9783642363023-c1

Toxin immunosensors and sensor arrays for foodquality control

Simone S. Moises & Michael Schferling

Received: 3 August 2009 /Accepted: 23 October 2009 /Published online: 26 November 2009# Springer-Verlag 2009

Abstract The global efforts to improve consumer protec-tion and public health lead to an increasing number ofanalytical approaches applicable to food analysis andprocess control. Biosensor systems are efficient analyticaltools to monitor production processes or storage of nutritionand to control contamination outbreaks as they are easy-to-use, fast, and with minimal effort on sample preparation.Relevant targets of immunosensors implemented to foodsafety are prevalent bacterial toxins (staphylococcal enter-otoxins and clostridial toxins), plant toxins (Ricin), myco-toxins (aflatoxins and ochratoxin A), marine toxins, andother pathogenic bacterial contaminations (Listeria, Salmo-nella, Staphylococcus aureus, or Escherichia coli). Thesecause acute intoxication and also chronic diseases inhumans consuming contaminated food. Promisingapproaches for the determination of different types oftoxins in food matrices will be outlined. The correspondingsensor systems use immunological receptor units such asantibodies or antigens and include optical (fluorescence andsurface plasmon resonance), electrochemical, or acousticalreadout methods. This review is focused on recent develop-ments of sensor formats devoted to food safety control andis structured according to the type of toxin or contaminantthat is recognized. It is intended to give an overview onemerging sensor technologies and their potential applica-tions for the rapid analysis of the most important foodpoisoning agents.

Keywords Food safety . Toxin . Sensor . Array . SPR .

ELISA

AbbreviationsAP Alkaline phosphataseATR Attenuated total reflectionBoNT Botulinum neurotoxinBSA Bovine serum albuminCCD Charge-coupled deviceCWC Chemical Weapons ConventionDON DeoxynivalenolDPV Differential pulse voltammetryEAPM Electrically active polyaniline-coated magneticECL ElectrochemoluminescenceEDC N-(3-dimethylaminopropyl)-N-

ethylcarbodiimide hydrochlorideELFA Enzyme-linked fluorescent assayELIMCL Enzyme-linked immunomagnetic

chemiluminescenceELISA Enzyme-linked immunosorbent assayFAO Food and Agriculture OrganizationFIA Flow injection analysisFITC Fluorescein isothiocyanateHPLC High-pressure liquid chromatographyHRP Horseradish peroxidaseIARC International Agency for Research on CancerISO International Organization for StandardizationLD50 Lethal dose (50% of the tested organisms die)LED Light-emitting diodeLOD Limit of detectionMED Multichannel electrochemical detectionMS Mass spectrometryNHS N-hydroxysuccinimideNPP NitrophenylphosphateOTA Ochratoxin A

S. S. Moises :M. Schferling (*)Institute of Analytical Chemistry, Chemo- and Biosensors,University of Regensburg,D-93040 Regensburg, Germanye-mail: [email protected]

Bioanal Rev (2009) 1:73104DOI 10.1007/s12566-009-0006-x

19Reprinted from the journal

OWLS Optical waveguide lightmode spectroscopyPCR Polymerase chain reactionPDMS PolydimethylsulfoxidePEG Polyethylene glycolPEMC Piezoelectric-excited, millimeter-sized

cantileverPMMA PolymethylmethacrylatePZT Lead zirconate titanateQCM Quartz crystal microbalanceRPLA Reversed passive latex agglutination assaySPE Screen-printed electrodeSPR Surface plasmon resonanceTMB TetramethylbenzidineTPA TripropylamineWHO World Health Organization

Introduction

Foodborne intoxications are an enduring risk for publichealth, and therefore, the feasibility of producing andconsuming safe foods is considered as one of the majorachievements of the last century. Over 200 known diseasesare transmitted via food consumption [1]. The spectrum offoodborne pathogens includes a variety of viruses, fungiand fungal toxins, chemicals, heavy metals, parasites,bacterial toxins, and bacteria, whereas bacteria-relatedpoisoning is the most prevalent. Only less than 20 differentbacteria act as originators. Every year, Staphylococcusaureus, Salmonella, Clostridium perfringens, Campylobac-ter, Listeria monocytogenes, Vibrio parahaemolyticus,Bacillus cereus, and enteropathogenic Escherichia coli arecausing more than 90 % of all food poisonings that arerelated to known pathogens. These bacteria are mainlyfound in raw foods [2, 3].

The Centers for Disease Control and Prevention [4] inthe USA are collecting data on foodborne disease outbreaksfrom all states and territories through the FoodborneDisease Outbreak Surveillance System to quantify theimpact of these diseases on health. The estimated numberof food-related diseases causes approximately 76 millionillnesses, 323,914 hospitalizations, and 5,194 deaths. Only14 million illnesses, 60,000 hospitalizations, and 1,500deaths are caused by known pathogens, while unknownagents are responsible for the remaining numbers [3].

Outbreak data reported internationally for source attri-bution were collected by Greig and Ravel [5]. Public reportsources reveal that 4,093 outbreaks have been registeredand analyzed between 1988 and 2007. According to thisstudy, 2,168 cases are allotted to the USA, 1,287 to theEuropean Union (EU), 246 to Australia and New Zealand,208 to Canada, and 184 to other countries. Based on a

study from the European Commission Rapid Alert Systemfor Food and Feed, a total of 11,403 reports were publishedbetween July 2003 and June 2007 [6].

Controlling bodies and guidelines are necessary due tothe large number of outbreaks and their impact on publichealth. The EU parliament assumes legislation in the formof directives and regulations for EU member states and isconsulted on food safety affairs by the European FoodSafety Authority [7]. The Food and Drug Administration ofthe USA publishes the Food Code which represents amodel set of guidelines and procedures that assists foodcontrol jurisdictions. It provides a scientifically soundtechnical basis for regulating the retail and food serviceindustry [8]. In 2003, the WHO and FAO published theCodex Alimentarius, which serves as a guideline to foodsafety [9]. In addition, Outbreak Alert Databases [10] andinformation tools for current outbreaks [11] in the USA areavailable online.

As food safety concerns consumers, food producers, andregulatory agencies, widespread concepts through the wholefeed and food chainfarm, transport, supply, and consump-tionare required to protect consumers from pathogeningestion. Hazard Analysis and Critical Control Points is asystematic preventive approach to food and pharmaceuticalsafety which addresses physical, chemical, and biologicalhazards [1214]. In Europe and the USA, a considerablenumber of research projects aiming for new tools for foodsafety were funded. Most of the projects unify research anddevelopment topics such as improved analytical and sam-pling methods with modeling and the compilation ofdatabases. Some projects have also a strictly food chain-dominated structure. The large number of generated data setsaffords refined statistical informatics. An introduction topractical biotraceability is given by Barker et al. [15].

The high relevance attributed to consumer protection isdocumented by the substantial number of integrated EUprojects such as BIOTRACER [16]. It has 46 projectpartners from 24 countries, including four INCO countries,and has a total budget of 15 million euro. Its objective is toprovide tools and computer models for the improvement oftracing accidental and deliberate microbial contaminationsof feed, food, and bottled water. PathogenCombat [17] isanother large integrated project that aims for better controland prevention of merging pathogens at cellular andmolecular level throughout the food chain. PathogenCom-bat includes 20 industrial partners and 24 research partnersfrom 17 countries. The main objective of the TRACE-BACK [18] project states the development of a functioninggeneric system for traceability of proliferations and infor-mation processing within food chains. It involves 28 projectpartners from 11 countries. Finally, CHILL-ON [19]researches for the improvement of quality, safety, andtransparency of the food supply chain. Further information

S.S. Moises and M. Schferling

20 Reprinted from the journal

on more related projects within different FrameworkProgrammes of the EU is available online [20].

The Food Safety Project [21] in the USA providesresources for consumers, food service operators, students,and educators to access research-based, unbiased informa-tion on food safety and quality. Its goal is to developeducational materials that give the public helpful tools forminimizing the risk of foodborne illness and disseminateinformation about food safety.

Rapid and reliable detection methods are essential toolsto process a large amount of samples that accumulate if aconsistent food control shall be achieved. Customarymicrobiological methods such as cell culture techniquesare often laborious and ineffective due to their incompat-ibility with the speed of the production chain and thedistribution of food, its endurance, and the operationalcosts. Furthermore, bacterial strains can fail regular growthprocesses and lead to false analysis results.

Quantitative polymerase chain reaction (PCR) is anaccurate, rapid, specific, and sensitive method for detectionof small amounts of pathogen deoxyribonucleic acid(DNA) in food samples. Unfortunately, DNA-based assayscan only detect the presence of toxin-producing organismsand do not quantify the amount of effective toxins. Onlinedetection with PCR methods is also expensive and requireswell-trained personnel [22].

Typical methods of instrumental analytical chemistrysuch as mass spectrometry, liquid chromatography, infrared,or UV/Vis spectrometry are powerful tools for a precisedetermination of pathogens, but they require time-consuming sample preparation and are usually not trans-portable devices, thus inapplicable for online monitoring,e.g., in the production process.

Sensor-based bioassays and microarray techniques arerapid and sensitive tools for online detection and automatedprocess control during food production and the supplychain. They can also be used in extensive research studies,mass tests, or to generate supporting data for modelingprograms. The results can be used to create new ISO orDIN norms that are significant for a huge number of foodproducers.

This review focuses on new developments in the field ofsensors and arrays for the detection of food pathogens. It isstructured corresponding to the type of targeted toxin toprovide an easier overview on already evaluated sensorapproaches for a certain type of contaminant. The article isfocused on the most relevant targets accounted for in foodsafety. It considers only antibody- and cell-based technol-ogies for the detection of food-contaminating bacteria andtheir corresponding toxins. Thus, it particularly includesimmunochemical recognition units. In this regard, it shouldbe noted that the term immunosensor is frequently usedto illustrate the operating mode, but strictly speaking, these

devices are not sensors as their responses are not reversibleand they are not suited for continuous measurements [23].Other types of biomolecular targets (viral or bacterial DNA)or recognition systems (e.g., molecular imprints) are out ofthe scope of this review.

Common ELISA methods and commercially availableassay kits

Many established analytical methods for a straightforwardand fast determination of toxins applicable to food,clinical, or environmental samples are based on enzyme-linked immunosorbent assays (ELISAs). These can beperformed with a high throughput of tested samples inmicrowell plate formats; hence, commercially availableready-to-use assay kits originate from the classicalsandwich-type ELISA.

ELISAs for toxin determination

Common techniques of instrumental analysis such as massspectrometry, gas chromatography, high-pressure liquidchromatography (HPLC), or gel electrophoresis are oftentime-consuming, expensive, and their processing demandsprofessionally trained personnel. Immunochemical methodssuch as ELISA fulfill the requirements of modern high-throughput analysis, including rapidness, specificity, sensi-tivity, multiplicity, and the availability of simple andcost-effective equipment. Moreover, the whole analysiscan be carried out in portable instruments. These advan-tages turned ELISAs into well-established, high-throughputassays with low sample volume requirements and often lessextensive sample preparation steps compared to conven-tional instrumental methods such as quantitative HPLC.Numerous assays targeting on the most frequent toxincontaminations in food matrices have been developed assummarized in Table 1. Figure 1 depicts a schematicoverview over the basic ELISA formats.

A variant of the classical ELISA is the dipstick sandwichassay (Fig. 2) which has been established for the detectionof staphylococcal enterotoxins (SEs). Capture antibodiesfor SE B are deposited on a polystyrene stick which isdipped in homogenized cheese samples contaminated withSE B. A typical sandwich assay is performed by subsequentaddition of primary antibodies conjugated to horseradishperoxidase (HRP). The assembly is transferred to a tubecontaining H2O2 and o-diaminobenzene dihydrochloride.This substrate is converted enzymatically into a blue-colored reaction product in the presence of HRP. Theoptical density of the solution can be read out at 490 nmwith a commercial fiber-optic probe after 510 min reactiontime [33].

Toxin immunosensors and sensor arrays for food quality control


The immunomagnetic separation assay (Fig. 3) utilizesdispersed magnetic particles (Magnetic Dynabead M280)[40] pre-coated with polyclonal capture antibodies specificfor C. perfringens enterotoxin A. For performance of theassay, target enterotoxins included in the sample wereplaced into microwell plates blocked with bovine serumalbumin (BSA) and incubated therein with the specificantibody-coated magnetic particles. After binding of toxinsto the particles, they were concentrated to the bottom of themicrotiter plates with a magnetic concentrator. Following

concentration of loaded particles and subsequent washingsteps, a biotinylated enterotoxin-specific target antibodywas added, building the base of the enzyme anti-enterotoxin complex. In order to ensure quantitativecomplex formation, a preformed avidinbiotin alkalinephosphatase complex (AP) or streptavidinbiotin HRPcomplex was added and incubated. To quantify the amountof toxin in the sample, p-nitrophenylphosphate or azino-benzthiozoline sulfonate substrate was used followingparticle concentration to the bottom of the wells with a

Table 1 ELISAs developed for toxin detection in food

Analyte ELISA type Food matrix LOD Ref.

Ochratoxin A Direct competitive Maize, barley, soy 0.5 ng/g [24]

Listeria monocytogenes Sandwich (direct, indirect,biotin-amplification)

Low-fat milk 1103 cells/mL;3103 cells/mL

[25]

C. perfringens Immunomagneticseparation (IMS)

Meat 2.5 ng/mL [26]A enterotoxin

SEA Sandwich andcompetitivea

Ham pat, turkey pat, milk, sausage,potato mayonnaise, cheese and lemon cake

0.5 ng/g [27]

Aflatoxin M1 Indirect competitive Milk, milk based-confectionery 0.5 ng/mL [28]

Streptomycin, S-derivate,Gentamycin, neomycin

Different Milk ns [29]

SEA Microsphere-packedcapillary/Sandwich

Ham, cheese, chicken, bean sprouts 1 ng/g [30]

SEH Sandwich Mashed potato (raw milk), sausage, bulk milk 55.5 ng/g [31]

SEA-SED Indirect double sandwich Cheese, chicken, salad, milk, cream 1 ng/g [32]

SEB Dipstick sandwich Cheese 0.51 ng/mL [33]

Aflatoxin M1 Supermagneticnanoparticles (directcompetitive)

Milk 4 ng/mL [34]

C. botulinum Amplified/sandwich Soft drinks, juices, bottled water, vanilla extract, ice cream,honey, milk, legumes, spices, turkey, sausages, beef

2 ng/mL [35]BoNT

A, B, E,F

C. botulinum BoNT/B endopeptidase Pate, cheese, cod, mince, sausage 10 pg/mL [36]BoNT B

C. botulinum BoNT/B endopeptidase Pate, cheese, cod, mince, sausage, yoghurt 5 pg/mL [37]BoNT B

Fumonisin B1 Indirect competitive Corn 5 ng/g [38]

Deoxynivalenol,Nivalenol, T-2, H-2

Competitive Wheat kernels 80 ng/g DON [39]30 ng/g T-2

LOD limit of detection, ns not specifieda Sandwich and competitive ELISA or used for different matrices

Fig. 1 Different types ofELISA assays



magnetic concentrator. The colored reaction mixture wastransferred to a new microtiter plate, and absorbance wasmeasured at 414 nm [26].

Sensitive assays for Botulinum neurotoxins use theirhigh endoprotease activity. These enzymes cleave differentisoforms of neuronal proteins, controlling the docking ofthe synaptic vesicles to the synaptic membrane. Syntheticpeptide substrates immobilized on a solid phase of acolumn can be used for the determination of Clostridiumbotulinum neurotoxin B (BoNT B) in food samples. Thetoxin cleaves the peptides, which are additionally biotiny-lated at the terminal position, after addition to the column.The eluate containing the released peptide fragment istransferred to streptavidin-coated microwell plates anddetected with specific antibodies conjugated to HRP [36].

Commercialized detection kits

The necessity of very rapid standardized tests raises thedemand for commercially available (semi-)automatedready-to-use assay kits. These are typically based onELISAs or related methods such as the enzyme-linkedfluorescent assay (ELFA) or the reversed passive latexagglutination assay (RPLA). There are only a few exampleswhere they were designed to distinguish between differenttypes of toxins of one species (e.g., the different Staphy-lococcus enterotoxins). The results of the RPLA can beread out by the naked eye. The agglutination of latex can bemonitored in case of toxin attendance. ELISAs and ELFAsare quantified by means of photodetectors, measuring theresulting absorbance or fluorescence intensity, respectively[41].

The widely used TECRA, RIDASCREEN, Veratox,AGRAQUANT, or Transia kits are generally derived fromthe ELISA principle, whereas the VIDAS system affiliatesthe category of ELFAs (Table 2).

Other typical sensor formats

Definitions, classifications, and essential components ofbiosensors are summarized by Wolfbeis [56]. In thischapter, the most important technologies applied in sensortechnology will be outlined.

Fiber-optic sensors

Fiber-optic sensors use optical waveguides connected to abiofunctionalized interface as sensing element or as signalprocessing element (transducer) receiving signals from aremote sensor. Fibers are preferentially used because oftheir small size, and they do not call for electrical power atthe point of measurement. Furthermore, many sensors canbe multiplexed along the fiber length using differentwavelengths for each sensor.

Waveguides with integrated optical frequency filters canbe used as sensor elements to measure temperature,pressure, and other physical quantities. Thermal or physicaldeformation of the fiber leads to a modulation of thewavelength, intensity, polarization, phase, or transit time ofthe light that leaves the fiber.

Fiber-optic sensors can also be coated with chemicalsensor layers, e.g., for oxygen or pH measurement [57], orbiological recognition elements such as proteins, antibodies,or DNA. Their binding to target molecules from the samplecan be monitored by fluorescence or ELISA-based methodsor by the change in the optical properties (e.g., the refractionindex) of the sensor interface [56, 58].

Surface plasmon resonance

If the surface of a glass substrate is coated with a thin metalfilm, a part of the incident light can refract into the metallicfilm. Typical coatings consist of noble metals such as silver,

Fig. 2 Dipstick sandwich assay scheme [33]

Fig. 3 Immunomagnetic sepa-ration assay scheme [26]



gold, copper, titanium, and chromium. In this assembly, asecond critical angle exists which is greater than the angleof total reflection. At this angle, the surface plasmonresonance angle, a loss of light appears and the intensityof reflected light reaches a minimum. This results from theinteraction of the incident light with oscillation modes ofmobile electrons at the surface of the metal film. Theseoscillating plasma waves are called the surface plasmons.The electron oscillations resonate if the wave vector of theincident light aligns this wavelength (surface plasmonresonance). This resonance of incident light with theplasmons reduces the intensity of reflected light due to aloss of energy. The resonance frequency of the surfaceplasma and the critical surface plasmon resonance (SPR)angle depend on the dielectric coefficient of an adjacentlayer on the metal surface. If this metal surface is coated bya thin layer of affinity ligands, the binding of biomolecules,

e.g., proteins, causes a change of the refractive index. Thisis detected by a shift in the resonance angle.

The frequently used Kretschmann configuration (Fig. 4)is based on a metal film which is evaporated on one face ofa glass prism. The light is coupled into the prism above thecritical angle of total reflection, and the resulting evanes-cent wave penetrates the metal film. The plasmons areexcited at the outside of the film. SPR reflectivity measure-ments can be used for the detection of specific molecularinteractions of bound receptor molecules on the metalsurface with their corresponding targets (e.g., DNA orproteins) [58, 59].

Evanescent wave methods

If a light beam traverses from a material with a high indexof refraction such as glass into one with a low index (water

Table 2 Examples of toxin determination in food with commercially available assay kits

Analyte KIT type Food matrix LOD Ref.

Staphylococcalenterotoxins

VIDAS SET2; Transia plate SET;SET-RPLA

Cake, liquid milk, cheeses from cow,goat, ewe, unspecific species

62.5500 pg/mL

[42]

SEA-SEC TECRA; RPLA Ham, cheese, mushrooms, salami, pasta 0.751.25 ng/mL

[43]

SEA-SEE RIDASCREEN SET; Noodle, ham, cheese, turkey, salami 200750 pg/g

[44]TECRA EIA

SEA-SEE VIDAS SET; VIDAS SET2; Transia plate;Staphylococcal enterotoxins Diffchamb

Cooked food, canned food, liquid food,raw meat, dehydrated food, Delicatessen meat

< 0.51 ng/g

[45]

Salmonella(typhimurium, derbi,enteritidis, typhi)

BAX system; TECRA unique Salmonella Raw pork, raw poultry, 9 major ready-to-eatfood groups

10 cfu/25 gsample

[46]

Ochratoxin A, T-2 Collaborative study kits Maize, wheat, barley, rye 4 mg/kgOTA

[47]

50 mg/kgT-2

Ochratoxin A RIDASCREEN OTA ELISA Coffee beans 25 ng/kg [48]

E. coli O157 TECRA E. coli O157 visual immunoassay Cheese, milk powder casein, whey products ns [49]

E. coli O157 VIDAS E. coli; Dynabeads Raw milk cheeses, poultry, raw sausage,ground beef

ns [50]

Listeria spp. TECRA Listeria visual immunoassay Lettuce, ice cream, fish fillets, cooked chicken,cooked ground turkey

ns [51]

Listeria spp. Palcam; VIDAS Raw dairy, raw egg, raw fish, raw meat, rawvegetables

ns [52]

Bacillus cereus BCET-RPLA; Noodles, spices, grains, legumes,legumes products, cooked foods

ns [53]enteroroxin TECRA

Deoxynivalenol RIDASCREEN DON; Veratox5/5 DON;Deoxynivalenol EIA; AgraQuantDON Assay 0.25/5.0 test kit

Beer, wheat ns [54]

Aflatoxin B1, B2, G1,G2

AgraQuant 4-40 ppb Corn, corn meal, corn gluten feed, corn germ,corn gluten meal, corn/soy blend,popcorn, sorghum, wheat, milled rice,soybeans, peanuts, cottonseed

2.216.4 ppb

[55]

ns not specified



or liquids), a part of the light is reflected from the surface.If the light strikes the interface at a certain angle that isgreater than the critical angle, it is completely reflected(total internal reflection). In this case, an evanescent wavepropagates perpendicular to the interface into the sur-rounding medium of the lower refractive index. Its fieldstrengths decay exponentially with penetration depthstypically in the magnitude of the wavelengths of theincident light [58].

Evanescent wave sensors can be applied in differentmanners. Total internal reflection fluorescence uses theevanescent field for the excitation of fluorophores bound onthe sensor surface [58]. The absorbance of the evanescentlight by molecules close to the sensor surface leads toattenuated total reflection (ATR). This can be measured bythe loss of the intensity of the reflected light coupled outfrom the ATR substrate. The resonance angle for thegeneration of evanescent waves is also an indicator of thechanges of the refractive index of the sensor surface.

The optical waveguide lightmode spectroscopy(OWLS) technique bases on the measurement of theresonance angle of polarized laser light coupled into athin wave guide. The incoupling resonance occurs atprecise angles which depend on the optical parameters ofthe sensor chip and the refractive index of the coveringsample medium. The sensor consists of an optical wave-guide layer on top of a glass support which includes a fineoptical grating. When a polarized laser beam strikes thegrating, it is diffracted. The diffraction angle does not onlydepend on the optical parameters of the sensor but also on therefractive index of the cover medium. The waveguide isrotated around its axis, and when the diffracted beam isincoupled into the waveguide, it is propagating toward theedge of the sensor through multiple internal reflections. Aphotodiode measures the intensity of the incoupled light(Fig. 5). Light intensity versus angle of incoupling isfollowing a sharp, resonant peak profile. A change in theeffective refractive index due to the binding of analytemolecules on the sensor surface results in a shift of theincoupling resonance angle. Similar to the SPR technique,association and dissociation of analyte molecules on thesensor surface can be followed in real time without labelingby continuous measurement of change in the incouplingangle. The method can be applied for direct sensing of

various types of biomolecules if appropriate probes orreceptors are immobilized on the waveguide.

Piezoelectric sensors

Quartz crystal microbalance Piezoelectric sensors repre-sent another option to build comparatively cheap sensorinstruments. These are based on the piezoelectric effect thatappears in anisotropic crystals like quartz. They provide alabel-free detection of molecular binding events in realtime. The crystals generate oriented dipoles by means ofpressure application. Accordingly, if external voltage isapplied, a mechanical deformation can be monitored. Thisallows probing an acoustic resonance by electrical power.Application of an alternating current to the crystal inducesoscillations, and a standing shear wave can be generated if aspecial cut crystal is placed between two disc electrodes.The oscillation frequency depends on the crystal thickness.Mass deposition on the surface of the crystal increases itsthickness and the oscillation frequency decreases. Thechange of the frequency can be correlated with the masschange using the Sauerbrey equation:

$f 2f20 $m

A rqmqp 2:3 106f 20

$m

A:

Fig. 4 Kretschmann configuration for SPR sensors

Fig. 5 Schematic arrangement of an optical waveguide lightmodespectroscopy immunosensor adapted from [160]



Accordingly, the change of the frequency, f, is directproportional to the crystals fundamental mode, f0, and thebound mass, m, on the crystal. Furthermore, it isindirectly proportional to the piezoelectrically active area,A. q represents the density of quartz (2.648 g/cm

3) and qthe shear modulus of quartz (2.9471011g/cms2) [60, 61].

PEMC sensors Cantilever biosensors exhibit the possibilityof label-free protein and pathogen detection with highsensitivity. They can be divided into two categories: staticbending or dynamic resonating cantilevers. Using thebending mode, the binding of antigens to the surface ofthe cantilever changes its surface stress and results in adeflection response proportional to the analyte concentra-tion. In the dynamic mode, the attachment of antigen iscausing a decrease in resonance frequency of the oscillatingcantilever due to an increase in mass. Damping occurs ifthis mode is used in liquids. A less damped response influidic experiments can be achieved with millimeter-widecantilevers. The piezo-excited, millimeter-sized cantileversensor is a macro-cantilever comprising a piezoelectriclayer (lead zirconate titanate, PZT) that is connected to aglass layer. The piezoelectric effect is used for theexcitation of the cantilever vibration, and the piezo filmalso senses the response. The sensor is modified byantibodies or antigens in the case of biosensor applications.The resonance frequency decreases as target analytes bindto these receptors, and this can be measured by means of animpedance analyzer. The significant advantages ofpiezoelectric-excited, millimeter-sized cantilevers (PEMCs)are the employment of label-free reagents, the highsensitivity (in the range of femtograms (protein) per liter)in complex matrices, and that only little sample preparationis required [62, 63].

Electrochemiluminescence sensors

Electrochemiluminescence (ECL) is based on the produc-tion of light near an electrode surface by the generation ofspecies that are capable of undergoing highly energeticelectron transfer reactions. The process involves theformation of electronically excited states via electron

transfer of electrochemically generated species. Customar-ily, ECL is originated from the annihilation due to anelectron transfer reaction between a reduced and anoxidized species that are produced at an electrode byalternating pulsing (3 V) of the electrode potential. It canbe formed in a single step reaction with a co-reactant.Ruthenium complexes such as Ru(bpy)3

2+ (bpy: 2,2-bipyridine) are applied as light-emitting molecules, andtripropylamine (TPA) serves as co-reactant (Fig. 6). Afterits deprotonation, TPA acts as a reducing agent. In thefollowing reaction of TPA with Ru(bpy)3

3+, the Ru dyeenters an excited state by high-energy electron light transferfrom TPA.

ECL is a very attractive method for the detection ofbiomolecular interactions because of its high sensitivity.The ECL luminophore is bound to a receptor molecule suchas an antibody, thus resembling commercially availableradioactive or fluorescent labels. The luminophore does notrequire an excitation light source, and therefore, ECL is notinterfered by luminescent impurities and scattered light[64]. Only the molecules that are in close proximity to theelectrode surface are stimulated to luminescence.

Biosensors and sensor arrays for screening toxinsand bacterial or fungal contaminants in food products

The applicability and acceptance of a sensor system ismainly determined by its robustness, the simple handling ofthe instrumental setup, and rapid sample preparation. It is ofcommon interest in industry, governmental food control,and research to be able to resort to sensor devices whichprovide reliable measurement results. This includes that thesensor has been thoroughly evaluated and standardizedoperation protocols are available. When glancing on themultiplicity of different kinds of reported sensors, manyresults are obtained from calibrations in buffered aqueoussolutions. A majority of the described assays have notbeen carried out in real food samples or still require someadditional development or evaluation steps before they areapplicable to real samples. Additionally, the study ofmatrix effects, cross-contaminations, and other interfer-ences are crucial points that have not always been

Fig. 6 Mechanism of ECL gen-erated from a ruthenium com-plex and TPA



considered consequently. The stability and specificity ofthe used receptor systems is also a major problem. Finally,any new sensor system has to overcome a criticalcomparison with established analytical methods.

Sensors for bacterial toxins and spores

This section gives an overview of recent approaches forthe detection of bacterial toxins and spores with signifi-cant impact on food safety and health using biosensors.These include Bacillus anthracis spores, C. botulinumneurotoxins (BoNTs), SEs, and aminoglycosides. Bacterialtoxins are often the cause of foodborne illness. Theyparticularly occur in dairy and fruit products. They can beproduced in purified form and be used as food additives interms of biological weapons. There is an urgent demandfor rapid detection and quantification methods due topublic health risks and the high number of samples.

B. anthracis spores

Bioterrorism which uses microorganisms as biologicalweapons has already been reported in history. It isreported that the dissemination of Yersinia pestis wasused in the fourteenth century as one of the firstbioweapons during the Mongolian siege of Kaffa on theCrimea peninsula. B. anthracis spores were deliberatelyreleased in 2001 through the United States postal system,which resulted in 22 cases of anthrax causing five cases ofdeath. B. anthracis is a large Gram-positive (11.5310 m), aerobic spore-bearing bacteria. It is an idealorganism to be employed as biological weapon. It formsspores that can be aerolized easily and survive under harshconditions. Spores are not formed in host tissues unlessinfected body fluids are exposed to air. When nutrients areexhausted, resistant spores form that can survive in soil fordecades. These spores then germinate when exposed to anutrient-rich environment (blood or tissue of human oranimal). Inhalation of the spores causes mortality rates near100% [6570]. The lethal dose LD50 is estimated to beabout 2,5005,000 inhalated spores [71]. Gastrointestinalanthrax results from ingestion of undercooked food fromanimals infected with B. anthracis (incubation period, 25 days).

Fiber-optic immunosensor A portable evanescent-wavefiber-optic biosensor, the Research International Analyte2000 fluorometer [72], can be used for the detection of B.anthracis spores in different products [73]. Biotinylatedcapture antibodies for B. anthracis were attached to atapered streptavidin-coated polystyrene fiber-optic wave-guide. The target analyte can be determined by means of aCy5-labeled detection antibody. The instrumentation

requires minimal sample preparation steps and an overallassay time of less than 1 h. No false positive results and alimit of detection (LOD) of 3.2105 spores per milligramwere obtained using this commercially available device.

Immunomagnetic sensor Lower LODs can be obtained byelectrically active polyaniline-coated magnetic particles(EAPM). These can be applied as transducers in a biosensorthat enables LODs of 4.2102 spores per milliliter inlettuce or beef and 4.2103 spores per milliliter in milksamples [74]. The sensor is based on a sandwich immuno-assay with capture antibodies immobilized on a nitrocellu-lose membrane pad and a detector antibody conjugated toEAPMs with a diameter of 100 nm. The particles are usedfor immunomagnetical concentration of spores in the foodmatrices and applied to the sample admission pad subse-quently. The antigenantibodyEAPM conjugate flows tothe capture pad and the sandwich complex is formed. Theconductive EAPM nanoparticles in the sandwich act as avoltage-controlled ON switch, resulting in decreasedresistance across the silver electrode, which is recordedelectrically. The mode of operation of the biosensor issummarized in Fig. 7.

Piezoelectric sensors In comparison to fluorescent orelectrochemical detection modes, cantilever sensors aredevoid of unfavorable effects like fluorophore quenching,noise, or high background signals. Monitoring of B.anthracis spores in water was achieved by means of amicroresonator mass sensor (microcantilever sensor) func-tionalized with capture antibodies. These enable a selectiveabsorption of the analyte on the surface [75]. Thequantification is based on the measurement of the decreaseof the resonance frequency as a result of the added mass ofthe bound spores by means of a laser Doppler vibrometer.A minimum of 50 spores (139 pg) can be determined usinga cantilever of 20-m length, 9-m width, and 200-nmthickness.

A complete sensing assay of Anthrax spores in liquidswas done in 20 min with an antibody-modified PEMCsensor in a flow cell [76]. PEMC immunosensors providemass change sensitivities in the range of picograms to sub-femtograms [77, 78]. The glass substrate is modified withantibodies using silane monolayers and EDC-NHS cou-pling. It is exposed to a pre-concentrated sample inphosphate-buffered saline, and the resonance frequency ismeasured directly in the sample containing the toxin. Theresonance frequency decreases linearly with the boundmass in the case of small mass quantities. Quantitativedetermination of bound spores can be carried out bymeasuring the change in resonance frequency with animpedance analyzer. A detection limit of 38 spores per litersample was attained.



C. botulinum toxins

BoNTs are considered as one of the most hazardous typesof toxins for humans with a lethal dose of 1 g toxin perkilogram body mass. The seven serologically dissimilartoxins AG are produced by C. botulinum bacteria [79].They selectively target cholinergic nerve endings and act aszinc-dependent endoproteases. BoNTs cleave proteinsinvolved in the release of acetylcholine, resulting in flaccidparalysis or death [80].

The majority of bacterial strains only produce one typeof toxin, but some strains are capable of multiple BoNTrelease [81]. Botulism has been involved in many foodcontamination cases. Products like meat, fish, dairy, sautedonions, chopped garlic in oil, and preserved vegetables aremostly affected. Estimated 58 cases of botulism aredocumented in the USA per year. Among these, 80% ofall cases require hospitalization. Hence, foodborne botulismis rare, but fatal. Contamination of food or drink withmicrobial toxins is not only a form of a bioterrorism butalso a global public health problem in general [82]. Rapiddetection of exposure to BoNTs is considered as animportant public health goal [83]. A LOD of 40 pg/mL orless has been recommended because of the high toxicity ofBoNTs [84, 85].

Surface plasmon resonance sensor The SPR techniqueallows a quantitative real-time detection of biomolecularinteractions including immunological assays. It is a label-free method and therefore avoids unwanted interferences ofthe fluorescent label with the antibody-antigen recognition.SPR is an appropriate alternative to ELISA methodscapable of multi-analyte and multiplexed sample screening.A four-channel SPR biosensor employing wavelengthmodulation has been reported for the determination ofBoNT A, B, and F in honey solutions [86]. A mixed self-

assembled monolayer of biotinylated alkanethiol and oligo(ethylene glycol)thiol is formed on a gold surface whichserves as binding matrix for streptavidin and, subsequent-ly, biotinylated capture antibodies. The assay provides anLOD of 5 ng/mL for each type of toxin after 1-h incubation time.

Hydrogel sensor Toxin-responsive hydrogel sensors are anadditional alternative to common detection methods used inorder to improve the sensitivity of Botulinum neurotoxin Adetermination. The response of the hydrogels relies on theenzymatic activity of the toxin and provides a highselectivity as the toxin includes a specific substratecleavage site. The sensor is based on a toxin-sensitive,peptide-modified hydrogel layer and requires only 20 L ofcontaminated sample for visual analysis of buffer, milk, andcream samples. Peptide-modified hydrogels can dissolve inthe presence of proteolytic enzymes. They represent anoptimized matrix for sensing a toxin possessing enzymaticactivity because of their simple fabrication and application.Toxin contamination is indicated in visually evident geldissolution. The degradation of the hydrogel can bemonitored by stereoscope imaging after removal of thesupernatant solution [87].

Immunomagneticelectrochemical sensor Novel assays us-ing a paramagnetic bead-based ECL technology were usedfor the detection of botulinum neurotoxins A, B, E, and F inmilk, apple juice, ground beef, pastry, and raw eggs [88].Biotinylated serotype-specific antibodies bound tostreptavidin-coated beads act as solid support and capturethe target from sample solutions. Analysis was carried outwith a BioVeris M-Series M1R analyzer [89] based on theuse of ruthenium chelate-labeled antibodies. On the surfaceof the working electrode, photon-emitting species aregenerated from stable precursors by anodic oxidation in

Fig. 7 Biosensor architecture (A) and schematic representation of the biosensor system (B) adapted from [74]



the presence of TPA. LODs of 50 pg/mL of BoNT A and800 pg/mL of BoNT F could be obtained.

S. aureus enterotoxins

S. aureus are Gram-positive bacteria that produce enterictoxins and grow in animal food products at ambienttemperatures [90]. S. aureus preferentially grows in foodswith low water activity (cooked or strong salted food)where they outcompete other microorganisms [91]. Out-breaks are frequently traced back to situations where foodpreparers did not comply with hygiene and safety regu-lations or to incidences where food, that has to stay frozen,was exposed to room temperature. Unfortunately, align-ments containing enterotoxins usually taste, smell, and lookunsuspicious and not adulterated [92]. SEs cause severegastroenteritis with symptoms like diarrhea and vomitingwithin 16 h after ingestion and act as superantigens thatstimulate non-specific T cell proliferation [93]. Enterotoxinsare very heat-stable and resistant to cooking, heating, andproteolytic decomposition; thus, only the bacterial cells willbe inactivated or killed during meal digestion [9496]. Tendifferent SEs have been identified (A, B, C1, C2, C3, D, E,H, I, G, and J) to date [9799]. Among these, SEA-SED arethe most common food pathogens. Particularly, enterotoxinA is involved in most of the disease outbreaks, whereasstaphylococcal enterotoxin B (SEB) is a potential bioterror-ism agent. Their physical and chemical properties weresummarized by Jay et al. [100]. SE levels of 1 ng/g foodwere reported to cause gastroenteritis in the case of adultpersons [101, 102]. Thus, assays which are able todetermine enterotoxin concentrations below 1 ng/g(1 ppb) reliably are required to guarantee the safety of food.

Fluorescent immunosensors Dong et al. [103] have shownthat microfluidic immunosensors for SEB can be integratedin polydimethylsiloxane (PDMS) chips. Supported phos-phatidylcholine bilayer membranes were generated inoxidized PDMS microchannels by vesicle fusion. Theyminimize the non-specific binding of biomolecules. Astreptavidin layer was attached to the membranes whichenables the immobilization of biotinylated antibodies. Thisfunctionalization results in reinforced supported bilayermembranes including the immobilized antibodies as recog-nition units for SEB. Their detection occurs via primaryand, subsequently, Cy3-labeled secondary antibodies. Alinear range of 11,000 ng/mL with a LOD of 1 ng/mL wasobtained in milk samples.

The rapid automated particle immunosensor from Sapid-yne [104] can be applied to analyze SEB contaminations incream. Polymethylmethacrylate (PMMA) beads conjugatedto antibodies were applied as the solid phase of the flow-

through device and trapped by a filter. Reagents for theassay could float the beads by means of a rotary valve.Fluorescein-labeled target antibodies are used for detection.The device is capable of determining 5 ng/g of SEB inapproximately 10-min incubation time [105].

Surface plasmon resonance sensors The immunologicaldetermination of SEB in milk can be carried out with awavelength modulation-based SPR sensor using two dif-ferent operation modes, either by direct toxin detection or asandwich assay. The sandwich assay mode uses secondaryantibodies that bind to the captured target toxins in order toamplify the attained response. SEB concentrations of as lowas 0.5 ng/mL could be determined in milk samples [106].The BIACORE SPR biosensor [107] can be adapted for theanalysis of SEA and SEB in various matrices and numerousconjunctions.

The multitoxin bioapplication interaction mass spec-trometry (BIA-MS) is a method which has been adapted forthe detection of SEB, e.g., in milk and mushrooms [108]. ABIACORE X instrument was combined with a sensor chipin a flow cell. Capture antibodies were immobilized on thecarboxymethyl dextran-derivatized surface of the sensorchip to determine the toxin concentration in the sample viaSPR, which is followed by the identification of the boundtoxin by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. A LOD of 1 ng/mL SEB in milkand mushrooms can be obtained.

An extended BIA-MS with two flow cells containinganti-SEB or anti-toxic shock syndrome toxin-1 wasperformed with the BIACORE X instrument and anantibody-derivatized sensor chip with an LOD of 1 ng/mLSEB in mushrooms [109].

Competitive immunoassays are also feasible with anupgraded BIACORE device employing two different sensorchips. SEB was covalently linked to the chip surface. Thesurface-bound toxins compete with the free toxins in thesample for the antibody binding sites. Antibodies bound totoxin molecules from the sample are washed away and themass increase on the surface measured. A LOD of 312 pg/mL of SEB in fresh whole milk and skimmed milk could beachieved [110]. SEA was verifiable in a similar approach inspiked raw egg supernatants with an LOD of 1 ng/mL and adynamic range of 140 ng/mL within 15-min of assay time[111].

Portable fiber-optic sensors Portable fiber-optic biosensors(Analyte 2000, see B. anthracis spores) can also beapplied for SEB determination [112]. This compact andlightweight biosensor system is capable of running fourtests simultaneously by means of using four individual fiberprobes. Light from a 635-nm diode laser is coupled into atapered 600-m fiber. Fluorescent molecules within 100-



nm distance from the fiber can be excited evanescently (seeEvanescent wave methods) and a portion of the fluores-cence emission recouples into the fiber. The fibers weresilanized and a heterobifunctional linker was attached forbinding the rabbit anti-SEB capture antibody. Cy5-labeledsheep anti-SEB detection antibodies were added afterincubating the fiber with aqueous ham extracts. Thesandwich assay forms a fluorescent complex at the surfaceof the fiber, generating only signals in the reach of theevanescent field. The assay can be performed within 45 minwith a LOD of 0.5 ng/mL of SEB in aqueous ham extract.

Piezoelectric sensors An approach by Lin and Tsai [113]uses an AT-cut quartz crystal (8 mm in diameter) which wascoated on both sides with gold electrodes (4 mm indiameter) as transducer element. The crystal surface wascovered with a polyethylenimide layer and anti-SEBcapture antibodies were subsequently cross-linked withglutaraldehyde. The piezoelectric sensor was exposed tosamples containing SEB for 1 h and the resonancefrequency measured. A LOD of 2.5 g/mL of SEB inspiked milk samples was attained.

PEMC (see Piezoelectric sensors) cantilever sensorsthat are excited by piezoelectric actuators are reliable andsensitive devices for the quantification of SEB. They wereapplied to beverage samples such as apple juice and milk.The LOD in milk was reported to be 10 fg of toxin withoutany sample preparation steps. The sensor device iscomposed of PZT cantilevers on glass slides. These wereconsecutively coated with polyurethane, gold, and proteinG for selective anti-SEB binding. The assays can beperformed in a flow cell [114].

Chemiluminescence sensor A new integrated sensor com-bining three detection elementscarbon nanotubes assupport for the immobilization of primary antibodies,enhanced chemiluminescence excitation for signal genera-tion, and a simple cooled charge coupled device (CCD)detectorhas been described by Yang et al. [115]. Anti-bodies against SEB attached to nanotube surfaces wereimmobilized onto a polycarbonate surface. SEB wasdetected by a sandwich-type ELISA. This combination ofsensor device and chemiluminescent ELISA provides aLOD of 10 pg/mL of SEB in soy milk, apple juice, andmeat baby food.

Immunomagneticelectrochemical sensors Sensitive andrapid detection of SEB in 5% skim milk in PBS can beperformed by a combination of immunomagnetic separationand electrochemiluminescent detection using the ORIGENsystem [89]. An assembly of biotinylated antibodiesimmobilized on streptavidin-coated magnetic beads wasused for capturing the antigens from the sample solution.

The magnetic beads were fixed in the electrochemical flowcell of the ORIGEN system. The detector antibody wasconjugated to ruthenium(II)-tris-bipyridine and binds to theimmunocomplex in the presence of toxin antigens. Light isgenerated electrochemically after addition of an excess ofTPA which is oxidized at the surface of the anode (seeElectrochemiluminescence sensors). The resulting chemi-luminescent emission of photons can be detected at 620 nmwith a linear range of 7 logs using a photomultiplier tube.This method requires a sample incubation time of 30 minand an assay time of 1 min. It assures fast results with100% specificity, 98% sensitivity, and an LOD of 1 pg/mL[116]. The reaction scheme is illustrated in Fig. 6.

Evanescent wave biosensor The evanescent wave biosensorIAsys [117] was applied to analyze SEA in potato salad,canned mushrooms, hot dogs, and dry milk in order todevelop a real-time detection system for the identificationof foodborne toxins. Carboxymethyl dextran-coated cuv-ettes were functionalized with capture antibodies. Asandwich-type assay can be carried out within 4 minmeasuring the change of the refractive index of the sensorsurface. The sensitivity is in the range of 10100 ng/gdepending on the type of food sample [118]. An overviewof sensor formats for bacterial toxins and spores is shown inTable 3.

Sensors for contaminations by aminoglycosides

Aminoglycosides are broad-spectrum antibiotics most com-monly used in veterinary drug medicine for the treatment ofinfections such as mastitis [119]. Gentamycin (GTM) isused for the therapy of bacterial infections in pigs andcattle, while neomycin (NEO) is indicated to treat gastro-intestinal infections of sheep, pigs, goats, cattle, and poultryand applied in case of mastitis. Streptomycin and itsderivate, dihydrostreptomycin (DHS), are used to treatbacterial diseases in sheep, pigs, poultry, and cattle. DHS isalso administered in combination with benzyl penicillins.Unfortunately, all aminoglycosides present some toxicolog-ical adverse reactions like nephrotoxicity and ototoxicity.Additionally, allergic reactions can be initiated by means ofaminoglycoside residues [120]. Injections of these substan-ces in cattle produce aminoglycoside residues that are apotential cause of food contamination [121, 122]. Maxi-mum residue limits (MRLs) have been constituted due tothe consumer risk connected with the presence of anti-biotics in food. The MRLs of aminoglycosides in fooddefined by the European Union range from 100 ng/mL(GTM) to 500 ng/mL (NEO).

Generally, screening of these substances is mainlyperformed by microbial tests [123], immunoassays [124,



125], and liquid chromatography [126]. A review of currentmethodologies for the detection of aminoglycosides ispresented by Stead [127].

Surface plasmon resonance sensors Different approachesfor the detection of streptomycin and other aminoglycosideshave been performed on the BIACORE SPR biosensorsystem (see last section). Competitive immunoassays forthe detection of streptomycin and dihydrostreptomycinresidues in whole milk, honey, pig muscle, and kidneyhave been described. A functionalized sensor chip contain-ing a covalently bound carboxylated dextrane matrix on agold surface was coated with a streptomycin derivative. Afixed concentration of antibody was mixed with the sampleand injected over the sensor chip surface. Cross-reactivitieswere evaluated by means of samples spiked with structur-ally related compounds and commonly used antibioticssuch as dihydrostreptomycin, neomycin, gentamycin, vir-giniamycin, kanamycin, penicillin G, sulfmethazine, orchlortetracycline. No significant cross-sensitivity higherthan 0.1% could be observed, except with dihydrostrepto-mycin. The LODs were determined from spiked samples.

They range between 30 and 70 g/kg for the different foodmatrices surveyed [128]. Furthermore, the BIACORE SPRsystems were evaluated in an interlaboratory study with 14participants. The analysis results of 12 samples spiked withstreptomycin and dihydrostreptomycin were compared toresults obtained by commercial ELISA kits [129].

Another competitive immunoassay using SPR sensorswith four flow channels and a mixture containing fourspecific antibodies was applied to reconstituted skimmedmilk [130]. Five relevant aminoglycosides (streptomycin,gentamycin, neomycine, kanamycin, and a streptomycinderivative) were determined in four single inhibition assayson a sensor chip containing immobilized haptens indifferent flow channels. All detection limits were found tobe below the maximum residue limits (100500 ng/mL)varying between 15 and 60 ng/mL.

Hapten microarray A regenerable immunochip for therapid determination of 13 different antibiotics (amongothers sulfamethazine, streptomycin, penicillin G, genta-mycin, enrofloxacin, and tetracyclin) in raw milk waspresented by Kloth et al. [131] using an automated

Table 3 Overview of sensor methods for the detection of bacterial toxins and spores

Analyte Sensor type Food matrix LOD Ref.

Anthrax spores Portable evanescent wave fiber-optic sensor Talc-based baby powder, cornstarch-basedbaby powder, sugar, baking soda

3.2105 spores/mg [73]

Anthrax spores Biosensor with electrically active polyanilinecoated magnetic (EAPM) nanoparticles

Lettuce, ground beef, whole milk 4.2102 spores/mL(lettuce, beef)

[74]

4.2103 spores/mL (milk)

Anthrax spores PEMC Air, water 50 spores/139 pg [75]

Anthrax spores PEMC Air, liquid 38 spores/L [76]

BoNT SPR Honey 5 ng/mL [86]A, B, F

BoNT A Toxin-responsive hydrogel sensor Milk, cream Visual [87]

BoNT Paramagnetic bead ECL Milk, apple juice, ground beef, pastryand raw eggs

50800 pg/mL [88]A, B, E, F

SEB Microfluidic immunosensor Milk 1 ng/mL [102]

SEB Fluorescence-based particle sensor (Sapidyne) 5 ng/g [103]

SEB Wavelength-modulated SPR Milk 0.5 ng/mL [105]

SEB SPR-BIA-MS (BIACORE X) Milk, mushrooms 1 ng/mL [93]

SEB SPR-BIA-MS (BIACORE X) Mushrooms 1 ng/mL [107]

SEB SPR (BIACORE 1000) Milk 0.312 ng/mL [108]

SEA SPR (BIACORE 1000) Raw egg 1 ng/mL [109]

SEB Portable fiber-optic sensor Ham extract 0.5 ng/mL [110]

SEB Piezoelectric crystal immunosensor Milk 2.5 g/mL [111]

SEB PEMC Milk, juice 10 fg (juice) [112]100 fg (milk)

SEB Carbon nanotube ECL sensor Soy milk, apple juice, meat baby food 0.01 ng/mL [113]

SEB IMS-ECL-ORIGEN system 5% skimmed milk 1 pg/mL [114]

SEB IAsys Potato salad, canned mushrooms, hotdogs, dried milk

10100 ng/g [116]



microarray platform reader. The chips consist of epoxylatedglass slides which are coated with polyethylene glycoldiamine and a diepoxy-polyethylene glycol layer. The lattercouples antibiotic haptens to the chip surface. Antibiotic levelscan be quantified by means of a competitive immunoassaywhere free antibodies were added to the sample. Signalgeneration occurs via chemiluminescent reaction of HRP-labeled secondary antibodies with a luminol-based substrate.The detection of the 13 analytes can be performed bymeans ofa CCD camera in less than 6 min. An overview of amino-glycoside biosensors is given in Table 4.

Sensors for mycotoxins

Mycotoxins represent a group of rather chemically differentcompounds produced by fungi, particularly Penicillium,Fusarium, and Aspergillus species, which grow on organicmatter. Nuts, cereals, oil seeds, or fruits are mainly affectedbecause of their high moisture content and richness ofnutrients. The major exposure routes are the intake ofcontaminated food, inhalation of spore-borne toxins, ordermal contamination by skin contact. The extent ofcontaminations by toxigenic mold species depends on severalfactors, e.g., temperature, storage conditions, or humidity[132]. The most significant toxins are aflatoxins, deoxyniva-lenol, ochratoxin, and fumonisins. An overview of theirchemical structures is given in Fig. 8.

The appearance of aflatoxins is quite common whileother mycotoxins are rare. They are all highly toxicmetabolites produced by Aspergillus species. Up to now,18 different types are identified, with aflatoxin B1, B2, G1,and G2 (AFB1, AFB2, AFG1, and AFG2) as the mostimportant ones [133, 134]. Aflatoxins are polycyclic

compounds which fluoresce strongly in UV light. TheInternational Agency for Research on Cancer (IARC) hasclassified all four aflatoxins as group 1 carcinogens [133].The maximum levels for aflatoxins set by the EuropeanCommission are 2 ng/g for AFB1 and 4 ng/g for totalaflatoxins in corn, groundnuts, nuts, dried fruit, and cereals[135]. AFB1 is the most prevalent and biologically activetoxin, and its carcinogenicity is related to the possibility offorming adducts with DNA [136].

When aflatoxin B1 is ingested by cows, excretion of thehydroxylated product aflatoxin M1 (AFM1) in milk occurs.Although it exhibits lower toxicity, this toxin is hazardousbecause of its hepatotoxic and carcinogenic effect. Unfor-tunately, it is relatively heat-stable and therefore resistantduring pasteurization and dairy production processes. Thelegal limit defined by the European Union is 0.05 mg/kg forAFM1 in milk [137141].

Table 4 Overview of sensor methods for the detection of amino-glycosides


Streptomycin BIACORE Q Milk, honey,kidney,muscle

30, 15, 50,and70 g/kg

[126]

Streptomycin BIACORE Q Bovine milk ns [29]S-Derivate

Streptomycin BIACORE 3000 Skimmedmilk

1560 ng/mL

[127]Gentamycin

Kanamycin

Neomycin

S-Derivate

Antibiotics Munich chipreader platformMCR3

Raw milk ns [128]

ns not specified

OO

O

OO

OMe OO

OO

OMe

OHO

O

O

CH2OHOH

O OH

CH2 CH NHCO O

OOH

ClCH3

COOH

O OH

O OH OH

NH2

O COOH

HOOC

O

COOH

COOH

Fig. 8 Chemical structures of the most important mycotoxins AFB1,AFM1, OTA, and DON



Ochratoxin A (OTA), a coumarin derivate, is producedby several Aspergillus and Penicillium molds [142]. Itcontaminates mainly cereals, but also coffee, dried fruits,grape juice, wine, beer, and nuts. OTA is considered as anephrotoxin, hepatotoxin, carcinogen, and teratogen. TheIARC has classified OTA as possibly carcinogenic inhumans (group 2B) [143]. The admissible limit for OTAin cereal products for human consumption is defined by theEuropean Commission as 3 g/kg.

Another relevant group of mycotoxins, Fumonisins, arereleased by Fusarium species. They also exhibit a potentialrisk for humans and animals [144]. Twelve Fumonisins areidentified; among these, Fumonisin B1 (FB1, macrofusin)is the most prevalent and toxic compound. Fumonisins arehepatotoxic and classified as potential carcinogenic agents.The European Union prescribes a maximum tolerable dailyintake of 2 g/kg body mass [145]. Deoxynivalenol (DON)is another hazardous substance produced by Fusarium inmoderate climate areas. Its tolerable daily intake isproposed by the European Scientific Committee on Foodas 0.5 g/kg bodyweight [146].

Electrochemical sensors A simple and fast disposableelectrochemical immunosensor using differential pulsevoltammetry (DPV) has been described for the detectionof AFB1 in barley. A graphite working electrode wascoated with a BSA-AFB1 conjugate following a blockingstep. Monoclonal AFB1 antibodies were added to thesample which is exposed to the electrode, forming acompetitive assay. A secondary AP-labeled antibody wasused to detect the bound target antibodies. 1-Naphthol wasdetected by means of DPV in a typical three-electrode cell,which is a product of the enzymatic conversion of 1-naphthylphosphate. A LOD of 2 ng of AFB1 in 1 g ofbarley could be obtained [147].

A similar approach includes disposable screen-printedelectrode systems (SPEs) for the quantification OTA inwheat extract. A direct electrochemical ELISA isapproached using a typical three-electrode assembly.Therefore, an OTA-AP conjugate is added to the sample.These compete for the binding to the monoclonal captureantibodies immobilized on the electrode surface. The APactivity again was measured by DPV recording theenzymatic formation of 1-naphthol. The method exhibits adetection limit of 400 ng/kg wheat sample [148].

An indirect competitive electrochemical immunosensorfor OTA uses avidin-coated microwell plates. These wereincubated with biotinylated OTA and blocked withskimmed milk. The competition reaction was carried outby incubation of OTA-contaminated samples with anti-OTA. This mixture was disposed to the wells followed byaddition of AP-conjugated secondary antibodies. Theiractivity was measured by means of SPEs as described

above using DPVor chronoamperometry. Concentrations of10 ng/mL of OTA in wine samples could be monitored[149].

Multichannel electrochemical detection systems can alsobe integrated into microwell plates. These immunosensorarrays have turned out to be a useful tool for thedetermination of mycotoxins such as AFB1 in corn samples[150]. Measurements were performed with an indirectELISA scheme in 96-well, screen-printed plates. Thesurfaces of the SPEs were coated with AFB1-BSAconjugate. A mixture of polyclonal antibodies and samplescontaining AFB1 is poured into the wells followed by theaddition of a secondary AP-labeled antibody. In this case,the amount of the enzymatically formed 1-naphthol wasdetected by means of intermittent pulse amperometry.AFB1 levels of 30 pg/mL can be determined in a dynamicrange between 0.05 and 2 ng/mL.

Volatile components of mycotoxin contaminations canbe detected with the help of electronic noses. These canquantify DON and OTA, e.g., in barley grains. A volatilecompound mapper is composed of ten metal oxidesemiconductor field effect transistor sensors, six SnO2-based gas sensors, and a Gascard O2 monitor. The artificialnose requires only 90 s of effective measurement time andcan be considered as a real alternative to conventionalsingle sensors if only semiquantitative results are required(e.g., if the OTA concentration is below or above a certainthreshold) [151].

Surface plasmon resonance sensors A SPR-based screen-ing technique for AFB1 and AFG1 in maize is demonstrat-ed by Cuccioloni et al. [152]. This biosensor represents anew approach for the detection of aflatoxins based on thereversible interaction between elastase, a proteolytic en-zyme, and the soluble toxin ligand in the sample. Elastasewas immobilized on functionalized cuvettes in a thermo-stated sensing chamber and AFB1 and AFG1 referencestandards were added in different concentrations. Associa-tion kinetics were followed up to equilibrium. Afterregeneration of the sensor surface, aflatoxin-containingsamples were added and each response was routinelyfollowed and analyzed. The major advantages of thismethod are its relative rapidity, the reusability of thesurface, and the low cost per single test. The LOD forAFB1 was 970 ng/kg.

Tudos et al. [153] used a BIACORE SPR system tomonitor DON in contaminated wheat and dry wheatpowder. The sensor surface was modified by a caseinDON conjugate. An excess of DON antibodies were addedto the sample to capture DON. Subsequently, the mixturewas exposed to the sensor to perform a competitive assay.Binding of free antibody is recorded by means of SPR. Thesensor could be reused more than 500 times without



significant loss of sensitivity, providing a dynamic range of2.530 ng/mL for DON determination.

The BIACORE system [107] was also adapted formultiplexed analysis of mycotoxins in food. A competitiveassay format is capable of determining DON, FB1, AFB1,and OTA in different samples in one measurement with thehelp of four flow channels. After sample extraction andincubation with a mixture of the corresponding antibodies,the results can be obtained within 25 min. LODs of 0.5, 0.2,50, and 0.1 ng/g could be achieved for DON, FB1, AFB1,and OTA, respectively [154].

Fiber-optic and portable immunosensors A portable immu-nosensor for the quantification of AFM1 in milk sampleswas developed by Sibanda et al. [155]. It consists of anylon membrane modified with a secondary antibody. Afterimmobilization of capture antibodies and incubation of thesample, a detector antibody conjugated to HRP was added.TMB is used as enzyme substrate, and colorimetric readoutyields a LOD of 0.05 ng/mL of AFM1 in milk samples.

Fiber-optic waveguides can be modified with monoclo-nal FB1 capture antibodies. Fluorescein-labeled FB1 wasadded to the sample to carry out a competitive assay. Thesensor works in a dynamic range from 10 to 1 g/mL ofFB1, and a LOD of 10 ng/mL was attained in ground corn[156].

Fiber-optic fluorescent sensors based on evanescentwave excitation were demonstrated by means of FB1 andAFB1 determination in maize. Again, a competitive formatwas applied. It uses monoclonal capture antibodies attachedto a silica fiber. In this case, FITC-conjugated AFB1 wasadded to the sample and its fluorescence is measured afterexposure to the fiber. The LOD in methanolic extractswas 3.2 g of FB1 in 1 g maize. AFB1 was detected inparallel with a direct assay via its intrinsic fluorescence[157].

The details of the Naval Research Laboratory (NRL)multi-analyte array biosensor have been already introducedin the last sections. It provides also a powerful platform forthe screening of mycotoxins. Barley, wheat pasta, corn-flakes, ground barley, cornmeal, red wine, and roasted

coffee were screened for OTA contaminations withoutextensive sample pretreatment. The NeutrAvidin-coatedsurface is used for the immobilization of biotinylatedOchratoxin A molecules. These compete with the freetoxin in the sample extract for binding to the added Cy5-conjugated OTA antibodies. Quantification occurs by themeasurement of the formation of fluorescent OTAanti-OTACy5 complexes on the waveguide by CCD imaging.The net assay time was approximately 3040 min. TheLODs in different food matrices range from 3 to 100 ng/g[158]. The same type of assay can be performed for thedetermination of DON [159].

Intrinsic fluorescence sensors A label-free method for theoptical determination of OTA contaminations is based onsilica sensor spots with integrated photodiodes as detectors.These monitor silica-bound OTA by means of its intrinsicfluorescence under UV radiation. The sensor assemblyconsists of a thin-layer chromatography plate which isplaced on a transparent conductive oxide material withintegrated a-Si:H photodiodes. The intrinsic fluorescence ofthe mycotoxin can be excited by UV radiation at 253 nm.The device is capable of detecting 0.25 ppb of toxin in asample volume of 10 mL (Fig. 9) [160].

The fiber-optic immunosensor for AFB1 described byCarter et al. [161] utilizes either its intrinsic fluorescence orcompetitive assays with labeled capture antibodies. Intrinsicfluorescence detection is carried out after binding of AFB1to an antibody-functionalized polymer membrane. A LODof 50 pg/mL of AFB1 in corn and peanut extracts could bedetected. The excitation and emission wavelengths were362 and 425 nm, respectively. The competitive assayrequires enzyme or fluorescent-labeled capture antibodies.In the latter case, AFB1 probes were immobilized on acellulose membrane. These compete with the free toxins inthe sample for the binding to the added FITC-labeledantibodies. This assay has a LOD of 1 ng/mL of AFB1 inpeanut extracts.

Immunosensors integrated in FIA systems The analysis ofmycotoxins in milk can be performed using a flow injection

Fig. 9 Schematic view of asensor for detection of intrinsicfluorescence with an integratedphotodiode [157]



immunoassay (FIA) method with amperometric detection.Badea et al. [162] mixed milk samples with definedamounts of anti-AFM1 and AFM1-HRP conjugate.AFM1-HRP acts as tracer in a competitive assay format.The mixture was injected to the flow system where theantibody-bound tracer is separated from the free tracer via aprotein G-modified column. The activity of the elutedenzyme label was detected amperometrically using TMB assubstrate.

Optical waveguide lightmode spectroscopy (seeEvanescent wave methods) was also shown as apromising approach to determine AFB1 and OTA in foodsamples such as barley and wheat flour [163]. OWLS usesthe evanescent field technique for in situ and label-freestudies of surface processes at molecular levels (see Fig. 5).The OWLS sensor chip can be integrated in a flow injectionanalysis system after immobilization of antigen conjugate.The sample was applied with antibodies and incubated for1 min following injection to the system. A competitiveassay format can be used for the simultaneous detection ofOTA and AFB1. Hapten-BSA conjugates are immobilizedon the waveguide as capture probes. Standards and samplescontaining the mycotoxin of interest are mixed withmonoclonal antibodies and injected in the OWLS system.

Hapten microarray A chemiluminescence-based, flow-through microarray system for the detection of OTA andaflatoxins was presented by Cervino et al. [164]. A haptenmicroarray combined with highly selective monoclonaldetector antibodies and an automated microfluidic systemwith chemiluminescence readout enable the identificationof aflatoxins in the sub-micrograms per kilogram range incomplex raw food extract matrices without prior cleanup. Aself-constructed microfluidic platform was used for theautomated competitive immunoassay sensor. The haptenswere bound to PEGylated and epoxy-modified glass slides.The detection of the bound primary antibody was per-formed by means of an enzymatic chemiluminescentreaction. The luminescence intensity was measured with aCCD camera. The different sensor approaches for myco-toxin detection are summarized in Table 5.

Sensors for plant toxins

Ricin is a highly toxic plant toxin which can be extractedfrom the beans of the castor oil plant, Ricinus communis.This toxic lectin is thousand times more poisonous thancyanide and 30 times more potent than nerve gases and canbe used to contaminate food or water. Ricin is aglycoprotein with a molecular weight of 6065 kDa andconsists of two glycoprotein chains. Its ribotoxic subunitinhibits protein synthesis in mammalian cells. [165173].

With an estimated lethal dose of 110 g/kg body mass,ricin leads to death within approximately 60 h [174]. Ricinis significant as a potential chemical warfare agent and aweapon of mass destruction for everyone because of itshigh availability and listed by the Chemical WeaponsConvention (CWC).

The detection of ricin in water was carried out by amagnetoelastic sensor integrated in a sealed disposablecontainer [175]. The sensor consists of Fe40Ni38Mo4B18coated with Cr and Au layers. The latter was modified withcystamine and the cross-linker glutaraldehyde which enablesthe covalent coupling of anti-ricinus capture antibodies onthe surface. A sandwich assay with AP-conjugated anti-ricinus detector antibodies is used in combination with aprecipitation reaction. AP converts 5-bromo-4-chloro-3-indolyl phosphate into an insoluble product that precipitateson the sensor surface. This amplifies the mass changeassociated with antigenantibody binding events on thesensor surface. Accordingly, a linear shift of the resonancefrequency of the magnetoelastic sensor occurs for ricinconcentrations ranging from 10 to 100 g/mL with a LODof 5 ng/mL.

An alternative approach makes use of an indirectcompetitive ELISA followed by amperometric detection.The assay is carried out on single-use polystyrenegraphiteSPEs. These sense the formation of 1-naphthol from 1-naphthyl phosphate in the presence of a detector antibodyconjugated to AP [176]. The response of the amperometricsensor is proportional to the logarithmic of the ricinconcentration from 100 to 3.2 g/mL. The sensor providesa LOD of 40 ng/mL of ricin which can be quantified in90 min of assay time. The assay flowchart is presented inFig. 10.

Sensors for marine toxins

Marine shellfish is a source of paralytic shellfish poisons.These are harmful because they block sodium channels andevoke poisoning of humans. Particularly, these fish speciescan comprise the toxic agents tetrodotoxin (TTX), saxitoxin(STX), and gonyautoxin (GTX). Their chemical structureshave been revealed and their threat for food safety has beenascertained [177180]. These substances are water-soluble,acid- and heat-stable, and not decomposed by ordinarycooking.

Nanomolar concentrations of TTX suffice to block thefast voltage-activated sodium channel in neurons. TTX isone of the most potent non-protein toxins with a mortaldose of 10 g/kg body mass. GTX and STX bind to thesame site of sodium channels. Moreover, STX is listed inschedule 1 of the CWC, and the lethal dose in a singleapplication is 200 g [181]. The chemical structures ofSTX and TTX are presented in Fig. 11.



Diarrheic shellfish poisoning is caused by the toxicsubstance ocadaic acid (OA). This fatty acid derivative iscytotoxic and inhibits certain protein phosphatases [182].OA is responsible for a notable number of intoxicationsevery year, and its quick detection during food productionand processing is highly demanded. The critical limit forOA defined by the EU is 4060 g/100 g mussel tissue[183]. The structure of OA is shown in Fig. 12.

Amnesic shellfish poisoning can be traced back todomoic acid (DA), an amino acid associated with harmfulalgal blooms [184]. In mammals, including humans,domoic acid acts as a neurotoxin. Most countries haveadopted a legal upper limit of 20 mg DA per gram shellfishfor consumer protection [185]. The structure of DA ispresented in Fig. 13.

Electrochemical sensors The electrochemical immunosen-sor presented by Micheli et al. [186] was adapted for theanalysis of DA. The sensor is based on a competitive

indirect ELISA and involves the use of disposable SPEs.The signal readout is performed by DPV. The workingelectrode was modified with a BSA-DA conjugate. DAantibodies were added to the sample. Enzyme-amplifieddetection with a secondary antibody conjugated to APenabled a LOD of 5 ng/mL of DA in spiked mussels tissueand shellfish in a linear range of 570 ng/mL.

Immunosensors integrated in FIA systems Semi-automatedanalysis of OA in mussels can be carried out by means of achemiluminescent immunosensor integrated to a FIA system[187]. The indirect competitive assay is carried out oncommercially available polyethersulfon membranes usingimmobilized haptens. The competing OA in the sample iscaptured by the addition of HRP-labeled anti-OA. An overallmeasurement time of 20 min and a dynamic measurementrange from 200 ng to 200 mg OA/100 g mussel homogenateassure a high applicability. A reproducible response wasobtained during the first 34 measurements.

Table 5 Sensors for the detection of mycotoxins


Aflatoxin B1 Electrochemical immunosensor Barley 2 ng/g [144]

Ochratoxin A Electrochemical immunosensor Wheat extract 400 ng/kg [145]

Ochratoxin A Electrochemical immunosensor Wine 10 ng/mL [146]

Aflatoxin B1 Electrochemical sensor Corn 30 pg/mL [147]

Deoxynivalenol Electronic nose Barley grains ns [148]Ochratoxin A

Aflatoxin B1/G1 SPR Maize 0.97 ng/g [149]

Deoxynivalenol SPR (BIACORE Q) Wheat, wheat powder ns [150]

Deoxynivalenol SPR (BIACORE 2000) 500 pg/g [151]Aflatoxin B1 20 pg/g

Fumonisin B1 50 ng/g

Ochratoxin A 100 pg/g

Aflatoxin M1 Portable field immunosensor Milk 50 pg/mL [152]

Fumonisin B1 Fiber-optic immunosensor Ground corn 10 ng/mL [153]

Fumonisin B1 Evanescent wave-based fiber-opticimmunosensor

maize 400 ng/g [154]Aflatoxin B1

Ochratoxin A Fluorescence-based NRL Wheat pasta, cornmeal,cornflakes, barley, coffee

7100 ng/g [155]

sensor Wine 38 ng/mL

Deoxynivalenol Fluorescence-based NRL Oats, barley, cornflakes,cornmeal, meat

150 ng/g [156]sensor

Ochratoxin A Amorphous silicon sensor Wine 100 pg [157]

Aflatoxin B1 Fiber-optic immunosensor Peanut meal, corn meal 1 ng/mL [158]

Aflatoxin M1 Amperometric immunosensor Milk ns [159]

Aflatoxin B1 Optical waveguide lightmodespectroscopy sensor (OWLS 1000)

Barley flour, wheat flour ns [160]Ochratoxin A

Aflatoxins Flow-through luminescence-detection microarray Raw food extracts ns [161]Ochratoxin A

ns not specified



Tissue biosensor An integrated tissue biosensor for mea-suring sodium channel blockers such as TTX and STX wasdescribed by Cheun et al. [188]. Muscle, testis, skin, liver,and ovary of Takifugu niphobles and muscle and skin ofTakifugu parudale (swellfish) have been examined for TTXand Zosims aeneus (crab) for STX. The tissue sensorconsists of a Na+-sensitive electrode in a flow cell. Theelectrode tip is covered with frog bladder membrane (Ranacatesbeiana membrane section from frog bladder) sand-wiched between two sheets of cellulose acetate. Thedistinctive features of this tissue biosensor are its long lifecycle (250-h reproducibility of analysis results), the shortassay times (5 min), and the low limit of detection (86 fg oftoxin). Measurements with an advanced tissue sensor werecarried out to determine the paralytic shellfish toxins suchas GTX [189]. GTX was determined in different seafoodsamples. Linear ranges of sensor response were achievedbetween 86 and 1,000 fg for TTX and 5 and 1,000 fg forSTX.

Sensors for major interesting food-contaminating bacteria

As the bacterial pathogens Salmonella spp., L. monocyto-genes, Campylobacter jejuni, and E. coli O157:H7 havebeen assessed to be responsible for approximately 67% oflethal food poisonings, their rapid and reliable detection is amajor task in consumer protection [190]. Water- andfoodborne infections are a problem that particularly affectsdeveloping countries. C. jejuni is the most prevalent causeof diarrheal and intestinal disease [191, 192]. Food, inparticular poultry, is the predominant source of its growth.An average of 6570% of retail poultry carcasses and 8095% of carcasses from processing plants is contaminated.Infection typically results from ingestions of non-pasteurized milk and milk products, undercooked meat, oreggs. Major symptoms of intoxication are diarrhea, fever,abdominal and muscle pain, nausea, and headache. Theminimum infective dose in clinical studies on humans wasfound to be 500 bacteria for C. jejuni [193].

E. coli O157:H7 is an enteropathogenic bacterium whichreleases Shiga toxins. Infection often leads to bloodydiarrhea, painful abdominal cramps, and, occasionally, tokidney failure [194196]. Most cases of contaminationhave been associated with undercooked ground meat or rawmilk. Infection can also occur after swimming in ordrinking sewage-contaminated water [194, 197, 198]. TheUS Food Safety Inspection Service established a zerotolerance threshold for E. coli O157:H7 in raw meatproducts [199]. The infectious dosage is ten cells, and theFederal EPA standard in water is 40 cells per liter [200].

NH2 O

ONH

N N

NH

NH2OHOH

HN

O O

OH

OH

OHOH

OH

NH

OHNH

NH

Fig. 11 Chemical structures of the marine toxines saxitoxin (left) andtetrodotoxin (right)

Fig. 10 Electrochemical immu-noassay for ricin determination(adapted from [173])



L. monocytogenes has been regarded an importantfoodborne pathogen since its discovery in 1926 and wasthe cause of some examined food poisoning outbreaks[201204]. Infections with Listeria cause Listeriosis whichcan induce spontaneous abortions, premature delivery,stillbirths, and infection of newborns [205]. Symptoms arediarrhea, nausea, vomiting, slight headache, and mild fever[206]. Listeriosis has been associated mainly with dairyproducts or chicken. The intoxication is accompanied by ahigh mortality rate, but its incidence is low compared toother food-associated infections.

Shigella dysenteriae are another non-spore forming typeof bacteria. They can cause shigellosis (bacillary dysentery)with a minimum human effective dose of ten bacteria [207209]. Illness arises through the ingestion of fecallycontaminated food and water, and therefore, insufficientpersonal hygiene and sanitation are the prevalent sources. S.dysenteriae causes the most severe dysentery because of itspotent and deadly Shiga toxin [210]. Infection symptomsinclude diarrhea, fever, abdominal and muscle pain, nausea,and headache.

A number of foodborne illness outbreaks have occurredwhere the source of contamination has been traced toSalmonella species. Salmonella is an enterobacterium thatcauses typhoid fever, paratyphoid fever, and salmonellosis.Contamination can occur via a wide range of food products[211, 212]. Although established microbial methods arepredominantly used for the determination of bacteria due totheir simple applicability and standardized operation proto-cols, there is also a demand for rapid and straightforwardsensing techniques.

Immunomagnetic sensors The combination of immuno-magnetic separation and ECL detection is a promisingapproach for an efficient screening of bacterial contami-nations in food samples as cell culture-based methods arerelatively slow. A commercially available sensor [213] wasevaluated for the determination of E. coli O157 andSalmonella typhimurium in food matrices [214]. Theimmunomagnetic separation was carried out in milk, juices,

supernatants from ground beef, chicken, and fish suspen-sions followed by ECL detection with the ORIGENanalyzer. The assay was processed as described in S.aureus enterotoxins [117] in less than 1 h. Antibody-coated magnetic beads are capable of binding high cellamounts. A limit of 1001,000 bacteria per milliliter couldbe detected.

An improved enzyme-linked immunomagnetic chemilu-minescence (ELIMCL) method has been developed tolower incidences of false positive results of E. coli O157:H7 in mixed cultures. Highly specific combinations of anti-O157 and anti-H7 antibodies in the sandwich immunoassayformat and a semi-automated version of ELIMCL areproviding reliable results. The total assay time is 6.5 h. Adetection limit of 400 cfu/g of E. coli cells in inoculatedground beef was attained [215]. The method scheme ispresented in Fig. 14.

Fiber-optic immunosensors An antibody-based fiber-opticimmunosensor for the determination of Listeria cells waspresented by Geng et al. [216]. Its principle relies on asandwich assay format and uses polyclonal capture anti-bodies which are immobilized on a polystyrene waveguidevia biotinstrepavidin binding. Cy5-conjugated detectorantibodies were incubated along with the sample. Thecomplete assay could be performed in less than 24 h fromsampling including an enrichment step. Since Listeria isproblematic in ready-to-eat meats, spiked hot dog andBologna sausage samples were tested after 20 h ofenrichment. Results showed that L. monocytogenes cellsare detectable directly from the enrichment broth even withlow levels of inoculations (10 to 1,000 cells per gram).

The occurrence of Listeria in sausages can also berevealed with the RAPTOR automated fiber-optic sandwichimmunosensor system [217]. It is based on an evanescentwave excitation generated by four 635-nm diodes coupledto four optical fibers. These are assembled in a disposablecoupon with a fluidic channel flow system for automatedoperation. The fluorescence emission of Cy5-conjugatedmonoclonal detector antibodies is measured by means ofphotodiodes at 670-nm wavelength. After inoculation ofsausage samples with Listeria cells and 20 h of cellenrichment, a minimum of 5105cfu/mL of Listeria cellscould be determined.

Another evanescent wave fiber-optic biosensor with highpotential for rapid, sensitive, and real-time microbial

Fig. 13 Chemical structure ofdomoic acid

OH

OH

O

OHOHO

H OO

O

HOH

OHO

O

Fig. 12 Chemical structure ofokadaic acid



analysis was developed by Geng et al. [218]. They wereable to detect low levels of E. coli O157:H7 in groundbeef. Their sensor is also based on a direct sandwichimmunoassay. Polyclonal capture antibodies specific forE. coli were immobilized on polystyrene fiber waveguidesvia biotinstreptavidin coupling. Indicator antibodieslabeled with AlexaFluor 647 were used for cell detection.The analysis is performed by means of an Analyte-2000system [219] employing evanescent field excitation of thefluorophores. The emitted light is transmitted by the fiberand quantified by a photodiode. A detection limit of 1 cfu/mL of E. coli in spiked ground beef could be attained afterpre-enrichment.

Piezoelectric sensors Campbell et al. [220] presented apiezoelectric millimeter-sized cantilever sensor for E. colianalysis in raw or sterilized ground beef. An E. coli O157:H7 specific antibody was immobilized on a PZT-layerbonded on borosilicate glass. Binding of target cellsinduces a significant resonance frequency change. In thisway, 50100 cells per milliliter could be determined in a 3-mL broth sample containing meat particles.

A piezoelectric quartz crystal microbalance sensor with agold-plated AT-cut surface was presented by Minunni et al.[221]. In this case, the surface was initially coated withprotein A or G. This facilitates the immobilization of heat-killed Listeria cells for a competitive assay. Listeria cells inthe sample compete for antibodies with immobilized cellson the crystal surface. The binding of antibodies on thesensor can be monitored. This very fast assay (15 min)provides a sensitivity of 3.19106 cells/crystal.

Amperometric sensors Listeria cells in low-fat and infantformula milk can be determined by means of an immuno-sensor SPE device. Sandwich assays with immobilizedcapture antibodies were employed. After the binding oftarget cells, a second primary antibody was incubatedfollowed by the addition of a secondary AP-conjugatedantibody. The enzymatic formation of p-aminophenol fromp-aminophenyl phosphate can be monitored amperometri-cally as the former can be oxidized at the electrode. The

assay provides a dynamic range of 1.3106 to 1.3103

cells per milliliter with a LOD of 9.3103 cells permilliliter milk and can be carried out within 3.5 h [222].

Surface plasmon resonance sensors The Plasmonic sur-face plasmon resonance device (HS Systeme, Wallenfels,Germany) [223] was adapted for a rapid, simple, andspecific screening of S. typhimurium in milk [224]. Acuvette was pla

9783642363023-c1

Documents

food matrices

contaminated food

marine toxins

clostridial toxins

keywords food safety

food safety control

sensor arrays

toxin immunosensors