The pathogenesis of E. coli is a multifactorial process depending on a variety of pathogenicity factors. A vast amount of already known and still unknown virulence determinants defines the virulence of a certain strain and thus the strength of the disease symptoms induced in the corresponding host organism. Although recent studies revealed considerable intersection between ExPEC pathotypes in general, the set of virulence genes present in pathogenic strains can differ considerably in terms of number and combination of genes [7, 8, 21]. Thus, the identification and characterization of additional virulence associated factors would still improve our understanding of the mechanisms underlying the pathogenicity and virulence of a certain group of E. coli strains.
Making use of two clinical strains, namely IMT5155 and CFT073, which differ with respect to host (avian versus human), pathotype (APEC vs. UPEC), O-type (O2 vs. O6), and multilocus sequence type (STC95 vs. STC73) in an SSH approach we identified an E. coli adhesin of the autotransporter family. The method of SSH enabled us to determine genes of the so far not sequenced APEC strain IMT5155 representing a well studied prototype strain isolated from a chicken in a German poultry flock which had experienced a severe outbreak of systemic E. coli infection [10, 16]. At the beginning of our studies, no sequence information was available for any APEC strain. Thus, SSH promised to be a useful tool to achieve sequence information about specific genes present in the avian pathogen but not in the human UTI strain albeit both being ExPEC strains. Indeed SSH has successfully been used in the past in many aspects, including the identification of virulence genes [22–25].
Among 28 DNA fragments that were specific for IMT5155 in our SSH approach, a 225 bp fragment, which showed similarity to putative adhesins, attracted our attention. Although in the run of our experiments a 98% identical adhesin gene as well as the functional role of its product in vitro and in vivo have been published by Li and colleagues , we still considered it important to complete our data as we observed some essential differences to the mentioned study.
Adhesins are involved in the first step of infection, allowing the primary and intimate contact of the pathogen with its host cell, initiating a pathogenic cascade. Their localization at the bacterial surface and their supposed function for enhanced gut colonization in terms of an intestinal reservoir make adhesins attractive candidates for vaccine developmental strategies .
APEC strain IMT5155 harbours the fim genes for the type 1 fimbriae, csg genes for curli fibers and the temperature-sensitive haemagglutinin (tsh) gene . It is interesting that IMT5155 lacks P, F1C, S and Dr fimbriae, known to be specifically involved in UPEC pathogenesis [16, 26, 27]. Thus, other, so far unidentified adhesins might play a role in IMT5155 pathogenesis. Indeed, we recently identified the yqi gene cluster encoding a fimbrial type of adhesin, called EA/I, that has been shown to confer an adhesive phenotype to a fim negative K-12 strain . Our data presented here show that autotransporter adhesin AatA might also play a certain role in the pathogenesis of APEC infections. In fact, few autotransporter type adhesins have been shown to be involved in APEC virulence to date. In 1994, Tsh which confers agglutination of chicken erythrocytes, was identified in APEC strain χ7122 . Later, Dozois and co-workers showed that Tsh probably contributes to the development of air sac lesions in birds . Furthermore, it turned out that the vacuolating autotransporter toxin Vat, identified in APEC strain Ec222 for the first time, was involved in the development of cellulitis in broiler chickens . Comparable to Tsh and Vat, AatA of APEC IMT5155 comprises all structural motifs characteristic for members of the family of autotransporter proteins: a signal peptide at the N-terminus, which would be recognized by the Sec secretion machinery; an autotransporter repeat, a passenger domain and a C-terminal translocation domain were predicted.
Adherence inhibition assay with a fusion protein containing the central part of the AatA protein confirmed the adhesive properties of AatA. This central part comprises the passenger domain, which is the secreted and surface-exposed protein part and thus the protein domain with supposed virulence function. While the translocation domain is highly conserved, the passenger domain demonstrates considerable sequence variation  making it a good candidate to gain specific antibodies against AatA.
By quantitative real-time PCR and immunoblot assays we could show that IMT5155 and APEC_O1 wildtype aatA are expressed under lab conditions, which stands in contrast to what Li et al. (2010) stated for APEC_O1 in their recent publication . These contradictory observations might in part be explained by different procedures used for antibody production, which could have led to a better detection of the AatA wild-type protein in our study. The failure to detect the very similar BL21-AatA protein with our antibody could be due to the low transcript level as indicated by qPCR experiments. Lower transcription might in turn have occurred due to sequence changes in the promoter regions in front of the IMT5155 and BL21 aatA ORFs, which in fact show only 70% identity. Adhesion assays with the fim negative strain AAEC189 expressing the IMT5155 AatA confirmed not only the adhesive properties for AatA but also the functionality of the predicted native promoter region. In addition, adhesion inhibition assays indicated a role for AatA as adhesin for IMT5155, which substantiates the findings of Li et al.  and indicates that the location of aatA, either on a plasmid or on the chromosome, does not seem to have any influence on the function of the adhesin, which has to be further investigated in the future.
The ability of bacteria to adhere to a diverse range of surfaces including different host tissues and abiotic elements is essential for colonization, survival and persistence [30, 31]. This is demonstrated by the enormous number of different adhesins known so far. It is assumed that a bacterial cell has such a huge set of diverse adhesive proteins to be able to adhere to different tissues and surfaces [15, 31]. Indeed the results of our adhesion inhibition assays supported this idea as blocking of IMT5155 and of DF-1 cells did not have a relevant effect on the adhesion property, showing that other adhesins are still effectively mediating adhesion.
An involvement of AatA in adhesion does not necessarily predict its vital importance for the virulence of a strain in vivo. However virulence, in particular with regard to ExPEC strains, is often a result of the interplay of several factors, with adhesion-related factors representing one of the most essential groups. Here, a number of adhesins are involved making it difficult to assess the contribution of one single adhesin to disease symptoms. However, for the 98% identical AatA of APEC_O1 its contribution to full virulence in chicken was shown .
One simple view is that one adhesin specifically mediates the adhesion to one specific receptor on the eukaryotic cell. This assumption led to the question if AatA isolated from APEC IMT5155, which enters the chicken via the respiratory tract, specifically recognizes proteins of the avian trachea and lung tissue. Interestingly, deduced from the amino acid sequence, AatA clustered together with Pertactin from B. pertussis, an adhesin which mediates binding to the lung epithelium of mammals (Figure 3; [32, 33]). As this is just a presumptive sequence-based finding, the identification of the host tissue receptor and its interaction with AatA has to be explored in future studies.
A number of publications claim that autotransporter adhesins are of special interest as they constitute an essential component of vaccines used in the medical area . Pertactin from Bordetella pertussis was the first autotransporter adhesin used as a vaccine . Also for Hap from H. influenzae elicitation of specific antibody titres was shown in mice . Indeed, new classes of these drugs are needed because of the increasing incidence of pathogenic organisms resistant to conventional antimicrobial agents; and it is believed that strains with genotypic resistance to the anti-adhesion agents will spread much slower than strains resistant to conventional drugs, such as antimicrobial substances . Here, we show that specific antibodies can be produced against AatA. Furthermore, we performed prevalence studies to verify if AatA fulfils criteria to serve as vaccine component from an epidemiological point of view. In contrast to the previously described novel adhesin gene yqi, initially identified in APEC strain IMT5155 , aatA was significantly associated with avian isolates, in that more than 90% of all positively tested strains were APEC and avian commensal strains, respectively, which is in accordance with the findings of Li et al. . Envisioning an intestinal prevention strategy that aims to combat pathogenic strains from colonizing the proposed intestinal reservoir, the frequent presence of aatA in avian commensal strains would basically contradict this idea, as the biological function of the physiological microbiota, including that of non-pathogenic E. coli strains, should not be diminished by such a vaccine. However, a high percentage of aatA positive strains was allocated to phylogenetic groups B2 and D. Avian commensal strains belonging to these groups have recently been shown to harbour an essential set of virulence genes and to be pathogenic for chickens . Thus, they represent pathogenic strains residing in the chicken intestine rather than fulfilling the criteria of non-pathogenic strains. In conclusion, AatA might not only be relevant to the adhesion of the upper respiratory tract of birds and subsequent pathogenic processes but seems to promote intestinal colonization, thereby contributing to the maintenance and transmission of pathogenic strains. A similar situation could be imagined for aatA positive E. coli strain B_REL606 that has been isolated from the human gut, but to our knowledge has not undergone further characterization in terms of potential extraintestinal virulence so far.
Li and colleagues found a significant association of aatA with isolates assigned to phylogenetic group D with 70% of APEC strains from this phylogenetic group being aatA-positive, and more than half of all aatA-positive strains belonging to phylogenetic group D . We observed a similar situation among our strain collection, while a distinction between different aatA-flanking region variants revealed that variant 1 (IMT5155) was more frequently observed in group B2 and D strains and variant 2 (BL21/B_REL606) in group A and D strains, while, although only rarely detected, the presumed episomal aatA variant 3 (APEC_O1) was linked with group B2 strains. Further large-scale analysis will have to rule out, whether the distribution of different aatA-flanking variants may be influenced by the phylogenetic background of the strains or by selective forces driven by environmental conditions, e.g. given in a certain host compartment.
The overall wide distribution of this supposed virulence associated gene in a large range of ExPEC and commensal isolates is indicative for a horizontal transfer of aatA. It is well established that virulence factors are often located on mobile elements, such as plasmids or pathogenicity islands and are thus often subjected to horizontal gene transfer . Sequence analyses of aatA and the flanking regions revealed a potential of mobility for the adhesin gene. In all completely sequenced E. coli genomes, where an aatA sequence was detected, the gene locus was enclosed by transposable elements. Furthermore, episomally located aatA variants might be transferred in the context of the whole plasmid, presuming the presence of functional transfer and mobility elements.
In addition, possible sequence variations among aatA genes of strains allocated to different phylogenetic groups might be reflected functionally, which has for example been shown for the genes of the fim cluster . Since aatA was retained in isolates of different phylogenetic groups, the discrete function of the protein in the respective strains, whether they commensally colonize the intestine or invade other internal organs of poultry and cause severe systemic infections, remains unsolved to date and should be subjected to thorough investigations in the future.
Many autotransporter adhesins are known to be relevant not only for adhesion but also for biofilm formation, invasion, aggregation and toxicity . Adhesins related to AatA, such as Hap, Ag43, AIDA and TibA, for example, contribute to bacterial aggregation by intercellular passenger domain interactions . Most trimeric autotransporter adhesins also seem to confer serum resistance by binding to components of the complement system . Although IMT5155 does not produce a biofilm under normal lab conditions, it remains to be determined if in vivo conditions might probably trigger this phenotype, enabling to investigate a possible role of AatA in this process. Although Li et al. suggested that AatA is not involved in autoaggregation or biofilm formation , it did not become evident whether they tested the wild-type and mutant strain, observing no difference, or whether the wild-type strain APEC_O1, comparable to IMT5155, did not show these phenotypes in general.