D

Ca

b

a

ARRAA

KTDUFE

1

ratttscicoslt[

ebett

0d

Analytica Chimica Acta 641 (2009) 59–63

Contents lists available at ScienceDirect

Analytica Chimica Acta

journa l homepage: www.e lsev ier .com/ locate /aca

irect toxicity assessment of toxic chemicals with electrochemical method

hang Liua,b, Ting Suna,∗, Xiaolong Xub, Shaojun Dongb,∗

College of Sciences, Northeastern University, Shenyang 110004, ChinaState Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun 130022, China

r t i c l e i n f o

rticle history:eceived 18 January 2009eceived in revised form 16 March 2009ccepted 16 March 2009vailable online 24 March 2009

eywords:oxicity

a b s t r a c t

Electrochemical measurement of respiratory chain activity is a rapid and reliable screening for the toxicityon microorganisms. Here, we investigated in-vitro effects of toxin on Escherichia coli (E. coli) that was takenas a model microorganism incubated with ferricyanide. The current signal of ferrocyanide effectivelyamplified by ultramicroelectrode array (UMEA), which was proven to be directly related to the toxicity.Accordingly, a direct toxicity assessment (DTA) based on chronoamperometry was proposed to detectthe effect of toxic chemicals on microorganisms. The electrochemical responses to 3,5-dichlorophenol(DCP) under the incubation times revealed that the toxicity reached a stable level at 60 min, and its 50%

−1

irect toxicity assessmentltramicroelectrode arrayerricyanidescherichia coli

inhibiting concentration (IC50) was estimated to be 8.0 mg L . At 60 min incubation, the IC50 values forKCN and As2O3 in water samples were 4.9 mg L−1 and 18.3 mg L−1, respectively. But the heavy metal ions,such as Cu2+, Pb2+ and Ni2+, showed no obvious toxicity on E. coli. With the exception of Hg2+, it showed40.0 mg L−1 IC50 value when E. coli was exposed to its solution for 60 min. The lower sensitivity of DTA forthe heavy metal ions could be attributed to the toxicological endpoint and the experimental conditions

hat th.

used. All results suggest twastewater toxic analysis

. Introduction

Besides rare but spectacular and media-reported accidents,apid development of industrial activities and chemical materi-ls results in increasing contamination of the environment [1]. Inhe last decade, various biological models have been developedo study toxicity and measure the risk of environmental pollu-ion. Existing conventional techniques rely on eucaryotic species,uch as daphnids, fish that sacrifice time, need high cost of spe-ialized equipment and trained personnel [2–6]. Consequently, its necessary to develop a simple, rapid and inexpensive analyti-al method for monitoring and detecting environmental pollutionr potential risk to human health. Microorganisms incorporatingome analytical tools meet these criteria because of their shortife cycle and sharp response to toxin, and rapidly adapted tooxicological studies for assessing organisms or ecosystem health7–14].

The Microtox® test based on the fluorescent fading, in the pres-nce of toxic materials, is considered to be a superior and rapid

acterial assay available. Toxicity is detected from 5 to 30 minxposure in this system, and is usually expressed as 50% effec-ive concentration (EC50) of the toxic material value correspondingo a 50% loss of luminescence. At present, this in-vitro bioassay

∗ Corresponding authors. Tel.: +86 431 85262101; fax: +86 431 85689711.E-mail addresses: Sun1th@163.com (T. Sun), dongsj@ciac.jl.cn (S. Dong).

003-2670/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2009.03.027

e DTA is a sensitive, rapid and inexpensive alternative to on-site water and

© 2009 Elsevier B.V. All rights reserved.

has been already successfully used to screen the acute toxicity ofa large number of chemicals [1]. However, the drawback of theMicrotox® system based on the measurement of light intensity isthat it restricts cell populations strictly, and is not suitable for sam-ples of high turbidity, which would cause the scattering from thebioluminescent analytical signal. Furthermore, the luminous bac-terial must work in 3% saline solution in order to maintain osmoticpressure of bacteria [15]. This would decrease the solubility of someorganic chemicals.

In order to complement the defects of Microtox® test, microor-ganisms combined electrochemical tests have been developed[7,11,15–19]. One model, a system of the direct toxicity assessment(DTA) originated from the MICREDOX, has been proposed by Pascoand co-workers [11]. The MICREDOX, developed by Lincoln Tech-nology, has been applied to biochemical oxygen demand (BOD)monitoring [20–24]. In the biocatalytic process, the synthetic co-substrate or mediator substitutes the nature co-substrate oxygenand facilitates the rapid reaction by combining high concentrationof microbial population [25]. The mediator carries the electronsand moves between intracellular space and the extracellular envi-ronment. Then the reduced mediator is eventually determined byelectrochemical methods. According to the principle of the method,

toxicity can be easily estimated as a deviation away from the elec-trochemical signal produced by uncontaminated cells. Scheme 1shows the principle of the DTA.

The successful development of new sensor for monitoring envi-ronment requires a transducer with sensitive response to the

60 C. Liu et al. / Analytica Chimica Acta 641 (2009) 59–63

f direc

masoet

iidFsiptiavta

2

2

sKSPi

1dyp1Iafciiu

2

cagtTv

output of limiting current at a appropriate concentration of toxi-

Scheme 1. Principle o

etabolic perturbations by pollutants [7]. The microelectrodemperometry combining with cell-based sensor rapidly achievesteady state in less than 10 s [26]. Thereby, the DTA that basedn combination of microorganism and microelectrode amperom-try provides the foundation for developing a successful on-siteoxicity-screening system.

Here, Escherichia coli (E. coli) was used as a model microorgan-sm for its well characterized respiratory pathway and regulationsn the mediated test [11,16,25]. Various chemicals, such as 3,5-ichlorophenol (DCP), KCN and As2O3, were chosen as toxicants.urthermore, some heavy metal ions are not stable in DTA system,uch as Cu2+, Pb2+, Ni2+ and Hg2+, which were carefully investigatedn this study. Amperometry incorporating ferricyanide as a redoxrobe was employed to determine the whole effects of chemicaloxicity on E. coli respiration rather than identifying the chemicalsncluded. Especially ultramicroelectrode array (UMEA), effectivelymplifying the signal from the total limiting currents of multi indi-iduals, was uesd to distinguish a little change in toxicity. Finally,he limiting currents were directly converted to inhibitory percent-ge by simple signal treatment.

. Experiment

.1. Chemicals and regents prepared

E. coli DH 5 � was purchased from Beijing Dingguo Chang-heng Biotechnology Co., Ltd. DCP was purchased from Aldrich.CN, As2O3, glucose and glutamic acid were purchased fromigma. Peptone and yeast extract were from OXOID Ltd. Cu(NO3)2,b(NO3)2, Ni(NO3)2 and HgCl2 were obtained from Sinophar Chem-cal Reagent Beijing Co., Ltd.

The BOD198 standard GGA solution (150 mg L−1 glucose and50 mg L−1 glutamic acid) was prepared according to APHA stan-ard methods [27]. The Luria Bertani (LB, 10 g L−1 tryptone, 5 g L−1

east extract, and 10 g L−1 NaCl) broth was adjusted to the desiredH with 2 mol L−1 HCl and sterilized in high-pressure steam at20 ◦C for 20 min. K3[Fe(CN)6] was freshly prepared before use.n the case of As2O3, mild heating, acidification and continuousgitation was necessary to obtain a complete dissolution [1]. Theollowing metal solutions for ICP (iCAP 6000 SERIFS, Thermo, USA)alibration of Cu2+, Pb2+, Ni2+ and Hg2+ (1000 mg L−1) were storedn 4 ◦C refrigerator. All chemicals used in this study were of analyt-cal grade, and all solutions were prepared with deionized water,nless otherwise stated.

.2. Cultivation of microorganisms

E. coli was maintained on nutrient agar plates at 4 ◦C. Bacterialultures were grown aerobically at 37 ◦C for 12 h in LB substrate on

shaking incubation (220 rpm). Cells were harvested by centrifu-

ation at 4500 rpm for 10 min at room temperature, then washedwice with phosphate buffer solution (PBS) and resuspended in PBS.he ultimate concentration of cells was adjusted to an absorbancealue of 24.0, measured at 600 nm using a Cary 500 Scan UV–vis-

t toxicity assessment.

NIR Spectrophotometer. The bacterial suspension was used for theexperiments on the day of harvesting.

2.3. UMEA voltammetry and amperometry

A characteristic of microelectrode, making it very attractivefrom an analytical perspective, is that its current in the diffusion-limiting region is independent of both time and potential [7]. TheUMEA fabricated by nine pieces of 25 �m single Pt ultramicroelec-trode was used as a working electrode, and was polished usingslurry of 0.05 �m � alumina powder before use. The voltammogramwas detected at a scan rate of 50 mV s−1 from 0 to 600 mV ver-sus an Ag/AgCl reference electrode (saturated KCl). Amperometricdetection was set at 450 mV that depended on results of voltam-metry. A Pt gauze auxiliary electrode was used to complete thethree-electrode system. All electrochemical measurements wereconducted using a CHI 832B electrochemical workstation (ChenHua, Shanghai).

2.4. Incubation of cells with mediator, substrates and toxicant

A total volume of 10.0 mL incubation suspension was preparedfor each trial. Replicate samples containing 45 mM ferricyanide,standard GGA198 solution, cells suspension with the absorbancevalue of ∼6.0 and toxicant with appropriate concentrations wereincubated anaerobically at 37 ◦C for desired incubation time. Con-trol incubation contained PBS in place of the toxicant, and positiveand negative controls refer to the presence and absence of GGAsolution, respectively. For investigating the influences of pH, PBSwas replaced by sterilized water. Different aliquots of 0.1 M NaOH or0.1 M HCl were used to adjust the desired pH. To terminate the reac-tion, solutions were withdrawn and centrifuged at 10,000 rpm for10 min. The supernatant solutions were maintained without oxygenand then taken for analysis.

2.5. Signal treatment

The limiting current of UMEA showed stable and amplified sig-nal for less than 10 s. The concentrations of ferrocyanide that wasproduced by the reduction of ferricyanide were directly estimatedby chronoamperograms. Furthermore, the limiting current can beconverted to equivalent inhibitory percentage values following Eq.(1):

%inhibition =(

1 − ilim(toxicant) − ilim(t−control)

ilim(p−control) − ilim(n−control)

)× 100 (1)

where: ilim (toxicant): output of limiting current at a appropriate con-centration of toxicant and standard GGA198 solution; ilim (t-control):

cant and no GGA198 solution; ilim (p-control): output of positive controlcurrent; ilim (n-control): output of negative control current

Eq. (1) is derived from the previous test that was investigatedby Pasco and his co-workers [11]. In their research, the effects oftoxicant on bacterial endogenous were negligible.

imica Acta 641 (2009) 59–63 61

3

3

bqesww

asoeaaGicirtS0wo

Fee

C. Liu et al. / Analytica Ch

. Results and discussion

.1. Feasibility of electrochemical detection

In previous studies, chronoamperometry and coulometry hadeen widely employed in the BOD detection [20,21,24,25], anduantification of the reduced mediator was achieved using bulklectrolysis as usual, which often resulted in not only time con-uming and cumbersome but destructive to the analyte [28]. Heree investigated the feasibility of voltammetry and amperometeryith the UMEA for quantifying the reduced ferricyanide.

In our study, the final concentration of 45 mM ferricyanide wasdopted instead of the normal concentration of 55 mM, becauseome effects of higher concentration of ferricyanide on E. coli wasbserved in our previous study [29]. DCP was chosen as the refer-nce for its toxicity has been widely studied using other biosensorpproaches [11]. Fig. 1a showed voltammetry results before andfter 1 h incubation of bacterial suspension in ferricyanide andGA standard solution containing DCP (DCP was replaced by PBS

n positive control solution). Prior to incubation, the ferrocyanideoncentration was zero, and negligible current was determinedn the potential range of 400–600 mV. After 1 h incubation, fer-ocyanide was accumulated by biocatalytic oxidation process, andhe anodic current increased obviously in the 400–600 mV range.

imultaneously, cathodic current decreased rapidly in range of–150 mV. Then 2 mg L−1 DCP was added, bacterial respirationas inhibited and the current of the ferrocyanide decreased obvi-usly comparing with the results of positive control. All results

ig. 1. i–E (a) and i–t (b) curves recorded electrochemical signals of a Pt UMEA (diam-ter 25 �m, nine pieces) before and after 1 h incubation of positive control, and theffect of 2 mg L−1 DCP on E. coli.

Fig. 2. Dose–response curve for the influences of pH. Changes of limiting currentsafter 60 min exposure of E. coli incubated with 45 mM ferricyanide at different pH.Each point plotted was the average of two samples.

indicated the influence of toxin on bacterial respiration could beestimated by voltammograms (VAs). The smooth signals of VAs fur-ther proved that no interference of electroactive species occurred inthis potential range. Fig. 1b showed the variation of limiting currentdepending upon time under the same conditions as voltammet-ric experiments. As expected, the UMEA provided the steady andstrong limiting current in less than 10 s and thus toxicity could bedistinguished sensitively even in a little difference of concentrationvariation.

3.2. Effects of pH and incubation time

Fig. 2 showed the changes of limiting current at different pHvalues (3.0–9.0) on E. coli after 1 h exposure to 45 mM ferricyanideand GGA standard solutions. There was a significant increase of the

limiting current at pH 3.0–5.0, and then the increase was attenu-ated at pH 5.0–7.0. In pH range of 7.0–9.0, the decrease in currentrecorded for the samples showed the evident negative effect. Thisresult was mainly due to the changes in chemical speciation of ferri-

Fig. 3. Inhibitory percentage estimated for E. coli after 30 min (�), 60 min (�) and120 min (�) of DCP exposure, respectively. Replicate samples containing 45 mM fer-ricyanide, standard GGA198 solution and cells suspension with the absorbance valueof ∼6.0.

62 C. Liu et al. / Analytica Chimica

F −1

Acia

cc

Do6ti

Fig. 5 showed the relationship between the inhibitory percent-ages and concentrations of heavy metal ions. Upon the additionof metal ions, no obvious changes of inhibition happened forPb2+, Ni2+ and Cu2+. This detention in toxicity continued in the

ig. 4. The i–E curves of 45 mM ferricyanide in the presence and absence of 20 mg Ls2O3, and no GGA and E. coli in solution (a). Inhibitory curves of E. coli at differentoncentrations of As2O3 (b) and CN− (c) were plotted. Replicate samples contain-ng 45 mM ferricyanide, standard GGA198 solution and cells suspension with thebsorbance value of ∼6.0.

yanide in the alkaline solution. Hence, pH 7.0 was used as optimumonditions for all further studies.

Fig. 3 showed the results of mean values by duplicate assays for

CP inhibitions on E. coli during 120 min incubation. There was anbvious distinction of toxicity on E. coli incubated between 30 and0 min. At 30 min, the inhibitory percentages were below 50% evenhe concentration of DCP reached 20 mg L−1. However, the toxic-ty rapidly increased to 50% inhibition when E. coli was exposed

Acta 641 (2009) 59–63

to 8.0 mg L−1 DCP at 60 min. At 120 min incubation, it was only0.5 mg L−1 decrease on respiratory inhibition comparing with thoseof 60 min. This meant that 60 min was enough for determining thetoxicity on E. coli in water samples including standard GGA198 solu-tion. In this investigation, the result of 60 min with 50% inhibitingconcentration (IC50) for DCP was more sensitive than that of San-tos reported, such as activated sludge and vibiro fischeri CellSensewith 9.8 and 37.5 mg L−1, respectively [30]. But our result was lesssensitive than that of the standard Microtox® [31].

3.3. Toxicity of KCN and As2O3 to E. coli

Toxicities of KCN and As2O3 to E. coli using the DTA were exam-ined, respectively (Fig. 4). Considering electroactivity of As3+, itspossible influence on electrochemical measurement was investi-gated before incubation with E. coli and the organic materials. Fig. 4ashowed the VAs of 45 mM ferricyanide with or without As2O3.The result demonstrated that there was no obvious influence ofAs3+ on electrochemical signals of the probe when the potentialwas set below 600 mV. Dose–response curve of As2O3 displayingan adverse influence was presented in Fig. 4b. The inhibitory per-centage increased rapidly when As2O3 concentration was below20 mg L−1 and then tended slowly. The slight difference of respira-tion inhibition induced between 20 and 100 mg L−1 As2O3 was only4.1%. In spite of this, the toxicity of As3+ for DTA was found moresensitive than that of the standard Microtox® [1].

Fig. 4c showed the inhibitory curve of E. coli exposed to CN− withdifferent concentrations. Accordingly, the value of IC50 for CN− wasestimated to be 4.9 mg L−1 with 60 min incubation. In addition, thedose–response curve gave a similar sigmoid curve with a slightlynegative inhibition at the low cyanide concentrations, followed bya rapid increase within a very narrow concentration range. Thismeant that the low level toxicant slightly accelerated the bacte-rial respiration, thereafter an obvious inhibition on microorganismswas provided because of the toxicity accumulation.

3.4. Toxicity of metal ions to E. coli

Fig. 5. Inhibition curves for Hg2+ (�), Pb2+ (�), Ni2+ (�) and Cu2+ (�). Changes ofinhibition on E. coli after 60 min exposure as a function of different heavy met-als. Replicate samples containing 45 mM ferricyanide, standard GGA198 solution andcells suspension with the absorbance value of ∼6.0.

imica

psieLattstathmia

4

dImefoefdDcft6wnirtasdo

A

RS

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[29] C. Liu, T. Sun, Y.M. Zhai, S.J. Dong, Talanta 78 (2009) 613–617.[30] L.F. Santos, L. Defrenne, A.K. Brown, Anal. Chim. Acta 456 (2002)

41–54.

C. Liu et al. / Analytica Ch

resence of 160 mg L−1 metal ions. With 60 min incubation, Hg2+-piked sample was significantly different from the other metalons. The inhibitory percentage increased rapidly throughout thexperiments, and IC50 extrapolated from Fig. 5 was 40.0 mg L−1.ow toxicity of Ni2+ and Pb2+ on Gram-negative bacteria has beenlready been reported [14,32]. In these assays, the low sensitivity ofhe growth inhibition for majority of the heavy metals was found,hough test bacteria were exposed to them over 6 h. The lower sen-itivity of the DTA for the heavy metal ions could be attributed to theoxicological endpoint and the conditions used. Our results were ingreement with Hsieh’s report on toxicity order of heavy metals,hough Cu2+ was found to be the most toxic metal among the foureavy metals [14]. It might be induced by different methods andicroorganisms used. Furthermore, we assumed that the sensitiv-

ty of inhibition was weakened by the presence of PBS, ferricyanidend neutral conditions by complexation or precipitation.

. Conclusion

A simple, rapid and reliable electrochemical method has beeneveloped successfully for the detection of various toxic substances.

n the present study, E. coli and ferricyanide were chosen as a modelicrobe and electrochemical probe, respectively. The UMEA used,

ffectively amplifying the current signal of ferrocyanide changedrom a zero background, which provided a significant advantagever other tests. The voltammetric results showed that no interfer-nce of electroactive species occurred at 0–600 mV potential rangeor the DTA. Based on the study of pH influences, E. coli was found toisplay excellent activity at pH 5.0–7.0. The obtained 60 min IC50 forCP was 8.0 mg L−1, which was more sensitive than other electro-hemical methods. Furthermore, the results showed that, exceptor accelerated bacterial respiration at low concentration of KCN,he IC50 values for As3+ and CN− were 18.3 and 4.9 mg L−1 with0 min incubation, respectively. Among the four metal ions, Hg2+

as found to be the most toxic, while Cu2+, Pb2+ and Ni2+ hado obvious toxicity on E. coli. It was attributed to certain negative

ons contained in samples, such as PO43−, CN− and OH−, which

educed heavy metal concentrations by complexation or precipi-ation. Therefore, an extension of this work in future needs to bessessed under different conditions of mediator, pH and biologicalystems. In general, the DTA described is a sensitive, rapid, repro-ucible and inexpensive alternative with no ethical argument forn-site toxicity of environmental water and wastewater samples.

cknowledgements

This project was supported by the State Plan for High-Techesearch and Development (2007AA061501), the National Naturalcience Foundation of China (No. 20820102037).

[

[

Acta 641 (2009) 59–63 63

References

[1] E. Fulladosa, J.C. Murat, M. Martínez, I. Villaescusa, Chemosphere 60 (2005)43–48.

[2] Y.V. Nancharaiah, M. Rajadirai, V.P. Venugopalan, Environ. Sci. Technol. 41(2007) 2617–2621.

[3] W. Drost, M. Matzke, T. Backhaus, Chemosphere 67 (2007) 36–43.[4] B. Burysková, K. Hilscherová, P. Babica, D. Vrsková, B. Marsálek, L. Bláha, Aquat.

Toxicol. 80 (2006) 346–354.[5] M.I. Arufe, J. Arellano, J.M. Moreno, C. Sarasquete, Ecotoxicol. Environ. Saf. 59

(2004) 209–216.[6] J.L. Levy, J.L. Stauber, D.F. Jolley, Sci. Total Environ. 387 (2007) 141–154.[7] A. Tizzard, J. Webber, R. Gooneratne, R. John, J. Hay, N. Pasco, Anal. Chim. Acta

522 (2004) 197–205.[8] E. Fulladosa, J. Debors, I. Villaescusa, J.C. Bollinger, J.C. Murat, Environ. Chem.

Lett. 5 (2007) 115–119.[9] M.D. Hernado, O. Malato, M. Farré, A.R. Fernandez, D. Barceló, Talanta 69 (2006)

370–376.10] A. Li, P.Y. Lee, B. Ho, J.L. Ding, T.C. Lim, Biochim. Biophys. Acta 1767 (2007)

411–418.[11] N. Pasco, J. Hay, J. Webber, Biomarkers 6 (2001) 83–89.12] M.R. Evans, G.M. Jordinson, D.M. Rawson, J.G. Rogerson, Pestic. Sci. 54 (1998)

447–452.13] M. Farré, O. Pasini, M.C. Alonso, M. Castillo, D. Barceló, Anal. Chim. Acta 426

(2001) 155–165.14] Y.C. Hsieh, H.M. Tsai, K.D. Ryan, C.O. Pancorbo, Sci. Total Environ. 320 (2004)

37–50.15] W. Liu, G.J. Jiang, G.Y. Shi, Y. He, Y. Liu, L.T. Jin, Chin. J. Chem. 25 (2007)

203–207.16] P. Ertl, E. Robello, F. Battaglini, S.R. Mikkelsen, Anal. Chem. 72 (2000) 4957–

4964.[17] J. Zhao, Z. Yang, M. Wang, Y. Lu, Z. Yang, J. Agric. Food Chem. 52 (2004) 7246–

7250.18] J. Zhao, Z. Yang, M. Wang, Y. Lu, Z.Y. Yang, Q.T. Gong, L.Y.Z. Yang, Biotechnol. Lett.

27 (2005) 207–211.19] J. Zhao, M. Wang, Z.Y. Yang, Z. Wang, H. Wang, Z. Yang, Anal. Chim. Acta 597

(2007) 67–74.20] N. Yoshida, K. Yano, T. Morita, S.J. Mcniven, H. Nakamura, I. Karube, Analyst 125

(2000) 2280–2284.21] N. Yoshida, J. Hoashi, T. Morita, S.J. Mcniven, K. Yano, A. Yoshida, H. Nakamura,

I. Karube, Analyst 126 (2001) 1751–1755.22] S.P. Trosok, B.T. Driscoll, J.T.H. Loung, Appl. Microbiol. Biotechnol. 56 (2001)

550–554.23] N. Yoshida, J. Hoashi, T. Morita, S.J. Mcniven, H. Nakamura, I. Karube, J. Biotech-

nol. 88 (2001) 269–275.24] H. Nakamura, K. Suzuki, H. Ishikuro, S. Kinoshita, R. Koizumi, S. Okuma, M.

Gotoh, I. Karube, Talanta 72 (2007) 210–216.25] N. Pasco, K. Baronian, C. Jefferies, J. Hay, Appl. Microbiol. Biotechnol. 53 (2000)

613–618.26] K. Catterall, K. Morris, C. Gladman, H. Zhao, N. Pasco, R. John, Talant 55 (2001)

1187–1194.27] Standard method for the examination of water and wastewater, 19th ed., Amer-

ican Public Health Association, American, 1997.28] K. Morris, K. Catterall, H. Zhao, N. Pasco, R. Johan, Anal. Chim. Acta 442 (2001)

129–139.

31] E. King, in: D. Liu, B.J. Dutka (Eds.), Toxicity Screening Procedures Using BacterialSystems, Marcel Dekker, New York, 1984.

32] G. Gellert, Ecotoxicol. Environ. Saf. 45 (2000) 87–91.