Tobacco Seed-Based Oral Vaccination against Verocytotoxic O138 Escherichia coli as Alternative Approach to Antibiotics in Weaned Piglets

Post-weaning diarrhoea and enterotoxaemia caused by Escherichia coli are serious threats in the pig (Sus scrofa domesticus) livestock industry and are responsible for economic losses related to mortality, morbidity and stunted growth. The aim of this study was to evaluate the effect of an engineered tobacco seeds-based edible vaccine in O138 Escherichia coli-challenged piglets throughout a multidisciplinary approach. Thirty-six weaned piglets were enrolled and randomly divided into two experimental groups, a control (C; n = 18) group and a tobacco edible vaccination group (T, n = 18), for 29 days of trial. At days 0, 1, 2, 5 and 14, piglets of the T group were fed with 10 g of the engineered tobacco seeds line expressing F18 and VT2eB antigens, while the C group received wild-type tobacco seeds. After 20 days, 6 piglets/group were orally challenged with the Escherichia coli O138 strain (creating four subgroups: UC = unchallenged control, CC = challenged control, UT = unchallenged tobacco, CT = challenged tobacco) and fed with a high protein diet for 3 consecutive days. Zootechnical, clinical, microbiological, histological and immunological parameters were assayed and registered during the 9 days of post-challenge follow up. At 29 days post-challenge, the CT group displayed a lower average of the sum of clinical scores compared to the CC group (p < 0.05), while the CC group showed a higher average sum of the faecal score (diarrhoea) (p < 0.05) than the CT group. A decreased number of days of shedding of the pathogenic strain was observed in the CT compared to the CC group (p < 0.05). Specific anti-F18 IgA molecules were significantly higher in the CT group compared to the CC group’s faecal samples during the post-challenge period (p < 0.01). In conclusion, edible vaccination with engineered tobacco seeds showed a protective effect on clinical symptoms and diarrhoea incidence during the post-challenge period, characterized by a limited time of pathogenic strain shedding in faeces.


Introduction
The weaning phase is critical for pig livestock, as it characterises the transition of piglets from lactation to self-sufficiency in feeding [1]. In particular, during the first two weeks after weaning, piglets are exposed to several stressors that could contribute to the occurrence of multifactorial diseases commonly associated with Escherichia coli infection [2,3]. Among Escherichia coli pathotypes categorized by the presence of virulence factors are verocytotoxinproducing E. coli (VTEC) O138, O139 and O141 strains, which cause important direct and The clinical scores showed a significant increase in the sum of epiphora, oedema, vitality, depression, hair and faecal scores during the post-challenge period (p < 0.05; Table  1). The average sum of the epiphora score was significantly higher in the CC group compared to the UC and UT groups (CC: 2.83 ± 0.35; UC 1.33 ± 0.50; UT: 0.50 ± 0.50; p < 0.01), while the CT group did not reveal statistical differences during the period from 20 to 23 days. The average sum of epiphora was significantly higher in the CC group compared to the UT and CT groups during the post-challenge period (CC: 8.38 ± 1.06; UT: 1.75 ± 1.51; CT: 4.00 ± 0.06; p < 0.01). The average sum of the oedema and hair scores was significantly higher in the CC group compared to the UT and CT groups from days 20 to 23 post-challenge (p < 0.01). The average of the oedema scores was increased in the CC group com- The results are presented as least squares means ± standard error.

Microbiological Analyses
Before the experimental infection procedures, no positivity was detected for the presence of O138 E. coli strains in the pigs' faeces. Faecal shedding of O138 E. coli was recorded from 21 to 25 days in the infected groups (CC and CT; Figure 2). In particular, the prevalence of faecal shedding of O138 E. coli was significantly higher at 23 days in the CC group compared to the CT group (33.33% and 0.00%, respectively; p < 0.01). The CC group showed a significantly longer period of prevalence of challenge strain faecal shedding. In particular, 33% of the piglets showed faecal shedding for 4 consecutive days, while no piglets showed more than 3 days of shedding of O138 E. coli in the CT group (p < 0.01). Σ3: sum of the individual daily scores from days 20 to 23 (n = 36); Σ9: sum of the individual daily scores from days 20 to 29 (n = 24). The comparisons were performed using the Kruskal-Wallis test and the means were separated with the Wilcoxon rank sum test for comparing independent groups. The results are presented as least squares means of the clinical scores sum for each animal per period ± standard error. * The asterisk indicates parameters in which a statistically significant difference was recorded. a-b Different lowercase letters indicate statistically significant differences (p < 0.05). UC: unchallenged control group, UT: unchallenged treatment group, CC: challenged control group, CT: challenged treatment group.

Microbiological Analyses
Before the experimental infection procedures, no positivity was detected for the pres ence of O138 E. coli strains in the pigs' faeces. Faecal shedding of O138 E. coli was recorded from 21 to 25 days in the infected groups (CC and CT; Figure 2). In particular, the preva lence of faecal shedding of O138 E. coli was significantly higher at 23 days in the CC group compared to the CT group (33.33% and 0.00%, respectively; p < 0.01). The CC group showed a significantly longer period of prevalence of challenge strain faecal shedding. In particular, 33% of the piglets showed faecal shedding for 4 consecutive days, while no piglets showed more than 3 days of shedding of O138 E. coli in the CT group (p < 0.01).

Histological Evaluations
The histological findings in the ileum did not reveal statistical differences among the experimental groups at 23 days (Table 2; Figure 3). Even if infiltrates of lymphocytes were found to be significant at 29 days, only the CC group tended to have increased levels compared to the UC group (p < 0.09). The expression of CD3 in the epithelium was significantly different among the experimental groups after 29 days (p = 0.0396). However, pairwise comparisons did not highlight any significant difference. Histology and immunohistochemistry showed significant differences within the experimental groups when comparing 23 to 29 days of the experimental period (Table 3; p < 0.05). In particular, infiltrated lymphocytes and CD3 in the lamina propria were decreased after 29 days as compared to 23 days in the UC group (p = 0.05). Secretory IgA(SIgA) levels in the luminal surface were elevated at 23 days compared to 29 days in the UT group (2.00 ± 0.00 and 0.00 ± 0.00, respectively; p = 0.05) Follicular hyperplasia was significantly higer in the CC group at 23 days compared to 29 days (0.50 ± 0.16 and 0.00 ± 0.11, respectively; p < 0.05).

Immune Response Parameters
Specific anti-F18 IgAs showed a significant increase in the UT and CT immunized groups compared to the CC group (p < 0.05) ( Table 4). Relative expression measured by real-time PCR and ELISA quantification did not reveal statistical differences for MHC-I, MHC-II, IFN-γ, IL-1β, TLR2, TLR4, serum IgA, IgA in an intestinal scrape, CXCL9 in an intestinal scrape, TNF-α in an intestinal scrape, IL-8 in an intesinal scrape or IL-1 in an intestinal scrape between the groups and comparing day 3 with day 9 post-challenge (Tables 5 and 6). Infiltrates of lymphocytes, epithelial regeneration, fusion of villi, oedema, T-atrophy, stroma and follicular hyperplasia were semi-quantitatively scored (0 = absent, 1 = slight, 2 = moderate, 3 = severe). CD3, CD20 and Iba1 immunoreactions were semi-quantitatively scored in the lamina propria (0 = absent, 1 = rare cells, 2 = some cells, 3 = numerous cells, 4 = very abundant cells). For IgG and IgA, the immunoreaction was semi-quantitatively scored in the epithelial luminal surface and lamina propria (0 = absent, 1 = slight, 2 = moderate, 3 = strong). The comparisons were performed using the Kruskal-Wallis test and the means were separated with Wilcoxon rank sum test for comparing independent groups. The results are presented as least squares means ± standard error. * The asterisk indicates parameters in which a statistically significant difference was recorded. a-b Different lowercase letters indicate statistically significant differences (p < 0.05). UC: unchallenged control group, UT: unchallenged treatment group, CC: challenged control group, CT: challenged treatment group.

Immune Response Parameters
Specific anti-F18 IgAs showed a significant increase in the UT and CT immunized groups compared to the CC group (p < 0.05) ( Table 4). Relative expression measured by real-time PCR and ELISA quantification did not reveal statistical differences for MHC-I, MHC-II, IFN-γ, IL-1β, TLR2, TLR4, serum IgA, IgA in an intestinal scrape, CXCL9 in an   The results are presented as least squares means of optical densities ± standard error. The comparisons were performed using the Kruskal-Wallis test and the means were separated with the Wilcoxon rank sum test for comparing independent groups. * The asterisk indicates parameters in which a statistically significant difference was recorded. a-b Different lowercase letters indicate statistically significant differences (p < 0.05). UC: unchallenged control group, UT: unchallenged treatment group, CC: challenged control group, CT: challenged treatment group.

Discussion
This study aimed to exploit an engineered tobacco seeds-based oral vaccine against VTEC O138 E. coli in weaned piglets. The animals enrolled in the study were susceptible to E. coli strains harbouring F18 fimbriae as proven by screening for FUT1 polymorphism.
The pigs in the vaccinated group were fed with 10 g of F18 (7 µg of F18 protein) and 10 g of VT2e-B (containing about 35 µg of VT2e-B), while the animals in the control group received 20 g of wild-type tobacco seeds.
The zootechnical performance did not show statistical differences among the challenged and unchallenged groups during the entire experimental period.
A clinical evaluation revealed significant differences between the groups after the experimental infection. The typical symptoms related to O138 E. coli infections are mostly caused by bacterial verotoxins and F18 adhesive fimbriae, which can impair the vascular endothelium of the small intestine, subcutis and brain and eventually cause subcutaneous oedema and neurological symptoms, such as convulsion, rear-leg ataxia, extensor rigidity, lateral recumbency, tremors, paralysis, dyspnea, opisthotonos and sudden death [16]. The infected groups showed a broad spectrum of clinical symptoms, such as epiphora, oedema and loss of vitality due to the E. coli challenge.
In particular, the average sum of the epiphora scores was found to be higher in the CC group compared to the UT group 3 days post-challenge. At 9 days post-infection, the average of the sum of the epiphora scores was increased in the CC group compared to the UT and CT groups. The average sum of the oedema scores was lower in the UT and CT groups compared to the CC group during the first three and nine days post-infection. The average of sum of the vitality scores registered a higher value (2 = bad) in the CC group compared to the other experimental groups three days post-challenge. Depression registered decreased scores in the UC and UT groups compared to the CC group three days post-challenge and at nine days, the CC group showed the highest values compared to the other groups.
The average sum of the hair scores revealed increased values in the CC group compared to the UC and UT groups 3 days post-infection. The piglets in the CC group displayed the highest values for hair scores when considering the nine days post-challenge.
The CC group expressed a plethora of clinical symptoms (specific and nonspecific) related to VTEC infection, displaying impaired health conditions during the 9 days postinfection. The CT group revealed values that were comparable to the unchallenged groups, suggesting the preventive effect of the oral vaccination that was able to limit the detrimental effect of the VTEC infection on animal health. In line with the previous study [15], the clinical evaluation allowed us to identify a panel of clinical symptoms related to VTEC infection that are useful in assessing the effectiveness of an edible vaccination.
The faecal score showed increased frequency of diarrhoea in the CC group compared to the CT group during the first 3 days post-challenge. Diarrhoea is the most evident clinical symptom related to VTEC infection and is caused by an alteration of the electrolyte homeostasis, contributing to fluid losses [17]. The results obtained indicate that oral vaccination with tobacco seeds limited diarrhoea occurrence during the first days after infection, underlining the role of vaccination as an intervention to reduce the need for antibiotic treatment [18].
The presence of O138 E. coli infection was verified by the faecal shedding of the pathogenic challenge strain. The CC and CT groups showed the presence of the pathogen, thus confirming the success of the experimental challenge procedure. However, faecal shedding of the O138 E. coli strain was significantly lower in the CT group compared to the CC group 3 and 4 days post-infection. These findings suggest that oral vaccination significantly reduced the number of days of faecal shedding of pathogenic strains. Such reduction of both pathogen excretion and shedding duration should be considered as a successful result of an alternative approach to the use of antibiotics [19].
A histological assessment of the intestinal sections showed significantly higher infiltrates of lymphocytes in the CC group compared to the UC group after 29 days of the trial. The detection of lymphocytic infiltrate indicates the presence of a triggering stimulus able to induce the pathological process. It is well known that enteric pathogens can induce an inflammatory response in the gut [20]. The CC group showed higher follicular hyperplasia compared to the UC group 9 days post-challenge. Follicular hyperplasia is generally the result of accumulation of plasma-cell in lymph nodes and is often characterized by a significant enlargement in one or several segments of the gastrointestinal tract or by a similar alteration of the lymphoid nodules of the Peyer's patches of the distal part of the small intestine [21]. In general, intestinal bacterial infections cause an increased number of lymphoid follicles [22]. The detected higher follicular hyperplasia in the CC group after 23 days of the trial might be due to virulence factors of the challenger strain that triggered the immune and inflammatory response, which was also confirmed by the highest cases of diarrhoea recorded in the same period. Nevertheless, the CT group showed comparable levels of lymphocytic infiltrate and follicular hyperplasia with the unchallenged groups, suggesting that the tobacco vaccination limited the detrimental effect of the challenge strain by regulating the intestinal inflammatory status. The piglets in the UC group displayed decreased levels of infiltrates of lymphocytes and CD3 in the lamina propria at 29 days compared to 23 days post-challenge.
The evaluation of specific anti-F18 IgAs showed that the vaccinated groups developed a higher concentration of immunoglobulins A compared to the CC group. In particular, the UT group displayed higher levels of specific anti-F18 IgAs compared to the CC groups. In addition, the UT group showed an increased presence of IgAs in the luminal surface at 23 days compared to 29 days post-challenge. A tendency toward the same trend was found in the UT group for IgAs in the intestinal crypts. Immunoglobulin A (IgA) is the predominant antibody isotype in the mucosal immune system, which widely exists in the gastrointestinal tract, respiratory tract, vaginal tract, tears, saliva and mammary gland [23].
This form of immunoglobulin can be secreted by different mucosal tissues in the body, and they can prevent initial pathogen attachment to the epithelium and subsequent entry [24]. The observed increase in specific anti-F18 IgAs and SIgAs in the luminal surface in the vaccinated groups compared to the CC group suggest that oral vaccination stimulated a protective response against VTEC infection. Mucosal surfaces serve as the primary portal of entry for most pathogens. The mucosal vaccination by edible vaccines provides a safe and efficient mechanism to induce mucosal and systemic immunity against pathogens [25]. Our findings are in agreement with those reported by other authors that indicate that oral vaccination mainly increases the secretory immunoglobulin A concentrations, promoting the local immune response [26][27][28]. The data obtained with this study indicate that oral vaccination with 10 g of F18 engineered tobacco seeds containing 7 µg of recombinant antigen may trigger the specific immune response by increasing the production of specific anti-F18 IgAs.
Compared to a previous study, the oral vaccination with tobacco seeds resulted in effective lowering of the clinical scores, decreasing of the faecal score and O138 E. coli shedding duration after the experimental infection [12]. In addition, the engineered tobacco seeds stimulated the immune response by increasing the concentrations of specific anti-F18 IgAs and SIgAs in the luminal surface in the vaccinated groups, indicating that oral vaccination efficiently stimulated the immune system against O138 E. coli.
In conclusion, our findings encourage scientific studies aimed to develop innovative prophylaxis tools based on engineered plants. Edible vaccines are interesting and may become fundamental for the sustainable development of a livestock system aimed at improving animal health and reducing the use of antibiotics.

Production and Characterization of Engineered Tobacco Seeds
In this study, three lines of tobacco seeds (Nicotiana tabacum L., cv. Xanthi) were considered: (i) engineered seeds expressing Fed-A, the major subunit of the F18 adhesive fimbriae of verocytotoxic O138 E. coli strain (F18s); (ii) engineered seeds expressing the B subunit of the verocytotoxin type 2 (VT2e) toxin of the verocytotoxic O138 E. coli strain (VT2eBs); (iii) wild-type tobacco seeds as controls (WTs). The engineered tobacco seeds were previously obtained by stable transformation of tobacco plants via Agrobacterium tumefaciens with chimeric constructs containing structural parts of bacterial genes under control of a seed-specific GLOB promoter [29]. The three lines of the experimental seeds were harvested from tobacco plants that were cultivated separately in a greenhouse under a polyethylene monofilament to avoid cross-pollination and to provide protection against parasites (Orto Botanico "G. Emilio Ghirardi", University of Milan, Toscolano Maderno, Italy).
The tobacco seeds were evaluated in order to verify the morphology and germination characteristics. Additionally, the presence of antigenic bacterial genes in the engineered seeds and the absence of the exogenous genes in the wild-type lines were verified by PCR using specific primer pairs (Table 7). The PCR assays were performed in 25 µL volume using 5 µL of a DNA template, 0.2 mM of deoxynucleotide triphosphates, 2.5 µL of 10× PCR buffer II, 3 mM MgCl 2 , 1 µM concentrations of each forward and reverse primer and 1.25 U of AmpliTaq Gold DNA Polymerase (Perkin Elmer, Norwalk, CT, USA). The thermal cycling conditions were set up as follows: initial denaturation at 95 • C for 5 min, followed by 35 cycles of denaturation at 95 • C for 1 min, annealing at 56 • C for 1 min and 20 s, elongation at 72 • C for 1 min and 30 s and a final extension at 72 • C. The total proteins were obtained from mature transformed tobacco seeds (from each line) by homogenization with liquid nitrogen in a mortar and extracted with the extraction buffer (50 mM Tris, pH 8.5 mM EDTA, 200 mM NaCl, 0.1% Tween 20). The seed expression of the major subunit Fed-A adhesive fimbriae was evaluated in F18s by indirect ELISA using anti-F18 antibody produced by GenScript (GenScript 860 Centennial Ave. Piscataway, NJ, USA) and purified F18 antigens [30] as a positive control. The seed expression of the B subunit of the VTe2 toxin was evaluated in the VTe2Bs by indirect ELISA using anti-VT2e-B polyclonal antibodies purchased from Plantechno s.r.l (Casalmaggiore, CR, Italy) and VT2eB protein expressed by the pET-system (Novagen) in E. coli BL21 as described by Rossi et al. [29].

Animals and Housing Conditions
Thirty-six piglets (Landrace × Large White) were enrolled from a conventional herd free from diseases according to the ex-A-list of the World Organization for Animal Health, and from Aujeszky's disease, atrophic rhinitis, transmissible gastroenteritis, porcine reproductive and respiratory syndrome and salmonellosis, without history of post-weaning diarrhoea (PWD) and oedema disease (OD) and after a bacteriological analysis of faeces negative for hemolytic Escherichia coli.
Piglets susceptible to verocytotoxic O138 E. coli were selected for the polymorphism of a 1,2 fucosyltransferase (FUT1) gene by PCR performed on genomic DNA extracted from duodenum tissue using the Wizard ® Genomic DNA Purification Kit (Promega Italia S.r.l., Milan, Italy), according to previous study [31].
The piglets, homogeneous for weight, age and sex (50% male, 50% female), weaned at 21 ± 2 days (to ensure the animals' sensitivity upon experimental infection) were transported to the Experimental Animal Research and Application Centre (pigs sector) of the University of Milan in Lodi.
The animals, identified by auricular tag, were housed in individual pens (100 × 50 cm) under controlled conditions (27-29 • C; 60% of relative humidity) with ad libitum access to water, feed and environmentally safe chewable enrichment materials.
The animals were fed basal diets in line with the nutritional requirements of weaned piglets [32] without antibiotics, with 50 ppm of zinc provided by Ferraroni S.p.A (Bonemerse CR, Italy). From the day of the challenge until the second day after the challenge, the same antimicrobial-free diet, containing 27% of crude protein on dry matter, was administered to all the experimental groups, according to Rossi et al. [15].
The experimental diets and tobacco seeds were analysed for the principal nutrient components following the official method of analysis of the Association of Analytical Chemists [33]. In particular, the following parameters were determined: dry matter (DM; method 930.15), crude protein (CP; method 2001.11); ether extract (EE; method 2003.05); crude fiber (CF; method Ba 6a-05) by the filtering bags technique, following the official method of the American Oil Chemists Society [34]; and ash (method 942.05). The composition of the diets is reported in Table 8.
The piglets were randomly divided into two experimental groups: the control group (C; n = 18) and the treatment group (T; n = 18).

Experimental Design and Treatments
The study was divided into three phases: (i) oral immunisation (days 0-20): administration of the engendered seeds (vaccine) by oral route on days 0, 1, 2, 5 and 14; The tobacco seeds were ground with a mill to obtain a uniform powder using a 0.75 mm grid under controlled temperature (max 50 • C). During the immunization phase (on days 0, 1, 2, 5 and 14), the tobacco seeds were administered with a commercial milk replacer powder (LACTOSOL) by mixing in a 1:2 ratio (w/w) in order to guarantee complete consumption. The specific pig milk replacer powder was composed of dairy products, vegetable oils, cereals, sugars and flavouring, containing 20.5% crude proteins.
The T group received Nicotiana tabacum seeds (F18s + VT2e-Bs) as a vaccine on days 0, 1, 2, 5 and 14 by feeding 10 g of F18s and 10 g of VT2e-B mixed with 40 g of milk replacer powder. On the same experimental days, the C group received 20 g of wild-type milled tobacco seeds mixed with 40 g of milk powder.

Challenge
Six days after a booster immunization, on day 20, 24 piglets (12 animals per group) were orally challenged with the O138 Escherichia coli strain, which was provided by the Experimental Zooprophylactic Institute of Lombardy and Emilia Romagna (IZSLER).
One hour before the challenge, the piglets were intramuscularly injected with azaperone (2 mL/pig, Stresnil TM , Janssen Cilag SpA, Milan, Italy) and 30 mL of a 10% NaHCO 3 solution was orally delivered to increase the challenge strain viability. After 10-15 min, a single dose of 5 mL of 10 10 CFU/mL of the O138 E. coli challenge strain was given through an intragastric catheter (16 G), according to Rossi et al. [15]. The animals fasted 3 h before and 3 h after challenge.
For each group, six animals were not subjected to the challenge and were separated from the infected pigs by applying biosafety barriers (two empty pens for the separation), changing gloves, boot covers and plastic coats during the passage of operators from the unchallenged to the infected groups. The animals were then divided into four experimental groups for the post-challenge phase as unchallenged control (UC), challenged control (CC), unchallenged treatment (UT) and challenged treatment (CT; Figure 4). placer powder (LACTOSOL) by mixing in a 1:2 ratio (w/w) in order to guarantee complete consumption. The specific pig milk replacer powder was composed of dairy products vegetable oils, cereals, sugars and flavouring, containing 20.5% crude proteins.
The T group received Nicotiana tabacum seeds (F18s + VT2e-Bs) as a vaccine on days 0, 1, 2, 5 and 14 by feeding 10 g of F18s and 10 g of VT2e-B mixed with 40 g of milk replacer powder. On the same experimental days, the C group received 20 g of wild-type milled tobacco seeds mixed with 40 g of milk powder.

Challenge
Six days after a booster immunization, on day 20, 24 piglets (12 animals per group were orally challenged with the O138 Escherichia coli strain, which was provided by the Experimental Zooprophylactic Institute of Lombardy and Emilia Romagna (IZSLER).
One hour before the challenge, the piglets were intramuscularly injected with azap erone (2 mL/pig, Stresnil TM , Janssen Cilag SpA, Milan, Italy) and 30 mL of a 10% NaHCO solution was orally delivered to increase the challenge strain viability. After 10-15 min, a single dose of 5 mL of 10 10 CFU/mL of the O138 E. coli challenge strain was given through an intragastric catheter (16 G), according to Rossi et al. [15]. The animals fasted 3 h before and 3 h after challenge.
For each group, six animals were not subjected to the challenge and were separated from the infected pigs by applying biosafety barriers (two empty pens for the separation) changing gloves, boot covers and plastic coats during the passage of operators from the unchallenged to the infected groups. The animals were then divided into four experi mental groups for the post-challenge phase as unchallenged control (UC), challenged con trol (CC), unchallenged treatment (UT) and challenged treatment (CT; Figure 4).

Zootechnical and Clinical Evaluation and Sampling
All the piglets were individually weighed weekly during the entire experimental period and twice a week during the post-challenge period (25 and 29 days); the feed intake (FI) was measured on a daily basis by weighing the feed refuse per pen (experimental unit for FI evaluation). In the same period, faecal consistency was evaluated daily through a four-point scale: 0 = normal faeces; 1 = soft consistency 2 = mild diarrhoea; 3 = severe diarrhoea [35].
Blood samples were collected from the jugular vein of each animal to determine the haematocrit value at 0, 7, 14, 21, 25 and 29 days. During the pre-challenge period, faecal samples were taken from the rectum weekly on 0, 7, 14 and 21 days to determine the total and specific IgA levels, as described below.
During the post-challenge period, faecal samples were collected in order to evaluate the shedding of the challenge strains. A rectal temperature was recorded daily. Clinical signs of VTEC infection were registered daily according to the adopted translational model [15]. In particular, palpebral oedema, epiphora, vitality, respiratory and neurological problems were monitored daily and scored using specific point scales. Epiphora is an overflow of tears due to the inflammation and the reduced motility of the eyelid or of the lacrimal pump and was scored as follows: 0 = normal; 1 = mild (moderate presence of eye discharge material in the corner of the eye); 2 = severe (abundant presence of brown discharge material in the corner of the eye). Palpebral oedema is defined as a pathological condition showing an accumulation of fluids in the inner part of the eyelids with different degrees of severity according to the adopted scoring: 0 = normal; 1 = mild (puffy eyelids giving a sleepy appearance); 2 = severe (prominent and closed eyelids). Vitality and depression were defined as disturbed behaviour or slow reactions, an unsteady and slow gait whilst walking and an inattentive response when encouraged to move. Vitality score: 0 = good; 1 = loose (failure to react to stimulus); 2 = bad (slow response to stimuli); depression score: 0 = high (lethargy); 1 = mild (slow reactions); 2 = normal status; perineal area: 0 = clean; 1 = smear; 3 = smear with flogosis (red area caused by inflammation); respiratory score: 0 = normal; 1 = slightly quick; 2 = quick; neurological score: 0 = normal; 1 = mild symptoms (incoordination); 2 = severe symptoms (lateral position, paddling limbs, central nervous symptoms); hair score: 0 = regular (smooth, clean, flat and uniform); 1 = slightly irregular (fuzzy hair coat and/or scaly skin); 2 = irregular (bald patches, or a rough, dull, uneven coat and reddened skin).
At day 3 post-challenge (23 days), 12 piglets (CC = 4, UC = 2, TC = 4, UT = 2) were euthanized for post-mortem examination; the other animals were sacrificed at 29 days (9 days post-challenge). Jejunum, mesenteric lymph nodes and intestinal scrapes were collected and stored in RNA Later (Quiagen, Hilden, Germany) for gene expression analysis. A part of the ileum samples was frozen for immunochemistry, and another one was fixed in 10% neutral buffered formalin for immunohistochemistry and histological evaluations. The samples of the ileum were collected and sealed with sterile cotton ligatures, 3 cm apart from the proximal and distal ends and used for microbiological analyses.

Microbiological Analysis
The faecal samples were analysed for animal selection and to confirm the success of experimental challenge procedures. A total of 1 g of faeces was homogenized in 1 mL of sterile saline buffer and plated onto 5% sheep blood agar plates (blood agar base no. 2, Oxoid, Altrincham, UK) and incubated overnight at 37 • C to assess the presence of haemolytic colonies. Up to 5 haemolytic colonies were selected from each plate and seeded on MacConkey agar (Oxoid, Altrincham, UK), triple sugar iron agar (Oxoid, Altrincham, UK), Simmons citrate agar (Oxoid, Altrincham, UK), and buffered peptone water broth (Oxoid, Altrincham, UK). The bacterial colonies positive for glucose oxidation-fermentation, fermentation of lactose, indole production and sodium citrate-negative were tested with the API system ® (API 20 NE-BioMerieux, Marcy-l'Étoile, France) in order to obtain a more precise identification. Lactobacillus spp. and Enterobacteriaceae were identified on adequate selective media and incubated following adequate conditions. The bacterial concentrations were determined with a semi-quantitative approach based on serial dilutions with medium (detection limit 1 CFU/g).
The Escherichia coli strains isolated from the faecal samples that were cultured on blood agar and MacConkey were further isolated on trypticase soy agar (Oxoid, Altrincham, UK) at 37 • C for 24 h and biochemically identified with API-20E method (BioMerieux, Marcyl'Étoile, France). Besides the evaluation of haemolytic activity, the E. coli were serotyped by using monospecific antisera for the O138 antigen and genetically characterized by PCR for Vt, Vt1, Vt2 and Vt2e. The Vt2e positive strains were assessed for the presence of the F18 gene.

Histological Evaluations
The organs fixed in formalin were paraffin wax-embedded, sliced with a microtome into 5 µm thick histological sections, stained with haematoxylin and eosin for 10 min and microscopically examined at 200× and 400× optical magnifications. The intestinal samples were semi-quantitatively scored (0 = absent, 1 = slight, 2 = moderate, 3 = strong) for the presence of inflammation in the villi and lamina propria (infiltrates of lymphocytes, plasma cells, histiocytes and eosinophils), epithelial regeneration (enterocytes with high nucleus/cytoplasm ratio on the intestinal layer), fusion of the villi, oedema in the deep lamina propria and T atrophy (T-dependent hypocellularity areas), stroma (fibroconnective and histiocytes) and follicular hyperplasia.
For the immunohistochemistry assays, the tissues were dewaxed and unmasked by heat-induced epitope retrieval and Buffer H (Bio-Optica, Milan, Italy); the endogenous peroxidase activity was blocked by incubation with 3% H 2 O 2 . After rinsing and treating with PBS (with normal serum to reduce nonspecific background staining), the sections were incubated with primary antibodies specific for Iba1 (Wako, Neuss, Germany), CD3 (Dako, Naestved, Denmark), CD20 (Thermo Scientific, Rome, Italy), IgG (Vector Laboratories, Burlingame, CA, USA) and IgA (Abcam, Cambridge, UK) at the suggested dilutions. The samples were subsequently incubated with biotinylated secondary antibody (rabbit anti-goat, Vector Laboratories, Burlingame, CA, USA) and labelled by the avidinbiotin-peroxidase (ABC) procedure (VECTASTAIN ® Elite ABC-Peroxidase Kit Standard, Vector Laboratories, Burlingame, CA, USA). The immunoreaction was visualized with a 3,3 -diaminobenzidine substrate (Peroxidase DAB Substrate Kit, Vector Laboratories, Burlingame, CA, USA) after counterstaining with Mayer's haematoxylin (Merck, Kenilworth, NJ, USA) and blindly evaluated with a light microscope by a veterinary pathologist. The known positive control sections were included in each assay. The CD3, CD20 and Iba1 immunoreactions were semi-quantitatively scored in the lamina propria (0 = absent; 1 = rare cells; 2 = some cells; 3 = numerous cells; 4 = very numerous cells), while the IgG and IgA immunoreactions were semi-quantitatively scored on the epithelial luminal surface and lamina propria (0 = absent, 1 = slight, 2 = moderate; 3 = strong).

Evaluation of Specific Intestinal Immunoglobulin-A Titer
For the evaluation of specific IgA (anti-F18 and anti-VT2eB), a double antibody sandwich ELISA system was developed. Fresh faeces (1 g) were vortexed in 10 mL of extraction buffer (0.01M PBS, 0.5% Tween, 0.05% sodium azide) and centrifugated at 1500× g for 20 min at 4 • C. A total of 2 mL of the supernatant were transferred to a sterile tube containing 20 µL of protease inhibitor (Roche Diagnostics, Mannheim, Germany) and mixed briefly. Finally, the samples were centrifuged at 10,000× g for 10 min and the supernatant was collected in sterile tubes.
The anti-pig IgA antibodies (Bethyl 1:500 in coating buffer, 50 mM sodium carbonate pH 9.6) were coated in a 96-well plate overnight at 4 • C. The plates were blocked with PBST and 3% BSA for 3 h at room temperature. The total immunoglobulins extracted from the faeces (included those linked to F18 and VT2eB) were incubated overnight at 4 • C in microtiter wells. After three washes with PBST, 100 µL of anti-F18 or anti-VT2eB rabbit IgGs were added to each well and incubated for 1 h at room temperature. After three washes with PBST, 100 µL of anti-rabbit IgG horseradish peroxidase (HRP) conjugate antibodies were added to each well for 1 h at room temperature. TMB substrate was used for the colorimetric reaction. The reaction was stopped by the addition of 50 µL of stop solution (H 2 SO 4 0.16 M). The absorbances were measured spectrophotometrically by a microplate reader at 450 nm (Model 680 Microplate Reader, Bio-Rad Laboratories Inc., Hercules, CA, USA). The data were expressed as optical densities (O.D.).
The titres of the serum and intestinal mucosa total IgA were determined using specific immunoenzymatic kits (Swine IgA ELISA Quantitation Set, Bethyl Laboratories Inc., Montgomery, TX, USA) according to the manufacturer's instructions. The intestinal mucosa TNF-α, IL-8 and CXCL9 (MIG) levels were measured using three specific ELISA kits (Kingfisher Biotech Inc., St. Paul, MN, USA), while the intestinal mucosa IL-1β was determined using a porcine interleukin 1β ELISA kit (Cusabio Life Science, Houston, TX, USA). Negative and blank (buffer) samples were included on each plate along with duplicates of each sample.

RNA Extraction, Reverse Transcription and Real-Time PCR Assays
The total RNA was isolated from the homogenized samples by a rotor-stator system (Ultra Turrax T25, Staufen im Breisgau, Germany) in 2 mL of TRI ® Reagent (Sigma-Aldrich, Saint Louis, MO, USA). The purity and concentration of the total RNA were spectrophotometrically evaluated (BioPhotometer Eppendorf, Hamburg, Germany) and 1 µg of total RNA from each sample was reverse transcribed to cDNA using the High-Capacity cDNA Archive kit (10 min at 25 • C, 60 min at 37 • C and 5 min at 95 • C) (Applied Biosystems, Waltham, MA, USA). The resulting cDNA was used for real-time PCR (7000 Sequence Detection System, Applied Biosystem, Waltham, MA, USA), according to a published protocol [36], to quantify the expression of the best-characterized swine innate immunity receptors TLRs (2 and 4) and pro-inflammatory cytokines IFN-γ and IL-1β in the jejunum, and the expression of the antigen-presenting molecules MHC (type I and II) in the mesenteric lymph node cells. All the primers were custom synthesized (Invitrogen, Waltham, MA, USA) and designed using the NCBI nucleotide sequences database ( Table 9). The porcine housekeeping gene beta-actin was used as an endogenous control.

Statistical Analysis
The zootechnical performance data were analysed using a mixed model, including the fixed effect of time, treatment, their interaction (treatment × time) and the random effect of animals using JMP ® Pro 15 (SAS Inst. Inc., Cary, NC, USA). Pairwise post-hoc comparisons were performed using the Tukey's honestly significant difference test (Tukey's HSD) or the Tukey-Kramer test, according to the sample size of the group. The faecal E. coli shedding prevalence was analysed using the chi-squared test. Other data were analysed using the Kruskal-Wallis test for unpaired samples and separating medians by the Wilcoxon test for each pairwise comparison. The differences were tested firstly between the control and the treated animals. Furthermore, the differences between animals sacrificed at 3 days and 9 days were verified within the control and treated groups, respectively. Finally, the differences between the control and treated animals sacrificed at 3 days and 9 days were also tested. All the measurements were assessed using the individual pig as an experimental unit. The statistical differences were considered when p ≤ 0.05. The results were presented as least squares means ± standard error.