In the present study, a total of 581 chicken Escherichia coli strains were isolated from 1045 liver samples of sick and dead chickens from 50 chicken farms at Shaanxi, Henan, Hebei and Shanxi provinces in China during 2007-2012 (Table 1). Samples were collected from enlarge livers of sick and dead chicken in sterile condition. In the present study, 581 chicken Escherichia coli strains were cultured on MaConkey agar plates, and incubated at 37˚C for 18 h, and then single bacterial colony was picked up and purified. All 581 chicken E .coli strains were then sent to China Institute of Veterinary Drug Control for serotyping.

581 chicken Escherichia coli strains were cultured in LB broth at 37˚C for 24 h. All tested Escherichia coli strains were individually suspended in sterile saline at a concentration of 10⁸ CFU/mL. In the pathogenic test, 581 groups of one-day-old White Leghorn specific-pathogen-free chickens supplied by Yangling Green Square Biological Engineering Co. (China) were used. Each group included five one-day-old White Leghorn specific- pathogen-free chickens and was reared in separate cages with food and water. Each chicken was inoculated subcutaneously with 0.2 mL of Escherichia coli suspension (1 × 10⁸ CFU/per mL). Deaths were recorded 4 times per day, and continuing for 7 days. Clinical signs of illness were recorded daily. The surviving chickens were killed at 7 day post-inoculation. The lesions were observed and bacteria were isolated and identified.

The pathogenicity of Escherichia coli strains was determined on the basis of lesions and mortalities as described previously [10] - [12] . The animal experiments were carried out in compliance with the regulations of the Guide for the Care and Use of Laboratory Animals prepared by the international animal welfare standards. All animal experiments in this study were approved by the committee on Research Animal Care of Northwest A and F University.

The different sets of primers (Table 2) used for PCR amplification were designed with Primer 5.0 software based on the previously published sequences [13] [14] . Primers were synthesized by Nanjing GenScript Biotechnology Co. (China).

All Escherichia coli strains and reference strains were grown on LB agar plates at 37˚C overnight. Escherichia coli colonies were suspended in 500 μL of deionized water and boiling for 10 min, followed by chilling on ice for 5 min and centrifugation at 10,000 ×g for 5 min, the supernatant was used as the DNA templates for PCR amplification. The PCR mixture contained 10 μL of 2 × PCR Master mix (including 2 × Taq DNA polymerase, 2 × PCR Buffer and 2 × dNTP mixture) (TaKaRa), 1 μL of primer pair, 4 μL of DNA template, and deionized water to a final volume of 25 μL. PCR was completed by an initial heat activation of 5 min at 95˚C; then 30 cycles of 30 s at 94˚C, 30 s at Tm, and 45 s at 72˚C; and an extension of 10 min at 72˚C. PCR products were separated by size by 1% agarose gel electrophoresis along with DL2000 DNA markers and visualized after staining with ethidium bromide on a UV transilluminator. PCR products amplified from the virulence genes in Escherichia coli strains were sequenced and their sequences were compared with previously published sequences by using DNAStar software.

Chi-square test were used to analyze the numbers and combination of virulence-associated genes in 320 chicken pathogenic and 50 non-pathogenic Escherichia coli strains and to analyze the numbers and combination of virulence-associated genes and their correlations with pathogenicity.

Among 581 Escherichia coli strains examined, 320 strains caused the death of chicken, suggesting that they are pathogenic Escherichia coli strains; 261 strains did not cause the death of chicken, suggesting that they are non-pathogenic Escherichia coli strains. Pathogenic test indicated that 320 chicken E. coli strains were classified into high , intermediate , and low pathogenicity groups based on the lesions and mortality of experimentally infected chickens [12] . Pathogenic test shows that the percentage of high pathogenic, intermediate pathogenic and low pathogenic strains was 60.3% (193/320), 30.6% (98/320), 9.1% (29/320), respectively.

Among 18 virulence-associated genes examined, 13 virulence-associated genes irp2, fyuA, iucA, iucD, iutA, FimA, FimC, papC, iss, hlyA, tsh, colV, and colBM in 581 Escherichia coli strains were detected by PCR. The

^aF: forward; R: reverse.

specific PCR products of expected sizes were revealed by agarose gel electrophoresis (Figure 1 and Figure 2). The sequences of these virulence-associated genes were determined and deposited with GenBank accession numbers, as shown in Table 3. Sequence analysis revealed that these 13 virulence-associated gene sequences detected in this study show high identity with those of published genes (Table 3).

Among 320 pathogenic and 50 non-pathogenic Escherichia coli strains examined, at least 2 to 13 genes were detected (Table 4). By examining 18 virulence-associated genes, the papA, vat, eaeA, stx1, and stx2 genes were not detectable in this study. Among the 13 virulence-associated genes detected, the most prevalent is colBM (100%) and the least one is hlyA (0.6%) (Table 4).

Our results reveal that the number of virulence genes in 193 high pathogenic Escherichia coli strains was significantly higher than that of 127 intermediate and low pathogenic strains (P < 0.01). Furthermore, the detection rates of the virulence genes for irp2, fyuA, iutA, papC, iss, colV, and iucD in 291 high and intermediate pa-

^aSequences determined in this study.

thogenic Escherichia coli strains were significantly higher than that of 29 low pathogenic and non-pathogenic Escherichia coli strains (P < 0.01). The detection rate of the virulence genes for iucA, fimA, and tsh in intermediate and low pathogenic Escherichia coli strains shows no significant difference (P > 0.05). The detection rate of virulence genes for FimC and colBM in three different pathogenic Escherichia coli strains shows no significant difference (P > 0.05). Interestingly, the hemolysin hlyA gene was detected only in highly pathogenic chicken Escherichia coli.

By examining 193 highly pathogenic Escherichia coli strains, it was found that all strains simultaneously carried at least 8 to13 virulence genes. As shown in Table 5, 14 strains (7.3%) simultaneously carried 12 virulence genes, causing the death of 4 - 5 chickens within 24 hours; 85 strains (44%) simultaneously carried 10 or 11 virulence genes, causing the death of 4 - 5 chickens within 72 hours. Among 13 virulence genes detected in 193 highly pathogenic Escherichia coli strains, a total of 51 different combinations of virulence-associated genes were observed. It was found that 134 of 193 (69.43%) highly pathogenic Escherichia coli strains belonged to 10 predominant patterns as indicated in Table 5. The most prevalence of combination patterns of virulence-asso- ciated genes (irp2 + fyuA + iucA + iucD + iutA + fimA + fimC + iss + tsh + colV + colBM and rp2 + fyuA + iucA + iucD + iutA + fimA + fimC + papC + iss + tsh + colV + colBM) in highly pathogenic Escherichia coli strains were 22.3% and 7.3%, respectively (Table 5).

By examining 98 intermediate pathogenic Escherichia coli strains, 12 virulence genes were detected in these strains which simultaneously carried at least 5-8 virulence genes. In the present study, 36 different combination patterns from these 12 virulence genes in 98 intermediate pathogenic Escherichia coli strains were found and 60.2% (59/98) of them belonged to the 7 predominant patterns (Table 5). The most prevalence combination patterns of virulence-associated genes (irp2 + fyuA + iutA + fimC + iss + colV + colBM and irp2 + fyuA + iucD + fimA + fimC + colV + colBM) in intermediate pathogenic Escherichia coli strains were 13.2% and 10.2%, respectively (Table 5).

By assessing 29 low pathogenic Escherichia coli strains, 9 virulence genes were detected in these strains which simultaneously carry at least 2 to 4 virulence genes. Twelve different combinations from these 9 virulence genes were observed, and 62% (18/29) of them belonged to 4 predominant patterns (Table 6). The most prevalence combination patterns of virulence-associated genes (iucD + iutA + fimC + colBM and fimA + fimC + colBM ) in low pathogenic Escherichia coli strains were 20.7% and 17.2%, respectively (Table 6). Furthermore, there are 7 virulence genes that were detected in 50 non-pathogenic Escherichia coli strains, and each carries at least 2 to 4 virulence genes. Four different combinations from 7 virulence genes were seen, and 70% (14/20) of them belong to 2 predominant patterns (Table 6). The major combination patterns of virulence-associated genes (colBM + fimA and colBM + fimC + tsh) in non-pathogenic Escherichia coli strains were 20.7% and 30%, respectively (Table 6).

Among 320 Escherichia coli strains examined, the detection rate of high pathogenicity island genes (irp2, fyuA, and colV) in high pathogenic strains were significantly higher than those in low and non-pathogenic strains (P < 0.01) (Table 7). Nine virulence-associated genes irp2, fyuA, iucA, iucD, iutA, papC, iss, tsh, and colV were significantly more often detected in high and intermediate pathogenic E. coli strains (Table 7). Furthermore, our results revealed that the combination patterns from five or more than five genes in high and intermediate pathogenic E. coli strains were highly correlated with its pathogenicity (P < 0.01).