Comparative whole genome sequence analysis of the carcinogenic bacterial model pathogen Helicobacter felis.

The gram-negative bacterium Helicobacter felis naturally colonizes the gastric mucosa of dogs and cats. Due to its ability to persistently infect laboratory mice, H. felis has been used extensively to experimentally model gastric disorders induced in humans by H. pylori. We determined the 1.67 Mb genome sequence of H. felis using combined Solexa and 454 pyrosequencing, annotated the genome, and compared it with multiple previously published Helicobacter genomes. About 1,063 (63.6%) of the 1,671 genes identified in the H. felis genome have orthologues in H. pylori, its closest relative among the fully sequenced Helicobacter species. Many H. pylori virulence factors are shared by H. felis: these include the gamma-glutamyl transpeptidase GGT, the immunomodulator NapA, and the secreted enzymes collagenase and HtrA. Helicobacter felis lacks a Cag pathogenicity island and the vacuolating cytotoxin VacA but possesses a complete comB system conferring natural competence. Remarkable features of the H. felis genome include its paucity of transcriptional regulators and an extraordinary abundance of chemotaxis sensors and restriction/modification systems. Helicobacter felis possesses an episomally replicating 6.7-kb plasmid and harbors three chromosomal regions with deviating GC content. These putative horizontally acquired regions show homology and synteny with the recently isolated H. pylori plasmid pHPPC4 and homology to Campylobacter bacteriophage genes (transposases, structural, and lytic genes), respectively. In summary, the H. felis genome harbors a variety of putative mobile elements that are unique among Helicobacter species and may contribute to this pathogen's carcinogenic properties.

Helicobacter felis is a close relative of the human gastric bacterial pathogen H. pylori, the causative agent of gastritis and gastric ulcers (Marshall and Warren 1984) and known risk factor for gastric adenocarcinoma and gastric lymphoma (Parsonnet et al. 1991, 1994). Helicobacter felis was originally isolated from cats (Lee et al. 1990). Its ability to persistently colonize laboratory mice was discovered in 1990 (Lee et al. 1990) and was subsequently exploited to generate convenient and highly reproducible mouse models of Helicobacter-induced chronic active gastritis (Lee et al. 1990, 1993), gastric atrophy (Fox et al. 2000), lymphoma (Enno et al. 1995, 1998), and adenocarcinoma (Fox et al. 2002). It remains the most broadly used strain for modeling gastric pathology associated with virulent Helicobacter infection to date (Houghton et al. 2004; Sayi et al. 2009; Craig et al. 2010a, 2010b; Toller, Altmeyer, et al. 2010; Toller, Hitzler, et al. 2010). Whereas fully annotated whole genome sequence information is now available for seven H. pylori isolates (Tomb et al. 1997; Alm et al. 1999; Oh et al. 2006; Giannakis et al. 2008; Baltrus et al. 2009; Farnbacher et al. 2010; Fischer et al. 2010), one strain each of H. hepaticus (Suerbaum et al. 2003), H. mustelae(O'Toole et al. 2010), and H. acinonychis (Eppinger et al. 2006) and for numerous other related species of the Campylobacterales family (Parkhill et al. 2000; Baar et al. 2003), the H. felis genome had not been sequenced to date. Due to its carcinogenic properties and complement of virulence factors that are suspected to differ from those of H. pylori, the genome sequence information of H. felis is expected to generate novel insights into the molecular pathogenesis of this model organism.

The genome sequence of H. felis CS1 (ATCC 49179) was determined at the Wellcome Trust Sanger Institute using a combination of 454 pyrosequencing (which generated a total number of 345,320 reads, 47 contigs, 54× genome coverage) and Solexa sequencing (28.4 million reads, 900× genome coverage). The average read lengths were 388 bp (454) and 54 bp (Solexa). The combined 454 and Solexa sequencing data were joined to generate a draft sequence based on the assembly scaffold information; gaps between contigs were closed by a single round of polymerase chain reaction-based finishing/gap closure. The resulting final H. felis genome assembly revealed a 1,672,681 bp genome that includes a 6,700 bp episomally replicating plasmid. The genome of H. felis is comparable in size to the sequenced H. pylori genomes (1.59–1.67 Mb) is slightly larger than the H. mustelae and H. acinonychis genomes (1.58 and 1.55 Mb) but smaller than the H. hepaticus genome (1.8 Mb). Its GC content of 44.51% is higher than that determined for H. pylori (39%) but well within the range of Campylobacterales generally (30.6–48.5%). Coding sequence predictions made by Orpheus, Glimmer2, and EasyGene software identified 1,671 genes and 1 pseudogene in a coding area of 92%, placing H. felis squarely in the range determined for other Helicobacter species (1,403–1,875 genes) and Campylobacterales (1,403–2,046 genes). The average length of an H. felis gene is 921 bp (0.998 genes per kb). We detected 35 tRNA genes using tRNAscan-SE (Lowe and Eddy 1997). The existence of a previously identified H. felis plasmid could be confirmed (see below) (De Ungria et al. 1998). The H. felis genome sequence was annotated using Artemis software (Rutherford et al. 2000) version 12 and visualized with DNA Plotter; protein domains and patterns were marked up using Pfam and Prosite (Bateman et al. 2004). Annotation was transferred from the previously annotated H. pylori P12 and H. mustelae genomes (Fischer et al. 2010; O'Toole et al. 2010) to orthologous genes and then manually curated using FASTA and Blast results. Orthologous proteins were identified as reciprocal best matches using FASTA; all-against-all FASTA searches were performed, and reciprocal best matches were defined if the top hit covered at least 80% of the length of both sequences with at least 60% identity for both proteins. Of the 1,671 genes, 83% could be annotated with high confidence using these criteria. The fully annotated genome is publicly available at GenBank (accession number FQ670179). All genes were assigned to COG categories using the integrated microbial (IMB) genomes tool (Markowitz et al. 2009); annotated genes are color coded according to COG category (fig. 1).

F

Circular genome atlas of Helicobacter felis CS1. Rings from outside to inside: 1, Selected H. felis orthologues of H. pylori factors associated with virulence, colonization, natural competence, and chemotaxis. 2, Nucleotide coordinates in bp. 3, open reading frame (ORF) distribution, plus strand. 4, ORF distribution, negative strand. ORFs are color coded based on COG classifications. Abbreviations: B2-10, comB2-10; coll., collagenase.

Helicobacter felis Factors Involved in Colonization, Motility, Chemotaxis, Virulence, and Natural Competence

Helicobacter felis shares numerous features with other Helicobacter species that facilitate colonization of the gastric acidic environment and are required for motility and chemotaxis (fig. 1 and table 1). The H. felis genome harbors a complete urease gene cluster (ureABIEFGH), which in H. pylori is essential for gastric colonization (Karita et al. 1995) (fig. 1). In line with the requirement for nickel as a cofactor of urease, the orthologue of the H. pylori nickel transporter NixA and another predicted high-affinity nickel transport protein are encoded directly downstream of the urease gene cluster. An additional ureA,B2 operon is present in H. felis, as was reported for H. mustelae (O'Toole et al. 2010), where UreA2 and UreB2 are known to be expressed under conditions of nickel limitation (Stoof et al. 2008). The H. felis genome harbors at least 40 motility/chemotaxis-related genes encoded by the fla, flg, flh, and fli gene families (table 1), which in H. pylori and other flagellated bacteria are involved in the regulation, secretion, and assembly of the flagellum(O'Toole et al. 2000). Helicobacter felis further possesses an extraordinary number of chemotaxis genes. At least 20 predicted methyl-accepting inner membrane chemotaxis proteins (MCPs) are present in the H. felis genome, many of which share homology with the Bacillus subtilis MCPs TlpA (present in 5 copies in the H. felis genome), TlpB (6 copies), and TlpC (3 copies) and tend to be clustered together on the chromosome (fig. 1). In other organisms, the binding of chemoattractants such as urea, bicarbonate, or amino acids to MCPs is transduced to the autophosphorylating kinase CheA via CheW. CheA donates a phosphoryl group from a histidine residue to an aspartate of CheY, which then interacts with switching proteins to change the direction of flagellar motor rotation. cheA, W, and Y orthologues are each present in one copy in the H. felis genome. Three additional chemotaxis genes share homology with the B. subtilis CheV protein (termed CheV, V1, and V2 in H. felis). Like H. pylori, H. felis lacks cheB and cheR, which in other bacteria are responsible for modulating the chemotactic response by addition and removal of methyl groups to/from MCPs. The abundance of predicted MCP-like chemotaxis sensors in H. felis, especially in comparison to other Campylobacterales (H. pylori: 4, H. hepaticus: 9, C. jejuni: 10), is striking and suggests an elaborate spatial orientation in a diverse habitat.

Table 1

Genes with Functions in Colonization, Motility, Chemotaxis, Natural Competence, and Virulence in the Helicobacter felis Genome, Compared with H. pylori

Trait	Helicobacter pylori Locus	Orthologous H. felis System	Role in Colonization/Virulence
Type IV secretion	Cag PAI	Absent	Severe inflammation and secretion elevated gastric cancer risk
	ComB	ComB2,3,4,6,8,9,10 present	Required for natural competence
Urease production	Urease gene cluster	UreABIEFGH present additional UreAB cluster present	Required for acid resistance and gastric colonization
Vacuolating cytotoxin	VacA	Absent	Vacuolization, cytotoxicity, T-cell inhibition
Neutrophil activation	NapA	NapA present	Neutrophil activation, Th1 polarization
Gamma-glutamy transpeptidase	GGT	GGT present	Oxidative DNA damage, colonization
Outer membrane proteins	Hop, Hor, Hof, Hom	Hop, Hor, Hof, Hom	Binding to various glycosylated host cell surface proteins
Motility	Fla, Flg, Flh, Fli	FlaABG; FlgBCEE2GG2HIKLM; FliADEFGHILMPQRSTWW2Y; FlhABF	Regulation, assembly, and function of flagella
Secreted serine protease	HtrA	Present	E-cadherin cleavage; access to intercellular space
Cytolethal distending toxin	Absent	Absent	csd1-3 in H. hepaticus; DNAse activity
Sigma factors	σ54, σ70,σσ28	σ54, σ70,σσ28/fliA	Very few σ factors present
DNA repair recombination	RecA, AddA,B	RecA,N present; AddB present RecB,C,D,G absent; mutS present	Required for gastric colonization
Collagenase	Present	Present	Required for gastric colonization
Iron uptake	Feo, Fec, Frn, Fur, TonB, ExbB, ExbD	FeoA,B, FecA, 2xFrpB, 2xTonB, ferritin, SodB, Fur, ExbB2,D,D2	Iron uptake likely critical for early colonization and persistence

Apart from a type III secretion system that exports the flagellar subunit components across both membranes (encoded in H. felis by predicted orthologues of the H. pylori genes flhA,B and fliH,I,P,Q,R), the H. felis genome harbors only one additional secretion system. The comB regulon encoding the type IV secretion components comB2,3,4,6,8,9,10 is required for natural competence of H. pylori (Hofreuter et al. 1998, 2003; Karnholz et al. 2006), and its orthologues are expected to encode the identical function in H. felis. Whereas the comB components of H. pylori are organized in only two operons encoding comB2-4 and comB6-10, respectively, the H. felis orthologues are dispersed across the genome in three operons (fig. 1). A second H. pylori-specific type IV secretion system, encoded by the Cag pathogenicity island, is clearly absent in the H. felis genome, as is the vacuolating cytotoxin VacA (Montecucco et al. 2001). Other virulence-associated genes of H. pylori, in contrast, are present in H. felis and typically share a high degree of similarity. An H. pylori virulence factor involved in immunomodulation, NapA (Satin et al. 2000), is present in H. felis, as are three enzymes recently implicated in H. pylori virulence—a collagenase, the secreted serine protease HtrA and the gamma-glutamyl transpeptidase GGT (Gong et al. 2010; Hoy et al. 2010). The cytolethal distending toxin shared by H. hepaticus and C. jejuni lacks an orthologue in H. felis. The H. felis genome encodes a total of 52 outer membrane proteins belonging to the Hor, Hop, Hof, and Hom gene families. Orthologues of the Sab and Bab adhesins could not be identified in the H. felis genome.

All Helicobacter genomes sequenced so far share several features that set the Helicobacter genus apart from other enteropathogenic bacteria; one striking characteristic is the scarcity of transcriptional regulators. We could identify only three sigma factors, σ54/rpoN, σ70/rpoD, and σ28/FliA. FliA, as well as the anti-sigma factor FlgM of H. felis contribute to the regulation of the fla/fli/flg/flh motility regulon. The CstA regulator of the “stringent response” to carbon starvation is present in the H. felis genome. Two-component systems consisting of a membrane histidine kinase sensor protein and a cytoplasmic DNA-binding response regulator are represented in the H. felis genome by only two sensors and two response regulators (in addition to the aforementioned CheA/Y system regulating chemotaxis), which all share high homology with their orthologues in other Helicobacter genomes. Additional transcriptional regulators identified in the H. felis genome include the ferric uptake regulator (Fur), a main regulator of iron acquisition, the nickel-responsive repressor NikR and the carbon storage regulator CsrA. Like other bacterial pathogens colonizing environments limited in ferric and ferrous iron, H. felis possesses a number of iron uptake and storage systems. In addition to Fur, the H. felis genome encodes orthologues of the Escherichia coli Fec and Feo siderophore-mediated iron uptake systems; multiple copies of frp genes encoding heme- or lactoferrin-binding proteins are present. The nonheme iron storage protein ferritin is also encoded in the H. felis genome. Overall, the iron uptake systems of H. felis are highly conserved within the Helicobacter genus (Tomb et al. 1997; Suerbaum et al. 2003), highlighting the evolutionary restrictions created by iron limitation in mammalian hosts.

Comparative Analysis of Sequenced and Annotated Helicobacter Genomes

Of the 1,671 genes identified in the H. felis genome, 1,033 are predicted to have orthologues in H. pylori P12 (61.8%), 1,048 have orthologues in H. pylori B8 (62.7%), 1,029 have orthologues in H. acinonychis (61.6%), 946 genes have orthologues in H. mustelae (56.6%), and 940 have orthologues in H. hepaticus (56.2%). A circular plot comparing the H. felis genome with these five Helicobacter genomes is shown in figure 2 along with GC content and GC skew maps. The genes that are H. felis-specific, that is, not shared with these representatives of the other sequenced Helicobacter species (purple ring, fig. 2), are enriched for very few gene classes. Genes encoding restriction/modification systems are particularly overrepresented among H. felis-specific genes. The H. felis genome encodes eight complete restriction/modification systems as well as methyltransferases for which no matching restriction endonuclease could be identified; several of the complete systems and most of the adenine methyltransferases lack orthologues in the other Helicobacter genomes (annotated in red, fig. 2). The number of restriction/modification systems is thus quite variable across the Helicobacter genus, with H. pylori encoding at least 11 and H. hepaticus encoding only 2 complete systems (Tomb et al. 1997; Suerbaum et al. 2003).

F

Helicobacter felis-specific genes, plasmid and genomic regions. (A) Circular plot showing the genomic differences and similarities of six Helicobacter species as indicated in the legend. Rings from outside to inside: 1, Positions of three putative horizontally acquired mobile elements (regions I–III) and the H. felis plasmid pHFS1 in the H. felis genome. 2, Positions of H. felis-specific genes; functional categories are color coded as indicated in the legend. 3, Nucleotide coordinates in bp. 4, H. felis-specific ORFs not present in any of the other five Helicobacter genomes (purple). 5, H. felis ORFs on the plus and minus strands (dark green). 6–10, ORFs of the indicated Helicobacter species with orthologues in H. felis. 11, GC content. 12, GC skew. The GC content and GC skew were calculated in Artemis with a window size of 2,000 and 5,000 bp, respectively, and an overlap of 200 bp between windows. (B) Schematic showing the positions of the five predicted ORFs of the H. felis plasmid pHFS1. (C) and (D) Schematics of the putative horizontally acquired regions I, II, and III with nucleotide coordinates; annotated genes are indicated using the following color code: dark green, phage structural, and morphogenesis genes; orange, phage lytic genes; red, (phage) transposases and resolvases; yellow, replication initiation proteins; blue, antibiotic resistance and modification genes; olive, mobilization proteins; gray, uncharacterized or putative genes. A schematic of the H. pylori plasmid pHPPC4 is shown in C for comparison.

Other classes of H. felis-specific genes include additional copies of the chemotaxis sensors tlpA, B, and C mentioned earlier and transposases encoded by putative insertion elements (annotated in fig. 2). As indicated above, H. felis possesses a plasmid of 6,712 bp that shares no homology with the plasmids characterized in H. pylori (fig. 2). The plasmid encodes five predicted proteins, two of which—a putative murein transglycosylase and an N-acetylmuramoyl L-alanine-amidase—are peptidoglycan-modifying enzymes with predicted autolytic activity. A replication initiation protein A (RepA) is not encoded by the plasmid; seven chromosome-encoded full-length copies of repA (Hfelis200, 16140, 16300, and 16400) may compensate for the plasmid’s repA deficiency and may regulate its replication in trans. Of the seven repA gene copies, two are encoded on two of three putative mobile elements (fig. 2), which differ in GC content from the rest of the genome (fig. 2). Region I (∼8 kb in length, positions 1596300–1604640) shows homology and synteny with an H. pylori plasmid isolated from a gastric cancer patient in Peru, pHPPC4. Two genes of region I encode an ABC-transporter-like multidrug resistance protein and a mersacidin-modifying enzyme (fig. 2), which introduces lanthionin rings into lantibiotics such as mersacidin. Both genes show >70% similarity with their H. pylori homologues encoded on pHPPC4 (accession number CP002075). pHPPC4 in turn shares sequence homology with many other H. pylori plasmids (pHel4, pHP69, pHPAG1, pHPG27, pHPP12), which in contrast to pHPPC4 lack both the multidrug resistance and mersacidin-modifying enzyme genes. Despite the fact that region I harbors a repA gene, we have no evidence that the region replicates episomally. A second region, region II (29.8 kb in length, positions 4000–33757, fig. 2), harbors elements with extensive homology to Campylobacter bacteriophage genes. At least eight region II-encoded phagic structural and morphogenesis proteins and resolvases share a high degree of homology (but no synteny) with Campylobacter jejuni and hominis phage sequences. Several of the bacteriophage genes (Hfelis0290, 0350, 0450, 0460, 0610) have homologues in either H. hepaticus and/or H. bilis, indicating that the sequenced strains of these two murine pathogens may also harbor prophages. Interestingly, region II is flanked by two IS605 transposases (Hfelis0370 and Hfelis0630), one of which is >60% identical to chromosomally encoded transposases of H. pylori. Region III (9.5 kb in length, positions 1624000–1633500, fig. 2), in contrast, neither bears resemblance to prophages of Campylobacter (despite sharing homology with nonphagic, chromosomally encoded Campylobacter genes) nor do the mobilization and replication initiation proteins encoded by region III resemble their functional counterparts on H. pylori plasmids. Again, we could not find evidence for episomal replication of region III, indicating that this region may represent an integrated, no longer autonomously replicating plasmid derived from Campylobacter.

In summary, the sequence of H. felis reaffirmed many known characteristics of the Helicobacter genus such as the general paucity of transcriptional regulators and the abundance of restriction/modification systems and chemotaxis sensors. Our evidence for the existence of at least one H. felis prophage makes it unique among Helicobacter species, which in contrast to Campylobacter and with the exception of H. acinonychis, are not known to harbor phages. Efforts are currently under way to establish the genetic tools to manipulate H. felis and will hopefully allow us to dissect experimentally which virulence factors are required for gastric carcinogenesis and lymphomagenesis induced in animal models by this interesting bacterial pathogen.