Phylogenetic relationship of Ornithobacterium rhinotracheale isolated from poultry and diverse avian hosts based on 16S rRNA and rpoB gene analyses

Ornithobacterium (O.) rhinotracheale is a relatively novel, emerging bacterial pathogen in turkeys and chickens that causes high economic losses to the commercial poultry production. It was first recognized and taxonomically classified in the early 1990’s [1, 2] and has been isolated from poultry flocks worldwide [3, 4].

The genus Ornithobacterium belongs to the family of the Flavobacteriaceae [5], which - besides others - also includes the genus Riemerella with R. anatipestifer [6, 7] and the genus Coenonia with C. anatina [8]. Both bacterial species are also important poultry pathogens mainly of domestic ducks and geese. Besides a new species proposed as Candidatus Ornithobacterium hominis sp. nov. [9], O. rhinotracheale is the only bacterial species described within the genus Ornithobacterium but not fully understood to date. Because of its veterinary importance particularly for the global turkey meat industry [10] the need for reliable diagnostic and characterization methods is obvious for early treatment. Its cultural characteristics and fastidious requirements (i.e., small colony size, slow growth, enriched media and capnophilic incubation), however, may challenge bacterial isolations and reduce the detection rates [11, 12]. Therefore, molecular detection of O. rhinotracheale DNA from tissues or swabs targeting the 16S rRNA gene with specific primers is frequently used in routine diagnostics, but 16S sequences often lack the resolution to capture heterogeneity among the strains [13–15].

Over the past few decades, the host range of birds infected by O. rhinotracheale or carrying the bacterium in their respiratory tract has constantly increased [16]. The presence of O. rhinotracheale in apparently healthy, captive and free-ranging non-galliform species raises important epidemiological and taxonomic questions for a better understanding of its diversity, ecology and transmission cycles. Multilocus sequence typing (MLST) recently established by Thieme et al. [3, 16] not only revealed specific phylogenetic relationships in non-galliform birds such as pigeons and birds of prey, but also identified few strains from turkeys and chickens that clearly differed from the main poultry cluster. The aim of the present study was to introduce partial rpoB gene sequencing for O. rhinotracheale into routine diagnostics to differentiate strains isolated from poultry and more diverse avian hosts and to compare the results with those from 16S rRNA gene and MLST analyses. The rpoB gene has been proved very useful and powerful for bacterial identification and phylogenetic studies [17–19].

Within the 16S rRNA gene analysis, the 65 O. rhinotracheale specific partial sequences showed a high level of homogeneity with similarity values ranging from 98.6 and 98.7% (ten strains from pigeons, GV37 and GV38) to 100% (strains from turkey or chicken and GV1) between non-galliform and poultry species. The structure of the phylogenetic tree (Fig. 1), however, was largely comparable to MLST results [16]. The majority of O. rhinotracheale strains were arranged in one cluster and separated from two subsets of three strains each isolated from poultry (family Phasianidae) or birds of prey (family Accipitridae), respectively. One of the O. rhinotracheale serotype reference strains (RefF) belonged to the aforementioned subset of the three strains from poultry (including GB 978/14/1 and GB 137/10/2) and formed a separate cluster in previous studies too [14, 15].

The main 16S cluster was split into two subclusters with 47 strains collected from poultry, birds of prey and corvids, which are more closely related to each other and comprise a high diversity of bird families and species. In contrast, the second subcluster only involved 11 strains from pigeons and one strain from a white-tailed eagle (GV37, family Accipitridae). The latter grouped together with two strains isolated from feral pigeons, which were separated from strains collected from pigeons of a pigeon loft [16].

The 65 partial rpoB gene sequences were heterogeneous and ranged in their similarity values from 85.1 to 100%. The structure of the phylogenetic tree was in strong correlation with MLST results [16]. Three distinct clusters A (n = 6; poultry and birds of prey), B (n = 47; poultry, birds of prey and corvids) and C (n = 11; only pigeon) were identified (Fig. 2) that corresponded well to the MLST clusters [16]. Strain GV37 formed a distinct lineage and only shared 86.2 to 87.4% sequence similarity with strains of cluster A, 86.8 to 87.5% similarity with strains of cluster B and 85.1 to 85.5% similarity with strains of cluster C (Table 2). Within cluster A, three strains from turkeys and chicken (98.7 to 100% similarity) were clearly separated from three strains collected from birds of prey (99.4 to 100% similarity). Both groups shared only 94.2 to 94.8% sequence similarity. In contrast, rpoB gene sequence similarity values among strains from cluster B or cluster C ranged from 98.5 and 98.0% to 100%, respectively.

Sequence analysis of a specific rpoB gene fragment is widely used in addition to the 16S rRNA gene for more reliable bacterial identification and taxonomic classifications at genus and species level [17]. For members of the family Pasteurellaceae, rpoB similarity cut-offs of 85 and 95% were recommended for the description of new genera and species, respectively [23]. For the family Flavobacteriaceae, corresponding similarity cut-offs have not been defined so far [24]. Partial rpoB gene analyses, however, proved to be very useful to reveal clear phylogenetic relationships of Riemerella strains and related members of Flavobacteriaceae [19]. Likewise, O. rhinotracheale strains from this study show strong differences in similarity among their rpoB gene sequences. Strains of cluster A and cluster C as well as strain GV37 were clearly separated from cluster B, a genetically homogenous group that includes the majority of O. rhinotracheale strains and the O. rhinotracheale type strain DSM 15997^T. Moreover, comparisons with O. rhinotracheale DSM 15997^T or respective with strains from cluster B revealed sequence similarities of ≤90.0% but above 85.0% (Table 2), which would only support bacterial identification at genus level.

By using partial sequences from a single gene, the rpoB gene analysis is in good agreement with MLST results with a slight decrease in resolution to distinguish more similar strains. Eight rpoB gene sequences were received from swab samples (all non-galliform birds) [20] extending the avian host range and phylogenetic relationships investigated by MLST. On a bird family-level, the Columbidae specific cluster C and two different subclusters of three strains from Accipitridae species (namely 165–2/2015, GV8 and GV13) and seven strains from Falconidae species (namely GV9 to GV12, GV55, GV143 and GV149) were seen likewise to MLST [16]. At the same time, however, several O. rhinotracheale strains from different poultry species, birds of prey and corvid species have a close genetic relationship (similarity ≥98.5%) pointing towards the risk of interspecies transmission. Moreover, our results provide strong evidence that traditional phenotypic and genetic methods used for identification may not properly represent the heterogeneous group of bacteria classified as O. rhinotracheale. From housekeeping gene analyses, it is very likely that the genus Ornithobacterium includes additional species and partial rpoB gene sequencing can be recommended as fast, cost-effective and readily available method to identify strains and differentiate between O. rhinotracheale and Ornithobacterium-like bacteria.