Skip to main content
ARS Home » Northeast Area » Wyndmoor, Pennsylvania » Eastern Regional Research Center » Molecular Characterization of Foodborne Pathogens Research » Research » Research Project #431165

Research Project: Development of Portable Detection and Quantification Technologies for Foodborne Pathogens

Location: Molecular Characterization of Foodborne Pathogens Research

2018 Annual Report

1: Develop rapid and efficient techniques that separate and concentrate and/or quantify targeted pathogens from food matrices. 1A. Apply rapid and high volume centrifugal flow concentration to the separation of bacteria from food matrices. 1B. Partition and concentrate bacteria using immunomagnetic separation with a new class of antibody-coated paramagnetic particles. 1C. Compare and contrast bacteria separation and concentration with flow-through filtration systems. 1D. Develop and validate procedures for the rapid and quantitative detection of multiple foodborne pathogens. 2: Develop and validate field testing kits that rapidly screen for the presence and quantification of pathogens and/or indicator microorganisms in foods at the initial processing level. 2A. Generate portable, label-free sensors (e.g., next generation cantilever microbalance) for rapid in-line or near-line screening of foods. 2B. Generate portable antibody and/or phage-based multiplex assays including integrated comprehensive droplet digital detection (IC 3D). 2C. Develop an AlphaLISA detection protocol for target pathogens. 2D. Develop a flow-through immunoelectrochemical detection device for field portable detection of target pathogens. 3: Develop and validate rapid methods for the identification of pathogens and/or indicator microorganisms in foods for application in either the field or testing laboratories. 3A. Generate phage and/or antibody typing arrays. 3B. Generate pathogen databases and improve the accuracy of the Beam (formerly BActerial Rapid Detection using Optical scattering Technology or BARDOT) system. 3C. Direct typing (colony isolates not required) of enriched samples using a targeted-sequencing method. 3D. Generate genome sequence-based typing and identification schemes using next-generation sequencing technology (e.g., MiSeq, Ion Torrent PGM, and MinION), and characterize virulence and antibiotic resistance of microbial pathogens.

The primary objective of the plan is to develop rapid screening and identification methods for top foodborne bacterial pathogens, including STECs, top Salmonella serotypes, and Listeria monocytogenes as well as those of intermittent concern. Novel or enhanced sample preparation techniques (e.g., flow-through centrifugation, hollow fiber filtration, immunomagnetic separation), most likely in conjunction with pre-filtration, will be key for rapid concentration of food-associated bacteria to readily de detectable levels by modern rapid methods. Subsequently, improved levels of detection sensitivity are expected, perhaps even to an extremely low goal of approximately 1 cell/100 mL of target pathogen as required for real-time testing in the field, processing plant, distribution center, or retail establishment. Total assay times are foreseen to be from a few minutes to = 2 hours. Also, enhanced detection systems will be needed in order to bypass growth enrichment and achieve the desired, quantifiable detection levels. Furthermore, numerous biomarkers and the potential for false positive results using cross-reacting biorecognition elements will require multiplex detection techniques (e.g., multiplex qPCR and microarrays) that may be employed to distinguish true positive results from interference by background matrix or flora. Methods will initially be developed with culture media or buffer as the sample matrix, and then extended to application with food (primarily ground meat) in multiple sample formats: N=60 samples, meat core samples, tissue homogenates, carcass rinses, etc. The efficacy of any newly developed methods will require comparison to current “gold standard” methods in order to validate assay performance. Initially, this will be accomplished by reliance on enumeration of known bacterial isolates, quantified in pure culture with total cell counting if a significant dead population is expected. For evaluation, artificially inoculated and unknown samples will be tested with new methods as assessed against selective enrichment followed by selective and differential plate agar analysis. Regulatory-based methods, such as biochemical testing, multiplex PCR, and serotyping, and possibly whole genome sequencing, may be invoked for additional comparison. Our sister agency, FSIS, will provide guidance as to the parameters and specifics regarding acceptable validation of desired rapid bacterial detection methods. We propose that our developed methods be initially tested at FSIS regional labs using inspector obtained samples, split/divided at the lab, and tested in parallel. Eventually, testing will move to the field- first off-line and near-line, then in-line for some analysis platforms (e.g., microcantilever balance biosensor) situated in the processing environment and/or retail establishments. It is expected that multitudes of tests will be conducted given that most samples will be negative. Regulatory, and perhaps legal guidance will be anticipated to be critical since validation testing may lead to recalls if “zero tolerance” organisms are detected or if threshold amounts of positive samples (e.g., for Salmonella) are discovered.

Progress Report
Progress was made on all three objectives and their associated subobjectives which fall under National Program 108, Component I, Foodborne Contaminants by ARS researchers (Wyndmoor, Pennsylvania) under Project Plan 8072-42000-084-00D, Development of Portable Detection and Quantification Technologies for Foodborne Pathogens. The Plan focuses on 3 major goals that address development of field-friendly methods for the rapid, real-time detection and identification of foodborne bacterial pathogens: 1) Rapid microbial sample preparation, 2) Rapid foodborne bacteria detection, and 3) Rapid bacterial identification. All Objective 1 and associated subobjective milestones were substantially met and include: 1A) processing of ground beef samples (250 g beef in 750 mL of buffer) with a benchtop flow-through centrifuge which was demonstrated to process the mixture (prefiltered with glass wool) at a rate of 50 mL/min, 1B) development of a novel antibody-based antigen separation device/method led to an invention disclosure that was approved for generation of a provisional U.S. patent application by ARS’ National Life Sciences Patent Committee, and 1C) the InnovaPrep “concentration pipette” flow-through filtration unit was tasked for separation and concentration of background flora in frozen mixed vegetables and inoculated pathogens (Shiga-toxin producing E. coli or STEC) in ground beef with emphasis on particulate analysis (with a Malvern particle analyzer) and capture efficiency. Approx. 74-79% bacterial recovery was observed for the Innovaprep (30 vs. 37°C conditions and either Tris- or phosphate-buffered saline in-house diluent and wash preparations were assessed revealing no significant differences). Recovery of bacteria in ground beef was determined to be futile as the hollow fiber filtration membranes were readily clogged. Finally, regarding Subobjective 1D, the following scenarios were considered: if each food-based pellet (an emulsion of various micro-organisms, lipids, proteins, and assorted debris) after centrifugation is 1 mL and the number of organisms per pellet is 10000, then the volume per particle after re-suspension and break-up of the pellet needs to be = 0.1 µL which is not readily obtainable. On the other hand, if 10 pathogens (or colony forming units or CFUs), or fewer, are present in such a pellet, the suspended particle size needs only to be = 100 µL. These differences in particle size and the resultant non-random distribution of associated pathogen CFUs are the basis for non-stochastic sampling error. As for a more quantitative example of this problem: Assuming one could count the total number of all suspended particles (n) as well as the number of these with at least one bacterium, or CFU, attached (x) then if n = 1000 after the breakup of the concentrated sample (pellet) and if x = 900, with = 1 CFUs attached, the most probable number (MPN) of bio-attachments per particle = ln[n/(n-x)] is over 2. By increasing n 10-fold (by decreasing the volume per particle) the MPN of organisms per particle would drop to 0.0943. Under this experimental condition, the counting of positive occurrences of bacterial attachment was more equivalent to the number of CFUs present in the sample regardless of the food particle associations. For Objective 2, substantial progress was made for all subobjectives and includes: 2A) empirical demonstration of successful coating of a CRADA partner’s piezoelectric membrane sensors (PEMS) with insulative (and covalent cross-linkable) 3-mercaptopropyltrimethoxysilane via lack of electronic short-circuiting upon exposure to aqueous mixtures and fluorescence immunoassay-based results of binding of various analytes (Salmonella, STEC, and microparticles) to sensor cross-linked antibody. Currently, the PEMs were determined to be too thick (approx. 200 µm as measured using a Keyence digital microscope) and therefore unlikely to exhibit any significant impedance-related response for binding by low concentrations of analyte [result empirically confirmed], 2B) As a draft MTRA with our potential IC 3D (integrated comprehensive droplet digital detection) collaborator (University of California, Irvine, California) is under review by their legal team, proof of concept of the key analytical aspect of IC 3D, i.e., ability to detect and characterize individual bacteria cells in complex food matrices was investigated with a comparable technology (droplet digital PCR or ddPCR) under an MTRA with the Food Science Division of Bio-Rad Laboratories (Marnes-la-Coquette, France). The objective of this collaboration is to determine if there are competitive advantages to ddPCR that can simplify and/or improve the detection of pathogens, specifically STEC in foods. Current screening PCR methods targeting stx and eae allow only independent detection of markers, while ddPCR enables the co-localization of STEC virulence genes which can help simplify the STEC identification process. Inherently quantitative, ddPCR does not rely on standard curves which can adversely affect the reliability and repeatability of the assay. The BioRad ddPCR system is in process of being benchmarked against the current FSIS MLG 5 standard, and Bio-Rad’s real time PCR assay, IQ Check for detection of STEC in ground beef. Preliminary results indicate that the assay is accurate and could serve as a more efficient screening protocol than the presently used BAX, real time PCR. Specifically, as the assay can determine if the target virulence factors are contained within one cell, as opposed to a mixture of cells, it will reduce the number of samples which need to be processed and further assessed. This could have a significant impact as FSIS relayed to ARS that 80% of the samples which test positive in the screening step of the MLG 5 are false positives. The device has also identified nonregulated serotypes which can cause foodborne illness that would not be identified by the present MLG protocol. The above data will be presented at the 6th qPCR & Digital PCR Congress at the 4Bio Summit (Fall 2018). For Subobjective 2C, in a collaboration with CRADA partner, AlphaLISAs were developed, optimized, and benchmarked (vs. a commercial ELISA) for STEC-generated Shiga toxin 2 (Stx2) in foods (Romaine lettuce and ground beef) allowing for more rapid analysis of Stx2 with less manual manipulation thus improving assay throughput and the ability to automate sample screening while maintaining detection limits of 0.5 parts-per-billion, and 2D) initial results focused on flow-through enzyme-linked immunoelectrochemical (FT-ELIEC) detection of Salmonella, and demonstrated that FT-ELIEC was two orders of magnitude more sensitive than ELISA, with a significantly larger dynamic range. FT-ELIEC was also demonstrated to accommodate sample volumes that are significantly larger than those used in conventional immunoassays; specifically the sensor can accommodate 60 mL sample volumes as opposed to 0.2 – 1 mL sample volumes used in ELISA platforms. For Objective 3, substantial progress was made on all subobjectives: 3A) bacteriophage (a virus that may specifically infect bacterial serotypes) isolated in the laboratory of our collaborator with the Center for Food Safety and Engineering at Purdue University (West Lafayette, Indiana) were tested for specificity to STEC (specifically E. coli serotype O157:H7 or STEC O157:H7) as it is the most common STEC causing foodborne illness. Previously an STEC O157:H7 detection system was developed employing the STEC-specific phage that was engineered to produce light when STEC O157:H7 is present. This year our collaboration further defined the binding requirements of the light-producing phage reporter to test if it is feasible to allow the detection method to be run while food samples are shipped from the producer to the testing laboratory. The success of these studies will minimize the likelihood of false-negative results and greatly improve the time-to-result for detection of STEC O157:H7, respectively. Specificity of the phage binding/infection was important for determining its effectiveness in array-based application. Currently we are working with our collaborator at Purdue University to fill a postdoctoral researcher vacancy with a predoctoral student who is also well-versed in the isolation, propagation, and manipulation of bacteriophage, 3B) a database for BARDOT scatter images was generated for Campylobacter species under various culture conditions that exploited the migratory/mobile-behavior of the bacteria, 3C) all milestones for this subobjective were unexpectedly met way under schedule and a manuscript on this work has already been published and recorded, and 3D) generation of genomic sequence databases for important foodborne pathogens focused on furthering investigations with Campylobacter spp., not only for its virulence, but also antimicrobial resistance properties. A total of 36 Campylobacter strains were isolated from retail chicken and beef liver samples by passive membrane filtration and confirmed to be either C. jejuni or C. coli by a multiplex real-time PCR assay previously developed in our laboratory. Genomic DNAs of all the strains were purified and sequenced using Illumina’s Miseq. Some of the genomes were also sequenced via PacBio’s Single Molecule Real-Time technology. Substantial progress was made on whole genome assembly of, annotation of, and determination of single nucleotide polymorphisms in the derived sequences. Multilocus sequence typing analysis of the strains, which provides genetic basis for the species identification and genotyping of the pathogen, was also performed. Comparative analyses at genomic and phenotypic levels have yielded a better understanding of the molecular mechanisms of virulence and antimicrobial resistance of the organism.

1. A novel test for detection of a bacterial toxin in food. Production of Shiga toxin (Stx) is both an important virulence factor for the pathogenic bacterium, Escherichia coli, and a distinguishing feature routinely screened for in meat samples by the USDA’s Food Safety and Inspection Service. Under a collaborative agreement between Abraxis, LLC (Warrington, Pennsylvania) and a team of ARS researchers in Wyndmoor, Pennsylvania, a novel antibody-based test (amplified luminescent proximity homogeneous assay or AlphaLISA) was developed for the detection of Shiga toxin 2 (Stx2) generated by Stx-producing E. coli (STEC) in foods (Romaine lettuce, ground beef). Efficacy and sensitivity trials showed that not only was AlphaLISA as sensitive as the industry standard test (enzyme-linked immunosorbent assay or ELISA), but it also demonstrated a superior signal-to-noise ratio with the ability to distinguish high concentrations of the toxin. These features in combination with the reduced hands-on workflow and amenability for automation, make the AlphaLisa method a more economically viable choice for Stx2 detection in foods by both regulatory agencies and food testing labs alike.

2. A new method for rapidly detecting harmful bacteria in processed food. The ability to quickly test food in a production facility for the presence of harmful bacteria like Salmonella requires a very sensitive, simple to use, and inexpensive method. This method must be very sensitive as it cannot rely on culture-based growth enrichment which takes too long and is too risky because there would be the potential to grow up high amounts of bad bacteria right where the food is being made or processed. ARS researchers in Wyndmoor, Pennsylvania have developed a new method that meets the needed criteria by combining existing electrochemical detection techniques with innovative sample concentration technology. This novel method was demonstrated to readily detect extremely low concentrations (approx. 10 cells/mL) of two major pathogens (E. coli O157:H7 and Salmonella) in up to 1 liter of ground beef samples in approx. 2 hours. An invention disclosure that describes this biosensor platform has been approved by the USDA’s National Mechanical & Measurements Patent Committee for submission as a patent application to the United States Patent and Trademark Office. This low cost and simple to use, disposable test may be applied for the rapid detection of foodborne pathogens in the field by production and regulatory personnel alike. With very minor changes, this method may also be applied to the rapid detection of other regulated substances including drug residues, chemical contaminants, toxins, and viruses as well as other bacterial pathogens and DNA.

Review Publications
Gurtler, J., Doyle, M.P., Kornacki, J.L., Fratamico, P.M., Gehring, A.G., Paoli, G. 2017. Advantages of virulotyping foodborne pathogens over traditional identification and characterization methods. Foodborne Pathogens Virulence Factors and Host Susceptability. New York, NY: Springer Publishing. p. 3-40.