Regional differences in presence of Shiga toxin-producing Escherichia coli virulence-associated genes in the environment in the North West and East Anglian regions of England

Signiﬁcance and Impact of the Study: Several outbreaks within the UK have highlighted the danger of contracting Shiga toxin-producing Escherichia coli from contact with areas recently vacated by livestock. This is more likely to occur for STEC infections compared to other zoonotic bacteria given the low infectious dose required. While studies have determined the prevalence of STEC within farms and petting zoos, determining the risk to individuals enjoying recreational outdoor activities that occur near where livestock may be present is less researched. This study describes the prevalence with which stx genes, indicative of STEC bacteria, were found in the environment in the English countryside. Abstract Shiga toxin-producing Escherichia coli is carried in the intestine of ruminant animals, and outbreaks have occurred after contact with ruminant animals or their environment. The presence of STEC virulence genes in the environment was investigated along recreational walking paths in the North West and East Anglia regions of England. In all, 720 boot sock samples from walkers’ shoes were collected between April 2013 and July 2014. Multiplex PCR was used to detect E. coli based on the ampliﬁcation of the uid A gene and investigate STEC-associated virulence genes eae A, stx 1 and stx 2. The eae A virulence gene was detected in 45 (cid:1) 5% of the samples, where stx 1 and/or stx 2 was detected in 12 (cid:1) 4% of samples. There was a difference between the two regions sampled, with the North West exhibiting a higher proportion of positive boot socks for stx compared to East Anglia. In univariate analysis, ground conditions, river ﬂow and temperature were associated with positive boot socks. The detection of stx genes in the soil samples suggests that STEC is present in the English countryside and individuals may be at risk for infection after outdoor activities even if there is no direct contact with animals.


Introduction
Shiga toxin-producing Escherichia coli cause diarrhoea, often bloody, that can progress to anaemia and kidney failure. An estimated 30% of cases require hospitalization during the course of infection (Byrne et al. 2015). In England, there is an average of 900 cases a year identified by Public Health England, but the real number of infections is probably much higher (Adams et al. 2016). The first outbreak of STEC was attributed to ground beef patties at a fast food restaurant chain in the United States in 1982 (Riley et al. 1983), and meat products are still associated with causing many STEC outbreaks (Heiman et al. 2015;Adams et al. 2016). However, several outbreaks have been attributed to the presence of people on fields recently vacated by ruminant animals (Crampin et al. 1999;Howie et al. 2003). Additionally, all published case-control studies performed on sporadic infections within the UK have identified contact with animals or their environments as a significant contributor to sporadic infections (Kintz et al. 2017). A recent study reports significant associations between livestock density and the spatial distribution of STEC infections in England (Elson et al. 2018). All this raises interesting questions about the likelihood of acquiring a STEC infection by participating in recreational outdoor activities in the countryside.
Cows and other ruminant animals are able to carry STEC asymptomatically as part of their normal intestinal flora since their epithelial cells lack the receptors for internalizing the Shiga toxin. Studies on cows in the UK have demonstrated that, at any time, 4-15% of the animals in a herd may be carrying STEC Mechie et al. 1997;Omisakin et al. 2003;Paiba et al. 2003). The bacteria can survive in cow pats up to 21 months (Kudva et al. 1998;Hutchison et al. 2005;Fremaux et al. 2007), and within soil, STEC can be detected up to 200 days after inoculation (Maule 2000;Jiang et al. 2002;Bolton et al. 2011). STEC requires only a small number of bacteria, between 10 and 100, to be ingested to cause illness (Tuttle et al. 1999;Strachan et al. 2002). Its extended survival time under different conditions and the low infectious dose increase the likelihood that individuals may become ill after encountering STEC in the environment.
STEC shares its zoonotic transmission with other gastrointestinal bacterial pathogens such as Campylobacter and Salmonella. Therefore, many techniques have been developed to screen for these pathogens within farms in an effort to curtail transmission from animals to consumers. One of these methods is using boot socks to cover the shoes and collecting samples from the floor or ground as the individual walks around the premises (Caldwell et al. 1998;Skov et al. 1999;McCrea et al. 2005). These boot socks can then be analysed for the presence of the pathogens, indicating colonized animals are present. This method has been demonstrated to be just as, if not more, sensitive than the older drag swab method developed for use on chicken farms (Buhr et al. 2007;Lungu et al. 2012). Boot socks have also recently been used to sample for pathogens in the wider environment (Brena et al. 2016;Jones et al. 2017). The aim of this research was to analyse boot socks generated during the Enigma project for Campylobacter to determine the frequency with which STEC-associated virulence genes were detected in the English countryside (Jones et al. 2017). Two different locations, the North West of England and East Anglia, encompassing different land uses, climates and geographies, were chosen for the study (Jones et al. 2017).

Results and discussion
Presence of STEC virulence genes in boot sock samples There were six walk locations: three in the North West and three in East Anglia. The walks were completed between April 2013 and July 2014, occurring every week from April through July and every 3 weeks during August through March, leading to a total of 40 walk dates for each location. For each walk, three different walkers wearing boot socks were present. This meant a total of 720 boot socks were collected (6 locations 9 40 walks per location 9 3 walkers). After the walks were completed, the walkers sent the boot socks in sterile plastic bags for processing. Further details on the walks and boot sock collection can be found in Jones et al. (2017).
Multiplex PCR was used to detect three different STEC virulence genes: the shiga-toxins stx1 and stx2 and the intimin eaeA as a marker for the locus of enterocyte effacement (LEE). This method was chosen as it has previously successfully detected the presence of STEC from a variety of different sources and would lend itself to quickly screening a large number of environmental samples (Deng and Fratamico 1996;Paton and Paton 1998a;Noll et al. 2015). The uidA gene was also included as an indicator that E. coli was present in the culture grown up from the frozen boot socks samples (McDaniels et al. 1996). Of the 720 samples, 592 (82Á2%) of the samples amplified uidA, indicating E. coli had grown in the overnight culture. Only 14 samples amplified virulence genes (10 eaeA and 4 stx2) in the absence of uidA; these were still included in the subsequent analyses given the low numbers of this occurrence compared to total boot sock samples. The breakdown of the virulence genes detected in the 720 boot socks samples is shown in Fig. 1. In total, 45Á5% of samples were positive for the eaeA and 12Á4% positive for either of the stx genes. For stx genes, stx2 was detected more often than stx1 with 9Á2% of the samples positive for only stx2 and 2Á1% of the samples positive for only stx1; 1Á1% of the samples contained both stx genes. Furthermore, the majority of the stx-positive samples were also positive for eaeA, with only 2Á2% of the samples containing only stx1 and/or stx2, but 10Á1% of the samples containing either stx gene or eaeA.
When considering only the stx-positive samples, a majority (83%) amplified the stx2 gene. Compared to stx1, presence of stx2 in STEC bacteria is associated with more severe disease and a higher likelihood to progress to haemolytic uremic syndrome (Boerlin et al. 1999;Tarr et al. 2005). Furthermore, 82% of stx-positive samples also amplified the eaeA gene. STEC strains that carry the LEE pathogenicity island are associated with more severe disease (Paton and Paton 1998b;Boerlin et al. 1999). If the virulence genes detected on these walks are associated with live bacteria, these results indicate that there would be the possibility for these strains to cause potentially severe disease if individuals became infected.
While the use of culturing from the boot sock samples suggests that the detected virulence factors came from live bacteria, it does not necessarily mean that these bacteria are an immediate threat to human health. One caveat to the method of detection for the boot socks as performed in this study is that there is no manner to quantify the amount of pathogenic bacteria that may have been encountered during a walk or that the bacteria would be ingested by the individual. Another caveat is that the detection method used does not guarantee that the detected virulence genes came from the same bacteria; eaeA is also associated with several other types of pathogenic E. coli that may have also transferred to the boot socks and grown under the culturing conditions. Additionally, it is possible that free stx phage was present on the boot sock and this then infected the E. coli growing in the overnight sample, leading to stx-positive samples that did not initially contain STEC bacteria.

Distribution of stx by region and walk location
As eaeA is associated with other pathogenic E. coli, further characterization of the boot sock samples focused on the stx virulence genes. Overall, 89 of the 720 boot socks were positive for at least one stx gene. The results were broken down by region and walk location to see whether there were any differences in the presence of the stx genes in the environment between the two regions. The regional analysis demonstrated that the number of stx-positive boot socks was much lower in East Anglia region compared to the North West, with 0Á8-5Á8% of the 120 boot socks per location demonstrating the presence of an stx gene in East Anglia compared to 15Á8-25% in the North West (Fig. 2). As far as the number of walks that exhibited at least one stx-positive boot sock, at least one of the walkers on 2Á5-15% of the walks in East Anglia walked through soil that later allowed for amplification of one of the stx genes. In the North West, this was between 32Á5 and 42Á5% of the walks.
More stx-positive boot socks were found positive for the North West region compared to East Anglia. This correlates with the higher amount of livestock that are present in the North West compared to East Anglia (Elson et al. 2018) and also suggests that there is a significant reservoir for stx-containing bacteria in such environments. There are also more STEC infections in the North West compared to East Anglia (Byrne et al. 2015;Visham 2019). It is reassuring for using boot socks as a sampling method that the results reflect both the trends in potential animal reservoirs and the recorded number of human infections.

Seasonality of STEC virulence genes in the environment
STEC infections exhibit a seasonal peak, with the number of human cases rising during summer months (Byrne et al. 2015). To see whether the detection of the stx virulence factor in the environment followed the seasonal trends for STEC infections, the number of positive boot socks from each region was plotted against the month of the walks occurring (Fig. 3). Too few boot socks in East Anglia were positive for stx to give any indication of seasonal differences of stx in the environment. However, the North West demonstrated an increase in the presence of stx over the summer months, with the number of positive boot socks dipping in October 2013 then increasing again in May 2014.

Environmental conditions associated with positive boot socks
Using information collected during the walks and data recorded from local weather stations on the days of the walks, regression analysis was used to determine whether there was any association with positive boot socks and environmental conditions. In univariate analyses, state of the ground, mean river flow and mean daily temperature were associated with the number of positive boot socks per walk. In particular boggy ground and river flow were negatively associated with stx-positive boot socks while the 7-day mean temperature was positively associated with stx-positive samples (Table 1). In the final model, only mean daily temperature was significantly associated with the number of positive boot socks (IRR = 1Á20 (1Á12-1Á28), P < 0.001); the only other variable in the final model after removing all predictors with a P > 0Á2 was mean river flow in the previous 7 days (IRR = 0Á89 (0Á777-1Á02), P = 0Á094).

Consistency of boot socks within walks
Since three different boot socks were worn during each walk, the number of walks where the boot socks all exhibited the same result (either all positive or all negative for stx) was calculated (Fig. 4). In all, 181 of all walks had zero positive boot socks while only seven of the walks, all occurring in the North West, found all three boot socks positive. This means that less than a quarter of all walks (21Á7%) exhibited variation in the presence or absence of STEC-related virulence factors on the boot socks collected.

Conclusions
Using multiplex PCR, we were able to successfully detect the virulence factors eaeA and stx genes, associated with STEC, from boot socks collected during walks in the English countryside. The samples for this study were initially collected to investigate the presence of Campylobacter in the environment (Jones et al. 2017). This initial analysis of the boot socks found 47Á1% positive for Campylobacter. Similar to the results of our study, Campylobacterpositive boot socks were also detected more frequently in the North West over East Anglia. Overall, 41 (5Á7%) of the 720 boot socks collected demonstrated the presence of Campylobacter and either the eaeA and stx virulence genes. The results of both of these studies indicate that boot socks can be used successfully for collecting environmental samples across a wide sampling area and then used to detect multiple different pathogens. Given the convenience of this method, it is now being used by Public Health England to assist with sampling during outbreak investigations (McFarland et al. 2017).

Preparation of samples from boot socks
Details on the choice of walking routes and use of citizen scientists for sample collection are shown in Jones et al. (2017). For processing the boot socks, 100 ml of room temperature buffered peptone water was added to the   sterile sample bag the boot socks were received in. These were palpated to re-suspend any material and 4 ml was removed and frozen in cryovials. This was undertaken for all boot socks received during the original study period from April 2013 through July 2014.

Isolation of genomic DNA and multiplex PCR
The samples frozen in peptone-buffered water were thawed and 0Á5 ml was added to 4Á5 ml modified tryptone soy broth (Oxoid CM0989, Thermo Scientific, Basingstoke, UK) to enrich for E. coli. Cultures were grown overnight at 37°C, 1 ml was removed and the bacteria were pelleted. Genomic DNA isolation was performed using Qiagen's DNA mini kit (Qiagen, Manchester, UK). Primers were purchased from Sigma-Aldrich (Haverhill, UK). The uidA primers were based on McDaniels et al. (1996) while the eaeA and stx primers were from Son et al. (2014) (McDaniels et al. 1996Son et al. 2014). Qiagen's HotStar Taq Plust Master Mix kit, supplemented to 2Á5 mmol l À1 MgCl 2 , was used to perform all multiplex PCR reactions according to the manufacturer's directions using 0Á2 µl of gDNA. Cycling conditions included an initial 10 min at 95°C to activate the polymerase followed by 25 cycles of 95°C for 1 min, 54°C for 1 min, and 72°C with a final extension step of 7 min at 72°C. Samples were run on 2% TAE gel electrophoresis with ethidium bromide and visualized with UVP's ChemiDoc-IT 2 imager. The expected sample sizes were 623 bp for uidA, 482 bp for stx2, 306 bp for stx1 and 245 bp for eaeA.

Statistical analysis
During the walks, information was collected on the condition of the footpath and the number of livestock seen by the walkers. Information was also collected from local weather stations on the flow of nearby rivers and the average temperatures and amount of rainfall in the 7 days up to and including the day of the walk. These were used in a longitudinal panel negative binomial regression univariate analysis using the number of boot socks as count data. All variables with P < 0Á2 in the univariate analysis were combined in a multiple variable analysis. Using a backwards step-wise process, variables were removed if P > 0Á2 until only variables with P < 0Á2 remained in the model. All statistical analyses were performed in STATA 14.