Scientists have demonstrated that the bacterium Escherichia coli – often associated with contaminated food – can be used as part of a system to detect heavy metal contamination in water.
E. coli exhibits a biochemical response in the presence of metal ions, a slight change that researchers – from the University of California, Irvine – say they were able to observe with chemically assembled gold nanoparticle optical sensors. Through a machine-learning analysis of the optical spectra of metabolites released in response to chromium and arsenic exposure, the scientists were able to detect metals in concentrations a billion times lower than those leading to cell death – while being able to deduce the heavy metal type and amount with higher than 96 percent accuracy.
The process, which the researchers said can be accomplished in about 10 minutes, is the subject of a study appearing in Proceedings of the National Academy of Sciences.
“This new water monitoring method developed by UCI researchers is highly sensitive, fast and versatile,” said co-author Regina Ragan, UCI professor of materials science and engineering. “It can be broadly deployed to monitor toxins at their sources in drinking and irrigation water and in agricultural and industrial runoff. This system can provide an early warning of heavy metal contamination to safeguard human health and ecosystems.”
In addition to proving that bacteria like E. coli can detect unsafe water, the researchers spotlighted the other necessary components – gold nanoparticles assembled with molecular precision and machine learning algorithms – which greatly enhanced the sensitivity of their monitoring system. Ragan said it can be applied toward spotting metal toxins – including arsenic, cadmium, chromium, copper, lead and mercury – at levels orders of magnitude below regulatory limits to provide early warning of contamination.
In the study, the scientists explained that they can apply trained algorithms to unseen tap water and wastewater samples, which means the system can be generalized to water sources and supplies anywhere in the world.
“This transfer learning method allowed the algorithms to determine if drinking water was within U.S. Environmental Protection Agency and World Health Organization recommend limits for each contaminant with greater than 96-percent accuracy and with 92-percent accuracy for treated wastewater,” Ragan said.
