2456 Environmental Toxicology and Chemistry, Vol. 20, No. 11, pp. 2456–2461, 2001  2001 SETAC Printed in the USA 0730-7268/01 $9.00 + .00 A NOVEL TOXICITY FINGERPRINTING METHOD FOR POLLUTANT IDENTIFICATION WITH lux-MARKED BIOSENSORS NIGEL L. TURNER,*² A LISON HORSBURGH,² G RAEME I. PATON,² K EN KILLHAM,² A NDREW MEHARG,² SANDY PRIMROSE,‡ and NORVAL J.C. STRACHAN² ²Department of Plant and Soil Science, Cruickshank Building, University of Aberdeen, Aberdeen AB24 3UU, Scotland, United Kingdom ‡AZUR Environmental, 540 Eskdale Road, Winnersh Triangle, Wokingham, Berks RG41 5TU, United Kingdom ( Received 10 October 2000; Accepted 9 April 2001) Abstract—A novel technique is described for the identification and quantification of environmental pollutants based on toxicity fingerprinting with a metabolic lux-marked bacterial biosensor. This method involved characterizing the toxicity-based responses of the biosensor to seven calibration pollutants as acute temporal–dose response fingerprints. An algorithm is described to allow comparisons of responses of an unknown pollutant to be made against the calibration data. This is based on predicting pollutant concentration at each of six different time points over the course of a 5-min assay. If the prediction is consistent between the unknown pollutant and a calibration pollutant at the 95% test level, this is considered to be a positive identification. All seven calibration pollutants could be successfully distinguished from each other with this technique. Environmental samples, individually spiked with single concentrations of pollutants, were compared in this way against the calibration pollutants. An 83% identification success was achieved, with no false positives at the 95% test level. This is a simple and rapid technique that potentially can be applied to monitoring of industrial wastewater or as a screening tool for regulators. Keywords—Toxicity fingerprinting lux-Marked biosensor INTRODUCTION Existing environmental legislation relating to chemical pol- lution in freshwater is mostly based upon total pollutant con- centration and thus does not take into account the effect of water chemistry on pollutant speciation and bioavailability. However, a movement is increasing toward the implementation of toxicity-based consents, where the direct toxicity of efflu- ents to biosensors is considered as the critical factor in deter- mining acceptable levels of release [1,2]. The recent imple- mentation of Part IIa of the United Kingdom 1990 Environ- mental Protection Act reflects this change in approach to pol- lution assessment [1]. At present, monitoring and identification of environmental pollutants in freshwater is performed predominantly by lab- oratory-based chemical analysis. Such an approach is often time-consuming and expensive and requires prior knowledge of the pollutants present for most practical purposes, because chemical identification techniques are generally specific to in- dividual pollutants or pollutant classes [3]. In addition, total pollutant load does not always accurately reflect environmental toxicity because pollutant bioavailability is dependent upon water chemistry. For these reasons, a movement is increasing toward the use of biological techniques for the monitoring of environmental pollution by industry and regulators alike [2]. Bioassays that estimate toxicity by determining the toxic effect on an organism are based primarily on higher organisms such as crustaceans or fish [3]. These assays are generally impractical for use as rapid monitoring or screening tools, because they are time-consuming to perform. Bioluminescent bacterial assays, including Microtox (AZUR Environmental, Carlsbad, CA, USA) [4,5] and lux-marked biosensors, can be * To whom correspondence may be addressed (n.l.turner@abdn.ac.uk). very rapid and are simple and inexpensive to perform [6,7]. The use of metabolic lux biosensors (where toxicity is related to a decrease in cellular luminescence) has been applied suc- cessfully to a range of environmental pollutants, including the assessment of copper bioavailability in malt whisky distillery effluent [8], assessing the toxicity of papermill effluents [9], and determining the relative toxicity of 20 metal ions [10], chlorobenzenes [11], and chlorophenols [12]. These assays have been found to be sensitive and reproducible [8], and because terrestrial bacteria are used they are not limited by a narrow pH range. Also, addition of NaCl is not needed, as is the case with the widely used Microtox assay. However, such assays generally are used only to quantify toxicity of envi- ronmental samples and therefore lack the qualitative potential of chemical analytical techniques [13]. Recent work [14], with a battery of seven Escherichia coli strains carrying lux genes fused to different promoters, iden- tified 23 out of 25 test chemicals within a 3-h timescale. These catabolic biosensors gave increasing light output with increas- ing concentration of toxicant, although because the promoters used are specific to individual chemicals, or groups of chem- icals, this technique is not well suited for universal blind tox- icity screening. To date, metabolic luminescent bacterial bio- sensors have not been used in a fully qualitative way for pol- lutant identification. Chemical fingerprinting techniques have been developed for pollutants such as hydrocarbons from gas chromatography–mass spectroscopy data [15,16], but such techniques are usually highly specific and generate data re- quiring a method of analysis considerably different to that from the temporal response curves of lux-marked biosensors. These require a new method of data analysis. This study describes the development and application of a method by which pollutants can be identified by characterizing their joint temporal–dose responses to a single nonspecific lux-