Towards Real-Time Bacterial Detection in Wastewater with Tailored Plasmon-Pathogen Interactions

Elevated pathogen levels in wastewater are one of the first indicators of disease outbreaks, making wastewater a powerful tool for surveilling the infections present in a community. Wastewater-based epidemiology (WBE) has gained popularity for viral infections, particularly to monitor ongoing epidemics of COVID-19, monkeypox, and polio. WBE monitoring of bacterial pathogens is likely to become increasingly important, especially as antimicrobial bacteria become a threat to public health. However, bacterial WBE presents several outstanding challenges. Foremost, the nucleic acid-based methods currently used to identify pathogens in wastewater require amplification or sequencing, and are limited by the need for pathogen-specific probes; therefore, is challenging to scale WBE to the many dozens of bacterial species and strains that can be present in a community. Further, high false negative rates can result from substances in wastewater, such as nucleases that degrade nucleic acids, heavy metals that bind to nucleic acids, and sugars that inhibit the activity of enzymes involved in DNA amplification and sequencing.<br/><br/>Here, we describe a new method to rapidly and accurately detect and identify bacteria in wastewater without the need for targeted probes or nucleic acids by combining surface-enhanced Raman spectroscopy (SERS) of<br/>pathogens with machine learning (ML). We: 1)maximize bacterial Raman signal in wastewater using plasmonic materials that selectively enhance the pathogen Raman spectra across a variety of pathogens; 2) create an ML algorithm that can discern between the most common bacterial species in wastewater; and 3) validate this identification system on real wastewater samples using an integrated microfluidic setup. Au nanorods with five distinct aspect ratios are synthesized using CTAB and NaOL as the surfactants. We also synthesize rods with polyethylenimine (PEI) and thiol polyethylene glycol (thiol-PEG) to vary the nanorod charge. Gram negative species such as <i>E.coli and S.marcescens, and </i>G positive species such as <i>S.epidermidis </i>and <i>S.aureus </i>are grown to log phase in Lysogeny broth (LB) culture medium and then spiked into wastewater with concentrations ranging from 10^5cells/mL-10^9 cells/ml. The nanorods are mixed into the wastewater, and liquid Raman spectra are collected in a miniaturized flow cell. By controlling the nanorod chemistry and concentration, we obtain uniform bacterial SERS enhancements of ~300X for all bacterial species, compared to the bacteria-only samples in both water and wastewater. We compare wastewater from various locations, each with their own unique compositions, pHs, and salinities, and show how machine learning based spectral data analysis enables high (&gt;88%) classification accuracy of both species and antibiotic susceptibility for these bacteria. Our work allows high signal-to-noise bacterial SERS measurements from wastewater, providing a foundation for rapid and label-free bacterial WBE.