Sensing Antimonite and Arsenite at the ... - ACS Publications

Anal. Chem. 1997, 69, 3380-3384

Sensing Antimonite and Arsenite at the Subattomole Level with Genetically Engineered Bioluminescent Bacteria Sridhar Ramanathan, Weiping Shi,† Barry P. Rosen,† and Sylvia Daunert*

Department of Chemistry, University of Kentucky, Lexington, Kentucky 40506

A highly sensitive and selective optical sensing system for antimonite has been developed using genetically engineered bacteria. The basis of this system is the ability of certain bacteria to survive in environments that are contaminated with antimonite, arsenite, and arsenate. The survival is conferred to the bacteria by the ars operon, which consists of five genes that code for three structural proteins, ArsA, ArsB, and ArsC, and two regulatory proteins, ArsD and ArsR. ArsA, ArsB, and ArsC form a protein pump system that extrudes antimonite, arsenite, and arsenate once these anions reach the cytoplasm of the bacterium. A method was developed for monitoring antimonite and arsenite by using a single plasmid that incorporates the regulatory gene of the extrusion system, arsR, and the genes of bacterial luciferase, luxA and luxB. In the designed plasmid, ArsR regulates the expression of bacterial luciferase in a manner that is dependent on the concentration of antimonite and arsenite in the sample. Thus, the bioluminescence emitted by luciferase can be related to the concentration of antimonite and arsenite in the sample. Concentrations for antimonite and arsenite in the order of 10-15 M, which corresponds to subattomole levels, can be detected. This bacterialbased sensing system is highly selective for antimonite and arsenite. It is well established that certain microorganisms can survive under adverse environmental conditions such as extreme temperatures, pressure, and high salt environments.1-3 Additionally, some strains of bacteria have the necessary genetic composition to survive in highly toxic environments. These bacteria can sense the presence of toxic compounds and produce proteins that can either convert the compounds to nontoxic products or selectively extrude them out of the cytoplasm. Bacterial strains that can survive in environments contaminated with mercury,4-6 naphthalene,7 arsenite/antimonite,8-10 and other heavy metals11 are among the bacterial systems that have been identified and characterized.

Sensors that use bacteria have been developed for mercury12,13 and naphthalene.14-17 In this article, we will discuss the development of an optical sensing system for antimonite and arsenite based on genetically designed bacteria. Resistance to arsenate, arsenite, and antimonite is conferred on bacteria by the ars operon. The ars operon codes for three structural proteins, ArsA, ArsB, and ArsC, and two regulatory proteins, ArsR and ArsD. The three structural proteins form a protein pump in which each of the proteins has a specific function. ArsB is a transmembrane protein18 and, along with ArsA,19 forms a channel for selectively extruding antimonite and arsenite out of the cell. The protein ArsC is an arsenate reductase, the function of which is to reduce arsenate to arsenite. Antimonite and arsenite are effluxed out of the cytoplasm by the ArsA-ArsB protein complex.20 The regulation of this protein pump is controlled by the ArsD and ArsR proteins as follows. In the absence of the inducer (arsenite or antimonite), the ArsR protein is expressed at basal levels,21 and it physically binds to the operator/promoter (O/P) region of the ars operon. This prevents the transcription and translation of the complete ars operon to occur. When arsenite or antimonite enters the cell, it binds to the binding site on the ArsR protein, which is highly specific for these ions.22,23 Then, the ArsR protein dissociates from the O/P region, and, as a consequence, the expression of the proteins encoded by the ars operon can commence. When the concentration of proteins expressed by the ars operon reaches a certain maximal limit, the ArsD protein inhibits the expression of the proteins coded by the

† Department of Biochemistry and Molecular Biology, Wayne State University, School of Medicine, Detroit, MI 48201. * Address correspondence to this author. Phone: (606) 257-7060. Fax: (606) 323-1069. E-mail: [email protected]. (1) Lowe, S. E.; Jain, M. K.; Zeikus, J. G. Microbiol. Rev. 1993, 57, 451-509. (2) Kato, C.; Inoue, A.; Horikoshi, K. Trends Biotechnol. 1996, 14, 6-12. (3) Scha¨fer, G.; Purschke, W. G.; Gleissner, M.; Schmidt, C. L. Biochim. Biophys. Acta 1996, 1275, 16-20. (4) Mergeay, M. Trends Biotechnol. 1991, 9, 17-24. (5) Silver, S.; Mishra, T. K. Annu. Rev. Microbiol. 1988, 42, 717-743. (6) Summers, A. O. Annu. Rev. Microbiol. 1986, 40, 607-634. (7) Yen, K.; Serdar, C. M. CRC Crit. Rev. Microbiol. 1988, 15, 247-268.

(8) Silver, S.; Ji, G.; Bro ¨er, S.; Dey, S.; Dou, D.; Rosen, B. P. Mol. Microbiol. 1993, 8, 637-642. (9) Kaur, P.; Rosen, B. P. Plasmid 1992, 27, 29-40. (10) Diorio, C.; Cai, J.; Marmer, J.; Shinder, R.; Dubow, M. S. J. Bacteriol. 1995, 177, 2050-2056. (11) Silver, S.; Walderhaug, M. Microbiol. Rev. 1992, 56, 195-228. (12) Selifonova, O.; Burlage, R.; Barkay, T. Appl. Environ. Microbiol. 1993, 59, 3083-3090. (13) Virta, M.; Lampinen, J.; Karp, M. Anal. Chem. 1995, 67, 667-669. (14) Burlage, R. S.; Sayler, G. S.; Larimer, F. J. Bacteriol. 1990, 172, 47494757. (15) King, J. M. H.; DiGrazia, P. M.; Applegate, B.; Burlage, R.; Sanseverino, J.; Dunbar, P.; Larimer, F.; Sayler, G. S. Science 1990, 249, 778-781. (16) Heitzer, A.; Webb, O. F.; Thonnard, J. E.; Sayler, G. S. Appl. Environ. Microbiol. 1992, 58, 1839-1846. (17) Heitzer, A.; Malachowsky, K.; Thonnard, J. E.; Bienkowski, P. R.; White, D. C.; Sayler, G. S. Appl. Environ. Microbiol. 1994, 60, 1487-1494. (18) Dey, S.; Rosen, B. P. J. Bacteriol. 1995, 177, 385-389. (19) Dey, S.; Dou, D.; Tisa, L. S.; Rosen, B. P. Arch. Biochem. Biophys. 1994, 311, 418-424. (20) Ji, G.; Garber, A. E.; Armes, L. G.; Chen, C.; Fuchs, J. A.; Silver, S. Biochemistry 1994, 33, 7294-7299. (21) Wu, J.; Rosen, B. P. Mol. Microbiol. 1991, 5, 1331-1336. (22) Shi, W.; Wu, J.; Rosen, B. P. J. Biol. Chem. 1994, 269, 19826-19829. (23) Wu, J.; Rosen, B. P. J. Biol. Chem. 1993, 256, 52-58.

3380 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

S0003-2700(97)00111-X CCC: $14.00

© 1997 American Chemical Society

ars operon.24 The exact mechanism of ArsD regulation is still being investigated. Thus, the ArsR protein regulates the basal expression, and the ArsD protein regulates the maximal expression of the ars operon. Given the specificity of the ars operon for antimonite and arsenite, a highly sensitive and selective sensing system for these ions could emerge if part or the entire ars operon is coupled to a sensitive detection method like bioluminescence. One such bioluminescent protein is bacterial luciferase, an enzyme that catalyzes the oxidation of FMNH2 and a long-chain aldehyde to FMN and the corresponding carboxylic acid; this process is accompanied by the emission of light at 490 nm.25-27 This enzyme has been used in a number of bioanalytical methods, including the monitoring of protein expression.25 Bacterial luciferase is a heterodimeric enzyme consisting of two subunits, LuxA and LuxB, that are coded for by the luxA and luxB genes of the lux operon. The long-chain aldehyde is synthesized in vivo through proteins encoded by the luxC, luxD, and luxE genes of the lux operon. If the aldehyde substrate (typically decanal) is introduced externally, then there is no need for the presence of the entire lux operon in order to produce bioluminescence in the bacteria. Indeed, the luxA and luxB genes are sufficient for the expression of the enzyme luciferase and, therefore, for the generation of the bioluminescence reaction. In this work, we describe the design of bacteria that contain a plasmid in which the genes that code for bacterial luciferase are fused to the arsD gene, which is downstream of the arsR gene. In this manner, the arsR gene controls the expression of bacterial luciferase. Another plasmid construct similar in design to our plasmid has been reported, where the arsB gene of Staphylococcus aureus has been fused to the lux genes. Although this plasmid was employed in studies that involve the elucidation of the regulation of the ars operon,28 it was not used in the development of sensing systems for either antimonite or arsenite. In the bacterial-based system described here, the arsR gene of pRLUX acts as the sensing element, and luciferase acts as the reporter/ transducer of a conventional sensor. When the bacteria are exposed to varying concentrations of antimonite or arsenite in a sample, varying amounts of luciferase are expressed. The activity of luciferase can be monitored by adding the substrate, decanal, and measuring the light emitted at 490 nm. The antimonite/ arsenite levels present in the sample can be determined from the amount of light emitted by the bacteria. EXPERIMENTAL SECTION Reagents. The restriction enzymes HindIII and PvuII were purchased from New England Biolabs (Beverly, MA). Potassium antimonyltartrate was purchased from Aldrich (Milwaukee, WI). n-Decyl aldehyde (decanal), bovine serum albumin (BSA), tris(hydroxymethyl)aminomethane (Tris), sodium salt of EDTA, sodium arsenite, dithiothreitol (DTT), sodium phosphate (monobasic and dibasic), and all other reagents were obtained from Sigma (St. Louis, MO). Luria Bertani (LB) broth was from BIO 101 (Vista, CA). All chemicals were reagent grade or better and were used as received. All solutions were prepared using deionized (24) Wu, J.; Rosen, B. P. Mol. Microbiol. 1993, 8, 615-623. (25) Campbell, K. Chemiluminescence; Ellis Horwood: Chichester, England, 1988. (26) Meighan, E. A. FASEB J. 1993, 7, 1016-1022. (27) Meighan, E. A. Microbiol. Rev. 1991, 55, 123-142. (28) Corbisier, P.; Ji, G.; Nuyts, G.; Mergeay, M.; Silver, S. FEMS Microbiol. Lett. 1993, 110, 231-238.

Figure 1. Schematic of the pRLUX plasmid.

(Milli-Q water purification system, Millipore, Bedford, MA) distilled water. Apparatus. Bioluminescence measurements were made on an Optocomp I luminometer from GEM Biomedical (Carrboro, NC) using a 100-µL fixed-volume injector. All experiments were conducted at room temperature unless specified otherwise. All luminescence intensities reported are the averages of a minimum of three replicates. Preparation of pRLUX plasmid. The pRLUX vector (Figure 1) was prepared by inserting a HindIII-PvuII fragment of pQF70 containing the luxA and luxB genes (for simplicity abbreviated as luxAB) between the HindIII and PvuII sites of plasmid pWSU1. The resulting plasmid retains the O/P region, arsR, and part of the arsD gene of the ars operon. The luxA gene is fused in frame to the arsD fragment, resulting in a fusion protein in which subunit A of luciferase is fused to ArsD. Detailed descriptions of pWSU1, from which the arsR and arsD were isolated, and pQF70, from which luxA and luxB were isolated, are found in refs 29 and 30, respectively. The construction of the pRLUX plasmid was verified by digesting the plasmid with HindIII and PvuII, and by confirming the lengths of the digested fragments on a 1% agarose gel. The pRLUX vector, which also contains the gene for ampicillin resistance, was transformed into Escherichia coli (strain JM109) using conventional protocols.31 Bioluminescence Emission. A single colony of the transformed bacteria was selected and grown overnight at 37 °C in 5 mL of LB broth with an ampicillin concentration of 50 µg/mL to a cell density that corresponds to an absorbance of 0.9 at 600 nm. A volume of 100 µL of the bacterial suspension was centrifuged at 2350g for 10 min at 4 °C, and the supernatant was discarded. The bacterial pellet was resuspended in 200 µL of a 10 mM TrisHCl, 1 mM EDTA, and 1 mM DTT, pH 8.0, buffer (Tris-EDTA). A volume of 50 µL of this bacterial solution was mixed with 250 (29) San Fransisco, M. J. D.; Hope, C. L.; Owolabi, J. B.; Tisa, L. S.; Rosen, B. P. Nucleic Acids Res. 1990, 18, 619-624. (30) Farinha, M. A.; Kropinski, A. M. J. Bacteriol. 1990, 172, 3496-3499. (31) Maniatis, T.; Fritsch, D. F.; Sambrook, J. Molecular Cloning: A Laboratory Manual; Cold Spring Harbor Laboratory: Cold Spring Harbor, NY, 1989.

Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

3381

µL of Tris-EDTA buffer and was placed in the luminometer. Then, the bioluminescence reaction was triggered by the injection of 100 µL of 63 µM decanal, and the bioluminescence signal generated was measured for a period of 1 min at 1-s intervals. Dose-Response Curves for Antimonite and Arsenite. A single colony of the transformed bacteria was selected and grown overnight at 37 °C in LB media containing 50 µg/mL ampicillin until the bacteria grew to a cell density corresponding to an absorbance of 0.9 at 600 nm. Then, 100 µL of the bacterial suspension was incubated with 100 µL of the inducer (antimonite or arsenite) at 37 °C. The blank corresponds to 100 µL of deionized distilled water incubated with 100 µL of the bacterial suspension. Antimonite and arsenite solutions were prepared fresh daily by serial dilution from a freshly prepared 1 × 10-3 M stock solution. To induce protein expression, the bacteria were incubated with antimonite for 30 min and with arsenite for 3 h. After the induction time, the bacterial solutions were centrifuged at 2350g for 10 min at 4 °C. The supernatant was discarded, and the bacteria were resuspended in 600 µL of Tris-EDTA buffer. All bacterial solutions were kept in an ice bath until their bioluminescence was measured. This was accomplished by mixing 50 µL of the bacterial solution with 250 µL of Tris-EDTA buffer, followed by injection of 100 µL of 63 µM decanal. After a delay time of 10 s necessary to reach maximal bioluminescence, the light emitted was collected for a period of 3 s. The same protocol with an induction time corresponding to 30 min was used to evaluate the selectivity of the bacterial system toward phosphate, sulfate, nitrate, arsenate, Cd2+, Bi3+, and Co2+. RESULTS AND DISCUSSION To couple the selectivity conferred to bacteria by the ars operon with the sensitivity associated with bioluminescence detection, a plasmid was designed that incorporates the genes of bacterial luciferase along with the arsR gene. This plasmid contains the O/P region and the gene for the ArsR protein from the ars operon upstream from the genes that code for luciferase, luxA and luxB (Figure 1). In the constructed plasmid pRLUX, the luxA gene is fused in frame to arsD in such a way that the resulting fusion protein contains part of ArsD fused to the N-terminal domain of subunit A of luciferase. It has been shown previously that fusion proteins with truncated ArsD do not retain the regulatory function of the full ArsD.32 Likewise, it has been demonstrated previously that fusions at the N-terminal domain of luciferase result in the formation of active enzyme.15,28,33,34 As will be demonstrated in this article, the fusion of truncated arsD with luxA yields a fusion protein that maintains luciferase activity, and its expression from the pRLUX plasmid is regulated by the ArsR protein. The pRLUX plasmid was introduced into E. coli to yield bacteria that bioluminesce in the presence of antimonite and arsenite. In the absence of antimonite/arsenite, the ArsR protein binds to the O/P site of the plasmid, preventing the transcription of the genes for ArsR and luciferase. When the inducer (antimonite or arsenite) is introduced, it binds to ArsR, causing a conformational change in the ArsR protein that results in its release from the O/P site of the plasmid. Consequently, protein (32) Shi, W.; Dong, J.; Scott, R. A.; Ksenzenko, M. Y.; Rosen, B. P. J. Biol. Chem. 1996, 271, 9291-9297. (33) Pazzagli, M.; Devine, J. H.; Peterson, D. O.; Baldwin, T. O. Anal. Biochem. 1992, 204, 315-323. (34) Condee, C. W.; Summers, A. O. J. Bacteriol. 1992, 174, 8094-8101.

3382 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

Figure 2. Schematic representation of the interactions between the operator/promoter (O/P) of the ars operon in plasmid pRLUX and ArsR. (a) In the absence of antimonite (or arsenite), no luciferase is being produced. (b) The presence of antimonite (or arsenite) results in the subsequent expression of luciferase.

Figure 3. Bioluminescence emission of bacteria with pRLUX plasmid. A volume of 100 µL of 63 µM decanal was injected into a solution containing 50 µL of bacteria in 250 µL of Tris-EDTA buffer.

expression of the bacterial luciferase can now commence (Figure 2). The reaction catalyzed by native bacterial luciferase employs three substrates, namely decanal, FMNH2, and O2. Oxygen is available in excess because the reaction is carried out in air, while FMNH2 is endogenous to the bacteria. Although decanal can be biosynthesized from the luxC, luxD, and luxE genes, in pRLUX these genes were omitted to allow us control of the triggering of bioluminescence emission. Figure 3 shows a typical bioluminescence emission profile measured for a period of 60 s after the injection of decanal. This profile demonstrates a glow-type bioluminescence, which is characteristic of native luciferase systems.25 From this figure and similarly obtained bioluminescence emission profiles at other decanal concentrations, it was determined that a period of 10 s is required to reach the maximum

Figure 4. Dose-response curve for antimonite performed after the bacteria with the pRLUX plasmid were incubated with potassium antimonyltartrate standard solutions for 30 min. A volume of 100 µL of 63 µM decanal was injected into 50 µL of bacteria in 250 µL of Tris-EDTA buffer. The bioluminescence signal was integrated over a period of 3 s and has been corrected with respect to the blank. Data are the average ( 1 standard deviation (n ) 3).

emission of light. After this period, the light intensity decays gradually. Therefore, for the subsequent experiments, the intensity of the emitted light was integrated over 3 s, after a 10-s delay from the time of injection of decanal. The induction of luciferase expression is time dependent. As the time of induction increases, there is an increase in the bioluminescence emitted by the bacteria containing the pRLUX plasmid. A 30-min time of induction was chosen because it yields a sufficiently high bioluminescence signal while maintaining a relatively short assay time. In the dose-response curve shown in Figure 4, the intensity of the bioluminescence signal emitted increases with the concentration of antimonite in the sample. The relative standard deviation of the signals was less than 10% in all cases. This dose-response curve indicates induction of luciferase, which is controlled by the ArsR protein. In that respect, the ArsR protein acts like a “switch” that turns on the expression of luciferase when antimonite is present in the sample. Based on this dose-response curve, antimonite can be detected at extracellular concentrations as low as 10-15 M, which corresponds to subattomole levels in the sample (the detection limit is defined as the antimonite concentration that corresponds to S/N ) 2). It should be noted that the concentration of antimonite inside the cell (which is what is sensed by ArsR) is not known, although it is related to the extracellular concentration. A step increase in the light intensity was observed (Figure 4), where the intensity increases to a certain level and is then constant over 3 orders of magnitude of antimonite concentration. After this, the intensity increases again with antimonite concentration, reaching a maximum at 10-8 M. The step increase in the doseresponse curve may be caused by the chromosomally encoded ars operon that has been recently discovered in E. coli.35-37 This version of the ars operon codes for three proteins, ArsR, ArsB, and ArsC. Though the chromosomal ArsR, ArsB, and ArsC proteins do not have the same amino acid sequence as those encoded by the plasmid ars operon, they are homologues of the (35) Carlin, A.; Shi, W.; Dey, S.; Rosen, B. P. J. Bacteriol. 1995, 177, 981-986. (36) Sofia, H. J.; Burland, V.; Daniels, D. L., Plunkett, G., III; Blattner, F. R. Nucleic Acids Res. 1994, 22, 2576-2586. (37) Xu, C.; Shi, W.; Rosen, B. P. J. Biol. Chem. 1996, 271, 2427-2432.

plasmid-encoded genes. It has been shown that, in the absence of the plasmid-coded ars operon in E. coli, the chromosomal operon is responsible for a basal level of antimonite/arsenite resistance.35 Thus, in E. coli that contains the pRLUX plasmid, there are two sources for the ArsR protein, a chromosomal source and a plasmid source. Another factor that affects the doseresponse curve in Figure 4 is the fact that plasmid DNA is present in multiple copies within E. coli, in contrast to the chromosomal ars operon, which is present as a single copy. The pRLUX plasmid is designed so that the ArsR protein regulates the expression of bacterial luciferase. As mentioned earlier, in pRLUX, all the genes of the ars operon that code for the protein pump have been deleted, and the luxAB genes have been inserted. Thus, the bacteria with the pRLUX plasmid can sense antimonite, but there is no formation of a plasmid-encoded pump. Since antimonite is toxic to the bacterial cell, when the accumulation of antimonite reaches certain levels, it kills the bacteria. This typically happens at concentrations of ∼10-4-10-3 M. The decrease in bioluminescence signal at 1.9 × 10-8 M of antimonite is, therefore, due not to cell death but rather to inactivation of luciferase by antimonite. This was verified by incubating antimonite with luciferase in vitro, followed by addition of the substrates decanal and FMNH2 (the latter was generated in situ using an oxidoreductase with NADH and FMN as substrates). It was found that, as the concentration of antimonite increased, the bioluminescence of luciferase decreased. A similar antimonite effect was observed using bacteria containing the pQF70 plasmid, in which the expression of luciferase is not regulated by ArsR. Using these bacteria, it was also found that, as the concentration of antimonite increases, a decrease in the emitted bioluminescence is observed. Therefore, in the concentration range of 10-8-10-10 M (Figure 4), there are two antimonitedependent competing processes: the induction of luciferase production the inactivation of luciferase. At this concentration range, the bioluminescence emitted masks the inactivation process. At concentrations higher than 1.9 × 10-8 M antimonite, the inactivation of luciferase by the oxoanion dominates, causing a significant decrease in the emission of bioluminescence, as shown in Figure 4. Although a luciferase inactivation-based system (e.g., bacteria containing the pQF70 plasmid) could be used to determine antimonite, it should be stressed that this inactivation is not specific to antimonite but rather could be caused by any toxic compound present in the sample. To the contrary, the sensing system based on bacteria containing the pRLUX plasmid is highly specific (see below), and this specificity is conferred by the ArsR regulatory protein. The response of the ArsR protein toward arsenite, the other oxoanion recognized by the ArsR protein, was also studied. Both arsenite and antimonite bind selectively to the ArsR protein. The arsenite dose-response curve is similar to the dose-response curve obtained for antimonite (Figure 5). The relative standard deviation of the signals was e10% for arsenite concentrations (the detection limit is defined as the arsenite concentration that corresponds to S/N ) 2). The two regions observed in the doseresponse curve for antimonite are also present in this curve. This indicates that both antimonite and arsenite behave similarly and can induce the same type of response from ArsR. From the ascending region of the curve, it can be determined that antimonite can be detected at concentrations as low as 10-15 M. Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

3383

Figure 5. Dose-response curve for arsenite performed after the bacteria with the pRLUX plasmid were incubated with sodium arsenite standard solutions for 3 h. A volume of 100 µL of 63 µM decanal was injected into 50 µL of bacteria in 250 µL of Tris-EDTA buffer. The bioluminescence signal was integrated over a period of 3 s and has been corrected with respect to the blank. Data are the average ( 1 standard deviation (n ) 3).

The ArsR protein has a binding site that is very specific toward antimonite and arsenite and can discriminate effectively against other species. The selectivity of the bacterial sensing system was tested with phosphate, sulfate, nitrate, and arsenate concentrations in the range of 10-16-10-4 M. Neither of these anions induced expression of luciferase above the levels induced by the blank. This implies that the selectivity of the bacterial sensing system (38) Chwastowska, J.; Zmijewska, W.; Sterlinska, E. J. Radioanal. Nucl. Chem. 1995, 196, 3-9. (39) Walcrez, M.; Bulska, E.; Hulanicki, A. Fresenius’ J. Anal. Chem. 1993, 346, 622-626. (40) Waller, P. A.; Pickering, W. F. Talanta 1995, 42, 197-204. (41) Cutter, L. S.; Cutter, G. A.; San Diego-McGlone, M. L. C. Anal. Chem. 1991, 63, 1138-1142.

3384 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

for antimonite over these anions is at least 1011-fold. The selectivity of the bioluminescence sensing system toward metal ions such as Bi3+, Co2+, and Cd2+ was also investigated. The response to these ions is not significantly different from that of the blank, suggesting that the ArsR protein binds very specifically to antimonite and arsenite. In summary, we have demonstrated that a sensing system for antimonite based on bacterial bioluminescence can be developed, where the expression of the bacterial luciferase is controlled by ArsR, the regulatory protein of the ars operon. By coupling the selectivity of the ArsR protein to a sensitive detection method like bioluminescence, a highly sensitive and selective system has been obtained. Indeed, concentrations of antimonite on the order of 10-15 M can be detected. This is at least 3 orders of magnitude better than detection limits reported for antimonite obtained by using conventional methods.38-41 Arsenite levels as low as 1 × 10-15 M can also be detected. In addition, we have demonstrated that this system is extremely selective for antimonite over a variety of metals and anions. Because of the improved detection limits and selectivity, this bioluminescence bacterial system could find applications in the sensing of antimonite/arsenite in a variety of environmental samples. ACKNOWLEDGMENT We thank the National Science Foundation (Grant CHE9502299 to S.D.), the U.S. Geological Survey (Grant KWRRI 9415 to S.D.), and the U.S. Public Health Service (Grant GM55425 to B.P.R.) for funding this work. The plasmid pQF70 was a gift from A. M. Kropinski.30 Received for review January 28, 1997. Accepted June 1, 1997.X AC970111P X

Abstract published in Advance ACS Abstracts, July 15, 1997.