Sm

FE

a

ARRAA

KCEME

1

hoOLpns1(a

b[Iusoahde

0d

Analytica Chimica Acta 649 (2009) 123–127

Contents lists available at ScienceDirect

Analytica Chimica Acta

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

ensitive and rapid chemiluminescence enzyme immunoassay foricrocystin-LR in water samples

. Long ∗, H.C. Shi, M. He, J.W. Sheng, J.F. Wangnvironmental Simulate and Pollution Control State Key Joint Laboratory, Department of Environment Science and Engineering, Tsinghua University, Beijing 100084, China

r t i c l e i n f o

rticle history:eceived 20 May 2009eceived in revised form 6 July 2009ccepted 8 July 2009vailable online 14 July 2009

a b s t r a c t

A highly sensitive, specific, simple, and rapid chemiluminescence enzyme immunoassay (CLEIA) wasdeveloped for the determination of microcystin-LR (MC-LR). Several physicochemical parameters suchas the chemiluminescent assay mediums, the dilution ratio of MC-LR–OVA conjugate, monoclonal anti-body concentration, and peroxidase labeled antibody concentration were studied and optimized. Under

−1

eywords:hemiluminescencenzyme immunoassayicrocystin-LR

nvironmental analysis

optimum conditions, calibration curve obtained for MC-LR had detection limits of 0.032 ± 0.003 �g L ,the 50% inhibition concentration (IC50) was 0.20 ± 0.02 �g L−1 and the quantitative detection range was0.062–0.65 �g L−1. The proposed methods was successfully applied to the monitoring of MC-LR in spikedwater samples without significant effect of the matrix, and the recovery of MC-LR added to water samplesat different concentrations ranged from 80% to 115% with the coefficients of variation (CVs) less than 9%.The LOD attained from the calibration curves and the results obtained for the real samples demonstrate

as a

the potential use of CLEIA

. Introduction

Microcystins (MCs) are a group of closely related toxic cycliceptapeptides secreted by freshwater cyanobacteria [1,2]. Amongver 80 microcystin variants found from Microcystis, Anabaena,scillatoria (Planktothrix), Nostoc and Anabaenopsis, micorocystin-R (MC-LR, MW = 995.2) containing leucine (L) and arginine (R) inosition 2 and 4, respectively, was the first identified MC species andow known to be most toxic [3,4]. The development of rapid andensitive methods for the determination of MC-LR at levels below�g L−1, the provisional guideline of World Health Organization

WHO) [5], has become very important to protect both human andnimal from exposure to MCs.

The several sophisticated analytical techniques (e.g. inverte-rate bioassays [3], protein phosphatase inhibition assays [6], HPLC7], etc) have currently been developed for microcystins analysis.mmunoassays based on monoclonal or polyclonal antibodies areseful as screening techniques because of their specificity, highlyensitivity, easy-to-use and rapidity [8–10]. While the sensitivityf an immunoassay strongly depends on the affinity of specific

ntibodies and the sensitivity of the detection method. Due to theigh sensitivity, rapidity of reaction, simple instrumentation, wideynamic range and possible analysis of difficult matrices withoutxtensive pre-treatment, chemiluminescence enzyme immunoas-

∗ Corresponding author. Tel.: +86 10 62773095; fax: +86 10 62771472.E-mail address: longf04@mails.tsinghua.edu.cn (F. Long).

003-2670/$ – see front matter © 2009 Published by Elsevier B.V.oi:10.1016/j.aca.2009.07.026

screening tool for the analysis of MC-LR in environmental samples.© 2009 Published by Elsevier B.V.

say (CLEIA) has been widely applied in different fields such astoxicological analysis, pharmaceutical analysis, bioanalysis, clini-cal chemistry and environmental analysis [11–14]. The sensitivityof CLEIA can be further improved by using enhanced chemilumi-nescent reactions [15]. For example, p-iodophenol (PIP) is the mostpopular enhancer of luminal–H2O2–horseradish peroxidase (HRP)CL reaction and applied to a wide variety of immunoassays andbiotoxins analysis [13,16,17].

In this paper, an enhanced chemiluminescence enzymeimmunoassay was established based on a monoclonal antibody(Clone MC8C10) with highly specificity against MC-LR. The con-centration of the MC-LR–OVA conjugate immobilized onto themicroplate and the concentration of monoclonal antibody MC8C10were optimized. Cross-reactivity of the immunodetection methodto compounds structurally similar to MC-LR was discussed. TheCLEIA had also been applied to the detection of MC-LR spiked infour water samples of different origin from China, and the matrixeffect on the immunoassay response was demonstrated.

2. Experimental

2.1. Materials

Bovine serum albumin (BSA), ovalbumin (OVA), Tween-20,luminol, 2-mercaptoethylamine HCl, and 1-ethyl-3(3-dimethylamino-propyl)carbodiimide hydrochloride (EDC) werepurchased from Sigma–Aldrich (Steinheim, Germany). MC-LR wasobtained from Alexis (Läufelfingen, Switzerland). All the other

124 F. Long et al. / Analytica Chimica Acta 649 (2009) 123–127

F ay meb ncenc −1, res

rAttiIf

pikIaP(b

2

MbC

2

g(a

ig. 1. The chemiluminescent intensity obtained from the optimization of the asseing 6 and 1.2 mmol L−1, respectively; (b) different H2O2 concentration with the cooncentration with the concentrations of luminol and H2O2 being 0.8 and 6 mmol L

eagents, unless specified, were supplied by the Beijing Chemicalgents; these were also of analytical grade and used without fur-

her purification. Distilled deionized water was used throughouthe investigation. 1 mg mL−1 MC-LR stock solutions were preparedn 0.01 mol L−1 phosphate buffered saline (1 PBS) and stored at 4 ◦C.n addition, standard concentrations of the analyte were preparedrom the stock solution by serial dilutions in 0.01 mol L−1 PBS.

Monoclonal anti-MC-LR antibody (MC-LR-MAb, MC8C10) wasroduced by our research group [18]. The preparation of coat-

ng conjugate MC-LR–OVA was previously described [19] and wasept frozen in PBS at −20 ◦C enzyme tracer: goat anti-mousegG-peroxidase (Sigma A4416); coating solution: sodium carbon-te/bicarbonate buffer (0.1 mol L−1, pH 9.6), stored at 4 ◦C; dilution:BS (0.01 mol L−1, pH 7.4), stored at 4 ◦C; wash solution: PBST0.01 mol L−1, pH 7.4, 0.05% (v/v) Tween-20 in PBS), stored at 4 ◦C;locking solution: 0.5% OVA, dissolved with dilution solution.

.2. Instrumentation

Chemiluminescent intensity was measured with Modulusicroplate Fluorometer (Turner Biosystems, USA). Opaque high

inding plates for chemiluminescent measurements were fromostar (Cambridge, USA).

.3. Chemiluminescent immunoassay

The basic process of CLEIA stepped as follows: the coating anti-en MC-LR–OVA was diluted to a concentration of 0.25 mg L−1

if no specially stated) with 0.1 mol L−1 pH 9.6 sodium carbon-te/bicarbonate buffer, and 100 �L of the solution were added to

dium: (a) varying luminol concentration with the concentrations of H2O2 and PIPtrations of luminol and PIP being 0.8 and 1.2 mmol L−1, respectively; (c) varying PIPpectively. Measurements were done in triplicate.

each well of a 96-well microtiter plate. The plate was covered andincubated overnight (>12 h). The wells were emptied, washed threetimes with PBST and blotted dry. Each well was blocked with 120 �L0.5% (w/v) OVA in PBS for 1 h. The wells were subsequently emptiedand washed again five times with PBST. 50 �L well−1 of MC-LR stan-dard solution in PBS or sample solution, followed by 50 �L well−1 ofMC-LR MAbs at the dilution ratio 1:16,000 (if not specially stated)were added. Inhibition standard curves were prepared with theconcentration of standard MC-LR ranged from 30 to 0 �g L−1. Thecompetitive reaction was allowed to take place for 50 min. Afterwashing, the peroxidase-labeled goat anti-mouse immunoglobu-lins at the dilution ratio 1:10,000 (if not specially stated) was added,and plates were incubated for 40 min. Plates were then washedand finally peroxidase activity was revealed by adding 50 �L well−1

of a freshly prepared substrate mixture (0.8 mmol L−1 luminol,1.2 mmol L−1 PIP, 6.0 mmol L−1 H2O2 in 0.2 mol L−1 borate buffer,pH 8.5). Chemiluminescent emission of each well was measuredafter the addition of the substrate for 30 s. All incubations wereperformed at room temperature.

2.4. Spiked water sample analysis

Water samples were fortified with MC-LR to evaluate potentialmatrix effects on CLEIA. Mean chemiluminescent intensity valuesobtained from triplicate wells were interpolated in a standard curve

run in the same plate. The water samples tested were the labora-tory tap water and three lake water samples collected from LakesTai (Jiangsu), Zao (Anhui), Beihai (Beijing), respectively. These sam-ples were spiked with MC-LR from the MC-LR stock solution atconcentrations of 0, 0.01, 0.05, 0.1, 0.5, 1, 8, and 30 �g L−1.

F. Long et al. / Analytica Chimica Acta 649 (2009) 123–127 125

Table 1The effect of MC-LR-OVA conjugate concentration on the immunoassay performances.

Dilution ratio LOD (�g L−1) IC50 (�g L−1) Linear response range (�g L−1)

1:8000 0.37 ± 0.02 0.82 ± 0.05 0.51–1.041:10,000 0.21 ± 0.02 0.59 ± 0.03 0.32–1.01

3

3

nwlccratr

3

ipa

3

mfiscms0saesi

Ft

the standard deviation of all the data points within 6%.The dynamic MC-LR detection range is described to exhibit

1:12,000 0.095 ± 0.0151:16,000 0.032 ± 0.0031:20,000 0.032 ± 0.0021:24,000 0.041 ± 0.005

. Results and discussion

.1. Optimization of chemiluminescent assay medium

To obtain maximized sensitivity enhancement in chemilumi-escent detection, the concentrations of luminol, PIP and H2O2ere firstly optimized by examining their effects on the chemi-

uminescent intensity (CLI). The PIP can obviously enhance thehemiluminescent strength of luminal associated with the HRP-atalyzed oxidation [20,21]. In Fig. 1, the chemiluminescentesponses as functions of varying concentrations of luminol, H2O2nd PIP were described. The results stated that the optimal concen-rations of luminol, H2O2 and PIP were 0.8, 6.0, and 1.2 mmol L−1,espectively.

.2. Optimization of the immunoassay procedure

The factors affecting the immunoassay procedure included coat-ng conjugate concentration, monoclonal antibody concentration,eroxidase labelled antibody concentration, buffer ionic strengthnd pH, etc.

.2.1. Optimization of coating conjugateIn environmental analysis, targets interest are usually small

olecule substances (molecular weight < 1000), and it is greatly dif-cult to directly immobilize them onto the bio-recognition sensingurface. Herein, competitive assays were performed using differentoncentrations of coating conjugate MC-LR–OVA as bio-recognitionolecules. Briefly, coating antigen MC-LR–OVA was diluted with

odium carbonate/bicarbonate buffer at concentrations of 0.10,.15, 0.25, 0.50, and 0.75 mg L−1, respectively. Then, 100 �L of the

olution of different concentrations were added to each well of

96-well microtiter plate, respectively. Thereafter, assays werexecuted as described above. In Fig. 2, the chemiluminescent inten-ity was plotted against the concentration of MC-LR–OVA. Threendependent experiments were carried out for each data points

ig. 2. Chemiluminescent intensity at MC-LR–OVA conjugate of various concentra-ions.

0.35 ± 0.02 0.15–0.820.20 ± 0.02 0.062–0.650.17 ± 0.03 0.059–0.520.21 ± 0.02 0.075–0.58

and demonstrated good reproducibility with a deviation of <8%(n = 3). As seen in Fig. 2, the chemiluminescent intensity rapidlyincreased with increasing the concentration of the conjugate whenthe concentration of conjugate was less than 0.25 mg L−1. Andover this concentration, the chemiluminescent intensity increasesonly a very little and reached a plateau. Therefore, 0.25 mg L−1

MC-LR–OVA conjugate were selected to immobilize on 96-wellmicrotiter plate in further experiments.

3.2.2. Optimization of monoclonal antibodyThe concentration of monoclonal antibody is one of the key

factors determining the sensitivity and working range of animmunoassay. For optimizing monoclonal antibody concentra-tion, MC8C10 solution was diluted with a series of dilution ratioof 1:8000, 1:10,000, 1:12,000, 1:16,000, 1:20,000, and 1:24,000,respectively. Fig. 3 showed the effect of monoclonal antibodyconcentration on standard assay curves, in which normalizedchemiluminescent signal, by expressing the signal of each standardpoint as the ratio of the maximum response [CLI/CLImax], were plot-ted against the logarithm of the concentration of MC-LR through afour-parameter logistic model as follows:

y = A1 − A2

1 + ([Ag]/[Ag0])p + A2 (1)

where [Ag] is the analyte concentration; A1, A2 are the upper andlower asymptote (background signal) to the titration curve; [Ag0]is the analyte concentration at inflection; and p is the slope at theinflection point. The error bars in the figure correspond to the stan-dard deviation of the data points in three-fold experiments, with

20–80% inhibition, and the limit of detection (LOD) for MC-LR wascalculated from the calibration curve, as the analyte concentrationproviding a 10% decrease of the blank signal. The dynamic MC-LR

Fig. 3. Variation of normalized chemiluminescent intensity with the concentra-tion of MC-LR in competitive immunoassay experiments at different MC-LR-MAbconcentrations.

126 F. Long et al. / Analytica Chimica Acta 649 (2009) 123–127

F MC-LRc

dsAttirLmft

3

rrtwtdoaw

3

oie

F

ig. 4. Variation of normalized chemiluminescence signal with the concentration ofoncentrations.

etection range and the LOD of the chemiluminescent immunoas-ays calculated from the calibration curve were shown in Table 1.lthough increasing the antibody concentration could increase

he chemiluminescent signal, decreasing the antibody concentra-ion could lead to decreasing the limit of detection and expandmmunoassay working range. On the other hand, when the dilutionatio was higher than 1:16,000, the linear response range and theOD rarely changed, and chemiluminescent signal decreased toouch and might lead to inaccuracy of the detection results. There-

ore, dilution ratio of 1:16,000 was selected, since it correspond tohe widest linear range and the highest sensitivity.

.2.3. Optimization of peroxidase labelled antibodyPeroxidase labelled antibody was first serially diluted to the

atio of 1:4000, 1:8000, 1:10,000, 1:12,000, 1:16,000 and 1:20,000,espectively. Then, 100 �L of the antibody solution at each dilu-ion ratio was applied to the assay. In Fig. 4a, the results statedhen the dilution ratios of peroxidase labelled antibody decreased,

he chemiluminescent signal increased. As seen in Fig. 4b, theynamic MC-LR detection range and the limit of detection (LOD)f the CLELISA rarely varied with the change of peroxidase labelledntibody concentration. Thus, 100 �L peroxidase labelled antibodyith 1:10,000 dilution ratio was selected in order to save reagents.

.2.4. Buffer ionic strength and pHBecause the binding of antigen and antibody depends mainly

n van der Waals forces and hydrophobic interactions, a changen either ionic strength or pH could affect this interaction [22]. Tovaluate the influence of ionic strength, calibration curves were

ig. 5. The influence of PBS concentration and pH on the CLImax/IC50 ratio for chemilunesce

in competitive immunoassay experiments at different peroxidase labelled antibody

performed with buffers of different ionic strength but at constantpH (7.5). Standard analyte curves were prepared with a constantconcentration of MAbs added to different concentration buffers.Thereafter, assays were executed as previously described.

It was observed that by increasing the ionic strength, the recog-nition of the conjugated hapten (CLImax) diminished, while therecognition of MC-LR (IC50) was improved. As shown in Fig. 5a, theCLImax/IC50 ratio increased with salt concentrations up to 1 PBSand reached a plateau, and then decreased. Therefore, 1 PBS wasselected for chemilunescent immunoassay. Due to buffer solutionsmay reduce the effect of other matrice (e.g. pH, and salt and so on)on the immunoassay, different concentrations of PBS may be usedto prepare the samples and/or antibody solution in the assay of thereal water samples.

In an immunoassay, the pH of solutions obviously do not affectonly the stability and biological activity of antibodies, but also thebinding efficiency between antibody and antigen, which may leadto the detection errors [23], if the effect of the pH is not taken intoaccount. To evaluate the influence of pH, standard analyte curveswere prepared with a constant concentration of MAbs in 1 PBS atdifferent pH values covering the range 3–11. Changes in pH haddifferent effects on the IC50 values and CLImax of the chemilunes-cent immunoassay. Between pH 6.5 and 7.5 the immunoassay wasmore sensitive, and the IC50 value and CLImax value of the MC-LR

assay did not change markedly. Fig. 5b shows the dependence of theCLImax/IC50 ratio on pH. This ratio reached a maximum at pH 6.5–7.5for MC-LR. These results showed that pH value of analyte solutionclose to neutral environment was the most favourable for the MC-LR chemilunescent immunoassay, and a pH of 7.5 was chosen as

nt immunoassay. Data were obtained from standard curves performed in triplicate.

F. Long et al. / Analytica Chimica

FP

oh(w(

3

iiisEacYb

3

afwcscooawo0t1h

iIcwttt

[

[

[[[

[

[[[

[

[

ig. 6. Standard calibration curves of MC-LR in different water types: 0.01 mol L−1

BS, tap water, Fuhai Lake water and Beihai Lake water.

ptimum. When the pH of the real water samples tested was tooigh or too low, we used a higher concentration of buffer solutione.g. 5× PBS or higher) to prepare samples and antibody solution,hich could effectively eliminate the effect of pH on immunoassay

data not shown).

.3. Cross-reactivity

Cross-reactivity describes the specificity of antibodies and is anmportant analytical parameter regarding specificity and reliabil-ty of immunoassays [24]. The specificity of the chemilunescentmmunoassay was evaluated using a wide range of compoundstructurally related to MC-LR. As previously shown using an ic-LISA [18] and a portable trace organic pollutant analyzer [19],ntibody MC8C10 exhibited high specificity for MC-LR, and theross-reactivity values of all the different interferences (e.g. MC-R, MC-RR, MC-LW, MC-LF, and Nodularin) were very small and fellelow 10% (data not shown).

.4. Water sample analysis

The evaluation of matrix effects is an important aspect in thessessment of newly developed methods [22]. The analytical per-ormance of CLEIA is commonly assessed by spiking matrix samples

ith the target analyte. Standard curves were generated in tripli-ate in the spiked water samples and in the 0.01 mol L−1 PBS bufferamples used as control to elucidate whether the calibration curveonstructed with standard solutions could be used in the analysisf real samples. As shown in Fig. 6, parallel calibration curves werebtained irrespective of the nature of the water sample. Addition-lly, the IC50 values obtained for MC-LR in 0.01 mol L−1 PBS, tapater, Lake Tai, Lake Zao, and Lake Beihai water were in the range

f 0.19–0.23 �g L−1, with the corresponding LODs of 0.032, 0.036,.041, and 0.052 �g L−1, respectively. The recovery of MC-LR addedo water samples at different concentrations ranged from 80% to15%. All coefficients of variation (CVs) were acceptable, never beingigher than 9%.

It should be noted that compared to the values obtained forc-ELISA [18] with the same monoclonal antibody, the LOD andC50 of chemilunescent immunoassay, by introducing an enhanced

hemiluminescence reaction as the end-point detection system,ere lower by a factor of 3 and 9, respectively. The decrease of

he LOD and IC50 value for MC-LR was a direct consequence ofhe decrease in the monoclonal antibody concentration and ofhe accurate optimization of assay parameters, because the dilu-

[[

[

Acta 649 (2009) 123–127 127

tion ratios of monoclonal antibody concentration for ic-ELISA andchemilunescence immunoassay were 1:6000 and 1:16,000, respec-tively. The sensitivity values (LODs) obtained by CLEIA is greatlylower than that of chemiluminescence immunosensor developedby other group [25], and is also comparable with other immunoas-say techniques such as electrochemical impedance [10], ELISA [8],TOPA [19], etc. Therefore, the CLEIA presented here could effec-tively be employed for the analysis of MC-LR in environmentalsamples. To the best of our knowledge, this is the first time sucha high sensitivity is reported for the detection of MC-LR employingchemiluminescent immunoassay.

4. Conclusions

Chemiluminescent immunoassay appears to be an effective ana-lytical technique for use in the monitoring of biotoxins due to itshigh sensitivity and ease-to-use. We developed and optimized ahighly sensitive CLEIA for MC-LR based on monoclonal antibody.The sensitivity of the proposed method was three-fold higher thanthat of the ic-ELISA and the consumption of immunoreagents waslower. Optimized CLEIA was also applied to detecting different realsamples. MC-LR in tap water and lake waters could be quantita-tively analyzed directly without the need for sample pre-treatment.Therefore, the CLEIA proposed here has great potential as a screen-ing tool for the analysis of MC-LR in environmental samples.

Acknowledgment

This project is supported by special fund of State Key JointLaboratory of Environment Simulation and Pollution Control(09Z01ESPCT) and Postdoctor Science Foundation (20080440035).

References

[1] E.M. Jochimsen, W.W. Carmichael, J.S. An, D.M. Cardo, S.T. Cookson, C.E. Holmes,M.B. Antunes, D.A. Melo Filho, T.M. Lyra, V.S. Barreto, S.M. Azevedo, W.R. Jarvis,N. Engl. J. Med. 338 (1998) 873–878.

[2] W.W. Carmichael, S.M. Azevedo, J.S. An, R.J. Molica, E.M. Jochimsen, S. Lau,K.L. Rinehart, G.R. Shaw, G.K. Eaglesham, Environ. Health Perspect. 109 (2001)663–668.

[3] D.L. Campbell, L.A. Lawton, K.A. Beattie, G.A. Codd, Environ. Toxicol. Water Qual.9 (1994) 71–77.

[4] J. McElhiney, L.A. Lawon, Toxicol. Appl. Pharmacol. 203 (2005) 219–230.[5] WHO, Guidelines for Drinking Water Quality, 3rd ed., World Health Organiza-

tion, Geneva, Switzerland, 2004.[6] C. Rivasseau, P. Racaud, A. Deguin, M.C. Hennion, Anal. Chim. Acta 394 (1999)

243–257.[7] C. Rivasseau, S. Martins, M.C. Hennion, J. Chromatogr. A 799 (1998) 155–169.[8] P. Lindner, R. Molz, E. Yacoub-George, A. Dürkop, H. Wolf, Anal. Chim. Acta 521

(2004) 37–44.[9] J.W. Sheng, M. He, H.C. Shi, Y. Qian, Anal. Chim. Acta 572 (2006) 309–315.10] M. Campàs, J. Marty, Biosens. Bioelectron. 22 (2007) 1034–1040.

[11] N. Kuroda, R. Shimoda, M. Wada, K. Nakashima, Anal. Chim. Acta 403 (2000)131–136.

12] T. Konry, A. Novoa, Y. Shemer-Avni, N. Hanuka, S. Cosnier, A. Lepellec, R.S. Marks,Anal. Chem. 77 (2005) 1771–1779.

13] Y. Zhou, Y. Zhang, C. Lau, J. Lu, Anal. Chem. 78 (2006) 5920–5924.14] J. Elbaz, B. Shlyahovsky, I. Willner, Chem. Commun. 13 (2008) 1569–1571.15] K.C. Ahn, P. Lohstroh, S.J. Gee, N.A. Gee, B. Lasley, B.D. Hammock, Anal. Chem.

79 (2007) 8883–8890.16] M.Y. Rubtsova, C. Wittman, Food Agric. Immunol. 9 (1997) 235–247.

[17] A.B. Dzgoev, I.G. Gazaryan, L.M. Lagrimini, K. Ramanathan, B. Danielsson, Anal.Chem. 71 (1999) 5258–5261.

18] J.W. Sheng, M. He, H.C. Shi, Anal. Chim. Acta 603 (2007) 111–118.19] F. Long, M. He, A.N. Zhu, H.C. Shi, Biosens. Bioelectron. 23 (2009) 2346–2351.20] P.M. Easton, A.C. Simmonds, A. Rakishev, A.M. Egorov, L.P. Candeias, J. Am. Chem.

Soc. 118 (1996) 6619–6624.21] Y.L. Kapeluich, M.Y. Rubtsova, A.M. Egorov, J. Biolumin. Chemilumin. 12 (1998)

299–308.22] S. Rodriguez-Mozaz, S. Reder, M. Lopez de Alda, G. Gauglitz, D. Barceló, Biosens.

Bioelectron. 19 (2004) 633–640.23] M.T. Matikainen, J. Immunol. Methods 75 (1984) 211–216.24] T. Kawaguchi, D.R. Shankaran, S.J. Kima, K.V. Gobi, K. Matsumoto, K. Toko, N.

Miura, Talanta 72 (2007) 554–560.25] P. Linder, R. Molz, E. Yacoub-George, H. Wolf, Anal. Chim. Acta 636 (2009)

218–223.