Introduction

Species belonging to the genus Pasteurella colonize mainly the upper respiratory tract and the oropharynx of animals, including cats and dogs [4]. Besides their potential pathogenicity to these hosts, they may behave as opportunistic pathogens to humans, causing bite wound infections, complications of underlying respiratory disorders, and systemic infections. Among this group, Pasteurella multocida is most frequently reported, although other species (P. canis, P. stomatis, and P. dagmatis) may also be involved [12]. In the medical literature there is a range of reports describing human infections caused by P. dagmatis. Most of these conditions have resulted from cat and dog bites [11, 23]. Other infections caused by P. dagmatis include peritonitis [3, 24], respiratory tract infection [1], and endocarditis [20].

Pasteurella dagmatis displays some phenotypic similarity (e.g., positive test for urease) to P. pneumotropica and Bisgaard Taxon 46, so that these three taxa might be easily misidentified [6, 11]. P. pneumotropica is associated predominantly with rodents [4], while bacteria belonging to Bisgaard Taxon 46 were isolated from leopard bite wounds in humans [6].

Accurate identification of the implicated pathogen is of great concern from an epidemiological point of view. In the case of pasteurellae, the phenotypic identification is quite cumbersome, even when automated identification systems are used [7, 11]. For this reason, many laboratories employ molecular methods for identification of Pasteurella species, especially sequence analysis of the 16S rRNA gene [16, 24]. Nonetheless, a prerequisite for reliable use of this method is a proven and comprehensive database which comprises a sufficient number of DNA sequences. As an example, it was shown, that strains of Pasteurella multocida originating from different animal species may differ genetically, even with regard to a highly conservative 16S rRNA gene [8]. A similar observation was made in the case of P. dagmatis [22].

DNA sequence comparison of housekeeping genes has proven to be another valuable tool for phylogenetic investigation and identification of different bacteria, including Pasteurellaceae [5]. It was found that the rpoB gene (encoding the ß-subunit of RNA polymerase) may have a higher discriminatory power than 16S rRNA sequences [17], constituting a reliable complement to the 16S rRNA phylogeny [13].

The objective of this study was to undertake a comparative sequence analysis of the 16S rRNA and rpoB genes of P. dagmatis (isolates of different host origin), to assess their phylogenetic relationship to Pasteurella sensu stricto, P. pneumotropica, and Bisgaard Taxon 46 as well as finding out molecular methods for differentiation of various P. dagmatis subpopulations.

Materials and Methods

Bacterial Strains and Growth Conditions

The study was performed on 15 field isolates, phenotypically identified as P. dagmatis, and the reference strain of this species (P. dagmatis CCUG 32658) obtained from the Culture Collection, University of Göteborg, Sweden. Most of the field P. dagmatis isolates originated from both diseased and healthy cats and dogs living in the Wrocław and Poznań areas of southwest Poland. Among them, eleven strains were isolated from cats and three from dogs. One strain was isolated from the oral cavity of a tiger kept in the Wrocław Zoo. All data concerning strains used are listed in Table 1. All isolates were grown on 5% sheep blood trypticase soy agar under aerobic conditions and subsequently identified phenotypically and genotypically.

Table 1 Bacterial strains used in this study
Full size table

Phenotypic Examination

Phenotypic identification included Gram-staining, catalase, oxidase, production of urease (in Christensen’s Medium supplemented with liver digest and glucose [10], production of indole (in Tryptophane Broth [Difco Laboratories, Detroit, MI] with subsequent addition of Ehrlich’s reagent), ornithine decarboxylase (using diagnostic tablets ODC, Rosco Diagnostica, Taastrup, Denmark), and production of acid from the following carbohydrates: glucose, sucrose, mannose, maltose, mannitol, sorbitol, and trehalose (in CTA Medium [Becton–Dickinson, Le Pont de Claix, France], supplemented with 1% of appropriate sugar). Results were observed for up to 3 days.

Extraction of DNA

After an overnight cultivation on blood agar, bacterial DNA was extracted using Genomic Mini (A&A Biotechnology, Gdynia, Poland) according to the manufacturer’s instruction.

Amplification and Partial Sequencing of the 16S rRNA and rpoB Genes

For amplification of 1403- and 560-bp fragments of the 16S rRNA and rpoB genes, respectively, previously described primers [13, 14] were used. The reaction mixture (25 μl) contained 10 mmol/l Tris–HCl, pH 8.8, 1.5 mmol/l MgCl2, 50 mmol/l KCl, 0.08% Nonidet P40 (Fermentas, Vilnius, Lithuania), 5 pmol of each primer (Institute of Biochemistry and Biophysics, Warsaw, Poland), 0.2 mmol/l of each deoxyribonucleotide (Fermentas), 2 U of Taq DNA polymerase (Fermentas), and 2 μl of DNA. Forty PCR cycles of denaturation at 94°C for 30 s, annealing at 50°C for 30 s, and elongation at 72°C for 120 s were performed. PCR products were purified (by adding 10 U of E. coli-Exonuclease I and 2 U of Shrimp Alkaline Phosphatase [Fermentas] to 5 μl of the reaction mixture) and sequenced—on both DNA strands—using the DYEnamic ET terminator cycle sequencing kit ABI Prism™ (Amersham Biosciences Europe GmbH, Germany).

Molecular Identification and Analysis of Nucleotide Sequences

Searches for homologous DNA sequences in GenBank were performed using the BLAST algorithm [2]. All sequences were subjected to multiple alignment using ClustalW program (http://www.mbio.ncsu.edu/bioedit/bioedit.html). For comparative studies, a 1308-bp fragment of the 16S rRNA gene and 501-bp fragment of the rpoB were selected (corresponding to positions 83–1390 of the E. coli 16S rRNA sequence, accession number J01859, and positions 1543–2043 of the E. coli rpoB gene, accession number AB488804, respectively). Phylogenetic analyses were performed using the MEGA version 3.1 software [15]. Dendrograms were generated by the neighbor-joining method, with bootstrap analysis (corresponding to 500 replications), included to assign confidence values.

Digestion of PCR Products using Restriction Endonuclease

A restriction map of 1403-bp 16S rRNA sequences, obtained from feline, canine, and tigrine P. dagmatis isolates, was constructed using BioEdit Software (http://www.mbio.ncsu.edu/bioedit/bioedit.html). For digestion, 476-bp fragments of this gene containing polymorphic site for TaqI restriction enzyme were selected (the following oligonucleotide primers were used: for, 5′-CGAACGGTAGCAGGAAGAAAGCTTG-3′; rev, 5′-GWATTACCGCGGCKGCTG-3′). Thirty-five PCR cycles of denaturation at 94°C for 30 s, annealing at 50°C for 30 s, and elongation at 72°C for 30 s were performed. The PCR products were subsequently digested in a mixture containing 1 μl (10 U) of TaqI restriction enzyme (Fermentas), 3 μl of 10× TaqI Buffer, 10 μl of PCR mixture, and 16 μl of water. Reaction was performed at 65°C for 60 min, and 15 μl of digestion mixture was resolved in 2.5% agarose gel stained with ethidium bromide.

Results

Phenotypic Identification

All isolates identified phenotypically as P. dagmatis were catalase-, oxidase-, indole-, and urease-positive. The isolates produced acid from glucose, sucrose, maltose, mannose (the feline isolate 6/4 was negative), and trehalose (with the exception of the reference strain CCUG 32658). Two strains (the feline isolate 6/4 and the strain from a tiger) also produced acid from sorbitol, and one canine isolate (149/1)—from mannitol. All strains (except for the isolate 505 recovered from a dog) were negative for ornithine decarboxylase.

Molecular Identification and Phylogenetic Study

Analysis of the 16S rRNA nucleotide sequences using the BLAST algorithm revealed that canine isolates have a high degree of sequence similarity (99.78–99.85%) with the P. dagmatis type strain AY362920. In contrast, all feline isolates were practically identical (99.47–99.85%) to the P. pneumotropica NCTC 10827 strain (accession number in GenBank AF224296). AF224296 was also the highest scoring sequence for the strain isolated from a tiger. In this case, the similarity was lower and amounted to 98.58%.

16S rRNA sequences of feline isolates were highly homologous, with the only polymorphic nucleotide site (A/T substitution) located in position 842 (relative to AF224296). It was found that feline P. dagmatis sequences display several unique nucleotide sites which distinguish it from canine P. dagmatis strains and distinguish it from the isolate recovered from a tiger (Table 2).

Table 2 Distribution of polymorphic nucleotide sites among 16S rRNA sequences of P. dagmatis of different host origin
Full size table

A phylogenetic analysis of 16S rRNA sequences was performed with 36 sequences, including 16 isolates determined in the present study and 20 sequences deposited in GenBank (14 sequences of Pasteurella sensu stricto core group, four of P. pneumotropica, and two of Bisgaard Taxon 46). An evolutionary tree constructed by the neighbor-joining method was shown on Fig. 1. Of the 19 nodes, 15 were also supported by bootstrap values higher than 50%. Analyzed sequences were divided into nine clusters. Cluster IA was composed of the P. pneumotropica strain NCTC10827, two strains recovered from cats, designated as P. dagmatis-like 5/8 and 8/4 (sequences retrieved from GenBank), and 11 feline P. dagmatis isolates analyzed in this study. Members of this group displayed only a 97.48–97.60% sequence similarity with the P. dagmatis type strain (AY362920). Cluster IB contained the strain 197 from a tiger and the feline P. dagmatis-like strain 1/1 (from GenBank). Pasteurella stomatis and P. canis constituted clusters II and III, respectively. Four canine P. dagmatis isolates, characterized by the present authors, as well as the type strain of this species (CCUG12397T, sequence obtained from GenBank), fell into cluster IV. Clusters V, VI, and VII were composed of Bisgaard Taxon 46, P. multocida subsp. septica, and P. multocida subsp. multocida/P. multocida subsp. gallicida complex, respectively. Three sequences of P. pneumotropica (other than AF224296) were grouped in separate clusters VIII and IX with low similarity to all other sequences of the Pasteurella sensu stricto core group.

Fig. 1
figure 1

Phylogenetic tree of 36 Pasteurella strains based on 16S rRNA sequences (1308 bp fragments). The tree was constructed using the neighbor-joining method. Bootstrap values of 500 replications are indicated as percent confidence values for particular branching. Sequences indicated in bold were determined in this study. CCUG Culture Collection, University of Göteborg, Sweden; NCTC National Collection of Type Cultures, UK; MCCM Medical Culture Collection Marburg, Germany; ATCC American Type Culture Collection, USA; CNP Centre National des Pasteurella, France; CDC Centers for Disease Control and Prevention, USA; T type strain; (F) isolates from cats; (C) isolates from dogs; (T) the isolate from a tiger

Full size image

Comparison of rpoB sequences confirmed the genetic distinctness of the feline subpopulation of P. dagmatis. All the isolates recovered from cats had the same nucleotide sequence of the rpoB gene that displayed a 97.4% similarity (12 mismatches on a total of 501 nucleotides compared) with the P. dagmatis type strain CCUG12397T (AY362966). On the other hand, sequences of the canine isolates showed a 98.6–99.6% similarity (2–7 mismatches out of 501 nucleotides) to AY362966. The isolate from a tiger was most distantly related and had only a 94.4% similarity with P. dagmatis CCUG12397T (28 mismatches).

The rpoB gene tree (Fig. 2) was strongly supported by high bootstrap values (higher than 80% accounted for 12 out of 17 nodes). In this dendrogram, canine and feline isolates of P. dagmatis also clustered in separated groups. However, unlike the phylogeny derived from the 16S rRNA gene, an analysis of the rpoB sequences revealed that both canine and feline subpopulations of this species are more closely related to each other than to other members of the Pasteurella sensu stricto core group. The isolate from a tiger constituted a distinct P. dagmatis-like taxon, less closely related to the above P. dagmatis subgroups. The remaining rpoB gene sequences analyzed in this study were grouped in separate taxon-specific clusters. In both 16S rRNA and rpoB evolutionary trees, strains of P. pneumotropica (other than NCTC10827) constituted a group clearly different from feline isolates of P. dagmatis.

Fig. 2
figure 2

Phylogenetic tree of 28 Pasteurella strains based on rpoB sequences (501 bp fragments). The tree was constructed using the neighbor-joining method. Bootstrap values of 500 replications are indicated as percent confidence values for particular branching. Sequences indicated in bold were determined in this study. CCUG Culture Collection, University of Göteborg, Sweden; NCTC National Collection of Type Cultures, UK; CDC Centers for Disease Control and Prevention, USA; T type strain; (F) isolates from cats; (C) isolates from dogs; (T) the isolate from a tiger

Full size image

Restriction Fragment Length Polymorphism of the16S rRNA Gene

The restriction nuclease TaqI was shown to cleave 476-bp fragments of the 16S rRNA gene from canine P. dagmatis isolates, giving products of 125 and 351 bp. Amplicons obtained from DNA of feline isolates (and the strain from a tiger) remained undigested (Fig. 3).

Fig. 3
figure 3

Agarose gel electrophoresis of a 476-bp fragment of the P. dagmatis 16S rRNA gene. Lanes 1–8 represent amplicons obtained from DNA of canine isolates, lanes 9–16 from DNA of feline isolates, and lanes 17–18 from DNA of the isolate from a tiger. For each isolate, intact amplicons (lanes with odd numbers) and TaqI-digested ones (lanes with even numbers) were run. M size marker (GeneRuler™100 bp DNA Ladder [Fermentas])

Full size image

Discussion

Comparison of 16S rRNA and rpoB gene sequences reveals that P. dagmatis is a heterogeneous species, containing at least two host-specific lineages. In this study, it was found that canine and feline isolates of P. dagmatis, recovered from animals living in Poland, differ genetically from one another. In addition, one strain from a tiger was found that seems to represent yet another taxon related to P. dagmatis. Key biochemical characteristics are very similar among particular groups of this species. This means that they are indistinguishable under routine phenotypic identification.

Although P. dagmatis has been isolated from different domestic animals [4, 9] as well as from human clinical specimens [23, 24], the genetic diversity of the species has not been explored extensively so far. Recently, atypical P. dagmatis isolates recovered from cats were described [22]. While the basic phenotypic properties of these strains were characteristic of P. dagmatis, they were referred to as “P. dagmatis-like” due to unusual colony morphology. Sequencing of the 16S rRNA gene of three selected isolates showed only a 97% similarity with known sequences of P. dagmatis; instead, they were strikingly similar to the reference strain P. pneumotropica NCTC10827 (GenBank accession AF224296). The results of the present investigation revealed that all feline P. dagmatis strains constitute a coherent group displaying almost 100% similarity of the 16S rRNA gene with sequences of Hungarian P. dagmatis-like strains 5/8 and 8/4 (GU177868 and GU177869) and with the sequence of P. pneumotropica NCTC10827. In contrast, all canine P. dagmatis isolates had the same 16S rRNA sequence that is consistent with sequences of P. dagmatis from GenBank. This might be of phylogenetic and epidemiologic significance, as it indicates that strains of P. dagmatis or P. dagmatis-like, isolated from cats (at least in Hungary and Poland) constitute a monophyletic group, differing from canine isolates. Whether or not this phenomenon is unique to Central Europe remains to be elucidated. Some evidence indicates, however, that feline-specific P. dagmatis strains have also been isolated in Western Europe. As an example, a case of peritonitis in a human caused by P. dagmatis was reported [24]. According to a case history, cats were suspected of being the source of infection. The 16S rRNA gene of the isolated strain was sequenced, giving only 97% similarity with sequence of P. dagmatis from GenBank.

The results of the investigation indicate that P. pneumotropica NCTC10827 may have been misclassified and its 16S rRNA gene sequence incorrectly labeled. Feline isolates that were identified phenotypically as P. dagmatis (both Hungarian [22] and those tested in this study), have over a 99.4% similarity of the 16S rRNA gene with the above strain, being identified as P. pneumotropica using the BLAST algorithm. P. pneumotropica is encountered predominantly in rodents yet its biovar Heyl was also isolated from a cat [4]. This species is positive for ornithine decarboxylase [11, 21], while all the feline isolates were ODC negative. Similarly, negative ODC-reaction was found in Hungarian P. dagmatis-like strains as well as in P. pneumotropica NCTC10827 [22]. NCTC10827 was reported to be an atypical strain, even within heterogeneous P. pneumotropica species [18, 21]. The similarity of the NCTC10827 16S rRNA gene with other P. pneumotropica sequences (deposited in GenBank) is relatively low, reaching at most 94%. Analysis of the sodA gene also confirms the genetic distinctness of the NCTC 10827 strain from other P. pneumotropica isolates [22], indicating that it should be placed within the Pasteurella sensu stricto core group. The strain NCTC10827 was isolated from human blood and identified based on its biochemical properties (mainly a positive urease reaction) [19]. The ability to produce gas from some carbohydrates, reported by Rogers et al. [19], could also argue that this strain belong to the P. dagmatis-like group rather than to P. pneumotropica.

In both phylogenetic trees based on the 16S rRNA and rpoB genes, all feline isolates investigated here constitute a well-separated cluster within the Pasteurella sensu stricto core group. Described genetic differentiation of P. dagmatis may be of some importance in epidemiologic studies, especially in cases of P. dagmatis-infections of unknown origin in humans. Feline and canine isolates were distinguishable using the restriction enzyme TaqI, which could be used for laboratory diagnostics.

It seems that the strain 197, isolated from a tiger, constitutes another evolutionary lineage within P. dagmatis. Based on 16S rRNA nucleotide composition, it has the highest level of similarity with the P. dagmatis-like isolate 1/1 GU177867 (99.08%) and P. pneumotropica NCTC10827 (98.58%). The isolate from a tiger differed phenotypically from Taxon 46 in positive sucrose fermentation. In addition, results of phylogenetic analyses based on both 16S rRNA and rpoB-trees indicate that the isolate from a tiger is more closely related to P. dagmatis than to Bisgaard Taxon 46.

In summary, it seems that there are several subpopulations within P. dagmatis, associated with specific hosts. Further studies on this organism which take into consideration isolates of various geographical distribution and host range should be performed. It would be interesting not only for phylogenetic analyses but also from epidemiological point of view, as P. dagmatis and P. dagmatis-like bacteria are involved in a number of bite wounds and other infections in humans.