Monitoring of mutagenic process with piezoelectric quartz crystal impedance analysis

Talanta 53 (2000) 525 – 533 www.elsevier.com/locate/talanta

Monitoring of mutagenic process with piezoelectric quartz crystal impedance analysis Jinzhong Zhang a,1, Wanzhi Wei a,*, Anhong Zhou a, Deliang He a, Shouzhuo Yao b, Qingji Xie b a

College of Chemistry and Chemical Engineering, Hunan Uni6ersity, Hunan, Changsha 410082, People’s Republic of China b Chemical Research Institute, Hunan Normal Uni6ersity, Hunan, Changsha 410081, People’s Republic of China Received 31 March 2000; received in revised form 18 July 2000; accepted 21 July 2000

Abstract A novel method for monitoring of mutagenic process of dimethyl sulfate to Salmonella typhimurium strain (TA100) was proposed by using piezoelectric quartz crystal impedance (PQCI) analysis technique. The time courses of responses piezoelectric impedance parameters for a quartz crystal in a culture system were simultaneously obtained and discussed. It was found that the motional resistance variation (DRm) increases and frequency shift (Df ) of PQC sensor decreases correspondingly during the mutagenic process of the bacteria. These parameters could reflect the variations of viscosity and density of culture system. By fitting DRm versus time curves toward Gompertz bacterial growth model, we obtained and discussed the bacterial growth parameters for both normal growth and mutagenic process. The experiments showed that the proposed method could provide real time and multidimensional impedance information to the monitoring of mutagenic process. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Mutagenic process; Dimethyl sulfate; Salmonella typhimurium; Piezoelectric quartz crystal impedance

1. Introduction Chemical factor in environment becomes an increasingly important influencing factor on public health. A great number of researchers found that some chemical components in airborne par-

* Corresponding author. Tel.: + 86-731-8823359; fax: +86731-8824525. E-mail address: [email protected] (W. Wei). 1 On leave from Southwest Agricultural University, Chongqing, China.

ticulate matter [1,2], contaminated soil [3,4] or drinking water [5,6] can be mutagenic. Ames et al. [7] indicated that about 90% of carcinogens have mutagenicity after they had studied some known chemical carcinogens. For this reason, it is possible to predict the carcinogenic activity of chemical pollutants by a mutagenicity test, which is of great significance in environmental medicine. In order to protect public health, many researchers have tried to develop mutagenicity test methods. Some methods with microorganisms have been proposed, such as Ames Salmonella/mammalian microsome test (Ames test) [8,9], microbial biolu-

0039-9140/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 9 1 4 0 ( 0 0 ) 0 0 5 3 0 - 0

526

J. Zhang et al. / Talanta 53 (2000) 525–533

minescence method (Mutatox test) [10], microdose forward mutagenic method [11] and microbial sensor methods [12,13]. But these methods only regard the study of mutagenic results, and can not provide the information of mutagenic process. To reveal the mutagenic mechanism, monitoring of mutagenic process is more important than mutagenicity test. To our best knowledge, monitoring of mutagenic process has not been reported. This paper is aimed at performing this aspect work. Piezoelectric quartz crystal impedance (PQCI) analysis is a practical method for studying the quartz crystal resonance and provides multidimensional information reflecting some physical and chemical properties of the investigated system [14 – 16]. For a lossless quartz crystal, PQCI analysis has been based on the Butterworth – Van Dyke (BVD) equivalent electrical circuit model (Fig. 1), which composed of a motional arm and static arm in parallel. The motional arm contains three equivalent circuit elements in series, namely, motional resistance (Rm), motional inductance (Lm) and motional capacitance (Cm), while the static arm only contains the static capacitance (C0), and all the four equivalent circuit parameters are of distinct physical meanings [14 – 18]. PQCI technique has been successfully applied to the field of life sciences, such as study of hemorheological characteristics [19,20], the detection of microbial growth and metabolism [21,22] and enzyme activity [23]. For monitoring bacterial growth, direct and indirect methods exist that were reviewed by Cooney [24], such as cell number estimation, measurement of biomass, optical density [25] and viscosity [26], and application of biosensors [27]. In this study, the growth situations of the bacteria were monitored by PQCI technique in the presence and absence of histidine. The study was based on the fact that Salmonella typhimurium auxotroph strain (TA100) could not grow without histidine. If mutagen was added into the culture medium of the strain, the strain could happen to mutate and revert to its ancestral type, and the physical and chemical properties of culture system would change accordingly due to the bacterial growth,

this causes the variations of equivalent circuit parameters of PQCI. The changing process of impedance parameters could reflect the mutagenic process. Dimethyl sulfate was used as a mutagen in this work, as it is a typical alkyl carcinogen [28].

2. Experimental

2.1. Reagents The composition of the medium for S. typhimurium TA100 was as follows, glucose, 20.0 g; citric acid, 2.0 g; K2HPO4 · 3H2O, 3.5 g; MgSO4 · 7H2O, 0.2 g; distilled water, 1000 ml. The medium was sterilized by autoclaving at 121°C for 15 min. The concentration of dimethyl sulfate solution was 50 mg l − 1. The buffer solution used was phosphate buffer solution (pH 7.2). All reagents used were of analytical grade except histidine as a biochemical reagent. Doubly distilled and sterilized water was used throughout.

2.2. Materials AT-cut 9 MHz piezoelectric quartz crystals (12.5 mm in diameter) were purchased from Staterun 707 factory (Beijing, China). The gold electrodes of the crystals were vacuum-evaporated by using an Eiko IB-3 ion coater and a highly pure gold foil purchased from Hitachi Inc., and a nearly equal apparent electrode area of 0.3 mm2 was obtained. The gold-coated quartz crystals were sealed with 704 silicon rubber to obtain one

Fig. 1. Electrical equivalent circuit for the PQC sensor. C0 is the static capacitance; Lm, Cm and Rm are the motional inductance, capacitance and resistance, respectively.

J. Zhang et al. / Talanta 53 (2000) 525–533

Fig. 2. Schematic representation of the experimental set-up for the PQCI measurements. (1) Personal computer. (2) HP4395A +16092 fixture. (3) PQC sensor. (4) Detection cell. (5) Biochemical incubator. (6) Gold electrode. (7) quartz crystal.

side of the crystal in contact with liquid. To remove possible surface adsorbate in air, the gold electrode surfaces were treated with H2SO4 + H2O2 (v/v 3:1) for 5 min, thoroughly rinsed with water and then cleaned with acetone- and water-wetted filter paper, successively. Before use, the electrodes were sterilized by autoclaving at 121°C for 15 min, then preserved in a refrigerator at 4°C.

2.3. Microorganism S. typhimurium TA100 strain was obtained commercially (Beijing, China). Four loops of TA100 on agar slant were inoculated into 100 ml of broth medium (pH 7.2), and were incubated at 37°C for 16 h under the condition providing oxygen. Then the mixture was preserved in the refrigerator at 4°C.

527

Rm, f0, 1/Cm and C0 were used as estimation parameters during the fitting, where the resonant frequency f0 is defined as f0 = 1/[2p(LmCm)1/2] [14]. The equivalent circuit parameters were obtained at a time interval of ca. 4.5 s, and displayed on the VB form during experiments. Admittance measurements were under conditions of 101 points, a frequency span of 40 kHz covering the resonant frequency of the PQC, IF BW of 10 kHz, source power of 0.5 dBm. HP 4395A was controlled by a PII 350 personal computer (PC). A biochemical incubator (Model SPX-250, Yuejin medical instrumental factory, Shanghai, China) was used to incubate the bacterial solution.

2.5. Procedures The culture solution was prepared by mixed with 20 ml of culture medium, 0.1 ml of dimethyl sulfate solution, 10 ml of TA100 bacterial solution and oxygen-saturated distilled water to 50 ml as a final volume, and then gold electrode was immersed. The detection cell was stuffed with a rubber plug and placed to incubate at 379 0.1°C in the biochemical incubator. Then the variations of impedance parameters were real time monitored by HP 4395A Impedance Analyzer.

3. Results and discussion

2.4. Instrumentation and data acquirement

3.1. Principle of the PQCI analysis

The experimental setup for the PQCI measurements is shown in Fig. 2. PQCI system comprises of a HP 4395A network/spectrum/impedance analyzer, a HP 43961A RF impedance test adapter, and a 16092A was directly connected to the terminal contacting liquid of the PQC. A user program was written in Visual Basic (VB) 5.0 to control the HP 4395A and to acquire admittance data (conductance G and susceptance B) via a HP 82341C high-performance HP-IB interface card for Windows 3.1/NT/95. Real time analysis of the admittance data based on the simultaneous nonlinear fitting of both G and B data to the BVD model were also achieved using the same VB program.

The equivalent circuit of the PQC sensor has been derived by Cady and Bottom [16], and as shown in Fig. 1. The PQC sensor responds to an applied voltage or current in the same way as the crystal itself. The four circuit elements with parameters Rm, Lm, Cm and C0 correspond to one mode of vibration of the quartz crystal. In the PQCI analysis, the spectra of conductance and susceptance were measured while the frequency span covered the complete resonant region. Then the equivalent circuit components can be calculated. The circuit elements discussed in this study are related to the properties of the quartz crystal are the following [29]:

J. Zhang et al. / Talanta 53 (2000) 525–533

528

Rm =

t 3Qr 8So 2

(1)

2

8So p 2tQc

(2)

ko0S tQ

(3)

Cm = C0 =

where tQ and S are the thickness and area of the quartz plate, respectively, o is the piezoelectric stress constant, r the dissipation coefficient, c the elasticity constant, k the dielectric constant of quartz and o0 is the permittivity of free space. So every circuit element has its corresponding physical significance. Rm corresponds to the loss in mechanical energy dissipated to the surrounding medium and the supporting structures. Cm corresponds to the mechanical elasticity of the crystal and surrounding medium. C0 originates from the two parallel plate metal electrodes on the quartz surfaces and stray capacitance to the supporting structure. Rm and Cm are related to the vibration caused by the piezoelectric effect while C0 is not related to the piezoelectric effect. In the PQC analysis, Sauerbrey equation [30] displays a linear relationship between the mass change (Dm) and the frequency shift (Dfm): 2f 20 Dm Dfm = 1/2 (rQmQ) S

(4)

where mQ and rQ are the shear modulus and density of quartz, respectively. In fact, the physical and chemical properties of the liquid may also cause the frequency to change [17]. A study by Kanazawa et al. [31] indicated that the effect of the viscosity (hL) and density (rL) of the liquid on the resonant frequency as follows: DfL = −

1/2 f 3/2 0 (rLhL) 1/2 (prQmQ)

(5)

Therefore, the total frequency shift (Dftot) is the sum of Dfm and DfL: DG oP = − 34 300− 21.5 × T (J mol − 1)

(6)

In the PQCI analysis, Muramatsu et al. [14] pointed out that the motional resistance (Rm) is related linearly to the (rLhL)1/2 of the liquid:

Rm =

(2pf0rLhL)1/2S k2

(7)

where k is the electromechanical coupling factor. If the total frequency shift (Dftot) is dominated by the net changes in solution viscosity and density, the relationship between DRm and DfL can be obtained from Martin’s equations [15] and it has been supported experimentally [32]: DfL 1 =− 4pLQ DRm

(8)

where LQ is the motional inductance for a quartz crystal in vacuum (or in air). As for a present 9 MHz crystal, LQ measured is ca. 7.86 mH. Therefore, the slope of DfL versus DRm is calculated to be 10.1 Hz·V − 1. In other word, the change in liquid viscosity and density equivalent to 1 V of DRm is able to cause 10.1 Hz of DfL. As for a practical system, Df/DRm may be a greater value because Df measured may contain the contributions of both mass and viscosity–density factors.

3.2. Monitoring of the growth of the auxotroph strain The typical response curves of PQCI parameters (Df, Rm, Cm and C0) during the growth of the auxotroph strain are shown in Fig. 3. It can be seen that Rm keeps constant value basically, and Df is ca. 180 Hz after 16 h’s incubation. The variations of the two parameters suggest that the viscosity and density of the culture system show no apparent changes in the incubation process. In other word, the auxotroph strain can not grow under these experimental conditions. The change in frequency shift is very small, and may be caused by the adsorption of the culture components onto the gold electrode. It can be found from Fig. 3 that both Cm and C0 increase slowly during the culture process. These phenomena indicate that both mechanical elasticity of the crystal and surrounding medium and dielectric constant of the quartz increase little. It should be pointed out the changes of Cm and C0 are related to bacterial growth process, similar to that in enzymatic reaction [33].

J. Zhang et al. / Talanta 53 (2000) 525–533

529

3.3. Monitoring of the normal growth of the bacteria The typical response curves of PQCI parameters (Df, DRm, Cm and C0) during the normal growth of the bacteria are shown in Fig. 4. It is observed that DRm appears a constant value in the initial 2 h, then increases greatly and finally reaches a stable level. Compared with that of DRm, Df shows a reverse changing trend. As shown in Eqs. (5) and (7), the changes in two parameters indicate viscosity and density variations of culture system during the normal growth. It is interesting to note that the shape of DRm curve resembles the theoretical growth curve. The section of DRm curve in the initial 2 h corresponds to the lag period of the strain, that of from 2 to ca.10 h corresponds to the exponential growth

Fig. 4. Time courses of simultaneous responses of DRm, Df, Cm and C0 during the normal growth of the bacteria (the volume of the bacterial solution with the addition of histidine is 10 ml, and the volume of the medium is 20 ml).

Fig. 3. Time courses of simultaneous responses of Rm, Df, Cm and C0 during the growth of the auxotroph strain (the volume of the bacterial solution is 10 ml, and the volume of the medium is 20 ml).

period, and after 10 h’s incubation, the bacterial growth goes to saturation. From Fig. 4, a decrease of ca. 985 Hz in Df and increase of ca. 36.5 V in DRm from 2 to ca. 12 h can be obtained. The slope of Df versus DRm can be calculated to be ca. 27.0 Hz·V − 1, this value is more than the theoretical value 10.1. This larger slope shows that the contributions to Df result not only from the increase in viscosity and density, but also from other factors, e.g. the increase in viscoelasticity of the crystal surface and the adsorption of the culture components onto the gold electrode. However, DRm reflects well the increase in viscosity and density of culture system according to Eq. (7). It can also be found from Fig. 4 that Cm and C0 increase by 3.8 fF and 1.2 pF, respectively. These results illustrate that both mechanical elasticity

530

J. Zhang et al. / Talanta 53 (2000) 525–533

and dielectric constant of the crystal increase during the normal growth of the strain. Attention should be paid to response curves that the variations of two capacitances show ahead trends than that of DRm and Df.

3.4. Monitoring of the mutagenic process The typical response curves of PQCI parameters (Df, DRm, Cm and C0) during the mutagenic process are shown in Fig. 5. It can be seen that DRm and Df curves both appear a platform during the first 8 h, then DRm increases and Df decreases greatly. As shown in Fig. 5, a decrease of ca. 474 Hz in Df and increase of ca. 13.5 V in DRm from 8 to 16 h can be obtained. The slope of Df versus DRm can be calculated to be ca. 35.1 Hz·V − 1. Comparing with Fig. 4, the results in Fig. 5 display that the similar changing trends of the two parameters can be observed. Thus, it can be

Fig. 5. Time courses of simultaneous responses of DRm, Df, Cm and C0 during the mutagenic process (the volume of the bacterial solution is 10 ml, the volume of the medium is 20 ml, and the concentration of dimethyl sulfate is 1 mg ml − 1).

demonstrated viscosity and density of culture system increase after 8 h’s incubation. It shows that the strain happens to mutate and regains its growth capability owing to the impact of dimethyl sulfate in the medium, so the strain shows similar growth situation with the addition of histidine. Furthermore, one could investigate the responses of DRm and Df in the two figures from three aspects. First, the times of great change in the two parameters are different, so the lag times expressed by the bacteria are different. It testifies that the auxotroph strain needs a period of time to regain its growth capability. In this way, bacterial growth caused by mutation is slower than normal growth. Secondly, the signal sizes of responses are different, DRm and Df in Fig. 4 are greater than those in Fig. 5. This may be that the number of induced revertants produced by mutation is less than that of normal growth bacteria. Thus the amount of metabolites in mutagenic case is less than that of normal growth, and the change in viscosity and density is lower than that of normal growth. Thirdly, the slopes of Df versus DRm are different in the time scope examined, Df/DRm in Fig. 5 is greater than that in Fig. 4. Although Df and DRm are both lower than that in Fig. 4, the ratio of Df to DRm is greater than that in Fig. 4. Attention should be paid to the fact that the differences between the growth in the presence of histidine (in Fig. 4) and that in the presence of mutagen (in Fig. 5) always exist. We observed these differences through multiple experiments. Comparing with Fig. 3, DRm and Df show evident variations in Fig. 5. It illustrates that mutation causes bacterial growth, and the variations of two parameters reflect the increases in viscosity and density of culture system. From Fig. 5, it can also be found that Cm increases after keeping constant in the first 5 h, the increase value is ca. 1.9 fF; and C0 increases during the whole culture process, the increase value is ca. 0.9 pF. The results illustrate that both mechanical elasticity and dielectric constant of the quartz crystal increase during the mutagenic process.

3.5. The parameters e6aluation of bacterial growth To examine the bacterial growth situation, we detected bacterial numbers in the course of the

J. Zhang et al. / Talanta 53 (2000) 525–533

531

impedance measurements at different culture time. The results showed that a linear relationship between DRm and ln N/N0 (the logarithm of the relative population size) could be obtained: DRm =K ln

N N0

(9)

where K is a coefficient in V. In general, a sigmoidal growth curve can be described according to Gompertz model [34], which is written as: ln

!



mme N = A exp −exp (l− t) +1 A N0

n"

(10)

where e is the natural logarithm (2.7183); t in h is culture time. Parameters A, mm and l have their own biological significance [34]: A (no dimension), mm in h − 1 and l in h are the asymptote, maximum specific growth rate and lag time, respectively. Moreover, from Eqs. (9) and (10), the relationship between DRm responses and the bacterial growth parameters can be described:

!  ! 

n" n"

m me (l −t) +1 A m e =A%exp −exp m (l− t) +1 A

DRm =KAexp −exp

(11)

where A% in V is the maximum response value of DRm. By taking A%, A, mm and l as estimation parameters, we fitted the DRm responses using the nonlinear fitting program embedded in Sigmaplot® Scientific Graphing Software Version 2.0, and the relative sum of residual square (qr) is used to reflect the validity of the fitting, defined as:

Fig. 6. Time courses of DRm values experimentally obtained and fitted during the normal growth.

close to the experimental ones. The satisfactory values of qr in the two cases show that the DRm versus time curve is suitable to reflect the bacterial growth situation, and the bacterial growth in the two cases can be described by Gompertz model. It should be noted that A%, mm and l correspond to the maximum response value of DRm, the slope of the great variation section and the time of the initial platform in DRm versus time curve, respectively. Comparing with the normal growth, mutagenic case shows small values of mm and A, long l. The results revealed that the present PQCI technique can monitor the variations of the physical and chemical properties of culture systems in the presence and absence of his-

n

%(DRfit −DRexp)2 qr =

1

(12)

n

% DR

2 m

1

where DRfit and DRexp denote the motional resistance variation fitted and experimentally obtained, respectively, n is the number of the response signal points. The results after fitting the responses of DRm are shown in Figs. 6 and 7, and summarized in Table 1. It should be noted that DRm versus time curves drawing from the fitted results are very

Fig. 7. Time courses of DRm values experimentally obtained and fitted during the mutagenic process.

J. Zhang et al. / Talanta 53 (2000) 525–533

532

Table 1 Parameters obtained by fitting responses of (in V) according to Eq. (11) Experiment

A% (V)

mm (g−1)

l (h)

A

qr

Normal growth Mutagenic process

34.7638 16.5231

6.9709 2.2397

2.8579 9.3832

35.0432 16.5095

1.70×10−3 9.00×10−4

tidine. At the same time, the growth parameters of the bacteria (such as mm, A and l) can also be obtained from PQCI technique. Moreover, the present method provides real time monitoring of process and multidimensional information (such as DRm, Df, Cm and C0).

4. Conclusion PQCI analysis as a novel technique was reported for monitoring the mutagenic process. Equivalent circuit parameters and resonant frequency had been obtained and discussed. The increase in the motional resistance reflected the viscosity and density variations of culture system during the bacterial growth process. It was also found that the mutagenic process displays similar growth situation in the presence of histidine. By fitting the exponential equation upon DRm, the normal growth and mutagenic cases gave different growth parameters values. The results demonstrated the validity of the proposed method for its ability of real time providing multidimensional information during the mutagenic process. Therefore, the present method will have a promising application in biological process studies.

Acknowledgements This work was supported by the National Natural Science Foundation of China.

References [1] S.S. Epstein, K. Fijii, S. Asahina, Environ. Res. 19 (1979) 163.

[2] P.H. Michael, R.C. Glen, W.P. Bruce, L.C. Charles, L.L. Arthur, F.B. William, Jr, G.T. William, R.T.S. Bernd, Environ. Sci. Technol. 32 (1998) 3502. [3] A.S. Jarvis, V.A. McFarland, M.E. Honeycutt, Ecotoxicol. Environ. Saf. 39 (1998) 131. [4] M.G. Zemank, S.J. Pollard, S.L. Kenefic, S.E. Hrudey, J. Air Waste Manage. Assoc. 47 (1997) 1250. [5] P. Wilcox, S. Williamson, Environ. Health Perspect. 69 (1986) 141. [6] A. Smedes, T. Vartiainen, J. MakiPakkanen, L. Kronbery, Environ. Sci. Technol. 31 (1997) 1033. [7] J. McCann, E. Choi, E. Yamasaki, B.N. Ames, Proc. Natl. Acad. Sci. 72 (1975) 5135. [8] B.N. Ames, J. McCann, E. Yamasaki, Mutat. Res. 31 (1975) 347. [9] D.M. Maron, B.N. Ames, Mutat. Res. 113 (1983) 173. [10] B. Hanser, G. Schrader, M. Bahadir, Ecotoxicol. Environ. Saf. 38 (1997) 224. [11] J. Lewtas, S. Goto, K. Williams, J.C. Chuang, B.A. Peterseen, N.K. Wilson, Atomos. Environ. 21 (1987) 443. [12] I. Karube, T. Matsunaga, T. Nakaharra, S. Suzuki, Anal. Chem. 53 (1981) 1024. [13] I. Kayube, T. Nakaharra, T. Matsunaga, S. Suzuki, Anal. Chem. 54 (1982) 1725. [14] H. Muramatsu, E. Tamiya, I. Karube, Anal. Chem. 60 (1988) 2142. [15] S.J. Martin, V.E. Granstaff, G.C. Frye, Anal. Chem. 63 (1991) 2272. [16] M. Thompson, A.L. Kipling, W.C. Duncan-Hewitt, L.V. Rajakovic, B.A. Cavic-Vlasak, Analyst 116 (1991) 881. [17] D.A. Buttry, M.D. Ward, Chem. Rev. 92 (1992) 1355. [18] T.A. Zhou, L.H. Nie, S.Z. Yao, J. Electroanal. Chem. 293 (1990) 1. [19] S.H. Si, D.Z. Shen, L.H. Nie, S.Z. Yao, Bioelectrochem. Bioeng. 36 (1995) 161. [20] S.H. Si, Y.J. Xu, L.H. Nie, S.Z. Yao, J. Biochem. Biophys. Methods 31 (1996) 135. [21] W.K. Hadley, G. Senyk, in: D. Schlessinger (Ed.), Microbiology, American Society for Microbiology, Washington, DC, 1975, p. 12. [22] S. Sepcter, J. Clin. Microbiol. 6 (1977) 489. [23] A.G.E. Saum, R.H. Cumming, F.J. Rowell, Biosens. Bioelectron. 13 (1998) 511. [24] L.C. Cooney, V. Chemie, Biotechnology, vol. 1, Weinheim, 1981, pp. 73 – 113.

J. Zhang et al. / Talanta 53 (2000) 525–533 [25] X.L. Huang, Microbiology, Higher Education Press, Beijing, 1998, pp. 141–142 in Chinese. [26] P. Rapp, H. Reng, D.C. Hempel, F. Wangner, Biotechnol. Bioeng. 26 (1984) 1167. [27] H. Neujahr, Biotechnol. Genet. Eng. Rev. 1 (1984) 167. [28] X.Y. Zhang, Handbook of Practical Chemistry, National Defence Industry Publishing House, Beijing, 1986, p. 1326 in Chinese. [29] A.L. Kipling, M. Thompson, Anal. Chem. 62 (1990) 1514.

.

533

[30] G. Sauerbrey, Z. Phys. 155 (1955) 206. [31] K.K. Kanazawa, J.G. Gordon, Anal. Chim. Acta 177 (1985) 99. [32] Q.J. Xie, J. Wang, A.H. Zhou, Y.Y. Zhang, H.W. Liu, Z.N. Xu, Y. Yuan, M.Y. Deng, S.Z. Yao, Anal. Chem. 71 (1999) 4649. [33] L.L. Bao, Q.J. Xie, Y.G. Xu, W.Z. Wei, Anal. Lett. 32 (1999) 885. [34] M.H. Zwietering, I. Jongenburger, F.M. Rombacts, K.V. Riet, Appl. Environ. Microbiol. 56 (1990) 1875.