Helicobacter hepaticus cholesterol-α-glucosyltransferase is essential for establishing colonization in male A/JCr mice

Background: Helicobacter pylori cholesterol‐α‐glucosyltransferase (cgt) is essential for survival of H. pylori in mice. Enterohepatic H. hepaticus, the cause of colonic and hepatocellular carcinoma in susceptible mouse strains, contains an ortholog of the H. pylori cgt. However, the role of cgt in the pathogenesis of H. hepaticus has not been investigated. Materials and Methods: Two cgt‐deficient isogenic mutants of wild‐type H. hepaticus (WT) 3B1 were generated and used to inoculate male A/JCr mice. Cecal and hepatic colonization levels of the mutants and WT 3B1 as well as select inflammation‐associated cytokines were measured by qPCR at 4 months postinoculation. Results: Both mutants were undetectable in the cecum of any inoculated mice (10 per mutant) but were detected in two livers (one for each mutant); by contrast, 9 and 7 of 10 mice inoculated with WT 3B1 were qPCR positive in the ceca and livers, respectively. The mice inoculated with the mutants developed significantly less severe hepatic inflammation (p < .05) and also produced significantly lower hepatic mRNA levels of proinflammatory cytokines Ifn‐γ (p < .01) and Tnf‐α (p ≤ .02) as well as anti‐inflammatory factors Il10 and Foxp3 compared with the WT 3B1‐inoculated mice. Additionally, the WT 3B1‐inoculated mice developed significantly higher Th1‐associated IgG2a (p < .0001) and Th2‐associated IgG1 responses (p < .0001) to H. hepaticus infection than mice dosed with isogenic cgt mutants. Conclusion: Our data indicate that the cholesterol‐α‐glucosyltransferase is required for establishing colonization of the intestine and liver and therefore plays a critical role in the pathogenesis of H. hepaticus.

Helicobacter hepaticus; virulence genes; bacterial load; cholesterol; mouse; model

Cholesterol is an essential component of mammalian cell membranes where it is required to establish proper membrane permeability and fluidity, but also plays a pivotal role in obesity and cardiovascular diseases. This molecule can be glycosylated into cholesteryl glucosides by glucosyltransferases in plants, most fungi, and some animals. Cholesteryl glucosides are also found in bacteria such as Helicobacter pylori and Acholeplasma axanthum, despite the fact that prokaryotes do not produce cholesterol. Three major cholesteryl glucosides, including cholesteryl‐α‐D‐glucopyranoside (αCG) and its two derivatives αCAG (cholesteryl—6′‐O‐tetradecanoyl‐α‐D‐glucopyranoside) and αCPG (cholesteryl—6′‐O‐phosphatidyl‐α‐D‐glucopyranoside), have been identified and characterized in H. pylori cells using two‐dimensional thin‐layer chromatography [1] . Subsequently, an H. pylori gene (HP0421) was identified to encode cholesterol‐α‐glucosyltransferase (Cgt) which catalyzes conversion of cholesterol and glucosides into cholesteryl glucosides such as αCG, αCAG, and αCPG [2] , [3] . In C57BL/6 mice and gerbils, inactivation of cgt in H. pylori increased host phagocytosis and T‐cell activation against infection, and the cgt mutants did not sustain colonization [3] , [4] . It was reported that gastric mucosa‐associated O‐glycans inhibited the growth and motility of H. pylori by disrupting its biosynthesis of αCG. In the presence of exogenous cholesterol, increased cholesterol glucosylation by H. pylori enhances its resistance to eight antibiotics and an antimicrobial peptide LL‐37 cathelicidin in vitro [4] . These lines of evidence indicate that cholesteryl glycosides protect H. pylori from the host immunity.

Helicobacter hepaticus, a prototype of enterohepatic Helicobacter species (EHS), induces chronic active hepatitis, hepatocellular carcinoma, colon cancer, and inflammatory bowel disease in susceptible mouse strains and has been widely used as infectious models for dissecting pathogenic mechanisms of similar human diseases [5] , [6] . H. hepaticus 16S rDNA are also detected in human subjects via PCR‐based assays [7] , [8] . Approximately 50% of the genes predicted from the genome of H. hepaticus are orthologs of H. pylori, including a cgt ortholog (Hh0676) [9] . In addition, the cgt gene is also present in several gastric helicobacters including H. felis, H. mustelae, and H. acinonychis. Among the recently sequenced genomes of five species of EHSs known to infect humans, this gene is identified only in H. bilis but is absent from the genomes of H. canadensis MIT 98‐5491, H. cinaedi CCUG 1881, H. pullorum MIT 98‐5489, H. winghamensis ATCC BAA‐430 (at www.broadinstitute.org/annotation/genome/Helicobacter%5fgroup). To explore the role of cgt in H. hepaticus‐induced liver disease in a mouse model of infectious hepatitis, we generated 2 isogenic H. hepaticus cgt mutants and characterized how cgt inactivation influences intestinal and hepatic colonization, proinflammatory responses, and induction of chronic hepatitis in male A/JCr mice.

Escherichia coli strain Top10 was used as a recipient for cloning, mutagenesis, and plasmid propagation and was cultured in LB broth or agar supplemented with antibiotics ampicillin (Amp, 50 μg/mL) and chloramphenicol (Cm, 25 μg/mL) when appropriate. Wild‐type H. hepaticus (WT) 3B1 (ATCC 51449) was cultured on blood agar (Remel, Lexignton, KY, USA) for 2–3 days under microaerobic conditions (10% H2, 10% CO2, 80% N2) [10] . Cm‐resistant H. hepaticus mutants were selected on tryptic soy agar supplemented with 5% sheep blood and 25 μg/mL of Cm (all from Sigma, St. Louis, MO, USA).

Chromosomal DNA from murine tissues and cultured bacteria was isolated using the High Pure PCR Template Preparation Kit (Roche Applied Science, Indianapolis, IN, USA). Plasmid DNA was prepared using the Aquick minipreparation kit (Qiagen Inc., Valencia, CA, USA). PCRs were conducted in a 50‐μL volume containing the following: 10–50 ng of DNA template, 1× commercial buffer, 100 μg/mL BSA, 500 nmol/L each of forward and reverse primers and 2.5 units of high‐fidelity DNA polymerase (Roche Applied Science). A thermocycling program of 35 cycles in a Thermocycler Genius (Technie Incorporated, Princeton, NJ, USA) included denaturation at 94 °C for 1 minute, followed by annealing at 50–60 °C (based on the respective primers) for 30 seconds and extension at 72 °C for 1 minute.

Approximately 600‐bp fragments upstream and downstream of Hh0676 (cgt) were amplified from WT 3B1 genomic DNA using primer pairs ZMG169/170 and 171/172, respectively (Table [NaN] ). PCR fragments were ligated with pBluescript II SK and subsequently inserted with a Cm acetyltransferase cassette lacking a transcriptional terminator (catNT) as illustrated in Fig. [NaN] . We previously demonstrated that the catNT has no polar effect on the downstream gene ureI[11] . Authenticity of the ligation and inserted DNA sequences was confirmed by sequencing (the ABI Genetic Analyzer 3500, Foster City, CA, USA). Isogenic H. hepaticus cgtΔcat mutants were generated as described previously [12] . CmR isogenic mutants of H. hepaticus were characterized by PCR and DNA sequencing. Two mutants, cgtΔcatNT1 (cat within the transcriptional orientation of cgt) and cgtΔcatNT2 (cat opposite to the transcriptional orientation of cgt), were selected for further characterization in vivo (Fig. [NaN] ).

Primer sequences used for generating cgt isogenic mutants of H. hepaticus

1 Numbers (+, downstream; −, upstream) denote starting nucleotide positions of primers in correspondence to the start codon of the H. hepaticus cgt (Hh0676).

Male A/JCr mice at 4–6 weeks of age free of known murine viruses, pathogenic bacteria including Helicobacter spp. and parasites, were obtained from NCI (Germantown, NY, USA). The mice were maintained in an Association for Accreditation and Assessment of Laboratory Animal Care, International‐accredited facility in static microisolater cages. Forty male A/JCr mice were divided into four groups of 10 mice dosed with either WT 3B1, cgtΔcatNT1, cgtΔcatNT2, or sham dosed with Brucella broth as a control. For oral gavage, bacteria were cultured on blood agar, suspended in Brucella broth, and adjusted to 109 organisms/mL as estimated by spectrophotometry at OD600 nm. Mice received 0.2 mL of fresh inoculum by gastric gavage every other day for three doses.

All mice were euthanized at 4 months postinoculation (MPI). Immediately after euthanasia, the ileocolic junction and two sagittal sections of each liver lobe were collected for routine histologic processing and sectioning. Hematoxylin and eosin of the liver and intestine were examined by a veterinary pathologist (SM) blinded to sample identity as previously described [13] . Hepatitis was graded on a 0–4 scale for lobular histologic activity (lobular hepatitis) and portal activity (portal and/or interface hepatitis).

Tissues for RNA/DNA isolation were frozen in liquid nitrogen immediately after sampling and stored at −70 °C prior to processing. For enumerating H. hepaticus organisms, chromosomal DNA from the ceca and livers was prepared using a High Pure PCR Template kit according to the manufacturer's protocol (Roche Applied Science). Levels of hepatic and cecal H. hepaticus were measured by qPCR in the 7500 Fast Sequence Detection System (Applied Biosystems) as described elsewhere [14] . The colonization level of H. hepaticus was estimated by normalizing to μg of mouse chromosomal DNA measured by qPCR using the 18S rRNA gene‐based primers and probe mixture (Life Technology, Grand Island, NY, USA).

For relative mRNA quantitation of selected genes, total RNA from livers of A/JCr mice was prepared using Trizol reagent according to manufacturer's recommendations (Invitrogen, Carlsbad, CA, USA). Five μg of total RNA from each sample was converted to cDNA using the High Capacity cDNA Archive kit (Applied Biosystems). Levels of Ifn‐γ, Tnf‐α, Il10, and Foxp3 mRNA were measured by qPCR using commercial primers and probes (TaqMan Gene Expression Assays) in the 7500 FAST Sequence Detection System. Transcript levels were normalized to the endogenous control glyceraldehyde‐3‐phosphate dehydrogenase mRNA (Gapdh) and expressed as fold change compared with sham‐dosed control mice using the comparative CT method (Applied Biosystems User Bulletin no. 2).

Sera were collected from all mice at necropsy (4 MPI) and evaluated by ELISA for Th1‐associated IgG2a and Th2‐associated IgG1 antibody responses to outer membrane antigens (OMPs) of WT H. hepaticus 3B1. Preparation of OMPs and ELISA were conducted as previously described [15] .

To identify and compare orthologs of bacterial Cgts, the putative H. hepaticus Cgt (deduced based on Hh0676) was used as a query sequence for searching homologs currently available in public databases using the Blastp at the website of the National Center of Biotechnology Information. Amino acid sequences of genes with ≥30% identity over 300 amino acids compared with the H. hepaticus Cgt were selected for phylogenetic analyses. Where putative Cgt orthologs are present in multiple species within the genus such as Lactobacillus, Bifidobacterium, or Clostridium, only two better characterized species were chosen. A phylogenetic tree was generated using Lasergene software (DNASTAR, Inc., Madison, WI, USA).

All statistical analyses were performed using the GraphPad Prism 5 software package. Histopathologic scores were compared using a Mann–Whitney nonparametric t‐test. Data on the colonization levels of H. hepaticus and cytokine mRNA in the tissues were analyzed using two‐tailed Student's t‐test. Serology results were compared using regression, ANOVA, and the Student's t‐test. Values of p < .05 were considered significant.

It was previously reported that inactivation of the H. pylori cgt resulted in failure of isogenic mutants to colonize stomachs of C57BL/6 mice at 12 hours postinoculation and gerbils by 1 month PI [3] , [4] . To characterize a role for cgt in the colonization and pathogenesis of H. hepaticus, colonization efficiencies of 2 cgt isogenic mutants cgtΔcatNT1 and cgtΔcatNT2 of WT 3B1 in parallel with WT 3B1 were examined by qPCR in male A/JCr mice (10 per group), a mouse model of infectious hepatitis and hepatocellular carcinoma. At 4 MPI, 9 of 10 WT 3B1‐dosed mice were qPCR positive for H. hepaticus in the ceca and 7 of 10 livers (Fig. [NaN] ). In contrast, all mice dosed with either mutant were negative for H. hepaticus in the ceca and qPCR positive only in one of 10 livers for each cgt mutant.

Infection with H. hepaticus leads to hepatic inflammation by 4 MPI in male A/JCr mice [11] , [13] , [16] , [17] . Consistent with these previous findings, 5 of 10 male mice infected with WT 3B1 developed subacute‐to‐chronic hepatitis characterized by multifocal lobular hepatitis with coagulative necrosis, and mild to moderate lymphocyte‐predominant portal and interface hepatitis (Fig. [NaN] A, 6–9 for hepatitis index scores). The observation that 5 WT 3B1‐infected mice did not develop overt hepatitis is in agreement with our previous observation that a subset of A/JCr mice appears resistant to H. hepaticus‐induced hepatic disease [13] . This resistance was supported by relatively lower levels of H. hepaticus colonization in the liver (∼100‐fold) compared with mice with hepatitis (Fig. [NaN] ). In contrast, no overt hepatic pathology was evident in male A/JCr mice inoculated with the cgt‐deficient H. hepaticus mutants, including one mouse from each group that was PCR positive for H. hepaticus in the liver (Fig. [NaN] ). Hepatitis indices in WT 3B1‐infected mice were significantly higher than in the cgt mutants‐infected mice or the sham controls (p < .05, Fig. [NaN] B); there were no significant differences in hepatitis indices between mice inoculated with the isogenic mutants or the sham‐dosed controls. These data indicate that infection with the cgt‐deficient H. hepaticus mutants did not lead to the development of chronic active hepatitis as observed in the WT 3B1‐dosed male A/JCr mice. This was consistent with the inability of the mutants to colonize the intestine and liver. Histopathologic evaluation of the intestines of all mice did not reveal significant lesions.

Previous studies showed that WT 3B1 infection significantly elevated expression of proinflammatory cytokines in splenocyte culture and in livers of male A/JCr mice [11] , [15] , [17] . To ascertain how infection with the cgt mutants influences hepatic proinflammatory responses compared with WT 3B1 infection, mRNA levels of proinflammatory Ifn‐γ and Tnf‐α, and anti‐inflammatory Il‐10 as well as Foxp3, a marker for natural regulatory T cells, were measured in cDNA prepared from hepatic RNA (Fig. [NaN] ). In the liver of mice infected with the cgt mutants, significantly lower mRNA levels of Ifn‐γ (p < .01), Tnf‐α (p < .05), Il10 (p < .05), and Foxp3 (p < .05) were observed compared with mice infected with WT 3B1. Consistent with a lack of overt hepatic pathology, mRNA levels of these hepatic cytokines in the two mice, which were positive for H. hepatics cgt mutants, were similar to those in the cgt mutant‐negative mice from the same groups. There were no significant differences in mRNA levels of these targets between the cgt mutant‐infected mice and the sham‐dosed controls (p > .2), except Ifn‐γ mRNA was lower in the cgtΔcat1‐dosed mice compared with the sham controls (p < .05) (Fig. [NaN] ).

Th1‐associated IgG2a and Th2‐associated IgG1 responses to WT 3B1 and cgt mutants infection were compared by ELISA (Fig. [NaN] ). WT 3B1‐dosed male A/JCr mice developed significantly higher levels of both IgG2a (p < .0001) and IgG1 (p < .0001) responses to infection than the cgt mutant‐dosed mice, consistent with the lack of colonization by cgt mutants (Fig. [NaN] ).

Twenty‐one Cgt‐like bacterial proteins were retrieved from the databases and phylogenetically analyzed. These gene products belong to the cd03178 family whose members are UDP‐glucose‐diacylglycerol glucosyltransferase‐like proteins and are most closely related to the GT1 super family of glycosyltransferases. Six gastric Helicobacter species and 2 EHS contain the Cgt‐like enzymes (Fig. [NaN] ). Interestingly, the sequence of the Cgt‐like protein in gastric H. mustelae is more similar to those from EHS H. hepaticus and H. bilis than those from other gastric helicobacters (Fig. [NaN] ). Additionally, the sequences of the Cgt‐like proteins in Clostridium thermocellum and Treponema brennaborenese are distantly related to those from C. methhylpentosum and T. azotonutricium, respectively, despite belonging to the same genus. It is also interesting to note that the genomes of multiple species in the genera Lactobacillus and Bifidobacterium, including probiotic L. acidophilus and B. bifidum, contain genes coding for Cgt‐like proteins.

Members of the genus Helicobacter are grouped into two groups based on their preferential colonization niches: gastric species and enterohepatic species [18] . The human pathogen H. pylori, the prototype of gastric helicobacters, can extract cholesterol from the epithelial membrane of host cells and convert it into cholesteryl‐α‐glucosides via Cgt. The loss of Cgt function leads to clearance of H. pylori by the host within 12 hours postinnoculation in C57BL/6 mice or 1 month in gerbils [3] , [4] . Intriguingly, orthologs of cgt are found in the genomes of six gastric Helicobacter species but only present in two of six EHS species analyzed, H. hepaticus and H. bilis. Our data demonstrate that H. hepaticus Cgt plays a pivotal role in establishing cecal colonization. In addition, only one of 10 mice infected with each of the two cgt mutants was PCR positive for H. hepaticus in the liver, suggesting that this enzyme is also important for hepatic colonization of this bacterium. Reduced colonization efficiency of the cgt mutants in the liver may result from inefficient translocation of the bacteria from the intestine to the liver due to their clearance in the intestine [19] . Alternatively, the cgt mutants may be cleared from the liver after their translocation. Interestingly, the cgt mutant‐colonized livers, like other H. hepaticus‐free cohorts in the same groups, did not develop hepatic inflammation nor upregulate transcription of proinflammatory cytokines, Ifn‐γ and Tnf‐α or anti‐inflammatory mediators Il10 and Foxp3; these two mice also induced less IgG1 and IgG2a responses, when compared to WT 3B1‐dosed mice. These results were not apparently caused by a polar effect or spontaneous mutations elsewhere in the genome for the following reasons. First, we previously showed that the allelic replacement of ureA/B with catNT lacking its transcription terminator did not downregulate the transcription of its downstream gene ureI in the ure operon of H. hepaticus[11] . Second, two cgt‐deficient mutants containing a catNT with opposite orientations conferred similar clinical manifestations. Third, a ‐35 (TTTATA), a 21‐nucleotide spacer and a ‐10 (TAAAAT) elements conserved among bacterial promoters precede HH0677 which locates immediately downstream of the H. hepaticus cgt (Hh0676), indicating that transcription of Hh0677 is likely controlled under its own promoter (Fig. [NaN] A).

Pathogenic bacteria, which are likely to have coevolved with their hosts, have developed multiple strategies for colonization success. Bacteria utilize various protein secretion systems, dynamic cell surface structures as well as virulence factors to evade host immunity, survive in complex ecosystems, and maintain pathogenicity. Cholesterol is present in the membranes of eukaryotic organisms and has various cellular functions including maintenance of the membrane structure, modulation of membrane fluidity, and involvement in cell signaling [20] , [21] . However, the mechanisms underlying the importance of the biosynthesis of cholesteryl‐α‐glucosides for the gastrointestinal and hepatic colonization of helicobacters require further studies. Accumulated evidence suggests that Cgt‐mediated cholesterol α‐glucosylation confers several advantages for H. pylori to survive in the host. First, this bacterium utilizes this process to escape phagocytosis and eventual ingestion by antigen presenting cells including macrophages, a first line of defense in innate immunity [3] . Second, this process helps H. pylori evade T‐cell‐mediated immune responses of the host via reduction in T‐cell proliferation [3] , [22] . Third, α‐glucosylation of cholesterol apparently increases resistance to bactericidal effects by the host mucus‐associated O‐glycans and cathelicidin LL‐37 as well as select antibiotics [1] , [4] , [23] . These advantages may be attributed to reduction in the availability of cholesterol for the host antibacterial response. We hypothesize that H. hepaticus, which preferentially colonizes the large intestine and the liver, also encounter cholesterol‐mediated antibacterial effects similar to that described for H. pylori colonizing the stomach. It is known that approximately 80% of total body cholesterol is endogenously produced primarily in the liver and intestine [24] .

Epidemiological data indicate there is an inverse association between H. pylori infection and prevalence of allergic asthma [25] , [26] . In vivo studies demonstrated that infection with H. pylori protected neonatal mice from experimentally induced asthma via expansion of regulatory T cells driven by dendritic cell‐derived Il‐18 or neutrophil activation protein‐mediated downregulation of Th2 responses [27] , [28] , [29] . Interestingly, a recent report indicated that treatment with a glycolipid P157, a chemically synthesized version of the H. pylori αCAG described previously [1] , [3] , protected suckling BALB/c mice against ovalbumin (allergen)‐induced airway hyperreactivity [30] . This protection resulted from activation of natural killer T cells mediated by CD1d (related to class I MHC molecules) [30] . Given that production of the H. pylori αCAG requires Cgt, the presence of cgt orthologs in EHS such as H. hepaticus and H. bilis as well as other enteric bacteria implies that bacterial glycolipids catalyzed by Cgt may have broad immune regulatory effects in human health and disease.

Blastp search and phylogenetic analyses suggest that diverse bacterial species have the potential to synthesize Cgt‐like proteins. Distribution of this gene in members of the same genus is not ubiquitous; for example, only some members of Helicobacter or Clostridium are positive for homologs of Cgt. Whether these predicted proteins have enzymatic activity to conduct α‐glucosylation of cholesterol remains unknown. However, significant sequence homology between H. pylori and other Cgt‐like proteins suggests that at least some of these proteins could be Cgts.

Taken together, the results from this study and previous reports demonstrate that cholesterol α–glucosylation plays a pivotal role in colonization of both gastric and enterohepatic helicobacters in the gastrointestinal and hepatic tissues of mice and gerbils. Given the role of H. pylori cholesteryl α‐glucosides in phagocytosis, immune evasion, and protection against allergic asthma, our results further highlight the physiological and pathogenic potential of these bacterial cholesteryl glucosides and warrant further investigations into expression and biologic functions of cholestoryl glucosides in medically important cgt‐positive bacteria, including additional Helicobacter, Clostridium, Lactobacillus, and Bifidobacterium species.

This study was partially supported by the NIH grants to JGF: R010D011141, P01CA026731, R01AT004326, P30‐ES002109.

Competing interests: We have no competing conflict interest.

Graph: Construction of cgt ‐deficient Helicobacter hepaticus mutants. ( A ), Genomic locus (indicated by numbers) of H. hepaticus cgt ( HH 0676) and its flanking genes. ( B ), Genotypes of WT 3B1 and 2 cgt mutants are schematically depicted. Locations of PCR primers used for generating and characterizing the mutants are denoted.

Graph: Deficiency in cgt impaired the ability of H. hepaticus to persistently colonize the intestine and liver of male A/ JC r mice (n = 10 per group). Numbers represent the copies of the H. hepaticus genome per μg mouse DNA in the individual samples L, liver; Ce, cecum.

Graph: Histology of the livers of male A/ JC r mice dosed by WT 3B1 o r cgt ‐deficient mutants cgtΔcat1 and cgtΔcat2 at 4 months postinoculation. ( A ). Representative histopathology. (No infection) Normal liver, sham‐inoculated mouse. ( cgtΔ1 ) and ( cgtΔ2 ) Minimal hepatic inflammation present in the livers of mice dosed with the isogenic mutants. (3B1) Subacute hepatitis with hepatocellular coagulative necrosis induced by WT 3B1. ( B ). Hepatitis index for male A/ JC r mice. The cgt mutant‐infected mice developed significantly less severe hepatitis compared to those infected with WT 3B1.

Graph: Mice dosed with the cgt mutants expressed lower m RNA levels of hepatic proinflammatory Th1 cytokine Tnf‐α and Ifn‐γ as well as anti‐inflammatory mediators Il10 and Foxp3 compared with the WT 3B1‐dosed mice. In each sample, the target gene m RNA was normalized to the transcript levels of Gapdh. Numbers on the y axis represent mean fold change of the individual m RNA levels in reference to the control group (0 defined as no change with its standard deviations denoted). p values versus WT 3B1‐infected samples: *<.05, **<.01, ***<.001.

Graph: Male A/ JC r mouse Th1‐associated IgG2a and Th2‐associated IgG1 responses to WT 3B1 outer membrane antigens measured by ELISA. WT 3B1‐dosed mice produced significantly higher levels of sera IgG2a and IgG1 compared with the cgt mutant‐dosed mice, which is consistent with the failure of these cgt mutants to colonize or persist in the host.

Graph: Phylogenetic analyses of bacterial Cgts. Majority of Cgt protein sequences were deduced from genome sequences. The phylogenetic tree was constructed using the DNASTAR software package. *, only N‐terminal sequences (∼390 aa) in these proteins, which display sequence similarity to helicobacter Cgts, are included for phylogenetic analysis.