Susan E. Maslanka, Gerald Zirnstein, Jeremy Sobel, and Bala Swaminathan

The consumption of contaminated food causes some 76 million illnesses, 325,000 hospitalizations, and 5,000 deaths in the United States each year. More than 200 known diseases are transmitted through foods; however, in only 18 percent of foodborne illnesses is an agent identified. Current laboratory methods used to identify etiologic agents are relatively slow, since they usually depend on the growth of an enrichment culture and subsequent characterization of an isolate by standard biochemical tests. While the combination of culture, isolation, and biochemical characterization of an isolate is the gold standard for identification of an etiologic agent, newer methods are needed for rapid detection so that prevention strategies can be implemented quickly to reduce the incidence of disease. Many immunoassay-based tests and/or systems show promise in their ability to correctly identify an agent quickly, particularly when the tests are used in conjunction with methods that concentrate the target agent. Some of these new technologies incorporate unique detection devices, such as biosensors, into immunoassay tests to increase the sensitivity of detection. Others, such as DNA chip technologies, have the capacity to screen rapidly and simultaneously for many infectious disease agents. Finally, a new technology may be on the horizon to identify a specific etiologic agent by assessing the change in expression of host immune response genes. This could make it possible to use a single universal testing protocol to determine the agent of disease.

Consumers increasingly demand high-quality, readily available, and safer food and water supplies. The United States has made a major commitment toward improving the nation's food supply through the interagency National Food Safety Initiative established in 1997. The establishment of this initiative resulted in a cooperative effort by federal and state partners to provide protection to consumers from contaminated food. Water will be included as a component of the initiative in 2001 by the addition of the Environmental Protection Agency to the federal partnership. The scope of these coordinated efforts needs to be broadened to include the development, validation, and deployment of sensitive, rapid methods for agent detection in investigations of intentional and unintentional contamination of our food and water supplies.

The spectrum of illnesses caused by consumption of contaminated foods may range from self-limiting mild gastroenteritis to life-threatening neurological, hepatic, and renal syndromes. Mead et al. (1999) have estimated the number of illnesses, hospitalizations and deaths in the United States using data from various national surveillance systems. Their estimates indicate that contaminated foods cause approximately 76 million illnesses, 325,000 hospitalizations, and 5,000 deaths in the United States each year. The economic burden is estimated to be $9 billion to $32 billion. More than 200 known diseases are transmitted through foods; the agents of foodborne illnesses include viruses, bacteria and their toxins, fungi and their toxins, parasites, poisonous plant components, marine biotoxins, heavy metals, and possibly prions. However, in 82 percent of foodborne illnesses, the identity of the pathogen is unknown. Of 1,500 deaths each year due to known pathogens, 75 percent are caused by Salmonella, Listeria monocytogenes, and Toxoplasma.

The epidemiology of foodborne diseases has undergone profound change in the past two decades. Some factors influencing this change are the global distribution of food supplies to meet increasing consumer demands for greater diversity of foods, centralization of food production, processing and distribution to improve efficiencies and reduce costs, demographic changes occurring in industrialized nations that have resulted in increases in the proportion of the population with heightened susceptibility to severe foodborne infections, changes in food-related behaviors by consumers, and dramatic increases in world travel (Kaferstein et al., 1997; Swerdlow and Altekruse, 1998). One negative effect of the high-

degree consolidation of food production, processing, and distribution is that food-safety-related failures may affect large numbers of people over large geographic areas and may have disastrous public health consequences. Because of the explosive increases in international travel, new and emerging pathogens from one corner of the world are able to arrive at a location thousands of miles away in a matter of hours. Transcontinental flights themselves offer many opportunities for transmission of foodborne disease (Tauxe et al., 1987). In addition, the manufacturers and/or distributors of a contaminated food are likely to encounter dire financial and public relations consequences following the implication of their products as a source of widespread illness.

Examples of large-scale (several thousands of cases) foodborne outbreaks are listed in Table 13.1. The 1985 outbreak of Salmonella ser. Typhimurium infections was most likely caused by improper switching of the stainless steel pipes in the milk-processing facility, which resulted in raw milk coming in contact with pasteurized milk (Ryan et al., 1987). Interestingly, the outbreak was first recognized as a potentially large one when clinical laboratories in the region ran out of laboratory supplies for culturing Salmonella from ill persons. The ice-cream-associated outbreak of Salmonella enteritidis infections in 1994 was caused by improper cleaning and sanitation of the ice cream premix tanker that was used previously to transport raw liquid eggs (Hennessy et al., 1996). The Japanese outbreak of Escherichia coli O157:H7 infections was most likely caused by contamination of seeds used for sprouting or contamination of water used in the sprouting process (Michino et al., 1999).

Some examples of large waterborne disease outbreaks are listed inTable 13.2. In the largest reported outbreak of typhoid fever in India in

1975 to 1976, 6.7 percent of the towns population was affected. The problem was caused by overflow of sewage into wells that were the sources of drinking water for the town. A large outbreak of hepatitis E infections occurred in Kanpur, India, in 1992, which was probably caused by fecal contamination of river water and inadequate chlorination. Waterborne outbreaks are not limited to areas with inadequate public health systems. In 1993 the Milwaukee outbreak of cryptosporidiosis rapidly depleted the available supply of over-the-counter antidiarrheal medications in that area and overwhelmed the health care system (Colley, 1995). In June 1978, 19 percent of the population of Bennington, Vermont, was affected by diarrheal illness caused by Campylobacter jejuni contamination of the town's drinking water source (Vogt et al., 1982). Despite the recent enhancement of conventional and laboratory-based surveillance for foodborne bacterial diseases in the United States, a 1999 waterborne outbreak of gastro-intestinal illness affected more than 900 persons who attended a county fair in New York state; 122 E. coli O157:H7 infections and 51 Campylobacter spp. infections were culture confirmed. Eleven persons developed hemolytic uremic syndrome and two died (Centers for Disease Control and Prevention, 1999).

Large outbreaks like those described above require enormous investments of federal, state, and local resources to investigate and control. These resources are needed for epidemiologic identification of the source of the disease and subsequent laboratory isolation, identification, and characterization of the etiologic agents. The National Food Safety Initiative (NFSI) provides a funding mechanism to improve foodborne disease surveillance and outbreak investigations (U.S. Department of Health and Human Services, Department of Agriculture, and Environmental Protection Agency, 1997). Key components of NFSI include enhanced foodborne disease surveillance, improved responses to foodborne disease outbreaks, improved risk assessment, development of new research methods, improved food safety inspection systems, and improved food safety education.

Intentional contamination of our food and water supplies is a real threat. There have been two well-documented instances of intentional contamination of foods with pathogenic microorganisms. In 1984, members of a religious commune in Oregon attempted to influence the outcome of a local election by intentionally contaminating salad bars in several restaurants with Salmonella ser. Typhimurium. The outbreak affected at least 750 persons, and S. Typhimurium was cultured from stool specimens of 388 persons (T örök et al., 1997). In 1996, 12 of 45 laboratory workers at a large medical center in Texas became infected with Shigella dysenteriae type 2; the outbreak was associated with eating pastries or doughnuts that had been placed in the staff break room on a specific day. Epidemiologic and laboratory investigations strongly suggested intentional contamination of pastries by someone who had access to the bacterial stock cultures in the medical center's laboratory and who was familiar with the methods of culturing the bacteria (Kolavic et al., 1997).

Unlike some potential threat agents (i.e., smallpox) for which the sources are limited, many foodborne agents such as Salmonella, E. coli O157, and even botulinum toxin, are relatively easy to obtain or produce. Many of the agents are stable under a variety of conditions and so could easily be added to food and water supplies before consumption. Although there was no reason to suspect foul play in any of the three foodborne outbreaks listed in Table 13.1, each could have easily been caused by intentional contamination by one or more persons involved in some way in food processing, preparation, or transport. As a vehicle for intentional dissemination of pathogens or toxins, water is particularly worrisome because it is an extremely efficient vehicle that could incapacitate large

numbers of people within a very narrow window of time. Although none of the outbreaks listed in Table 13.2 were due to intentional contamination of the water supply, the data in this table illustrate the magnitude of problems that could be created by intentional acts of bioterrorism in which water supply is used as the vehicle for contamination.

Some foodborne disease agents require only a small inoculum to cause disease. Shigellosis can be caused by just a few hundred organisms; the infective dose of E. coli O157:H7 is thought to be even less (Hornick, 1998). Botulinum toxin is one of the most potent toxins known; it has been estimated that 1 gram of botulinum toxin is enough to kill 1.5 million people (McNally et al., 1994). Introduction of botulinum toxin into a food source would severely strain the resources of the health care system (e.g., antitoxin, hospital support, mechanical ventilators). Although perhaps less deadly, other pathogens intentionally introduced into food and/or water supplies could also negatively affect the ability of a community to respond to the disease. Widespread disease could easily overburden the health care system (hospitals, doctors, medical supplies), the public health system (epidemiologists, diagnostic testing laboratories), and emergency response teams (police, paramedics, decontamination crews). In addition, lack of consumer confidence in the quality of the food and water supplies would be an additional burden on community governments. Capacity for early detection of intentional contamination of the nation's food and water supplies is vital to minimize the impact on community health.

There is a great deal of overlap between laboratory support for surveillance for foodborne (or waterborne) diseases, outbreak investigations, and bioterrorism preparedness. The methods used for all these public health response activities must be sensitive and specific in order to accurately identify cases. Ideally, the diagnostic method used for any of these activities also would be rapid, easy to use, and cost effective. In reality, some of these additional requirements are sacrificed to support the most important purpose of a particular activity while maintaining accuracy in case identification. In general, the requirements of a method are dictated either by a need to respond rapidly to an event or to perform extensive characterization of isolates. Table 13.3 lists some of the additional attributes that must be considered, other than sensitivity and specificity, which are critical for all activities, when evaluating a laboratory method for a specific purpose.

Laboratory-based surveillance activities aim to capture all confirmed cases of a given disease. As centralized production and wide distribution

of food become more common, foodborne disease outbreaks increasingly occur over dispersed areas. Identifying increases in specific subtypes of pathogens through ongoing comparison of surveillance data with historical baselines may be the only way to detect outbreaks among large numbers of persons that lack a clear geographic focus. A good example of this is the ice-cream-associated Salmonella outbreak of 1994 (Hennessy et al., 1996). Because surveillance, by its nature, entails capturing the universe of all cases in a population, it requires the processing of high volumes of samples. Surveillance data continue to be used to determine long-term trends in pathogen incidence. However, incidence is documented by sporadic case counts as well as through acute outbreaks (Centers for Disease Control and Prevention, 2000; Glynn et al., 1998).

Laboratory-based outbreak response activities aim to rapidly identify the disease agent. The chosen method should be capable of identifying the suspected pathogen rapidly so that investigations can occur, food vehicles can be identified, and intervention strategies can be implemented to limit the number of cases of disease. Next in importance are ease of use and cost. The tests are usually performed by highly skilled laboratory workers and are more likely to be conducted in state and federal laboratories. Unlike surveillance, identification of the pathogen in a foodborne outbreak may not require testing large numbers of specimens. The majority of unintentional foodborne outbreaks are due to a single pathogen, and confirmation of a single pathogen in 10 to 20 epidemiologically linked patients with a compatible clinical syndrome is usually sufficient. Subtyping of pathogens may be important for epidemiologic investigations of outbreaks, especially when relatively high background rates of infection occur. Identification of a subtype associated with the outbreak

allows exclusion of nonoutbreak cases from analytical studies, thereby increasing the power of such studies to implicate a food vehicle.

Laboratory diagnostic response to bioterrorism, as in unintentional outbreak response, also aims to rapidly identify the disease agent. The constraints on appropriate methods for response to a bioterrorism event are great. The method(s) must be rapid to help minimize the potentially large number of cases, and it must be easy to use in order to be effective. The most appropriate method potentially could be used by minimally trained first responders to the event. The need to have a large number of tests available at local sites throughout the country mandates that the method be highly cost effective. To increase the burden of illness or to confound investigators, a bioterrorist might attempt to contaminate food(s) with more than one pathogen. This means that additional tests with different methods may be required even after identifying one agent. Nevertheless, it is appropriate to test a limited number of samples, as in outbreak response to unintentional contamination, with the option of testing a representative sample of cases occurring subsequent to identification of an initial pathogen to rule out “second waves” of contamination with other pathogens. Similar to outbreak investigations is the need for discriminatory methods that could provide data to trace the event back to the source and would provide law enforcement personnel with information to link the event to a perpetrator(s).

The Centers for Disease Control and Prevention, in collaboration with the Association of Public Health Laboratories, the Department of Defense, and other federal partners, has established a Laboratory Response Network for Biological Terrorism. The U.S. Congress allocated funds to establish this network in 1998, which links response to bioterrorism events by various laboratories (local, state, federal). The intent of the network is to provide standardized protocols to appropriate testing laboratories that are as close as possible to the geographic area of the release of the biological agent. Laboratories are given designations (Level A, B, C, or D) based on their capacity to handle a specific agent and perform specific diagnostic tests. Some agents and/or protocols that can be handled without special safety equipment, such as a biological safety cabinet, can be processed and identified at the lowest level (Level A) in a hospital or clinical microbiology laboratory. Other agents or protocols, because of their particular hazardous status, must be handled at a federal laboratory level (Level D). With such a network, laboratory response time to an announced or unannounced bioterrorism event should be reduced because emergency response personnel know the appropriate mechanisms for submitting specimens for testing.

There are a number of different immunologic technologies that fit the requirements of clinical laboratories. Although some are more laborious than others, each has a definite role in analysis of pathogens and toxins likely to be encountered in outbreak investigations and in response to bioterrorist events. Many of these technologies detect only a few pathogens or toxins and need to be adapted for the detection of multiple foodborne/waterborne agents.

The majority of immunological test formats for rapid automated analysis of clinical and food samples are based on the noncompetitive “sandwich ” enzyme linked immunosorbent assay (ELISA). This method is generally favorable for most clinical and food samples because materials present in the sample, including proteases and noncompetitive enzyme inhibitors that could substantially alter enzyme activity, are removed in the washing step prior to detection (Swaminathan et al., 1985; Clark and Engvall, 1985). Both the Alert system (Neogen, Lansing, Mich.) and the RIDASCREEN (R-Biopharm GmbH, Darmstadt, Germany) are sandwich ELISAs that use antibody-coated microtiter wells but different enzyme/substrate detection systems. Both can detect the presence of Staphylococcus aureus enterotoxins. However, the RIDASCREEN microtiter strips can differentiate between toxins A, B, C, D, and E.

The Eia Foss system (Foss North America, Eden Prairie, Minn.) also utilizes sandwich ELISA technology but with fluorescence detection. However, the Eia Foss system is unique in that it combines automated ELISA and immunomagnetic separation (IMS) methods in one system. The Eia Foss system can accommodate up to 27 samples per run and can complete the tests in less than 2 hours. Immunomagnetic separation improves the sensitivity and specificity of the test.

Two examples of fully automated microplate and sample handling systems are the Transia Elisamatic II (Diffchamb AB, Gothenburg, Sweden) and the TECRA MINILYSER processor (TECRA International, Willoughby, NSW, Australia). With these systems, results are available within approximately 3 hours, including the sample preparation and the ELISA analysis. Both can detect staphylococcal enterotoxins (SET). TECRA also produces a TECRA SET ID VIA kit for identification of individual toxins in samples found to be positive with the TECRA SET VIA kit, providing the same valuable advantages as the RIDASCREEN product with toxin-specific identification capability but with the option of fully automating the assay.

The VIDAS (Vitek Immuno Diagnostic Assay System) and the mini VIDAS (bioMerieux, Hazelwood, Mo.) also are automated systems that have some unique capabilities not found in other products. A reagent strip with plastic wells containing all of the individual reagents and washes needed for a particular immunoassay is punctured during the course of testing by a solid-phase receptacle. The receptacle is coated on its inside surface with antibody directed toward the target antigen in the sample. Additionally, the receptacle serves as a pipette tip to transfer sample and reagents during the assay after puncturing the foil covering protecting the contents of the individual wells. The VIDAS module, which can run 30 tests at a time, is controlled by a computer module that executes all of the robotic procedures. As many as four VIDAS modules can be operated by a single computer module, for a total capacity of 120 simultaneous tests. A unique advantage of the VIDAS module over other forms of automated immunological testing instruments is that the 30-sample capacity of the module is divided among five chambers with six slots for reagent strips in each chamber. Each chamber can be programmed to run assays with incubation temperatures, hold times, and reagents completely different from the conditions used to perform assays in an adjacent chamber. This gives the VIDAS system a significant degree of flexibility when testing simultaneously for multiple analytes.

Commercial immunological assays for mycotoxin detection generally rely on rapid detection methods because of the need to test agricultural commodities for toxins and quickly release the product for further processing or shipment. Agri-Screen, Veratox, and Veratox HS (Neogen, Lansing, Mich.) are rapid tests for mycotoxins that use a competitive direct ELISA. The assay is based on competition between free toxin in the sample and the control toxin (enzyme-labeled conjugate) for the antibody binding sites. Testing can be completed in 5 to 10 minutes using the AgriScreen and Veratox products, while the Veratox HS product, which has a higher level of sensitivity (< 2 ppb), takes 20 minutes to perform. AgriScreen is sold as a field kit and is a qualitative test that allows a sample to be scored as positive or negative for the analyte by visual comparison with known controls; therefore, it does not require the use of laboratory equipment. Veratox and Veratox HS require the use of microwell readers. Several companies have developed immunoassay cards for pathogens and toxins that are similar in design to pregnancy test kits. The VIP test (BioControl Systems, Bellevue, Wash.), the Transia Card (Diffchamb AB, Gothenburg, Sweden), the REVEAL test (Neogen, Lansing, Mich.), and the Listeria Rapid Test and the Clostridium difficile Toxin A test (both by

Oxoid, Basingstoke, Hampshire, England) are immunoassay cards based on the sandwich ELISA method. All of the immunoassay cards rely on the formation of an antigen-antibody complex, as the test sample (usually 100 to 135 µL) is pulled by capillary action through the test card. The antigen-antibody complex becomes trapped on an immobilized line of antibody. The line becomes visible because of the presence of a dye, chromogen, or colloidal gold bound to the antigen-antibody complex. Nothing is required of the operator except the preparation and loading of the sample to be tested. Results can be read within 5 to 20 minutes, depending on the construction method for the card.

The rapid immunological tests are easy to perform, relatively inexpensive, reasonably fast, and able to be performed immediately upon receipt of a sample rather than after all the samples are run in a large batch. Therefore, they are particularly well suited for field use during outbreak investigations or in first tier response laboratories in the bioterrorism response network.

A biosensor can be thought of as a biorecognition assay or “sensor” in close proximity to or linked to a signal receptor system or “transducer” (Goldschmidt, 1999). When the sensor reacts with its target, the transducer records the changes in intensity that are related to the concentration and/or activity that occurs. Sensors can be made to detect microorganisms, tissue culture cells, and a variety of biomolecules such as enzymes, antibodies, antigens, DNA, and RNA. The technology behind the operation of the transducer may be based on electrochemical/electric, optical, thermal, or other properties, and most instruments can be automated by interfacing them with a computer and recording device.

One of the newer immunology-based detection systems is the BIACORE system (Biacore International AB, Uppsala, Sweden). This is a microchip-based system that permits the detection of biomolecules and the association/dissociation kinetics between macromolecular complexes. The detection principle of the BIACORE microchip is based on surface plasmon resonance, which allows the measurement of changes in the refractive index at the sensor chip surface as they occur. With one reactant immobilized on the sensor chip, a second reactant can be injected over the chip 's surface. Changes in mass at the chip surface due to the interaction of reactants alter the refractive index and affect the resonance angle. A variety of assays are possible with this technology, including a direct assay (an antibody bound to the sensor chip surface captures an antigen), sandwich assays similar to ELISA methods, and inhibition assays (sensor-bound antibody competes with antibody added to the sample; Robinson,

1997). In addition, a method for thermodynamic analysis of biomolecular interactions using this technology has been described (Roos et al., 1998). Among other applications, the BIACORE system has been used to examine the interactions between antibodies and antigens (Malmqvist, 1993) and to detect Salmonella (Haines and Patel, 1995). A major pharmaceutical company is adopting biosensor assays using the BIACORE 2000 to support preclinical and clinical trials for the detection of antibodies. One noted advantage of the biosensor assay is increased throughput, since detection is label free and occurs in real time. Also, direct detection of binding interactions may enable assessment of antibody avidity (Swanson et al., 1999).

Another fully automated biosensor system based on optical detection is the IAsys Auto+ Advantage (Affinity Sensors, Cambridge, United Kingdom). The IAsys system utilizes resonant mirror (also referred to as evanescent wave) technology, which can detect changes in refractive index (or mass) due to binding of molecules at the surface of a biosensor cuvette. Changes in mass (binding) result in changes in the resonant angle, which are linear with respect to mass. This system uses an open cuvette design, which is intended to minimize sample contamination and aid sample recovery. This technology has been used successfully to detect bacterial toxins in food samples by utilizing a sandwich biosensor with two antibodies. Little or no background interference was noted, even when detecting toxin in complex food matrices (Rasooly and Rasooly, 1999).

The use of biosensors for analysis of clinical and food samples for surveillance, outbreak investigations, and response to bioterrorism events is most certainly in its infancy. In the future, the use of low-cost biosensors combined with efficient telemetry should enable rapid monitoring of environmental contamination in theaters of war and during political and social events with increased risks of terrorist attack.

With the invention of microarray (DNA chip) technology, researchers who used to spend many hours performing tedious manual experiments to study one gene at a time (Zirnstein et al., 1999; Mustapha et al., 1995; Heruth et al., 1993) will be able to examine thousands of genes at a time. As the manufacture and use of DNA chips becomes more common, the price of this type of molecular genetic analysis will become more accessible for routine use by public health laboratories. Nucleotide base pairing (i.e., A-T and G-C for DNA, A-U and G-C for RNA), or hybridization of DNA and RNA molecules, is the scientific principle by which micro-arrays function. Microarrays generally have nucleic acid probe spot sizes of 200 microns or less in diameter, and these arrays generally have thousands of spots. The “probe” is a tethered or fixed nucleic acid of known

sequence, while the “target” is the free nucleic acid in a sample. The primary uses for DNA microarray technology are to identify gene sequences and mutations in gene sequences and to determine the expression level or abundance of genes.

Two general formats of DNA microarray technology are in use. The first format uses probe cDNA (500 to 5,000 bases in length) immobilized on a solid surface, usually glass, using robotic spotting. The array is then hybridized to a set of target nucleic acids, either one at a time or as a mixture of targets. This method is generally referred to as “DNA microarray” (Ekins and Chu, 1999). The second format is an array of oligonucleotide (20- to 25-mer oligos) or peptide nucleic acid probes synthesized in situ on the chip or by conventional synthesis of the oligonucleotide probes followed by immobilization of the probes onto the chip. The array is hybridized to labeled sample DNA, and the identity and abundance of complementary sequences are determined. This second format for microarrays is generally termed DNA chips or GeneChip arrays since they were first developed at Affymetrix, Inc. (Santa Clara, Calif.), a company that sells an integrated GeneChip Instrument System that automates the loading, processing, and analysis of premanufactured GeneChip microarray cartridges. The GeneChip Fluidics station automatically loads the nucleic acid target sample onto the probe microarray cartridge. The fluidics station also controls the delivery of reagents and the timing and temperature for hybridization of the nucleic acid target to the probe array. As many as four probe arrays can be processed independently at one time. Once the hybridization step is complete, messages are displayed on the screen of the controlling computer station indicating that the probe array is ready for scanning. Probe arrays are scanned with a Hewlett-Packard Gene Array Scanner (Hewlett-Packard, Palo Alto, Calif.), which uses an argonion laser to excite fluorescent molecules incorporated into the nucleic acid target to generate a quantitative hybridization signal. The fluorescent data from thousands of 3-micron spots within the probe cells of the GeneChip result in a high-resolution image of the probe array and are stored in a file. Data are analyzed using GeneChip Analysis Suite software. GeneChip technology has been useful for species identification in the genus Mycobacterium (determined from 16S rRNA sequences) and for detecting rifampin resistance (rpoB alleles) in Mycobacterium tuberculosis (Troesch et al., 1999), and it should be extremely useful for the detection of bacteria in general.

At this time the cost of premanufactured Affymetrix DNA chips is prohibitively high for many applications (approximately $2,000 per chip). However, a GeneMachines microarray printer (GeneMachines, San Carlos, Calif.) is available for those who wish to produce their own DNA microarray slides. Estimated cost is reduced to $30 to $50 per slide.

GeneMachines currently produces a high-performance microarrayer capable of arraying biological samples from standard 96- or 384-well microwell plates onto a variety of substrates, including glass slides and nylon membranes. Separate hybridization chambers can be purchased to process the arrays. A GenePix 4000A microarray scanner and controlling software (Axon Instruments, Inc., Foster City, Calif.) can be used to collect the fluorescent data generated by the hybridized microarray slides.

The technologies mentioned above rely on passive hybridization of target molecules in solution with the immobilized probe DNAs. Another approach used by Nanogen (San Diego, Calif.) uses electronically mediated active hybridization to move and concentrate target DNA molecules, which accelerates hybridization. Hybridization occurs in minutes rather than the hours required for passive hybridization techniques. This method relies on the fact that negatively charged DNA molecules can be attracted to a positively charged electrode, and if the electrode has DNA probes attached to it that are complementary to the target molecules, the target molecules are “captured.” Electronics, in a microchip format, move and concentrate target molecules to one or more test sites on the chip. To remove unbound or nonspecifically bound DNA from each test site, the polarity or charge of the site is reversed to negative, forcing non-specifically bound and unbound molecules back into solution and away from the capture probes. A dye-labeled reporter binds to a specific target-DNA sequence at the test site, thus allowing the target DNA in the sample to be quantified. This technology can be applied to many types of analyses, including antigen-antibody, enzyme-substrate, cell-receptor, and cell separation techniques. Another advantage to this technique is that the concentration of target molecules over the test site enables a lower concentration of target DNA molecules to be used, resulting in reduced time and labor required for pretest sample preparation.

Flow cytometry was developed in the 1970s for eukaryotic cell analysis. Flow cytometers operate by the movement of particles (cells, beads, nuclei) in single file in an aqueous stream. The cells flow through a focused, high-intensity light beam, which is usually produced by a laser (Raybourne, 1997). Flow cytometry has been used to detect Listeria monocytogenes in milk (Donnelly and Baigent, 1986), to distinguish bacterial species with preferential A-T or C-G DNA-binding dyes (Van Dilla et al., 1983), to identify specific species of bacteria using fluorescent in situ hybridization probes for 16S rRNA (Amann et al., 1990), and to differentiate live versus dead cells (Raybourne, 1997). The latter method has been used for antibiotic susceptibility testing and may be applicable to food

microbiology or determination of the viability of organisms used in bioterroristic events.

One of the problems confronted when trying to analyze food and environmental samples is the need to detect very small numbers of pathogen in a large volume of sample or in a sample that contains large amounts of interfering contaminants. Currently used official methods for pathogen detection in foods, such as contained in the Bacteriological Analytical Manual (1996), and similar references, have a minimum detectable limit of one cell per 25-g sample. Improvements in pathogen detection should include approaches that decrease the time required for preenrichment, enrichment, and postenrichment to recover pathogens prior to the application of one of the assays described above. For example, methods currently available for Salmonella require a day or two of processing before a rapid diagnostic test can be performed.

Using magnetic beads coated with antibodies is one method that shows potential for reducing the time required for recovery (Holt et al., 1995; Fierens and Huyghebaert, 1996; Mitchell et al., 1994). Antibodies for a specific pathogen are attached to the surface of the superparamagnetic microspheres and effectively capture any pathogen present in the sample. Pathogen-specific magnetic beads have been developed by at least two commercial companies (Dynal AS, Oslo, Norway, and Vicam, Somerville, Mass.). The pathogens captured on the beads can then be cultured or used for further rapid testing by methods such as ELISA or DNA/RNA probes.

A relatively novel approach for the concentration of whole bacterial cells from ground beef, bovine carcass samples, and bovine feces involves the use of hydroxyapatite for the adherence and recovery of viable pathogens (Berry and Siragusa, 1997). Bacterial adhesion to hydroxyapatite is mediated by nonspecific van der Waals and electrostatic attractions (Cowan et al., 1987; Nesbitt et al., 1982). Bacteria, which possess a net negative charge, are adsorbed to hydroxyapatite in a manner similar to the attraction of a negatively charged protein to the positively charged calcium ions of hydroxyapatite (Berry and Siragusa, 1997; Bernardi et al., 1972; Rölla and Melsen, 1975). Kinetic studies revealed that maximum adherence of bacteria to hydroxyapatite takes place within 5 minutes and that both high (109) and low (103) concentrations of cells yielded comparable percent adherence values within this time.

The use of biotinylated oligonucleotide probes to permit the specific capture of pathogenic DNA in clinical samples containing large amounts of extraneous DNA and inhibitors has also been demonstrated

(Mangiapan et al., 1996). Nonspecific inhibition can be minimized by mechanical disruption and proteinase K digestion. The pathogen-specific DNA that hybridizes with the probes is captured with strepavidin-coated magnetic beads and subsequently amplified using PCR. This approach is reported to be 10 to 100 times more sensitive than procedures in which total DNA is extracted and purified before PCR amplification (Mangiapan et al., 1996). Although sensitivity may be sacrificed, biotinylated oligo-nucleotides can be directly coupled to the magnetic beads (a direct capture technique), which subsequently decreases the time for identification of the pathogen (Muir et al., 1993).

Current methods for the detection of etiologic agents of foodborne diseases are pathogen/toxin specific. For example, methods for the examination of patient and implicated food specimens in cases of suspected botulism are very different from those in which E. coli O157:H7, Salmonella sp., or Listeria monocytogenes is the suspected etiologic agent. There are significant differences even in the type of specimen(s) collected as well as in the methods for specimen collection and storage protocols for food-borne bacterial and viral pathogens. The variability in agent-specific protocols is likely to significantly delay laboratory identification of the etiologic agent in the event of an unannounced bioterrorism event in which the clinical symptoms alone do not allow investigators to narrow the list of suspected agents to a small manageable number. In the past, rapid screening of patient specimens for immunological response to a broad spectrum of infectious agents has greatly facilitated the identification of a newly emerging etiologic agent; however, because specimens must be processed and screened independently for each potential agent, this process may be labor intensive and time consuming.

This is best illustrated by the approach used in the identification of hantavirus as the etiologic agent of a new respiratory disease syndrome in the southwestern United States (Chapman and Khabbaz, 1994). A variety of clinical specimens were collected during the investigation of this outbreak. These specimens were triaged through several laboratories with expertise in the various potential agents. As one agent was ruled out, specimens were transferred to another laboratory for further testing, until finally the correct identification was obtained. This traditional stepwise protocol used during an outbreak can significantly increase the time to identification of an agent of disease and as a result may hinder the implementation of public health measures to reduce the incidence of disease.

The technologies described earlier may provide some improvement in the rapid diagnosis of foodborne disease; however, assumptions still must be made about the potential agent involved before tests can be performed. It may be possible in the future to characterize and quantify changes in the expression of host genes induced by various etiologic agents using the new microarray technology rather than to directly identify the agent (Manger and Relman, 2000). This technology has the potential to use a universal platform for screening large numbers of clinical specimens for a wide range of etiologic agents within hours or possibly minutes. The advantages would be a uniform protocol for specimen collection, a uniform method for specimen processing and testing, early identification of a suspected agent, and a narrow field of possible agents so that the most effort can be directed toward appropriate specimens and methods for isolation and characterization of a particular agent. This method for agent detection is in its infancy and requires further evaluation to determine the utility in disease diagnostics.

There has been a recent surge in the development of new technologies for the detection of foodborne pathogens and their toxins. This has been in direct response to the change in the epidemiology of foodborne disease. As public health awareness of new and newly emerging foodborne disease increases, research and development of new methods for agent detection also increase. Although foodborne agent detection has improved dramatically in the past few years, many of the most promising techniques, such as DNA chip technology, are currently too costly to implement effectively on a large scale. However, consumers increasingly demand higher quality and more readily available foods of all types. In addition, they expect protection from disease while consuming these products. Consumers' expectations are fast outgrowing the ability of the public health community to protect them adequately from disease.

As food processing becomes more complex and the diversity of available foods increases, the need for more rapid, sensitive, and sophisticated diagnostic techniques will continue to rise. With a dramatic increase in centralized production and global distribution of foods, outbreaks may affect people in many states or even countries without the geographic clustering that formerly allowed astute clinicians, laboratorians, and public health officials to detect clusters of diseases. The integration of rapid diagnostic tests into laboratory-based public health surveillance is essential to meet this global challenge. In some widespread diffuse outbreaks, cases may be linked and the outbreak recognized only after comparison of agent subtypes with surveillance data available in molecular

databases. In addition, the waterborne outbreaks in the United States demonstrate that not only are these techniques needed to help protect consumers of food but that efforts also need to be directed toward the nation's water supplies.

The specter of bioterrorism aimed at food and water supplies, with the ominous possibility of deliberate contamination of foods and water with highly virulent pathogens, further underscores the need for integration of rapid diagnostic capacity into routine surveillance activities, in addition to the parallel rapid-diagnosis infrastructure described in this paper. This is especially important because a bioterrorist attack on the food or water supply may not be readily distinguished from an unintentional event and might well be first detected through routine laboratory surveillance. Further collaboration between government, university, and commercial laboratories is needed to make these promising techniques available for global use. Only then can we meet the increasing demand for even safer food and water supplies.

The United States has made a major commitment toward improving the safety of the nation's food supply by developing and funding the NFSI. This funding was provided to improve foodborne disease surveillance and outbreak investigations, enhance the regulatory control of foods through more frequent inspections, and encourage a prevention-oriented approach to food safety by implementing Hazard Analysis and Critical Control Points (HACCP) programs. The NFSI also requires various federal agencies with responsibility for food safety to coordinate their efforts. This interagency investment in food safety is now beginning to produce results; the addition of efforts specific to water supplies also should provide improvement in consumer protection. Much needs to be done to improve the nation's diagnostic capabilities to achieve a state of readiness to appropriately address bioterroristic events involving either food or water. Our ability to keep pace with rapid technological developments and marketing practices in the food production sector, as well as protect the public from the threats of emerging pathogens and intentional contamination of the food and water supplies, cannot be met without intensive integration of improved rapid diagnostics; enhanced and continually supported public health infrastructure (federal, state, and local agencies), including surveillance and investigative entities; and improved communication components.

Amann, R. I., B. J. Binder, R. J. Olson, S. W. Chisholm, S. W. Devereux, and D. A. Stahl. 1990. Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Applied and Environmental Microbiology, 56:1919-1925.

Bernardi, G., M. G. Giro, and C. Gaillard. 1972. Biochimica Biophysica Acta, 278:409-420.

Berry, E., and G. R. Siragusa. 1997. Chromatography of polypeptides and proteins on hydroxyapatite columns: Some new developments. Applied and Environmental Microbiology, 63:4069-4074.

Centers for Disease Control and Prevention. 1999. Morbidity and Mortality Weekly Report, 48:803.

Centers for Disease Control and Prevention. 2000. Morbidity and Mortality Weekly Report, 49:73-79.

Chapman, L. E., and R. F. Khabbaz. 1994. Etiology and epidemiology of the Four Corners hantavirus outbreak Infectious Agents Disease, 3:234-244.

Clark, B. R., and E. Engvall. 1980. Enzyme Linked Immunosorbent Assay (ELISA): Theoretical and Practical Aspects. Boca Raton, Fla.: CRC Press.

Colley, D. G. 1995. Waterborne cryptosporidiosis threat addressed. Emerging Infectious Diseases, 1:67-68.

Cowan, M. M., K. G. Taylor, and R. J. Doyle. 1987. Energetics of the initial phase of adhesion of Streptococcus sanguis to hydroxylapatite. Journal of Bacteriology, 169:2995-3000.

Donnelly, C. W., and G. J. Baigent. 1986. Method for flow cytometric detection of Listeria monocytogenes in milk. Applied and Environmental Microbiology, 52:689-695.

Ekins, R., and F. W. Chu. 1999. Microarrays: Their origins and applications. Trends in Biotechnology, 17:217-218.

Fierens, H., and A. Huyghebaert. 1996. Screening of Salmonella in naturally contaminated feeds with rapid methods. International Journal of Food Microbiology, 31:301-309.

Food and Drug Administration. 1996. Bacteriological Analytical Manual. Food and Drug Administration.

Glynn, M. K., C. Bopp, W. Dewitt, P. Dabney, M. Mokhtar, and F. J. Angulo. 1998. Emergence of multidrug-resistant Salmonella enterica serotype typhimurium DT104 infections in the United States. New England Journal of Medicine, 338:1333-1338.

Goldschmidt, M. 1999. Pp. 286-289 in The 19th International Workshop on Rapid Methods and Automation in Microbiology, D. Y. C. Fung, ed. Manhattan: Kansas State University.

Haines, J., and P. D. Patel. 1995. Detection of foodborne pathogens using BIA. BIA Journal, 2:31.

Hennessy, T. W., C. W. Hedberg, L. Slutsker, K. E. White, J. M. Besser-Wiek, M. E. Moen, J. Feldman, W. W. Coleman, L. M. Edmonson, K. L. MacDonald, and M. T. Osterholm. 1996. A national outbreak of Salmonella enteritidis infections from ice cream: The investigation team. New England Journal of Medicine, 334:1281-1286.

Heruth, D. P., G. W. Zirnstein, J. F. Bradley, and P. G. Rothberg. 1993. Sodium butyrate causes an increase in the block to transcriptional elongation in the c-myc gene in SW837 rectal carcinoma cells. Journal of Biological Chemistry, 268:20466-20472.

Holt, P. S., R. K. Gast, and C. R. Greene. 1995. Rapid detection of Salmonella enteritidis in pooled liquid egg samples using a magnetic bead-ELISA system. Journal of Food Protection, 58:967-972.

Hornick, R. B. 1998. Pp. 736-741in Infectious Diseases, P. D. Hoeprich, M. C. Jordan, and R. C. Ronald, eds. Philadelphia: J. B. Lippincott.

Kaferstein, F. K., Y. Motarjemi, and D. W. Bettcher. 1997. Foodborne disease control: A transnational challenge. Emerging Infectious Diseases, 3:503-510.

Kolavic, S. A., A. Kimura, S. L. Simons, L. Slutsker, S. Barth, and C. E. Haley. 1997. An outbreak of Shigella dysenteriae type 2 among laboratory workers due to intentional food contamination. Journal of the American Medical Association, 278:396-398.

Malmqvist, M. 1993. Biospecific interaction analysis using biosensor technology. Nature, 361:186-187.

Manger, I. D., and D. A. Relman. 2000. How the host “sees” pathogens: Global gene expression responses to infection. Current Opinion in Immunology, 12:215-218.

Mangiapan, G., M. Vokurka, L. Schouls, J. Cadranel, D. Lecossier, J. van Embden, and A. J. Hance. 1996. Sequence capture-PCR improves detection of mycobacterial DNA in clinical specimens. Journal of Clinical Microbiology, 34:1209-1215.

McNally, R. E., M. B. Morrison, J. E. Berndt, J. E. Fisher, J. T. Bo'Berry, V. Puckett, and N. J. Simini. 1994. Effectiveness of Medical Defense Interventions Against Predicted Battlefield Levels of Botulinum Toxin A. Joppa, Md.: Science Applications International Corp.

Mead, P. S., L. Slutsker, V. Dietz, L. F. McCaig, J. S. Bresee, C. Shapiro, P. M. Griffin, and R. V. Tauxe. 1999. Food-related illness and death in the United States. Emerging Infectious Diseases, 5:607-625.

Michino, H., K. Araki, S. Minami, S. Takaya, N. Sakai, M. Miyazaki, A. Ono, and H. Yanagawa. 1999. Massive outbreak of Escherichia coli O157:H7 infection in schoolchildren in Sakai City, Japan, associated with consumption of white radish sprouts. American Journal of Epidemiology, 150:787-796.

Mitchell, B. A., J. A. Milbury, A. B. Brookins, and B. J. Jackson. 1994. Use of immunomagnetic capture on beads to recover Listeria from environmental samples. Journal of Food Protection, 57:743-745.

Muir, P., F. Nicholson, M. Jhetam, S. Neogi, and J. E. Banatvala. 1993. Rapid diagnosis of enterovirus infection by magnetic bead extraction and polymerase chain reaction detection of enterovirus RNA in clinical specimens. Journal of Clinical Microbiology, 31:31-38.

Mustapha, A. R., R. W. Hutkins, and G. W. Zirnstein. 1995. Cloning and characterization of the galactokinase gene from Streptococcus themophilus. Journal of Dairy Science, 78:989-997.

Nesbitt, W. E., R. J. Doyle, K. G. Taylor, R. H. Staat, and R. R. Arnold. 1982. Positive cooperativity in the binding of Streptococcus sanguis to hydroxylapatite. Infection and Immunity, 35:157-165.

Rasooly, L., and A. Rasooly. 1999. Real time biosensor analysis of staphylococcal enterotoxin A in food International Journal of Food Microbiology, 49:119-127.

Raybourne, R. B. 1997. Flow Cytometry in Food Microbiology: Detection of Escherichia coli 0157:H7. New York: Marcel Dekker.

Robinson, B. J. 1997. Pp. 87-89 in Food Microbiological Analysis: New Technologies, M. L. Tortorello and S. M. Gendel, eds. New York: Marcel Dekker.

Rölla, G., and B. Melsen. 1975. Desorption of protein and bacteria from hydroxyapatite by fluoride and monofluorophosphate. Caries Research, 9:66-73.

Roos, H., R. Karlsson, H. Nilshans, and A. Persson. 1998. Thermodynamic analysis of protein interactions with biosensor technology Journal of Molecular Recognition, 11:204-210.

Ryan, C. A., M. K. Nickels, N. T. Hargrett-Bean, M. E. Potter, T. Endo, L. Mayer, C. W. Langkop, C. Gibson, R. C. McDonald, and R. T. Kenney. 1987. Massive outbreak of antimicrobial-resistant salmonellosis traced to pasteurized milk. Journal of the American Medical Association, 258:3269-3274.

Swaminathan, B., J. A. G. Aleixo, and S. A. Minnich. 1985. Enzyme immunoassays for Salmonella: One day testing is now a reality. Food Technology, 39:83-89.

Swanson, S. J., S. J. Jacobs, D. Mytych, C. Shah, S. R. Indelicato, and R. W. Bordens. 1999. Applications for the new electrochemiluminescent (ECL) and biosensor technologies. Developments in Biological Standardization, 97:135-147.

Swerdlow, D. L., and S. F. Altekruse. 1998. Pp. 273-294 in Emerging Infections 2, W. M. Scheld, W. A. Craig, and J. M. Hughes, eds. Washington, D.C.: ASM Press.

Tauxe, R. V., M. P. Tormey, L. Mascola, N. Hargrett-Bean, and P. A. Blake. 1987. Salmonellosis outbreak on transatlantic flights; foodborne illness on aircraft: 1947-1984. American Journal of Epidemiology, 125:150-157.

Török, T. J., R. V. Tauxe, R. P. Wise, J. R. Livengood, R. Sokolow, S. Mauvais, K. A. Birkness, M. R. Skeels, J. M. Horan, and L. R. Foster. 1997. A large community outbreak of salmonellosis caused by intentional contamination of restaurant salad bars. Journal of the American Medical Association, 278:389-395.

Troesch, A., H. Nguyen, C. G. Miyada, S. Desvarenne, T. R. Gingeras, P. M. Kaplan, P. Cros, and C. Mabilat. 1999. Mycobacterium species identification and rifampin resistance testing with high-density DNA probe arrays. Journal of Clinical Microbiology, 37:49-55.

U.S. Department of Health and Human Services, Department of Agriculture, and Environmental Protection Agency. 1997. Food Safety from Farm to Table: A National Food Safety Initiative (A Report to the President.) Washington, D.C.: Department of Health and Human Services, Department of Agriculture, Environmental Protection Agency.

Van Dilla, M. A., R. G. Langlois, D. Pinkel, D. Yajko, and W. K. Hadley. 1983. Bacterial characterization by flow cytometry. Science, 220:620-622.

Vogt, R. L., H. E. Sours, T. Barrett, R. A. Feldman, R. J. Dickinson, and L. Witherell. 1982. Campylobacter enteritis associated with contaminated water. Annals of Internal Medicine, 96:292-296.

Zirnstein, G., Y. Li, B. Swaminathan, and F. Angulo. 1999. Ciprofloxacin resistance in Campylobacter jejuni isolates: Detection of gyrA resistance mutations by mismatch amplification mutation assay PCR and DNA sequence analysis. Journal of Clinical Microbiology, 37:3276-3280.