E. coli serogroup O78:K80 strain EC34195 was isolated from a chicken that died of colibacillosis and has been extensively studied 33 . Deletion of 100 bp of the aroA gene of this strain gave rise to the live-attenuated vaccine Poulvac® E. coli presently commercialized by Zoetis for the control of avian colibacillosis 22 . The lyophilized Poulvac® E. coli vaccine strain was re-constituted in sterile water to ca. 10⁹ colony-forming units (CFU) per mL prior to use. A whole cell formalin-inactivated vaccine based on strain EC34195 was prepared without adjuvant (hereafter designated bacterin), or with a licensed aluminium hydroxide adjuvant, by a supplier of emergency poultry vaccines according to standard procedures (Ridgeway Biologicals Ltd., Compton, UK). A spontaneous nalidixic acid resistant derivative of EC34195 was produced by plating approximately 10 log₁₀ colony-forming units (CFU) of EC34195 on MacConkey agar (Oxoid, Basingstoke, UK) supplemented with 25 μg/mL nalidixic acid (Mac + Nal). The subsequent derivative (EC34195nal^R) was passaged in pure culture and confirmed to possess an identical growth rate, phenotypic characteristics and panel of virulence-associated genes (cvi/cvaC iss, iucD and tsh) to the parent strain. Strain EC34195nal^R was stored at −70 °C in 15% (v/v) glycerol in Luria Bertani (LB) broth until required. The inoculum of this strain was prepared as described 26 .

Animal experiments were conducted according to the requirements of the Animal (Scientific Procedures) Act 1986 (licence no. 30/2463) with the approval of the Local Ethical Review Committee. Male Big5FLX turkeys were obtained from Aviagen Ltd. (Tattenhall, Cheshire, UK) as day-old poults, housed and reared as previously described 26 .

Thirty 1-day old turkey poults were randomly housed in groups of 10 birds per room on floor pens to simulate commercial practice. For the delivery of a coarse aerosol spray of live vaccine, birds were restrained in one corner of the room at a density of 10 birds per 0.25 m², and the vaccine was administered with a syringe fitted with a device that creates a coarse aerosol directed at the faces of individual birds. Group A (control) received 10 mL sterile saline as coarse spray (per group, not individually) while group B received 10 mL of saline containing 10⁷ CFU per mL of the Poulvac® E. coli live-attenuated vaccine. Each bird in group C received 0.2 mL (containing the equivalent of approximately 2 × 10⁹ CFU) of the EC34195 inactivated vaccine via the subcutaneous route as per commercial practice and recommended by the manufacturer. Birds were monitored for adverse reactions every 3 h for the first 24 h and thereafter twice daily until the booster vaccinations. Birds in group B were given a booster of ca. 10⁷ CFU per mL in drinking water at 7 days of age. Excess vaccine-treated water was discarded after 18 h. Birds in group C were given a booster (0.4 mL) of the same inactivated vaccine at 14 days of age via the subcutaneous route in the neck region. The control group was given sterile saline in drinking water mixed at the same proportions as for the live vaccine. Birds were monitored twice daily for the remainder of the experiment. On day 41 post-primary vaccination, all the birds were inoculated into the left caudal thoracic airsac with 10⁷ CFU of the EC34195nal^R strain, on which both vaccines are based, in 100 μL of saline and monitored overnight for clinical signs. After 24 h, all the birds were killed humanely and subjected to post-mortem examination essentially as described 26 . Part of the spleen was collected for T cell re-stimulation assays. Lung, liver, kidney and the remaining spleen tissue were sampled to enumerate tissue-associated bacteria whilst blood and bile were collected for serology.

Having established vaccine efficacy in pilot studies (above), one hundred and thirty-five 1-day old turkey poults were randomly divided into 3 groups of 45 birds and housed separately. Birds were vaccinated as described above with sterile saline (control group), Poulvac® E. coli or inactivated vaccine with aluminium hydroxide adjuvant. Boosters were given at the respective time-points as above. On days 1, 3, 5, 7, 14, 21 and 28 post-primary vaccination, five birds were randomly removed from each group, killed humanely and subjected to post-mortem examination. From each bird, sections of liver, lung and spleen were collected in RNAlater (Ambion, Warrington, UK) for cytokine analysis. In the live vaccine group, the remainder of the lung was used to enumerate the vaccine strain as described below. Selected colonies were subjected to PCR analysis spanning the aroA gene using the primers (aroA-F-TTGAGTTCGAACGTCGTCAC and aroA-R-GCAATGTGCCGACGTCTTTG) and agglutination with specific anti-O78 serum (AHVLA, Weybridge, UK) to confirm recovery of the vaccine strain. At day 28 post-primary vaccination, the spleen was collected into ice-cold RPMI1640 medium for use in T cell re-stimulation assays and blood and bile were collected for serology. On day 41 post-primary vaccination, all the remaining birds were challenged, subjected to post-mortem examination 24 h later and tissues sampled as described above.

Total RNA extraction and cytokine mRNA analyses were performed essentially as described 26 . Data are expressed in terms of the cycle threshold value (Ct), normalised for each sample using the Ct value of 28S rRNA product for the same sample, as described previously 34 , 35 . Final results are shown as corrected ΔCt, using the normalised value.

Viable tissue-associated bacteria were enumerated essentially as described 4 . The lower limit of detection by this method was 1.79 log₁₀ CFU/g of tissue. Enrichment cultures were also performed in which 1 mL of tissue homogenate was cultured in 9 mL brain heart infusion (BHI) broth for 18 h at 37 °C aerobically, prior to plating on Mac + Nal, giving a theoretical lower limit of detection by enrichment of 1 log₁₀ CFU/g of tissue.

Preparation of APEC soluble antigens for ELISA, quantification of total protein content, storage and subsequent antibody assays were performed essentially as described 26 . Cell-mediated responses were determined by splenocyte proliferation assays again as previously described 26 and data is expressed as mean counts per minute (cpm).

The Th1-stimulating adjuvant CASAC: (combined adjuvant synergistic activation of cellular immunity) was formulated by an adaptation of the method described 36 . Briefly, the first dose comprised 50 μg of monophosphoryl lipid A (MPL; Invivogen, Toulouse, France), 50 μg of Pam3CysSerLys4 (Pam3CSK4; EMC Microcollections, Tuebingen, Germany) and 50 μg of chicken interferon-γ (2B Scientific, Upper Heyford, UK) homogenized with 100 μL (same dose as previously used) of formalin-inactivated EC34195 supplied to commercial specification by Ridgeway Biologicals Ltd. but without adjuvant (bacterin). The booster comprised the same ingredients of CASAC mixed with 200 μL of bacterin. Bacterin formulated with a second Th1-biasing adjuvant was separately prepared using synthetic cytosine phosphodiester-guanine (CpG)-containing oligodeoxynucleotides (ODN) as described 37 . The sequence for CpG-ODN, TCGTCGTTTGTCGTTTTGTCGTT (bold motifs enhance CpG-ODN activity), was synthesized with a phosphothionate backbone (Sigma). The first dose was prepared by mixing 20 μg of synthetic CpG-ODN (in sterile pyrogen-free water) with 100 μL of the non-adjuvanted EC34195 bacterin. The booster dose comprised the same amount of CpG-ODN in 200 μL of non-adjuvanted EC34195 bacterin.

For comparison, the formalin-inactivated bacterin without adjuvant was also formulated with Th2-stimulating adjuvants. The first of these comprised a dose of 100 μL of EC34195 bacterin thoroughly mixed with an equal volume (ratio of 1:1) of Titermax® Gold (Sigma) as per manufacturer’s instructions. The booster dose was 200 μL of the homogenous mixture. EC34195 bacterin formulated with aluminium hydroxide adjuvant was also tested as another Th2-stimulating adjuvant; 100 μL for primary vaccination and 200 μL for the booster (same dose as in the pilot studies).

For vaccination, thirty 1-day old turkey poults were randomly divided into 6 groups of 5 birds and housed separately as before. Group 1 received sterile saline (negative control group), group 2 received formalin-inactivated EC34195 bacterin without adjuvant (control group), group 3 received the same bacterin formulated with CpG-ODN, group 4 received the same bacterin with CASAC, group 5 received the same bacterin mixed with aluminium hydroxide, group 6 received the same bacterin mixed with Titermax® gold. All vaccines were administered by the subcutaneous route with booster doses for the respective vaccine formulations given at 14 days of age via the same route. Birds were monitored twice daily for the duration of the experiment. On day 41 post-primary vaccination, all the birds were inoculated with 10⁷ CFU of strain EC34195nal^R in 100 μL of saline into the left caudal thoracic airsac as above and monitored overnight for clinical signs. All the birds were killed humanely and subjected to post-mortem examination and collection of organs and blood as above.

Cytokine data was analysed for statistical significance using a one-way ANOVA test with an ad hoc Tukey’s test. Bacterial counts recovered from tissues, antibody levels and splenocyte recall responses were analysed by Student’s t-test. P values ≤ 0.05 were considered significant.

To evaluate the protective efficacy of existing vaccines and establish models in which the basis of protection can be dissected, turkeys were vaccinated using Poulvac® E. coli or an inactivated vaccine based on the parent strain of Poulvac® E. coli and compared to mock-vaccinated turkeys. In birds given the live vaccine, up to ca. 4 log₁₀ CFU per gram lung tissue of the vaccine strain was detected 1, 3, 5 and 7 days post-primary vaccination (data not shown). Recovery of the vaccine strain was confirmed by a specific PCR for the aroA gene in E. coli isolated from the lung homogenates at these times (data not shown). The live-attenuated vaccine strain was not detectable at any other time-points in any group.

Upon intra-airsac challenge of vaccinated birds with a nalidixic acid resistant derivative of the strain used to prepare the vaccines, significant protection was observed in the groups that received either Poulvac® E. coli or the inactivated vaccine (Figure 1). Approximately 4 log₁₀ CFU/g EC34195nal^R were recovered from the lung, liver, spleen and kidney of mock-vaccinated turkeys. In contrast, nalidixic acid resistant E. coli were only detectable in the spleens, livers and kidneys of turkeys that received the inactivated vaccine by enrichment, where the theoretical lower limit of detection was 1 log₁₀ CFU/g (Figure 1; P < 0.05). The number of nalidixic acid resistant E. coli recovered from the lung, liver, spleen and kidney of turkeys that received the live vaccine was significantly reduced by around 2 log₁₀ CFU/g (Figure 1; P < 0.05). Consistent with the lower numbers of bacteria recovered from the organs of vaccinated birds after challenge with EC34195nal^R, none of the vaccinated birds exhibited clinical signs or pathological lesions indicative of colibacillosis as previously defined 26 . By contrast, all mock-vaccinated birds challenged with EC34195nal^R exhibited clinical signs typical of colibacillosis (respiratory distress, hunched posture, ruffled feathers, reduced response to stimulus) and gross post-mortem lesions including airsacculitis, pericarditis, perihepatitis and fibrin deposits on serosal surfaces at 24 h post-challenge. It can therefore be inferred that the strain is virulent and that live-attenuated or inactivated vaccines derived from it confer protection against homologous re-challenge.

To assess cytokine responses as signatures of innate and adaptive immune responses induced by the live-attenuated and inactivated vaccines, turkey poults were vaccinated as above and killed at intervals. We analysed the levels of transcripts encoding a pro-inflammatory cytokine (IL-1β), a pro-inflammatory chemokine (CXCLi2), a signature Th1 cytokine (IFN-γ), a signature Th2 cytokine (IL-13) and two signature T regulatory cytokines (IL-10 and TGF-β4), at intervals post-primary and -secondary vaccination.

The normalised levels of transcripts encoding these cytokines are shown over time in the spleen (Figure 2), liver (Figure 3) and lung (Figure 4). In the spleen, the live-attenuated and inactivated vaccines significantly induced transcription of IL-13 in the first week post-primary vaccination (Figure 2; P < 0.05). In the same period, transcription of the signature Th1 cytokine IFN-γ was significantly repressed following vaccination relative to levels in the mock-vaccinated controls (Figure 2; P < 0.05), and transcription levels of the anti-inflammatory cytokine TGF-β4 (the chicken equivalent of mammalian TGF-β1) were significantly induced following vaccination (Figure 2; p < 0.05). Together, these data indicate a bias towards a Th2 response at a key site of immune priming during the typical period of induction of an adaptive immune response to primary vaccination.

In the other two organs, the pattern is less clear. Elevated IL-13 mRNA expression levels were also detected in the liver (Figure 3) and lung (Figure 4), but only at a couple of time-points and with no discernible pattern in the first week post-primary vaccination with live-attenuated or inactivated vaccines. In the lung, IFN-γ transcription was generally lower in turkeys given the live-attenuated vaccine compared to mock-vaccinated birds, significantly so at multiple time-points (Figure 4). However, the inactivated vaccine only significantly repressed IFN-γ transcription in the lung at 21 and 28 days post-primary vaccination, time-points which would represent the resolution of a normal adaptive immune response to primary vaccination (Figure 4). No repression of IFN-γ transcription was detected in the liver at any time-point (Figure 3). In both organs, we detected significant induction of TGF-β4 mRNA expression in turkeys vaccinated with the live-attenuated or inactivated vaccines, particularly in the first two weeks post-primary vaccination (Figures 3 and 4; P < 0.05).

Differences between groups in transcription of the other cytokines analysed did not show consistent trends across organs or time in vaccinated turkeys relative to controls, with the exception of CXCLi2 expression, which was unaltered following vaccination in the spleen (Figure 2), consistently down-regulated in the lung (Figure 4; P < 0.05) and generally up-regulated in the liver (Figure 3; P < 0.05). CXCLi2 mainly chemo-attracts monocytes 35 , and this differential expression may reflect traffic of these cells in response to vaccination.

To identify adaptive immune responses associated with protection against homologous challenge induced by each vaccine, we measured humoral and cellular responses at intervals after primary and secondary vaccination of turkeys using samples from the same birds as used for cytokine analysis. Immunoglobin Y (IgY) levels and splenocyte proliferation responses were measured in birds killed at day 28 and after challenge (day 42) whereas IgA in bile was measured 24 h after challenge. Both vaccines induced elevated IgY levels post-primary vaccination relative to the mock-vaccinated group, for the inactivated vaccine at both intervals and for the live-attenuated vaccine at day 42 (Figure 5A; P ≤ 0.05). APEC-specific IgA was elevated in bile from poults given the inactivated vaccine relative to the mock-vaccinated group (Figure 5B; P ≤ 0.05).

The ability of splenocytes recovered from each group to proliferate in response to mitogen or soluble EC34195 antigen was also evaluated at day 28 post-primary vaccination and at post-mortem examination on day 42 (24 h after challenge). The response to ConA was comparable between the birds given mock- or inactivated vaccine at day 28, and was not significantly different from control birds in the group given Poulvac® E. coli (Figure 6A). After challenge, almost no proliferation of splenocytes from the control group was seen in response to ConA (Figure 6A), consistent with a suppressive effect of APEC O78 infection detected following primary and secondary infection 26 . Splenocytes recovered from birds given live-attenuated or inactivated vaccines at day 42 proliferated in response to ConA (Figure 6A).

Splenocyte proliferation in response to APEC soluble antigen was significantly elevated both vaccinated groups at day 42 relative to levels in the mock-vaccinated group (Figure 6B, P < 0.05). Responses were lower in the group given the live-attenuated vaccine relative to the control group at day 28 and relative to the inactivated vaccine group at both time-points. At day 42, no response to EC34195 soluble antigen was seen in the control group.

To examine the effects of Th1- and Th2-stimulating adjuvants, we formulated the EC34195 inactivated bacterin with two compounds with a reported bias toward Th1 responses (CASAC or CpG) and two with a reported bias to Th2 responses (aluminium hydroxide or TiterMax®Gold) 36 , 38 , 39 . The vaccines were given to day-old turkey poults and again 14 days later by the subcutaneous route, with intra-airsac challenge with EC34195nal^R at day 41, as described in pilot studies where protection conferred by the inactivated vaccine was shown (Figure 1). A bacterin without adjuvant and mock-vaccination with saline were used as controls.

At 24 h post-challenge, statistically significant reductions in the numbers of EC34195nal^R recovered from the liver and kidney were observed in birds vaccinated with bacterin formulated with either CASAC or CpG oligonucleotides as adjuvants, and in the spleen of birds vaccinated with CpG oligonucleotides, relative to levels in the mock-vaccinated group (Figure 7A; P < 0.05). In the groups that received the bacterin formulated with CASAC or CpG, there was also a significant reduction in bacterial numbers in the liver relative to the group that received the bacterin alone, but this difference was not seen in the spleen, kidney or lung (Figure 7A).

In the turkeys vaccinated with the bacterin formulated with the Th2-stimulating adjuvants aluminium hydroxide or TiterMax®, no nalidixic acid resistant E. coli were isolated from the liver, kidney and spleen on direct plating (Figure 7B), but they were detected by enrichment of homogenates of these organs (a theoretical limit of detection of 1 log₁₀ CFU/g). Differences in bacterial numbers were statistically significant in all organs from the group that received bacterin formulated with TiterMax® relative to either the bacterin alone or mock-vaccinated control groups (P < 0.05). Significant reductions in bacterial numbers were also seen in the liver, spleen and kidney of birds that received the bacterin with aluminium hydroxide relative to the groups that received mock, bacterin alone, or bacterin formulated with CASAC or CpG (Figure 7A and B).

On analysis of APEC-specific IgY at 42 days post-primary vaccination, the bacterin alone induced significantly elevated APEC-specific IgY levels relative to mock vaccinated birds (Figure 7C; P < 0.05). The magnitude of the APEC-specific IgY response was no greater when the bacterin was formulated with CpG and marginally lower when formulated with CASAC relative to bacterin alone (Figure 7C). Levels of APEC-specific IgY were no greater in the groups that received the bacterin formulated with aluminium hydroxide and marginally greater when formulated with TiterMax® compared to the bacterin alone (Figure 7D). There is therefore a suggestion that these Th2-biasing adjuvants might generate greater levels of APEC-specific IgY response than Th1-biasing adjuvants.