After ABT-737 purchase washing, the plate was developed using 3,3′,5,5′-tetramethylbenzidine as the HRP substrate. The reaction was terminated with the addition of 0.25% (w/v) hydrofluoric acid. Absorbance was measured at 630 nm in an ELISA reader (BioTek spectrophotometer, Winooski, VT). End-point titers were calculated as the reciprocal of the last serum dilution that was two-fold higher than the control. We isolated porcine neutrophils and investigated opsonophagocytosis based on the studies of Zhang et al. (2009). The number of viable cells was counted by trypan blue staining and adjusted to 4 × 106 cells mL−1 in Dulbecco’s Modified Eagle’s Medium.

ExPEC PCN033 was grown to log phase and adjusted to 4 × 104 CFU mL−1. To facilitate interactions between bacteria and antibodies, bacterial cells were preincubated in 10% mouse serum at 37 °C for 30 min. The reaction mixture consisted of aliquots of 500 μL

bacteria, 500 μL neutrophils and 100 μL healthy piglet serum as a complement source. The mixture was incubated at 37 °C for 1 h with rotation. After phagocytosis, the neutrophils were lysed with sterile water and serially diluted 10-fold. Dilutions find more were plated on LB plates and incubated at 37 °C overnight to determine viable counts. The sera from mice immunized only with adjuvant were used as a control. The efficiency of bacterial killing was estimated by the Clomifene following formula: [1 − (number of bacteria recovered in presence of phagocytes/number of bacteria recovered in absence of phagocytes)] × 100% (Zhang et al., 2009). Data of the opsonophagocytosis

assay are summarized as mean ± standard deviation. The differences in antibody titers from the ELISA and the percentage of bacteria killed in the opsonophagocytosis assay were determined using the Mann–Whitney–Wilcoxon test. The significance cutoff was set to 0.01. The complete coding regions of E. coli porin genes ompC and ompF sequenced in the present and previous studies were collected to detect evidence of recombination and selective pressure (Table 1). Multiple sequence alignments were carried out for the translated protein sequences using the program t-coffee (Notredame et al., 2000). The aligned amino acid sequences were then mapped onto the corresponding codon sequences. A maximum likelihood phylogenetic tree was reconstructed using phyml (Guindon & Gascuel, 2003). Recombination events in our datasets were tested using the Single Break-Point (SBP) and the Genetic Algorithms for Recombination Detection (GARD) methods in the hyphy package (Pond et al., 2006). To infer selective pressure on the coding genes, the ratio of nonsynonymous substitutions (dN) to synonymous substitutions (dS) was estimated using the fixed effects likelihood (FEL) method via the Datamonkey webserver (http://www.datamonkey.org/) (Pond & Frost, 2005).