317
Review
Application of Electrochemical Biosensors for Detection of Food Pathogenic Bacteria Dmitri Ivnitski,+ Ihab Abdel-Hamid,+ Plamen Atanasov,+ Ebtisam Wilkins,*+ and Stephen Stricker++ + ++
Department of Chemical and Nuclear Engineering, University of New Mexico, Albuquerque, New Mexico 87131, USA; e-mail:
[email protected] Department of Biology, University of New Mexico, Albuquerque, New Mexico 87131, USA
Received: September 29, 1999 Final version: November 16, 1999 Abstract
Current practices for preventing microbial diseases rely upon careful control of various kinds of pathogenic bacteria in food safety and environmental monitoring. The main disadvantages of conventional bacterial detection methods are the multistep procedure and long time requirements. This article gives an overview of alternative electrochemical biosensors for detection of pathogenic bacteria in the food industry. Focus has been on new microbial metabolism-based, antibody-based and DNA-based biosensors. The underlying principles and applications of these biosensors are discussed. Recent developments in ¯ow-injection biosensor systems with an electrochemical detection are also presented. Keywords: Electrochemical immunosensors, DNA recognition, Immuno®ltration, Pathogenic bacteria, Food quality control
1. Introduction Food processing quality control is a national priority. Contamination of foods by bacterial pathogens (such as Escherichia coli, Salmonella typhimurium, Campylobacter jejuni, Legionella pneumophila, Staphylococcus aureus, Streptococci, etc.) results in numerous foodborne diseases [1±4]. Infectious diseases account for nearly 40 % of the total 50 million annual estimated deaths worldwide. Microbial diseases constitute the major cause of death in many developing countries. A growing number of bacterial pathogens have been identi®ed as important food- and waterborne pathogens [5±8]. Estimates of the yearly incidence of foodborne illness vary widely from one million cases to 81 million cases in the USA, with bacterial foodborne outbreaks accounting for 91 % of the total outbreaks [2, 9]. In fact, the incidences of human diseases caused by foodborne pathogens, such as Salmonella typhimurium, Escherichia coli, Staphylococcus aureus, Campylobacter jejuni, and Bacillus cereus has not decreased since 1994. Examples of these are the 1997 Hudson ground beef recall and the 1996 incident where more than 9000 fell ill and 313 died due to E.coli O157:H7 contamination [6]. Current practices for preventing microbial diseases rely upon careful control of various kinds of pathogenic bacteria in clinical medicine, food safety and environmental monitoring. Approximately 5 million analytical tests (for Salmonella only) are performed annually in the United States [10±12]. Conventional bacterial identi®cation methods usually include a morphological evaluation of the microorganism as well as tests for the organism's ability to grow in various media under a variety of conditions. Although standard microbiological techniques allow the detection of a single bacteria, ampli®cation of the signal is required through growth of a single cell into a colony. Microbiological methods generally have 4 distinct phases [12]: 1) preenrichment, to allow growth of all organisms; 2) selective enrichment, to allow growth of the organism under investigation and to increase bacterial population to a detectable level; 3) isolation, by using selective agar plates; and 4) conformation, serological and biochemical tests to con®rm the identi®cation of Electroanalysis 2000, 12, No. 5
a particular pathogenic organism. The main disadvantages of the conventional methods are the multistep assay and the time consuming process. Completion of all phases requires at least 16 h and can take as long as 48 h. The detection limit is usually 105± 106 cells=mL without pre-enrichment. Thus, conventional techniques are not suitable for fast and direct analysis of bacteria without pre-enrichment. To meet expectations of users, analytical instruments for bacteria must satisfy the requirements outlined in (Table 1): the analytical devices must have the speci®city to distinguish between different bacteria, the adaptability to detect different analytes, the sensitivity to detect bacteria on-line and directly in real samples without pre-enrichment. Time and sensitivity of analysis are the most important limitations on the usefulness of microbiological testing. Effective screening of food samples requires a rapid, selective and highly sensitive analytical device. The presence of even a single pathogenic organism in the body or food may be an infectious dose. The device must also be simple Table 1. Summary of the requirements for bacterial biosensors. Low detection limit Species selectivity Strain selectivity Assay time Precision Assay protocol Measurement Format Operator Viable cell count Size
WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2000
Ability to detect single bacteria in a reasonably small sample volume (from 1 to 100 mL). Ability to distinguish individual bacterial species in the presence of other microorganisms or cells. Ability to distinguish an individual bacterial strain from other strains of the same species. 5±10 min for a single test. 5±7 %. No reagent addition needed. Direct, without pre-enrichment Highly automated format (``single button device''). No skill required to use the assay. Should discriminate between live and dead cells. Compact, portable, hand-held, design for ®eld use. 1040±0397/00/0503-0317 $17.50.50=0
318
and inexpensive to design and manufacture. These requirements are satis®ed using biosensor technology. Biosensors for bacterial detection generally involve a biological recognition components such as receptors, nucleic acids, or antibodies attached to an appropriate transducer. Depending on the method of signal transduction, biosensors may be divided into four basic groups: optical, mass, electrochemical, and thermal [13±17]. In addition, biosensors can be classi®ed into two broad categories: for direct detection of the target analyte and label-based sensors. Direct detection biosensors are designed in such a way that the biospeci®c reaction is directly determined in real time by measuring the physical changes induced by the formation of the complex. Label-based biosensors are those in which a preliminary biochemical reaction takes place and the products of that reaction are then detected by a sensor. Electrochemical biosensors have some advantages over other analytical systems in that they can operate in turbid media, offer comparable instrumental sensitivity, and more amenable to miniaturization. Modern electroanalytical techniques have very low detection limits (typically10ÿ9 M) that can be achieved using small volumes (1± 20 mL) of samples [18]. Furthermore, the continuous response of an electrode system allows for on-line control. The equipment required for electrochemical analysis are simple and cheap compared to most other analytical techniques.
2. Microbial Metabolism-Based Biosensors Microorganisms are able to transduce their metabolic redox reactions into quanti®able electrical signals by oxidoreductase reactions using an appropriate mediator, the microbial content of a sample can be determined by monitoring microbial metabolism. Various combinations of biosensors based on the monitoring of microbial metabolism have been reported [19±26]. These techniques are based on direct measurements of a physical phenomena occurring during the biochemical reactions on a transducer surface. Signal parameters such as pH change, oxygen consumption, ion concentrations, potential difference, current, or resistance, can be measured by electrochemical transducers. The transducer can either detect consumption of oxygen or the appearance=disappearance of an electrochemical active metabolite. Electrochemical biosensors based on CO2 and the Clark-type oxygen electrode have been introduced for the determination of bacteria [24]. This simple system, however, suffered from low sensitivity due to the high background noise. The assay time for detection of 106 cells=mL varied from 2 to 4 h. Electrochemical biosensors based on the Clark-type oxygen electrode have been introduced for the determination of E. coli, S. aureus and Enterococcus serolicida [27]. The cell suspension was ®ltered through a cellulose nitrate membrane (pore size, 0.45 mm). The membrane along with the captured cells was set on the platinum working cathode of a Clark oxygen electrode and covered with a dialysis membrane. The microbial electrode was immersed in 0.05 M phosphate buffer until the output current became stable. The electrode was then taken out and placed in a solution containing sodium azide, which suppressed the growth of most microorganisms except Enterococcus serolicida. The sensor output was linear in the range of 1.46107 to 7.26107 cells=mL with an assay time of 2 hours. The linear range of the oxygen electrode is limited because of low oxygen concentrations. Also, the response ¯uctuated as a result of variations in circulating oxygen concentrations. Electroanalysis 2000, 12, No. 5
D. Ivnitski et al.
Takayama et al. [23] demonstrated mediated electrocatalysis based on the bacterium Gluconobacter industrius. It was shown that the bioelectrocatalytic behavior of the Gluconobacter industrius-BQ-electrode system is very similar to that observed with a glucose oxidase-BQ-electrode system. Glucose dehydrogenase (GDH), an enzyme present in the bacterial cell membrane, was the catalyst for producing the BQ-mediated electrocatalytic current. A current magnitude as high as 200 mA=cm2 was obtained with 2.46106 cells in the presence of BQ and 10 mM glucose. A steady-state current was reached in about 30 seconds. Mediators such as K3Fe(CN)6 and dichlorophenol indophenol are capable of accepting electrons from GDH or from some other part of the respiratory chain in the cell membrane. An interesting approach has been developed for rapid determination of E. coli using a ¯ow-injection system [28]. Electrochemical measurement of K3Fe(CN)6, reduced by microbial metabolism allowed the quantitative determination of fungi and bacteria in 20 minutes. E. coli was detected in the range of 4.76106 to 2.46109 CFU=mL when the microorganisms were not separated from the cultivation medium. Hitchens et al. [29] illustrated a method for enumerating microorganisms which combines electrochemical detection with ¯ow injection analysis. This method is based on the measurement of electrical currents generated by active microbes in the presence of redox mediators (i.e., low molecular weight electron acceptors) that can diffuse through the bacterial cell membrane. Analysis of samples could be performed in 10 min at a lower detection limit of 105 cells=mL. The results also provided new insights into factors that limit the lower limits of sensitivity of the mediated amperometric detection system. Microbial metabolism usually results in an increase in both conductance and capacitance, while causing a decrease in impedance. Therefore, impedance, conductance, capacitance and resistance are only different ways of monitoring the test system and are all interrelated [30, 31]. The impedance method was accepted by the Association of Of®cial Analytical Chemists, Intl., (AOAC) as a ®rst action method [32]. This method is well suited for detection of bacteria in clinical specimens, to monitor quality and to detect speci®c food pathogens, also for industrial microbial process control, and for sanitation microbiology [33]. This technique has been used for estimating microbial biomass, for detecting microbial metabolism [34, 35], the concentration and physiological state of bacteria [36±38]. Current instruments usually detect active metabolizing bacteria when 106 to 107 bacteria per milliliter are present in the culture media. A very important parameter is the enumeration of viable cells. Viable cells are commonly measured microscopically after suspending the cells in a dye such as Trypan Blue. A new biosensor for real-time monitoring of concentration, growth and physiology of cells in a culture media was proposed [38]. This biosensor is based on impedance measurement of adherently growing cells on the electrode surface. Cell density, growth, and long-term behavior of cells on the electrodes change the impedance of the biosensor. The main effect of cells on the sensor signal is due to the insulating property of the cell membrane. The biosensor provides information about spreading, attachment and morphology of cultured cells. A reusable bulk acoustic wave (BAW) ± impedance sensor has been developed for continuous detection of growth and numbers of Proteus vulgaris on the surface of a solid medium under ordinary conditions [39, 40]. The proposed sensor relies on the fact that bacteria can transform uncharged or weakly charged substrates into highly charged end products causing an alteration
Detection of Food Pathogenic Bacteria
in the conductance of the medium. The sensor is simple, rapid and bacteria can be detected using the proposed method in the range of 3.46102 to 6.76106 cells=mL. The main drawbacks of biosensors based on the monitoring of microbial metabolism are related to their poor selectivity because of the possible presence of enzymes from sources other than the bacteria of interest. This may be avoided, however, by using antibody modi®ed sensing elements.
3. Antibody-Based Biosensors Almost all microorganisms can now be sensed electrochemically by their involvement in a bioaf®nity reaction. Amperometric immunosensors aimed at microbial analysis have recently been reported [41±47]. In the work of Nakamura et al. [42], an electrode system consisting of a basal-plane pyrolytic graphite (BPG) electrode and a porous nitrocellulose membrane ®lter to trap bacteria was used for the detection of bacteria in urine. The peak current of a cyclic voltammogram increased with increasing initial cell concentration of Escherichia coli in urine. Urine containing 56102 to 56105 cells=mL was measured with this system. Mirhabibollahi et al. [41] utilized an enzyme-linked amperometric immunosensor for the detection of Staphylococcus aureus and Salmonella in pure cultures and in foods. This immunosensor could detect 104 to 105 CFU=mL of Staphylococcus aureus. However, the electrochemical detection step was awkward to perform, and there were variations in the signals produced by different strains of bacteria. This approach was modi®ed in a later work by the same authors [43] utilizing alkaline phosphatase as the enzyme-marker and phenyl phosphate as the substrate followed by the amperometric detection of phenol. They also proposed another system which incorporated an enzyme ampli®cation step and relied on the amperometric detection of reduced mediator (ferrocyanide). Both systems were able to detect low numbers (1±5 CFU=g or =mL) of Salmonella in food after nonselective (18 h) and selective (22 h) enrichment steps. A novel liposome-based amperometric biosensor for the detection of haemolytic microorganisms was developed (Fig. 1) [44]. Haemolytic organisms are able to attack and rupture the lipid bilayer of liposomes to release mediator (2,6-dichlorophenolindophenol) and generate a signal, whereas a weak or negligible response is produced by non-haemolytic species. The potential of this approach was illustrated for detection of various
Fig. 1. Schematic illustration of the principle of amperometric signalgeneration by haemolytic bacteria. 1) Liposome containing entrapped oxidized mediator; 2) action of bacterial enzymes releases mediator, which is reduced by the bacteria; 3) reduced mediator is reoxidized at a platinum working electrode. Adapted from Kim et al. [44].
319
strains of Listeria monocytogenes, Listeria welshimeri and Escherichia coli. Bacterial concentrations were measured in the range of 4.76106 to 2.46109 CFU=mL. Recently immunomagnetic beads have been applied in immunoelectrochemical assays for the detection of Salmonella typhimurium [45±47]. This technique combines the selectivity of antibody-coated superparamagnetic beads with the rapidity and sensitivity of electrochemical detection of bacteria in a format termed enzyme-linked immunomagnetic electrochemistry. In this case, heat-killed Salmonella typhimurium were sandwiched between antibody-coated magnetic beads and an enzyme-conjugated antibody. With the aid of a magnet, the beads were localized onto the surface of disposable graphite ink electrodes in a multiwell plate format. After magnetic separation, the liquid was removed by aspiration and p-aminophenylphosphate was added to the electrochemical cell, p-aminophenol, the product of the enzymatic reaction, was measured using square-wave voltammetry. Using this technique, a minimum of 86103 cells=mL of Salmonella typhimurium in buffer was detected in approximately 80 min. This device is faster than standard microbiological methods. However, its sensitivity is not enough for direct detection of a speci®c pathogen in real samples and foods must ®rst undergo traditional culture enrichment. Therefore, even though the identi®cation steps may be rapid, the overall time of analysis remains dependent on lengthy growth procedures. The main parameters affecting biosensors for bacterial detection are [48, 49]: the binding af®nity of the antibody for antigen; the nature of the enzyme-marker; the concentration of conjugate and the nature of the transducer. In many practical situations, the mass transport of analytes from the bulk solution to the solid surface of a sensing element is a crucial parameter in manufacturing immunosensors with a rapid response time and high sensitivity [50±52]. Experimental data on the effect of mass transport controlled antibody-antigen binding reactions at solid surfaces have been reported in the literature [48]. The time to reach equilibrium in the diffusion controlled process is extremely long. Therefore, the general approach to achieve signi®cantly short assay time is to reduce transport limitations across the unstirred layer of solvent (Nernst layer) to the electrode surface. The acceleration of the diffusion-controlled rate of immunological and enzymatic reactions on the solidsolution interface has been accomplished by: intensive mixing of the solution (liquid phase) [53]; conducting immunoreactions in an ultranarrow microcapillary immunoreactor or porous material [54]; the utilization of highly dispersed carbon-based immunosorbents as electrode materials [55]; or by using an enzymechanneling approach which allows in situ generation of the substrates of the enzyme-label [56]. A separation-free, amperometric enzyme-channeling immunosensor that does not involve a washing step was developed for Staphylococcus aureus [56]. The ampli®cation of the analytical signal was achieved by combining enzyme-channeling reactions, optimizing hydrodynamic conditions, and electrochemical regeneration of mediators within the membrane layer of an anionexchange polyethylenimine-glucose oxidase-antibody modi®ed electrode. This approach is based on using sequential enzymatic reactions, where the product of one enzyme is a substrate for a second one. A disposable graphite working electrode (made from pencil lead) is covered by PEI ®lm with coimmobilized glucose oxidase (GOD) and bacteria-speci®c antibodies (Abs). When the bacteria sample, conjugate, and glucose are added to the electrochemical cell, an immunological reaction occurs between bacteria and Ab, from one side, and an immunological reaction Electroanalysis 2000, 12, No. 5
320
occurs between bacteria and conjugate from the other side, brings the two enzymes (GOD and peroxidase) in close physical proximity to the electrode surface. The ampli®cation of the analytical signal is ®rst achieved by an enzyme-channeling effect. The product of glucose oxidase reaction (H2O2) is used as a substrate for peroxidase. Then, the positively charged PEI ®lm on the electrode surface is effective not only for immobilizing the biomolecules, but also for accumulating high concentrations of mediators (iodide ions) close to the electrode surface. Finally, the cyclic electrochemical regeneration of the iodide ions within the PEI ®lm is another ampli®cation factor. The immunosensor enables preferential measurement of surface-bound conjugate molecules relative to the excess enzyme-labeled reagents in the bulk sample solution. Staphylococcus aureus cells were detected in pure culture at concentrations as low as 1000 cells=mL in a relatively short assay time of 30 min. A novel approach based on using partially immersed immunoelectrodes has been demonstrated for fast immunoassay of E.coli O157:H7 [57]. Antibodies against E.coli O157:H7 were immobilized on the surface of carbon rods (2 mm diameter) (Fig. 2). As the electrode is raised from the fully immersed position, a liquid meniscus is formed at the gas=liquid=solid interface. In addition to this meniscus the hydrophilic protein layer (antibody
D. Ivnitski et al.
modifying the electrode surface) facilitates formation of the supermeniscus an additional thin liquid ®lm, supermeniscus, on the electrode surface (placed above the meniscus). This supermeniscus is characterized by a thickness of approximately 0.4± 0.8 mm. The meniscus and supermeniscus play an important role in providing hydrodynamic conditions of the thin ®lm in the process of current signal generation. It has been shown that the sensitivity of the partially immersed electrode was 6 fold greater than that of the fully immersed one. The partially immersed immunoelectrode allows determination of Escherichia coli cell concentrations in the range of 150 to 7000 cells=mL. The sensitivity increased on the partially immersed electrodes may be explained by facilitated diffusion of reagents (antigen, conjugate, substrate and product of enzymatic reaction) to the electrode surface in the upper part of the meniscus and in the supermeniscus. Since the thickness of the supermeniscus is usually much less than the thickness of the diffusion layer in the bulk solution, the effective working volume of the electrolyte participating in the measurement becomes very small. This may result in increased local concentration of the product of the enzymatic reaction which increases the analytical signal and reduces the response time. The capillary phenomena associated with the observed effects could also be explored in other system formats such as a high-surface-area dispersed electrode material, porous electrodes or hollow capillary tubes. This new immunoassay approach can be easily extended to the detection of other bacteria and may be a basis for creating new, highly sensitive and rapid biosensors for bacteria.
4. Antibody-Based Flow-Injection Biosensors
Fig. 2. Schematic illustration of the principle of enzyme immunoassay of E.coli O157:H7 with partially immersed antibody-modi®ed electrode. Electroanalysis 2000, 12, No. 5
In many situations and in order to increase assay sensitivity, it would be desirable to concentrate the bacteria into a smaller volume prior to the assay. Several possible formats for concentrating cells in analytical systems were described by Wyatt (Fig. 3) [58]. Of these, the most attractive technique for the concentration of bacteria is membrane ®ltration in conjunction with ¯ow systems [59, 60]. A ¯ow-injection con®guration of immunosensor has such important properties as [61, 62]: transferring the injected or aspirated sample to the sensor; conditioning the sample (pH adjustment, mixing with other reagents) for optimal development of the reaction and detection, regeneration of the sensor between samples; facilitating reliable calibration and increasing the sensor selectivity and sensitivity via a continuous separation module. The typical output of a ¯owinjection system is a peak that results from the dispersion of the injected sample. Perez et al. [63] proposed an amperometric ¯ow-injection system for the measurement viable Escherichia coli O157 (Fig. 4). This system is based on the selective immunological separation of Escherichia coli O157 using antibody-coated magnetic particles and the generation of a signal by bacterial cells. For the immunological step, immunomagnetic beads were selected as the immunocapture reagent. Electrochemical detection was carried out using redox mediators, potassium hexacyanoferrate(III) and 2,6-dichlorophenolindophenol. The measurement was performed using a FIA system. The detection limit was 105 CFU=mL, and the complete assay was performed in 2 h. This technique could easily be automated and the analysis can be performed quickly and continuously. The renewal of the immunosensor sensing surface was accomplished by removing
Detection of Food Pathogenic Bacteria
Fig. 3. Some possible formates of immunocapture techniques. i) Initial mixed population of bacteria; ii) capture on solid phase; iii) puri®ed (or enriched) bacterial population; iv) detection stage. Adapted from Wyatt [58].
Fig. 4. Schematic model of the immunomagnetic separation with mediated ¯ow injection analysis of viable Escherichia coli O157. A) Selective capture of E.coli O157 using antibody-derivatized magnetic particles; B) reaction of bacteria with a mediator; (c) electrochemical measurement of the reduced mediator using an amperometric method. Adapted from Perez et al. [63].
321
the magnet and the washing down the magnetic particles. The immunosensor was then ready for the injection of new antibody modi®ed magnetic particles for another analytical cycle. The system has been applied to detect Bacillus anthrax spores, Escherichia coli O157 and Salmonella typhimurium [45±47, 64± 67]. This format offers several advantages for automatic immunoassays in that the many required washing steps are inherent and the immunoassays can be carried out rapidly. This format can be automated easier than formats using tubes, microtiter plates or other similar reaction vessels. In many analytical systems, the separation and biosensing are performed in several steps with a manual transfer step in between. The multistep assay and manual transfer can result in sample loss and errors due to contamination. An alternative approach is the on-line combination of a biosensor with a separation method. The combination of prior separations (electrophoresis, chromatography or immuno®ltration) with biosensor technology can dramatically improve speed, selectivity, and sensitivity of assays with minimal sample manipulation. The advantage of an on-line separation=biosensor system is that individual bacterial cells within a complex mixture are ®rst puri®ed and then detected by selective interactions with analysis times typically around 30 minutes. Immuno®ltration-based separations offer an attractive approach that may be performed in the subsecond time regime. Flow immuno®ltration sensors can be an excellent alternative for detection of bacterial pathogens because it not only overrides the effects of diffusional limitations, but also allows the concentration of bacteria on the membrane by ®ltering a large volume of the sample. Heterogeneous ¯ow immuno®ltration assays offer extremely accelerated binding kinetics. First, there is a high surface area to volume ratio in the immunosorbent. Second, the ¯owing stream actively brings the sample in contact with the solid-phase antibody. This factor results in a greatly enhanced antigen-antibody encounter rate and in nearly quantitative immunobinding during the short immunoreaction time. This approach has also dramatically increased the potential for automation of immunoassays. Clark et al. [59] described an apparatus for use in an enzyme linked immuno®ltration assay (ELIFA) which incorporated a peristaltic pumping system allowing continuous ®ltration of reagents through a nitrocellulose membrane clamped between two 96-well plates. An on-line biosensor assay system based on the combination of several signal-ampli®cation systems has been developed [68, 69]. The acceleration of the diffusion-controlled rate of immunological, enzymatic and electrochemical reactions has been achieved by combining a ¯ow-injection immuno®ltration technique with an electrochemical detection. The ¯ow-injection immuno®ltration sensor for bacteria has four main components (Fig. 5): a peristaltic pump, a low pressure sample injection valve (which is supplied with a ®xed volume loop), the amperometric immunosensor assembly, and an electrochemical=data recorder interface. The immunosensor consists of a disposable antibodymodi®ed ®lter membrane resting on top of the working electrode. Total E. coli, E. coli O157:H7, and Salmonella were selected as model organisms. A sandwich scheme of immunoassay was employed. A sample is injected by the pump through the porous membrane of the immunosensor. Antibodies coupled to porous membrane are used to capture and accumulate speci®c bacterial pathogens. Bacteria are concentrated on the membrane surface from a relatively large volume of sample, thereby eliminating the need for selective enrichment. This is followed by a washing step, and then a conjugate solution is injected through the membrane. The bound and unbound peroxidase labeled conjugate on the Electroanalysis 2000, 12, No. 5
322
D. Ivnitski et al.
Fig. 6. Detail of surface topography of porous nylon membranes with E.coli O157:H7 cells.
element in all channels, or the simultaneous real-time control of several types of pathogenic bacteria in the same sample using different biorecognition elements in individual channels.
Fig. 5. Schematic illustration of the ¯ow-injection electrochemical biosensor for bacteria.
5. DNA Biosensors
membrane surface are separated by washing. The last step is the electrochemical measurement of the total peroxidase activity on the porous membrane surface. The analytical signal of the enzymatic reaction was estimated as the initial rate of currentchange (nA=s). The ¯ow-injection amperometric immuno®ltration system demonstrated high sensitivity and good reproducibility. The coef®cient of variation between membranes prepared from the same batch was less then 8 %. The working range for E. coli O157:H7 detection is from 50 to 300 cells=mL and the overall analysis time was 30 min. The sensitivity of the amperometric ¯ow-injection immunoassay system to E. coli O157:H7 is 1000 fold greater than that of the standard ELISA assay. A number of factors contributed to this ampli®cation. First, immuno®ltration accelerates the diffusion-controlled rate of immunological and enzymatic reactions to within the surface of the sensing element of immunosensor. Second, microporous membranes offer 100 to 1000 times more available surface area for immobilization of antibodies than solid wells currently used in ELISA [70]. The bacteria may be concentrated on the membrane surface from a large volume of sample. It was found that cell adhesion to the antibody modi®ed membrane surface was signi®cant. At the same time, the degree of adhesion of bacteria to the unmodi®ed membrane surface was low. The results were con®rmed by direct examination of the cell adhesion of E. coli O157:H7 to the antibody modi®ed porous nylon membranes surface (1.2 mm pore size) using scanning electron microscopy (Fig. 6). The immunosensor may be easily transformed to a multichannel analytical device for the detection of different microorganisms in food and water. In this case, the disposable sensing elements could be engineered in the form of multichannel cartridge to analyze either several samples for the content of one selected bacteria using the same biorecognition
In recent years, various kind of electrochemical biosensors based on identi®cation of the bacterial nucleic acid have been developed [71±80]. DNA biosensors are analytical devices that contain immobilized DNA probes that speci®cally hybridize to their complementary sequences in a DNA sample. The basic principle of a DNA biosensor is to detect the molecular recognition provided by the DNA probe and to transform it into the signal using a transducer. Applications of gene probes are associated with ultrasensitive determination of microorganisms, viruses and trace amounts of special chemicals in various environments. Gene probes are already ®nding applications in detection of disease-causing microorganisms in water supplies, food, and plant, animal and human tissues [71±80]. Bacterial and viral pathogens responsible for disease states are detectable because of their unique nucleic acid sequences. Through application of molecular ``probes,'' labeled DNA sequences that are complementary to unique portions of bacterial DNA can be detected and identi®ed [74]. The term nucleic acid (gene) probe describes a segment of nucleic acid which speci®cally recognizes and binds to a nucleic acid target. The recognition is dependent upon the formation of stable hydrogen bonds between the two nucleic acid strands. This contrasts with interactions of antibody-antigen complex formation where hydrophobic, ionic and hydrogen bonds play a role. The bonding between nucleic acids takes place at regular (nucleotide) intervals along the length of the nucleic acid duplex, whereas antibodyprotein bonds occur only at a few speci®c sites (epitopes). Samples containing bacterial nucleic acid are treated (usually by heating) to cause the double strands of nucleic acids to separate and thus become open to hybridization with the nucleic acid probe. Because bacterial nucleic acid may be present in very small quantities, the probing process is preceded by a technique called polymerase chain reaction (PCR), which ampli®es the amount of nucleic acid present [75]. The gene ampli®cation
Electroanalysis 2000, 12, No. 5
Detection of Food Pathogenic Bacteria
323
Table 2. Features of bacterial sensors. BAWI: bulk acoustic wave impedance sensor; IMT: immunomagnetic technique; LBB: liposome-based biosensor; ECI: enzyme-channeling immunosensor; PII: partially immersed immunoelectrode; FIIT: ¯ow-injection immuno®ltration technique. Detection technique
Bacteria
Detection limit [cells=mL]
Assay time
References
106 107 106 36102
2±3 h 2h 20 min ±
[30] [27] [28] [39, 40]
Salmonella Salmonella
1 cfu=g 86103
18 h 80 min
[43] [45±47]
Listeria monocytogenes
107cfu=ml
30±40 min
[44]
Staphylococcus aureus Staphylococcus aureus
1 cfu=g 103
18 h 30 min
[41] [56]
Escherichia coli O157:H7
200
40 min
[57]
Escherichia coli O157
105 cfu=ml
2h
[63]
Escherichia coli O157:H7
50
30 min
[68, 69]
Cryptosporidium parvum DNA M. tuberculosis DNA Escherichia coli, Giardia DNA
ng=mL 200 ng=mL
20 min 20 min
[78] [79]
Microbial Metabolism-Based Biosensors Impedimetry Staphylococcus aureus Amperometry E. seriolocida Amperometry Escherichia coli BAWI Proteus vulgaris Antibody-Based Biosensors Amperometry Amperometry (IMT) Amperometry (LBB) Amperometry Amperometry (ECI) Amperometry (PII) Amperometry (IMT) Amperometry (FIIT) DNA Biosensors Chronopotentiometry Chronopotentiometry
method (the polymerase chain reaction) enhances the sensitivity of DNA probes by at least three orders of magnitude [76]. This technique uses the heat-stable DNA polymerase of Thermus aquaticus and allows short lengths of a double-stranded target DNA (template) to be copied in vitro thousands or millions of times, very quickly. According to Jones et al. [75], a PCR-gene probe based assay has high potential for improving monitoring of foodborn bacteria. To date, only methods involving the polymerase chain reaction (PCR) have been employed to detect foodborne pathogens. Using PCR, bacteria can be detected directly, without cultivation, by extraction and isolation of nucleic acids from real samples, followed by hybridization with speci®c probes. Without any enrichment steps, the PCR method detects less than 40 cells=gram of a given food sample [74]. Different concepts for electrochemical biosensing of DNA sequences (direct electrochemical detection of DNA hybridization, adsorptive stripping analysis, indicator-based and renewable DNA probes) have been considered in a number of excellent reviews [71, 72, 80, 81]. Wang et al. [80] has exploited the sensitivity of the guanine oxidation signal to the DNA structure for detecting the formation of the surface duplex. The decreased guanine response of the immobilized probe was used for detecting the DNA hybridization. A more attractive approach is the use of an inosine-substituted (guanine-free) probe [82]. Recently a DNA hybridization electrochemical biosensor for the detection of DNA fragments of the waterborne pathogen Cryptosporidium have been developed [78]. The sensor relies on the immobilization of an oligonucleotide unique to the Cryptosporidium DNA onto the carbon-paste transducer and employs a highly sensitive chronopotentiometric transduction mode for monitoring the hybridization event. Very short (3 min) hybridization periods give rise to well-de®ned hybridization signals at mg=mL concentrations of the Cryptosporidium DNA sequences, while 20±30 min hybridization periods permit ng=mL detection limit. Similar hybridization=chronopotentiometric schemes are
currently being developed for other pathogens, such as Escherichia coli, Giardia and Mycobacterium tuberculosis [79]. A brief summary indicating some of the biosensors covered in this review is presented in Table 2.
6. Conclusions Analysis of published literature has shown that it is a challenge to create electrochemical biosensors with the necessary properties for reliable and effective use in routine applications. The main reasons for this are both technology and market related. The biosensor system must have the speci®city to distinguish the target bacteria in a multiorganism matrix, the adaptability to detect different analytes, the sensitivity to detect bacteria directly and on-line without pre-enrichment, and the rapidity to give realtime results [83]. At the same time, the biosensor must have relatively simple and inexpensive con®gurations. Another obstacle is the tendency to focus only on the scienti®c basis of the technology while excluding the other equally important aspects. There are a number of practical and technical issues which must be overcome in the development of bacterial biosensors for their commercialization. One of the problems facing the production of biosensors for direct detection of bacteria is the sensitivity of assay in real samples. The infectious dosage of pathogens such as Salmonella or E. coli O157:H7 is 10 cells and the existing coliform standard for E. coli in water is 4 cells=100 mL. The Environmental Protection Agency regulations specify the minimum frequency of water sampling and the maximum number of coliform organisms allowed. Treated drinking water should contain no coliforms in 100 mL [84, 85]. Hence, a biosensor must be able to provide a detection limit as low as a single coliform organism in 100 mL of potable water, with a rapid analysis time at a relatively low cost. Only in this case will the biosensor be convenient for on-line testing of bacterial pathogens Electroanalysis 2000, 12, No. 5
324
in real samples. Thus, sensitivity is another issue that still requires improvement. Technical problems facing biosensor development include the interaction of matrix compounds, methods of sensor calibration, the requirements for reliable and low maintenance functioning over extended periods of time, sterilization, reproducible fabrication of numerous sensors, the ability to manufacture the biosensor at a competitive cost, disposable format, and a clearly identi®ed market. Biosensors for bacteria should minimize human participation (to avoid contamination) and hence, automation must be an inherent attribute of the biosensor. It is our understanding that in the near future the second generation of electrochemical biosensors will be fully automated analytical systems based on combining multisensor technology with arti®cial neural networks ( as in the case of the electronic nose) or with other analytical and discriminative mathematical methods. The potential of the electronic nose within the food and drinks industry lies in the speed and simplicity of the method and in the nondestructive determination of food quality [86-89]. Arti®cial neural networks do not require any expert knowledge once programmed, and the only task of the operator is to indicate the objects to be recognized after which the network functions on its own.
7. Acknowledgements This research was partially supported by grants from the National Science Foundation and the Waste-Management Education & Research Consortium of New Mexico.
8. References [1] M.P. Doyle, L.R. Beuchat, T.J. Montville, Food Microbiology: Fundamentals and Frontiers, ASM Press, Washington DC, 1997, ch. 3, pp. 127±390. [2] G.W. Beran, H.P. Shoeman, K.F. Anderson, Dairy Food Environ. Sci. 1991, 11, 189. [3] R.M. Atlas, Critical Rev. Microb. 1998, 24, 157. [4] Malcolm Dando, Biological Warfare in the 21st Century, Brassey's (UK) London, NY, Macmillan Pab.Co. 1994. [5] B. Swaminathan, P. Feng, Annu. Rev. Microbiol. 1994, 48, 401. [6] USDA, Hudson Foods Recalls Beef Burgers Nationwide for E.coli O157:H7, News Release No.0272.97, Washington DC, Aug.12, 1997. [7] A.M. McNamara, J. Urban Health-Bulletin of the N.Y. Academy of Medicine 1998, 75, 503. [8] L. Slutsker, S.F. Altekruse, D.L. Swerdlow, Infectious Disease Clinics of North America 1998, 12, 199. [9] M.E. Potter, S. Gonzalez-Ayala, N. Silarug, in Food Microbiology: Fundamentals and Frontiers (Eds: M.P. Doyle, L.R. Beuchat, T.J. Montville). ASM Press, Washington, DC. 1997, pp. 376±390. [10] J. Meng J., M.P.Doyle, Bulletin Institut Pasteur 1998, 96, 151. [11] P. Feng, J. Food Protection 1992, 55, 927. [12] K. Helrich, Of®cial Methods of Analysis of the Association of Of®cial Analytical Chemists, Vol. 2, 15th ed., AOAC, Arlington, VA 1990, pp. 425±497. [13] E. Kress-Rogers, Handbook of Biosensors and Electronic Noses: Medicine, Food, and the Environment, CRC Press, Boca Raton 1997. [14] N.S. Hobson, I.Tothill, A.P.F. Turner, Biosens. Bioelectron. 1996, 11, 455. [15] F. Basile, M.B. Beverly, K.J. Voorhees, Trends in Anal. Chem. 1998, 17, 95. [16] A.L. Ghindilis, P. Atanasov, M. Wilkins, E. Wlikins, Biosens. Bioelectron. 1998, 13, 113. Electroanalysis 2000, 12, No. 5
D. Ivnitski et al. [17] D. Ivnitski, J. Rishpon, Biosens. Bioelectron. 1996, 11, 409. [18] S.H. Jenkins, W.R. Heineman, H.B. Halsall, Anal. Biochemistry 1988, 168, 292. [19] T. Kalab, P. Skladal, Electroanalysis, 1994, 6, 1004. [20] J.R. Wilkins, J. Appl. Environ. Microbiol. 1978, 36, 683. [21] T. Matsunaga, I. Karube, S. Suzuki, Appl. Env. Microbiol. 1979, 37, 117. [22] R.L., Holland, B.H. Cooper, N.G.P. Hegelson, A.W. McCracken, J. Clin. Microbiol. 1980, 12, 180. [23] K. Takayama, T. Kurosaki, T. Ikeda, J. Electroanalyt. Chem. 1993, 356, 295. [24] H. Suzuki, E. Tamiya, I. Karube, Electroanalysis 1991, 3, 53. [25] J. Piehler, A. Brecht, K.E. Geckeler, G. Gauglitz, Biosens. Bioelectron. 1996, 11, 579. [26] C.L. Morgan, D.J. Newman, C.P. Price, Clinical Chemistry 1996, 42, 193. [27] H. Endo, K. Fujisaki, Y. Ohkubo, T. Hayashi, E. Watanabe, Fisheries Science 1996, 62, 235. [28] T. Ding, U. Bilitewski, R.D. Schmid, D.J. Korz, E.A. Sanders, J. Biotechnol. 1993, 27, 143. [29] G.D. Hitchens, D. Hodko, D.R. Miller, O.J. Murphy, T.D. Rogers, Russian J. Electrochem. 1993, 29, 1344. [30] P. Silley, S. Forsythe, J. Appl. Bacteriol. 1996, 80, 233. [31] K.R. Milner, A.P. Brown, D.W.E. Allsopp, W.B. Betts, Electron. Lett. 1998, 34, 66. [32] D.M. Gibson, P. Coombs, D.W. Pimbley, J. Assoc. Off. Anal. Chem. 1992, 75, 293. [33] B. Swaminathan, P. Feng, Annu. Rev. Microbiol. 1994, 48, 401. [34] E. Palmqvist, K.C. Berggren, M. Khayyami, B. Danielsson, P.O. Larsson, K. Mosbach, D. Kriz, Biosens. Bioelectron. 1994, 9, 551. [35] T. Dezenclos, M. Asconcabrera, D. Ascon, J.M. Lebeault, A. Pauss, Appl. Microbiol. Biotechnol. 1994, 42, 232. [36] M.S. DeSilva, Yu Zhang, P.J. Hesketh, G.J. Maclay, S.M. Gendel, J.R. Stetter, Biosens. Bioelectron. 1995, 10, 675. [37] J. Dupont, D. Menard, C. Herve, F. Chevalier, B. Beliaeff, B. Minier, J. Appl. Bacteriol. 1996, 80, 81. [38] R. Ehret, W. Baumann, M. Brischwein, A. Schwinde, K. Stegbauer, B. Wolf, Biosens. Bioelectron. 1997, 12, 29. [39] L. Deng, L.L. Bao, Z.Y. Yang, L.H. Nie, S.Z. Yao, J. Microbiol. Methods 1996, 26, 197. [40] L. Deng, H. Tan, Y. Xu, L.H. Nie, S.Z.Yao, Enzyme & Microbial Technol. 1997, 21, 258. [41] B. Mirhabibollahi, J.L. Brooks, R.G. Krool, Appl. Microbiol. Biotechnol. 1990, 34, 242. [42] N. Nakamura, A. Shigematsu, T. Matsunaga, Biosens. Bioelectron. 1991, 6, 575. [43] J.L. Brooks, B. Mirhabibollahi, R.G. Kroll, J. Appl. Bacteriol. 1992, 73, 189. [44] H.J. Kim, H.P. Bennetto, M.A. Halablab, Biotechnol. Tech. 1995, 9, 389. [45] J.D. Brewster, A.G. Gehring, R.S. Mazenko, L.J. Vanhouten, C.J. Crawford, Anal. Chem. 1996, 68, 4153. [46] J.D. Brewster, R.S. Mazenko, J. Immunological Meth. 1998, 211, 1. [47] A.G. Gehring, C.G. Crawford, R.S. Mazenko, L.J. Van Houten, J.D. Brewster, J. Immunological Meth. 1996, 195, 15. [48] M.J. Eddowes, Analytical Proc. 1989, 26, 152. [49] A. Sadana, D. Sii, Biosens. Bioelectron. 1992, 7, 559. [50] J.M.W. Ducey, M.E. Meyerhoff, Electroanalysis 1998, 10, 157. [51] J. Wang, P.V.A. Pamidi, C.L. Renschler, C. White, J. Electroanal. Chem. 1996, 404, 137. [52] I.Iliev, A. Kaisheva, F. Scheller, D. Pfeiffer, Electroanalysis 1995, 7, 542. [53] D. Huet, C. Gyss, C. Bourdillon, J. Immunol. Methods 1990, 135, 33. [54] M. Defrutos, S.K. Paliwal, F.E. Regnier, Meth. Enzymology, 1996, 270, 82. [55] I. Abdel-Hamid, A.L. Ghindilis, P. Atanasov, E. Wilkins, Anal. Lett. 1999, 32, 1081. [56] J. Rishpon, D. Ivnitski, Biosens. Bioelectron. 1997, 12, 195. [57] I. Abdel-Hamid, D. Ivnitski, P. Atanasov, E. Wilkins, Electroanalysis 1998, 10, 758.
Detection of Food Pathogenic Bacteria [58] G.M. Wyatt, Food & Agricultural Immunology 1995, 7, 55. [59] C.R. Clark, K.K. Hines, A.K. Mallia, Biotechnol. Tech. 1993, 7, 461. [60] S.M. Paffard, R.J. Miles, C.R. Clark, R.G. Price, J. Immunological Meth. 1996, 192, 133. [61] R. Puchades, A. Maquieira, Cr. Rev. Anal. Chem. 1996, 26, 195. [62] C.E. Stager, J.R. Davis, Clin. Microbiology Rev. 1992, 5, 302. [63] F.G. Perez, M. Mascini, I.E. Tothill, A.P.F. Turner, Anal. Chem. 1998, 70, 2380. [64] H. Yu, P.J. Stopa, in Environmental Immunochemical Methods: Perspectives and Applications (Eds: Van Emon, J.M., Gerlach, C.L. & Johnson, J.C.). ACS, Washington, DC, 1995, pp. 297±306. [65] H. Yu, J.G. Bruno, Appl. & Environmental Microbiology 1996, 62, 587. [66] C. Vernozy-Rozand, C. Mazuy, S. Ray-Gueniot, S. Boutrand-Loei, A. Mayrand, Y. Richard, Lett. Appl. Microbiol. 1997, 25, 442. [67] P. Bouvrette, J.H.T. Luong, Microbiology 1995, 27, 129. [68] I. Abdel-Hamid, D. Ivnitski, P. Atanasov, E. Wilkins, Proc. First NSF Int. Conf. on Food Safety, Albuquerque, NM, USA, November 16±18, 1998, 141. [69] I. Abdel-Hamid, D. Ivnitski, P. Atanasov, E. Wilkins, Biosens. Bioelectron., 1999, 14, 309. [70] Pall Gelman Sciences Corporation, Diagnostic Application Guide 1997, pp. 1±15. [71] S.R. Mikkelsen, Electroanalysis 1996, 8, 15. [72] J. Wang, G. Rivas, X. Cai, E. Palecek, P. Nielsen, H. Shiraishi, N. Dontha, D. Luo, C. Parrado, M. Chicharro, P. Farias, F.S. Valera, D.H. Grant, M. Ozsoz, M.N. Flair, Anal. Chim. Acta 1997, 347, 1. [73] J.H. Zhai, H. Cui, R.F. Yang, Biotechnol. Adv. 1997, 15, 43.
325 [74] M. Tietjen, D.Y.C. Fung, Critical Rev. Microb. 1995, 21, 53. [75] D.D. Jones, R. Law, A.K. Bej, J. Food Sci. 1993, 58, 1191. [76] R.K. Sailki, S. Scharf, F. Faloona, K.B. Mullis, G.T. Horn, H.A. Erlich, N. Arnheim, Science, 1985, 230, 1350. [77] De Lumley-Woodyear, C.N. Campbell, E. Freeman, A. Freeman, G. Georgiou, A. Heller, Anal. Chem. 1999, 71, 535. [78] J. Wang, G. Rivas, C. Parrado, C. Xiaohua, M. Flair, Talanta 1997, 44, 2003. [79] J. Wang, G. Rivas, X.H. Cai, Electroanalysis 1997, 9, 395. [80] J. Wang, Chem. Eur. J. 1999, 5, 1681. [81] S. Palanti, G. Marrazza, M. Mascini, Anal. Lett. 1996, 2309. [82] J. Wang, G. Rivas, J. Fernandes, J. Paz, M. Jiang, R. Waymire, Anal. Chim. Acta 1998, 375, 197. [83] A. Otero, M-L. Garcia-Lopez, B. Moreno, Meat Sci. 1998, 49, S179. [84] Fderal Register, Drinking water: National primary drinking water regulations; total coliform proposed rule. Federal Register 1991, 54, 27544. [85] A.E. Greenberg, R.R. Trussel, L.S. Clesceri, M.A.H. Franson, Standard Methods for the Examination of Water and Wastewater, American Pablic Health Association, Washington, DC 1992. [86] Y. Blixt, E. Borch, Int. J. Food Microbiology 1999, 46, 123. [87] C. Di Natale, A. Macagnano, F. Davide, A. D'Amico, R. Paolesse, T. Boschi, M. Faccio, G. Ferri, Sens. Actuators B 1997, 44, 521. [88] M. Holmberg, F. Gustafsson, E.G. Hornsten, F. Winquist, L.E. Nilsson, L. Ljung, I. Lundstrom, Biotechnology Tech. 1998, 12, 319. [89] E. Schaller, J.O. Bosset, F. Escher, Food Sci. Technol.±Lebensm.Wiss. Technol. 1998, 31, 305.
Electroanalysis 2000, 12, No. 5