Enzyme and Microbial Technology 32 (2003) 3–13

Review

Advances in biosensors for detection of pathogens in food and water

Paul Leonarda, Stephen Heartya, Joanne Brennana, Lynsey Dunnea, John Quinna,b,Trinad Chakrabortyc, Richard O’Kennedya,∗

a School of Biotechnology and National Centre for Sensor Research, Dublin City University, Dublin 9, Irelandb Texas Instruments Inc., Austin, TX, USA

c Institute of Med. Microbiology, Justus-Leibig-Universität Giessen, Frankfurter Street 10, D-35392 Giessen, Germany

Received 19 February 2002; received in revised form 23 August 2002; accepted 6 September 2002

Abstract

While most microbes play an important role in nature, certain potentially harmful microbes can contaminate food and water, andcause a plethora of infectious diseases in both animals and humans. Conventional methods for detecting microbial contamination haveprimarily relied on time-consuming enrichment steps, followed by biochemical identification, having a total assay time of up to 1 weekin certain cases. Over the last decade, a great deal of research has focused on the development of biological sensors for the detection ofmicro-organisms, allowing rapid and “real-time” identification. This paper reviews some of the most commonly used biosensor systemsbased on their transducer properties, which include surface plasmon resonance (SPR), amperometric, potentiometric, and acoustic wavesensors and their applications for the detection of pathogens in food and water. It also highlights some of the limitations of applyingbiosensors for the detection of pathogens, such as sensitivity, cost and the need for sample pre-treatment.© 2002 Elsevier Science Inc. All rights reserved.

Keywords: Biosensor; Surface plasmon resonance (SPR); Piezoelectric; Amperometric; Potentiometric

1. Introduction

Micro-organisms, such as bacteria and viruses, are foundwidely in the environment, in food, marine and estuarinewaters, soil, and the intestinal tracts of humans and ani-mals. Many of these organisms have an essential functionin nature, but certain potentially harmful micro-organismscan have profound negative effects on both animals andhumans, costing the food industry (and indirectly, the con-sumer) many millions of dollars each year. It is estimatedthat infectious diseases cause about 40% of the approxi-mately 50 million total annual deaths world-wide[1]. Asummary of the estimated annual foodborne illnesses, hos-pitalisations and deaths caused by selected pathogens in theUS is shown inTable 1. Waterborne pathogens cause 10–20million of these deaths and, in addition non-fatal infectionof more than 200 million people each year[2]. The recentwell publicised million-dollar food recalls due to food poi-soning bacteria such asListeria monocytogenes [3,4] hasincreased the need for more rapid, sensitive and specificmethods of detecting these microbial contaminants.

∗ Corresponding author. Tel.:+353-1-7005319; fax:+353-1-7005412.E-mail address: richard.okennedy@dcu.ie (R. O’Kennedy).

Conventional methods for the detection and identifica-tion of bacteria mainly rely on specific microbiologicaland biochemical identification. While these methods can besensitive, inexpensive and give both qualitative and quan-titative information on the number and the nature of themicro-organisms tested, they are greatly restricted by as-say time, with initial enrichment needed in order to detectpathogens which typically occur in low numbers in foodand water. Some standard methods, such as, the NF EN ISO11290-1 method for the detection ofL. monocytogenes canrequire up to 7 days to yield results, as they rely on the abil-ity of micro-organisms to multiply to visible colonies[5,6].Even though, newer microbiological-based tests, such asthe ALOA® method (AES laboratoire), which uses a chro-mogenic medium in conjunction with a Listeria monodiskfor the detection ofL. monocytogenes, can reduce detectiontime down to 3 days[6], this still can present difficultiesin the quality control of semi-perishable foods. In addition,viable bacterial strains in the environment can enter a dor-mancy state where they become non-culturable (viable-butnon-culturable (VBNC)) which can subsequently lead to anunderestimation of pathogen numbers or a failure to isolatea pathogen from a contaminated sample[7–9].

Detection of micro-organisms by DNA amplification hasbeen shown to be more suitable. The polymerase chain

0141-0229/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved.PII: S0141-0229(02)00232-6

4 P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13

Table 1A summary of estimated foodborne illnesses, hospitalisations and deaths caused by selected pathogens in the US annually as calculated by the USDA’seconomic research service

Bacteria Estimatedannual cases

Estimated annualhospitalisations

Estimatedannual deaths

Onset Infectiousdose (CFU)

Salmonella 1,342,532 16,102 556 6 h to 28 days 104–107

Listeria monocytogenes 2493 2298 499 A few days to 3 weeks 400–103

Campylobacter spp. 1,963,141 10,539 99 2–5 days 400–106

Escherichia coli (0157:H7 and other types) 173,107 2785 78 12 h to 3 days 101–102

Clostridium perfringens 248,520 41 7 18–36 h >108

Staphylococcus food poisoning 185,060 1753 2 1–7 h >106

reaction (PCR) can be used to enhance the sensitivity ofnucleic acid-based assays. Target nucleic segments of de-fined length and sequence are amplified by repetitive cyclesof strand denaturation, annealing, and extension of oligonu-cleotide primers by the thermostable DNA polymerase,Thermus aquaticus (Taq) DNA polymerase[10]. PCR hasdistinct advantages over culture and other standard meth-ods for the detection of microbial pathogens and offers theadvantages of specificity, sensitivity, rapidity, accuracy andcapacity to detect small amounts of target nucleic acid ina sample[9]. PCR has been shown to accurately detectlow numbers of microbes such as viruses[11–13], bacte-ria [14–16], protozoa[17,18] and helminthes[19,20] andmultiple primers can be used to detect different pathogensin one multiplex reaction. However, problems such as thesensitivity of the polymerase enzyme to environmental con-taminants, difficulties in quantification, the generation offalse positives through the detection of naked nucleic acids,non-viable micro-organisms, or contamination of samplesin the laboratory, may limit the use of PCR for the direct de-tection of microbial contamination[9]. In addition, nucleicacid-based assays are limited in that they can only indicatethe genetic potential of a micro-organism to produce toxinor to express virulence and do not give any informationon toxins in foods or environmental samples[5]. Use ofselective enrichment media overcomes the difficulties thatmay arise in direct PCR by increasing the amount of targetDNA and decreasing the amount of inhibitors of PCR thatmay be present in samples. However, from an industrialpoint of view, routine detection of microbes using PCRcan be expensive and complicated, requiring highly skilledworkers to carry out the tests.

Immunological detection with antibodies is perhaps theonly technology that has been successfully employed forthe detection of cells, spores, viruses and toxins alike[21].Polyclonal antibodies can be raised quickly and cheaplyand do not require the time or expertise associated with theproduction of monoclonal antibodies[22]. However, poly-clonal antibodies are limited both in terms of their speci-ficity and abundance. Since the development of hybridomatechniques[23] and the emergence of recombinant antibodyphage display technology, developed during the past decade,immunological detection of microbial contamination has

become more sensitive, specific, reproducible and reliablewith many commercial immunoassays available for the de-tection of a wide variety of microbes and their products.While nucleic acid-based detection may be more specificand sensitive than immunological-based detection, the lat-ter is faster, more robust and has the ability to detect notonly contaminating organisms but also their biotoxins thatmay not be expressed in the organism’s genome[21]. Eventhough both antibody-based and nucleic acid-based detec-tion have greatly decreased assay time compared to tradi-tional culture techniques, they still lack the ability to detectmicro-organisms in “real-time”. The need for a more rapid,reliable, specific and sensitive method of detecting a targetanalyte, at low cost, is the focus of a great deal of research,especially for applications outside the laboratory environ-ment. Since its inception in the 1970s, Hazard AnalysisCritical Control Point (HACCP) methodology has evolvedas the leading food safety strategy used by the food indus-try. HACCP indentifies where potential contamination, timeand temperature problems can occur (the critical controlpoints). However, key technologies needed to successfullyimplement any HACCP program are real-time microbialdetection, tracebility and source indentification. Biosensorsoffer the potential of detecting pathogens in real-time but,however, still require the time-consuming pre-enrichment inorder to detect low numbers of pathogens in food and wa-ter. Advances in antibody production and the recent emer-gence of phage-displayed peptide biosensors[24–27] offerincreased possibilities for the rapid detection of pathogens.

Biosensors use a combination of biological receptorcompounds (antibody, enzyme, nucleic acid, etc.) and thephysical or physico-chemical transducer directing, in mostcases, “real-time” observation of a specific biological event(e.g. antibody–antigen interaction) (Fig. 1). They allowthe detection of a broad spectrum of analytes in complexsample matrices, and have shown great promise in areassuch as clinical diagnostics, food analysis, bioprocess andenvironmental monitoring[1,28]. Ideal characteristics for abiosensor are outlined inTable 2. Biosensors may be dividedinto four basic groups, depending on the method of signaltransduction: optical, mass, electrochemical and thermalsensors[29–31]. Optical transducers are particularly attrac-tive as they can allow direct “label-free” and “real-time”

P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13 5

Fig. 1. Schematic diagram showing the main components of a biosensor[24]. The biological event, e.g. antibody–antigen interaction ellicits aphysico-chemical change at the biointerface, e.g. change of mass, heat change or change in electrical potential, which is converted by the transducer toan electrical signal. The output from the transducer is then amplified, processed and finally, displayed as a measurable signal.

Table 2Ideal requirements for biosensor-based microbial detection assay

Accuracy False-positive and false-negative results must be low or preferably zero, especially when detecting pathogenic organisms.Assay time The biosensor should produce a ‘real-time’ response, especially when perishable foods are being tested.Sensitivity Failure to detect false-negative results, lowers the sensitivity of the assay, which cannot be tolerated in food microbiology.Specificity The biosensor should easily discriminate between the target organism or toxin and other organisms.Reproducible Each assay should be highly reproducible and easy to calibrate.Robust The biosensor must be able to resist changes in temperature, pH, ionic strength and be sterilisible.User friendly The assay should be fully automated and require minimal operator skills for routine detection.Compatible interface The biointerface should be compatible with the transduction principle, resist non-specific binding and

should be freely accessible in three-dimensional space.Validation The biosensor assay should be evaluated against current standard techniques and a LOD obtained.

detection of bacteria. Current studies focusing on opticallybased transduction methods aim to achieve a more robust,easy to use, portable, and inexpensive analytical system.The phenomena of surface plasmon resonance (SPR), hasshown good biosensing potential and many commercialSPR systems are now available (e.g. BIAcoreTM) [32].

1.1. Surface plasmon resonance-based sensors

SPR is a phenomenon that occurs during optical illumina-tion of a metal surface and it can be harnessed for biomolec-ular interaction analysis (BIA),[33]. It is best described as acharge density oscillation at the interface between two me-dia with oppositely charged dielectric constants. Plasmons

Fig. 2. Diagrammatic illustration of the SPR principle[36], showing the Kretschmann[37] prism arrangement of the type used in BIAcore instrumentation.

represent the ‘excited’ free electron portion of the surfacemetal layer. This resonant excitation is provided by com-patible light energy photons. The amplitude of the resultingplasmon electromagnetic or evanescent wave is maximal atthe interface between the plasmon generating (metal) andthe emergent (ambient) medium[34]. The ambient mediumis generally aqueous phase and thus, less dense with cor-respondingly lower refractive indices and is penetrated bythe evanescent wave to a depth of approximately one wave-length. Typically, guided waves propagate in a confiningstructure such as an optical fibre, whereas the surface plas-mon wave (SPW) is guided by the metal–dielectric interface[35]. Fig. 2 provides a simplified overview of the detectionprinciple.

6 P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13

Conventional SPR requires that only the p-polarised orTM (i.e. trans-magnetic, with magnetic vector normal to thedirection of wave propagation) component of the incidentlight is coupled into the plasmon oscillation. Experimentally,resonance is achieved either by varying the incident lightwavelength or the frequency at a fixed angle at or abovethe critical angle or alternatively by varying the angle ata fixed wavelength[34]. When these resonance conditionsare satisfied, incident light is absorbed with a concomitantdecrease or ‘attenuation’ of the reflected light. Interactionof contaminant material (e.g. bacterial cell or toxin) withspecific antibody immobilised on the surface results in achange in mass at the surface, which in turn causes a changein refractive index. This correspondingly alters the resonancestate and is reported as shift in the angular position of thereflectance minimum (I to II,Fig. 2).

1.2. Commercial SPR-based biosensors

The pioneers of commercial SPR-based biosensingwere Pharmacia BiosensorAB[38], now BIAcoreAB, wholaunched the original BIAcore system in 1990[33]. Thiswas one of the first commercial SPR biosensing devices.The company now has a large range of biosensors whichincludes several generations of the original BIAcore (series1000, 2000 and 3000) as well as the BIAliteTM (1994),BIAcore XTM (1996), BIAQuadrantTM, BIAcore S51 andBIAcore J (2001) systems offering varying degrees ofautomation and parameter specifications.

Artificial Sensing Instruments (ASI AG) launched Bios-1shortly after the first BIAcore systems in 1991. This in-strument is based on a grating coupler format. It uses flowthrough cells to facilitate biological interaction analysis oflarger molecular entities such as bacteria and other wholecells. Fisions Applied Sensor Technology (FAST), now

Fig. 3. Schematic outline of the TI SPREETATM biosensing element. The sensing element is an integrated disposable chip encased in a precision mouldedplastic housing. The flow cell is comprised of teflon and is detachable.

Affinity Sensors[39], was established with the sole purposeof exploiting the growing interest in evenescent wave-basedbiosensing. Systems currently available include IASysplusTM and IASys Auto+ AdvantageTM. The latter is a fullyautomated system designed for higher productivity. TheIASys system employs a resonant mirror configuration anduses stirred, disposable cuvettes in place of flow cells to en-sure sample mixing and minimise mass transport problems.Another resonant mirror configured device has recentlybeen marketed by XanTec bioanalytics GBR[40]. TexasInstruments (TI) USA[41] have developed several systemsbased on the same angular interrogation mode as BIAcoreinstruments. The miniaturised SPREETA device (Fig. 3)appeals to a broad market base, especially the medical di-agnostics and food industries, by enabling the fabricationof biosensors tailored for routine testing, remote testing andbasic research and development applications[42]. In Au-gust 1999, Biotul[43] offered the Kinomics PlasmoonTM

device. Quantech[44] have also recently begun developingan SPR-based instrument to be applied to routine medicaldiagnostics.Table 3compares several of the systems men-tioned. It is common to measure transducer performance interms of limit of detection (LOD) values. Low LOD is nor-mally achieved by increasing the signal to noise (S/N) ratio.However, SPR-based biosensors tend to use very small sam-ple volumes, requiring a pre-enrichment step to concentratelow numbers of pathogens occurring in a large sample vol-ume. SPR-based transducers are also particularly sensitiveto ambient temperature drift and, for maximal performancethis parameter must be controlled. Other disadvantages ofSPR-based biosensors such as cost and flow cell stabilitycan be overcome by sensor miniaturisation (e.g. SPREETA)and choice of coupling chemistry or sensor surface used(e.g. BIAcore supply a range of sensor chips with differentsurface chemistries for different applications).

P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13 7

Table 3Comparison of several commercially available SPR-based biosensors

Principle BIAcore 3000(prism-based SPR)

IBIS (vibratingmirror SPR)

Plasmoon(broad-range SPR)

SPREETA(prism-based SPR)

IASys(resonant mirror)

Flow-injection analysis (FIA) system Yes Yes No Yes No

Temperature control Yes Yes Yes Noa Yes

Autosampler Yes No Yes No Yes

Microfluidics Yes No No No No

Disposable sensing element Yes Yes Yes Optional Yes

Refractive index range 1.33–1.40 1.33–1.43 1.33–1.48 1.33–1.40 –

Limit of detection (RIU) 3× 10−7 2 × 10−6 6 × 10−6 0.3 × 10−6 >1 × 10−6

a Does not offer temperature control but offers temperature compensation by correcting the signal for temperature fluctuations.

Successful detection of micro-organisms requires opti-misation of several parameters. Firstly, suitably specificantibodies must be available. Microbial cell membranesand/or walls contain a plethora of protein and carbohydratematerial but ideally, the antibody used should be directedagainst a single target molecule, which is species-specific,surface exposed and constitutively expressed. One of theearliest attempts to use BIAcoreTM for direct microbial de-tection[45] achieved a detection limit forEscherichia coliO157:H7 of 5–7× 107 CFU/ml. Koubava et al.[46], re-ported successful detection ofSalmonella enteritidis andL.monocytogenes using the conventional Kretschmann prismarrangement. Cells could be detected at concentrations of106 ml−1, which although still quite insensitive, was com-parable to ELISA using the same antibodies. Immobilisingthe antibodies in cross-linked dextran was preferred as itwas thought to allow greater cell–ligand interaction.

Optimal immobilisation is the second major parame-ter of importance. For cellular interactions it may be thatdextran-based hydrogels have a negative effect on sensitiv-ity since they can inhibit diffusion of larger cell entities andrestrict cells to the extreme of the propagating evanescentfield. Whole cells were also anticipated to have potentiallyproblematic effects on the complex microfluidics of themarket leading BIAcore device[47]. However, Quinn andO’Kennedy [48] demonstrated the successful detection ofdifferent red blood cell types. Even allowing for the vis-coelastic properties of RBC they are an order of magnitudelarger than bacterial cells. We anticipate that the presenceof extraneous material from complex food matrices may bemore problematic. This can only be overcome by a selectiveenrichment in defined growth media such as that describedin ISO 11290-1 and/or by immunoseparation such, as themethod described by Kaclikova et al.[49], for the separa-tion of Listeria from cheese. Thus, sample preparation is thethird parameter that needs to be considered. Assay design isthe final stage, where all the above considerations are usedto decide the most suitable assay format for the particularbiosensing device being employed. A competitive inhibitionassay is often employed where, for any or all of the criteria

outlined before, a direct assay is unsuitable. This essentiallyinvolves incubating specific antibody with analyte, sepa-rating and measuring the proportion of unreacted antibody,thus inversely estimating the amount of analyte present.Haines and Patel[50] employed such an assay for detectionof Salmonella andListeria. Essentially, bacteria were incu-bated with specific antibody; cells and bound antibody wereconcomitantly removed with a 0.22�m filter. Free antibodywas determined by passing the filtered solution over an im-mobilised anti-Fab antibody layer thus generating an indi-rect estimation of cell concentration sensitive down to levelsof 104 cells/ml. This type of assay is particularly suitablewhere the availability of pure antigen in sufficient quantitiesfor traditional antigen inhibition assays is limiting.

Perkins and Squirrell[51], have described the use of lightscattering in combination with conventional SPR-based plat-forms with potential for the enhanced detection of bacterialcells. The example given was for the detection ofBacil-lus subtilis var. niger spores down to a concentration of107 spores/ml. It is suggested that such a system could beroutinely used for the concomitant quantification of solubleproteins and the detection of bacterial cells as particulateentities. An additional possibility is the incorporation of di-electrophoresis[52] to minimise non-specific interference.

1.3. Acoustic wave-based biosensors

Acoustic wave biosensors are based on the detection ofmechanical acoustic waves and incorporate a biologicalcomponent. These are mass sensitive detectors, which areoperated on the basis of an oscillating crystal that resonatesat a fundamental frequency[53]. After the crystal has beencoated with a biological reagent (such as an antibody) andexposed to the particular antigen a quantifiable changeoccurs in the resonant frequency of the crystal, which cor-relates to mass changes at the crystal surface[54]. The vastmajority of acoustic wave biosensors utilise piezoelectricmaterials as the signal transducers. Piezoelectric materialsare ideal for use in this application due to their ability to gen-erate and transmit acoustic waves in a frequency-dependent

8 P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13

manner[54]. The physical dimensions and properties ofthe piezoelectric material influence the optimal resonantfrequency for the transmission of the acoustic wave. Themost commonly used piezoelectric materials include quartz(SiO2) and lithium niobate (LiTaO3) [55]. In order to ac-quire an active surface for use in a piezoelectric biosensorthe surface must be stable chemically, contain a high num-ber of the actively immobilised biological elements and thecoating surface should also be as thin as possible[53].

Acoustic wave biosensors offer label-free, on-line anal-ysis for antigen–antibody interactions, and also providethe option of several immunoassay formats, which allowincreased detection sensitivity and specificity. Other advan-tages include cost effectiveness combined with ease of use.Disadvantages associated with these sensors include rela-tively long incubation times for the bacteria and biosensorsurface, problems with crystal surface regeneration and thenumber of washing and drying steps required[1].

Limitations may also be encounted due to difficulties incoating and immobilisation on the crystal surface. Severalimmobilisation methods have been tested. However, since nosingle immobilisation method has proven to be optimum forall piezoelectric sensors[1], this would suggest that a suit-able immobilisation method needs to be established for eachbiological reagent and possibly each sensor format. Babacanet al.[53] have shown that immobilisation of anti-Salmonellaantibodies onto a gold electrode of a piezoelectric quartzcrystal (PQC) through Protein A coupling has proven morereproducible and more stable than coupling of the antibodywith polyethylenimine. Mass balance acoustic wave trans-ducers can be classified into two main groups: (1) bulk wave(BW) devices and (2) surface acoustic wave (SAW) devices.

(1) Bulk wave (BW)The BW device, which is also known as a PQC, is

the oldest and simplest acoustic wave device in opera-tion. This type of acoustic wave transducer may also bereferred to as a quartz crystal microbalance (QCM) ora thickness shear mode (TSM) resonator. QCM is usedin the description of the mass sensitivity of the crystaland TSM refers to the motion of the crystals vibration.This acoustic wave device characteristically transmitsthe acoustic wave from one crystal face to another.

Bulk wave devices consist of parallel circular elec-trodes placed on both sides of a thin cut piece of crystal.An alternating electric field is applied, which results in apotential difference between the two electrodes causingshear deformation of the crystal. This results in me-chanical oscillation of a standing wave, at a character-istic vibrational frequency, across the bulk of the cutquartz. The frequency of the vibrations is dependent onthe physical properties of the crystal, such as size anddensity, and on the properties of the phase in contactwith the crystal surface. The TSM mode of vibration,commonly used by AT and BT cut crystals, is mostsensitive to changes in mass and it results in a displace-

ment parallel to the cut surface. BW devices generallyconsist of a thin disk of AT cut quartz, cut at an angleof +35◦15′ along thez-axis[56], because they are morestable than BT cut crystals over a large range of tem-peratures. This sensor was introduced in immunoassayformat in 1972 by Shons and associates who coatedthe quartz crystal surface with an antigen for the de-tection of specific antibodies[57]. The combination ofPQCs and a biologically active element saw the intro-duction of a new generation of biospecific interactionanalysis[58].

Over the past few years a number of piezoelectricbiosensors formats have been developed for the detec-tion of several microbial contaminants. RecombinantS.enteritidis proteins have been coated onto AT cut quartzcrystals and used in the detection of anti-S. enteritidisantibodies in chicken serum and egg white samples forthe development of a piezoelectric biosensor forS. en-teritidis. This sensor provided a label-free method forthe diagnosis ofS. enteritidis with a clinical sensitivityof 100% and a specificity of 92.9%. When comparedwith ELISA, it was also shown that this sensor wascapable of estimating anti-S. enteritidis antibody titres[58]. Pathirana et al.[57] have also been working on thedevelopment of a piezoelectric biosensor for the detec-tion of Salmonella. In this case, a polyclonal antibodydirected againstS. typhimurium was immobilised ontothe quartz crystal surface using the Langmuir–Blodgettmethod. This sensor was capable of detecting a fewhundred cells in the less than 100 s[57]. A sandwichand enzymatically amplified piezoelectric sensor wasdeveloped for the detection ofHelicobacter pylori.In this case recombinantH. pylori antigens were im-mobilised onto the surface of AT cut quartz crystals.Following incubation with positive human serum sam-ples a less than significant signal was obtained whencompared with the negative background. In order toincrease the positive signal and reduce the negativebackground incubation with anti-human conjugates fol-lowed, resulting in a sandwich immunocomplex. Thepositive signal was further enhanced on addition ofthe conjugates substrate, which resulted in depositionof precipitates on the crystal surface following theenzymatic reaction, leading to greater sensitivity[59].

Work has also focused on the development of piezo-electric biosensors based on the detection of microbialDNA sequences. In one case a 23 bp probe, specificto Aeromonas hydrophila, was immobilised onto thesurface of a streptavidin-coated gold surface of a quartzcrystal. This sensor was capable of detecting a PCRproduct amplified from a specific gene ofA. hydrophilaand of distinguishing between samples that containedthe gene and samples that did not. The hybridisationwas detectable using a fragment of DNA of 205 bp withsynthetic oligonucleotides of 23 bp[60]. This type ofsensor format may offer an ideal opportunity for the

P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13 9

detection of species/strain-specific microbes by choos-ing different DNA probes.

Piezoelectric sensors surfaces can be immobilisedwith antibodies specific to whole microbial cells,surface-expressed antigens or even microbial toxins. Inone case a flow-injection sensor was set up for the de-tection of theStaphylococcal Enterotoxin B using poly-clonal antibodies[61]. Sensor surfaces have also beencoated with specific antigens for the detection of serumantibodies and with microbial DNA probes for thedetection of strain-specific PCR amplified sequences.These sensors have also been used in the detection ofseveral other microbial contaminants includingCan-dida albicans [62], L. monocytogenes [63,64]andVibrocholerae [65].

(2) Surface acoustic wave (SAW) devicesThis type of acoustic wave device transmits along a

single crystal face from one location to another (Fig. 4).In this case the electrodes are on the same side of thecrystal and the transducer can act as both transmitterand receiver. An excited wave travels across the crys-tal face and the physical deformation of the wave isrestricted to the crystal surface. SAW sensors have theability to directly sense changes in mass and mechanicalproperties. SAW have proven to be more sensitive thanPQC sensors. However, numerous problems have beenencountered when applying it to a biological sensingsystem because the surface acoustic wave can becomeseverely decreased in biological solutions[56]. For thisreason SAW sensors were rarely used in the past forthe detection of microbial contaminants.

Research on the use of a dual channel surface acous-tic wave device for the detection ofLegionella andEscherichia coli was reported[66]. In this case the bac-terial cells were coated on the sensor surface and used

Fig. 4. Surface acoustic wave sensor[68], consisting of a cut crystal with two electrodes on the same crystal face acting as both transmitter and receiver.

in the detection of specific antibodies. Further workhas also been done in an attempt to rule out the prob-lems associated with the use of SAW devices in liquidphases. In one case a SAW sensor surface was coatedwith a layer of SiO2, protecting the electrodes fromliquid, resulting in a contactless surface acoustic wavebiosensor. However, further work will be required to in-crease the sensor sensitivity before it can be applied tothe detection of specific microbial contaminants[67].

1.4. Amperometric sensors

Amperometric detection of micro-organisms involvesthe measurement of the current generated through electro-oxidation/reduction catalysed by their enzymes, or by theirinvolvement in a bioaffinity reaction at the surface of theworking electrode. The potential of the working electrodeis maintained with respect to a reference electrode, usu-ally Ag/AgCl, which is at equilibrium. The most commonworking electrodes are noble metals, graphite, modifiedforms of carbon or conducting polymers[1]. Amperom-etry provides a linear concentration dependence over adefined range. These systems offer the advantage that theycan be very sensitive, are usually small and robust, arerapid and economical, and are easily used outside the lab-oratory environment. However, amperometric sensors cansuffer from poor selectivity, especially when applying po-tentials for oxygen/hydrogen peroxide detection, but thiscan usually be overcome by using mediators (e.g. iodine,ferrocyanide, etc.) and perm-selective membranes. Hasebeet al. [69] describe a tyrosinase-based chemically ampli-fied biosensor for the detection ofE. coli. They employ atyrosinase-coupling electrode to detect polyphenolic com-pounds produced microbially from salicyclic acid. Thepolyphenolic compounds are enzymatically oxidised to

10 P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13

Fig. 5. Ion channel formed by antibody embedded in planar lipid bilayerfor detection ofCampylobacter species[72].

o-quinones by dissolved oxygen in a reaction catalysed bytyrosinase. Theo-quinones are then recycled to the originalpolyphenols byl-ascorbic acid and repeatedly reoxidised,which amplifies the signal. The sensor has shown promise inthe capability to carry out sample analysis when presentedwith wastewater samples. However, if the samples containany inherent tyrosinase-active compounds, these could haveto be eliminated prior to analysis. The sensor is capable ofdetecting 103–104 cells/ml but requires a 3 h pre-enrichmentperiod. Various microbial metabolic pathways also producepolyphenolic compounds, which may cause interference,but no cross-reactivity studies were carried out to determinethe specificity of the sensor forE. coli.

A variation to the traditional enzyme-based sensors wasintroduced over 20 years ago by Gehring et al.[70], wherethe specificity of an immunological reaction is coupled withthe simplicity and rapidity of electrochemical detection. Im-munosensors offer the potential to increase sensitivity andselectivity but lack a form of inherent signal amplification[71]. Ivnitski et al. [72] have developed an amperometricimmunosensor based on a supported planar lipid bilayer forthe detection ofCampylobacter (Fig. 5). They employ astainless steel working electrode, whose tip is covered withan artificial bilayer membrane, in which anti-Campylobacterantibodies are embedded, creating a channel. Upon forma-tion of an antibody–antigen (bacteria) complex, the mem-brane alters, allowing ions to pass through. The ions migrateaccording to the difference in voltage and ion concentrationon both sides of the bilayer, creating a current, which canbe amperometrically detected. The bilayer lipid membraneserves several functions. In addition to acting as a matrixfor antibody, it also acts as a very thin electrical insulatorand suppresses non-specific binding of ligand. The biosen-sor has a strong signal amplification effect, defined as thetotal number of ions transported across the bilayer. Thisis 1010 ions/s, giving it a theoretical detection limit of onebacterium in an assay time of just 10 min. The sensor mayhave potential for the detection of other micro-organismsprovided that high quality antibodies are available that arespecifically reactive against them.

Brewster and Mazenko[73] have developed an immu-noelectrical sensor, that coupled with filtration capture,

allows rapid detection ofE. coli O157:H7. Cells are in-cubated with an enzyme-labelled antibody for 15 min andthe antibody–antigen (bacteria) complex, captured on acellulose acetate filter. The filter is brought in contactwith the electrode surface and 5 min after substrate addi-tion, cells can be detected by the conversion of substrate(para-aminophenyl phosphate) to an electroactive product(para-aminophenol). The sensor has a detection limit of5000 cells/ml in an assay time of 25 min.

Immunomagnetic beads have also been used to increasethe selectivity of amperometric biosensors[70]. This tech-nique requires thatSalmonella typhimurium is sandwichedbetween antibody coated magnetic beads and an enzyme(alkaline phosphatase) labelled antibody. A magnet is thenused to localise the beads onto the surface of a disposablegraphite ink electrode in a multiwell plate format. Cells aredetected by the oxidation of the electroactive enzyme prod-uct. This offers a LOD of 8×103 cells/ml in buffer in a totalanalysis time of 80 min.

Rishpon and Ivnitski[74] have developed a separation-freeamperometric enzyme-channelling immunosensor for thedetection ofS. aureus. The immunosensor consists of a car-bon electrode with anti-Protein A antibodies, which detectS.aureus, and the enzyme glucose oxidase immobilised on itssurface. The test solution containingS. aureus, a horseradishperoxide (HRP)-labelled anti-Protein A antibody and iodideis then added. Upon antibody–antigen–antibody sandwichformation, hydrogen peroxidase is brought into contactwith glucose oxidase at the electrode surface. A glucosesolution is then added which reacts with glucose oxidaseresulting in the formation of hydrogen peroxidase (H2O2).HRP catalyses the reaction of H2O2 and iodide ions. Thereduction of iodide ions at the electrode surface allowingdetection ofS. aureus to a limit of 1000 cells/ml of pureculture in 30 min. This assay format offers the advantageof being ‘pseudohomogenous’, i.e. without a wash step, asthe polyethylenimine (PEI) membrane used discriminatesbetween the analytical signal from the bound immunocom-plex and background noise from HRP–H2O2 reacting insolution away from the electrode. Signal amplification isalso facilitated through enzyme channelling.

Flow injection increases the potential for assay automa-tion. Pérez and Mascini[75] developed an amperometricflow-injection sensor with immunomagnetic separation forthe detection of viableE. coli O157:H7 cells. Antibody-coated beads are used to capture theE. coli cells from thematrix. Amperometric detection was carried out by the ox-idation of two mediators (potassium hexacyanoferrate (III)and 2,6-dichloro-phenolindophenol) in phosphate bufferedsaline–potassium chloride (PBS–KCl). The LOD was105 CFU/ml in a time of 2 h with pre-enrichment necessaryfor low numbers of bacteria. This has the advantage of de-tecting viable cells and labelled antibodies are not required.

Flow-injection immunofiltration increases the ratio ofsurface area of immunosorbent to sample volume, offer-ing increased antibody–antigen encounter and accelerated

P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13 11

binding kinetics. This sensor format offers greater antibodybinding capacity, higher sensitivity, a reduction in assaytime and simplicity of operation. Abdel-Hamid et al.[76]developed a flow-injection amperometric immunofiltrationassay for the detection ofE. coli O157:H7. In this systemthe test solution containing the antigen (bacteria) is filteredthrough an antibody-coated filter membrane. A sandwichimmunoassay scheme is created by the addition of an en-zyme (HRP)-conjugated antibody. The amperometric signalproduced by the reduction of iodine at the working electrodeis proportional to the antibody–antigen–antibody sandwichcomplexes present on the filter. Using this technique a lowerLOD for E. coli O157:H7 of 100 cells/ml, with a workingrange of 100–600 cells/ml, in an overall analysis time of30 min could be achieved. The sensor was later adaptedfor other micro-organisms and was further developed forthe detection of totalE. coli and totalSalmonella. A LODof 50 cells/ml for each organism could be achieved in ananalysis time of 35 min[77].

1.5. Potentiometric sensors

A potentiometric biosensor consists of a perm-selectiveouter layer and a bioactive material, usually an enzyme. Theenzyme-catalysed reaction generates or consumes a species,which is detected by an ion selective electrode. Potentiome-try provides a logarithmic concentration dependence. A fieldeffect transistor (FET) is a device where the transistor ampli-fier is adapted to be a miniature transducer for the detectionand measurement of the potentiometric signal produced by asensor process at the gate of the FET. This allows miniatur-isation and increased sensitivity due to the minimal amountof circuit wiring. The recently developed light addressablepotentiometric sensor (LAPS), based on the FET has provedto be suitable for detection of microbial contamination[1].It consists of an n-type silicon semiconductor-based sensorand an insulating layer that is in contact with an aqueous so-lution where an immunoreaction takes place. Changes in po-tential at the silicon-interface are detected by the differencein charge distribution between the surface of the insulatorand the FET. A LAPS measures an alternating photocurrentgenerated by a light source, such as a light emitting diode(LED), so that changes in potential can be transduced intovoltage per time differentials.

Using this technology a commercially available LAPScalled the Threshold Immunoassay System® is availablefrom molecular devices (Fig. 6). This exploits the naturalaffinity between streptavidin and biotin. An immunocomplexis formed in the solution phase between sample (antigen),a labelled antibody, biotinylated antibody and streptavidin,which is then captured onto solid phase biotinylated filter.Upon addition of substrate the filter is brought in contactwith the silicon semiconductor and the enzyme generates apotentiometric signal.

Dill et al. [78] have exploited this technique for the de-tection ofY. pestis andBacillus globigii spores with a lower

Fig. 6. Schematic representation of immunocomplex formed in theThreshold® Immunoassay System.

LOD of 10 cells or spores per sample. They later usedthe same technology to detectS. typhimurium to levels aslow as 119 CFUs[79]. Real sample analysis was simulatedwith chicken carcass washings spiked withSalmonella. Al-though, problems were encountered with sample debris a re-covery of 90% was achieved indicating a potential for rapid(15 min) field monitoring. Gehring et al.[80] have used thistechnique to assay for liveE. coli O157:H7 with a LOD of2.5 × 104 cells/ml.

The US Department of Defence have introduced the Bio-logical Integrated Detection System for providing detectionof airborne biological threats. One of the instruments incor-porated in this is the Biological Detector, which works onthe same principle as the Threshold Immunoassay System®.This system is capable of detecting eight agents simulta-neously within 15 min. The LOD forB. subtilis is 3 ×103 CFU/ml [81]. This technique offers potential for in thefield detection although limits of detection need some im-provement (Figs. 5 and 6).

2. Conclusion

Although conventional methods for the detection andidentification of microbial contaminants can be very sen-sitive, inexpensive and present both qualitative and quan-titative information, they can require several days to yieldresults. Biosensors offer an exciting alternative to the moretraditional methods, allowing rapid “real-time” and multipleanalyses that are essential for the detection of bacteria infood, especially perishable or semi-perishable foods. While

12 P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13

the sensitivity of each of the sensor systems discussed in thisreview may vary depending on the transducer’s properties,and the various biological elements, improvements in flowcell design, transducer sensitivity, miniaturisation, and anti-body design and production, will see an increase in highlysensitive, fully automated and inexpensive biosensors beingroutinely used both in the field and the laboratory, for therapid detection of micro-organisms.

Acknowledgments

We wish to acknowledge the support of Enterprise Ireland,the EU funded INCO COPERNICUS project, PL 979012,and the National Centre for Sensor Research (funded underthe Higher Educational Authority Programme for Researchin Third Level Institutions).

References

[1] Invitski D, Abdel-Hamid I, Atanasov P, Wilkins E. Biosensors for thedetection of pathogenic bacteria. Biosens Bioelectron 1999;14:599–624.

[2] Anon. Waterborne pathogens kill 10M–20M people/year. WorldWater Environ Eng 1996 June;6.

[3] Donnelly CW.Listeria monocytogenes: a continuing challenge. NutrRev 2001;59:183–94.

[4] Schlech III WF. Foodborne listeriosis. Clin Infect Dis 2000;31:770–5.

[5] De Boer E, Beumer RR. Methodology for detection and typing offoodborne microorganisms. Int J Food Microbiol 1999;50:119–30.

[6] Artault S, Blind JL, Delaval J, Dureuil Y, Gaillard N. DetectingListeria monocytogenes in food. Int Food Hyg 2001;12(3):23.

[7] Xu HS, Roberts N, Singelton FL, Attwell RW, Grimes DJ, ColwellRR. Survival and viability of non-culturableEscherichia coli andVibrio cholerae in the estuarine and marine environment. MicrobiolEcol 1982;80:313–23.

[8] Porter J, Edwards C, Pickup RW. Rapid assessment of physiologicalstatus inEscherichia coli using fluorescent probes. J Appl Bacteriol1995;79:399–408.

[9] Toze S. PCR and the detection of microbial pathogens in water andwastewater. Water Res 1999;33:3545–56.

[10] Bsat N, Weidmenn M, Czajka J, Barany F, Piani M, Batt CA.Food safety applications of nucleic acid-based assays. Food Technol1994;48:142–5.

[11] Schwab KJ, De Leon R, Sobsey MD. Immunoaffinity concentrationand purification of waterborne enteric viruses for detection by reversetranscriptase PCR. Appl Environ Microbiol 1996;63:4401–7.

[12] Dubois E, Le Guyader F, Haugarreau L, Kopecka H, CormierM, Pommepuy M. Molecular epidemiological study of rotavirusesin sewage by reverse transcriptase seminested PCR andrestriction fragment polymorphism assay. Appl Environ Microbiol1997;63:1794–800.

[13] Traore O, Arnal C, Mignotte B, Maul A, Laveran H, Billaudel S,et al. Reverse transcriptase PCR detection of astrovirus, hepatitisA virus and poliovirus in experimentally contaminated mussels:comparison of several extraction and concentration methods. ApplEnviron Microbiol 1998;64:3118–22.

[14] Jensen M, Webster JA, Straus N. Rapid identification of bacteriaon the basis of polymerase chain reaction-amplified ribosomal DNAspacer polymorphisms. Appl Environ Microbiol 1993;59:945–52.

[15] Fach P, Popoff MR. Detection of enterotoxigenicClostridiumperfringens in food and fecal samples with duplex PCR and slidelatex agglutination test. Appl Environ Microbiol 1997;63:4232–6.

[16] Tsen HY, Lin CK, Chi WR. Development and use of 16S rRNAgene targeted PCR primers for the identification ofEscherichia coliin water. J Appl Microbiol 1998;61:554–60.

[17] Rochelle PA, De Leon R, Stewart MH, Wolfe RL. Comparisonof primers and optimization of PCR conditions for detection ofCrypotosporidium parvum and Giardia lamblia in water. ApplEnviron Microbiol 1997;63:106–14.

[18] Stinear T, Matusan A, Hines K, Sandery M. Detection of a singleviable Crypotosporidium parvum oocyst in environmental waterconcentrates by reverse transcription-PCR. Appl Environ Microbiol1996;62:3385–90.

[19] Geary TG. Approaching helminth biology from the moleculardirection. Parasitol Today 1996;12:373–5.

[20] Mathis A, Deplazes P, Eckert J. An improved test system forPCR-based specific detection ofEchinococcus multilocularis eggs. JHelminthol 1996;70:219–22.

[21] Iqbal SS, Mayo MW, Bruno JG, Bronk BV, Batt CA, ChambersP. A review of molecular recognition technologies for detection ofbiological threat agents. Biosens Bioelectron 2000;15:549–78.

[22] Cahill D, Roben P, Quinlan N, O’Kennedy R. Production ofantibodies. In: Townsend A, editor. Encyclopedia Analytical Science,vol. IV. New York: Academic Press; 1995, p. 2057–66.

[23] Köhler G, Milstein C. Continuous cultures of fused cells secretingantibody of predefined specificity. Nature 1975;310:792–4.

[24] Chaplin MF, Bucke C. Enzyme technology. 1st ed. England:Cambridge University Press; 1990.

[25] Goodridge L, Griffiths M. Reporter bacteriophage assays as a meansto detect foodborne pathogenic bacteria. Food Res Int 2002;35:863–70.

[26] Benhar I, Eshkenazi I, Neufeld T, Opatowsky J, Shaky S, Rishpon J.Recombinant single chain antibodies in bioelectrochemical sensors.Talanta 2001;55:899–907.

[27] Goldman ER, Pazirandeh MP, Mauro JM, King KD, Frey JC,Anderson GP. Phage-displayed peptides as biosensor reagents. J MolRecognit 2000;13(6):382–7.

[28] Fitzpatrick J, Fanning L, Hearty S, Leonard P, Manning BM, QuinnJG, et al. Applications and recent developments in the use ofantibodies for analysis. Anal Lett 2000;33(13):263–2609.

[29] Goepel W. Chemical sensing, molecular electronics andnanotechnology: interface technologies down to the molecular scale.Sens Actuat B 1991;4:7–21.

[30] Seyhi RS. Transducer aspects of biosensors. Biosens Bioelectron1994;9:243–64.

[31] Goepel W, Heiduschka P. Interface analysis in biosensor design.Biosens Bioelectron 1995;10:853–83.

[32] Homola J, Yee SS, Gauglite G. Surface plasmon resonance sensors:review. Sens Actuat I 1999;54:3–15.

[33] Liedberg B, Nylander C, Lundstrom I. Biosensing with surfaceplasmon resonance—how it all started. Biosens Bioelectron1995;10:1–9.

[34] Salamon Z, Brown MF, Tollin G. Plasmon resonance spectroscopy:probing molecular interactions within membranes. Trends Biochem1999;24:213–9.

[35] Tubb AJC, Payne FP, Millington RB, Lowe CR. Single modeoptical fibre surface plasmon wave chemical sensor. Sens Actuat B1997;4:71–9.

[36] Quinn JG, O’Kennedy R. Transduction platforms and biointerfacialdesign of biosensors for real-time biomolecular interaction analysis.Anal Lett 1999;32(8):1475–517.

[37] Kretschmann E. The determination of the optical constants of metalsby excitation of surface plasmon resonans. Z Phys 1971;241:313–24.

[38] http://www.BIAcore.com.[39] http://www.affinity-sensors.com.[40] http://www.xantec.com.

P. Leonard et al. / Enzyme and Microbial Technology 32 (2003) 3–13 13

[41] http://www.ti.com/spr/.[42] Elkind JL, Stimpson DI, Strong AA, Bartholemew JL, Melendez JL.

Integrated analytical sensors, the use of the TISPR-1 as a biosensor.Sens Actuat B 1999;54:182–90.

[43] http://www.Biotul.com/.[44] http://www.biosensor.com.[45] Fratamico PM. Detection ofEscherichia coli O157:H7 using a surface

plasmon resonance biosensor. Biotechnol Tech 1998;12(7):571–6.[46] Koubava V, Brynda E, Karasova L, Skvor J, Homola J, Dostalek

J, et al. Detection of foodborne pathogens using surface plasmonresonance biosensors. Sens Actuat B 2001;74:100–5.

[47] BIAcore AB biotechnology note; 1995.[48] Quinn JG, O’Kennedy R. Detection of whole cell:antibody

interactions using BIAcore’s SPR technology. BIAJ 2001;1:22–4.[49] Kaclikova E, Kuchta T, Kay H, Gray D. Separation ofListeria from

cheese and enrichment media using antibody-coated microbeads andcentrifugation. J Immunol Methods 2001;46:63–7.

[50] Haines J, Patel PD. Detection of foodborne pathogens using BIA.BIAJ 1995;2(2):31.

[51] Perkins EA, Squirrell DJ. Development of instrumentation to allowthe detection of microorganisms using light scattering in combinationwith surface plasmon resonance. Biosens Bioelectron 2000;14:853–9.

[52] Hughes MP, Morgan H. Dielectrophoretic trapping of single sub-micrometre scale bioparticles. J Phys D: Appl Phys 1998;31:2205–10.

[53] Babacan S, Pivarnik P, Letcher S, Rand AG. Evaluation of antibodyimmobilisation methods for piezoelectric biosensor application.Biosens Bioelectron 2000;15:615–21.

[54] Griffiths D, Hall G. Biosensors—what real progress is being made?TIBTECH 1993;11:122–30.

[55] http://www.sensorsmag.com/articles/1000/68/main.html.[56] Bunde RL, Jarvi EJ, Rosentreter JJ. Piezoelectric quartz crystal

biosensors. Talanta 1998;46:1223–36.[57] Pathirana ST, Barbaree J, Chin BA, Hartell MG, Neely WC,

Vodyanoy V. Rapid and sensitive biosensors forSalmonella. BiosensBioelectron 2000;15:135–41.

[58] Su X, Low S, Kwang J, Chew VHT, Li SFY. Piezoelectric quartzcrystal based veterinary diagnosis forSalmonella enteritidis infectionin chicken and egg. Sens Actuat B 2001;75:29–35.

[59] Su X, Li SFY. Serological determination ofHelicobacter pyloriinfection using sandwiched and enzymatically amplified piezoelectricbiosensor. Anal Chim Acta 2001;429:27–36.

[60] Tombelli S, Mascini M, Sacco C, Turner APF. A DNA piezoelectricbiosensor assay coupled with a polymerase chain reaction forbacterial toxicity determination in environmental samples. Anal ChimActa 2001;418:1–9.

[61] Harteveld JLN, Nieuwenhuizen MS, Wils ERJ. Detection ofStaphylococcal Enterotoxin B employing a piezoelectric crystalimmunosensor. Biosens Bioelectron 1997;12:661–7.

[62] Maramatsu H, Kajiwara K, Tamiya E, Karube I. Piezoelectricimmunosensor for the detection ofCandida albicans microbes. AnalChim Acta 1986;188:257–61.

[63] Jacobs MB, Cater RM, Lubrano GJ, Guilbault GG. A piezoelectricbiosensor forListeria monocytogenes. Am Lab 1995;27(11):26–8.

[64] Vaughan RD, O’Sullivan CK, Guilbault GG. Development of aquartz crystal microbalance (QCM) immunosensor for the detectionof Listeria monocytogenes. Enzyme Microb Tech 2001;29:635–8.

[65] Cater RM, Mekalanos JJ, Jacobs MB, Lubrano GJ, GuilbaultGG. Quartz crystal microbalance detection ofVibro cholerae 0139serotype. J Immunol Methods 1995;187:121–5.

[66] Howe E, Harding G. A comparison of protocols for the optimisationof detection of bacteria using a surface acoustic wave (SAW)biosensor. Biosens Bioelectron 2000;15:641–9.

[67] Freudenberg J, Schelle S, Beck K, von Schickfus M, Hunklinger S.A contactless surface acoustic wave biosensor. Biosens Bioelectron1999;4:427–9.

[68] http://www.sensorsmag.com/articles/1000/68/main.shtml.[69] Hasebe Y, Yokobari K, Fukasawa K, Kogure T, Uchiyama S. Highly

sensitive electrochemical determination ofEscherichia coli densityusing tyrosinase-based chemically amplified biosensor. Anal ChimActa 1997;357:51–4.

[70] Gehring AG, Crawford CG, Mazenko RS, Van Houten LJ, BrewsterJD. Enzyme-linked immunomagnetic electrochemical detection ofSalmonella typhimurium. J Immunol Methods 1996;195:15–25.

[71] Tiefenauer LX, Kossek S, Padeste C, Thiébaud P. Towards ampero-metric immunosensor devices. Biosens Bioelectron 1997;12(3):213–23.

[72] Ivnitski D, Wilkins E, Tien HT, Ottova A. Electrochemical biosensorbased on supported planar lipid bilayers for fast detection ofpathogenic bacteria. ELECOM 2000;2:457–60.

[73] Brewster JD, Mazenko RS. Filtration capture and immunoelec-trochemical detection for rapid assay ofEscherichia coli 0157:H7.J Immunol Methods 1998;211:1–8.

[74] Rishpon J, Ivnitski D. An amperometric enzyme-channelingimmunosensor. Biosens Bioelectron 1997;12(2):195–204.

[75] Pérez FG, Mascini M. Immunomagnetic separation with mediatedflow injection analysis amperometric detection of viableEscherichiacoli 0157. Anal Chem 1998;70:2380–6.

[76] Abdel-Hamid I, Ivnitski D, Atanasov P, Wilkins E. Flow-throughimmunofiltratin assay system for rapid detection ofE. coli 0157:H7.Biosens Bioelectron 1999;14:309–16.

[77] Abdel-Hamid I, Ivnitski D, Atanasov P, Wilkins E. Highly sensitiveflow-injection immunoassay system for rapid detection of bacteria.Anal Chim Acta 1999;399:99–108.

[78] Dill K, Song JH, Blomdahl JA, Olson JD. Rapid, sensitive andspecific detection of whole cells and spores using the light-addressable potentiometric sensor. J Biochem Biophys Methods1997;34:161–6.

[79] Dill K, Stanker LH, Young CR. Detection of salmonella in poultryusing a silicon chip-based biosensor. J Biochem Biophys Methods1999;41:61–7.

[80] Gehring AG, Patterson DL, Tu S. Use of a light addressablepotentiometric sensor for the detection ofEscherichia coli 0157:H7.Anal Biochem 1998;258:293–8.

[81] Uithoven KA, Schmidt JC, Ballman ME. Rapid identification ofbiological warfare agents using an instrument employing a lightaddressable-potentiometric sensor and a flow-through immunofil-tration-enzyme assay system. Biosens Bioelectron 2000;14:761–70.