ABSTRACT
Hemolysins have been found to possess a variety of functions in bacteria, including a role in virulence. Helicobacter pylori demonstrates hemolytic activity when cultured on unlysed blood agar plates which is increased under iron-limiting conditions. However, the role of an H. pylori hemolysin in virulence is unclear. Scrutiny of the H. pylori 26695 genome sequence suggests the presence of at least two distinct hemolysins, HP1086 and HP1490, in this strain. Previous studies have shown that the in vitro hemolytic activity of H. pylori is reduced when it is coincubated with dextran 5000, suggesting the presence of a pore-forming cytolysin. HP1086 has homology to pore-forming cytolysins (TlyA) from other bacterial species, and the introduction of the clonedH. pylori tlyA gene into a nonhemolyticEscherichia coli strain conferred hemolytic activity. AnH. pylori tlyA defined mutant showed reduced in vitro hemolytic activity, which appears to be due to pore formation, as the hemolytic activity of the wild-type strain is reduced to the same level as the tlyA mutant by the addition of dextran 5000. The mutant also showed reduced adhesion to human gastric adenocarcinoma cells and failed to colonize the gastric mucosa of mice. These data clearly suggest a role in virulence for H. pyloriTlyA, contrary to the suggestion that hemolytic activity is an in vitro phenomenon for this pathogen.
The gram-negative bacteriumHelicobacter pylori is a human-specific gastric pathogen that survives largely within the gastric mucus layer in the stomach (5). Infection with H. pylori is associated with the development of duodenal and gastric ulcers, gastric adenocarcinoma, and mucosa-associated lymphoid tissue lymphoma (18). Many of the factors involved in H. pylori virulence have been studied in detail, including urease, motility, the VacA cytotoxin, CagA and the cag pathogenicity island, the neutrophil-activating protein NapA, adhesins, iron acquisition, and lipopolysaccharide (for a review, see reference22). Despite this range of virulence determinants, VacA is the only toxin so far identified, and the role of this toxin in in vivo pathogenesis has been questioned (22). Furthermore, the H. pylori determinants responsible for inducing inflammation, a hallmark of active gastritis, remain obscure (16). The availability of the genome sequences for H. pylori strains 26695 (34) and J99 (1) provides a powerful tool not only to investigate new potential virulence factors but also to identify genes responsible for known phenotypic characteristics. H. pylori is hemolytic when grown on unlysed blood agar plates, and hemolytic activity is increased under iron-limiting conditions (32). Six chromosomal fragments from H. pylori ATCC 49503 have been identified as containing hemolytic factors based on the capacity to confer on a nonhemolyticEscherichia coli strain the ability to lyse red blood cells (RBC) (10). However, no further characterization of these putative genes has been reported, and the role of H. pylori hemolytic activity in pathogenesis is unclear, to the point that it has been suggested that the hemolytic activity of H. pylori is not a significant virulence factor in human infection (26).
Hemolysins are defined as bacterial toxins that lyse erythrocytes by cell wall disruption and are often more correctly referred to as cytolysins. Hemolysins have been demonstrated in a number of pathogens, including streptococcal and staphylococcal species, E. coli, Serpulina hyodysenteriae, Mycobacterium tuberculosis, Trypanosoma cruzi, and Listeria monocytogenes (2, 4, 24), and some of these have been shown to be important virulence factors (6). Hemolytic activity can be demonstrated in vitro by the ability to lyse erythrocytes. This phenotype is easily measured colorimetrically by quantitating the release of hemoglobin into solution. The in vivo significance of RBC lysis by hemolysins is unclear, although erythrolysis has been proposed as a mechanism for iron acquisition in an iron-deficient microenvironment (29). For example, the hemolysin produced by Vibrio vulnificus can lyse erythrocytes and eucaryotic cells, which in turn may free heme-containing compounds to serve as a source of iron during sepsis and wound infection (20). Alternatively, hemolysins may lyse or disrupt membranes of other cell types, for example, leukocytes or gastric epithelium cells, thus enhancing bacterial survival and making preferred metabolites more accessible (3, 28). H. pylori hemolysins could lyse the cytoplasmic or vacuolar membranes of phagocytic cells it encounters or damage epithelial cell membranes.
Hemolysins can be separated into three categories based on the mechanism of action against target cell membranes: enzymatic (which includes phospholipases), pore forming, and surfactant (28). There are at least two putative hemolysin gene sequences in the H. pylori 26695 genome, HP1086 and HP1490 (34), though it is probable that H. pylori possesses a number of proteins with hemolytic activity. For example, the H. pylori phospholipase PldA (HP0499) has been shown to possess hemolytic as well as phospholipase activity (9). HP1086 has homology to the pore-forming cytolysins from S. hyodysenteriae and M. tuberculosis(25, 38). Pore-forming cytolysins function by first attaching to the cell membrane. Once attached, the cytolysin penetrates and disrupts the membrane by forming a pore, leading to alteration of membrane permeability and hence cytolysis (28). Binding is usually temperature independent and can occur at 4°C, although some pore-forming cytolysins may require higher temperatures to function. For example, Streptococcus pneumoniae streptolysin O is active only at 37°C, which may relate to the fluidity of the cell membrane (28). Pore-forming cytolysins are sensitive to the presence of sugars, since solutes with molecular diameters larger than the size of the pore formed in the target cell membrane can prevent target cell lysis, as has been demonstrated with the pore-forming cytolysins from both S. hyodysenteriae andM. tuberculosis (25; R. A. Stabler and B. W. Wren, unpublished data). Smaller solute molecules can pass through the pores, resulting in an osmotic gradient and ultimately cell lysis. Osmotic protection experiments can be used to define discrete, finite pore sizes lying between the diameters of two solutes, using the transition from a state of little protection against cytolysin-induced lysis by one solute to full protection provided by a larger solute (23). The hemolysis of sheep RBC by H. pylori has been shown to be inhibited by the presence of dextran 5000, suggesting that one possible mode of action is by pore formation (32).
In this study we investigated the role of the H. pyloripore-forming cytolysin orthologue HP1086 (termed tlyA) by the construction and characterization of an isogenic H. pylori tlyA mutant and the initial characterization of the recombinant TlyA protein. Mutation of the tlyA gene resulted in a reduction in the in vitro hemolytic activity which was unaffected by the presence of dextran 5000. However, coincubation of wild-type H. pylori with dextran 5000 reduced the level of hemolytic activity to that of the tlyA mutant, suggesting that the mode of action of TlyA is by pore formation. Finally, we present evidence that the expression of TlyA is a prerequisite for the colonization of H. pylori in the mouse model, thus suggesting the importance of TlyA as a virulence determinant.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.The bacterial strains and plasmids used in this study are listed in Table1. H. pylori strains were stored at −80°C in brain heart infusion broth (Oxoid, Basingstoke, United Kingdom) containing 15% (vol/vol) glycerol and 10% (vol/vol) fetal calf serum (Sigma-Aldrich, Poole, United Kingdom). Strains were grown in brain heart infusion broth supplemented with 10% (vol/vol) fetal calf serum or on Helicobacter selective agar (DENT), consisting of Blood Agar Base No. 2 (Oxoid) supplemented with 7% (vol/vol) lysed defibrinated horse blood (TCS Microbiology, Botolph Claydon, United Kingdom) and DENT selective supplement (Oxoid), in a microaerobic atmosphere at 37°C. Blood agar plates consisted of Columbia agar (Oxoid) supplemented with 7% (vol/vol) defibrinated horse blood. Stab agar motility plates consisted of Mueller-Hinton broth (Oxoid) supplemented with 10% fetal calf serum and 0.3% (wt/vol) Agar Bacteriological No. 1 (Oxoid) and were inoculated as described previously (15). E. coli strains were routinely grown in Luria-Bertani broth or on Luria-Bertani agar. The antibiotics used for selection purposes were ampicillin (100 μg/ml) and kanamycin (20 μg/ml for H. pylori and 50 μg/ml for E. coli).
Bacterial strains and plasmids used in this study
DNA manipulations.Unless otherwise stated, plasmid and chromosomal DNA extractions, restriction enzyme digests, DNA ligations, and transformations were performed by standard procedures (30) using enzymes supplied by Promega (Southampton, United Kingdom). Transformations into the E. coli XL2-Blue MRF′ strain (Stratagene Europe, Amsterdam, The Netherlands) were performed following the manufacturer's protocol. All chemicals were purchased from Sigma-Aldrich. The oligonucleotide primers used for PCRs were purchased from Genosys Biotechnologies (Europe) Ltd. (Pampisford, United Kingdom) and are summarized in Table2. Sequencing of cloned DNA was performed by the dideoxynucleotide chain termination method with a PRISM sequencing kit (Applied Biosystems, Warrington, United Kingdom).
Oligonucleotides used for PCR
Construction of a defined H. pylori hemolysin mutant.Specific primers RS7F and RS7R were designed from theH. pylori 26695 genome sequence to amplify the putativetlyA gene (HP1086) from H. pylori 26695 chromosomal DNA. The amplified PCR product was cloned into pUC19. A defined deletion and unique BglII site were introduced into the cloned gene by inverse PCR mutagenesis (IPCRM) using the primer pair RS7 TIF and RS7 TIR shown in Table 2, as described previously (8, 37). A 1.4-kb BamHI restriction fragment of plasmid pJMK30 containing a gene encoding resistance to kanamycin (aph3′-III) (13) was cloned into the uniqueBglII site. This construct was introduced into theH. pylori SS1 wild-type strain by electroporation (31). Double-crossover mutants were selected and screened as described previously (9, 14).
Assay for hemolysis of RBC.The hemolysis assay was performed basically as described previously (9). Briefly, RBC from defibrinated horse blood were washed in phosphate-buffered saline (PBS) and resuspended to the initial volume with sterile PBS (pH 7.4). This washed RBC suspension was taken to be 100% RBC. Twenty-four-hour cultures of SS1 and RS7 were resuspended in PBS to obtain a 20% (wt/vol) suspension. Duplicate reactions containing 4 and 2% H. pylori were incubated with a final concentration of 2% RBC for 2 h at 37°C with shaking at 200 rpm. Controls, replacing the bacterial suspension with either PBS (0% hemolysis) or H2O (100% hemolysis), were also included. The samples were centrifuged at 3,000 × g for 5 min, and readings of optical density at 540 nm were recorded in duplicate. Each strain was tested in triplicate, and the results were expressed as a percentage of 100% hemolysis. Dextran sulfate (average molecular weight, 5,000) was purchased from Sigma-Aldrich, dissolved in PBS to a final concentration of 20% (wt/vol), and referred to as dextran 5000.
Preparation of H. pylori soluble and insoluble fractions. H. pylori wild-type cells grown on DENT agar for 48 h were washed in PBS and resuspended in 1 ml of sonication buffer (0.1 M sodium phosphate buffer, pH 8.0) for every 0.4 g of wet cells, to give a final suspension of approximately 40%. The cells were lysed by five 30-s bursts of ultrasound (Ultrasonic Processor; Jencons, Leighton Buzzard, United Kingdom) with 30-s cooling periods on ice between each burst. Sonicates were centrifuged at 10,000 × g for 20 min, and the soluble fraction and insoluble debris were separated and stored at −20°C.
Recombinant H. pylori TlyA studies.The cloned H. pylori tlyA gene was isolated from plasmid pRS7 by digestion with restriction enzymes EcoRI andPstI and ligated into the pRSET-A, -B, and -C expression vectors (Invitrogen, Groningen, The Netherlands), which were previously digested with the same restriction enzymes. These three separate ligations were then transformed into E. coli XL2-Blue MRF′ cells to form plasmids pCM-A, pCM-B, and pCM-C, respectively. Clones containing the correctly sized inserts were identified using vector-specific primers pRSETI and pRSETII. Plasmid DNA was purified from a positive clone for each of the three expression vectors and transformed into E. coli BL21(DE3)(pLysS).
Cultures of the strains containing the three different expression vectors were grown to mid-log phase, and then IPTG (isopropyl-β-d-thiogalactopyranoside) was added to a final concentration of 1 mM. The cultures were grown at 37°C with shaking at 200 rpm. Samples were taken at time points over a 4-h period. Whole-cell extracts were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) performed using the Mini-Protean II system (Bio-Rad Laboratories, Hemel Hempstead, United Kingdom) using a 12% acrylamide resolving gel. For Western blot analysis, proteins were transferred to nitrocellulose (Hybond-C pure; Amersham Pharmacia Biotech, St. Albans, United Kingdom) using the Mini Trans-Blot system (Bio-Rad Laboratories). Blots were incubated with Anti-Xpress monoclonal antibody (dilution of 1:5,000; Invitrogen). Bound antibodies were detected using polyvalent anti-mouse immunoglobulin horseradish peroxidase conjugate (dilution of 1:2,000; Sigma-Aldrich).
Recombinant TlyA was purified from the E. coliBL21(DE3)(pLysS)(pRSET) strain expressing the recombinant H. pylori TlyA using Ni-nitrilotriacetic acid columns (Qiagen, Crawley, United Kingdom) according to the manufacturer's instructions.
Adherence to AGS cells.The adherence assay was carried out as described previously (11). H. pyloristrains (≈5 × 108 organisms) were incubated with human gastric adenocarcinoma (AGS) cells (≈5 × 106cells) at 37°C for 1 h with agitation (150 rpm). Nonadherent bacteria were removed by washing with 10 ml of 15% sucrose solution. Cells were washed once with PBS and subsequently incubated with a 1:5 dilution of polyclonal anti-H. pylori antibody (SkyTek Laboratories, Logan, Utah) on ice for 30 min. After being washed with 15 ml of PBS, the cells were then incubated for an additional 30 min on ice in a 1:20 dilution of fluorescein isothiocyanate-conjugated goat anti-rabbit immunoglobulin G (Sigma-Aldrich). The cells were subsequently washed and resuspended in 1 ml of 1% formaldehyde for flow cytometric analysis. A FACScan flow cytometer (Becton Dickinson, San Jose, Calif.) was used to measure bacteria adhering to AGS cells and was gated to include single cells and to exclude cell debris and unbound bacteria. Each strain was tested in triplicate, and the results are presented as a percentage figure calculated from the number of AGS cells adhered to by H. pylori strains and the total number of AGS cells analyzed by flow cytometry.
Animal infection with H. pylori SS1 strains.The mouse model of H. pylori infection was performed as described previously (21). Briefly, 5-week-old specific pathogen-free CD1 male mice (Charles River, Calco, Italy) were maintained on a 12-h light-dark schedule in an air-conditioned animal facility. Water and food were provided ad libitum. Before inoculation and sacrifice, mice were subjected to fasting for 24 h, having access to water only. H. pylori SS1 wild-type and tlyA mutant strains were grown on DENT agar plates for 48 h. Bacterial suspensions were prepared immediately before administration to mice. Bacteria from a single plate were collected with a sterile cotton swab and resuspended in the appropriate volume of PBS to give a suspension containing approximately 1010 CFU/ml. Prior to inoculation, fasting mice received 0.2 ml of 0.2 M NaHCO3 to neutralize gastric acidity. All inoculations (0.1 ml ≈ 109 CFU) were performed intragastrically with a Luer lock stainless steel gavage with a round tip and were repeated twice on days 3 and 5.
At days 12 and 33 after the first bacterial challenge (1 and 4 weeks after the final inoculation), mice were sacrificed by cervical dislocation, and the stomachs were removed and cut along the lesser curvature. The whole stomach was homogenized in 1 ml of PBS. Serial dilutions of the obtained suspension were plated onto DENT agar plates and incubated for 7 days. Identity of H. pyloricolonies was confirmed by visual inspection (colony morphology), urease test, and carbolfuchsin staining.
Statistical analysis.The statistical significance of experimental data was determined using the unpaired t test performed with the InStat 2.03 statistical package (GraphPad Software, San Diego, Calif.), unless otherwise stated.
RESULTS
Construction of a defined H. pylori hemolysin mutant.The putative H. pylori hemolysin HP1086 shows 34.9% identity and 55.9% similarity to the M. tuberculosis cytotoxin (or hemolysin) TlyA (38) and 35.3% identity and 53.0% similarity to the S. hyodysenteriae hemolysin A (TlyA) (25). An orthologue was identified from the recently sequencedCampylobacter jejuni 11168 genome with 40.8% identity and 61.9% similarity (27). The putative H. pylori hemolysin HP1086 was thus termed TlyA. PCRs with primers RS7F and RS7R consistently amplified a single band of 705 bp fromH. pylori 26695 chromosomal DNA, which represents 100% of the H. pylori tlyA gene. The nucleotide sequences of two independently derived tlyA clones were determined and found to be identical to each other and to the sequence of HP1086 (data not shown). The predicted molecular mass of the TlyA protein is 26,643 Da (http://www.tigr.org/tdb/CMR/ghp/htmls/SplashPage.html ). AnH. pylori tlyA mutant was constructed by allelic replacement as summarized previously (9, 14) and was termed RS7. The H. pylori tlyA cloned gene fragment was mutated by IPCRM through introduction of a 53-bp deletion and the insertion of a kanamycin resistance cassette (8, 37). Typically, electroporation resulted in 250 kanamycin-resistant colonies per experiment. PCR analysis using primers RS7F and RS7R confirmed that a double-recombination event had occurred (data not shown). Southern blot analysis of DNA from the putative mutant using a 705-bpH. pylori tlyA probe confirmed that the mutated gene had undergone a double-recombination event (data not shown). To check for possible polar effects caused by the insertion of the kanamycin cassette into the tlyA gene, reverse transcription-PCR and cDNA analysis were performed, showing transcription of the adjacent genes HP1085 and HP1087 in both SS1 and the tlyAmutant (data not shown).
No differences in colony morphology or growth rate were observed between the tlyA mutant and the wild-type strain SS1 when grown on DENT plates. The standard 0.3% stab agar motility test was used to assess motility (17). Both the wild-type strain SS1 and the tlyA mutant formed diffuse colonies with large swarming halos (Fig. 1A). However, thetlyA mutant showed reduced zones of clearing (hemolysis) when grown on blood agar plates, in contrast to SS1 (Fig. 1B).
(A) Stab agar testing of H. pylori SS1 wild-type (w/t) and tlyA mutant strains. Plates consisted of 0.3% agar and were incubated for 5 days after inoculation. (B) Blood agar clearing by H. pylori SS1 wild-type strain. ThetlyA mutant strain shows reduced hemolytic activity. Plates contained 7% (vol/vol) blood and were incubated for 3 days after inoculation.
in vitro hemolytic activity is reduced in the tlyAmutant. H. pylori suspensions (4 and 2% [wt/vol]) were analyzed for the quantitative hemolysis of RBC, and the results are presented as a percentage of 100% hemolysis (Fig.2). Hemolytic activity was reduced in thetlyA mutant compared to the wild-type SS1 strain. Statistical analysis showed the differences in hemolytic activity observed with the two different concentrations of H. pylori suspensions to be highly significant (4% [wt/vol], P = 0.0001 ; 2% [wt/vol], P = 0.0003 ). Hemolysis assays were performed on a secondtlyA mutant constructed independently from RS7. Similar results were observed, strongly suggesting that the phenotype was due to mutation of the tlyA gene and not to an unrelated mutation elsewhere on the chromosome (data not shown). Hemolysis assays performed on wild-type samples (whole cells and soluble and insoluble fractions) showed hemolytic activity only for whole cells and the insoluble fraction.
Quantitative determination of hemolytic activity ofH. pylori strains, showing the level of hemolysis of RBC after 2 h at 37°C when coincubated with either 4 or 2%H. pylori whole-cell suspensions. The level of hemolysis is presented as a percentage of 100% hemolysis. The reduction in hemolysis observed with the tlyA mutant is extremely significant at both the 4% (wt/vol) ( P = 0.0001 ) and the 2% (wt/vol) ( P = 0.0003 ) bacterial concentrations. w/t, wild type.
The hemolysis of sheep RBC by H. pylori has been shown to be inhibited by the presence of dextran 5000 (32). The hemolysis assays were repeated in the presence of increasing concentrations of dextran 5000 (Fig. 3). The hemolysis of horse RBC by the wild-type SS1 strain was reduced to a level similar to that of the tlyA mutant, which was unaffected by the presence of dextran 5000. These data would suggest that the hemolysis attributable to H. pylori TlyA is due to pore formation.
Quantitative determination of hemolytic activity ofH. pylori strains in the presence of dextran 5000, showing the level of hemolysis of RBC after 2 h at 37°C when coincubated with 2% H. pylori whole-cell suspensions. The level of hemolysis is presented as a percentage of 100% hemolysis. The addition of dextran 5000 reduces the level of hemolysis observed with the SS1 wild-type (w/t) strain to a level which is not significantly different ( P > 0.5 ) from that of the tlyA mutant.
Recombinant H. pylori TlyA studies.Further evidence for the hemolytic activity of the H. pyloriTlyA was provided by the expression and analysis of the recombinant protein. The H. pylori tlyA gene was successfully cloned into the three pRSET expression vectors. To check which of the pRSET clones contained the tlyA gene inserted in the correct reading frame for expression, clones were screened for the expressed protein with Anti-Xpress monoclonal antibody. Only the pCM-A clone expressed the recombinant TlyA, which had an approximate molecular mass of 30 kDa (which includes ≈3 kDa for the His6 tag) (Fig.4). The predicted size of TlyA is 26.6 kDa. The plasmid pCM-A was sequenced to confirm that the completetlyA gene had been successfully cloned in the correct reading frame for expression. Western analysis of induced protein expression over a 4-h period showed the amount of protein expressed decreased over time (Fig. 5), suggesting that the H. pylori TlyA may be toxic forE. coli.
SDS-PAGE analysis of E. coliBL21(DE3)(pLysS)(pCM-A, -B, and -C) cell extracts after IPTG induction. pCM-A, -B, and -C are the pRSET-A, -B, and -C expression vectors containing the cloned H. pylori tlyA gene in the three different reading frames. 1 and 2, 1 and 2 h after induction, respectively; NC, negative control (no H. pylori tlyAgene insert).
Western blot of E. coliBL21(DE3)(pLysS)(pCM-A) cell extracts (50 μg) expressing the His6-tagged H. pylori TlyA protein at different time points after IPTG induction. The primary antibody used was the Anti-Xpress monoclonal antibody. Bound antibodies were detected using polyvalent anti-mouse immunoglobulin horseradish peroxidase conjugate.
Purification of the recombinant TlyA from the pCM-A clone was investigated. After a preliminary separation of fractions, only the insoluble fraction showed hemolytic activity in the hemolysis assay. A level of 78.7 ± 2.6% hemolysis was recorded from the insoluble fraction from a 40-ml culture of E. coliBL21(DE3)(pLysS)(pCM-A). However, further purification of the recombinant protein using a Ni-nitrilotriacetic acid column led to the loss of hemolytic activity, probably due to the loss of protein conformation caused by purification under denaturing conditions.
Recombinant H. pylori TlyA confers hemolytic activity to the nonhemolytic E. coli strain BL21(DE3)(pLysS).Due to difficulties in purifying an active recombinant H. pylori TlyA, the activity of the protein in a nonhemolytic E. coli strain was studied. The hemolysis assay was performed on E. coliBL21(DE3)(pLysS)(pCM-A) (host E. coli strain plus vector expressing H. pylori TlyA), usingE. coli BL21(DE3)(pLysS) (host E. coli strain) and E. coli BL21(DE3)(pLysS)(pCM-B) (host E. coli strain plus vector containing the clonedH. pylori tlyA gene but not expressing H. pylori TlyA) as negative controls to exclude any possible hemolytic activity due to the E. coli host strain or the vector containing the cloned tlyA gene. Cell extracts ofE. coli BL21(DE3)(pLysS)(pCM-A) possessed significant hemolytic activity when induced by IPTG compared to the control cell extracts (Table 3). Statistical analysis showed this increase in hemolytic activity to be highly significant ( P = 0.0018 and 0.0017, respectively).
Hemolytic activity of 4% (wt/vol) E. coli cell extracts from IPTG-induced cultures
Adherence to AGS cells is reduced in the tlyAmutant.The ability of the tlyA mutant to adhere to cultured human AGS cells was analyzed. The tlyA mutant showed a reduced level of adherence to AGS cells compared to SS1. The results are presented as a percentage calculated from the number of AGS cells adhered to by H. pylori strains and the total number of AGS cells analyzed by flow cytometry. The tlyAmutant adherence was measured as 71.4 ± 7.5% compared to 92.8 ± 2.6% observed with SS1. Statistical analysis showed the difference in adherence to be highly significant ( P = 0.0017 ). The reduction in adherence to AGS cells is not a result of the introduction of the kanamycin resistance cassette during mutagenesis, as H. pylori SS1 pldA, ureB,and flhB mutants constructed by the same strategy show wild-type levels of adherence (9, 14). The results suggest that the adherence of H. pylori to epithelial cells is affected by a mutation in the tlyA gene.
The H. pylori tlyA mutant shows a reduced ability to colonize mice.The murine model for H. pylori colonization (21) was used to assess the role of TlyA in virulence (Table 4). At the 10-day time point (1 week after the final challenge), only 1 of 34 (2.9%) mice inoculated with the tlyA mutant were colonized by H. pylori. In contrast, 21 of 29 (72.4%) mice challenged with the SS1 wild-type strain were colonized. At the 31-day time point (4 weeks after the final challenge), all five mice inoculated with the SS1 wild-type strain were colonized. However, none of the five mice challenged with the tlyA mutant were colonized at this time point.
Colonization data from specific pathogen-free CD1 mice challenged with either the SS1 wild-type or the tlyA mutant strain
DISCUSSION
Hemolysins of many gram-negative and gram-positive bacteria are thought to contribute to the pathogenesis of infections caused by these organisms. Although hemolysins by definition lyse RBC, the lysis of other host cells also probably contributes to pathogenesis. For example, human leukocytes exposed to E. colialpha-hemolysin demonstrate an oxidative burst and decreased phagocytic and chemotactic abilities (7). Several studies suggest that H. pylori is a hemolytic bacterium, but to date the nature and mode of action of its hemolysins are unclear (3, 10, 32).
In this study, we present evidence for the role of a pore-forming cytolysin orthologue, TlyA, in H. pylori virulence. It has been suggested that the hemolytic activity of H. pylori is not a significant virulence factor (26) because this activity is directed against erythrocytes of different animals and not against human erythrocytes (3, 10). However, Vibrio cholerae cytolysin (VCC), which is also a pore-forming cytolysin, interacts with a high-affinity binding site present on rabbit RBC prior to oligomerization and pore formation (39). This as yet unidentified binding site is absent on human RBC, which are less susceptible to VCC action (39). VCC binds to human intestinal cells, causing cell death, possibly resulting in diarrhea (40). This raises the possibility that the in vitro hemolytic phenotype of H. pylori is not directly related to the in vivo function.
A hallmark of H. pylori colonization is chronic inflammation of the gastric mucosa and infiltration of inflammatory cells (5). The process by which H. pyloriattracts mononuclear and polymorphonuclear leukocytes to the site of colonization is still unclear. Recently, the inflammation induced byH. pylori has been shown not to be associated with CagA, VacA, or PicB (CagE) (16). Activation of human neutrophils and monocytes by H. pylori with different virulence genotypes showed that cagA-negative,vacA-S2, and picB-negative strains retain their inflammatory capacity. Other H. pylori factors are likely to be involved in this proinflammatory activation, such as the neutrophil-activating protein NapA (12). Interestingly, anE. coli hemolysin causes the release of inflammatory mediators at sublytic concentrations (36). It is possible that H. pylori proteins with hemolytic activity, such as TlyA or PldA, may have an analogous function in the release of inflammatory mediators and inflammation in H. pylori-associated disease.
Other studies have shown that H. pylori produces at least two proteins with homology to the family of alpha-hemolysins (RTX cytotoxins) (32). These proteins, 42 and 72 kDa in size, were identified using Western analysis with monoclonal antibodies produced against the Bordetella pertussis adenyl cyclase toxin. Using a polyclonal anti-E. coli alpha-hemolysin antibody produced a broader reactivity but included the 42- and 72-kDa bands. The predicted size of TlyA is 26.6 kDa, which was verified by SDS-PAGE analysis of the recombinant TlyA protein. The size of the other putative hemolysin (HP1490), identified from analysis of theH. pylori 26695 genome sequence, is predicted to be 50.4 kDa. This suggests that the H. pylori hemolysin orthologues are not the proteins identified using the monoclonalB. pertussis antibodies as orthologues to RTX cytotoxins. However, six chromosomal fragments from H. pylori ATCC 49503 appear to contain genes which code for proteins with hemolytic activity (10), which may account for the broader reactivity observed using the polyclonal E. coliantibody. Mutation of the tlyA gene reduced the in vitro hemolytic activity of H. pylori to approximately half that of the wild-type strain when 2% (wt/vol) whole-cell suspensions were used. However, the hemolytic activity of the tlyAmutant was increased to about two-thirds of that of the wild-type strain with 4% (wt/vol) whole-cell suspensions. This indicates the presence of other H. pylori proteins which have in vitro hemolytic activity. In a similar assay, mutation of theH. pylori pldA gene (which encodes a phospholipase) results in an even greater reduction in hemolytic activity (9). Taken together, these data clearly suggest that the in vitro hemolytic activity observed with H. pylori is the result of the action of multiple proteins.
The close contact between the bacterium and the gastric epithelium suggests that cell-bound rather than secreted factors will contribute more to the pathogenicity of H. pylori. In this study, different fractions (whole cells and soluble and insoluble fractions) of wild-type H. pylori were analyzed for hemolytic activity. Only the whole cells and insoluble fractions possessed hemolytic activity, suggesting that TlyA is attached in some way to the cell wall and not actively secreted. Interestingly, the tlyA mutant shows a reduced ability to adhere to AGS cells, in contrast to the pldA mutant (9). Pore-forming cytolysins function by initially attaching to the host cell membrane. It is tempting to speculate that TlyA plays a role in the adherence of H. pylori to gastric epithelial cells. Certainly the fact that H. pylori hemolytic activity is increased under iron-limiting conditions (32) would point to a significant in vivo function. Interestingly, H. pylori hemolysin mutants (specific genes not specified) cannot induce tyrosine phosphorylation of CagA in host cells but can still induce interleukin 8 expression from a human gastric epithelial cell line (33). NeithertlyA nor pldA (9) mutants are able to colonize the H. pylori mouse model, providing further evidence for the importance of this hemolytic activity in colonization and pathogenesis.
We have shown one of the hemolysin orthologues from the H. pylori 26695 genome sequence to encode an active hemolysin, which appears to act by pore formation, capable of conferring hemolytic activity when expressed in a nonhemolytic E. colistrain. Mutation of the tlyA gene abrogates the ability ofH. pylori to colonize mice and also reduces the bacterium's ability to adhere to gastric epithelial cells. Whether these two observations are linked is unclear. These results strongly indicate that H. pylori hemolysin TlyA is an important virulence determinant and highlight the importance of deciphering the role of hemolytic activity in H. pylori pathogenesis.
ACKNOWLEDGMENTS
We gratefully acknowledge Lynne Batty for technical assistance, Richard Ferrero for the generous gift of SS1, Giuseppe Del Giudice for assistance with the H. pylori mouse model, and Katherine Fielding for assistance with statistical analysis.
This work was supported by the MRC and the Joint Research Board of St. Bartholomew's Hospital.
Notes
Editor: J. D. Clements
FOOTNOTES
- Received 18 October 2000.
- Returned for modification 16 November 2000.
- Accepted 13 December 2000.
- Copyright © 2001 American Society for Microbiology
