Rapid Detection of Shiga-Toxin-Producing (STP) Escherichia coli Using Surface-Enhanced Raman Spectroscopy (SERS) Combined with Chemometric Methods

Shiga-toxin-producing (STP) Escherichia coli is a significant foodborne/clinical pathogen because it can cause serious gastrointestinal illness and fatal systemic complications. Rapid, label-free, and sensitive detection of STP is considered of high priority because the pathogen has a low infectious dose and hemolytic uremic syndrome develops rapidly. Surface-enhanced Raman spectroscopy (SERS) has great potential for biochemical fingerprinting of pathogenic bacteria in terms of their molecular composition. In the current study, SERS has been utilized for the analysis of the samples of cell masses of STP and Shiga-toxin-non-producing (STNP) strains of E. coli by utilizing the Ag-NPs as a SERS substrate. Different SERS spectral fingerprints, such as proteins, lipids, carbohydrates, and nucleic acid-associated biomarkers, enable them to distinguish between toxin-producing and non-toxin-producing strains. To classify the SERS spectra, chemometric techniques including principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA) were used. PCA showed differentiation of different groups of SERS spectra, where PC1 explained 85.16% of the total variance. The PLS-DA model was evaluated using sample-level data partitioning to avoid pseudo-replication. A 60:40 sample-level split achieved 90.0% accuracy with an area under the curve (AUC) of 0.956, while leave-one-sample-out cross-validation provided a more conservative sample-level accuracy of 80.0%. Monte Carlo cross-validation showed a mean accuracy of 65%, reflecting the variability associated with the limited number of independent biological replicates. These findings demonstrate that SERS combined with chemometric analysis provides a promising proof-of-concept approach for differentiating STP and STNP E. coli strains under controlled experimental conditions.