Moving towards on-site detection of Shiga toxin-producing Escherichia coli in ready-to-eat leafy greens

Abstract

Rapid identification of Shiga toxin-producing Escherichia coli, or STEC, is of utmost importance to assure the innocuousness of the foodstuffs. STEC have been implicated in outbreaks associated with different types of foods however, among them, ready-to-eat (RTE) vegetables are particularly problematic as they are consumed raw, and are rich in compounds that inhibit DNA-based detection methods such as qPCR. In the present study a novel method based on Loop-mediated isothermal amplification (LAMP) to overcome the limitations associated with current molecular methods for the detection of STEC in RTE vegetables targeting stx1 and stx2 genes. In this sense, LAMP demonstrated to be more robust against inhibitory substances in food. In this study, a comprehensive enrichment protocol was combined with four inexpensive DNA extraction protocols. The one based on silica purification enhanced the performance of the method, therefore it was selected for its implementation in the final method. Additionally, three different detection chemistries were compared, namely real-time fluorescence detection, and two end-point colorimetric strategies, one based on the addition of SYBR Green, and the other based on a commercial colorimetric master mix. After optimization, all three chemistries demonstrated suitable for the detection of STEC in spiked RTE salad samples, as it was possible to reach a LOD50 of 0.9, 1.4, and 7.0 CFU/25 g for the real-time, SYBR and CC LAMP assays respectively. All the performance parameters reached values higher than 90 %, when compared to a reference method based on multiplex qPCR. More specifically, the analytical sensitivity was 100, 90.0 and 100 % for real-time, SYBR and CC LAMP respectively, the specificity 100 % for all three assays, and accuracy 100, 96 and 100 %. Finally, a high degree of concordance was also obtained (1, 0.92 and 1 respectively). Considering the current technological advances, the method reported, using any of the three detection strategies, demonstrated suitable for their implementation in decentralized settings, with low equipment resources.