Surface-enhanced Raman spectroscopy and machine learning based profiling of β-lactam antibiotic-induced biochemical responses in methicillin-resistant and methicillin-susceptible Staphylococcus aureus strains.

The global escalation of antimicrobial resistance in Staphylococcus aureus (S. aureus) necessitates rapid analytical tools to reliably distinguish between methicillin-susceptible (MSSA) and methicillin-resistant (MRSA). In this study, the surface-enhanced Raman spectroscopy (SERS) was utilized to antibiotic induced biochemical response in MSSA and MRSA cell pellets following exposure to oxacillin (1 μg/ml) and amoxicillin (25 μg/ml). Distinct SERS spectral features at 566, 796, 857, 1004, 1094, 1280, 1374, 1706, and 1588 cm-1 were observed, corresponding primarily to alterations in proteins, nucleic acids, and cell wall associated biomolecules under antibiotic stress. Principal Component Analysis (PCA) revealed clear unsupervised segregation of susceptible and resistant strains, while Partial Least Discriminant Analysis (PLS-DA) enabled robust classification with a cross-validation of 99% sensitivity, 87% specificity, 99.5% accuracy. These findings demonstrate that SERS combined with multivariate analysis can demonstrate potential for distinguishing antibiotic susceptibility phenotypes and capturing resistant-associated biochemical signatures in S. aureus. These findings represent a proof-of-concept and require validation across a broader range of clinical isolates.