Minimum Inhibitory Concentration of Glyphosate and of a Glyphosate-Containing Herbicide Formulation for Escherichia coli Isolates – Differences Between Pathogenicand Non-pathogenic Isolates and Between Host Species

Glyphosate is the most extensively used herbicide in the world. However, concerns regarding its safety, side effects, and impact on other organisms have increased in recent years. This is the first study to analyze a large set of recent and historical Escherichia coli isolates varying in pathogenicity and beta-lactam resistance from different host species for their susceptibility to glyphosate isopropylamine salt (IPA), the active ingredient of the herbicide, and to a complete glyphosate-containing formulation (Roundup LB Plus). For this, minimum inhibitory concentrations (MIC) were determined for 238 E. coli isolates by broth microdilution in Mueller Hinton I media followed by the statistical analyses using Mann-Whitney-U test, multivariable analysis of variance (ANOVA) and a multivariable proportional-odds ordinal regression model. While the overall MIC distribution was narrow and lacked a highly resistant sub-population for both substances, statistical analyses revealed small but significant associations between glyphosate resistance levels and different factors tested. Mean MIC values for the entire dataset showed a higher level of resistance to the complete glyphosate-containing formulation (40 mg/ml IPA) than to pure glyphosate (10 mg/ml IPA) in E. coli. Isolates that originated from poultry had significantly higher MIC values for both pure glyphosate and the complete formulation. Pathogenic and non-extended-spectrum beta-lactamase (non-ESBL) E. coli isolates each showed significantly higher MIC values compared to commensals and ESBL-producing E. coli in pure glyphosate, but not in the complete formulation. Recently sampled isolates showed statistically higher MICs than the isolates of the historic standard E. coli collection of reference in pure glyphosate, when tested by nonparametric Mann-Whitney-U test, but not in the multivariable model. Further investigations are necessary to confirm whether these associations have a casual relationship with the glyphosate use or are the consequence of co-selection due to the increased application rates of antibiotics, heavy metals or other biocides. A possible accumulation of pathogenic bacteria in livestock animals fed with glyphosate-containing feed should also be considered.