsABC constructs

Plasmid	Aerobic culture		Anaerobic culture
	Enzyme activitya (U/g [dry wt])	Relative activityb	Enzyme activity (U/g [dry wt])	Relative activity
Controld	138 ± 26	1	218 ± 22	1.6
pSB74	324 ± 49	2.3	684 ± 40	5.0
pSB77	209 ± 31	1.5	627 ± 27	4.5
pSB103	191 ± 27	1.4	529 ± 27	3.8
pSB107	187 ± 22	1.4	498 ± 31	3.6

^aOne unit is equivalent to 1 μmol of sulfite produced/min in a 1.0-ml assay system. Mean values from three replicate experiments are reported. 

^bThe relative activity is the ratio of the thiosulfate reductase activity in the cell extracts from E. coli strains harboring phsABC constructs grown aerobically or anaerobically to that in the control cells grown aerobically. 

Hydrogen sulfide production from thiosulfate.

The native and engineered phsABC genetic cassettes (pSB74, pSB77, pSB103, and pSB107) were expressed in E. coli DH5α in the presence of thiosulfate. All four cultures turned black (due to FeS precipitation) when grown in the sulfide detection medium supplemented with 2.5 mM FeCl₂·4H₂O and 3 mM Na₂S₂O₃·5H₂O (data not shown). In contrast, E. coli cells harboring the pTrc99A vector alone did not turn black in the medium. This result was an indication that E. coli DH5α expressing the phsABC cassettes produced a functional thiosulfate reductase and generated sulfide from inorganic thiosulfate.

A more quantitative measure of sulfide production was performed using a sulfide electrode. The cultures were grown at 37°C without agitation. The resazurin indicator revealed that the cultures remained aerobic for the first hour and then became microaerobic and anaerobic thereafter. E. coli DH5α harboring pSB74 (native phsABC operon inserted into pTrc99A) produced the most sulfide of all of the strains tested: 173 μM in 1 h, 377 μM in 5 h, and 389 μM in 24 h (Fig. 2). The second highest sulfide production was observed with the cells harboring pSB77 (native phsABC operon in pJB866): 156 μM in 1 h and 257 μM in 24 h. E. coli DH5α harboring pSB103 and pSB107 (modified phsABC operon on pTrc99A and pJB866) generated 210 and 152 μM sulfide in 24 h, respectively. Sulfide production by E. coli DH5α harboring pTrc99A remained below the limit of detection with the sulfide electrode (approximately 50 μM).

FIG. 2.

Sulfide production by E. coli DH5α expressing the phsABC genetic cassettes. The sulfide concentration was determined using a sulfide electrode. The reported values are the means of triplicate measurements; the standard errors were less than ±6%.

Cadmium removal by engineered E. coli.

Cadmium removal by E. coli cells expressing phsABC was investigated in the modified MOPS medium in the presence of 3 mM Na₂S₂O₃·5H₂O and various concentrations of CdCl₂. To reduce individual sample errors, all culture and sample preparations were performed simultaneously. Triplicate samples were collected and prepared for cadmium concentration determination by ICP-OES. All of the data reported are mean values. The standard deviation in all cases was 2% or less.

When 50 μM CdCl₂ was present in the culture medium, all four strains expressing the phsABC operon removed nearly all of the cadmium within 24 h (Fig. 3). A bright yellow precipitate developed in all four cultures, an indication of CdS precipitation. In contrast, the negative control (E. coli DH5α harboring pTrc99A) removed less than one-quarter of the total cadmium and did not turn yellow. In general, the percentage of cadmium removed from solution decreased as the cadmium concentration in the medium increased. E. coli DH5α(pSB74) outperformed all of the other constructs; it removed nearly all of the cadmium at 100 and 150 μM CdCl₂ and most of it at 200 μM. At the high cadmium concentrations of 300 and 400 μM, it removed 46 and 25%, respectively (data not shown). The cells harboring pSB77 removed nearly all of the cadmium at concentrations of up to 100 μM and slightly less than E. coli DH5α(pSB74) at all other concentrations. All of the other constructs removed significantly less cadmium than did E. coli DH5α(pSB74) at cadmium concentrations of 100 μM and higher. There was little difference in the growth of the strains at any particular cadmium concentration.

FIG. 3.

Cell growth and percent cadmium removal by E. coli DH5α expressing the phsABC genetic cassettes. (A) Cell growth in the presence of various concentrations of CdCl₂. OD600, optical density at 600 nm. (B) Cadmium removed by the cells after 24 h. The reported values are the means of triplicate measurements; the standard errors were less than ±2%. Symbols: solid bars, pSB74; diagonal bars, pSB77; grid bars, pSB103; horizontal bars, pSB107; empty bars, pTrc99A.

EDXS analysis.

Electron microscopy showed extensive precipitation in the culture medium. EDXS was used to determine the nature of the precipitate (Fig. 4). The predominant elements in these precipitates were cadmium and sulfur (osmium and gold result from the fixation process and the grid). Multiple spectra were acquired along the particle axis, and the areas for the peaks corresponding to cadmium and sulfur were integrated. The relative amounts of cadmium and sulfur are constant along the trajectory of the electron beam through the particle.

FIG. 4.

EDXS analysis of metal sulfide complexes. (a) Electron micrograph of cells and cadmium sulfide granules in a sample. The line across the dark granule indicates the 30-nm path of the electron beam. The position in the electron micrograph where the spectrum in panel b was taken is indicated by the dot on the line. Three windows, one corresponding to the energy associated with Cd, one corresponding to S, and the other containing no peaks (to be used as background), are shown on the spectra. (c) Integrated areas under the peaks in the Cd (squares), S (circles), and background (triangles) windows are plotted as a function of position along the line in panel a. Each point along the line was scanned for 30 s. The solid vertical line on the graph corresponds to the position of the dot on the line in panel a.

DISCUSSION

There are many approaches for the use of bacteria to remove heavy metals from the environment: bioaccumulation and biosorption, oxidation and reduction, methylation and demethylation, and ligand degradation by bacteria (5). A common choice for bioremediation of heavy metals that readily precipitate as sulfides has been the use of SRB to generate hydrogen sulfide from sulfate and to precipitate metals as insoluble metal sulfides (10, 12, 24).

In this study, we used recombinant DNA technology to engineer the thiosulfate reductase operon (phsABC) from S. enterica serovar Typhimurium to overproduce hydrogen sulfide from inorganic thiosulfate and precipitate cadmium as cadmium sulfide. This sulfide-producing genetic system could be transferred to certain bacteria (such as organic pollutant degraders that are sensitive to toxic metals), enabling them to tolerate or to remove heavy metals, in addition to mineralizing organic pollutants as a carbon source.

Thiosulfate reductase enzyme assays revealed that E. coli cells harboring phsABC showed comparatively higher activity than a control under both aerobic and anaerobic conditions, suggesting that thiosulfate reductase encoded by phsABC was functional. The cells harboring the native phsABC constructs (retaining the phsABC putative promoter and regulatory regions) showed more activity than those harboring the engineered phsABC constructs (retaining only the structural genes of phsABC), which may be due to the involvement of the putative promoter and associated regulatory regions. In general, cells harboring the phsABC construct on high-copy-number plasmids had more enzyme activity than the cells harboring the same construct on medium-copy-number plasmids, indicating the copy number effect on enzyme production. Overall, E. coli cells grown under anaerobic conditions showed higher activity than those grown under aerobic conditions, suggesting additional regulation at the level of gene expression or enzyme activity in response to anaerobic conditions. The enzyme activity of the extracts of E. coli harboring the phsABC constructs correlates well with the results from the sulfide production and cadmium removal experiments. Although the thiosulfate reductase assay revealed that the extracts of the cells grown aerobically showed substantial enzyme activity, it was not practical to use the cells grown aerobically for cadmium removal due to rapid sulfide oxidation and loss as hydrogen sulfide gas under such conditions.

Our results in the sulfide production experiments demonstrated that all four phsABC constructs encoded functional thiosulfate reductase that generated sulfide from thiosulfate. There was a definitive correlation between the amount of sulfide produced and the amount of cadmium removed by cells expressing the phsABC genetic cassettes. Cells that generated more sulfide from thiosulfate could efficiently remove more cadmium from the medium. Among the four phsABC cassettes, the cells expressing the entire operon under the control of P_trc on a high-copy plasmid (pSB74) produced the most sulfide and removed the most cadmium.

While the cell numbers of all of the strains tested remained in a close range within 24 h, including the negative control, sulfide production and cadmium removal were significantly greater in the cells expressing phsABC than in the control. This result demonstrated that the cadmium removal was due to expression of phsABC, not to biosorption to the cells. In addition, electron microscopy and EDXS analysis revealed extensive precipitates in the surrounding medium containing both cadmium and sulfur. While the EDXS analysis does not definitively prove that the precipitates are cadmium sulfide, these results, along with the color of the precipitate and the measurement of sulfide production in the absence of cadmium, are highly suggestive of cadmium sulfide.

Our results showed that the rate of sulfide production and cadmium removal varied depending on the expression systems and which phsABC fragments were expressed. The primary reason for constructing pSB103 and pSB107 was to eliminate any regulatory region in the phsABC operon that might be involved in its control. However, it appears that the cells harboring pSB103 and pSB107 indeed produced less sulfide than those expressing pSB74 and pSB77, which retained the entire native phsABC operon. It is not clear whether the undefined upstream region of phsABC or the downstream ORF plays an important role in the expression of the phsABC genes or enzyme activity. It has been suggested that the truncated ORF downstream of phsABC is not required for phsABC expression and that anaerobic conditions and/or thiosulfate are required for full expression (14). E. coli cells harboring the pTrc99A derivatives pSB74 and pSB103 showed higher sulfide production and cadmium removal than those harboring the pJB866 derivatives pSB77 and pSB107. This may be due to the strong expression system on the pTrc99A vector (P_trc and transcriptional terminators of rrnBT1T2) or the high copy number of the vector. In contrast, pJB866 is a medium-copy-number vector and carries the relatively weaker P_m promoter. While sulfide was produced and cadmium was removed by all of the constructs, the best results were achieved using the entire native phsABC region carried on a high-copy-number expression vector.

In summary, we demonstrated the use of the thiosulfate reductase gene (phsABC) from S. enterica serovar Typhimurium to overproduce hydrogen sulfide and remove cadmium from solution. The four phsABC cassettes constructed in this study encoded functional thiosulfate reductases that produced hydrogen sulfide from inorganic thiosulfate. All four constructs showed cadmium removal from the medium, demonstrating the potential use of phsABC genetic cassettes for bioremediation of heavy metals in waste streams.