Environ. Sci. Technol. 2010, 44, 1049–1055
Development of a Multistrain Bacterial Bioreporter Platform for the Monitoring of Hydrocarbon Contaminants in Marine Environments ROBIN TECON,† SIHAM BEGGAH,† KAMILA CZECHOWSKA,† VLADIMIR SENTCHILO,† PANAGIOTA-MYRSINI CHRONOPOULOU,‡ TERRY J. MCGENITY,‡ AND J A N R O E L O F V A N D E R M E E R * ,† Department of Fundamental Microbiology, University of Lausanne, Lausanne, Switzerland, and Department of Biological Sciences, University of Essex, Essex, United Kingdom
Received September 21, 2009. Revised manuscript received November 21, 2009. Accepted November 25, 2009.
Petroleum hydrocarbons are common contaminants in marine and freshwater aquatic habitats, often occurring as a result of oil spillage. Rapid and reliable on-site tools for measuring the bioavailable hydrocarbon fractions, i.e., those that are mostlikelytocausetoxiceffectsorareavailableforbiodegradation, would assist in assessing potential ecological damage and following the progress of cleanup operations. Here we examined the suitability of a set of different rapid bioassays (2-3 h) using bacteria expressing the LuxAB luciferase to measure the presence of short-chain linear alkanes, monoaromatic and polyaromatic compounds, biphenyls, and DNA-damaging agents in seawater after a laboratory-scale oil spill. Five independent spills of 20 mL of NSO-1 crude oil with 2 L of seawater (North Sea or Mediterranean Sea) were carried out in 5 L glass flasks for periods of up to 10 days. Bioassays readily detected ephemeral concentrations of short-chain alkanes and BTEX (i.e., benzene, toluene, ethylbenzene, and xylenes) in the seawater within minutes to hours after the spill, increasing to a maximum of up to 80 µM within 6-24 h, after which they decreased to low or undetectable levels. The strong decrease in short-chain alkanes and BTEX may have been due to their volatilization or biodegradation, which was supported by changes in the microbial community composition. Two- and three-ring PAHs appeared in the seawater phase after 24 h with a concentration up to 1 µM naphthalene equivalents and remained above 0.5 µM for the duration of the experiment. DNAdamage-sensitive bioreporters did not produce any signal with the oil-spilled aqueous-phase samples, whereas bioassays for (hydroxy)biphenyls showed occasional responses. Chemical analysis for alkanes and PAHs in contaminated seawater samples supported the bioassay data, but did not show the typical ephemeral peaks observed with the bioassays.
* Corresponding author phone: (+41) 21 692 5630; fax: (+41) 21 692 5605; e-mail:
[email protected]. † University of Lausanne. ‡ University of Essex. 10.1021/es902849w
2010 American Chemical Society
Published on Web 12/14/2009
We conclude that bacterium-based bioassays can be a suitable alternative for rapid on-site quantitative measurement of hydrocarbons in seawater.
Introduction In November 2007, at least 10 ships sank in the Black Sea as a result of storms, releasing thousands of tons of oil and sulfur. In particular, the oil tanker Volganeft-139 ran aground and released half of its 4800 ton load of heavy fuel oil into the sea (1). Large oil spills in the marine environment resulting from transportation that reach public attention still occur strikingly often (2). In addition, the marine environment is constantly exposed to smaller oil releases, and these represent a major fraction of the total quantity of petroleum hydrocarbons entering the marine environment (3). It has been estimated that several millions of tons of petroleum hydrocarbons contaminate marine environments annually, which makes crude oil one of the most abundant and widespread organic pollutants of the sea (3, 4). Effective management of cleanup operations after largemagnitude oil spills necessitates an accurate and rapid estimation of the magnitude of the pollution, of its effects on the ecosystem, and of (bio)remediation possibilities. Visible oil slicks and their devastating effects on sea mammals and birds have rightfully drawn much of the attention, but the area of potential contamination through dissolution of oil components into the water column may be much larger. The early dissolution of low molecular mass compounds from oil into the aqueous phase is significant. However, typically this phenomenon is not sampled, because response teams and scientists usually do not arrive on the scene of an accidental oil spill until several hours, or even days, after the initial release event (so-called ephemeral data (5)). However, since those compounds are the ones that have the highest aqueous solubility, they may reach high and immediately toxic concentrations (5). Also, after weathering of the oil surface layer by evaporation, dispersion, and dissolution and by sedimentation of heavier slicks or after physical removal of the major part of oil slicks, dissolved hydrocarbons in the water may still be toxic for organisms, ranging from picoplankton to fish (3). For these reasons, it would be helpful to have rapid field monitoring tools that assess compound bioavailability, which can be used in conjunction with precise and conventional chemical sampling and analytics. Whole-cell bacterial bioreporters, i.e., bacteria that have been genetically modified to produce an easily quantifiable signal in the presence of one or a group of target analytes (6), have often been proposed as a useful tool for the multitarget analysis of environmental contaminants (7-10). Bacterial bioreporters are interesting for a number of reasons. First, they are relatively easy to engineer, and a number of broad chemical detection specificities can be obtained. Admittedly, most bioreporters are not 100% selective, but react to a range of related chemical compounds with different relative responses (see, e.g., Table S1, Supporting Information). Second, bioassays with whole-cell bacterial bioreporters consist mostly of simple incubations of the cells with an aqueous or gaseous sample for a period of up to a few hours, after which the reporter signal is read out and compared to a calibration curve. Literature data indicate that the method detection limit (MDL) in most bacterial bioreporter assays is in the nanomolar to micromolar range, which is sufficiently sensitive for a first screening of samples for the presence of contaminants at levels that are likely to be toxic (6). Third, bacterial reporter-based bioassays may be particularly useful VOL. 44, NO. 3, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1049
TABLE 1. Bioreporter Assay Specificitiesa bacterial strain E. coli DH5R
reporter plasmid
pGEC74, pJAMA7 pPROBE-LuxAB-TbuT pHYBP109 pHYBP103M3 E. coli MG1655 pJAMA8-cda B. sartisoli RP007 pPROBE-phn-luxAB
assay selection abbreviation marker reference
chemicals detected alkanes (C6-C10) benzene, toluene, xylene, ethylbenzene 2-hydroxybiphenyl 2-hydroxybiphenyl, biphenyl DNA-damaging agents naphthalene, (di)methylnaphthalene, phenanthrene
OCT BTEX HBP HBP CDA NAH
Ap + Tc 17 Km this study Ap 25 Ap 16 Ap this study Km this study
a Antibiotics used for plasmid selection: Ap, ampicillin at 100 µg/mL; Tc, tetracycline at 10 µg/mL; Km, kanamycin at 50 µg/mL. For estimation of relative response factors, see Table S1 in the Supporting Information.
in determining compound bioavailability as opposed to total chemical concentration, which can help in interpreting immediate ecotoxicological risks of contaminants (11). Despite numerous proofs of principle for the functionality of bioassays with whole-cell bacterial bioreporters, there are still very few practical examples with more realistic sample sets and simultaneous application of multiple bacterial bioreporter specificities. Here we developed and tested a quantitative bacterial bioreporter assay platform for the screening of marine oil spills. The platform was based on four bacterial bioreporter strains that could specifically detect a number of watersoluble compounds from oil (e.g., short-chain linear alkanes, monoaromatics, and two- to three-ring aromatics) and one strain that was sensitive to DNA damage to test the potential presence of any water-soluble compound from oil that would exert such a toxicity response. The performance of the multistrain assay platform was examined in individual calibrations and against chemical analysis of the contaminated aqueous phase in a number of laboratory-sized oil spills with Norwegian Geochemical Standard North Sea Oil-1 (NSO-1) (12) in marine water, taken either from the North Sea or from the Mediterranean.
Experimental Section Chemicals. Octane (99% purity), naphthalene crystals (99% purity), toluene (99% purity), and 2-hydroxybiphenyl (99% purity) were purchased from Fluka (Buchs, Switzerland). North Sea standard crude oil (NSO-1) was obtained from the Norwegian Petroleum Directorate. Bacterial Bioreporter Assays. Strains and Culture Conditions. Escherichia coli strains were routinely grown on Luria-Bertani (LB) medium at 37 °C in the presence of appropriate antibiotics. Burkholderia sartisoli RP007 (pPROBE-phn-luxAB) was routinely grown at 30 °C on tryptone yeast (TY; contains 3 g/L yeast extract and 5 g/L Bacto tryptone) agar plates or TY liquid medium amended with 50 mM NaCl and 50 µg/mL kanamycin. For biosensor assays, E. coli and B. sartisoli strains were resuspended in MOPS medium (for the composition see the Supporting Information). The bacterial bioreporter strains are listed in Table 1. Details of their genetic construction and their selectivity of detection are described in the Supporting Information (Table S1). Bioluminescence Reporter Assays. Aliquots (0.5 mL) of frozen bioreporter cells (see the supplementary methods in the Supporting Information) were thawed in a water bath at 25 °C for 2 min and diluted in 4.5 mL of MOPS medium to obtain the final bioreporter cell suspension for the bioassay. The bacterial cell suspension (150 µL) was transferred to the individual 0.5 mL glass vials of a 96-well Multi-Tier glass plate (Wheaton Science Products), as outlined in Figure S1, Supporting Information. MOPS medium (75 µL) was added to each well, after which 75 µL of either standard solution or contaminated seawater sample, or of a dilution thereof, was added. Seawater samples had to be diluted at least 4-fold 1050
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 3, 2010
TABLE 2. Figures of Merit for the Bioreporter Assaysa assay OCT BTEX HBP NAH CDA
calibration compd
MDL (µM)
octane 0.039 ( 0.037 toluene 0.24 ( 0.22 2-hydroxybiphenyl 0.30 ( 0.27 naphthalene 0.17 ( 0.13 nalidixic acid 34 ( 6.9
n
r2
20 (4) 24 (5) 16 (3) 8 (2) 16 (3)
0.96 ( 0.02 0.97 ( 0.03 0.97 ( 0.04 0.98 ( 0.02 0.96 ( 0.01
a MDL ) method of detection limit as the calculated equivalent concentration of pure calibration compound giving rise to a bioluminescence value of the blank plus 3 times its standard deviation. n ) number of assays used for calculation with, within parentheses, the number of independent spills to which this assay was applied. r 2 ) average regression coefficient over all assays assuming linearity.
in the reporter assay to avoid inhibition of the cells by the salt. Standard solutions with pure target compounds were prepared in uncontaminated water of the same origin as used for the spill, to which 3 µL of a concentrated chemical stock solution was added. For simplicity, standard calibration of the response of each bioreporter strain was performed with a single pure chemical that was representative of the target components in oil (Table 2) or as reported in the literature (13). All chemicals were dissolved in dimethyl sulfoxide (DMSO), except nalidixic acid, which was dissolved in water. Pure compound concentrations in standard series typically ranged from 0 to 0.05, 0.1, and 0.25 µM for octane, from 0 to 0.25, 0.50, 1.0, and 3.0 µM for toluene, from 0 to 0.10, 0.20, and 0.40 µM for naphthalene, from 0 to 5, 10, and 20 µM for nalidixic acid, and from 0 to 0.05, 0.1, 0.25, 0.50, and 1.0 µM for 2-HBP. To the blank was added the same quantity of pure DMSO (for all organic contaminants) or water (for nalidixic acid). The microtiter plate was finally sealed with an adhesive plate seal (Thermo Scientific, Epsom, U.K.) and incubated at 30 °C in a Thermostar orbital shaker (BMG Labtechnologies, Germany) at 400 rpm for 2 h for E. coli- and 3 h for B. sartisoli RP007-based assays. Bioluminescence Measurements. After assay incubation the luciferase activity (bioluminescence) was measured by transferring 200 µL of each incubation into a clean well of a white 96-well plate (Cliniplate, Thermoscientific, Finland) and adding n-decanal substrate solution. For E. coli assays we used 20 µL per 0.2 mL assay of an 18 mM n-decanal solution (1:1, v/v, ethanol/water solution), whereas for RP007 this was 1 µL. After 3 min (for E. coli assays) or 5 min (for B. sartisoli assays) of incubation at room temperature, the bioluminescence was measured using a microtiter plate luminometer (Centro LB 960 luminometer, Berthold AG, Switzerland) with an integration time of 0.1 s. Spiking. To test for inhibition in the bioassays, all contaminated seawater samples were additionally analyzed by spiking 3 µL of inducer stock solution to the assay to achieve spiking concentrations of 100 nM for octane, 500
nM for toluene, 10 µM for nalidixic acid, 500 nM for 2-HBP, and 100 nM for NAH. If these concentrations resulted in a response of the cells outside the range of the calibration curve, the assay was repeated with half that amount of spiking concentration. Bioassay Data Analysis. Light emission values for the standard series were used to calculate a calibration curve for each of the reporter assay types. We used linear or binomial regression calculated in IGOR Pro (version 5, Wavemetrics, Inc., Oregon) to fit bioluminescence emission data as a function of the compound concentration. This equation was then subsequently used to calculate corresponding “equivalent target concentrations” from the bioluminescence values in the samples or dilutions thereof. For quantification, the standards were measured at the same time as the samples. The MDL was calculated as the equivalent target compound concentration using the derived standard fitting curve equation corresponding to the light emission value of the blank plus 3 times the standard deviation of the blank. Samples with light emissions below the MDL were considered as nonsignificantly different from the blank. To correct for possible sample inhibition, we used the light emission value in the spiked sample assay as explained in the Supporting Information. Artificial Oil Spill. In a 5 L glass flask, 2 L of seawater (coastal Mediterranean or North Sea water) were artificially contaminated with 1% (20 mL) crude oil (NSO-1). The flask was open at its top and gently rocked at 20 rpm on an automated platform at room temperature (Figure S1, Supporting Information). The rotation did not produce a mixing (emulsion) of organic and aqueous phases. A tap located at the bottom of the bottle permitted the sampling of seawater (Figure S1). Samples were taken 15 min and 2 and 6 h after the spill and subsequently after 1, 3, 7, or 10 days. Triplicate 4 mL glass vials were filled to the top with seawater sample and stored at 4 °C for a maximum of 1 day before bioreporter assay analysis. Additionally, triplicate 4 mL seawater samples were filled to the top in amber glass vials, acidified to pH < 2 with two drops of 1 N HCl to block any biological activity, and stored at 4 °C until chemical analysis of the hydrocarbon content. Storage for chemical analysis included a period of 24 h of transport on ice from the sampling station (Lausanne, Switzerland) to the place of analysis (Essex, U.K.). After each sampling, the volume of seawater sampled was replaced in the bottle by addition of uncontaminated seawater to keep the total volume constant. Hydrocarbon Analysis. Oil hydrocarbons were extracted from 4 mL seawater samples using solid-phase extraction (SPE) tubes as recommended by the manufacturer (SPE Supelclean Envi-18, Supelco Bellefonte, PA). Initially, SPE columns were conditioned by being passed through 2 mL of methanol and then 2 mL of acidified water (pH 2). Subsequently, the seawater samples were introduced into the SPE tubes, and extracts were eluted with 2 mL of hexane/ dichloromethane (1:1). A 700 µL volume of the eluted extract was used for hydrocarbon analysis. Conditions of GC-MS were as described by Coulon et al. (14). Statistical analysis of the GC-MS results was performed with Microsoft Office Excel 2003.
Results Design of a Multistrain Bacterial Bioreporter Platform Assay. To produce a streamlined multistrain bacterial bioreporter platform assay for detection of oil compounds, we decided to re-engineer a number of previously reported bioreporters into the bacterium E. coli as a living chassis for all genetic constructions. Bioluminescence was used as the reporter signal that the cells would produce in contact with the target chemical compounds. Table 1 specifies the final set of bioreporter strains and their detection specificities.
Only the NAH bioreporter strain could not be produced in E. coli but instead was engineered in the bacterium B. sartisoli RP007 (15). Reporter assays were generally calibrated with three standard concentrations and a blank (Figure S2, Supporting Information), each in triplicate and newly performed for every series of samples. As evident from Figure S2, standard curves varied somewhat depending on the exact amount and activity of cells incubated in the assay. Summarized MDLs and coefficients of variation (r 2, based on linearity assumptions) over all assays are presented in Table 2. Measurement variation of triplicates in the standard curve was more important in assays for those compounds which are rather volatile (e.g., octane and toluene), compared to more soluble and less volatile compounds (e.g., 2-HBP and nalidixic acid). Not all assays produced the same light output in relative units. For example, the BTEX, CDA, and NAH assays generally resulted in less light produced in the assay than OCT and HBP (Figure S2). One reason for this is different plasmid copy numbers with the reporter gene constructions, with CDA and BTEX having lower copy number plasmids than OCT and HBP. The other reason is the use of the B. sartisoli strain for the NAH assay, a strain which generally seemed to produce less light output than E. coli. This difference in the absolute level of luciferase activity did not influence the biosensor’s sensitivity (i.e., the slope of the induction curve). Analysis of Oil-Contaminated Seawater with the Bacterial Biosensor Platform Assays. To test the performance of the selected bioreporter strains for quantifying bioavailable dissolved oil components in seawater, we mimicked oil spills in the laboratory. Hereto, 20 mL of crude NSO-1 standard oil was poured on top of 2 L of uncontaminated seawater in an open 5 L glass flask that was gently rocked without forming emulsions (Figure S1A, Supporting Information). After different incubation periods during 1 week, we sampled the aqueous phase via a tap located at the bottom of the glass bottle to avoid including oil phase in the samples. The experiment was repeated five times sequentially and with seawater of different origins (e.g., North Sea and Mediterranean). Figures 1 and 2 show the ephemeral evolution of equivalent octane and toluene concentrations, respectively, in the contaminated seawater. It should be noted that these responses were likely elicited by mixtures of chemicals, since the bioreporter assays are not selective for a single target compound only (Table S1, Supporting Information). The OCT and BTEX assays produced the most consistent response over time for four (for OCT) and five (for BTEX) crude oil spill replicates in different seawater origins and on different occasions during the year, although the magnitude of the responses varied. Both OCT and BTEX assays showed a rapid increase of detectable compounds within the first 6 h after the spill (Figures 1 and 2), which then gradually decreased during the following 50 h to very low or nondetectable levels. The maximum equivalent octane concentration in the aqueous phase amounted to between 200 and 600 nM. The calculated equivalent octane concentrations were in general in good agreement between spiked and nonspiked samples, suggesting that in these assays the bacterial reporter cells were not critically inhibited by other chemicals present in the sample. On the other hand, for some samples the light emissions in the spiked and nonspiked assays were very close, which makes it impossible to calculate the inhibition factor. For these samples no spiking correction was made (indicated by asterisks in Figure 1), and therefore, the calculated equivalent octane concentration is likely an underestimation. The calculated toluene equivalent concentrations in seawater in the first 6 h after the crude oil spill were much higher than the octane equivalent concentrations (Figure 2). In four out of five assays we detected values between 4 and 12 µM toluene equivalents, which were even likely to be an VOL. 44, NO. 3, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1051
FIGURE 1. Time development of equivalent octane concentrations in the aqueous phase after the oil spill as outlined in Figure S1 in the Supporting Information on four different repetitions, here labeled with (A) Med1, (B) Med2, and (D) Med3 (seawater from the Mediterranean) and (C) N.Sea1 (seawater from the North Sea). Squares indicate the average equivalent octane concentration from triplicate sample measurements plus and minus the calculated standard deviation and corrected for spiking inhibition, if necessary. Sample points marked with an asterisk are those in which no spiking correction was possible. The zone in gray underneath the measurement points is the zone below the calculated MDL for every individual standard curve. Note that all indicated sample measurements were significantly different from the blank, even when lying below the calculated MDL. underestimation because of poor spiked-sample control calculations (see the asterisks in Figure 2) and reporter assay values outside the standard curve. In a fifth repetition of the oil spill in North Sea water (experiment N.Sea2), we therefore concentrated on more intensive assaying with the BTEX biosensor, in particular analyzing different sample dilutions (3- and 10-fold). Interestingly, measurements now showed toluene equivalent concentrations of up to 80 µM after 6 h (Figure 2, panel N.Sea2). Toluene equivalent concentrations in the samples decreased 1-2 days after the spill, but remained significantly above the MDL in the assays for the rest of the experiments (7-10 days). There was no significant induction of the genotoxicity biosensor E. coli MG1655 (pJAMA8-cda) compared to the nonexposed seawater control in any of the spills at any time point (data not shown). In contrast to the CDA assay, the HBP assay produced reporter output significantly above the MDL limit in a number of time points in various spills, but did not show any particular trend (Figure 3A). It has to be mentioned, though, that for one spill (Med1) we used the bacterial strain E. coli DH5R (pHYBP109), which has a specificity of detection only for hydroxylated biphenyls. The slight increase of signal observed near the end of that spill suggests that perhaps hydroxylated metabolites were produced by biological transformation during the course of time. Two other spills (i.e., Med2 and N.Sea1) were followed with a bacterial reporter strain, E. coli DH5R (pHYBP103M3), which covers a compound range extending into nonhydroxylated biphenyls (16). Assays with this bioreporter strain produced a very slight peak of detection in samples taken during the first 2-6 h, followed by a rapid decrease, on two spill occasions (Figure 3A). Although those signals were higher than the signals in the blanks of the corresponding calibration curves, they were not significantly above the calculated MDL. 1052
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 3, 2010
FIGURE 2. Time development of equivalent toluene concentrations in the aqueous phase after the oil spill as outlined in Figure S1 in the Supporting on five different repetitions, here labeled with (A) Med1, (B) Med2, and (D) Med3 (seawater from the Mediterranean) and (C) N.Sea1 and (E) N.Sea2 (seawater from the North Sea). For other explanations, see the caption to Figure 1. Note that N.Sea2 assays were performed with a larger number of sample dilutions. Naphthalene Reporter Assays. Light emission in assays with the NAH reporter strain B. sartisoli RP007 (pPROBEphn-luxAB) was linearly proportional to increasing naphthalene concentrations up to 3 µM (r 2 ) 0.9817 ( 0.0169, Table 2). Above that concentration, which is far below the aqueous solubility of naphthalene (∼250 µM), luciferase activity did not further increase (not shown). Relative errors based on standard deviations in triplicate assays ranged from 9% to 18%. The NAH bioreporter assay was used on samples coming from two experimental spills (Figure 3B). In one of those (N.Sea2) the samples produced values slightly above or below the MDL of the assay, which on that particular occasion was relatively high due to large variations in the blank. In addition, significant assay inhibition was detected in a number of samples, which forced us to dilute them, but the result of that was that the reporter signal became too close to the MDL. For the other spill (Figure 3, panel Med3), however, most of the samples produced a clear signal in the NAH assays above the MDL. Over time, this produced a trend for appearance of naphthalene equivalent material in the aqueous phase, which was different from that observed in the OCT and BTEX assays. The naphthalene-equivalent profile had a maximum concentration of ∼1 µM 2-3 days after the spill and a much slower disappearance of the signal for the remaining period. PAHs could still be detected at concentrations above 0.5 µM equivalent naphthalene after 5 days, at which time point alkanes were below the MDL in the OCT assay and BTEX had decreased to around 200 nM toluene equivalent. Chemical Analysis. To corroborate bioreporter assay data, we performed chemical analyses for PAHs and n-alkanes in the aqueous-phase samples from three of the oil spills (Figure S3, Supporting Information). Unfortunately, it was not possible to measure short-chain alkanes below C10, BTEX,
Bacterial Community Changes in the Seawater upon Oil Spill. Although this study did not focus per se on bacterial degradation of oil components during time in the assay (7-10 days), we monitored bacterial community changes over time in two of the spills by using molecular techniques based on 16S rRNA gene amplification. Figure S4 in the Supporting Information shows the results from DGGE analysis of the variable V3 regions of bacterial 16S rRNA genes from experiment Med2. In the absence of added oil the bacterial community in the seawater changed little over the course of 7 days, while oil addition caused a big change in community composition, even after 1 day (Figure S4A). The biggest change was between days 1 and 3, after which the community was stable. These changes may have been caused in part by loss of sensitive species, but mostly by the growth of new hydrocarbon-degrading organisms. The latter is supported by the detection of several dense bands deriving from Marinobacter species, which are common marine oildegrading microbes (18). Similar shifts in microbial community composition were observed for the Med1 experiment, except that the identified organisms were largely different (data not shown). Therefore, some of the differences seen in hydrocarbon concentration between the various experiments may reflect the abundance and activity of different hydrocarbon-degrading bacteria in the samples at different times of year and their locations. FIGURE 3. Time development of equivalent 2-HBP or naphthalene concentrations in the aqueous phase after the oil spill as outlined in Figure S2 on five different repetitions, here labeled with Med1, Med2, and Med3 (seawater from the Mediterranean) and N.Sea1 and N.Sea2 (seawater from the North Sea): (A) HBP assay, (B) naphthalene assay. For other explanations, see the caption to Figure 1. and hydroxylated compounds with the same analysis methodology. Quantification of the total (methyl)naphthalenes plus phenanthrene in the samples produced a similar trend over time of the spill as detected with the NAH reporter assay with a maximum concentration of between 0.2 and 1 µg/mL in the aqueous phase between 24 and 72 h. The most abundant dissolved PAHs detected by GC-MS were the low molecular mass PAHs naphthalene, methylnaphthalenes, dimethylnaphthalenes, and phenanthrene. Assuming that the 1 µM naphthalene equivalent detected in the bioreporter assay consisted solely of naphthalene, this would correspond to 0.13 µg/mL, whereas the chemical analysis measured between 0.18 and 0.80 µg/mL in three spills (Figure S3). As seen for the total PAHs, the total n-alkane quantification showed no obvious trend on three occasions with an average concentration in the seawater of between 0.5 and 2.8 µg/mL (Figure S3, Supporting Information). The most abundant n-alkanes were undecane, tetradecane, pentadecane, hexadecane, and octadecane. Pristane and phytane were also detected in high amounts in the seawater samples in concentrations up to 100 ng/mL. Previous work had indicated that the reporter strain used for the OCT assay is capable of detecting linear alkanes in the range of C6-C10 with the highest relative reporter signal generation for octane and nonane (Table S1, Supporting Information) (17). When concentrating on quantification of undecane plus dodecane only, there was a trend in chemical analysis similar to that of the bioreporter assay of a maximum appearance in the water phase 2-6 h after spillage (Figure S3). Concentrations of undecane plus dodecane in the aqueous phase reached up to 0.04 µg/mL, which corresponds to ∼0.2 µM assuming the molecular mass of undecane. The quantified data for available linear alkanes of the reporter assay thus showed the same order of magnitude as the chemical analysis.
Discussion We tested here the utilization, practicality, and accuracy of bioassays for the detection of several key petroleum hydrocarbons in seawater using a set of five whole-cell bacterial bioreporters. Specifically, we were interested (i) to verify the capacity of the bioassays to detect and measure the concentration of dissolved oil components in seawater and (ii) to study the dynamics of the hydrocarbon dissolution and the fate of the dissolved chemicals after several days of incubation in a model oil spill experiment. We did not specifically focus on the sampling strategy, although we acknowledge that such strategies are important to compare bioreporter assays in the field to more convential methods of sampling/analyzing oil hydrocarbons in the marine water column (19). Figures of merit for the bioassays were satisfactory, in terms of overall coefficients of regression (r 2 > 0.96), MDL (
