Anal Bioanal Chem (2002) 373 : 696–703 DOI 10.1007/s00216-002-1313-z
S P E C I A L I S S U E PA P E R
M. Farré · C. Gonçalves · S. Lacorte · D. Barceló · M. F. Alpendurada
Pesticide toxicity assessment using an electrochemical biosensor with Pseudomonas putida and a bioluminescence inhibition assay with Vibrio fischeri Received: 5 December 2001 / Revised: 11 February 2002 / Accepted: 11 April 2002 / Published online: 4 July 2002 © Springer-Verlag 2002
Abstract Two different toxicity tests, an electrochemical biosensor Cellsense and a bioluminescence inhibition assay ToxAlert were performed in order to establish and compare the acute toxicity responses of different types of raw and spiked water for a selected group of pesticides. The selected compounds were endosulfan, chlorfenvinphos, dimethoate, fenamiphos, ametryn, deltamethrin and α-cypermethrin; all of them are used in large quantities for agricultural purposes. In the first step, the study of the toxicity responses for each individual pesticide with Milli-Q water was carried out. Next, the toxic responses of different mixtures of these pesticides in different water matrices, i.e., Milli-Q water, surface water, groundwater and wastewater were studied in order to evaluate (i) device advantages and limitations for the toxicity evaluation of real environmental samples, (ii) antagonistic or synergistic effects and (iii) the influence of the water matrices. The survey of pesticides in real samples was carried out using a combined method involving both chemical analysis and toxicity bioassays. Chemical analysis involved the use of solid-phase micro-extraction (SPME) followed by gas chromatography with electron capture detection (GC/ECD) or thermoionic specific detection (GC/TSD) with mass spectrometric confirmation (GC/MS).
Introduction Tons of pesticides are produced and used every year for control of pests, and in agriculture and horticulture. Wastewater from greenhouses and runoff from agricultural land can reach and contaminate rivers, lakes and groundwater. M. Farré · S. Lacorte (✉) · D. Barceló Department of Environmental Chemistry, IIQAB-CSIC, Jordi Girona 18-26, 08034 Barcelona, Spain e-mail:
[email protected] C. Gonçalves · M.F. Alpendurada Laboratory of Hydrology, Faculty of Pharmacy, University of Porto/R. Anibal Cunha 164, 4050-047 Porto, Portugal
Pesticides and other organic pollutants have been detected in receiving bodies of water, at µg/L levels. However, one of the limitations of chemical analysis is that when used alone, it does not provide an entire response of the effects organic pollutants may have on the environment. Therefore, its use in conjunction with techniques that can measure biological effects, such as toxicity tests, is highly recommended [1,2] in order to establish risk assessment practices. The present paper reports on a combined method using toxicity tests and chemical analysis, in order to evaluate the occurrence and aquatic toxic impact of a selected group of pesticides of different chemical families. This protocol has been applied to different types of water to evaluate the performance of both toxicological and analytical methods. The compounds studied were organochlorine, organophosphorus, a triazine, and pyrethroid pesticides, all of them of widespread use in many countries. The pesticides fenamiphos, chlorfenvinphos and endosulfan are very toxic to mammals and are classified by the Environmental Protection Agency (EPA) as restricted use pesticides (RUP) or class I, whereas dimethoate, deltamethrin and cypermethrin are moderately toxic to mammals, i.e. EPA toxicity class II. Ametryn is an unrestricted pesticide, toxicity class III, that means slightly toxic [3].It is relatively nontoxic to mammals and fish [4] but highly toxic to crustaceans and mollusks [5]. From the selected compounds, chlorfenvinphos and endosulfan are priority pollutants included in the new Framework Directive on Pollution of the European Union, and in 1984 endosulfan was classified as a hazardous chemical by the World Heath Organization (WHO) [6]. Its use as an insecticide in a wide variety of food crops including fruits, vegetables and cereals has somewhat increased recently due to decline in the use of other insecticides such as endrin. During 1999, 85 t of this compound were applied in Portugal alone. In many countries the use of endosulfan is restricted due to acute toxicity towards aquatic animals [7] and has been replaced by organophosphorus pesticides. Among the most common ones, chlorfenvinphos and dimethoate are widely used as insecticides and acaricides whereas fenamiphos is one of
697
the most prominent nematicide. About 7 t of chlorfenvinphos and 68 t of dimethoate were applied in Portugal during 1999. The group of organophosphate compounds are nowadays the most widely used group of insecticides in the world, especially in developing countries because they are cheaper than the newer alternatives like synthetic pyrethroids. As many other triazines, ametryn is a selective herbicide used on corn and potatoes for general weed control. The main problem presented for this pesticide is its persistence in groundwater. It moves both vertically and laterally in soil due to its high water solubility. It may leach as a result of high rainfall, floods, and furrow irrigation [8]. In general, the widespread application of s-triazine herbicides often leads to their presence in fresh water [9,10] and marine coastal environments [11]. Two synthetic pyrethroids were included in the present work, deltamethrin and α-cypermethrin. The synthetic pyrethroids are widely used for control of public-health pest, household insects, insects on crops and animal ectoparasites, etc., because of their general low toxicity to birds and mammals. Cypermethrin has become one of the most important insecticides in wide-scale use. Nevertheless, it is highly toxic to fish and aquatic invertebrates. The 50% lethal concentration (LC50) (96 h) for cypermethrin in rainbow trout is 0.0082 mg/L and in bluegill sunfish 0.0018 mg/L [12]. The 50% effective concentration (EC50) in Daphnia magna, a freshwater crustacean, is 0.0002 mg/L [10]. Under normal environmental temperatures and pH, cypermethryn is stable to hydrolysis, with a half-life greater than 50 days and to photodegradation, with half-life greater than 100 days [13], and it is more persistent in anaerobic conditions, such as occurs in groundwater. Like cypermethrin and other pyrethroids, the use of deltamethrin increases every year, in particular in developed countries for agricultural, public health and livestock applications, but it is highly toxic to fish under laboratory conditions and has a high impact on aquatic fauna, particularly on crustacean. All these compounds are present alone or as complex mixtures with other compounds in different types of water , according to the formulations or uses. Therefore, the first step of this work was the toxicity evaluation of each of the selected pesticides by studying toxic responses, defined by the EC50 and the toxicity units (TU). In order to do this, two different systems Cellsense and ToxAlert100 were compared using different test species, Pseudomonas putida immobilized on screen printed electrodes and Vibrio fischeri, respectively. Next, the toxic responses of mixtures of these pesticides were evaluated by both tests in different water matrices such as, Milli-Q water, surface water, groundwater and wastewater in order to evaluate the limitations of the toxicity tests for the analysis of environmental samples. At the same time, antagonistic or synergistic effects between different mixtures of pesticides and the influence of the water matrices were evaluated. All studies involving pesticide determination in surface, ground and wastewater samples were performed by a combined method involving toxicity evaluation and
chemical analysis based on solid-phase micro-extraction (SPME) coupled with gas chromatography with electron capture detection (GC/ECD) or thermoionic specific detection (GC/TSD), with mass spectrometric confirmation (GC/MS). This work is part of the Toxicity Identification Evaluation (TIE) Program that has been developed at our Department, intended to combine effect-related assays with chemical analysis. The final goal is to establish correlations between toxicity values and specific pollutants. To our knowledge, such a study on real environmental waters has not been previously performed.
Experimental Chemical and reagents Pure pesticides were supplied by Riedel-de-Haën (Seelze, Germany). Individual standard stock solutions at a concentration about 1000 mg/L were prepared in methanol, that is the solvent recommended for use with toxicity tests and can be tolerated by the bacteria at a max. conc. of 10%. All solvents used were of LiChrosolv gradient grade purchased from Merck KGaA (Darmstadt, Germany). Ultrapure Milli Q water (Millipore SA, Molsheim, France) and NaCl of analytical reagent grade from Panreac (Barcelona, Spain) were used to prepare a 0.85% saline solution for Cellsense experiments. Pseudomonas putida (ST02/I.NB/280300) biosensors and freeze-dried substrate were obtained from Terra Nova Systems, (Cambridge, UK). Liquid-dried photo-bacteria reagent Vibrio fischeri NRRL B-111 77 was kindly supplied by Merck KGaA (Darmstadt, Germany). Toxicity tests Two types of acute toxicity measurements based on the use of bacteria as a test species were used to assess the toxic effect of selected pesticides. Cellsense is an amperometric biosensor [14] which measures the electrical current produced by the bacteria’s electron transport chain, the current generated being proportional to the level of metabolic activity of the cells. An electron mediator is added to the sample in order to provide a current flow between the bacterial cells immobilized on a screen-printed electrode and the transducer. This device is able to give a continuous signal monitored at very short time intervals over a chosen length of time. The toxicity of the standard/sample is determined by measuring the degree of inhibition of the biosensor signal after a fixed time of exposure, with respect to the control biosensor. Different bacterial species can be immobilized on the electrodes. Further details and operation of the equipment can be found elsewhere [14,15]. In this work we used P. putida as the biological component, described for the first time by M. Farré et al. [16], and all toxicity measurements refer to a 30 min exposure.
698
The second type of acute toxicity measurements were based on the bioluminescence inhibition of Vibrio fischeri NRRL B-111 77 by means of the ToxAlert 100 system from Merck. This is a well-established test extensively described in the literature [17, 18,19] and similar in principle to the better-known Microtox test. Toxic substances will cause changes in cell structures and/or metabolic pathways of marine V. fischeri, which are rapidly reflected in a decrease of bioluminescence. Inhibition due to the toxic effect of the sample is calculated against the response given by a saline control solution, which accounts for natural decrease in light. Data were collected after 15 and 30 min exposure times. In all cases, individual pesticides were diluted in methanol first and working solutions between 0.3 and 14 mg/L were prepared in HPLC water in order not to exceed the maximum allowed percentage of methanol (10%). The inhibition curve of each individual compound was fitted determining the inhibition produced over a wide range of concentrations. The EC50 of each substance can then be determined along with TUs of standard substances, calculated according to the formula of Sprague and Ramsay (1965) [20]:
Study of matrix effect
TUs = (EC50 )−1 ×100
Five groundwater samples from Póvoa de Varzim, an area of intensive agricultural in Portugal, were collected by pumping water from a well. The wells were from 4 to 10 m in depth These samples were analyzed following the analytical protocol indicated above and their acute toxicity was assessed in terms of ToxAlert and Cellsense analysis.
(1)
Chromatographic analysis Determination of pesticides in water samples was performed by SPME using a 60 µm PDMS/DVB-coated fiber, coupled with GC/ECD andGC/TSD. Extractions were achieved by immersion of the fiber in 3 mL of sample, with permanent stirring at 60 C. Neither pH adjustment nor ionic strength correction was needed. Analytes were allowed to adsorb onto the fiber for 60 min, and afterwards desorbed in the hot injection port of the gas chromatograph, for 5 min. Chromatographic analyses were carried out in a Varian 3400 CX (Walnut Creek, USA) gas chromatograph. The injector and detector temperatures were set at 250 and 310°C isothermal, respectively. All compounds were resolved in a MDN-5 column (30 m×0.32 mm id×0.25 µm film) (Supelco, Bellefonte, USA) using helium as the carrier gas and detected either by ECD or TSD operating at 3.2 A intensity. An adjustable splitter (SGE Europe Ltd., Milton Keynes, UK) was inserted at the column exit in order to divert about a tenth of the effluent flow to ECD and the remaining to TSD. This instrumental configuration was optimized for the determination of 34 pesticides in a single 30-min chromatographic run following a single extraction procedure. Confirmatory analysis was performed in a Varian Saturn 2000 GC/MS equipped with a CPSil 8 CB low bleed MS column (30 m×0.25 mm id×0.25 µm film) (Varian Chrompack Int., Middelburg, Netherlands). Following a previous extraction and concentration procedure according to the same SPME protocol described above, chromatographic separation took place and mass spectrometric data were acquired in full scan mode for spectra analysis.
Water samples with different organic carbon content, namely surface water from the river Llobregat (TOC of 29.7 mg/L), groundwater from a well in Barcelona (TOC 2.0 mg/L) and WWTP effluent (treated wastewater) from Igualada (Spain) (TOC 230.3 mg/L) were chemically characterized using SPME and GC-ECD, and GC-MS, as described. These natural and spiked water samples were used to determine possible matrix effect with both Cellsense and ToxAlert. The spiking experiments were carried out with different mixtures in order to obtain the following final concentrations in each type of water. Mixture M1: 1.5 mg/L endosulfan, 14 mg/L dimethoate, 8 mg/L fenamiphos and 5 mg/L ametryn. Mixture M2: 1.5 mg/L endosulfan, 14 mg/L dimethoate, 8 mg/L fenamiphos, 5 mg/L ametryn and 5 mg/L chlorfenvinphos, and mixture M3: 0.364 mg/L endosulfan, 0.364 mg/L dimethoate, 0.364 mg/L fenamiphos and 0.364 mg/L ametryn. Groundwater assessment
3. Results and discussion ToxAlert 100 and Cellsense performance The use of luminescent bacteria as a toxicity test species presents many advantages such as, rapid, simple, sensitive and cost-effective analysis, however, some disadvantages and some particular characteristics must be mentioned. For the good performance of the luminescence assay (ToxAlert), a filtration step is necessary prior to every test, and it must work in 2% saline solution in order to maintain the osmotic pressure of V. fischeri. As a consequence of salinity, the insolubility of some organic substances are enhanced producing turbid solutions. In these cases, the luminometer’s measure is lower than the luminescence produced by V. fischeri, because photons are partially dispersed when they interact with colloids. This was inconvenient for synthetic pyrethroids because the more concentrated solutions exhibited significant turbidity. In addition, precaution has to be taken with the solvent used and its concentration in the solution. The maximum percent of methanol accepted for the test is 10%. As regards to Cellsense, the bacterial biosensor, it allows the use of different test species to investigate specific groups of toxins, and proper selection of the organism is essential. Some bacteria species are especially sensitive to specific groups of substances, nevertheless the same species can present high resistance to others sub-
699 Fig. 1 Pseudomonas putida and Vibrio fischeri inhibition curves for standard substances after 30 min exposure. The curves show the percent of inhibition vs. the logarithm concentration (c), expressed in µg/ml
stances. Measurements using the Cellsense system are not disrupted by turbidity, even when measuring suspensions, which is an advantage, especially for wastewater samples. On the other hand, the measurable solutions must be free of particles, which can precipitate on the electrode or electrochemical active sites. Taking into consideration the precautionary actions mentioned above, Fig. 1 shows the inhibition curves obtained by Cellsense and ToxAlert 100 for four selected pesticides after a 30-min exposure time. It can be observed that P. putida was more rapidly saturated than V. fischeri, indicating a smaller linear range, but P. putida was much more sensitive to the compounds studied. Table 1 indicates the EC50, TU, goodness of fit (R2), standard deviation (Sy, x) and the theoretical equations corresponding to the fitted curves, for V. fischeri and P. putida after 30 min exposure. As regards to the quality parameters indicated in Table 1, the toxicity responses using ToxAlert 100 and V. fischeri showed high reproducibility with a standard deviation of 8.8%, while this value for Cellsense was 19.9%. For the selected pesticides, P. putida gave values ranging between 2.8 and 113 TUs, whereas, for V. fischeri the TUs were between 0.9 and 18. With V. fisheri, the highest TU corresponded to endosulfan, followed by ametryn, indicating high acute toxicity. As expected, the pyrethroids presented the lowest values, a result which was corroborated by the Cellsense analysis. The low acute toxicity of pyrethroids supports their safer use in more and more applications, especially domestic ones. Using Cellsense, the compounds which showed the highest toxicity were the organophosphorus pesticides due to inhibition of the cellular response. However, the organophosphorus pesticide chlorfenvinphos could not be measured using the bacterial biosensor Cellsense due to the fact that this compound is electrochemically active. Chlorfenvinphos acts as a redox substance interfering in the electronic interchange between the mediator and the transducer producing an over
signal and in such cases, inhibition cannot be measured. In mixtures containing this kind of compound the inhibition measured will be lower than the real value. This constitutes a limitation of the system and it is necessary to use electrodes without bacteria immobilized on them in order to check if a substance, real sample, or mixture is electrochemically active. For this reason, chlorfenvinphos was eliminated from the mixtures tested using Cellsense. Matrix effect Subsequently, the toxic responses using both systems for four different water matrices, Milli-Q, groundwater, river water and a wastewater effluent from a wastewater treatment plant (WWTP) was determined. The toxicity values are reported in Table 2, expressed as %I of these water samples. Chemical analysis of these water samples was also performed, and compounds encountered and their concentration are also indicated in Table 2. Milli-Q and groundwater samples were non-toxic, with inhibition values lower than 10% and the results were similar for both systems. Nevertheless, river water contained 0.15 µg/L diazinon which was not detected by the toxicity tests. In contrast, raw wastewater showed inhibition using both systems (18.2% with ToxAlert 100 and 15% with Cellsense). This increase in toxicity was attributed to the presence of 0.19 and 0.74 µg/L diazinon and fenitrothion, as confirmed with chemical analysis. The higher value with ToxAlert 100 could be partially influenced by the slight yellow color of the wastewater effluent. These 4 different types of water were then spiked with the mixtures M1, M2 and M3 in order to evaluate the matrix effect using ToxAlert 100 and Cellsense (only M3). By performing the analysis with spiked water, two different phenomena were observed: on one hand, the inhibition of mixtures of pesticides was lower, in all types of
%I=–0.0806 +(102.6248/[1+10(-0.0145–LogC)0.928743]) 4.74 31.63 Fenamiphos
3.2
0.9791
5.12
%I=–0.3604 +(104.3080/[1+10(1.5001–LogC)0.6352])
0.928
109
0.997
%I=–1.8042 +(103.8334/[1+10(0.5304–LogC)1.007316]) 7.01 5.63 Endosulfan
17.8
0.9760
7.16
%I=–2.1418 +(90.8013/[1+10(0.7503–LogC)3.5190])
3.38
29.6
0.993
%I=–2.0323 +(94.9771/[1+10(–0.08545–LogC)2.4511]) 19.89 113 0.888 %I=–1.2716 +(103.5049/[1+10(1.744–LogC)0.7854]) 6.11 0.9723 55.46 Dimethoate
1.8
28.0 %I=–0.7164 +(100.5356/[1+10(2.0036–LogC)2.34508]) 7.54 0.9748 100.84 Deltamethrin
0.99
36.1 %I=–0.6063 +(102.2582/[1+10(2.0428–LogC)1.7789]) 8.82 0.9596 110.36 Cypermethrin
0.91
0.869
%I=4.2458 +(102.0965/[1+10(1.4465–LogC)6.2783]) 17.58 3.6
0.964
%I=0.0738 +(104.1865/[1+10(1.5579–LogC)9.3358]) 11.89 2.8
0.913
–––––%I=1.2211 +(103.3184/[1+10(24.5867–LogC)0.9501]) 3.13 0.994 24.59 Chlorfenvinphos
4.1
%I=–1.9306 +(94.5500/[1+10(0.00119–LogC)0.7017]) 6.40 0.988 45.3 2.21 %I=–1.8030 +(94.482/[1+10(1.270–LogC)1.270]) 4.24 0.984 5.4 18.64 Ametryn
Equation Syx R2 TU EC50 (µg/ml) EC50 (µg/ml)
TU
R2
Syx
Equation
Cellsense P. putida ToxAlert 100 V. fischeri Compound
Table 1 50% Effective concentration (EC50), toxicity units (TU), goodness of fit (R2), standard deviation (Sy.x) and theoretical equations corresponding fitted curves for V. fischeri and for P. putida after 30 min exposure time
700 Table 2 Toxicity values expressed as %I for four different water matrices using ToxAlert 100 with V. fischeri and Cellsense with P. putida. Chemical identification of pesticides was performed with solid-phase micro-extraction (SPME) followed by gas chromatography with electron capture detection (GC/ECD) or thermoionic specific detection (GC/TSD) with mass spectrometric confirmation (GC/MS). n.d.=not detected Compound
ToxAlert 100 V. fischeri %I
Cellsense P. putida %I
Chemical characterization (µg/L)
Milli-Q Groundwater River water Wastewater
