Relating growth potential and biofilm formation of Shigatoxigenic Escherichia coli to in planta colonisation and the metabolome of ready-to-eat crops

Contamination of fresh produce with pathogenic Escherichia coli, including Shigatoxigenic E. coli (STEC), represents a serious risk to human health. Colonisation is governed by multiple bacterial and plant factors that can impact on the probability and suitability of bacterial growth. Thus, we aimed to determine whether the growth potential of STEC for plants associated with foodborne outbreaks (two leafy vegetables and two sprouted seed species), is predictive for colonisation of living plants as assessed from growth kinetics and biofilm formation in plant extracts. Fitness of STEC was compared to environmental E. coli, at temperatures relevant to plant growth. Growth kinetics in plant extracts varied in a plant-dependent and isolate-dependent manner for all isolates, with spinach leaf lysates supporting the fastest rates of growth. Spinach extracts also supported the highest levels of biofilm formation. Saccharides were identified as the major driver of bacterial growth, although no single metabolite could be correlated with growth kinetics. The highest level of in planta colonisation occurred on alfalfa sprouts, though internalisation was 10-times more prevalent in the leafy vegetables than in sprouted seeds. Marked differences in in planta growth meant that growth potential could only be inferred for STEC for sprouted seeds. In contrast, biofilm formation in extracts related to spinach colonisation. Overall, the capacity of E. coli to colonise, grow and internalise within plants or plant-derived matrices were influenced by the isolate type, plant species, plant tissue type and temperature, complicating any straight-forward relationship between in vitro and in planta behaviours. Importance Fresh produce is an important vehicle for STEC transmission and experimental evidence shows that STEC can colonise plants as secondary hosts, but differences in the capacity to colonise occur between different plant species and tissues. Therefore, an understanding of the impact of these plant factors have on the ability of STEC to grow and establish is required for food safety considerations and risk assessment. Here, we determined whether growth and the ability of STEC to form biofilms in plants extracts could be related to specific plant metabolites or could predict the ability of the bacteria to colonise living plants. Growth rates for sprouted seeds (alfalfa and fenugreek) exhibited a positive relationship between plant extracts and living plants, but not for leafy vegetables (lettuce and spinach). Therefore, the detailed variations at the level of the bacterial isolate, plant species and tissue type all need to be considered in risk assessment.

Importance (149 / 150 word) 21 Fresh produce is an important vehicle for STEC transmission and experimental evidence 22 shows that STEC can colonise plants as secondary hosts, but differences in the capacity to 23 colonise occur between different plant species and tissues. Therefore, an understanding of the 24 impact of these plant factors have on the ability of STEC to grow and establish is required for 25 food safety considerations and risk assessment. Here, we determined whether growth and the 26 ability of STEC to form biofilms in plants extracts could be related to specific plant metabolites 27 or could predict the ability of the bacteria to colonise living plants. Growth rates for sprouted 28 seeds (alfalfa and fenugreek) exhibited a positive relationship between plant extracts and living 29 plants, but not for leafy vegetables (lettuce and spinach). Therefore, the detailed variations at Introduction 34 Contamination of fresh produce from Shigatoxigenic Escherichia coli (STEC) presents a 35 serious hazard as a cause of food-borne illnesses, diarrhoea and enterohemorrhagic disease. 36 Fresh produce is a major vehicle of transmission of STEC, with foods of plant origin accounting 37 for the majority of E. coli and Shigella outbreaks in the USA (50). Fresh produce is often eaten 38 raw or minimally processed and contamination of the produce can occur at any point along the 39 food chain from farm to fork, with major outbreaks e.g. spinach (30) and sprouted seeds (6). although proliferation is well known to be influenced by physio-chemico factors (5, 23, 57, 58), 49 risk assessments for STEC on fresh produce tend to consider plants as a homogenous whole 50 (12, 21, 51). 51 STEC preferentially colonise the roots and rhizosphere of fresh produce plants over leafy 52 tissue and have been shown to internalise into plant tissue, where they can persist in the 53 apoplastic space as endophytes (13, 72). The apoplast contains metabolites, such as solutes, 54 sugars, proteins and cell wall components (54) and as such provides a rich environment for 55 5 many bacterial species, both commensal bacteria and human pathogens (20, 28). The rate of 56 STEC internalisation is dependent on multiple factors including the plant species and tissue 57 (73) and how plants are propagated (17-19). Specificity in the response of STEC to different 58 plant species and tissue types has been demonstrated at the transcriptional level (9, 38). 59 Therefore, there is a need to take into account specificity of the STEC-plant interactions that 60 could impact risk.

79
Growth rate parameterisation 80 To relate growth potential to colonisation of STEC in fresh produce plants, in vitro growth rates 81 were first measured in plant extracts. Representative edible species associated with food-82 borne outbreaks were used: two leafy greens (lettuce, spinach) and two sprouted seeds 83 (fenugreek, alfalfa). Plant tissues used were to represent edible, non-edible and internalised 84 tissues of the leafy greens from total lysates of leaves or roots, and apoplastic washing 85 recovered from leaves, respectively, while total sprout lysates were used to represent edible 86 sprouts. A panel of five E. coli was assessed (Table 1)  apoplastic extract at all temperatures tested (Fig. 1B). All isolates grew consistently faster in 101 fenugreek sprout extracts than in alfalfa, and either sprout extract supported faster growth than 102 defined medium (RDMG) (Fig. 1C) fenugreek sprout extracts, followed by lettuce apoplast and lettuce leaf lysates (Table 2).

123
Sucrose was the most abundant sugar in all species and cultivars, except for alfalfa, which had 124 high levels of fructose and glucose. Arabinose was only detected in the apolastic fluid of 125 spinach and lettuce, accounting for 0.36 % and 0.23 % of all sugars, respectively. A two-way 126 ANOVA found significant differences for tissue types (F (7, 60) = 16.5; p < 0.0001).

127
The levels of amino acids and other metabolites were determined from identification of 116 128 polar metabolites, of which 60 were assigned and mapped onto a simplified polar metabolite 129 pathway for plants to visualise metabolite availability for the bacteria (Fig. S2). The abundance 130 ratio of each compound against the internal standard ribitol, generated a response ratio (RR) 131 to allow semi-quantitative comparison (Table S2). Differences occurred between species and 132 tissue types in a similar pattern to the mono-and disaccharides (   The main metabolite groups were then investigated as groups that could influence bacterial log (cfu g -1 ) on alfalfa sprouts and 3 log (cfu g -1 ) on fenugreek sprouts, between 0 and 2 dpi.

207
Viable counts for isolate ZAP1589 were generally lower on both sprouted seeds compared to 208 isolate Sakai, but still reached 6.00 ± 0.253 log (cfu g -1 ) on alfalfa 2 dpi.

209
Internalisation was also assessed since endophytic behaviour is a feature of E. coli O157:H7 log (cfu g -1 )) and fenugreek (1.53 log (cfu g -1 )) on day 1, and isolate ZAP1589 in alfalfa (1.87 216 log (cfu g -1 )) on day 2. The prevalence was 7.1 % (1/14 samples positive), although the viable 217 counts were close to the limit of detection by direct plating. Therefore, internalisation of E. coli 218 O157:H7 isolates Sakai and ZAP1589 appeared to be a rare event on sprouted seeds, 219 although they colonised the external sprout tissue to higher levels than on lettuce or spinach.

220
Correlating in planta colonisation with plant extract growth rate kinetics 221 To relate growth kinetics in extracts with in planta growth, growth rates were estimated for in 222 planta growth. This was possible for sprouted seeds since colonisation levels increased over 223 time (Fig. 4). Alfalfa plants supported significantly faster growth rates for both E. coli O157:H7 224 isolates compared to fenugreek, at 2.23 ± 0.213 log cfu g -1 per day (R 2 = 0.720) and 1.50 ± 225 0.0913 log cfu g -1 (R 2 = 0.863) for Sakai on alfalfa and fenugreek sprouts, respectively, and for 226 isolate ZAP1589, rates of 2.24 ± 0.159 log cfu g -1 (R 2 = 0.822) and 0.710 ± 0.116 log cfu g -1 (R 2 227 = 0.464) per day on alfalfa and fenugreek sprouts, respectively. The difference in growth rate 228 between the isolates on fenugreek sprouts was significant (p < 0.0001). Although in planta 229 growth rates for E. coli isolates Sakai were estimated on spinach tissues (leaves, roots or 230 internalised in leaf apoplast) or lettuce (leaves, roots) from low inoculation dose (10 3 cfu ml -1 )

231
(73) these were non-significant since growth over the 10 day period was minimal or completely 232 constrained, with a high degree of plant-to-plant variation. Growth rate estimates were not 233 13 made when a high starting inoculum was used since the colonisation levels decreased over 234 time (Fig. 4).

235
Comparison of the in planta and extract growth rate estimates were made for both E. coli 236 O157:H7 isolates on sprouted seeds (at 25 C) or in spinach and lettuce (at 18 C) (Fig. 5). A 237 positive correlation occurred for growth rate estimates in the sprouted seeds (R 2 = 0.516), 238 although this was not significant. Since in planta growth in spinach or lettuce tissues was 239 minimal, there was no correlation with growth rates in corresponding extracts. Therefore, the 240 restrictions in bacterial growth that occurred with living plants meant that growth rates in 241 extracts could not be extrapolated to in planta growth potential for leafy vegetables, but did 242 bear a positive relationship for sprouted seeds. The potential for food-borne bacteria to grow in fresh produce food commodities is a key 246 consideration in quantitative risk assessment. Factors that influence bacterial growth are the 247 plant species and tissue, the bacterial species or isolate, and the surrounding environment.

248
The growth potential of a bacterial population consists of proportion of the growing sub- show an average as high as ~ 1000 mg / 100 g fresh weight (47)  In conclusion, growth potential in planta was described in part, by growth rates in plant   'sprout extract' media was generated by adding each group of constituents (Table 3)   diluted to OD 600 of 0.02 (equivalent to 10 7 cfu ml -1 ) in SDW, which partially submerged pots.

425
After 1 h inoculation, the pots were transferred to the growth chamber until sampling. Sprouts  Tables and Figures   678   Tables   679   Table 1 Bacterial isolates used in this study 680 ST = sequence type, Stx = Shiga toxin presence, nd = not determined, n/a = not applicable.

681
Isolate Sakai used here is the stx-inactivated derivative (10