Helicobacter pylori Adhesin Binding Fucosylated Histo-Blood Group Antigens Revealed by Retagging

See allHide authors and affiliations

Science  16 Jan 1998:
Vol. 279, Issue 5349, pp. 373-377
DOI: 10.1126/science.279.5349.373


The bacterium Helicobacter pylori is the causative agent for peptic ulcer disease. Bacterial adherence to the human gastric epithelial lining is mediated by the fucosylated Lewis b (Leb) histo-blood group antigen. The Leb-binding adhesin, BabA, was purified by receptor activity–directed affinity tagging. The bacterial Leb-binding phenotype was associated with the presence of the cag pathogenicity island among clinical isolates ofH. pylori. A vaccine strategy based on the BabA adhesin might serve as a means to target the virulent type I strains ofH. pylori.

Helicobacter pylori, a human-specific gastric pathogen, was first isolated in 1982 (1) and has emerged as the causative agent of chronic active gastritis and peptic ulcer disease (2). Most infected individuals show no clinical symptoms, implicating additional factors, such as genetic predisposition and the genotype of the infecting strain, in disease pathogenesis. Chronic infection is associated with the development of gastric adenocarcinoma, one of the most common types of cancer in humans (3), and H. pylori was recently defined as a class 1 carcinogen (4).

The bacterium colonizes the human gastric mucosa by adhering to the mucous epithelial cells and the mucus layer lining the gastric epithelium (5). These adherence properties protect the bacteria from the extreme acidity of the gastric lumen and displacement from the stomach by forces such as those generated by peristalsis and gastric emptying. The fucosylated blood group antigens Lewis b (Leb) and H-1 (Fig. 1A) mediate adherence of H. pylori to human gastric epithelial cells in situ (6).

Figure 1

Biochemical characterization of the blood group antigen–binding (BAB) activity of H. pylori. (A) The fucosylated blood group antigens. The H antigen (defining group O in the ABO blood group system) presents the Fucα1.2 residue (no. 1) in the core chain, the Leaantigen instead presents the Fucα1.4 residue (no. 2), and the difucosylated Leb antigen presents both fucose residues. The equivalent H-2, LeX, and LeYantigens differ respectively by having a central β1.4 linkage, compared with the β1.3 linkage in the previous series (30). The fucosylated blood group antigens are typically found on red blood cells where they define the ABO blood group system, but they are also expressed on the epithelial cell surfaces as histo-blood group antigens (30). Gal, galactose; GlcNAc,N-acetylglucosamine; Glc, glucose. (B) Bacterial binding to soluble blood group antigens. Five H. pylori strains (7) were incubated with125I-labeled blood group antigen glycoconjugates (8). Solid bars, Leb; open bars, H-1; striped (third) bars, Lea; hatched (fourth) bars, H-2 plus LeX plus LeY. (C) Receptor displacement assay. Strain CCUG17875 was incubated for 1 hour with 10 ng of 125I-labeled Leb conjugate, and the resulting complex was then challenged with an excess of unlabeled Leb or Lea conjugate. The remaining radioactivity in the bacterial pellet was measured (8). The Leb conjugate, but not the Lea conjugate, displaced the 125I-labeled Leb antigen from BabA. Concentrations of unlabeled conjugate ranged from 50 ng to 8 μg. (D) Scatchard analysis of the H. pylori–Leb antigen interaction. Binding of strain CCUG 17875 to the Leb antigen (8) was measured at Leb conjugate concentrations of 10 to 260 ng/ml, yielding a K a value of 8 × 109 M−1 (15).

We have now biochemically characterized and identified theH. pylori blood group antigen–binding adhesin, BabA. Various strains of H. pylori were analyzed for binding to125I-labeled fucosylated blood group antigens (Fig. 1B) (7, 8). Three of the five strains examined bound Leb and H-1. The receptor specificities of these strains for the soluble blood group antigens correlate with their adherence properties in situ (6). The prevalence of blood group antigen–binding (BAB) activity was also assessed among 95 recent clinical isolates of H. pylori, and 66% (63 isolates) bound the Leb antigen (7). None of the reference strains or the 95 recent isolates bound to the related Lea, H-2, LeX, or LeY antigens (Fig. 1, A and B). These results support previous observations of the receptor specificity of H. pylori for the Leb and H-1 blood group antigens (6) and, in addition, demonstrate the high prevalence of BAB activity among clinical isolates.

Isolates of H. pylori are thought to differ in virulence and those from individuals with peptic ulcers most often are type I strains that express the vacuolating cytotoxin A (VacA) and the cytotoxin-associated gene A (CagA) protein (9). By definition, type II strains express neither marker. Twenty-one strains of previously defined type (10) and 73 of the 95 recent isolates were analyzed for cag genotype (11) and binding to the Leb antigen. The presence of cagAwas associated with bacterial binding to the Leb antigen; 73% (54/74) of CagA+ strains, compared with 5% (1/20) of the CagA strains, were positive for binding. The cagA gene is located in the 40-kbcag pathogenicity island (PAI), which contains genes that encode proteins with similarities to components of secretion systems (12). However, a deletion of the entire PAI that we engineered into a type I Leb antigen–binding strain resulted in no reduction in Leb antigen–binding activity (13). Thus, the epidemiological association betweenCagA+ status and Lebantigen–binding activity is not mechanistic.

We next determined the affinity constant ( Ka) for the BabA-Leb interaction by performing receptor displacement analyses (Fig. 1C). These results showed that the receptor-adhesin complex was formed under conditions of equilibrium. Most of the cells (>90%) of the bacterial population exhibited BAB activity, as determined with the use of confocal microscopy and fluorescent Leb antigen (14). The Ka value for formation of the Lebantigen–BabA complex was ∼1 × 1010M−1 (Fig. 1D) (15). The number of Leb glycoconjugate molecules bound to BabA was calculated as ∼500 per bacterial cell, similar to the number of fimbriae organelles on the surface of Escherichia coli(16).

The localization of BabA on the bacterial cell surface was investigated by immunogold electron microscopy. The BabA adhesin was detected on the bacterial cell outer membrane by probing with the Lebantigen, but not with the Lea antigen (Fig. 2, A and B) (14). No gold particles were located on the flagellar sheath, suggesting that, despite their continuity, the membranes of the cell surface and sheath differ in protein composition.

Figure 2

Localization and characterization of the BabA adhesin. (A and B) Electron microscopy of cells of H. pylori strain CCUG17875 exposed to biotinylated Leb or Lea glycoconjugates, respectively. After washing, bacteria were incubated with 10-nm gold-labeled antibodies to biotin (ICN, Costa Mesa, California), counterstained, and air-dried onto formvar-coated copper grids (14). Bar, 1 μm. (C) Characterization of the molecular mass of the BabA adhesin by receptor overlay analysis (17). SDS-solubilized protein extracts of strain CCUG17875 were separated by SDS-PAGE and transferred to a PVDF membrane, which was then incubated with biotinylated Lebglycoconjugate (lane 1) or biotinylated albumin (lane 2), followed by peroxidase-streptavidin. The positions of molecular size markers (in kilodaltons) are indicated on the left. (D) Receptor overlay analysis of BabA adhesins from various strains. Lanes 1 to 4: A5 (Sweden), P466 (South America), CCUG17875 (Australia), and MO19 (United States), respectively. The lack of a Lebantigen–binding band with MO19 is consistent with this strain's lack of BAB activity (Fig. 1B). (E) Receptor activity–directed affinity tagging (ReTagging) (20) of BabA from various strains. Lanes are as in (D). Results are consistent with those in (D).

The molecular mass of the BabA protein was characterized by receptor overlay analysis. BAB activity corresponded to a single 75-kD band (Fig. 2C) (17); a 40-kD band also detected is probably endogenous peroxidase, possibly the 39-kD HP1461 peroxidase (18), because it stained without the Lebconjugate overlay. The panel of strains from Sweden, Australia, and South America showed a conserved molecular mass of BabA (Fig. 2D).

Because the BabA protein is not abundant (Fig. 1D), we developed a combined ligand identification and purification technique, termed receptor activity–directed affinity tagging (ReTagging). Cross-linking agents with radiolabeled donative tags have previously been used for characterization of receptor ligands (19). However, for ReTagging, the Leb glycoconjugate was equipped with an affinity tag–donating cross-linker structure. The modified Leb glycoconjugate directed the targeted transfer of the affinity tag (biotin) to the BabA protein by virtue of its receptor activity (Fig. 3, A and B). After cross-linking, the covalently attached biotin tag was used to identify the adhesin with streptavidin (Fig. 2E) (20). One biotin-tagged protein of 75 kD was detected in several strains, consistent with the results of the overlay analysis (Fig. 2, C and D). More generally, ReTagging should prove useful for diverse studies of interactions whether in infectious disease, inflammation processes, or cell differentiation and development.

Figure 3

ReTagging and purification of the BabA adhesin. (A) A multifunctional cross-linking agent was conjugated to the Leb glycoconjugate. The Sulfo-SBED cross-linker contains an NHS group for conjugation to the protein core of the glycoconjugate, a central disulfide bond, a photoreactive aryl azide group, and a biotin side group (star) (20). (B) The cross-linker–labeled Lebglycoconjugate is incubated with H. pylori so that binding to the bacteria brings the cross-linker molecule close to the adhesin protein. When the cells are subjected to ultraviolet (UV) irradiation, the photoreactive group forms a covalent bond to structures in the immediate vicinity (that is, the adhesin protein). Exposure to reducing conditions (DTT) results in cleavage of the disulfide bond in the cross-linker. The Lebglycoconjugate is subsequently released and washed away. (C) Consecutive steps in the purification of BabA protein. Lane 1, SDS-PAGE and Coomassie blue staining of untreated bacterial protein extract (control). Lanes 2 and 3, protein extract after the ultraviolet-activated cross-linking reaction; lane 2 shows the Coomassie blue–stained gel and lane 3 shows blot detection of biotin-tagged proteins with peroxidase-streptavidin [the 75-kD BabA protein (asterisk) and some remaining Leb glycoconjugate of >100 kD are apparent]. Lane 4, blot analysis of the protein extract after treatment with streptavidin-coated magnetic beads; no detectable biotin-tagged adhesin protein remained in the protein extract. Lanes 5 and 6, Coomassie blue–stained gel and blot analysis, respectively, of BabA protein eluted from the streptavidin-coated magnetic beads. Lanes 7 and 8, Coomassie blue–stained gel and blot analysis, respectively, of the protein preparation after final fractionation by preparative SDS-PAGE; BabA is now the dominant protein band (21). This band was excised for NH2-terminal and COOH-terminal sequencing (21, 23).

The high specificity of the ReTagging technique provided a means for affinity-purification of the adhesin protein. After cross-linking, bacteria were solubilized in SDS sample buffer, streptavidin-coated magnetic beads were added to the solubilized proteins, and biotin-tagged BabA protein was extracted (Fig. 3C) (21). The sequences of the 20 NH2-terminal amino acids of the BabA adhesins from an Australian and a Swedish strain were found to be identical, and were used to construct degenerate polymerase chain reaction (PCR) primers for cloning purposes (22). Two sets of clones were identified that encode two proteins with almost identical NH2-terminal domains and completely identical COOH-terminal domains (∼300 amino acids), but with divergent central regions (Fig. 4A). For the identification of the functional babA gene, the BabA adhesin was subjected to large-scale purification by ReTagging, which provided sufficient protein for determination of the sequence of the 41 NH2-terminal residues. The DNA sequence of one set of clones encoded this 41–amino acid sequence, and the corresponding gene was designated babA. The gene corresponding to the second set of clones was designated babB.

Figure 4

(A) The translated babA andbabB sequences showing regions of amino acid sequence similarity and heterogeneity in black and white, respectively. The BabA2 signal peptide starts at position −20. Position −18 of BabB indicates the predicted translational start position. Positions 721 for BabA (78 kD) and 689 for BabB (75 kD) indicate the ends of the open reading frames. (B) Nucleotide sequence of the upstream and signal peptide regions of the functional adhesin genebabA2. The putative Shine-Dalgarno sequence is underlined. The signal peptide sequences predicted by babA1 andbabA2 are also shown. The 10-bp insert with a repeat motif (RM) is absent from babA1, which is otherwise identical tobabA2, resulting in elimination of the start codon. A UAA termination codon is present at codon position −24 of the signal peptide region in babA1. The accession numbers forbabA1 and babA2 are AF001388 and AF033654, respectively, with both predicted proteins being 91% identical to HP1243 (18). The accession number for babB isAF001389, with the predicted protein being 95% identical to HP896 (18). Abbreviations for the amino acid residues are: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

The genes corresponding to both sets of clones appear to encode proteins with an 18- to 20-residue signal peptide sequence that would be cleaved during secretion to produce the determined NH2-terminal EDD sequences. The calculated molecular size of the mature babA product is 78 kD (23). However, the cloned babA gene lacked an initiation codon at the start of the open reading frame (Fig. 4B). To localize additionalbabA gene alleles, we screened an ordered cosmid library, and two babA genes and one babB gene were mapped (24). Gene inactivation experiments identified the functional babA gene in strain CCUG17875, which expresses the BabA adhesin; inactivation of the second babA gene, now denoted babA2, resulted in a loss of Lebantigen–binding activity, whereas inactivation of the originalbabA gene (babA1) did not affect Lebantigen–binding activity (24). The functionalbabA2 gene was subsequently amplified by PCR and sequenced (without cloning). The coding region was found to be identical to the previously cloned and sequenced babA1 with the exception of an insert of 10 base pairs (bp) with a repeat motif in the signal peptide sequence, which resulted in the creation of a translational initiation codon (Fig. 4). Sequence analysis of the babA2gene obtained by PCR amplification and cloning in a plasmid in E. coli revealed frequent deletion of the repeat motif and convergence to the silent babA1 gene, suggesting the presence of hot spots for phenotypic (phase) variation within thebab gene family (25).

The babA and babB coding sequences are highly similar to open reading frames in the sequenced genome of strain 26695 (open reading frames HP1243 and HP896, respectively) (18). The genomic location of babA1 corresponds to that of open reading frame HP1243, as revealed by an almost perfect match between the upstream open reading frame HP1244 (ribosomal protein S18) and the sequence upstream of babA1 from strain CCUG17875. No equivalent genomic location for the babA2 upstream sequence was detected. Absence of the babA2 allele might explain the lack of Leb antigen–binding activity in strain 26695 (Fig.1B). Although the cag PAI locus has a lower G+C content than the average of 39% for H. pylori (18),babA has a G+C content of 46%. Differences in G+C content may indicate the acquisition of genes and DNA loci from different sources.

The bab genes belong to a family of ∼30 genes whose products show extensive amino acid sequence homology in the NH2-terminal and COOH-terminal domains (18) (Fig. 4A), suggesting possibilities for recombination and consequent changes in the positions of individual genes. Evidence supporting this possibility is provided by the map positions of genes in several strains. In strain 26695, babB is located 5.3 kb from thevacA gene (18). Pulsed-field gel mapping also placed babB near vacA in strain NCTC11637 (26). In contrast, in strain NCTC11638 (24),babA2 is located close to vacA. Recombination between duplicate segments would allow adhesin synthesis to be readily switched on or off. Such a mechanism might be important in determining host specificity during colonization and in bacterial persistence during chronic infection (27).

We propose that BabA-mediated adherence of H. pylorito gastric epithelium plays a critical role in efficient delivery of bacterial virulence factors that damage host tissue either directly or through inflammatory or autoimmune reactions, eventually leading to ulcer disease. Immunization experiments with adhesins of uropathogenicE. coli have demonstrated the potential for the generation of antibodies that inhibit adhesion (28). A vaccine strategy based on the BabA adhesin might possibly target the virulent type I strains of H. pylori, sparing the less virulent strains that may be constituents of the commensal flora (29).

  • * These authors contributed equally to this work.

  • Present address: Max-von Pettenkofer Institut für Hygiene und Medizinische Mikrobiologie, Pettenkoferstrasse 9A, D-81677 Munich, Germany.

  • Present address: Swedish Institute for Infectious Disease Control, S-105 21, Stockholm, Sweden.

  • § To whom correspondence should be addressed. E-mail: thomas.boren{at}


View Abstract

Navigate This Article