High genetic diversity among extraintestinal Escherichia coli isolates in pullets and layers revealed by a longitudinal study

Background Various information about the genetic diversity of Escherichia coli isolates from chickens are available but a detailed epidemiological investigation based upon isolates obtained from interrelated pullet and layer flocks is still missing. Therefore, in the course of a longitudinal epidemiological study on pullets and layers, 144 E. coli isolates from chickens with or without pathological lesions of the reproductive tract were serotyped and genotyped with pulsed-field gel electrophoresis (PFGE). These isolates were collected during rearing, peak and at the end of production. The actual study is the first of its kind so as to elucidate genetic relatedness among extraintestinal E. coli isolated from chickens with varying pathological conditions in interrelated layer farms/flocks at different stages of rearing. Results Serotyping revealed that 63.19 % of the isolates could not be assigned to any of the three serotypes tested whereas 30.55 % of the isolates belonged to serotype O1:K1, 4.86 % to O2:K1 and 1.38 % to O78:K80. After macrorestriction digest with XbaI, 91.66 % of the isolates were typeable resulting in 96 distinct PFGE profiles. Among them, five PFGE types included isolates collected from diseased chickens as well as from birds without pathological lesions. This finding shows that pathogenicity of E. coli in layers seems to be largely influenced by concurrent susceptibility factors. Furthermore, in six out of eight cases where two isolates were collected from each of eight birds, different PFGE types were found in the same or different organs of the same bird. The existence of predominant or persistent E. coli genotypes was only observed in two cases. Conclusions It is concluded that extraintestinal E. coli genotypes and serotypes in pullets and layers are heterogenous and also do not maintain a single clonality within the same bird. The facts that E. coli strains did not show any definite clonal population structure based on geographical region, age of the host and pathological lesions should have relevance in further epidemiological studies and control strategies.


Background
Escherichia coli isolates that are extraintestinal in nature are associated with the disease named colibacillosis that can infect all aged groups of chickens [1]. In layers, the pathogen is able to cause a systemic infection leading to fibrinous polyserositis, pericarditis, perihepatitis, salpingitis, peritonitis, salpingoperitonitis and a decrease in egg production ultimately leading to severe economic losses [2][3][4][5][6][7][8]. Despite serological diversities, serogroups such as O1, O2 and O78 are mostly implicated in disease conditions [9][10][11]. Until now, the pathogenicity of E. coli infection in chickens is not well understood. Several putative virulence and virulence-associated genes have been reported in avian pathogenic E. coli (APEC) [1,11,12]. However, the fact that a single genetic trait cannot separate disease-associated E. coli from commensal intestinal isolates raised certain concern on the definition of APEC as a single pathotype [13,14].
From an epidemiological point of view, understanding the clonal population structure of extraintestinal E. coli involving a longitudinal sampling scheme in interrelated rearing and laying flocks has a high priority. Thus we performed a longitudinal study in order to characterize the relatedness among E. coli isolates from systemic organs of pullets and layers kept in alternative housing systems in Austria. Beside the determination of the serotype, pulsed-field gel electrophoresis (PFGE) was applied for genetic fingerprinting which has higher discriminating power compared to other methods such as multilocus sequence typing [15]. PFGE is more applicable to investigate large-scale genomic diversity within a distinct population and has also been previously applied to infer molecular relatedness among APEC isolates in other geographical locations [16][17][18].

Methods
Flock history, sampling and E. coli isolation The present investigation was focused on extraintestinal E. coli isolates from pullets and laying hens kept in alternative husbandry system that were located in different provincial states of Austria. Six rearing and eight related layer farms comprising 15 layer flocks were included in the longitudinal study. Rearing farms are designated with letter "R" along with farm numbers as R I -R VI (e. g. R I is rearing farm 1). The layer flocks are designated with letter "L" along with the flock number and the corresponding rearing farm (e. g. L 1/I indicate for layer flock 1 that comes from rearing farm 1). Detailed information on farms and flocks is provided in Table 1.
In total, 188 birds were sampled for extraintestinal E. coli based on the sampling scheme as shown in Fig. 1. Sampling was performed during rearing (age of birds: 16-19 weeks), at the peak of production (age of birds: 37-42 weeks) and at the end of production (age of birds: 64-80 weeks). In each of the sampling events, five birds per rearing farm/layer flock were necropsied and sampled for extraintestinal E. coli. In two flocks of one layer farm (L 2/IV and L 3/IV ), additional samplings were included Most of the isolates were collected from ovary or oviduct (number of isolates n = 106) followed by liver (n = 25), lung (n = 10) and heart (n = 3). Details on E. coli isolates included in the present study are shown in Table 2. Isolates from rearing farms are marked with letter "R" along with farm number and bird number (e. g. R I -1 denotes for E. coli isolate collected from rearing farm 1 and bird number 1). Likewise, isolates from layer flocks are labelled with letter "L" along with flock number/corresponding rearing farm numbertime of sampling (A: peak of production, B: end of production, Z1 or Z2: first or second additional samplings)bird numberorgans (only in those birds from where two samples were collected). For instance, L 1/I -A-1 denotes for the isolate collected from layer flock 1 that originated from rearing farm 1; at the peak of production; bird number 1. Generally, one E. coli isolate per bird was included for further characterization. However, in the case of eight birds, two isolates per bird from the same or different organs were collected at the same sampling event: L 1/III -B-2-ovary1, Additionally, gross pathological lesions of the reproductive tract were recorded. Pullets from all six rearing farms did not show any gross pathological lesions. At the peak of production, some E. coli isolates originated from birds showing lesions in the reproductive tract, including egg peritonitis, inflammation of ovary and/or oviduct and degeneration of ovary and/or oviduct in 5, 17 and 7 birds respectively. Also, at the end of production, egg peritonitis, inflammation of ovary and/or oviduct and degeneration of ovary and/or oviduct were recorded in 15, 40 and 9 birds, respectively. In additional samplings, gross pathological lesions found in the oophoritis and salpingitis O1:K1 S10

E. coli confirmation of non-typeable genotypes
Partial sequencing of 16S rRNA gene was done in PFGE non-typeable isolates (n = 12) as described previously [19]. For this purpose, strains were grown on COS agar plates at 37°C for 24 h. DNA extraction was done from two to three colonies using DNeasy Blood and Tissue Kit (QIAGEN, Hilden, Germany) following manufacturer's recommendation. PCR was performed with a set of primers: 16S F 5'-GGCGGCRKGCCTAAYACATGC AAGT-3' and 16S R 5'-GACGACARCCATGCASC ACCTGT-3'. Amplification was carried out in 25 μl reaction volume consisting of 12.5 μl of HotStarTaq Master Mix (Qiagen, Hilden, Germany), 8 μl of nuclease free distilled water, 1 μl of each forward and reverse primers (10pmol/μl) and 2.5 μl of DNA template. The PCR thermocycler was programmed as: initial denaturation at 95°C for 15 min followed by 40 cycles of heat denaturation at 94°C, annealing at 60°C for 1 min and extension at 72°C for 1.5 min. Final elongation was performed at 72°C for 10 min. The PCR products were visualized by agarose gel electrophoresis. The gel slices were cut and purified using QIAquick® gel extraction kit (QIAGEN, Germany). Samples were then dispatched to LGC genomics GmbH (Berlin, Germany) for sequencing. The data obtained were processed with software Accelrys Gene v2.5 (Accelrys Inc) and analyzed with BLAST search in NCBI database.

Antimicrobial resistance (AMR)
Sixteen E. coli isolates originating from eight birds (two isolates per bird from the same or different organs) were investigated for the potential difference in AMR among strains isolated from the same organ (2 birds) or from different organs of the same bird (6 birds). The antimicrobial susceptibility test was performed using the disk diffusion method on Mueller-Hinton Agar (BioMeriéux, Vienna, Austria) according to Bauer et al. [20]

Subtyping of E. coli isolates
Serotyping revealed that 44 isolates (30.55 %) were grouped as O1:K1 while 7 (4.86 %) and 2 (1.38 %) strains belonged to O2:K1 and O78:K80, respectively. Furthermore, 91 isolates (63.19 %) could not be assigned to a definite serotype using these three antisera as they did not show agglutination (n = 79) or reacted positive with more than one anti-serum used (n = 12). Isolates that did not show agglutination with any or reacted positive with more than one anti-serum were assigned as nontypeable ( Table 2). The PFGE analysis of 132 E. coli isolates resulted in a heterogenous PFGE cluster: 96 E. coli profiles were obtained after macrorestriction digest applying XbaI while 12 isolates were non-typeable. The dendrogram obtained from the cluster analysis is shown in Fig. 2. The most abundant E. coli PFGE-profile was LA25 (n = 8) which included strains from three layer flocks (L 1/IV -B, L 2/IV -B and L 3/IV -B) that originated from a single rearing farm (R IV ). All these isolates were associated with lesions in the reproductive tract. Likewise, B1 included four isolates from birds with inflammation of ovaries in the same flock (L 3/III -A). Furthermore, E. coli genotypes which caused reproductive tract lesions in more than one laying bird at one sampling occasion from the same flock were: B9 (n = 2), LA11 (n = 2), LA18 (n = 3), LA21 (n = 2), S19 (n = 2), S33 (n = 2), S34 (n = 2) and S9 (n = 2).

DNA sequencing
The non-typable isolates were confirmed by partial sequencing of 16S rRNA gene as E. coli (99-100 % identity).

Discussion
An infection with E. coli in layers is regarded as one of the major problems in global poultry industry that might cause reproductive disorders referred as salpingitis/ peritonitis/salpingoperitonitis and peritonitis syndrome ultimately leading to severe economic losses on commercial farms [6]. In this regards, an epidemiological knowledge of the disease and disease causing agent is fundamental in order to develop effective control and prophylactic strategies. Here, we studied molecular epidemiology of E. coli isolates collected from pullets and layers in a longitudinal sampling study in Austria. Data obtained from genetic fingerprinting by PFGE were analyzed together with serotypes, geographical regions of isolation, and concurrent pathological lesions in each of the sampled birds.
In total, more than half of the E. coli isolates (n = 91/144) could not be assigned to a single serotype using antibodies against O1:K1, O2:K1 and O78:K80. Furthermore, for those isolates that could be assigned to one of the named serotypes, no correlation was found between a specific serotype and the occurrence of lesions in birds. In previous studies, it was also shown that E. coli isolates collected from diseased birds display a high serological diversity [16,21,22], demonstrating as high as 62 different O serogroups [21]. Thus classifying E. coli strains into a definite serotype might sometimes be somewhat challenging. Hence, our finding is in agreement with a previous notion that serotyping alone might not be helpful as a tool for characterization of E. coli [16].
In this study, the PFGE subtyping of E. coli isolates (n = 132) resulted in 96 XbaI profiles. Exclusively in two events, the same PFGE profile was seen in isolates from different sampling dates in mutually related farms/ flocks, indicating potential E. coli persistence. The PFGEtype S7 (n = 3) included isolates from pullets (n = 2, rearing farm R V ) without pathological lesions and from one layer in the corresponding flock L 1/V suffering from egg peritonitis and fibrinous oophoritis at the peak of production. In the second case, PFGE type S32 contained two isolates from the same layer flock (L 2/VI ) at the peak and end of production. One bird sampled at the peak of production showed inflammation of the ovary whereas egg peritonitis was diagnosed in the other birds necropsied at the end of production. These results indicate that some E. coli genotypes may retain in certain flocks at different stages of rearing but the associated pathological outcomes in birds can vary.
The genomic profile of extraintestinal E. coli with PFGE further revealed that strains collected from birds with pathological lesions can have 100 % genetic identity with strains that were collected from healthy birds. For instance, in PFGE type S10 (n = 3) in flock L 2/V , two birds did not have any lesions while one had oophoritis and salpingitis. Likewise in PFGE type B7 (n = 2) in L 2/III , one bird showed no lesions while in contrast, the other had egg peritonitis. Also, remaining isolates could not be grouped into distinct clonal clusters based on presence or absence of pathological lesions in sampled birds. This finding is in agreement with a previous study in broilers where authors have reported a high heterogenecity of E. coli isolates in broilers [13,23]. It can be hypothesized that pathogenicity of extraintestinal E. coli in chickens is highly dependent on concurrent environmental and host susceptibility factors. Providing a suitable opportunity in certain circumstances, E. coli residing in clinically healthy chickens might turn up into pathogenic. The hypothesis is further supported by an earlier finding in broiler that many collibacillosis associated isolates might not be clearly distinguished solely on the basis of presence of virulence associated genes as compared to intestinal commensal E. coli [13].
In the present study, we found no evidence for clonality of E. coli with respect to geographical locations of farms. Previously, Ewers et al. (2004) found only a limited number of E. coli clones to be distributed in poultry production in Germany [16]. In another study, it was reported that chickens with peritonitis in a single flock were likely to be infected by the same E. coli strain [24]. Different to this, we did not find clonality of E. coli isolates in birds from the same flock showing gross pathological lesions in the reproductive tract thus maintaining a high heterogenicity of PFGE types. Interestingly, we further noticed that a single bird can harbour two different PFGE types of E. coli in the same or different organs. Thus, the study demonstrated that a layer can be infected simultaneously by different E. coli genotypes. A similar finding was previously reported in broilers [18]. However, in another study in layers, one PFGE type was found to be present in bone marrow of an individual bird [17]. It might be that in some organs E. coli isolates possess less or no genetic diversity due to an adaptation process, which should however be further elucidated. In the present study, we also tested antibiotic susceptibility of 16 isolates that were collected from eight birds. All the isolates were sensitive to ceftiofur, colistin, gentamicin and spectinomycin but the resistant rate to tylosin was found 100 %. Mixed results were obtained for other antibiotics tested. MDR was seen in 3/ 16 isolates showing resistance to as high as five different antibiotics used. Although the number of isolates included for antimicrobial susceptibility test in the actual study is (See figure on previous page.) Fig. 2 PFGE cluster analysis of Escherichia coli isolates from pullets and layers (restriction enzyme XbaI). The TIFF images were compared using BioNumerics 6.6 software (Applied Math NV, Sint-Martens-Latem, Belgium), and normalized using the PFGE global standard Salmonella ser. Braenderup H9812. Pattern clustering was performed using the unweighted pair group method using arithmetic averages (UPGMA) and the Dice correlation coefficient was applied with a position tolerance of 1.0 %. Information provided adjacent to the dendrogram include the PFGE-type in combination with areas of isolation (S: Styria, LA: Lower Austria, Ua: Upper Austria, Sa: Salzburg, B: Burgenland, Ca: Carinthia), number of isolates (n) in each PFGE type, organs of isolation (source) and serotype [not applicable (na) are untypeable isolates]. Furthermore, E. coli isolation was classified according to the absence (1) or presence (2) of lesions in the reproductive tract Each isolate ID is designated with letter "L" along with the number of layer flock and rearing farm -A (sampling at the peak of production) or B (sampling at the end of production)number of sampled birdorgan of isolationisolate number (in case when two isolates were collected from the same organ). Antibiotic resistance pattern of two isolates from the same bird are in bold letter and highlighted if they showed different sensitivity to antimicrobials used. S: sensitive, I intermediate, R: resistant not very high, it already provides an indication for the problem of antibiotic resistance in E. coli towards commonly used antimicrobials. In a recent report from China, E. coli isolates collected from chickens were sensitive to relatively newer antibiotics such as cephalosporin but MDR rate was as high as 80.25 % [11]. The results from the present study further indicate that isolates collected from the same bird may not necessarily have identical antibiotic sensitivity profiles. Thus it can be suggested that testing of the antibiotic sensitivity profile from just one isolate per bird might not be enough to decide the most appropriate treatment.

Conclusions
Serotyping, antibiotic resistance test and genotypic fingerprinting of extraintestinal E. coli revealed that isolates exhibit high diversities within and between birds. As one bird can harbour different E. coli types an appropriate number of isolates should be considered for epidemiological studies and antibiotic sensitivity test.