Extension of the species Helicobacter bilis to include the reference strains of Helicobacter sp. flexispira taxa 2, 3 and 8 and Finnish canine and feline flexispira strains

The taxonomy and diversity of the genus Helicobacter has been extended during the last 10 years to contain 23 species with validly published names (; last updated January 2004). Helicobacter aurati, Helicobacter bilis, Helicobacter canis, Helicobacter cinaedi, Helicobacter cholecystus, Helicobacter fennelliae, Helicobacter hepaticus, Helicobacter marmotae, Helicobacter muridarum, Helicobacter pametensis, Helicobacter pullorum and Helicobacter trogontum colonize the lower intestinal tract of various animal species (Solnick & Schauer, 2001; Fox et al., 2002). Some of these species seem to be strictly host-adapted, such as H. canis in dogs and H. pametensis in birds (Solnick & Schauer, 2001). Certain enterohepatic Helicobacter species, e.g. Flexispira rappini and H. pullorum, have shown potential for zoonotic transmission (Archer et al., 1988; Romero et al., 1988; Stanley et al., 1994). Moreover, DNA of H. bilis, H. hepaticus, F. rappini, H. pullorum and Helicobacter pylori has been detected by PCR from the bile and gall bladder of human patients with various hepatobiliary diseases (Fox et al., 1998; Ljungh & Wadström, 2002; Silva et al., 2003). An increased seropositivity for non-gastric Helicobacter species has also been detected in serum of patients with autoimmune liver diseases (Nilsson et al., 2003). The host spectrum of several intestinal Helicobacter species is not fully known, due to limited studies on a large variety of animals. Cultivation of intestinal Helicobacter species requires fresh samples and adequate culture techniques, limiting the knowledge gained on their taxonomy, ecology and epidemiology.

F. rappini is a provisional name given by Bryner and colleagues (Bryner, 1987; Bryner et al., 1988) to a group of organisms originally isolated from aborted sheep fetuses (Kirkbride et al., 1986). Flexispiras, which have spindle-shaped cells surrounded by periplasmic fibrils and bipolar tufts of sheathed flagella, were later shown to be members of the genus Helicobacter (Paster et al., 1991; Dewhirst et al., 2000a). The flexispira group includes the four named species H. bilis (Fox et al., 1995), H. trogontum (Mendes et al., 1996), H. muridarum (Lee et al., 1992) and H. aurati (Patterson et al., 2000), originally isolated from mice, rats, mice and hamsters, respectively, and several unnamed taxa based on 16S rRNA gene sequence analysis (Dewhirst et al., 2000a). Recently, the species H. trogontum was extended when a polyphasic approach in studies of their taxonomy was used to include some porcine Helicobacter strains as well as the reference strains of provisional taxa 1, 4 and 5 (Hänninen et al., 2003). H. bilis and taxa 2, 3 and 8 have been shown to produce cytolethal distending toxin but no toxin production was found in H. trogontum (Kostia et al., 2003). Dewhirst et al. (2000a) proposed the provisional name Helicobacter sp. flexispira taxa 1 to 8 and 10 for unnamed Helicobacter species with flexispira morphology based on 16S rRNA gene sequences. A group of Helicobacter sp. flexispira without validly published taxonomy has been isolated from pigs (taxon 2), sheep (taxon 3), dogs (taxa 7 and 8) and humans (taxon 8) (Dewhirst et al., 2000a). Bacteria with flexispira morphology have also been isolated from blood of patients with bacteraemia (Tee et al., 1998; Sorlin et al., 1999; Weir et al., 1999) and in faecal samples of patients with diarrhoea, and their DNA has been detected in bile and liver samples in association with various hepatobiliary diseases (Fox et al., 1998; Matsukura et al., 2002).

Analysis of 16S rRNA gene sequences has been one of the most common methods used for studies of phylogeny and taxonomy of bacteria, including Helicobacter species (Vandamme et al., 1996; Dewhirst et al., 2000b; Stackebrandt et al., 2002). However, several examples among Helicobacter indicate that 16S rRNA gene sequences are not good markers for species identification of certain Helicobacter species. Helicobacter bizzozeronii and Helicobacter salomonis, for instance, have sequences that can be more than 99 % similar (Jalava et al., 1997). By contrast, the intraspecies diversity within H. cinaedi and H. trogontum can be 4 %, which is higher than the interspecies diversity between some enterohepatic Helicobacter species (Vandamme et al., 2000; Hänninen et al., 2003). The name flexispira is sometimes used for bacteria that do not have spiral morphology or fibrils outside the cell and that have only one flagellum at each end of the cell, based on the high similarity of their 16S rRNA gene sequence to a strain of Helicobacter sp. flexispira taxon 8 (Iten et al., 2001).

Here, we report, using a polyphasic approach, that Helicobacter sp. flexispira taxa 2, 3 and 8, some canine and feline flexispira strains and H. bilis constitute a species. In addition, we found that a flexispira strain isolated from a dog differs from H. bilis in several genetic characteristics.

Bacterial strains.
Bacteria with flexispira morphology were isolated from 14 dogs and cats using either a filtration technique on Brucella blood agar or culture of faecal material on freshly prepared Brucella blood agar medium containing (ml¹) 2·5 U polymyxin B, 5 µg trimethoprim, 10 µg vancomycin and 2 µg amphotericin B (Hänninen et al., 2003). The medium was incubated at 37 °C for up to 7 days in a microaerobic atmosphere containing 5 % oxygen, 10 % CO₂ and 3 % H₂ and observed for spreading growth. Bacteria with flexispira morphology were identified using light microscopy and certain cultures (F56, F57, KI900, KO220, KI520, KI311 and KO214) were later confirmed by transmission electron microscopy to have typical flexispira morphology (Hänninen et al., 2003). During earlier studies, the strains were tested several times for catalase, oxidase and urease production (Hänninen et al., 1996, 2003). In addition, the strains were characterized for the following features: nitrate reduction, urease, γ-glutamyl transferase and esterase using API CAMPY (bioMérieux), growth at 25 and 42 °C and on 1 % glycine and sensitivity to nalidixic acid and cephalothin (Hänninen et al., 1996, 2003).

Sequencing.
For sequencing studies, DNA was isolated as described elsewhere (Hänninen et al., 1996, 2003; Mikkonen et al., 2004). The nearly complete (at least 1400 bases sequenced) 16S rRNA genes of eight Finnish canine and feline flexispira strains were amplified by PCR with universal primer pair F19-38 (5'-CTGGCTCAGGAYGAACGCTG-3') and R1541-1522 (5'-AAGGAGGTGATCCAGCCGCA-3'). Sequencing of the purified (QIAquick PCR purification kit; Qiagen) PCR products was performed by Sanger's dideoxynucleotide chain-termination method using primers F19-38, R1541-1522, F908-926 (5'-AACTCAAAGGAATTGACGG-3') and R536-519 (5'-GTATTACCGCGGCTGCTG-3'). Samples were run in a Global IR² sequencing device with e-Seq 1.1 software (LiCor) according to the manufacturer's instructions. Overlapping complementary sequences were joined using the Align IR 1.2 program (LiCor). The consensus sequences of strains belonging to Helicobacter species and Campylobacter jejuni (outgroup) [retrieved from/deposited in GenBank () using BLASTN 2.2.6 (Altschul et al., 2001)] were aligned and a phylogenetic tree was constructed from the global alignment by the neighbour-joining algorithm using the BioNumerics 3.0 software package (Applied Maths). Bootstrap probability values were calculated from 1000 resampled trees. Fig. 1 shows the distance matrix tree based on 16S rRNA gene sequences and the accession numbers of the 16S rRNA gene sequences used/deposited. The specificity of the primers designed for identification of H. bilis based on 16S rRNA gene amplification was tested by the method of Fox et al. (1995).

(44K): Fig. 1. Phylogenetic tree of representative enterohepatic Helicobacter species based on 16S rRNA gene sequences (>1400 nucleotides). The tree was constructed using a neighbour-joining method rooted with Campylobacter jejuni. GenBank accession numbers are given in parentheses. Numbers at each node represent percentages of bootstrap support calculated from 1000 probable trees. Bar, 2 % sequence divergence.

Partial sequences of the ureB gene of 12 Helicobacter sp. flexispira strains (Table 1) were produced by automated cycle sequencing with Big Dye terminators (ABI 377XL; PE Applied Biosystems). The sequenced area (519 bp) corresponded to positions 139658 of the partial 732 bp ureB gene sequence of Helicobacter sp. flexispira ATCC 49315. The primers (ureBFF, 5'-GAGATGGTATGGCGCAATCT-3'; ureBFR1, 5'-cccaatcttcatgcaccttt-3') amplified a fragment of the expected size from all Helicobacter sp. flexispira strains, except H. trogontum. Using another reverse primer (5'-tcaatgccaccagcagtaac-3'), a 247 bp fragment was also amplified from H. trogontum. Additional sequences were available from public databases (Fig. 2). After pairwise and multiple alignments, a phylogenetic tree was constructed by the neighbour-joining method using the JukesCantor coefficient (Fig. 2). The topology of the tree was evaluated by 1000 trials of bootstrap analysis. The ureB sequences were translated to the corresponding protein sequences with the TRANSEQ program (EMBOSS). The resulting protein sequences were aligned and the neighbour-joining tree was calculated with CLUSTAL W. The phylogenetic tree based on the protein sequences was drawn by the TREEVIEW program 1.6.6 ().

Table 1. Helicobacter strains used in the study

(36K): Fig. 2. Phylogenetic tree of representative enterohepatic and gastric Helicobacter species based on partial ureB gene sequences (518 nucleotides). The tree was constructed using a neighbour-joining method rooted with Klebsiella aerogenes. GenBank accession numbers are given in parentheses. Numbers at each node represent percentages of bootstrap support calculated from 1000 probable trees. Bar, 5 % sequence divergence.

DNADNA hybridization.
Dot-blot hybridization of whole genomic DNA was performed for the reference strains of taxa 2, 3 and 8, H. bilis ATCC 50160^T, H. trogontum ATCC 700114^T, H. canis NCTC 12739^T and H. cinaedi CCUG 18818^T as well as for 14 canine and feline flexispira strains (Table 1) by the technique employed in our previous studies (Hänninen et al., 1996; Jalava et al., 1997, 2001). The concentrations of DNA used were 50, 5 and 0·5 ng. The strains used as probes were reference strains of taxa 2 (ATCC 49314), 3 (ATCC 49320) and 8 (CCUG 23435) (Table 1), H. bilis ATCC 50160^T, H. trogontum ATCC 700114^T and strains FL56 and FL57. The probes were labelled with digoxigenin with a DNA labelling and detection kit (Roche). Hybridization and all washings were performed at 58 °C. The intensity of hybridization was assessed by comparing the intensities of the colour that developed as described previously (Hänninen et al., 1996; Jalava et al., 1997, 2001). Morphology
Light and electron microscopy characterization of the strains showed that they all had cigar-like morphology with external fibrils outside the cell and tufts of sheathed flagella at both ends of the cell, findings typical of organisms (Weir et al., 1999; Sorlin et al., 1999; Jalava et al., 1998; Hänninen et al., 2003) given the provisional name flexispira by Bryner et al. (1987).

Phenotypic characteristics
The strains were all urease-positive, negative in nitrate reduction, had esterase and γ-glutamyl transferase activity and grew at 42 °C but not at 25 °C and did not grow in 1 % glycine (Table 2). In catalase tests, repeated several times, 10/14 isolates were catalase-negative. In the taxonomic description of the species H. bilis, the type strain ATCC 51630^T and all other isolates were described as reducing nitrate to nitrite (Fox et al., 1995). In our studies, the type strain ATCC 51630^T did not reduce nitrate, even after several separate experiments. Whether the type strain had lost its nitrate reduction capacity during subculture is unknown. Earlier studies have shown that canine flexispiras can be either catalase-negative or catalase-positive (Dewhirst et al., 2000a). These results were confirmed here.

Table 2. Characteristics of selected non-gastric Helicobacter species Taxa: 1, H. bilis, 14 Finnish strains; 2, H. bilis ATCC 53630T; 3, Helicobacter sp. flexispira, taxon 8; 4, Helicobacter sp. flexispira, taxon 2; 5, Helicobacter sp. flexispira, taxon 3; 6, Helicobacter sp. flexispira, taxon 7; 7, H. aurati; 8, H. trogontum; 9, H. pullorum; 10, H. muridarum; 11, H. hepaticus; 12, H. cinaedi; 13, H. canis; 14, Helicobacter sp. flexispira, taxon 10; 15, H. rodentium. Data were obtained from Dewhirst et al. (2000a), Fox et al. (1995), Hänninen et al. (2003), Mendes et al. (1996), Patterson et al. (2000), Saunders et al. (1999), Shen et al. (1997) and Robertson et al. (2001). ND, Not determined.

Phylogeny based on the 16S rRNA gene
Sequences of H. bilis ATCC 51630^T, Helicobacter sp. flexispira taxa 2, 3 and 8 reference strains and Finnish canine and feline flexispira strains and sequences of flexispira strains deposited in GenBank under the names H. bilis (HR17 and MIT 93-4659) and F. rappini (AF034135, AF118017, AF225550, AF336948) were widely dispersed into a cluster, which also included some other enterohepatic Helicobacter species (H. canis NCTC 12739^T, H. cinaedi CCUG 18818^T and ATCC 30365 and H. hepaticus ATCC 51448^T). H. trogontum was in the same cluster as H. muridarum, H. aurati and H. fennelliae. The bootstrap values supporting the topology of the branching of enterohepatic species were usually low. These results show trends similar to those of several previous studies (Sorlin et al., 1999; Dewhirst et al., 2000a; Vandamme et al., 2000; Hänninen et al., 2003), indicating extensive diversity within enterohepatic Helicobacter species and revealing that the phylogeny of the 16S rRNA gene is inconsistent with the cellular phylogeny and taxonomy of these enterohepatic Helicobacter species. Finnish canine and feline flexispira strains had similarity of 94·896·8 % to H. trogontum, 98100 % to H. bilis ATCC 51630^T, 97 % to the reference strains of taxa 2 and 3 and 9799·5 % to taxon 8. Taxa 2 and 3 were in their own branch, although the bootstrap value supporting their position was low (Fig. 1). The sequence of strain FL56 had only 91 % similarity to taxon 8, H. bilis ATCC 51630^T and Finnish canine and feline strains. It had the highest similarity to the taxon 7 reference strain (98·5 %). H. aurati had 9596 % similarity to the other flexispira strains. The incoherence of the evolution of the 16S rRNA gene with that of enterohepatic Helicobacter species studied by a polyphasic taxonomic approach could be explained by frequent genetic exchanges between enterohepatic species in this genomic region. Several enterohepatic Helicobacter species may simultaneously colonize the intestines of animal hosts, e.g. cats, dogs, mice and rats (Solnick & Schauer, 2001), enabling genetic recombination between different Helicobacter species. We also tested the specificity of the H. bilis 16S rRNA gene primers (Fox et al., 1995) and found that they produced fragments from H. trogontum ATCC 700114^T, H. bilis ATCC 51630^T, H. cinaedi CCUG 18818^T, the taxon 2 reference strain and five of the Finnish canine flexispira strains examined. These studies revealed that the primers were not specific to H. bilis but amplified fragments from other related enterohepatic Helicobacter species as well. These results are concordant with the observation of rather high interspecies 16S rRNA gene similarity. High intraspecies diversity of the 16S rRNA gene sequence (Vandamme et al., 2000; Hänninen et al., 2003) also suggests that design of a species-specific PCR test for identification of related enterohepatic Helicobacter species will be a difficult task.

Insertion sequences in the 16S rRNA gene
Insertion sequences have been demonstrated to be rather common in both the 16S and 23S rRNA genes in Helicobacter species (Hurtado et al., 1997). They seem to be identical or highly similar within a Helicobacter species and interspecies similarities also exist (Hurtado et al., 1997). The insert usually starts in the same region. In our studies, an identical or highly similar (98100 % similarity) 187 bp insertion sequence, starting at position 220, was detected in 11 of the flexispira sequences (Fig. 1). Insertion sequences were lacking in six sequences, including the reference strain of taxon 8, ATCC 43880 (Table 1). Insertion sequences of strain FL56 and the taxon 7 reference strain were highly similar and differed from other studied intervening sequences. Previous studies have shown that bacteria with flexispira morphology possess intervening sequences (Dewhirst et al., 2000a). The high similarity of the intervening sequences of H. bilis, taxa 2, 3 and 8 and some Finnish canine and feline flexispira strains suggests that these bacteria are also taxonomically highly related, as has been shown to be the case for other Helicobacter species (Linton et al., 1994; Fox et al., 1995). A PCR amplification test specific for intervening sequences can be used as a diagnostic tool to identify those H. bilis strains that have an intervening sequence in their 16S rRNA gene, as earlier proposed by Fox et al. (1995).

Phylogeny based on ureB sequences
ureB has been shown in some previous studies to be suitable for studying the phylogeny and evolution of gastric Helicobacter species (Gueneau & Loiseaux-De Goër, 2002). Evaluation of the suitability of the ureB gene for the taxonomy of enterohepatic Helicobacter species has been limited by the small number of sequences available. ureB sequence analysis could not, in any case, be used for the whole genus because several urease-negative enterohepatic species exist (Table 2; Solnick & Schauer, 2001). As expected, the fragment was amplified from the reference strains of taxa 2, 3 and 8 (CCUG 223435). In addition, the same product was amplified from H. bilis ATCC 53160^T and from all eight Finnish canine and feline strains studied.

The 479 bp ureB sequences were used for phylogenetic analysis and a tree was constructed (Fig. 2). The sequences of H. bilis, taxa 2, 3 and 8 reference strains, Finnish canine and feline flexispira strains and the four flexispira sequences from GenBank [F. rappini ATCC 43879 (AF116549), F. rappini ATCC 49315 (AF1165509), F. rappini NIH (AF116551) and Flexispira sp. (AJ130884)] clustered tightly together, having 97100 % similarity. The sequence of H. bilis ATCC 51630^T was 9798·7 % similar to the flexispira sequences listed above. The flexispira group mentioned above had similarity of 64·465·8 % to H. hepaticus ATCC 51448^T. These results indicate a close genetic relationship between the studied bacteria with flexispira morphology as well as their clear separation from the studied urease-positive enterohepatic Helicobacter species. The only exception was the sequence of strain FL56, which had a similarity of only 91·493·5 % to other flexispira strains. In the phylogenetic tree, strain FL56 was also distinguished from the other flexispira strains, having a high bootstrap value of 100 (Fig. 2). H. hepaticus was represented by six sequences (five from GenBank and a sequence from our study) and they formed their own cluster in the tree. The similarity of the six H. hepaticus sequences varied from 99·2 to 99·4 %, indicating a higher intraspecies similarity than that found for the flexispira group. Helicobacter mustelae clustered with the enterohepatic branch and not with the gastric Helicobacter species (Fig. 2), even though it is a gastric-area-colonizing species (Fox et al., 1990). The type strains of the gastric species H. pylori, Helicobacter felis, H. bizzozeronii and H. salomonis formed a gastric branch. Other studies concentrating more on the analysis of ureB gene phylogeny of gastric Helicobacter species have also shown that gastric and enterohepatic species are located in separate clusters (Gueneau & Loiseaux-De Goër, 2002). Comparison of the respective amino acid sequences indicated 98100 % similarity for H. bilis, Helicobacter sp. flexispira taxa 2, 3 and 8 strains, seven Finnish canine and feline flexispira strains and the four F. rappini sequences from GenBank. FL56 had a similarity of 9697 % to other strains. Our results suggest that, based on the high similarity of their ureB sequences, H. bilis, taxa 2, 3 and 8 strains and canine and feline flexispira strains are members of the same species. The lower similarity of FL56 to other flexispira strains suggests that this strain might belong to another species.

Comparison of the sequences of H. trogontum with sequences of other flexispira strains was not possible because the primers used failed to amplify its ureB sequence. We therefore performed a limited study on seven strains by developing primers that amplified both canine and feline flexispiras (KO220, FL57, KI311), H. bilis ATCC 51630^T, taxa 2 and 8 reference strains and H. trogontum. Sequencing and analysing this 247 bp fragment revealed that H. trogontum ATCC 700411^T had only 81·582·1 % similarity to H. bilis and other strains studied, suggesting that the phylogeny of this fragment of ureB was concordant with speciation. The sequence similarity within the other flexispira strains was high, from 96·5 to 99·5 %. In conclusion, these studies showed that ureB sequence analysis can differentiate H. trogontum from other flexispiras, including H. bilis. PCR with primers ureBFF and ureBR1 can be used for specific amplification of H. bilis DNA.

Whole genome DNADNA hybridization
Dot-blot hybridization using three concentrations of genomic DNA revealed a high level of hybridization between H. bilis ATCC 51630^T, taxa 2, 3 and 8 reference strains and 14 canine and feline flexispira strains. This indicates high whole genomic DNA similarity (results not shown) and suggests that all of these strains are members of a single species, i.e. H. bilis. Because 16S rRNA gene sequences of H. canis and H. cinaedi can show high similarity to H. bilis and other flexispiras (Fig. 1), DNAs of H. canis and H. cinaedi were hybridized with probe DNAs as well. The probes hybridized only weakly with the DNAs of H. canis and H. cinaedi, indicating only low relatedness and revealing that the flexispira group is not part of either H. canis or H. cinaedi. Whole genomic DNADNA hybridization has been one of the key taxonomic markers in the delineation of a bacterial species (Stackebrandt et al., 2002). DNADNA hybridization studies have not been performed extensively for Helicobacter species. In our previous studies, we used the present dot-blot technique, which requires less than 100 ng DNA, to describe two novel Helicobacter species, H. bizzozeronii (Hänninen et al., 1996; Jalava et al. 2001) and H. salomonis (Jalava et al., 1997), as well as to describe the extension of H. trogontum to include reference strains ATCC 43968, ATCC 43966 and ATCC 49310 (Hänninen et al., 2003).

Phylogeny of HSP60
Using approximately 600 bp sequences of the HSP60 gene as a phylogenetic marker, we have earlier shown that evolution of the HSP60 gene tends to correspond much better to the taxonomy of the genus Helicobacter than does the 16S rRNA gene (Mikkonen et al., 2004). The study also included H. bilis ATCC 51630^T, taxa 2, 3 and 8 reference strains, some of the canine and feline flexispira strains included in the present study (KO214, KO534B, KO220 and FL56), H. cinaedi and H. canis, as well as several H. trogontum strains. Analysis of HSP60 sequences showed that H. bilis, the taxon 8 reference strain and Finnish flexispira strains were in the same branch of the phylogenetic tree and had highly similar sequences, supporting the results of the present study. The sequences of taxa 2 and 3 strains had 88·9, 96 and 95·4 % similarity to H. trogontum, H. bilis and the taxon 8 reference strain, respectively. However, the respective amino acid sequences of taxa 2 and 3 differed from the other flexispiras by only one amino acid, indicating close interrelatedness. Strain FL56 had only 90·391·2 % sequence similarity to the H. bilis cluster (Mikkonen et al., 2004), supporting the results of 16S rRNA and ureB gene analysis. H. canis and H. cinaedi clustered with other enterohepatic species and were separated from H. bilis and H. trogontum.

Conclusions
Some confusion exists in the taxonomy and naming of organisms with flexispira morphology. H. trogontum is a named species associated with rat, sheep and pig, and Kirkbride et al. (1986) and Bryner et al. (1987) have shown that this species is strongly associated with ovine abortions. Our present and previous studies have revealed that a group of flexispira strains isolated from faecal samples of humans, dogs, cats and sheep as well as from the porcine stomach had highly similar phenotypic and genotypic characteristics based on their 16S rRNA, ureB, cdtB and HSP60 gene sequences, had high whole genomic relatedness in DNADNA hybridizations and differed from other named Helicobacter species. These facts lead us to conclude that reference strains of taxa 2, 3 and 8 and Finnish canine and feline flexispira strains constitute a single species, H. bilis. The methods in our studies fulfil the minimum requirements presented by Dewhirst et al. (2000b) for description of a novel Helicobacter species, as well as the later recommendations of the ad hoc committee for the revaluation of the definition of the bacterial species (Stackebrandt et al., 2002). Strain FL56 had different characteristics from other flexispiras and its position requires further studies. Bacteria resembling H. bilis have been isolated from blood cultures of human patients in association of bacteraemia (Tee et al., 1998; Sorlin et al., 1999; Weir et al., 1999), and similar DNA has been detected in bile and tissue samples of patients with various hepatobiliary diseases (Fox et al., 1998; Matsukura et al., 2002). These studies suggest that the clinical significance of H. bilis in gastrointestinal diseases should be further evaluated using adequate diagnostic methods. Based on its wide spectrum of hosts, H. bilis has a capacity for zoonotic transmission. Some case studies have shown potential transmission from dogs to humans (Romero et al., 1988). Our results provide diagnostic tools for further studies on the role of H. bilis in various diseases of animals and humans and confirm the power of the polyphasic approach in the taxonomy of species with a complex evolutionary background. More strains are required for studies on the taxonomy of strains FL56 and Helicobacter sp. taxon 7.

The financial support of the Research Funds of the University of Helsinki and the Academy of Finland are acknowledged.