Incidence and Methods of Cyclospora cayetanensis Detection in Environmental Waters

Abstract

The coccidian parasite Cyclospora cayetanensis is the causative agent for foodborne outbreaks of cyclosporiasis and multiple annual fresh produce recalls. In recent years during outbreak investigations, this organism has been found in irrigation waters prompting a call for more research on its environmental prevalence in the United States. Currently, there is little information available on methods to conduct this research, including methods for DNA extraction for molecular testing. Extraction methods for environmental samples must be robust due to the limited amount of target available in the sample, and the potential for molecular inhibitors present in the sample. The AllPrep PowerViral DNA/RNA Kit, RNeasy PowerSoil Total RNA Kit (with DNA extraction modification) and the Universal Nucleic Extraction (UNEX) were included in this assessment to determine which gives the highest recovery of DNA to be used in molecular testing. Environmentally prevalent organisms (Salmonella enterica and Murine Norovirus) and multiple sample types (surface water, produce wash water, and sewage sludge) were assessed to identify a method that can be used universally for more than one organism and sample type. In terms of extraction for C. cayetanensis, the PowerViral and the UNEX method a with heated incubation performed comparably for tap water, produce wash water and sludge sample. The PowerViral and UNEX methods both had superior recovery and detection rates of C. cayetanensis than the PowerSoil method for sludge samples. We observed the highest recovery rates, for Murine Norovirus and S. enterica from water using the PowerViral method. Overall, for all water sample types and targets, we observed the highest detection rates and greatest recovery using the PowerViral method, as compared to the PowerSoil and UNEX methodologies.

The recent occurrence of outbreaks of cyclosporiasis associated with fresh produce originating from the United States suggests that environmental prevalence C. cayetanensis in agricultural areas needs evaluation to better characterize the risk of C. cayetanensis contamination in produce grown in the United States. Here, we assess the presence of C. cayetanensis in samples of municipal wastewater sludge, on-farm portable toilets, irrigation pond water, and spent packing house produce wash water in a Southern Georgia agricultural area. We used the extraction methods showing the greatest recover to assess risk of contamination. Water samples were tested using human-specific fecal molecular markers Bacteroides HF183 and crAssphage. The rate of C. cayetanensis detection ranged between 27 and 52% in samples of irrigation pond water using qPCR methods with difference master mixes. In contrast, there were detection in less than 1% of dump tank water samples by qPCR. Notably, none of the qPCR detections in water samples were positive for C. cayetanensis with confirmatory sequencing, suggesting the qPCR results were potential false positive results. Rates of HF183 detections were 33% in irrigation pond water samples and 4% in dump tank water samples, while crAssphage was detected in 6% of pond water samples. Presence of human-specific fecal markers in these samples points to the likelihood of irrigation water and produce encountering human fecal contamination in this region. The finding of a 9% rate of C. cayetanensis detection in samples from municipal wastewater sludge suggests that this could potentially be a source of C. cayetanensis contamination in the environment.

Based on these results, we aimed to identify potential cross-reacting species for the C. cayetanensis 18S rRNA assay and an additional new MIT1C gene target qPCR assay. The environmental samples evaluated were irrigation water, produce wash water, and wastewater treatment sludge from the previous study with qPCR detections of C. cayetanensis by the 18S rRNA gene target qPCR. From these samples, we deep-sequenced longer regions of the 18S rRNA gene and the mitochondrial cytochrome c oxidase subunit III gene (cox3). Out of 65 irrigation water samples with detection results using the C. cayetanensis 18S rRNA gene qPCR assay, none had MIT1C qPCR assay detections or sequences that clustered with C. cayetanensis based on deep sequencing of the cox3 and 18S rRNA gene. Instead, sequences from these samples clustered around coccidia sequences found in bird, fish, reptile, and amphibian hosts. Out of 26 sludge samples showing detections by either qPCR assay, only 14 (54%) could be confirmed to contain C. cayetanensis by deep sequencing of cox3 and 18S rRNA gene regions. In a portion of the remaining sludge samples, sequenced reads clustered around coccidia from rodents. This study demonstrated that caution should be taken when interpreting qPCR C. cayetanensis detection data in environmental samples, and confirmatory steps or a multiple-tool testing panel consisting of molecular (qPCR and sequencing), microscopic, or statistical methodologies will likely be needed for confirmation.

Produce irrigated with water contaminated with microbes contributes to foodborne illness in the U.S. Testing agriculture source water for human fecal indicators can provide farm operators, regulatory, and academic organizations an improved understanding of the quality and contamination sources of water used in produce production. Using two previously collected datasets, we developed and assessed the efficacy of logistic regression and machine learning models in predicting human specific and general fecal contamination of southern Georgia produce irrigation water as represented by human specific Bacteroides (HF183) and crAssphage and generic Escherichia coli and coliphage, respectively. Using a training dataset, the risk factors identified by the models as significant contributors to human fecal contamination to irrigation surface water sources included rainfall (in the previous 2-7 days) and being located within 500 feet of a building. Rainfall was the only predictor identified as a significant contributor to generic fecal contamination of irrigation water. However, the models could not predict generic fecal indicators nor human fecal indicators across multiple datasets in the same geographic area. While the models demonstrate that nearness to buildings and rainfall likely increase the risk of fecal contamination of agricultural waters, caution should be taken when using statistical models of fecal contamination of agricultural waters from a single study to predict contamination risk in other settings including within the same growing region.