Development and on-site evaluation of an easy-to-perform and low-cost food pathogen diagnostic workflow for low-resource communities

Food-borne diseases are a leading cause of illness and death in many developing countries and thus, there is a real need to develop affordable and practical technologies that can help improve food safety in these countries. The ability to efficiently identify food pathogens is essential to allow national regulatory authorities to monitor food quality and implement safety protocols. In this study, we have developed a simple, low-cost ($0.76 (USD)) complete food pathogen diagnostic workflow ideally suited for deployment in low-resource environments that uses a simple four step process (sample enrichment, cell lysis, DNA amplification, and naked-eye readout). The minimal number of steps and equipment involved in our diagnostic workflow, as well as the simplicity of the yes/no flocculation readout, allows non-technical personnel to perform and interpret the assay. To evaluate the system’s performance, we tested the entire system on fresh produce samples collected from local farms and markets in Cambodia for the presence of the E. coli O157 O-antigen polymerase, wzy. Although this was a proof-of-concept study, our system successfully revealed a clear correlation between the origin and condition of the produce collected and their likelihood of contamination. In conclusion, we believe that our easy-to-perform diagnostic system can have a significant impact on improving food quality and human health if adopted by regulatory authorities in developing countries due to the assay’s simplicity, affordability, and adaptability.


41
Over the last 20 years, there has been a significant increase in the incidence of food-42 borne diseases worldwide [1]. The risk of infection by a food-borne disease is 43 significantly higher in developing countries due to a combination of factors including 44 access to clean water and bathroom facilities, poor hygiene education, inadequate 45 food production and storage practices, and either insufficient food safety legislation 46 or poor implementation of existing legislation [2]. Most food-borne disease infections 47 in these countries result from the consumption of perishable foods sold in informal 48 markets [3] and, as such, food-borne diseases have become a leading cause of 49 illness and death in developing countries [4,5]. Hence, there is a need for low-cost 50 and simple technologies that can help health authorities to monitor and enforce 51 adequate levels of food safety. In addition to the health benefits, increased food 52 safety standards would likely have economic benefits as a result of increased 53 demand for fresh produce exports [3]. human pathogen as it can still cause illness even when the initial source of 75 contamination has been significantly diluted [14]. Thus, countries typically adopt a 0 76 CFU limit of E. coli O157:H7 or other STEC strains in food [15]. The standard 77 method of identifying E. coli O157:H7 is an involved process that requires a trained 78 microbiologist and takes days to complete [16,17]. While this method is well 79 established and highly reliable, it is not always practical, especially for developing

103
Sample processing and bacterial enrichment 104 In modern food microbiology laboratories, a stomacher machine is typically 105 employed to macerate tissue in growth media prior to microbial enrichment [21]. This  is as low as a few hundred cells [7,14,22], an initial enrichment step in buffered 120 peptone media for 16-24 hours is required before detection. We reasoned that the   To further test the specificity of the wzy-1 and stx2-1 primer sets, we examined their 155 ability to differentiate between E. coli O157:H7 and Salmonella enterica, a 156 genetically similar pathogenic species to E. coli [28]. Alfalfa sprouts were inoculated  (Table S1), were able to specifically identify the 160 Salmonella infected alfalfa sprouts while no amplification products were detected in 161 the E. coli O157:H7 inoculated samples and non-inoculated controls (Fig 2A). The   Procurement of amplification reagents in many developing countries is problematic 177 largely due to erratic power supplies and inadequate cold storage [29,30]. Freeze 178 drying of reactions could facilitate room temperature transport to, and storage at 179 remote locations. Trehalose is commonly used as a stabilizer of biomolecules during 180 the freeze-drying process [31,32], however our initial attempts to freeze dry a 181 complete LAMP reaction failed to produce an amplification product upon 182 reconstitution with water either in the presence or absence of trehalose ( Fig 3A). 183 Further tests revealed that the presence of betaine, an essential component of the 184 LAMP reaction, negatively affected the activity of the rehydrated reaction ( Fig 3A). The individual components developed in this study were combined to create a 208 complete diagnostic workflow suitable for testing fresh produce for E. coli O157 209 contamination (Fig 4). To test the full food pathogen diagnostic system in our   Tissues were processed following the above described workflow (Fig. 4) and positive 263 tests were obtained at all sites. In total, 25%, 39% and 26% of samples tested 264 positive for wzy on farms #1, #2 and #3, respectively.  The goal of this study was to develop a complete diagnostic workflow for food 287 pathogens tailored to countries with limited resources to enhance food biosecurity 288 capability. Our focus was to create a robust, simple and low-cost system that is safe 289 to perform by people with limited training and equipment. The successful testing of 290 our system in Cambodia on samples collected from the local farms and markets, 291 suggests that we have developed a practical system that meets these requirements 292 and is capable of providing meaningful data that can support food safety initiatives. The diagnostic workflow developed here can be easily adapted to detect the 307 presence of target genes from different food pathogens using highly specific LAMP 308 primer sets readily available in the literature [37][38][39]. Unlike specificity, the sensitivity 309 of the primers is less critical to the success of the assay as the overnight enrichment 310 step significantly increases the pathogen levels in the tested sample. Consistent 311 with this, our assay has successfully detected the presence of 1 CFU of E. coli O157 312 on a 1 g sample of Alfalfa sprouts (Fig 2B) emphasizing that the simplicity of the 17 313 detection workflow presented here does not limit its capacity to detect trace amounts 314 of pathogens on produce.

316
The focus of our investigation in Cambodia was to examine the performance of our 317 workflow rather than perform a comprehensive survey of Cambodian produce. Thus, 318 we purposely biased the sampling by seeking out produce with increased likelihood 319 of E. coli contamination such as containing mud splashes or selecting farms with free 320 roaming animals [10,23,40,41]. The data obtained in this study revealed a clear [44]. Collectively, these findings suggest that there is a real need for simple, low-cost 337 diagnostic systems to help the local authorities to improve food safety.

339
All the surveyed farms had a similar proportion (25-38%) of produce that tested 340 positive for wzy ( Fig 5C). As the farms were located within a 10 km radius of each 341 other, these findings suggest that they are exposed to similar levels of pathogenic 342 bacteria through common water supplies [40] or airborne particulates that can move 343 between farms [45]. In this proof-of-concept study, we demonstrate that our simple 344 workflow is capable of obtaining meaningful data on the prevalence of harmful food 345 pathogens in specific locations; such information is critical to improve regional food 346 safety.

348
There are many diagnostic systems for food pathogens described in the scientific 349 literature or commercially available. However, our diagnostic workflow has a number 350 of advantages over many of the available systems. Compared to some commercially 351 available diagnostic systems (e.g. lateral flow strips ($6.80 USD each, Romer labs)) 352 or systems that involve disposable electronics or custom-made microfluidic parts [46, 19 353 47]; our diagnostic system is considerably more affordable; costing $0.76 USD, 354 including all tubes and reagents (S2 Table), which can be further reduced to $0.53 355 USD if the ball bearings and 50ml tubes are carefully decontaminated and reused.

356
Cost is a critical factor for developing countries with very limited budgets devoted to 357 food safety. Lowering the cost of assays will allow increased testing and thereby 358 boosting its efficiency as a biosecurity tool [48].   amplification, and naked eye readout (Fig 4). The minimal number of steps and 393 equipment involved, as well as the simplicity of the presence/absence flocculation 394 readout, allows almost anyone, including those with limited scientific training, to 395 perform the assay. Furthermore, the low cost of the system ($0.76 USD) and broad 396 availability of its reagents, makes the system accessible to countries or institutions 397 with limited resources who might not otherwise be able to afford regular food testing.

398
Every step in the diagnostic workflow has been designed with the World Health 399 Organization's ASSURED philosophy in mind, that is, Affordable, Sensitive, Specific, 400 User-friendly, Rapid, Equipment-free, and Deliverable to those who need it [51]. As 401 the system is based on DNA amplification using specific primers, our diagnostic 402 system can be easily modified to identify a large variety of pathogens including those 403 that infect humans, crops or animals. Therefore, we anticipate that the incorporation 404 of our diagnostic system into food safety programs of developing countries will 405 facilitate improvements to both their food quality and human health.  Table). A CLUSTALW alignment was performed for each of each of these genes 417 using E. coli O157:H7 sequences found on the Genbank database. These 418 alignments were used to design LAMP primer against conserved regions of these 419 target genes using Primer Explorer software V4 (http://primerexplorer.jp/e/).   minutes to both kill any bacteria present and to release their DNA into the media.

470
One microliter of culture that had been diluted 15-fold with water was added to a 471 freeze-dried LAMP reaction that had been rehydrate with 9 µl 0.89M betaine. The