Avian pathogenic Escherichia coli (APEC), the etiologic agent of avian colibacillosis, is one of the most important bacterial diseases of domestic poultry causing huge economic losses worldwide (1). The disease is characterized by various conditions including septicemia, swollen head syndrome, yolk sac infection (YSI), cellulitis, as well as inflammation of different organs such as pericarditis, airsacculitis, and perihepatitis (2). APEC infections lead to increased morbidity, mortality, and carcass condemnation (3).

Virulence traits are important in the pathogenesis and epidemiology of APEC infections (4). Generally, the E. coli strains isolated from lesions of colibacillosis and that harbor virulence-associated genes (VAGs), such as those encoding adhesins, toxins, invasins, iron-scavenging systems, and serum survival, are classified as APEC (5, 6).

Combinations of several VAGs are needed to confer pathogenicity upon APEC and usually no single VAG alone is attributed to disease in chickens (5). According to the literature, certain serotypes appear to be associated with virulence among APEC as most O78 and O2, followed by O18 and O1 are responsible for more than half of the cases of colibacillosis worldwide (5).

From a one health perspective, APEC also appears to have a role in human disease. APEC is a subgroup of extraintestinal pathogenic E. coli (ExPEC), which is distinct from commensal and diarrheagenic E. coli groups. Different studies reported that APEC and ExPEC strains share common characteristics in terms of serotypes, phylogenetic groups, virulence factors, and ability to cause disease in various models of human and animal disease (7–9) and that E. coli with ExPEC characteristics occur on retail poultry meat targeted for human consumption (10). Therefore, APEC strains may represent a zoonotic threat either by causing disease in human hosts or via horizontal gene transfer of plasmid-linked VAGs to human commensal strains (11, 12). Importantly, many of the VAGs known to contribute to APEC virulence are linked to pathogenicity islands (PAIs) found on large transmissible plasmids (13, 14). Such plasmids may enable commensal strains to cause disease in models of human and animal disease (15, 16).

Another reason for the importance of APEC strains is the increasing antimicrobial resistance (AMR) among them. Increasing AMR, in both animal and human bacteria, has led to growing concerns worldwide (17). Multi-drug resistant (MDR) bacteria are frequently reported in poultry (17, 18). Antibiotics are administered as growth promoters and for the treatment, control, and prevention of bacterial infectious diseases in the poultry industry (19). Altogether, these factors have resulted in the higher AMR index in the poultry industry (3). These resistant strains could be the source of AMR for humans via chicken products or direct contact with infected birds (20).

There are different viewpoints on the relationship between VAGs and AMR. Some researchers believe that VAGs of E. coli strains are more associated with AMR (21, 22), whereas other researchers have believed that VAGs may even be weakly or negatively linked to AMR (23, 24).

In the present study, we aimed to reach a broader perspective on the presence of VAGs and AMR by comparisons of characteristics among APEC and avian fecal E. coli (AFEC) from apparently healthy birds.

A total of 100 E. coli strains were isolated from broiler chickens with different conditions. The number of identified strains related to colisepticemia, YSIs, and healthy chickens was 32, 32, and 36, respectively.

Colibacillosis is an important disease in the poultry industry caused by E. coli. However, unlike most enteric pathogenic E. coli, the extraintestinal infections are not specifically attributed to few well-known virulence-associated determinants. In the present study, we aimed to reach a more integrated perspective on APEC strains by considering various genotypic and phenotypic traits and performing statistical analysis to search for any possible associations. Any observed association could be considered a primary clue that could be further evaluated in future studies. For this purpose, we characterized and compared the strains isolated from sick and healthy birds according to AMR phenotype, VAGs, phylogenetic group, and ERIC fingerprints. All investigated E. coli strains were susceptible to fosfomycin, ceftriaxone, and cefixime; however, the rate of resistance against other antimicrobial agents was alarming. Despite the prohibition of some antibiotics in veterinary medicine such as chloramphenicol, 65% of the investigated strains were resistant to this drug. In the latest edition of World Health Organization Critically Important Antimicrobials lists (WHO CIA lists; 6th version), the highest priority CIAs are follows: third- and higher-generation cephalosporins, polymyxins, quinolones, macrolides-ketolides, and glycopeptides (39). Ceftriaxone and cefixime are the third generation of cephalosporins and no resistance was found to them. However, more than 80% of E. coli strains tested were resistant to fluoroquinolones (enrofloxacin and flumequine), which seems to be alarming. The use of antimicrobial agents as growth-promoting additives in poultry is not recommended by the Iranian Veterinary Organization, but we have no official information about illegal antibiotic usage as growth promoter in the poultry industry. However, Shahroozian and Khoshgoftar (40) found gentamicin residue in 78.57% of poultry meat samples of Semnan province (40). In another study, enrofloxacin residue was detected in poultry meat (41). Such antibiotic residues may be due to the unofficial use of drugs with the unfortunate outcome of selection of resistance among intestinal commensal E. coli strains. Our results about high AMR and MDR rates in AFEC and APEC strains were consistent with previous studies in Iran (42, 43). Importantly, the resistance rates were significantly higher in fecal commensal strains (AFEC) against some antimicrobials such as gentamycin and enrofloxacin; this observation indicates the necessity of surveillance of AMR in animals in both commensal and pathogenic strains particularly where antimicrobial stewardship strategies have not been rigorously implemented.

In the present study, we investigated the frequency and scores of VAGs among different E. coli strains and the association between these genes. E. coli strains isolated from chickens with typical signs of colibacillosis were divided into two groups of colisepticemic and YSI. Statistical analysis revealed that colisepticemic strains significantly have more APEC predictors genes (iss, iutA, hly, iroN, and ompT) introduced by Johnson et al. (36) compared to YSI strains (Table 2). In addition, the tsh gene was also attributed to the colisepticemic strains. However, more than 63% of AFEC strains also harbored the tsh gene. This gene is known as a putative virulence gene (44), which is located on the variable region of ColV plasmids, non-ColV plasmids, or within PAIs on the bacterial chromosome such as PAI III₅₃₆ that was originally identified in UPEC strains (14). Compression of the frequency of VAGs and VAG scores show that YSI-causing strains have less virulence traits than colisepticemic strains; therefore, the disease caused by these strains could be secondary in nature. The astA gene was more prevalent in the AFEC strains compared to APEC. Although astA has been reported in ExPEC (45), it appears to be more related to intestinal (commensal or diarrheagenic) strains (46–48). astA is a virulence gene also associated with the enteroaggregative E. coli (EAEC) isolates (49–51), although it is not restricted to this pathotype (52). The presence of this gene does not appear to contribute to disease by E. coli unless combined with other virulence traits (53).

The frequency of colV gene was not statistically significant between two groups of APEC and AFEC strains, and this gene was not detected in 69% of strains with predictor VAGs of APEC. Although the colV operon is the namesake trait of ColV plasmids and encodes the production of colicins, it is not necessarily related to APEC pathogenicity. The absence of colV gene in some APEC strains with all predictor genes indicates that these markers may not be solely carried on the ColV plasmids. Johnson et al. (13) found that the ColBM plasmids can be the location of VAGs occurring in ColV-negative strains. Therefore, the absence of colV gene does not indicate that the strains are not pathogenic.

Horizontal gene transfer or carriage within the mobile genetic elements, or co-selection of virulence genes leads to the simultaneous appearance of these genes in a particular strain (32, 54, 55). Since several virulence factors are usually involved in causing infection and disease, understanding these relationships is very important to unraveling the pathogenesis of colibacillosis. Correlation analysis revealed the associations among iss, hly, iutA, iroN, and ompT VAGs. This gene profile appears to play an important role in the pathogenicity of APEC, as previously identified in research by Johnson et al. (14). We also found a positive association between tsh and other APEC predictor genes; however, the correlation between tsh with iutA was stronger than the other genes. iutA encodes the ferric aerobactin receptor, which is involved in acquisition of iron. A group of strains that were positive for both iutA/tsh and negative for other predictor genes were placed in the AFEC group. Therefore, the association of these two genes seems insufficient to predict the pathogenic potential of APEC. Very strong association was found between papG and papC genes, which are part of the pap operon within a PAI on the chromosome (56).

Regarding the relationship between AMR and VAGs, two possibilities have been raised. First, resistant strains are more virulent because many genetic elements such as plasmids, integrons, and composite transposons may carry both VAGs and AMR genes simultaneously. On the other hand, some researchers believe that AMR strains are less virulent than susceptible ones because AMR and virulence have not necessarily evolved simultaneously (23, 57). In the present study, no relationship was found between VAGs and resistance phenotype. Only a weak positive association was found between ompT and resistance to tetracycline and tsh and furazolidone. The particular host, geographical origin, and type of antimicrobial agent used, can probably affect the relationship between AMR and VAGs (58).

A strong cross-resistance phenomenon was observed between chloramphenicol with florfenicol. Florfenicol is an analog of chloramphenicol that has not been approved for human use. It was reported that cross-resistance to chloramphenicol and florfenicol is due to non-enzymatic activity of E. coli flo gene. Also, a strong cross-resistance was found between enrofloxacin (quinolone) and flumequine (fluoroquinolone). Resistance against quinolones and fluoroquinolones is acquired through mutations in some chromosomal genes such as gyrA and parC or by acquiring a wide range of plasmid-encoded genes (59).

Classification of strains based on phylogenetic group showed that AFEC strains mostly belonged to groups A and E, while colisepticemic and YSI strains belonged to groups C and B1, respectively. Phylogroup A had the lowest VS among other phylogroups of AFEC strains. In contrast, while only three colisepticemic strains were present in the A group, their VS mean (8.3) was high, which indicates inconsistency of VSs within a particular phylogroup. The highest mean of VS was found in group C and most strains of this phylogroup were isolated from colisepticemic lesions. Although few reports indicated that this phylogroup is close to groups B1 or A (60, 61), the present study showed that the strains related to phylogroup C are virulent and play an important role in development of colibacillosis. Instead, phylogroup B1 is more common in commensal strains and has less virulence. The B1 group strains isolated from YSI lesions had the lowest VS and had a negative association with two main APEC virulence genes (iss and ompT). Within the B2 phylogroup, the papC and papG genes were significantly more prevalent. pap gene clusters are located on the bacterial chromosome and are common virulence genes of group B2 UPEC strains. It should be noted that recent studies comparing the genomic features of APEC showed the multilineage evolution of pathogenic strains that makes comparison difficult. For example, O78 belongs to two lineages in phylogroups C and G (ST-23 and ST-117) and O1, and O2 belongs to three subpopulations within the third lineage in the B2 phylogroup (ST-95, ST-140, and ST-428/429) (62). Several studies have found that the APEC and UPEC strains are phylogenetically close and share some of the same VAGs and therefore should be considered potential zoonoses (7, 62–64).

Cluster analysis by UPGMA was performed on obtained data of ERIC-PCR, VAG-AMR traits, and VAGs to determine which assay is more suitable for the classification of strains. By ERIC-PCR, 15 clusters were identified. An integrated analysis revealed that, in many cases, the strains belonging to one cluster have high similarity in terms of virulence, resistance, source, and phylogenetic group. Cluster 7 contained the majority of phylogenetic group C that were isolated from colisepticemic lesions and also had high scores of virulence and AMR. Cluster 9 was mostly composed of AFEC strains and belonged to group A, with relatively high virulence and resistance scores. The strains that belonged to clusters 2 and 5 shared a similar origin and phylogenetic group and almost similar VAGs. Therefore, ERIC-PCR could be a practical approach for the initial classification of the strains in epidemiological studies. However, there are different viewpoints on ERIC-PCR discriminatory power for clonal differentiation. Some studies suggested the ERIC-PCR for investigation of clonal diversity and association between the phenotype and genotype of APEC strains (65, 66), while other researchers disagree (67).

In the present study, we tried to have a bird's eye view on the complicated APEC pathotype. Our finding indicates that virulence and AMR are not essentially related traits in avian E. coli and commensal strains may show greater resistance in some instances. This shows that resistance and virulence could evolve divergently, and therefore, AMR monitoring of poultry farms cannot only rely on the recovered strains of diseased birds. This is of particular interest where antimicrobial stewardship strategies have not been rigorously implemented. In the present study, strains related to phylogenetic group C had higher virulence and AMR scores compared to other groups, which indicates the importance of this phylogroup in both disease burden and AMR problem. Shared VAG reservoirs of APEC/AFEC and UPEC strains propose that avian E. coli could be potential reservoirs of VAGs for ExPEC strains in humans and therefore is a potential zoonotic agent. Further integrated studies on a larger number of strains along with assessment of more detailed genotypic features could potentially help us to upgrade our knowledge on virulence, resistance, and evolution of ExPEC.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

Ethical review and approval was not required for the animal study because Since we did not work on animals and the isolates were available from previously approved study re-submission for new approval was not necessary.

SR and MA designed the study and analyzed the data. MA, AN, and MS performed the study. SR wrote the first draft. SR, MA, and SP contributed to writing and critically reviewed the manuscript and assisted in analysis of data. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.