Divergence between the Highly Virulent Zoonotic Pathogen Helicobacter heilmannii and Its Closest Relative, the Low-Virulence “Helicobacter ailurogastricus” sp. nov.

Abstract

Helicobacter heilmannii naturally colonizes the stomachs of dogs and cats and has been associated with gastric disorders in humans. Nine feline Helicobacter strains, classified as H. heilmannii based on ureAB and 16S rRNA gene sequences, were divided into a highly virulent and a low-virulence group. The genomes of these strains were sequenced to investigate their phylogenetic relationships, to define their gene content and diversity, and to determine if the differences in pathogenicity were associated with the presence or absence of potential virulence genes. The capacities of these helicobacters to bind to the gastric mucosa were investigated as well. Our analyses revealed that the low-virulence strains do not belong to the species H. heilmannii but to a novel, closely related species for which we propose the name Helicobacter ailurogastricus. Several homologs of H. pylori virulence factors, such as IceA1, HrgA, and jhp0562-like glycosyltransferase, are present in H. heilmannii but absent in H. ailurogastricus. Both species contain a VacA-like autotransporter, for which the passenger domain is remarkably larger in H. ailurogastricus than in H. heilmannii. In addition, H. ailurogastricus shows clear differences in binding to the gastric mucosa compared to H. heilmannii. These findings highlight the low-virulence character of this novel Helicobacter species.

INTRODUCTION

Helicobacter pylori is considered one of the most successful human pathogens. Infection with this agent has been associated with a wide range of gastric disorders. However, H. pylori is not the only Helicobacter species causing gastric disease in humans. Helicobacter heilmannii (sensu stricto), a zoonotic bacterium naturally colonizing the stomachs of cats and dogs, has been associated with gastritis, peptic and duodenal ulcers, and mucosa-associated lymphoid tissue (MALT) lymphoma in humans (1–6). This Helicobacter species is highly prevalent in the stomachs of clinically healthy cats and dogs as well as in those of animals showing chronic active gastritis (1, 4). Its pathogenic significance in these animals remains unclear and is probably strain dependent or related to host differences (1).

Little information is available regarding the pathogenesis of H. heilmannii infections in humans (1, 7). A recent experimental infection study, using Mongolian gerbils as an in vivo model to study Helicobacter-related gastric pathology in humans, investigated the colonization capacities and virulence of nine different Helicobacter strains (8). These helicobacters had been isolated from the gastric mucosae of stray cats and had been classified as H. heilmannii on the basis of the ureAB and 16S rRNA gene sequences (9). At 9 weeks postinfection, the induction of an antrum-dominant chronic active gastritis associated with the formation of lymphocytic aggregates and upregulation of the proinflammatory cytokine interleukin 1β (IL-1β) was shown for seven strains. However, differences in the expression of IL-1β were noted, together with differences in the intensity of the observed gastritis. High-level antral colonization was seen for four strains, while the colonization levels of the other strains were lower in the antrum and the fundus of the stomach. Based on the differences seen in colonization capacity and virulence, these Helicobacter strains were divided into a highly virulent group and a low-virulence group (8).

In the present study, we sequenced the genomes of the highly virulent and low-virulence H. heilmannii strains in order to investigate their phylogenetic relationships, to define this species' gene content and diversity, and to determine whether the presence or absence of specific virulence-associated genes might help to explain the differences in pathogenicity between these helicobacters. Because of the differences seen in gastric colonization between the highly virulent and low-virulence H. heilmannii strains, several in vitro binding assays were carried out to investigate possible differences in the capacity for adhesion to the gastric mucosa.

MATERIALS AND METHODS

Bacterial strains and whole-genome sequencing.

Five highly virulent (ASB1^T, ASB2, ASB3, ASB6, and ASB14) and four low-virulence (ASB7^T, ASB9, ASB11, and ASB13) feline Helicobacter strains (8) were cultivated in biphasic medium as described previously (9), and their genomic DNA was extracted by using the Qiagen, (Venlo, Netherlands) Blood & Tissue kit according to the manufacturer's guidelines. The genomes of ASB1^T and ASB7^T were obtained as described previously (10). An improved new assembly of the ASB1^T genome resulted in a genome size of 1,638,988 bp (Table 1), which is approximately 200 kb smaller than the previously published ASB1^T genome (10). An erratum will be submitted to adjust the previous publication. For whole-genome sequencing of the other seven strains, genomic DNA was normalized to 0.2 ng/μl, and a total of 1 ng was used for library generation. Sequencing libraries were prepared by using Nextera XT chemistry (Illumina Inc., San Diego, CA, USA) in accordance with the manufacturer's recommendations. Libraries were sequenced for a 250- or 300-bp paired-end sequencing run using the MiSeq personal sequencer (Illumina) (11). All genomes were assembled with the CLC Genomics Workbench, version 7. Gene finding and automatic annotation were performed using the RAST server (12, 13).

TABLE 1.

General features of the genomes of H. heilmannii and H. ailurogastricus

Strain	Genome size (bp)	No. of contigs	% GC	No. of coding sequences	No. of hypothetical proteins	No. of RNAs
H. heilmannii
ASB1	1,638,988	6	37.40	1,740	567	40
ASB2	1,570,832	93	37.46	1,783	627	38
ASB3	1,671,206	136	37.11	1,892	682	39
ASB6	1,606,820	84	37.50	1,781	642	40
ASB14	1,574,711	73	37.54	1,762	622	37
H. ailurogastricus
ASB7	1,675,643	9	37.34	1,706	564	42
ASB9	1,584,457	71	37.35	1,688	531	38
ASB11	1,578,737	56	37.38	1,677	532	38
ASB13	1,580,529	68	37.33	1,697	539	38

Phylogenetic and evolutionary analyses.

The list of fully annotated complete or draft genomes of different gastric and enterohepatic Helicobacter species obtained from the NCBI ftp server is shown in Table S1 in the supplemental material. The GET_HOMOLOGUES software package (14) was used to cluster genes in groups of orthologs. The bidirectional best-hit algorithm was used to define the core genes of these genomes. The GET_HOMOLOGUES output was further filtered by removing genes containing ambiguous nucleotides and selecting unique nucleotide sequences present in all genomes. Subsequently, a phylogenetic tree was created based on 303 concatenated core genes, as described previously (15–17). The phylogenetic tree was built using PhyML (18) by applying the -b 2, -m GTR, -f e, -c 6, -a e, -s BEST, and -o tlr parameters and was visualized by MEGA6 software (19). A distance matrix of the concatenated aligned core genes was calculated with DISTMAT implemented in jEMBOSS using the Kimura 2-parameter model (20).

The ASB genomes were submitted to the Genome-to-Genome Distance Calculator (GGDC; http://ggdc.dsmz.de) in order to calculate whole-genome distances and define the degree of DNA-DNA hybridization (DDH) between them. Additionally, the average nucleotide identity (ANI) values among the ASB genomes were calculated using the online “average nucleotide identity calculator” tool (enve-omics.ce.gatech.edu/ani/index) (21).

TEM.

The morphologies of two low-virulence strains, ASB7^T and ASB11, were characterized by means of transmission electron microscopy (TEM) as described previously (22). Semithin sections (2 μm) were cut and stained with toluidine blue. Thereafter, selected regions were chosen for ultrathin sectioning (90 nm) with an ultramicrotome (Ultracut E; Reichert-Jung, Nussloch, Germany). The sections were stained with uranyl acetate and lead citrate solutions before examination under a JEOL EX II transmission electron microscope (JEOL, Tokyo, Japan) at 80 kV. The morphologies of these two low-virulence strains were also studied by negative staining of bacterial culture samples with 2% (wt/vol) uranyl acetate.

Biochemical and tolerance tests.

The isolates were examined for catalase activity by adding a 3% H₂O₂ solution and observing the reaction within 5 s. Oxidase activity was tested with Bactident Oxidase strips (Merck, Overijse, Belgium). The API Campy identification system (bioMérieux, Marcy L'Etoile, France) was used to study urease activity, nitrate reduction, esterase activity, hippurate hydrolysis, γ-glutamyltransferase activity, triphenyltetrazolium chloride (TTC) reduction, alkaline phosphatase activity, and pyrrolidonyl, l-arginine, and l-aspartate arylamidase activities. Tests were read after 24 h of incubation at 37°C under an aerobic atmosphere.

Comparative proteomic analyses.

The complete set of predicted proteins from the Helicobacter species used in this study (see Table S1 in the supplemental material) and those from the ASB strains were clustered in groups of orthologs by using the GET_HOMOLOGUES software and applying the OrthoMCL algorithm. The compare_cluster.pl and parse_pangenome_matrix.pl Perl scripts were then used to find proteins that are absent in other Helicobacter species and thus unique to H. heilmannii.

To identify proteins present in the highly virulent strains but absent in the low-virulence strains, and vice versa, all proteins of the highly virulent strain ASB1 were compared with those of the low-virulence strain ASB7 by reciprocal BLASTP. Orthologous proteins were identified using the BLAST score ratio, a powerful tool for determining the probability of sharing a recent common ancestor. The general “acceptable” BLAST score ratio cutoff of 0.4 (equivalent to 40%) was used to define two proteins as homologs. Subsequently, the presence or absence of these specific proteins was checked and confirmed in the other highly virulent (ASB2, ASB3, ASB6, and ASB14) and low-virulence (ASB9, ASB11, and ASB13) strains as well. The proteins obtained were then used as queries for BLASTP homology searches against the total NCBI database in order to find related sequences present in other gastric Helicobacter species.

The putative outer membrane protein (OMP) sequences were extracted from the ASB genomes by using the HHomp tool (23) and BLASTP against the OMP database (24) and a set of well-known H. pylori OMPs using a BLASTP score ratio of 0.45.

Evolutionary analysis of the putative vacA-like genes present in all gastric Helicobacter species was performed using MUSCLE software (25). Neighbor-joining tree data were calculated on the basis of the VacA-like amino acid sequence alignment.

In vitro binding assays. (i) Binding to human gastric mucins.

All ASB isolates, cultured for 24 h, were harvested, centrifuged at 2,500 × g for 4 min, and resuspended in 1% Blocking Reagent for ELISA (Roche, Stockholm, Sweden), containing 0.05% Tween 20 (blocking buffer). Two human gastric mucin samples were used, one derived from a healthy stomach and one from a patient with a gastric tumor. Mucin samples were diluted in 4 M guanidinium chloride to 4 mg/ml and were used to coat 96-well plates (PolySorp; Nunc A/S, Roskilde, Denmark) overnight at 4°C. The plates were washed three times with phosphate-buffered saline (PBS) containing 0.05% Tween 20, and the wells were blocked for 1 h with blocking buffer. After the blocking buffer was discarded, the bacteria with an optical density at 600 nm (OD₆₀₀) of 0.1 were diluted 1:10 in blocking buffer containing 10 mM citric acid (at pH 2 and pH 7), and the dilutions were added to the wells. The 96-well plates were incubated for 2 h at 37°C in a shaker at 120 rpm. The plates were washed three times with PBS (plus 0.05% Tween 20) and were incubated with rabbit anti-H. pylori serum (1:1,000 dilution in blocking buffer) for 1 h at room temperature. Subsequently, the plates were washed three times and were incubated with horseradish peroxidase-conjugated anti-rabbit IgG (1:10,000 dilution in blocking buffer) for 1 h at room temperature. After further washing steps, a 3,3′,5,5′-tetramethylbenzidine (TMB) liquid substrate (Sigma-Aldrich, Diegem, Belgium) was added to the wells, and the plates were incubated for 20 min. The reaction was stopped with an equivalent amount of 0.5 M H₂SO₄, and the absorbance was measured in a microplate reader at 450 nm.

(ii) Binding to gastric epithelial cells.

The human gastric epithelial cell line MKN7 (Riken Cell Bank, Japan) was cultured in RPMI 1640 medium with 10% (vol/vol) heat-inactivated fetal bovine serum (FBS; HyClone, Logan, UT, USA), 2 mM l-glutamine (Invitrogen, Carlsbad, CA, USA), penicillin (50 U/ml), and streptomycin (50 μg/ml) (Invitrogen) at 37°C under 5% CO₂.

GSM06 cells (a murine gastric surface mucous cell line) were cultured in Ham's F-12 medium (Invitrogen) and Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% (vol/vol) heat-inactivated FBS, 1% (vol/vol) insulin-transferrin-selenium-A supplement (ITS; Gibco, Life Technologies, Erembodegem-Aalst, Belgium), penicillin (50 U/ml), streptomycin (50 μg/ml), and 0.2% (vol/vol) epidermal growth factor (EGF; Sigma-Aldrich) at 37°C under 5% CO₂.

For cocultures with the highly virulent and low-virulence Helicobacter strains, the cell medium was changed to antibiotic-free medium.

Bacteria adhering to gastric epithelial cells were visualized by scanning electron microscopy (SEM). For this purpose, MKN7 cells and GSM06 cells were seeded at a concentration of 10⁴/ml on coverslips in 24-well plates and were incubated overnight at 37°C. After incubation, cells were washed twice with Hanks' balanced salt solution with Ca²⁺ and Mg²⁺ (HBSS+; Life Technologies). Five hundred microliters of a bacterial suspension (at a concentration of 10⁸ viable bacteria/ml of cell medium) at pH 2 or pH 7 was added to the cells, and they were further incubated for 1 h at 37°C under microaerobic conditions. Thereafter, coverslips were again washed twice with HBSS+. Finally, coverslips were fixed in 500 μl HEPES fixative (2% paraformaldehyde) and were prepared for SEM as described previously (26). Briefly, the coverslip samples were fixed overnight in a HEPES-buffered glutaraldehyde solution. Samples were postfixed in 1% buffered osmium tetroxide for 2 h and were dehydrated in an increasing alcohol series, followed by an increasing ethanol-acetone series up to 100% acetone. The samples were then dried to the critical point with a Balzers CPD 030 critical point dryer (Sercolab BVBA, Merksem, Belgium) and were further mounted on metal bases and sputtered with platinum using the JEOL JFC-1300 Auto Fine Coater (JEOL Ltd., Zaventem, Belgium). The samples were examined with a JEOL JSM 5600LV scanning electron microscope (JEOL Ltd.). The mean number of binding bacteria per cell was calculated by counting spiral and coccoid Helicobacter bacteria attached to 5 cells selected at random.

A quantitative fluorescence-based adherence assay was performed in order to confirm the SEM results. For this purpose, MKN7 and GSM06 cells were seeded at a concentration of 10⁴/ml in 200 μl antibiotic-free cell medium in 96-well plates (Greiner Bio-One, Vilvoorde, Belgium) and were incubated overnight at 37°C. Helicobacter ASB strains were fluorescently labeled with fluorescein isothiocyanate isomer I (FITC; excitation wavelength, 492 nm; emission wavelength, 518 nm; Sigma-Aldrich). Briefly, bacteria (concentration, 10⁸ viable bacteria/ml of Brucella broth) were harvested, washed three times by centrifugation at 2,000 × g for 5 min, and resuspended in PBS-0.05% Tween. Subsequently, pellets were resuspended in 0.1 M carbonate and 0.15 M NaCl buffer (pH 9.0). Ten microliters of FITC (10 mg/ml dimethyl sulfoxide [DMSO]) was added to 1 ml of the bacterial suspension, followed by incubation for 30 min in the dark. FITC-labeled bacteria were washed three times in blocking buffer (1% bovine serum albumin [BSA] in PBS–0.05% Tween). The viability of all FITC-labeled strains was examined by checking their motility using light microscopy. The antibiotic-free cell medium was removed from the 96-well plates, and 150 μl of the FITC-labeled bacterial suspension was added to the cells (5 replicates per strain), followed by incubation for 1 h at 37°C under microaerobic conditions (9). Thereafter, the cells were washed twice with HBSS+, and the emission of fluorescent light at a λ of 527 nm was measured with a fluorometer (Fluoroskan Ascent FL microplate fluorometer and luminometer; Thermo Scientific, Erembodegem-Aalst, Belgium). Wells without cells (bacterial suspension only) and wells without the bacterial suspension (cells only) were included as controls to correct for any possible background signal. The adherence assay was performed immediately after fluorescent labeling of the Helicobacter strains in order to minimize the possible loss in viability of the labeled helicobacters over time. Finally, the relative levels of FITC labeling of all ASB strains were analyzed by flow cytometry (FCM) on a BD FACSCanto II flow cytometer (Becton Dickinson, Erembodegem, Belgium). The mean fluorescence intensity of each labeled strain, measured by FCM, was used as a correction factor for differential FITC labeling of the nine different strains. For each strain, the ratio of the mean fluorescence intensity measured by the fluorometer (indicating bacterial adhesion) to the mean fluorescence intensity measured by FCM (indicating the relative FITC labeling per strain) was calculated.

(iii) Binding to the gastric mucosae of Mongolian gerbils and stray cats.

Paraffin-embedded stomach tissues of 20 euthanized Mongolian gerbils (8) and 5 euthanized stray cats (9) were used. The ASB1^T, ASB7^T, and ASB11 isolates were cultured for 24 h, harvested, washed twice by centrifugation at 2,500 × g for 4 min, and resuspended in PBS. The bacterial concentration was adjusted to an OD₆₀₀ of 0.1 in PBS, and bacteria were labeled by incubation with 100 μg/ml FITC for 5 min at room temperature. Labeled bacteria were recovered by centrifugation at 800 × g for 7 min, washed three times with PBS, and resuspended in blocking buffer.

Paraffin-embedded tissue sections were deparaffinized and were washed twice in water and once in PBS with 0.05% Tween 20. The slides were incubated with Blocking Reagent for ELISA (Roche) for 30 min at room temperature. The labeled bacteria were diluted 1:20 in blocking reagent containing 10 mM citric acid (pH 2 and pH 7), and 200 μl was added to each slide, followed by incubation in a humidified chamber for 1 h at room temperature. The slides were washed twice in PBS with 0.05% Tween 20 and once in water and were then mounted with ProLong Antifade Reagent containing 4′,6-diamidino-2-phenylindole (DAPI) (Life Technologies).

Statistical analysis.

The normality of data was assessed using the Shapiro-Wilk normality test, which is appropriate for smaller sample sizes up to 2,000. The variance homogeneity of data was analyzed by using Levene's test for homogeneity of variances (SPSS Statistics, version 22, Command Syntax Reference; IBM).

The quantities of bacteria binding MKN7 cells and GSM06 cells were compared between H. heilmannii- and H. ailurogastricus-infected cells by using Kruskal-Wallis analysis, followed by a Mann-Whitney U test. An unpaired t test was applied to compare the quantities of H. heilmannii and H. ailurogastricus bacteria bound to gastric mucins and DNA. The numbers of bacteria binding gastric glands and surface epithelium on paraffin-embedded tissue sections were analyzed by Kruskal-Wallis analysis, followed by a Mann-Whitney U test. Differences were considered statistically significant at a P value of <0.05. SPSS Statistics software, version 22 (IBM), and GraphPad Prism, version 6, were used for the analyses.

Ethics statement.

All experimental procedures were approved and carried out in accordance with the regulation and guidelines of the Ethical Committee of the Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium (approval number EC2011/090), the IMIM-Hospital del Mar, Barcelona, Spain, and the Lund University Hospital, Lund, Sweden.

Nucleotide sequence accession numbers.

The genome sequences determined in this study have been deposited in the EMBL database, and the accession numbers can be found via BioProject record numbers PRJEB7933 and PRJEB7975. (The accession numbers of the individual strains are shown in Table S1 in the supplemental material.)

RESULTS

Phylogenetic relationships and phenotypical characterization of the highly virulent and low-virulence strains.

The draft genomes of the five highly virulent strains (ASB1^T, ASB2, ASB3, ASB6, and ASB14) and the four low-virulence strains (ASB7, ASB9, ASB11, and ASB13) (8) were 1.57 to 1.67 Mb. General features of the genomes are listed in Table 1, and the phylogenetic positions of the ASB strains are shown in Fig. 1. The nodes in this phylogenetic tree were supported with Chi2-based parameter branch values of 99%. All ASB strains belonged to the clade of the gastric non-H. pylori Helicobacter species and clustered in the sister clade of Helicobacter suis, but the highly virulent group was separated from the low-virulence group by a long branch. The branch length was comparable to that separating H. pylori from Helicobacter cetorum.

FIG 1.

Phylogenetic tree. Shown is a phylogram representing a maximum-likelihood tree of gastric and enterohepatic Helicobacter spp. based on 303 aligned and concatenated core genes. All nodes are supported with approximate likelihood ratio test (aLRT) values of >99%, and the topology, branch length, and parameters of the starting tree were optimized. The enterohepatic Helicobacter spp. were used as an outgroup.

The Kimura-2 corrected distance value between the two ASB groups, calculated on the basis of the 303 core genes, was approximately 19 substitutions per 100 bp (19%). The average distances among the highly virulent group and the low-virulence group were approximately 3% and 1%, respectively. Thus, at 19% substitution, these two ASB groups differed far more from each other than was expected on the basis of the 16S rRNA and ureAB genes. Compared to the H. pylori clade (Fig. 1), the average distance values between both the highly virulent ASB group and the low-virulence ASB group and H. pylori, Helicobacter acinonychis, or H. cetorum were approximately 50%. The average distance between the ASB group and H. suis was approximately 35%. Subsequently, additional tests were used to further investigate the relationship of the ASB strains to each other and, in particular, to determine whether they belong to the same species.

By use of the GGDC tool, it was shown that all the isolates previously called “low-virulence strains” did not belong to the H. heilmannii species. DNA-DNA hybridization (DDH) parameters, estimated in silico by calculating whole-genome distances, yielded a probability via logistic regression of 96.30% that the five highly virulent isolates ASB1^T, ASB2, ASB3, ASB6, and ASB14 belong to the H. heilmannii species. The DDH estimates for the four low-virulence isolates ASB7, ASB9, ASB11, and ASB13 resulted in a probability of only 0.01% that they belong to the H. heilmannii species. DDH parameters among these low-virulence strains yielded a probability of ca. 98%, indicating that these strains belong to the same species.

The average nucleotide identity (ANI) value among ASB1^T, ASB2, ASB3, ASB6, and ASB14 was 97.7%, whereas ASB7, ASB9, ASB11, and ASB13 shared an even higher ANI value of 99%. The ANI value between the two groups was only 84%, which is lower than the generally accepted threshold of 95% for belonging to the same species (21).

The morphology of the low-virulence strains was characterized by transmission electron microscopy. In our previous experimental infection study with Mongolian gerbils (8), strains ASB7 and ASB9, with low colonization capacity, did not cause antral inflammation, whereas chronic active gastritis was seen in the antra of the stomachs of gerbils infected with the low-colonization strains ASB11 and ASB13. Based on these differences among the four low-virulence strains, we selected ASB7 and ASB11 for the TEM study. As shown in Fig. 2, ASB7 and ASB11 presented as spiral bacteria with 4 to 5 turns that are 3.0 to 5.5 μm long and 0.5 to 0.7 μm wide and have 6 to 8 sheathed blunt-end flagella at both ends. No periplasmic fibrils were observed. These morphological characteristics are similar to those of H. heilmannii as described by Smet et al. in 2012 (9), with the exception that H. heilmannii has more spiral turns (as many as 9) and presents more bipolar flagella (as many as 10). The biochemical characteristics of ASB7, ASB9, ASB11, and ASB13 were similar to those of H. heilmannii as well. Biochemical analysis revealed that these low-virulence strains were oxidase, catalase, and urease positive. They reduced nitrate and triphenyltetrazolium chloride (TTC) and tested positive for esterase, hippurate, and γ-glutamyltransferase. No pyrrolidonyl arylamidase activity, l-aspartate arylamidase activity, or indoxyl acetate hydrolysis was detected. In contrast to the result for H. heilmannii, alkaline phosphatase activity was present. This enzyme is produced by most gastric Helicobacter species, but not by H. heilmannii (9). An overview of the morphological and biochemical characteristics of the highly virulent and low-virulence ASB isolates is shown in Table 2.

FIG 2.

Transmission electron microscopic images of H. ailurogastricus strains ASB7^T (A and C) and ASB11 (B and D). (A and B) Negatively stained cells of H. ailurogastricus ASB7^T (A) and ASB11 (B) showing cells with as many as 5 turns (arrows) and bipolar blunt-end flagella (bp). (C and D) Uranyl acetate and lead citrate staining of H. ailurogastricus ASB7^T (C) and ASB11 (D).

TABLE 2.

Morphological and biochemical characteristics of H. heilmannii and H. ailurogastricus compared to those of other gastric Helicobacter species

Helicobacter speciesa	Cell size (μm)		Periplasmic fibril	Flagella		Urease activity	Nitrate production	Alkaline phosphatase activity	Hydrolysis of indoxyl acetate	Growth at 42°C
	Length	Width		No. per cell	Distributionb
H. heilmannii	3–6.5	0.6–0.7	−	4–10	BP	+	+	−	−	−
H. ailurogastricus	3–5.5	0.5–0.7	−	6–8	BP	+	+	+	−	−
H. felis	5–7.5	0.4	+	14–20	BP	+	+	+	−	−
H. bizzozeronii	5–10	0.3	−	10–20	BP	+	+	+	+	+
H. salomonis	5–7	0.8–1.2	−	10–23	BP	+	+	+	+	−
H. cynogastricus	10–18	0.8–1.0	+	6–12	BP	+	+	+	−	−
H. baculiformis	10	1	+	11	BP	+	+	+	−	−
H. suis	2.3–6.7	0.9–1.2	−	4–10	BP	+	−	+	−	−
H. pylori	2.5–5.0	0.5–1.0	−	4–8	MP	+	−	+	−	−

^aData for H. ailurogastricus are from this study. Data for the other species are from references 9 (H. heilmannii), 57 to 59 (H. felis), 57 (H. bizzozeronii), 59 (H. salomonis), 60 (H. cynogastricus), 61 (H. baculiformis), 62 (H. suis), and 57 and 59 (H. pylori).

^bBP, bipolar; MP, monopolar.

All these data support the reclassification of these low-virulence strains as a novel species, for which we propose the name Helicobacter ailurogastricus sp. nov., with strain ASB7 as the type strain.

In silico proteome analysis.

Examination of the annotated genomes of different gastric and enterohepatic Helicobacter species (see Table S1 in the supplemental material) yielded a total of 50,899 predicted protein sequences. Based on OrthoMCL clustering, these proteins were divided into 12,216 groups of orthologs. A total of 132 (7.25%) H. heilmannii proteins and 82 (4.91%) H. ailurogastricus proteins had no orthologs in the other available genome-sequenced Helicobacter species (see Table S1) and thus might be unique to H. heilmannii and H. ailurogastricus, respectively. Reciprocal BLASTP analysis between the ASB1^T and ASB7^T proteomes identified 394 ASB1^T proteins (59 with putative function and 335 hypothetical proteins) with no significant homology to any H. ailurogastricus ASB7^T protein. Conversely, 303 ASB7^T proteins (59 with putative function and 244 hypothetical proteins) with no significant homology to any H. heilmannii ASB1^T protein were identified. An overview of the proteins with predicted functions is shown in Table S2 in the supplemental material. The majority of genes present in H. heilmannii ASB1^T but absent in H. ailurogastricus ASB7^T and vice versa play roles in DNA replication, recombination, and repair, protection against the uptake of foreign DNA, chemotaxis (bacterial signaling), outer membrane and lipopolysaccharide (LPS) synthesis, lipid metabolism, and transport and metabolism of nucleotides, amino acids, and carbohydrates (see Table S2). Additionally, genes encoding proteins involved in DNA binding and transfer and associated with ulcer development were identified in ASB1^T but not in ASB7^T, whereas proteins that play roles in fermentation, iron uptake, cell division, and various metabolic processes were found in ASB7^T but were absent in ASB1^T. Similar results were obtained when other H. heilmannii strains were compared with the H. ailurogastricus strains (data not shown).

Genes possibly associated with differences in virulence and colonization capacity between H. heilmannii and H. ailurogastricus.

Genes implicated in bacterium-host interactions that differ between H. heilmannii and H. ailurogastricus or that are present in H. heilmannii but absent in H. ailurogastricus merit special attention. The absence or presence of these genes in other gastric Helicobacter species was verified as well.

(i) iceA1.

In H. pylori, the ulcer-associated protein restriction endonuclease (IceA) has been identified as a virulence factor associated with peptic ulcer disease and is induced by contact of the bacterium with epithelial cells. The iceA gene exists as two distinct genotypes, iceA1 and iceA2, and only iceA1 RNA is induced following adherence (27, 28). The iceA allele is part of a restriction-modification (R-M) system and is located upstream of the ulcer-associated adenine-specific DNA methyltransferase (hpyIM) (28). R-M systems function in self/nonself recognition and protect against genomic adulteration by foreign DNA. These systems also promote homologous recombination of species-specific or closely related DNA and thereby provide a rapid mechanism of genetic adaptation (29, 30). The H. heilmannii strains each contain an intact iceA1 homolog (HHE01_10510 [see Table S2 in the supplemental material], HHE02_06610, HHE03_16620, HHE06_03450, HHE014_17360), showing approximately 90% amino acid identity to one another and only 54% identity to H. pylori IceA1 homologs. As in H. pylori, this gene is located next to a homolog encoding an ulcer-associated adenine-specific DNA methyltransferase. Interestingly, an iceA homolog is absent in H. ailurogastricus, and only a gene encoding the DNA methyltransferase is present. The iceA1-DNA methyltransferase locus has also been found in the available genomes of H. acinonychis and Helicobacter bizzozeronii (see Table S1 in the supplemental material) but is absent in H. cetorum, H. suis, and Helicobacter felis.

(ii) Putative hrgA.

Another DNA R-M system described for H. pylori is the hrgA/hpyIIIR system, of which the endonuclease-replacing gene (hrgA) has been described as a clinical marker for virulence (31). A hrgA-like gene, though in the absence of its methyltransferase enzyme, is also present in the genomes of H. heilmannii and H. bizzozeronii but not in H. ailurogastricus and other non-Helicobacter pylori helicobacters. H. heilmannii HrgA shows 50% protein-level identity with H. pylori HrgA. Interestingly, this protein exhibited limited sequence identity among the HrgA proteins of the different H. heilmannii strains (HHE01_08490 showed sequence identities of 98.26% with HHE02_00920, 42.32% with HHE03_02210, 40.44% with HHE06_03280, and 42.01% with HHE014_12870; HHE02_00920 showed sequence identities of 44.79% with HHE03_02210, 42.71% with HHE06_03280, and 44.44% with HHE014_12870; HHE03_02210 showed sequence identities of 64.06% with HHE06_03280 and 99.38% with HHE014_12870; HHE06_03280 showed a sequence identity of 63.75% with HHE014_12870). Also for H. pylori, virulence factors are highly diverse between strains, and this diversity has been associated with different disease outcomes (32).

(iii) jhp0562-like glycosyltransferase.

Besides proteins with functions in a R-M system, we identified a putative homolog of the H. pylori jhp0562 glycosyltransferase in the H. heilmannii strains (HHE01_14290, HHE02_07300, HHE03_07750, HHE06_15850, HHE014_01210) that is absent in H. ailurogastricus. The H. pylori LPS biosynthesis enzyme jhp0562 glycosyltransferase functions in the synthesis of both type I and type II Lewis (Le) antigens, which are present on the LPS of the bacterial outer membrane (33). Via intragenomic recombination of jhp0562, diverse Le antigens are generated, and this glycosyltransferase contributes to the process of phase variation in H. pylori (33–35). Phase variation is one of the mechanisms used by H. pylori to escape the host immune response and to persist in the stomach. It creates phenotypical variation in a bacterial population by the reversible process of switching a gene on and off. Phase-variable bacterial genes, such as LPS biosynthesis genes, play roles in bacterial pathogenesis and virulence (34, 35). Moreover, the presence of jhp0562 has been associated with peptic ulcer disease in children (33, 36). The jhp0562-like glycosyltransferases of the five H. heilmannii strains showed approximately 99% protein-level identity to one another but only 36% identity with H. pylori jhp0562 homologs. Homologs are also present in H. bizzozeronii, H. felis, H. suis, and H. cetorum, with protein-level identity between 35 and 40% with the jhp0562-like glycosyltransferase of H. heilmannii.

(iv) OMPs.

Several outer membrane proteins (OMPs) of H. pylori play important roles in adhesion to and colonization of the human stomach (37). H. heilmannii strains ASB1^T, ASB2, ASB3, ASB6, and ASB14 contain 53, 55, 56, 56, and 55 OMP-encoding genes, respectively, whereas the OMP repertoire of H. ailurogastricus consists of approximately 60 OMP-encoding genes. This gene number is in agreement with the ∼64 well-annotated OMP-encoding genes described for H. pylori (38). The H. pylori (J99), H. heilmannii (ASB1^T, ASB2, ASB3, ASB6, and ASB14), and H. ailurogastricus (ASB7^T, ASB9, ASB11, and ASB13) OMPs were clustered in groups of orthologs. The results are shown in Table S3 in the supplemental material. The analysis showed that H. heilmannii and H. ailurogastricus share only a few homologs of the H. pylori Hop, Hor, and Hom proteins. Remarkably, the well-studied H. pylori adhesins BabA and BabB (HopS and HopT), SabA (HopP), AlpA and AlpB (HopB and HopC), OipA (HopH), HopZ, HopQ, and HomB are absent in H. heilmannii and H. ailurogastricus. Only genes encoding homologs of the H. pylori Hof proteins, except for HofB, are present in H. heilmannii and H. ailurogastricus. The different Hof proteins each exhibit 98% identity among the different H. heilmannii strains. Similar findings were made for the H. ailurogastricus Hof proteins. The average levels of amino acid identity of the H. heilmannii Hof proteins to the H. ailurogastricus and H. pylori Hof proteins are about 88 to 90% and 55%, respectively. In contrast to those of H. pylori, the H. heilmannii and H. ailurogastricus hof genes are located in a ∼10-kb locus. This locus is also present in other canine, feline, and porcine gastric helicobacters. Additionally, H. heilmannii and H. ailurogastricus harbor several unique putative OMPs that are absent in H. pylori and whose biological function (e.g., interaction with the gastric mucosa) is unknown. Moreover, H. heilmannii harbors 6 putative OMPs that are absent in H. ailurogastricus. Two of these hypothetical OMPs are located in close proximity to each other on the same locus of approximately 13 kb (HHE01_09750 and HHE01_09730, HHE02_11280 and HHE02_11300, HHE03_06010 and HHE03_06030, HHE06_13950 and HHE06_13930, and HHE014_02600 and HHE014_02620 for ASB1, ASB2, ASB3, and ASB14, respectively [see Table S3 in the supplemental material]).

(v) VacA-like autotransporter.

One of the major protein toxins secreted by H. pylori is the vacuolating cytotoxin A (VacA), which belongs to an additional family of OMPs called autotransporters (39). The VacA toxin binds to host cells and is internalized, causing severe “vacuolation” characterized by the accumulation of large vesicles that possess hallmarks of both late endosomes and early lysosomes (40). No homologs of the H. pylori vacA gene are present among the H. heilmannii and H. ailurogastricus genomes. This vacA gene is also absent in the other canine, feline, and porcine non-H. pylori Helicobacter species. Intact homologs of this gene have been reported only for H. cetorum (15). Additionally, H. pylori contains three genes annotated as putative vacA paralogs, because the C-terminal autotransporter domains of the proteins they encode show approximately 30% identity to that of VacA. These three VacA-like autotransporters each enhance the capacity of H. pylori to colonize the stomach (41). The H. heilmannii and H. ailurogastricus strains also contain a gene encoding a VacA-like autotransporter (HHE01_12480, HHE02_13180, HHE03_15470, HHE06_06350, HHE014_11640, HAL07_13640, HAL09_00010, HAL011_16100, HAL013_02860). The VacA-like autotransporters of the different H. heilmannii strains exhibit approximately 95% identity to one another and 82% identity to those of H. ailurogastricus. An average identity of 98% was seen among the VacA-like autotransporters of the H. ailurogastricus strains. The H. pylori VacA-like autotransporters possess a conserved domain structure consisting of an N-terminal signal peptide, a nonconserved central passenger domain, and a C-terminal β-barrel domain. The presence of similar conserved domains in the VacA-like autotransporter proteins of H. heilmannii and H. ailurogastricus was predicted in silico and is shown in Fig. 3. These proteins show typical hallmarks of an autotransporter. The H. heilmannii autotransporter protein contains a short N-terminal cytoplasmic tail (ca. 1 to 53 amino acids [aa]) but without a predicted signal sequence, a transmembrane helix (ca. 23 aa), and a large noncytoplasmic part. The latter part contains the passenger domain with three VacA2 regions and a well-conserved C-terminal autotransporter (β-barrel) domain (Fig. 3). A similar structure was predicted for the H. ailurogastricus VacA-like autotransporter protein but with a larger passenger domain. Interestingly, all H. ailurogastricus strains showed four VacA2 regions in their passenger domain (Fig. 3). The passenger domain represents the surface-exposed component of the protein and adopts an extended right-handed β-helix structure (39). Also other canine, feline, and porcine gastric Helicobacter species harbor a copy of a vacA-like autotransporter gene. Phylogenetic analysis of the VacA-like autotransporter proteins present among the different gastric Helicobacter species highlighted high divergence among these autotransporters between species; only their C-terminal parts were well conserved (see Fig. S1 in the supplemental material). The H. heilmannii VacA-like protein exhibits approximately 40 to 50% protein-level identity to the H. felis, H. bizzozeronii, and H. suis homologs, while only 30% identity to the H. pylori, H. acinonychis, and H. cetorum homologs was noted.

FIG 3.

Schematic representation of the conserved domains present in the VacA-like autotransporter of H. heilmannii (ASB1^T) and H. ailurogastricus (ASB7^T and ASB11). No N-terminal signal sequence could be predicted by SignalP, version 3.0. The passenger domain of the ASB1^T VacA-like autotransporter contains three VacA2 regions, whereas those of ASB7^T and ASB11 harbor four VacA2 regions (small red rectangles). The main block of homology present in the autotransporter proteins of the two species is the C-terminal β-barrel domain (large red rectangle). (Top) The autotransporter of H. heilmannii ASB1^T is 8,583 bp long, with three VacA2 regions at bp 1494 to 1659, bp 2946 to 3120, and bp 5190 to 5358 and a β-barrel at bp 7809 to 8541. (Center) The autotransporter of H. ailurogastricus ASB7^T is 10,788 bp long, with four VacA2 regions at bp 1083 to 1248, bp 3897 to 4056, bp 5166 to 5340, and bp 7374 to 7542 and a β-barrel at bp 9972 to 10746. (Bottom) The autotransporter of H. ailurogastricus ASB11 is 11,040 bp long, with four VacA2 regions at bp 1335 to 1500, bp 4149 to 4308, bp 5418 to 5592, and bp 7626 to 7794 and a β-barrel at bp 10224 to 10998. The white triangle represents the region in the passenger domain that is absent in H. heilmannii ASB1^T but present in H. ailurogastricus ASB7^T and ASB11.

In vitro binding to the gastric mucosa.

Because of the differences seen in gastric colonization between H. heilmannii and H. ailurogastricus (8), we also investigated if there were differences in their capacities for binding to gastric mucins and epithelial cells.

(i) In vitro binding to human gastric mucins.

The capacities of H. heilmannii and H. ailurogastricus for binding to human gastric mucin samples, derived from a healthy stomach and from a patient with a gastric tumor, were tested at pH 2 and pH 7. The binding of H. pylori to mucins at acidic pHs has been shown to be dependent on charge (42). Mucins carry on the order of 100 different carbohydrate structures, including negatively charged carbohydrates. To distinguish between the charge-dependent binding mechanism and binding to other structures present on mucins, binding to DNA (as a marker for a negative charge) was also investigated. The results for the five H. heilmannii and four H. ailurogastricus isolates are displayed in Fig. 4. In general, the binding of H. heilmannii and H. ailurogastricus to the mucins was very weak at both pH values, whereas the level of binding to DNA at pH 2 was 10-fold higher (Fig. 4E). H. heilmannii had a higher capacity for binding to DNA than H. ailurogastricus (Fig. 4E and F) (P, <0.05 by an unpaired t test), and there also was a trend toward higher binding of H. heilmannii to the gastric mucins (Fig. 4B through D) (P, 0.07 to 0.11). There was no clear difference between binding to mucins at pH 2 and binding to mucins at pH 7.

FIG 4.

Capacity for in vitro binding to human gastric mucins and DNA. Shown is the in vitro binding of H. heilmannii (ASB1^T, ASB2, ASB3, ASB6, and ASB14) and H. ailurogastricus (ASB7^T, ASB9, ASB11, and ASB13) isolates to two human gastric mucins and to DNA at pH 2 (A, C, and E) and pH 7 (B, D, and F). Binding was quantified by measuring the OD at 450 nm. (A and B) Binding to a mucin sample derived from a gastric tumor; (C and D) binding to a mucin sample derived from a healthy stomach; (E and F) binding to DNA as a marker for a negative charge. Significant differences in DNA binding between the two species are indicated by asterisks (P, <0.05 by an unpaired t test). P values given above bars indicate non-statistically significant differences in mucin binding between the two species (B through D).

(ii) In vitro binding to gastric epithelial cells.

The in vitro capacities of H. heilmannii and H. ailurogastricus for binding to gastric epithelial cells were studied by scanning electron microscopy (SEM). An explicit difference between the capacities of H. heilmannii ASB1^T and H. ailurogastricus ASB7^T for binding to human-derived MKN7 cells and mouse-derived GSM06 cells was observed and is illustrated in Fig. 5A to D. Quantification of bacteria bound to cells showed higher numbers of H. heilmannii than of H. ailurogastricus bacteria binding to gastric epithelial MKN7 and GSM06 cells at both pH 2 and pH 7 (Fig. 5E to H) (P, <0.05 by the Mann-Whitney U test). At pH 7, more H. heilmannii bacteria were able to bind to MKN7 and GSM06 cells than at pH 2 (P, <0.05 by the Mann-Whitney U test). For each H. heilmannii and H. ailurogastricus isolate, the mean numbers of bacteria binding to MKN7 cells and GSM06 cells at pH 7 and pH 2 are shown in Table S4 in the supplemental material. The H. heilmannii type strain, ASB1, had the highest capacity for binding to both cell lines.

FIG 5.

In vitro binding of H. heilmannii and H. ailurogastricus isolates to gastric epithelial cells. (A to D) Scanning electron microscopy was used for visualization of the binding of H. heilmannii strain ASB1^T (A and C) and H. ailurogastricus strain ASB7^T (B and D) to MKN7 (A and B) and GSM06 (C and D) cells. Binding bacteria are indicated by red arrows for spiral Helicobacter cells and blue arrows for coccoid Helicobacter cells. Bars, 10 μm. (E to H) The mean numbers of H. heilmannii (ASB1^T, ASB2, ASB3, ASB6, and ASB14) and H. ailurogastricus (ASB7^T, ASB9, ASB11, and ASB13) bacteria binding MKN7 cells and GSM06 cells at pH 7 (E and F) and at pH 2 (G and H) were calculated (binding bacteria per cell). The results showed higher numbers of H. heilmannii than of H. ailurogastricus bacteria binding gastric epithelial cells. Significant differences in binding between the two species are indicated by asterisks (P, <0.05 by the Mann-Whitney U test).

A quantitative fluorescence-based adherence assay was performed to validate the SEM results. As shown in Fig. 6, a higher mean intensity of fluorescent light was emitted after the binding of FITC-labeled H. heilmannii bacteria to gastric epithelial MKN7 and GSM06 cells than after the binding of FITC-labeled H. ailurogastricus bacteria. Given that the intensity of emitted light is proportional to the quantity of cell-bound bacteria, this experiment confirms that H. heilmannii has a higher capacity for binding to gastric epithelial cells than H. ailurogastricus.

FIG 6.

Fluorescence-based assay of the adherence of H. heilmannii and H. ailurogastricus isolates to gastric epithelial cells. The binding of FITC-labeled H. heilmannii (ASB1^T, ASB2, ASB3, ASB6, and ASB14) and H. ailurogastricus (ASB7^T, ASB9, ASB11, and ASB13) isolates to gastric epithelial MKN7 (A) and GSM06 (B) cells was quantified by measuring the emission of fluorescent light at a λ of 527 nm. To correct for differential labeling of H. heilmannii and H. ailurogastricus with FITC, the relative levels of FITC labeling of all ASB strains were analyzed by FCM. Therefore, data are presented as the mean intensity of the emitted light normalized to the relative level of FITC labeling of H. heilmannii and H. ailurogastricus strains. Significant differences in binding between H. heilmannii and H. ailurogastricus are indicated by asterisks (P, <0.05 by the Mann-Whitney U test).

(iii) In vitro binding to the gastric mucosae of Mongolian gerbils and stray cats.

In vitro binding experiments (at pH 7 and pH 2) were also performed on paraffin-embedded gastric tissue samples from the antrum and the corpus of the stomach. Samples from Mongolian gerbils were used because this is a model used for studying Helicobacter-related gastric pathology in humans. Stomach samples from cats, the natural host of H. heilmannii and H. ailurogastricus, were included as well. Overall, similar binding patterns were obtained in the gastric mucosae of Mongolian gerbils and cats, and H. heilmannii and H. ailurogastricus showed equally strong capacities for binding to the samples. Results from Mongolian gerbils and cats were pooled in order to study the capacities for binding to the corpus versus the antrum of the stomach and to the surface epithelium versus the glands of the gastric mucosa. A clear difference in binding specificity between H. heilmannii and H. ailurogastricus (Fig. 7), which was more pronounced at pH 2, was observed. H. heilmannii ASB1^T bound mainly to the glandular cells of both the antrum and the corpus of the stomach (Fig. 7A and D), whereas H. ailurogastricus ASB7^T (Fig. 7B and E) and ASB11 (Fig. 7C and F) had a higher capacity for binding to the surface epithelium lining the gastric mucosa.

FIG 7.

In vitro binding to the gastric mucosae of Mongolian gerbils and stray cats. The number of binding bacteria was counted in 10 randomly chosen high-power fields at the level of the surface epithelium and the gastric glands (magnification, ×40) in the antra and corpora of gerbil and cat stomachs, both at pH 7 (A to C) and at pH 2 (D to F). Results from gerbils and cats were pooled for each stomach region, and data are shown as means + standard deviations. Significant differences in binding are indicated by asterisks (P, <0.05 by the Mann-Whitney U test). (A and D) Binding of H. heilmannii ASB1^T; (B and E) binding of H. ailurogastricus ASB7^T; (C and F) binding of H. ailurogastricus ASB11.

DISCUSSION

In the present study, comparative genomics and phylogenetic and phenotypical analyses of nine feline Helicobacter strains identified as H. heilmannii on the basis of their 16S rRNA and ureAB genes revealed that the previously defined low-virulence strains (8) belong to a novel species, closely related to H. heilmannii, that has not been described before and for which we propose the name Helicobacter ailurogastricus sp. nov. H. ailurogastricus cannot be distinguished from H. heilmannii by means of its 16S rRNA and ureAB gene sequences, which have frequently been used for differentiation between gastric Helicobacter species (2, 9). This implies that the discriminatory capacity of these gene sequences is not high enough for distinguishing between closely related gastric Helicobacter species. In this respect, our study underlines that genome-sequencing-based approaches are superior to traditional 16S rRNA sequence analysis for studying phylogeny, because they are based on the complete genome content and because they have better resolution for distinguishing between both distantly and closely related bacteria (43).

Phenotypically, H. ailurogastricus and H. heilmannii are also similar. Both species presented a spiral morphology with bipolar flagella but without periplasmic fibrils. The biochemical properties of H. ailurogastricus are very similar to those of H. heilmannii. Only the alkaline phosphatase activity differed between the two species: it was absent in H. heilmannii but present in H. ailurogastricus. The similar phenotypical characteristics contribute to the difficulty of distinguishing between these species. This reinforces the suggestion that a full genome sequence, combined with a minimal description of phenotypic characteristics, should become sufficient for the description of a novel species (43).

Our proteomic analyses revealed that H. ailurogastricus lacks homologs of the H. pylori IceA1, HrgA, and jhp0562 glycosyltransferase proteins, which have been reported to be involved in the disease outcome of H. pylori infection (31, 33, 44, 45). The absence of these virulence factors in H. ailurogastricus might thus contribute to the low-virulence character of this species. In contrast, genes encoding homologs of IceA1 and HrgA and a putative jhp0562-like glycosyltransferase are present in the H. heilmannii genomes. H. heilmannii has been associated with a number of different gastric disorders in humans, and the risk of developing mucosa-associated lymphoid tissue (MALT) lymphoma is higher after infection with H. heilmannii than after infection with H. pylori (1, 46, 47). The biological function of these putative virulence-associated proteins and their exact role in the disease outcome of H. heilmannii infection remain to be investigated.

Bacterial outer membrane proteins (OMPs) are directly involved in the interactions of pathogenic bacteria with their hosts. H. heilmannii and H. ailurogastricus harbor several OMPs, but only a few are members of the H. pylori Hop, Hor, or Hom family. This suggests the presence of other OMP classes in the genus Helicobacter. Interestingly, both species lack all H. pylori adhesins described so far. Only genes encoding homologs of the H. pylori Hof proteins were well conserved in both species. In contrast to those of H. pylori, their hof genes are located in a large locus. This locus seems to be unique for the canine, feline, and porcine gastric helicobacters (48). Preliminary results obtained with H. heilmannii deletion mutants demonstrated that the H. heilmannii Hof locus plays a role in gastric colonization (unpublished results). Moreover, six putative unique OMPs were predicted in the H. heilmannii genome, whereas they were absent in H. ailurogastricus. These OMPs might be involved in the difference between the colonization capacities of the two species. Further research is necessary to investigate the similarities and differences in these OMP genes and their expression, as well as the evolutionary events that are involved in their acquisition (e.g., gene conversion and phase variation).

Another virulence factor that merits particular attention is the VacA-like autotransporter. H. pylori harbors three such autotransporters, which belong to an additional OMP family and have been shown to enhance the ability of H. pylori to colonize the stomach (39, 41). Only one copy is present among the canine, feline, and porcine gastric helicobacters. Although no signal sequence was predicted using available tools, whether these autotransporters really lack a signal sequence remains to be elucidated. Interestingly, the VacA-like autotransporters were highly divergent between the different gastric Helicobacter species at the amino acid level. Possibly, horizontal transfer events are involved in their acquisition. Further studies are necessary to reveal (i) how their divergent sequences affect the transport, actions, and interactions of the proteins they encode and (ii) the selective forces that drive their evolution. In H. pylori, the passenger domain of the VacA-like autotransporter, which likely confers the effector function of this protein, contains three VacA2 regions. Although these VacA2 regions show low similarity to the VacA toxin, they do not correspond to a functional portion of VacA (39). In our study, the passenger domains of the VacA-like autotransporters of all H. heilmannii strains also contained three VacA2 regions, whereas in H. ailurogastricus strains, which have low colonization capacity in Mongolian gerbils (8), this passenger domain contained four VacA2 regions. In general, the major difference between the H. heilmannii and H. ailurogastricus VacA-like autotransporters was the size of the complete passenger domain, which was remarkably larger for the H. ailurogastricus VacA-like autotransporter. This highlights the potential role of the size of the passenger domain in gastric colonization, which merits further investigation.

In vitro binding assays revealed that H. ailurogastricus had a lower ability to bind to human- and mouse-derived gastric epithelial cells than H. heilmannii. The level of binding of H. heilmannii bacteria to gastric epithelial cells was higher at pH 7 than at pH 2. This is consistent with the physiological pH gradient in the stomach, ranging from pH 1 to 2 in the gastric lumen to pH 6 to 7 at the epithelial cell surface (49–51). Following this theory, we would expect better binding to human gastric mucins at pH 2, since they are located in the acidic lumen of the stomach. BabA- and SabA-independent adhesion of H. pylori to gastric mucins has indeed been shown to be more pronounced at low pHs, a phenomenon that is dependent on a charge/low-pH-dependent mechanism (42, 52, 53). In the present study, however, the in vitro capacities of H. heilmannii and H. ailurogastricus for binding to human gastric mucins were very low, and there was no difference in binding at low or neutral pHs. The binding capacities of H. heilmannii and H. ailurogastricus were 20- to 100-fold lower than the capacity of H. pylori for binding to the same human gastric mucins (42, 52, 54). Mucins can carry on the order of 100 different carbohydrate structures, which provide the mucins with a bottle brush appearance and make them act as receptors for microorganisms (55). The absence of carbohydrates acting as receptors for H. heilmannii and H. ailurogastricus in the mucin samples used might explain the low binding capacities of both species. On the other hand, binding to DNA as a marker for a negative charge was markedly higher at pH 2 than at pH 7. Thus, the role of a charge/low-pH-dependent mechanism in the binding of H. heilmannii and H. ailurogastricus to gastric mucins needs to be further investigated.

Both in vivo (1, 8) and in vitro in our study, H. heilmannii is found mostly in the gastric glands. A recent study also highlighted the potential role of the gland mucin MUC6 in the early colonization process of H. heilmannii (46). In contrast, the binding of H. ailurogastricus to the surface epithelium is more pronounced, like that of H. pylori (56).

In conclusion, we have described a new feline gastric H. ailurogastricus species, which is closely related to H. heilmannii. H. ailurogastricus lacks several homologs encoding H. pylori virulence and colonization factors and has a lower capacity for binding to gastric epithelial cells in vitro. This may explain why its virulence is lower than that of H. heilmannii.

Description of Helicobacter ailurogastricus sp. nov.

Helicobacter ailurogastricus (ai.lu.ro.gas′tri.cus. N.L. n. ailurus, cat, from Gr. n. ailouros, cat; N.L. masc. adj. gastricus, of the stomach, from Gr. n. gaster, stomach; ailurogastricus, N.L. masc. adj., of a cat's stomach).

Cells are tightly coiled spirals with as many as 5 turns that are approximately 3.0 to 5.5 μm long and approximately 0.5 to 0.7 μm wide. They have no periplasmic fibrils, and coccoid cells predominate in older cultures. They are motile by means of tufts of as many as 8 sheathed blunt-end flagella at both ends of the cells. Cells are Gram negative and nonsporulating.

Growth is detected under microaerobic conditions at 37°C, but not at 25 or 42°C. The organism is oxidase, catalase, and urease positive, reduces TTC and nitrate, and tests positive for esterase, γ-glutamyltransferase, hippurate, l-arginine arylamidase, and alkaline phosphatase. Activity of pyrrolidonyl arylamidase, l-aspartate arylamidase, and indoxyl acetate hydrolysis is not detected. Its clinical significance in cats and humans is so far unknown. The type strain, ASB7^T (also referred to as DSM 100489^T or LMG 28648^T), was isolated from the gastric mucosa of a stray cat.

Funding Statement

Agentschap voor Innovatie door Wetenschap en Technologie (IWT) provided funding to Myrthe Joosten under grant number SB-121092. Fonds Wetenschappelijk Onderzoek (FWO) provided funding to Annemieke Smet. Research Fund Ghent University provided funding to Richard Ducatelle and Freddy Haesebrouck under grant number GOA 01G00408. Svenska Forskningsrådet Formas (Swedish Research Council Formas) (grant numbers 221-2011-1036 and 221-2013-590), Vetenskapsrådet (Swedish Research Council; grant number 521-2011-2370), the Swedish Cancer Foundation, and the Söderbergs Foundation provided funding to Sara K. Lindén, Emma Skoog, and Médea Padra. National Health and Medical Research Council (NHMRC) (grant number 572723) and the University of Western Australia (UWA) provided funding to Alfred Chin Yen Tay, Fanny Peters, Tim Perkins, and Barry J. Marshall.