Pathogen levels in wastewater are correlated with rates of infectious diseases, making wastewater-based epidemiology (WBE) a powerful tool. With the rise of antibiotic-resistant bacteria, WBE monitoring of bacteria will become more important. Surface-enhanced Raman spectroscopy (SERS) has potential as a technique for bacterial identification in wastewater: because it directly measures intrinsic properties of a sample, it does not require targeted probes. Our group has previously used Raman combined with machine learning (ML) to identify the species, strain, and antibiotic susceptibility of bacteria. Moreover, we have achieved SERS using plasmonic gold nanorods (AuNRs) that electrostatically bind to the cell surface, allowing rapid analysis of liquid samples in deionized water.

Here, we achieve SERS from liquid samples of clinically-relevant bacterial species Staphylococcus epidermidis, Staphylococcus aureus, Escherichia coli, and Serratia marcescens spiked into filter-sterilized wastewater. In this wastewater matrix, we observe enhancement of biologically-relevant peaks describing proteins, nucleic acids, and lipids comparable to enhancement in deionized water, suggesting that the complex mixture of dissolved components in wastewater do not interfere with bacteiral detection. We use transmission cryoelectron microscopy to quantify AuNR binding to each species in wastewater and correlate this binding efficiency with Raman signal. Additionally, we show that bacterial SERS spectra collected in wastewater can be used to predict bacterial species present: we train a convolutional neural network to achieve >85% species identification accuracy in a wastewater matrix. Additionally, we show a limit of SERS-based detection of 10⁵ cells/ml, which nears realistic concentrations for bacterial bacterial pathogens during epidemics. Finally, we build a microfluidic Raman flow cytometer that flows single cells through a channel while collecting their Raman spectra, which we intend to use for high-throughput single-cell detection to characterize complex bacterial mixtures in wastewater. Our approach demonstrates key steps toward use of SERS and ML to rapidly detect bacterial outbreaks through wastewater before they become crises.

Fareeha Safir

Appearances