High-resolution analysis of spatiotemporal virulence gene regulation during food-borne infection with Escherichia coli O157:H7 within a live host

Food-borne infection with enterohemorrhagic Escherichia coli (EHEC) is a major cause of diarrheal illness in humans, and can lead to severe complications such as hemolytic uremic syndrome. Cattle and other ruminants are the main reservoir of EHEC, which enters the food-chain through contaminated meat, dairy, or vegetables. However, how EHEC transitions from the transmission vector to colonizing the intestinal tract, and how virulence-specific genes are regulated during this transition, is not well understood. Here, we describe the establishment of a vertebrate model for food-borne EHEC infection, using the protozoan Paramecium caudatum as a vector and the zebrafish (Danio rerio) as a host. At 4 days post fertilization, zebrafish have a fully developed intestinal tract, yet are fully transparent. This allows us to follow intestinal colonization, microbe-host cell interactions, and microbial gene induction within the live host and in real time throughout the infection. Additionally, this model can be adapted to compare food- and water-borne infections, under gnotobiotic conditions or against the backdrop of an endogenous (and variable) host microbiota. Finally, the zebrafish allows for investigation of factors affecting shedding and transmission of bacteria to naïve hosts. High-resolution analysis of EHEC gene expression within the zebrafish host emphasizes the need for tight transcriptional regulation of virulence factors for within-host fitness. IMPORTANCE Enterohemorrhagic Escherichia coli (EHEC) is a food-borne pathogen which can cause diarrhea, vomiting and in some cases, severe complications such as kidney problems in humans. Up to 30% of cattle are colonized with EHEC, which can enter the food-chain through contaminated meat, dairy and vegetables. In order to control infections and stop transmission, it is important to understand what factors allow EHEC to colonize its hosts, cause virulence and aid transmission. Since this cannot be systematically studied in humans, it is important to develop animal models of infection and transmission. We developed a model which allows us to study food-borne infection in zebrafish, a vertebrate host that is transparent and genetically tractable. Using the zebrafish host, we can follow the bacterial infection cycle in real time, and gain important information regarding bacterial physiology and microbe-host interactions. This will allow us to identify potential new targets for infection control and prevention.


INTRODUCTION 46
Enterohemorrhagic Escherichia coli (EHEC) are a major cause of food-borne infections worldwide. EHEC are transmitted through consumption of water, meat, dairy or vegetables 48 contaminated with fecal matter, or hand-to-mouth, which is common in school and nursery 49 settings. EHEC infection usually presents with bloody diarrhea, vomiting and stomach cramps, 50 but in rare cases can lead to hemolytic uremic syndrome (HUS), a severe clinical complication 51 resulting in kidney damage and often life-long morbidity, or mortality. Antibiotics are contra-52 indicated, since antibiotic treatment can increase toxin production and exacerbate toxin-mediated 53 disease pathology. Depending on geographical location, up to 30 % of cattle are colonized by 54 EHEC, which presents a considerable environmental reservoir. One of the main virulence factors 55 associated with colonization of ruminants as well as human hosts is the locus of enterocyte 56 effacement (LEE), a horizontally acquired pathogenicity island encoding for a type 3 secretion 57 system (T3SS). The LEE also encodes for the adhesion intimin and its host translocated receptor 58 Tir (translocated intimin receptor), which in volunteer studies with the closely related 59 enteropathogenic E. coli (EPEC) have shown to be a key factor for the development of diarrheal 60 symptoms (1). 61 Ongoing studies of EHEC focus on understanding how the LEE is regulated during the EHEC 62 life cycle, and how the LEE encoded genes and other virulence factors, such as Shiga toxin 63 (STx) contribute to colonization, disease pathogenesis and transmission. Another area of interest 64 is how a host's endogenous microbiota interacts with EHEC, and how this affects host fate 65 following EHEC ingestion. Several infection models exist to study EHEC virulence factors, most 66 notably pigs, rabbits and mice. And although no single model host is capable of reproducing the 67 full clinical presentation of human EHEC infection, each has its own distinct advantages and 68 Spatiotemporal in vivo analysis of EHEC virulence gene regulation Following two hours of co-incubation with EHEC, P. caudatum were transferred to medium without bacteria.

140
Food-borne E. coli colonize larval zebrafish more efficiently than water-borne EHEC. 141 Initially, colonization by water-borne EHEC was characterized in gnotobiotic zebrafish larvae, 142 which were acquired from bleached eggs and reared under sterile conditions as previouly 143 described (18). EHEC were administered to larval zebrafish at 4 dpf, at which point larvae had a 144 fully developed intestinal tract with a functional anal opening. E. coli constitutively expressing 145 mCherry were used to visualize colonization in vivo. Bacterial burden was initially assessed 146 following 2 hours of exposure, and increased dependent on the amount of ingested EHEC (Fig.  147 2A). Colonization was mostly localized to the mid-intestine (Fig. 2B), and accumulation and 148 secretion of EHEC-containg fecal matter was observed soon after ingestion (Fig. 3). 149 Next, paramecia loaded with EHEC Sakai:mCherry were administered to larval zebrafish at 4 150 dpf, at which point larvae were able to swim freely, and prey on paramecium ( Figure S2 and 151 Movie S1). To administer a consistent number of E. coli to zebrafish, paramecia were kept in co-152 culture with E. coli at a ratio of 1:500, gently washed to remove extracellular E. coli and added 153 Spatiotemporal in vivo analysis of EHEC virulence gene regulation to fish medium to result in a defined initial concentration as described above, and offered to 154 zebrafish as a prey. 155 156 Figure S2. Stills from Movie S1, showing a zebrafish larva at 4 dpf preying on P. caudatum loaded with EHEC 157 Sakai:mCherry.

159
Bacterial burden was determined in fish exposed to matched doses of food-borne or water-borne 160 EHEC, following two hours of exposure. At all three doses tested, colonization levels were 161 approximately 10 times higher following food-borne infection ( Zebrafish larvae were exposed to 10 8 CFU/mL of food-borne (magenta) or water-borne (purple) EHEC for two 12 Spatiotemporal in vivo analysis of EHEC virulence gene regulation hours, transferred to fresh sterile medium, and bacterial burden was determined by dilution plating on EHEC 179 selective agar at indicated time points. Individual data points, means (n=10 for each condition) and stdev are shown.  Although LEE1:gfp activity relative to the constitutively active mCherry was enhanced for tolA 272 compared to wild type bacteria within zebrafish, the bacterial burden of the tolA mutant was 273 significantly reduced compared to the wild type ( Fig. 4 and Fig. 6A-F). This was not due to 274 Spatiotemporal in vivo analysis of EHEC virulence gene regulation altered growth or degradation within the P. caudatum vector (Fig. 6G, H). Overall, these data 275 suggest that the tight regulation of T3SS expression is essential for within-host fitness. colonization, we asked if EHEC ingestion would cause mortality in zebrafish. We infected 290 zebrafish at 4 dpf with a dose of 10 9 CFU/ml of food-borne EHEC or left them uninfected 291 (control). Fish were assessed for vital signs (movement, heart beat and circulation) daily for a 292 total of six days post infection (dpi), and survival was analysed using the Kaplan-Meier estimator 293 (Fig. 7). Despite showing similar colonization levels ( Fig. 5 and 6), Sakai and TUV 93-0 wild 294 type strains displayed significantly different pathogenicity, with a mean survival of 76% for 295 Sakai and 56% for TUV 93-0 at the experimental endpoint. The attenuation of the tolA and adhE 296 mutants observed in terms of intestinal colonization (Fig. 4-6) was also reflected by a reduction 297 in pathogenicity. The tolA mutant was non-pathogenic in the zebrafish model, with 100% host 298 Spatiotemporal in vivo analysis of EHEC virulence gene regulation survival and no observable morbidity up until day 6 post infection. The adhE mutant was 299 similarly attenuated, with a mean host survival of 87% at day 6 post infection (Fig. 7).

308
Zebrafish as a model to study fecal shedding and transmission. While fecal shedding of 309 EHEC is often used as a proxy for bacterial burden in rodents, fecal-oral transmission is rarely 310 studied in these models. The zebrafish model allows simultanous analysis of fecal shedding and 311 fecal-oral transmission from infected fish to naïve recipients in one experiment. AB fish infected 312 with food-borne EHEC for two hours were transferred into fresh media together with a naïve 313 recipient (Fig. 8A). Tg(mpo:gfp) fish were used as recipients to be able to visually distinguish 314 donor and recipient (green fluorescent). EHEC was continutally shed from donor fish following 315 Spatiotemporal in vivo analysis of EHEC virulence gene regulation transfer into fresh media, and levels of shed bacteria increased steadily up until 24 hours post 316 transfer (Fig. 8B). Onward transmission to naïve fish was first observed between 12-24 hours 317 post transfer (Fig. 8C), at which point EHEC:mCherry could be visualized in the foregut and 318 mid-intestine of recipient embryos (Fig. 8D). These data demonstrate that the dynamics of fecal 319 shedding and fecal-oral transmission to naïve hosts can be studied using the zebrafish model. 320 It has been reported that zebrafish acquire a microbiota which rapidly diversifies during early 337 development (18, 36, 37). To test whether EHEC infection can be studied in the zebrafish model 338 against the backdrop of the endogenous microbiota, we compared levels of EHEC in gnotobiotic 339 fish and conventionalized fish, which were transferred into a mixture of E3 and tank water 340 following hatching. Compared to gnotobiotic fish, initial colonization with EHEC was 341 significantly decreased, but not entirely eliminated, in conventionalized fish (Fig. 9A). 342 Surprisingly, colonization levels in conventionalized fish were very consistent, even though the 343 composition as well as levels of the colonizing microbiota differed considerably between 344 Spatiotemporal in vivo analysis of EHEC virulence gene regulation individual animals (Fig. 9B). In conventialized fish, the bacterial burden expanded with 345 increased incubation time, and could easily be visualized from 16 hpi (Fig. 9C). Upon ingestion, EHEC rapidly colonize the mid-intestine, and although bacteria were initially 374 observed both in fore-gut and mid-intestinal tract, EHEC shows a distinct preference for 375 colonizing the mid-intestine, and the site of infection displays sharp boundaries, with temporary 376 colonization of the fore-gut during early infection, and no colonization of the posterior intestine 377 (Fig. 2, 4). In contrast, colonization of infant rabbit is not as localized, with similar bacterial 378 burdens found in ileum, cecum and colon even at later time points (5). 379 Human EHEC infection is known to cause a strong pro-inflammatory response, and neutrophil 380 infiltration of the lamina propria and transmigration through the intestinal epithelium into the gut 381 lumen has been described in monkey, piglet, and rodent models (5, 40, 41). This is a response of 382 increased IL-8 production by the intestinal mucosae, although it has been a point of contention 383 whether Stx or TLR5 recognition of H7 flagellin was the major factor inducing IL-8 secretion. In 384 vitro and ex vivo studies using flagellar mutants demonstrated that IL-8 secretion is driven by 385 exposure of epithelial cells to flagellar antigen, and to some extent, TLR4-driven responses to 386 Spatiotemporal in vivo analysis of EHEC virulence gene regulation LPS (42, 43). In our model, Stx negative EHEC is still capable of mounting a strong neutrophilic 387 inflammation, which supports these data. While neutrophil depletion has been shown to increase 388 the bacterial burden of Citrobacter rodentium in mouse (44) and targeting leukocyte adhesion 389 factor with antibody reduces disease symptoms in rabbits (3), no direct link between neutrophil 390 recruitment and bacterial clearance has been established for EHEC. The zebrafish immune 391 system displays many similarities to that of mammals, with counterparts for most human 392 immune cell types (45). The zebrafish innate immune system starts to develop as early as 24 hpf 393 with primitive macrophages followed by neutrophils at 32-48 hpf. The development of the 394 adaptive immune system lags behind, taking another 4 weeks to fully develop (46). This feature 395 provides an opportunity to exclusively observe the innate immune reaction to an EHEC infection 396 in our experimental system. Real-time imaging of EHEC infected zebrafish allowed us to 397 simultaneously follow and establish a temporal link between neutrophilic inflammation and 398 EHEC persistence in the intestine. We observed a significant increase in neutrophil recruitment 399 to the infection site as early as 2 hpi, with a peak around 8 hpi. After that, the response gradually 400 diminished until the experimental endpoint (12 hpi). This coincided with bacterial burden, which 401 was significantly decreased 3 hpi, and gradually diminished thereafter. However, a low level of 402 persistence was still observed at the experimental endpoint, 24 hpi. Although only 403 circumstantial, the timing of these two events suggests bacterial clearance may, at least in part, 404 be mediated by neutrophils. 405 EHEC infection in infant rabbits is self-limiting, with a peak in bacterial burden approximately 7 406 dpi and a decline in inflammation and bacterial burden thereafter (5). We observe a similar, 407 albeit accelerated pattern in zebrafish, with peak inflammation approximately 8 hpi. Bacterial 408 burden declines over time, and interestingly, there are two distinct colonization patterns: 409

Food-and water-borne fish infections.
For infection experiments bacteria were harvested by 500 centrifugation at 6000 g for 2 min and adjusted to an OD600 of 1.0 ( concentration of 2x10 8 501 bacteria/ml). P. caudatum were quantified using a hemocytometer and added to the suspension to 502 give a concentration of 10 4 paramecia/mL, and incubated for 2 h at 32 °C. Following pre-503 incubation, paramecia were washed and added to zebrafish larvae (4 dpf) housed in 6-well plates, 504 to give the indicated bacterial concentrations. For water-borne infections, EHEC concentrations 505 as indicated were directly prepared in E3 and added to zebrafish larvae (2ml/10 zebrafish 506 larvae/6-well). Following infections, zebrafish larvae were either anaesthetized by adding 507 tricaine (final conc. 160 µg/mL) to 20 mL E3 and 2 mL of a 0.1 M sodium-bicarbonate solution 508 to buffer the medium, or euthanized by adding 1.6 mg/mL tricaine to the buffered medium. For 509 microscopy, larvae were transferred to a 1.5 mL microcentrifuge tube, washed in PBS, fixed by 510 adding 1 mL of a 4 % para-formaldehyde solution in PBS, and stored at 4 °C in the dark until 511 used. 512 513 Imaging of infected fish. For live imaging, infected anaesthetized larvae were positioned in 96-514 well glass-bottom plates and covered and immobilized with 1 % low-melting-point agarose 515 solution. 200 µl E3 containing 160 µg/mL tricaine was added to cover the immobilized larvae. 516 Live imaging was performed at 32 °C and 80 % humidity. A Zeiss Axio Observer inverse 517 microscope equipped with a 10x objective was used for acquisition of 2 fluorescent channels and 518 1 differential interference contrast (DIC) channel. The 4D images produced by the time-lapse 519 acquisitions were processed, clipped, examined and interpreted using the Zen 2 software (Zeiss). 520 Spatiotemporal in vivo analysis of EHEC virulence gene regulation Maximum intensity projection was used to project developed Z-stacks and files were exported in 521 tiff format for images or mov format for QuickTime movies. We thank A. Roe for sharing strain TUV 93-0 and its derivatives. We thank S. Renshaw,S. 536 Johnston and R. Wheeler for sharing Tg(mpo:gfp) eggs. We thank members of the Krachler and 537 Voelz labs for critical reading and comments on the manuscript. 538