Light-Scattering Sensor for Detection and Identification of "Indicator Bacteria" Enterobacteriaceae Family for Process Verification and Hygiene Monitoring

Marcela Martinez, Purdue University

Abstract

The Enterobacteriaceae (EB) family and its most well-known members, coliforms, are used as “indicators” of hygiene monitoring, sanitization practices, and process verification of food products. The traditional detection methods rely on bacterial growth on selective and differential media that contain different types of sugars as energy source, such as glucose for EB and lactose for coliforms. However, differentiation within the family based on ambiguous biochemical attributes is challenging and emphasizes the need for reliable identification techniques. In this study, we used a laser scatterometer, BARDOT (bacterial rapid detection using optical light scattering technology) that employs a red diode laser (635 nm) beam to generate scatter patterns of colonies directly on a plate to differentiate the members of EB family. The media tested in this study included violet red bile glucose agar (VRBGA), rapid’ Enterobacteriaceae (REB), glucose bromocresol purple agar (GBPA) and modified Kingler iron agar (mKIA). We selected REB as an optimal growth medium for this study, since REB was highly selective, resulted in highest PPV (91.4%), and generated optically transparent colonies which allow capture of the scatter patterns. Though other EB media, such as VRBGA were more selective, the members of the Enterobacteriaceae family produced high amounts of chromogens that interfered with laser propagation through colonies. Therefore, REB was used to build the EB genera scatter image library. The EB scatter pattern library contains approximately 7,000 patterns belonging to the top 15 medically important EB genera. The EB scatter image library was used for the screening and detection of artificially inoculated whole milk, egg whites, and dry infant formula, and identification of natural EB isolates from bovine raw milk and slaughter house environmental samples. In parallel, ISO and 3M methods were used as verification methods. A total 30 natural isolates were further confirmed by MALDI-TOF and 16S rRNA gene sequencing. When BARDOT libraries (Lib-1, coliform group; Lib-2, major pathogens; Lib-3, miscellaneous/opportunistic pathogens) were matched against spiked food samples with mixed cultures, variably successful identification was observed. PPV ranged from 33% to 96.5%. Subsequent tests involved individual culture analysis where PPV values ranged from 83-100%. Nonetheless, the increase in PPVs had no correlation with an accurate identification; BARDOT was unable to correctly identify colonies. Similar results were obtained from the matching of isolates from naturally contaminated samples against the EB scatter pattern groups. A positive result was established for each isolate and a BARDOT ID was obtained; however, subsequent MALDI-TOF and sequencing results confirmed these isolates were different than what BARDOT identified them as. In the same line, isolate analysis with individual genus libraries was not in agreement with the results obtained with confirmatory tests. The results overall indicate that BARDOT might accurately identify a single microorganism. However, the possibility exists that overlapping similarity of scatter patterns within EB genera affects the quality of identification.

Degree

M.S.

Advisors

Bhunia, Purdue University.

Subject Area

Food Science|Microbiology

COinS