Development of a Faradic Impedimetric ... - ACS Publications

Jan 25, 2008 - several days for presumptive results and confirmation.3,4 Many ... (2) Terrorist Threats to Food; Guidance for Establishing and Strengt...
0 downloads 0 Views 263KB Size
Anal. Chem. 2008, 80, 1169-1175

Development of a Faradic Impedimetric Immunosensor for the Detection of Salmonella typhimurium in Milk Aikaterini G. Mantzila,† Vassiliki Maipa,‡ and Mamas I. Prodromidis*,†

Laboratory of Analytical Chemistry, Department of Chemistry, and Department of Hygiene, Medical School, University of Ioannina, 45 110 Ioannina, Greece

The development of a faradic impedimetric immunosensor for the detection of S. typhimurium in milk is described for first time. Polyclonal anti-Salmonella was cross-linked, in the presence of glutaraldehyde, on gold electrodes modified with a single 11-amino-1-undecanethiol (MUAM) self-assembled monolayer (SAM) or a mixed SAM of MUAM and 6-mercapto-1-hexanol at a constant 1 + 3 proportion, respectively. The mixed SAM was also deposited in the presence of triethylamine, which was used to prevent the formation of interplane hydrogen bonds among amine-terminated thiols. The effect of the different surface modifications on both the sensitivity and the selectivity of the immunosensors was investigated. The alteration of the interfacial features of the electrodes due to different modification or recognition steps, was measured by faradic electrochemical impedance spectroscopy in the presence of a hexacyanoferrate(II)/(III) redox couple. A substantial amplification of the measuring signal was achieved by performing the immunoreaction directly in culture samples. This resulted in immunosensors with great analytical features, as follows: (i) high sensitivity; the response of the immunosensors increases with respect to the detection time as a consequence of the simultaneous proliferation of the viable bacteria cells in the tested samples; (ii) validity; the response of the immunosensors is practically insensitive to the presence of dead cells; (iii) working simplicity; elimination of various centrifugation and washing steps, which are used for the isolation of bacteria cells from the culture. The proposed immunosensors were successfully used for the detection of S. typhimurium in experimentally inoculated milk samples. The effect of different postblocking agents on the performance of the immunosensors in real samples was also examined. Salmonella typhimurium is a Gram-negative foodborne pathogen that affects the abdomen causing infection, diarrhea, and pain, and it is recognized as the second most common serotype (after * To whom correspondence should be addressed. E-mail: mprodrom@ cc.uoi.gr, Tel: +30-26510-98301/412. Fax: +30-26510-44831. † Laboratory of Analytical Chemistry, Department of Chemistry. ‡ Department of Hygiene. 10.1021/ac071570l CCC: $40.75 Published on Web 01/25/2008

© 2008 American Chemical Society

Salmonella enteritidis) of Salmonella found in humans.1 Beyond the inadvertent contamination of foods and water, public concern also arises from cases of adulteration, such as the contamination of salad bars with S. typhimurium in Oregon.2 Hence, the detection of this pathogen is extremely important and highly desirable for the safety of food products and biosecurity. Bearing in mind that the pathogen has a very high proliferation rate and exists in very complex matrixes, there is a need for advanced monitoring methods and techniques. Although the standard techniques in pathogen detection are sensitive and selective enough, involving enrichment of samples and culture plating procedures, are usually elaborate, and need several days for presumptive results and confirmation.3,4 Many types of Salmonella detection tests have been developed during the last decades, involving enzyme-linked immunosorbent assay (ELISA)5-8 and nucleic acid-based polymerase chain reaction (PCR) technology.9-11 However, ELISA requires extra enzymelabeled antibodies, while the PCR-based assay is often more laborintensive than ELISA.12 In addition, some components of food and chemicals required for selective enrichment of cells may influence the effectiveness of the PCR and cause inhibitory effects.4 A variety of alternative methods have been also reported for the detection of Salmonella including the monitoring of oxygen (1) Mahon, C. R.; Manuselis, G. Textbook of Diagnostic Microbiology, 2nd ed.; Saunders: Philadelphia, PA, 2004; Chapter 6. (2) Terrorist Threats to Food; Guidance for Establishing and Strengtheninng Prevention and Response Systems; WHO: Geneva, Switzerland, 2002; pp 1-46. (3) Andrews, W. H.; June, G. A.; Sherrod, P.; Hammack, T. S.; Amaguana, R. M. Food and Drug Administration Bacteriological Analytical Manual, 8th ed.; AOAC International; Gaithersburg, MD, 1995; pp 5.01-5.20. (4) Jofre, A.; Martin, B.; Garriga, M.; Hugas, M.; Pla, M.; Rodriguez-Lazaro, D.; Aymerich, T. Food Microbiol. 2005, 22, 109-115. (5) Cudjoe, K. S.; Hagtvedt, T.; Dainty, R. Int. J. Food Microbiol. 1995, 27, 11-25. (6) Tian, H.; Miyamoto, T.; Okabe, T.; Kuramitsu, Y.; Honjoh, K. C.; Hatano, S. J. Food Prot. 1996, 59, 1158-1163. (7) Beckers, H. J.; Tips, P. D.; Soentoro, P. S. S.; Delfgou-Van Asch, E. H. M.; Peters, R. Food Microbiol. 1998, 5, 147-156. (8) Mansfield, L. P.; Forsythe, S. J. Food Microbiol. 2001, 18, 361-366. (9) Luk, J. M. C. Biotechnology 1994, 17, 1038-1042. (10) Bennett, A. R.; Greenwood, D.; Tennant, C.; Banks, J. G.; Betts, R. P. Lett. Appl. Microbiol. 1998. 26, 437-441. (11) Chen, W.; Martinez, G.; Mulchandani, A. Anal. Biochem. 2000, 280, 166172. (12) Wray, C.; Wray, A. Salmonella in Domestic Animals; Oxford University Press: New York, 2000.

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008 1169

consumption with cyclic voltammetry,13 acoustic wave sensors based on filamentous bacteriophages,14 and indirect fiber-optic biosensors.15 Other approaches, which are based on quartz crystal microbalance piezoimmunosensors,16-20 conductometric immunosensors,21 magnetoelastic resonance biosensors,22 and electrochemical ELISA immunosensors23-25 have been also proposed. Of the methods available for the detection of Salmonella, impedance microbiology has been successfully used in both standard and real samples by measuring either the impedance of selective growth media or the interfacial impedance of the electrodes.26-28 However, these methods do not provide the analytical simplifications of biosensors. Electrochemical impedance spectroscopy (EIS)-based immunosensors are particularly attractive compared with the aforementioned approaches, due to their ability for miniaturization, the low cost of electrode mass production, and the cost-effective instrumentation.29 In addition, they exhibit low detection limits, especially when the main antibody-antigen interaction signal is being enhanced by case-specific amplification schemes.30-32 In general, these schemes follow the main immunoreaction step and include the application of extra enzyme-labeled antibodies,32 biotinstrept)avidin complex,30 or enzyme labels that precipitate an insoluble compound on the sensing interface.31 However, the majority of the published works have not been tested in real matrixes, since fundamental issues still exist, and this in turn has brought the reliability of impedimetric immunosensors into question. The aim of this work was the selection of a suitable immobilization platform for anchoring anti-Salmonella, so that one can preserve its biological activity and selectivity of the target analyte. In order to amplify the antibody-bacteria cell interaction signal, antibody functionalized electrodes-bacteria cells immunoreaction step was performed directly in the culture samples. This led to considerable increasing signals compared with those which would have been obtained if the immunoreaction was (13) Ruan, C.; Yang, L.; Li, Y. J. Electroanal. Chem. 2002, 519, 33-38. (14) Olsen, E V.; Sorokulova, I. B.; Petrenko, V. A.; Chen, I-H.; Barbaree, J. M.; Vodyanoy, V. J. Biosens. Bioelectron. 2006, 21, 1434-1442. (15) Ko, S.; Grant, S. A. Biosens. Bioelectron. 2006, 21, 1283-1290. (16) Babacan, S.; Pivarnik, P.; Letcher, S.; Rand, A G. Biosens. Bioelectron. 2000, 15, 615-621. (17) Fung, Y. S.; Wong, Y. Y. Anal. Chem. 2001, 73, 5302-5309. (18) Wong, Y. Y.; Ng, S. P.; Ng, M. H.; Si, S. H.; Yao, S. Z.; Fung, Y. S. Biosens. Bioelectron. 2002, 17, 676-684. (19) Kim, G-H.; Rand, A. G.; Letcher, S. V. Biosens. Bioelectron. 2003, 18, 9199. (20) Su, X-L.; Li, Y. Biosens. Bioelectron. 2005, 21, 840-848. (21) Muhammad-Tahir, Z.; Alocilja, E. C. Biosens. Bioelectron. 2003, 18, 813819. (22) Guntupalli, R.; Hu, J.; Lakshmanan, R. S.; Huang, T. S.; Barbaree, J. M.; Chin, B. A. Biosens. Bioelectron. 2007, 22, 1474-1479. (23) Croci, L.; Delibato, E.; Volpe, G.; Palleschi, G. Anal. Lett. 2001, 34, 25972607. (24) Delibato, E.; Volpe, G.; Stangalini, D.; De Medici, D.; Moscone, D.; Palleschi, G. Anal. Lett. 2006, 39, 1611-1625. (25) Croci, L.; Delibato, E.; Volpe, G.; De Medici, D.; Palleschi, G. Appl. Environ. Microbiol. 2004, 70, 1393-1396. (26) Yang, L.; Ruan, C.; Li, Y. Biosens. Bioelectron. 2003, 19, 495-502. (27) K’Owino, I. O.; Sadik, O. A. Electroanalysis 2005, 17, 2101-2113. (28) Yang, L.; Li, Y.; Griffis, C. L.; Johnson, M. G. Biosens. Bioelectron. 2004, 19, 1139-1147. (29) Katz, E.; Willner, I. Electroanalysis 2003, 15, 913-947 (30) Pei, R.; Cheng, Z.; Wang, E.; Yang, X. Biosens. Bioelectron. 2001, 16, 355361. (31) Ruan, C.; Yang, L.; Li, Y. Anal. Chem. 2002, 74, 4814-4820. (32) Balkenhohl, T.; Lisdat, F. Analyst 2007, 132, 314-322.

1170

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

carried out in a buffer solution, where the population of the cells remains unchangeable during the course of the immunoreaction. It is also important to point out that the proposed method is insensitive to dead cells as, in contrast to viable cells, they cannot proliferate and thus their presence has almost no contribution to the measuring signal. An additional advantage of this method is the elimination of time-consuming cleanup sample procedures in pathogen detection (centrifugation and washing steps), which may also reduce the original concentration of the viable cells in the tested samples leading to false negative results. Finally, the proposed immunosensors were successfully used for the detection of S. typhimurium in experimentally inoculated milk samples. The effect of different postblocking agents on the performance of the immunosensors in real samples was examined and discussed. EXPERIMENTAL SECTION Chemicals and Solutions. 11-Amino-1-undecanethiol hydrochloride (MUAM) was from Dojindo . 6-Mercapto-1-hexanol (MH), glutaraldehyde (GA) (∼25% in water; kept in sealed vials under argon at +4 °C), and bovine serum albumin (BSA) were purchased from Sigma. Absolute ethanol, denaturated ethanol, potassium ferrocyanide, potassium ferricyanide, and triethylamine were from Merck. Tryptone soy broth (TSB) and affinity-purified rabbit polyclonal antibodies against Salmonella (∼4 mg mL-1 in 10 mM PBS pH 7.2) were purchased from LAB M and Biodesign, respectively. Milk powder (Regilait) and fresh, skimmed, pasteurized milk were purchased from the local market and used as received. All other chemicals were from Merck and Sigma, and double-distilled water (DDW) was used throughout. A 10 mM phosphate-buffered saline (PBS) solution containing 140 mM NaCl, pH 7.2, was used for the dilution of the antibody stock solution and postblocking steps. A 50 mM phosphate buffer (PB) solution, pH 7, was used for the activation of amine groups with glutaraldehyde and in various washing steps, whereas a 50 mM PBS solution containing 100 mM KCl, pH 7, was used for the removal of nonbound antibodies from the surface of the immunosensors and as electrolyte of the redox couple in EIS measurements. All buffer solutions were filtered through a 0.2µm pore size membrane (Millipore) and stored at +4 °C for up to 2 weeks. Bacteria and Colony Forming Unit (cfu) Determinations. The tested bacteria were S. typhimurium (4,5:i:1,2) (from National School of Public Health, Laboratory of Microbiology, Athens, Greece) and Escherichia coli NCTC 9001 (from Public Health Laboratory Service-North, Newcastle, England). The bacteria cultures were grown in TSB at 37 ( 1 oC for 24 h before use, and the number of viable cells was determined by a microbial plate count method. Antigen samples were then prepared by serial dilutions of the stock culture with the broth solution. Apparatus. EIS experiments were performed with the electrochemical Analyzer PGSTAT12/FRA2 (Eco Chemie) in a onecompartment three-electrode cell. Gold electrodes were used as working electrodes, and a platinum wire (diameter of 1 mm and length of 3 cm) served as the auxiliary electrode. The reference electrode was a Ag/AgCl/3 M KCl (IJ Cambria, UK) electrode, and all potentials reported hereafter refer to the potential of this electrode. The impedance spectra were recorded over a frequency range of 10-1-105 Hz, using a sinusoidal excitation signal,

superimposed on a dc potential of +0.200 V. Excitation amplitude of 10 mV (rms) was used throughout. All measurements were performed in a solution of 5 mM hexacyanoferrate(II)/(III) (1 + 1 mixture) in PBS solution (pH 7) at room temperature. Formation of SAMs and Immobilization of the Antibodies. Gold electrodes were constructed by using the commercial kit EasyCon (EasyCon Hellas, provided by Eco Chemie). Before use, gold electrodes of 2-mm active surface were polished with Al2O3 (0.01-mm grain size) and sonicated for 3 min in DDW. After polishing, gold surfaces were cleaned by dipping into a solution of 1 + 1 + 5 (v/v), NH4OH + 30% H2O2 + H2O for 10 min, washed thoroughly with DDW and absolute ethanol, dried under argon, and immersed immediately in a mixture of 0.5 mM MUAM and 1.5 mM MH in absolute ethanol or in a solution of 0.5 mM MUAM in the same solvent for 16 h at room temperature in the dark. Deposition of a mixed self-assembled monolayer SAM was also performed in an ethanolic mixture of 0.5 mM MUAM and 1.5 mM MH with 3% (v/v) triethylamine (TEA).33 The thiol SAM-modified electrodes were then thoroughly rinsed in fresh baths of absolute ethanol, 4 × 5 min (electrodes that had been modified in the presence of TEA were additionally washed with 10% (v/v) acetic acid in ethanol after the first washing step in absolute ethanol), dried under argon, and immersed immediately in well-stoppered vials containing a degassed solution of 2.5% glutaraldehyde in PB solution for 1 h under mild stirring. Then electrodes were rinsed several times with the same buffer solution to remove the physically absorbed glutaraldehyde. The Salmonella antibodies (anti-SA) were introduced onto the Au/thiol/GA-activated electrodes by dropping 10 µL of ∼2 mg mL-1 antibody in PBS (pH 7.2) and allowed to incubate at 4 °C for at least 12 h in a humidified glass chamber. After the immobilization of the antibody, the electrodes were rinsed with PBS solution (pH 7) to remove the excess of physically bound antibody and immersed in the solution of the blocking agent [TSB, or BSA (1.6 or 3.2 mg mL-1) or 1.6 mg mL-1 milk powder in PBS solution (pH 7.2)] for 2 h at room temperature to block nonspecific binding sites. Procedures. Optimization Studies. The stock culture was serially diluted with TSB, and samples of different initial concentrations of S. typhimurium (typically in the range of ∼10-106 cfu mL-1) were thus prepared. The 2-mL portions of them were introduced in sterilized glass vials. Ready-to-use immunosensors [fully functionalized (Au/thiol/GA/anti-SA) electrodes after their incubation in the solution of the blocking agent] were immersed in the standard culture samples for a specified time interval at room temperature under mild stirring. Immunosensors were then thoroughly rinsed with PB solution and transferred to the measuring cell. During the immunoreaction, the initial transparent or semitransparent (depending on the initial cell number of the bacteria) samples became turbid as a result of the proliferation of the bacteria cells. Application to Milk Samples. The 0.9-mL portions of fresh pasteurized milk were inoculated with 100 µL of the bacteria culture containing 107 or 2 × 104 cfu mL-1 S. typhimurium (positive controls) or E. coli (negative controls) and kept under natural conditions for 1 h. Then aliquots (100 µL) of inoculated and uninoculated (used as blank to evaluate the effect of the matrix components on the measuring signal) milk samples were mixed (33) Wang, H.; Chen, S.; Li, L.; Jiang, S. Langmuir 2005, 21, 2633-2636.

with 1.9 mL of TSB in sterilized glass vials. The rest of the procedure was similar to that followed during the optimization studies. Throughout this study, the relative change of the signal of the ready-to-use immunosensors before and after the immunoreaction, ∆Rct (%), which is expressed as ∆Rct ) {[Rct (after the immunoreaction) - Rct (before the immunoreaction)]/Rct (before the immunoreaction)} × 100 was taken as a measure of the concentration of bacterial cells in standard and real samples. Correspondingly, ∆Rct (%) values before and after each modification step were used as a measure to evaluate the effect of the different modification steps on the measuring signal. Safety Considerations. All glassware in contact with bacteria must be sterilized for 1.5 h at 115 °C before and after use. Before heating, glassware was immersed sequentially in baths of 0.1 M HCl and denatured ethanol. Waste solutions containing bacteria must be sterilized before disposal. RESULTS AND DISCUSSION Evaluation of Different Surface Modifications. A determining factor in the performance of biosensors is the design of the immobilization platform, as it greatly affects both the sensitivity and the specificity of the biointerfaces. Our aim was to immobilize the anti-Salmonella onto the surface of gold electrodes after their modification with an amine-terminated SAM, and glutaraldehyde cross-linking. MUAM was chosen among other amine-terminated thiols, as it provides stable coverage over time.34,35 In order to define the most suitable protocol for the immobilization of antibodies, three different electrode assemblies were examined: (i) Au/MUAM, (ii) Au/MUAM-MH, and (iii) Au/MUAM-MH in the presence of TEA. MH was used at a 3-fold higher concentration compared with that of MUAM, in order to be served as a “dilutor” of the main thiol (MUAM) that brings the functional headgroup. In general, the use of two-thiol mixed SAMs has been shown to improve the bioactivity of a protein immobilized on such layers compared with that on a single thiol-based SAM.36-38 The second thiol reduces the surface concentration of the functional groups and thus minimizes steric hindrances and can also lead to different conformation of the immobilized molecules.36,39 TEA was used to prevent the formation of interplane hydrogen bonds between amine groups of MUAM on gold and free MUAM molecules in the bulk that could lead to a partial second layer of MUAM atop of the gold-bound monolayer (Scheme 1). More details and scientific evidence of this behavior are provided in the excellent work of Wang et al.33 Fully functionalized immunosensors that are based on Au/ MUAM, Au/MUAM-MH, and Au/MUAM-MH(TEA) electrodes were blocked with the TSB solution and then incubated in both positive and negative control samples containing the same initial cell number of bacteria. Comparative results (Table 1) reveal that (34) Mirsky, V. M.; Riepl, M.; Wolfbeis, O. S. Biosens. Bioelectron. 1997, 12, 977-989. (35) Lasseter, T. L.; Cai, W.; Hamers, R. J. Analyst 2004, 129, 3-8. (36) Briand, E.; Salmain, M.; Herry, J. M.; Perrot, H.; Compe`re, C.; Pradier, C. M. Biosens. Bioelectron. 2006, 22, 440-448. (37) Frederix, F.; Bonroy, K.; Laureyn, W.; Reekmans, G.; Campitelli, A.; Dehean, W.; Maes, G. Langmuir 2003, 19, 4351-4357. (38) Ge, B.; Lisdat, F. Anal. Chim. Acta 2002, 454, 53-64. (39) Guiomar, A. J.; Guthrie, J. T.; Evans, S. D. Langmuir 1999, 15, 11981207.

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1171

Scheme 1. Tentative View of the Different Modification and Recognition Steps of the BSA-Blocked Au/MUAM-MH/GA/Anti-SA Immunosensorsa

a

Drawings are not to scale.

Table 1. Buildup of Different Au/Thiol/GA/Anti-SA Electrode Assemblies and Comparative Results of Their Response (after Postblocking with TSB) to Positive and Negative Controls Containing an Initial Concentration of 106 CFU mL-1 Bacteriaa electrode assembly

a

MUAM

MUAM-MH

MUAM-MH TEA

Au, Rct Au/thiol, Rct Au/thiol/GA, Rct , (θ) ∆Rct (%) after glutaraldehyde Au/thiol/GA/anti-Salmonella, Rct ∆Rct (%) after anti-Salmonella

Buildup of Different Electrode Assemblies 0.20 0.21 1.96 0.15 9.71 (0.98) 0.46 (0.54) +395 +210 18.7 1.1 +93 +135

0.19 0.16 1.86 (0.89) +1089 3.21 +73

∆Rct (%) S. typhimurium (P) ∆Rct (%) E. coli (N) ∆Rct (P)/∆Rct (N)

After Immunoreaction. Detection Time 2 h +33 +213 +22 +16 1.5 13.3

+125 +29 4.3

Rct in kΩ. The values in parentheses are of the surface coverage, calculated as θ ) 1 - [(Rct(Au)/ Rct(Au/thiol/GA)].

MH has a beneficial effect on the analytical features of the immunosensors. Indeed, in the absence of it, the resulted immunosensors exhibit a poor performance. On the other hand, the presence of TEA seems to have a rather negative effect on the performance of the mixed SAM-based immunosensors. Trying to elucidate these results, the tested electrodes assemblies were repeatedly examined and faradic impedimetric measurements after each step of the sensor buildup were performed. Impedance data were fitted to a Randles equivalent circuit (Figure 1A, side panel), and electrical parameters were determined using the FRA2 software (Eco Chemie). To ensure the best reproducible conditions, every series of the tested electrodes was run in parallel and various surface modification and immunoreaction steps were made from the same stock solutions of thiols, glutaraldehyde, antibody, and bacteria. Received results can be rationalized as follows: (i) The presence of TEA seems to have no effect on the Rct values of Au/MUAM-MH electrodes indicating that Rct values are mostly determined on the presence of the “dilutor” (MH). The fact that the Rct values of Au/MUAM-MH electrodes are lower than those of bare gold electrodes (Figure 1A and Table 1) is attributed to the observed enhancement of the electrocatalytic 1172 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

efficiency of short-chain thiol-modified electrodes to the used redox probe, in accordance with previous studies.40 On the other hand, the increase of Rct from 0.20 to 1.96 kΩ at Au/MUAM electrodes indicates the formation of Au/MUAM and Au/MUAM‚‚‚MUAM layers,33 which provide an effective barrier to electron transfer of the redox species in solution. (ii) In the next step of the sensor buildup, free terminal amine groups react with glutaraldehyde to create amine-reactive aldehyde ends that will function as binding sites for the immobilization of the antibody molecules (Scheme 1). The extremely high relative signal changes (1089%) that were observed at Au/MUAM-MH (TEA) electrodes constitutes great evidence for the high population of free amine groups in these electrode assemblies. On the other hand, in the absence of TEA, the considerably lower values of ∆Rct, (395 and 210% at Au/MUAM and Au/MUAM-MH electrodes, respectively) are probably attributable to the low percentage of free amine ends due to the partial formation of interplane MUAM layers. The value of ∆Rct at Au/MUAM-MH electrodes (210%) is lower compared with that observed at Au/ MUAM electrodes, as the presence of the “dilutor” in the former (40) Mantzila, A. G.; Strongylis, C.; Tsikaris, V.; Prodromidis, M. I. Biosens. Bioelectron. 2007, 23, 362-369.

Figure 2. Response and specificity of BSA-blocked Au/MUAMMH/GA/anti-SA electrodes. BSA, 1.6 mg mL-1 protein in 10 mM PBS solution, pH 7.2. Detection time, 2 h. Measuring conditions, 10-1105 Hz at +0.200 V bias (10 mV rms). Electrolyte, 5 mM hexacyanoferrate(II)/(III) (1 + 1 mixture) in PBS solution, pH 7.

ones reduces the population of the free amine groups. At this point, it is important to state that the observed increase of the absolute values of Rct(Au/thiol/GA) in all the tested assemblies is not solely attributable to the attachment of glutaraldehyde (a relative small molecule of MW 100.1 Da) but also to the spontaneous polymerization of it,41 although the necessary measures for the minimization of this effect have been taken (the type of the stock material and performance of the reaction in degassed conditions). (iii) An interesting finding, which is revealed by this study, is that the ∆Rct(%) values after the immobilization of the antibodies are irreversibly proportional to those calculated after the treatment with glutaraldehyde (Table 1). In this regard, factors that control the glutaraldehyde loadings are rather determinant of the analytical performance of the resulted immunosensors. High glutaral-

dehyde loadings (as they are also defined by the corresponding values of the surface coverage, Table 1) may result in the immobilization of antibodies through multiple binding sites, thus reducing the flexibility (rigidity of binding) and the binding capacity of the antibodies as well as the accessibility of the target analyte to them. The experimental results support this assumption, as it revealed from the comparison of the selectivity ratios (Table 1), expressed as, P/N ) [∆Rct (%) S. typhimurium (P)/∆Rct (%) E. coli (N)], that have been calculated for each tested electrode assembly. The buildup of the optimum electrode assembly and the selectivity of the resulted immunosensors to both positive and negative control samples are illustrated in Nyquist plots in Figure 1A,B, respectively. Antibody Loading. Bioreceptor loading for anti-Salmonella was tested in order to define the antibody loading to obtain the maximum response. A saturation study was made using the glutaraldehyde-activated Au/MUAM-MH electrodes and four discrete antibody concentrations, that is, 0.5, 1, 2, and 4 mg mL-1 in PBS solution (pH 7.2). The sensitivity of the corresponding immunosensors, after their blocking, to a standard sample containing an initial concentration of 106 cfu mL-1 S. typhimurium (data not shown) was taken as the criterion for this study. The profile of the system sensitivity reaches a plateau at an antibody concentration of 2 mg mL-1, and therefore, further experiments were performed, applying the useful antibody content onto the Au/MUAM-MH/GA electrodes. Performance of the Immunosensors. The recognition properties of Au/MUAM-MH/GA/anti-SA electrodes were examined after their incubation for 2 h in 1.6 mg mL-1 BSA solution (blocking agent) in PBS, pH 7.2. The efficiency of the resulted immunosensors in a series of standard culture samples over the (initial) concentration range 102-106 cfu mL-1 S. typhimurium and to a discrete concentration of E. coli are shown in Figure 2. The data obtained indicate that, for a detection time of 2 h, it is possible to detect S. typhimurium at a concentration level 3 orders of magnitude lower than the infectious dosage, that is, 105 CFU mL-1.42

(41) Gillet, R.; Gull, K. Histochemie 1972, 30, 162-167. Technical Data Sheet No.124, Polysciences Inc., 1999; pp 1-3.

(42) Murray, P. R.; Rosenthal, K. S.; Kobayashi, G. S.; Pfaller, M. A. Medical Microbiology, 3rd ed.; Mosby: St. Louis, MO, 1998; Chapter 29.

Figure 1. (A) Nyquist plots showing the buildup of the immunosensors: (a) Au/MUAM-MH, (b) Au bare, (c) Au/MUAM-MH/GA, and (d) Au/MUAM-MH/GA/anti-SA electrodes. Side panel, Randles equivalent circuit Rs(Qdl[RctW]), where Rs is the ohmic resistance of the electrolyte, Rct is the charge-transfer resistance, due to electron transfer of the redox probe to the electrode, W the Warbung impedance resulting from the diffusion of the redox couple toward the electrode surface and Qdl is the double layer capacitance, which is represented by a constant phase element [CPE ) (1/Q)-1(jω)-n]. (B) Impedimetric spectra of TSB-blocked Au/MUAM-MH/GA/anti-SA electrodes (a) before and after their incubation with (b) E. coli and (c) S. typhimurium culture samples for 1 h. Initial concentration of bacteria, 106 cfu mL-1. Measuring conditions, 10-1-105 Hz at +0.200 V bias (10 mV rms). Electrolyte, 5 mM hexacyanoferrate(II)/(III) (1 + 1 mixture) in PBS solution, pH 7.

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1173

Figure 3. Nyquist plots of BSA-blocked Au/MUAM-MH/GA/antiSA electrodes (a) before and (b) after their incubation for 2, (c) 6, and (d) 20 h in a culture containing an initial concentration of 102 cfu mL-1 S. typhimurium. Measuring conditions, 10-1-105 Hz at +0.200 V bias (10 mV rms) Electrolyte, 5 mM hexacyanoferrate(II)/(III) (1 + 1 mixture) in PBS solution, pH 7.

Figure 4. Nyquist plots of BSA-blocked Au/MUAM-MH/GA/antiSA electrodes (a) before and after (b) 2-h incubation in a culture sample containing an initial concentration of 106 cfu mL-1 S. typhimurium and (c) 2-h incubation in the culture supernatant solution (see text for details). Measuring conditions, 10-1-105 Hz at +0.200 V bias (10 mV rms). Electrolyte, 5 mM hexacyanoferrate(II)/(III) (1 + 1 mixture) in PBS solution, pH 7.

The effectiveness of the proposed amplification scheme is proven in Figure 3. The lowest signal response that was received (Figure 2, 102 cfu mL-1 S. typhimurium, ∆Rct ) 14% for a detection time of 2 h) can be substantially increased to 89 and 665% for detection time intervals of 6 and 20 h, respectively. Correspondingly, ∆Rct values of 33 and 131% were achieved at a sample containing an initial concentration of 10 cfu mL-1 of the target pathogen for detection time intervals of 6 and 10 h, respectively. The nature of the proposed measuring methodology raised a concern regarding the mechanism that produces the observed signal changes. During the proliferation of Gram-negative bacteria, a number of exotoxins (enterotoxins and cytotoxins with a MW ∼70-88 kDa) are also produced and liberated in the culture medium.43 In order to evaluate signal contributions originated from the attachment of the exotoxins onto the surface of the BSAblocked immunosensors, the following diagnostic experiment was performed. A culture sample of S. typhimurium was divided into two equal parts, one of which was used to perform the proposed methodology as it has been described so far (2-h incubation time), while the other was kept aside for 2 h and then was centrifuged to separate the cells from the culture medium. The supernatant solution was then transferred to a sterilized vial and a new BSAblocked immunosensor was immersed in it for 2 h. The Nyquist plots in Figure 4 (curves b and c) indicate that any contribution to the measuring signal due to exotoxins, is lower than 4% and the overwhelming contribution is due to the specific antibodyantigen interactions. This also justifies the high selectivity of the proposed immunosensors to other Gram-negative bacteria that also produce exotoxins. Application to Milk Samples. This part of the study was focused on the qualitative discrimination of S. typhimurium and E. coli inoculated milk samples, which are used as positive and negative controls, respectively. As can be seen from the results presented in Table 2, TSBblocked immunosensors succeed in the discrimination of positive

Table 2. Application of Au/Thiol/GA/Anti-SA Electrodes in Inoculated Milk Samples (5 × 104 and 102 CFU mL-1 Initial Concentration of Bacteria Cells)a

(43) Hirst, T. R. Assembly and secretion of oligomeric toxins. In Sourcebook of Bacterial Protein Toxins; Alouf, J. E., Freer, J. H., Eds.; Academic Press: London, 1991; pp 75-100.

1174 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

∆Rct (%) E. coli inoculated milk samplesb

P/N ratio

blocking agent/ detection time

milk samples

S. typhimurium inoculated milk samplesb

TSB/5 h milk powder/5 h BSA (mg mL-1) 3.2/5 h 1.6/5 h 1.6/2 h 1.6/10 h

70 28

320 224

52 239

∼6 ∼1

29 44 34 59

506 1449 117 254c

14 17 25 40c

∼36 ∼85 ∼5 ∼6

a The standard deviation of the mean ranges from 12 and 18%, n ) 5. b 5 × 104 cfu mL-1. C 102 cfu mL-1.

and negative milk samples with a calculated P/N ratio of 6. This ratio should not be compared with that calculated in standard culture samples (P/N 13.3, Table 1), as the magnitude of ∆Rct(%) values in real samples is affected to a certain degree by background signals (Table 2). The lower signal response, which was observed at E. coli inoculated real samples, compared with that observed in milk samples (blank), could probably be explained by a screening effect of the nonspecifically bound E. coli cells on the various components of the matrix, thus reducing the surface coverage of the immunosensors. Aiming to achieve higher P/N ratios, thus allowing the successful application of the method in shorter detection times, the effect of various blocking agents on the background signal was also examined. Unexpected results were obtained for milk powder-blocked immunosensors. Bulk micelles of casein probably block the specific binding sites as well, thus resulting in practically inactive (P/N ∼ 1) immunosensors. The observed ∆Rct(%) values in both inoculated samples could possibly be attributed to nonspecific electrostatic interactions between the casein micelles and the bacterial cells. However, the exact mechanism of this behavior

Finally, it is important to point out that none of the analytical trials we run gave false negative (S. typhimurium) or positive (E. coli) results, thus constituting strong evidence for the validity of the proposed immunosensors for the detection of S. typhimurium in milk samples.

Figure 5. Bode plots illustrating the response of (a) BSA-blocked Au/MUAM-MH/GA/anti-SA electrodes before and after their incubation for 5 h in milk samples inoculated with 5 ×104 cfu mL-1 (b) E. coli, and (c) S. typhimurium. Filled symbols, impedance profile; empty symbols, phase profile. Measuring conditions, 10-1-105 Hz at +0.200 V bias (10 mV rms). Electrolyte, 5 mM hexacyanoferrate(II)/(III) (1 + 1 mixture) in PBS solution, pH 7.

cannot be explained with the given data, and further studies should be performed. On the other hand, BSA-blocked immunosensors exhibited an excellent performance. As can be seen in Table 2, for low- and high-concentration BSA-blocked immunosensors, the background signal dropped from 70 (blocking with TSB) to 44 and 29%, respectively. This effect in combination with the higher sensitivity provided by the BSA-blocked immunosensors (see values in Table 1 and Figure 2), results in a remarkable increase of the corresponding P/N ratios (85 and 36 for the low and high BSA-blocked immunosensors, respectively), as can be seen in Table 2. The calculated P/N ratios ensure safe results and certify the applicability of the method in milk samples (Figure 5). Sufficiently high P/N ratios of 5-6 were also observed for a detection time interval of 2 h (Table 2), or at milk samples being inoculated with a low concentration of bacteria (100 cfu mL-1) for a detection time interval of 10 h.

CONCLUSION This study employs functional immunosensors, based on gold electrodes modified with a mixed thiol-based SAM, as probes to detect S. typhimurium in milk samples for first time. Immunosensors that were developed on a mixed SAM of MUAM-MH at a 1 + 3 proportion were found to exhibit better sensitivity and selectivity compared with those developed on the same SAM in the presence of TEA or on a single SAM of MUAM. A very simple strategy for the amplification of the signal, based on the performance of the immunoreaction directly in the culture media, was successfully tested in both standard and real samples. Due to this amplification scheme, low detection limits can be achieved at the cost of course of prolonged times of detection. For a detection time of 2 h, BSA-blocked immunosensors can provide reliable analytical signals for a concentration of 3 orders of magnitude lower to the infectious dosage of S. typhimurium. In addition, the proposed method is insensitive to dead cells. In our opinion, the present work offers a true alternative to the existing ELISA and PCR methods, incorporating the simplicity and advantages of biosensors. ACKNOWLEDGMENT The research project is cofunded by the European Unions European Social Fund (ESF) & National Sources, in the framework of the program “Pythagoras II” of the “Operational Program for Education and Initial Vocational Training” of the third Community Support Framework of the Hellenic Ministry of Education. The authors thank Dr. Alexandra Koutsotoli for her valuable assistance. Received for review July 25, 2007. Accepted December 15, 2007. AC071570L

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1175