Helicobacter pylori and Complex Gangliosides

Abstract

Recognition of sialic acid-containing glycoconjugates by the human gastric pathogen Helicobacter pylori has been repeatedly demonstrated. To investigate the structural requirements for H. pylori binding to complex gangliosides, a large number of gangliosides were isolated and characterized by mass spectrometry and proton nuclear magnetic resonance. Ganglioside binding of sialic acid-recognizing H. pylori strains (strains J99 and CCUG 17874) and knockout mutant strains with the sialic acid binding adhesin SabA or the NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ-binding neutrophil-activating protein HPNAP deleted was investigated using the thin-layer chromatogram binding assay. The wild-type bacteria bound to N-acetyllactosamine-based gangliosides with terminal α3-linked NeuAc, while gangliosides with terminal NeuGcα3, NeuAcα6, or NeuAcα8NeuAcα3 were not recognized. The factors affecting binding affinity were identified as (i) the length of the N-acetyllactosamine carbohydrate chain, (ii) the branches of the carbohydrate chain, and (iii) fucose substitution of the N-acetyllactosamine core chain. While the J99/NAP⁻ mutant strain displayed a ganglioside binding pattern identical to that of the parent J99 wild-type strain, no ganglioside binding was obtained with the J99/SabA⁻ mutant strain, demonstrating that the SabA adhesin is the sole factor responsible for the binding of H. pylori bacterial cells to gangliosides.

Helicobacter pylori is a human- and primate-specific pathogen found in the gastric mucus layer or attached to the gastric epithelium. H. pylori infection, which affects about half the world population, results in chronic active gastritis and is a risk factor for the development of peptic ulcer disease, gastric adenocarcinoma, and gastric lymphoma (23).

In order to initiate and maintain infection, microbes must first bind to receptors present on their target tissue. Therefore, interest has been directed to the elucidation of receptors for different microorganisms, many of which have been demonstrated to be glycosphingolipids (14, 25). Initial studies of potential receptors for H. pylori suggested that acid glycosphingolipids, such as the GM3 ganglioside and sulfatide (7, 27), can function as receptors for the bacterium. (The glycosphingolipid nomenclature follows the recommendations of the International Union of Pure and Applied Chemistry-International Union of Biochemistry Commission on Biochemical Nomenclature [Commission on Biochemical Nomenclature for Lipids] [5a, 5b, 5c]. It is assumed that Gal, Glc, GlcNAc, GalNAc, and NeuAc are of the d configuration; that Fuc is of the l configuration; and that all sugars are present in the pyranose form. In the shorthand nomenclature for fatty acids and bases, the number before the colon refers to the carbon chain length and the number after the colon gives the total number of double bonds in the molecule. Fatty acids with a 2-hydroxy group are denoted by the prefix h before the abbreviation, e.g., h16:0. For long-chain bases, d denotes dihydroxy and t denotes trihydroxy. Thus, d18:1 designates sphingosine [1,3-dihydroxy-2-aminooctadecene] and t18:0 designates phytosphingosine [1,3,4-trihydroxy-2-aminooctadecane].) Other receptors subsequently reported include gangliotetraosylceramide (17), the Le^b antigen (3), NeuAcα3-neolactotetraosylceramide (20), lactosylceramide (1), and lactotetraosylceramide (38). In a separate series of studies, the binding of H. pylori to sialic acid-containing glycoconjugates from a variety of origins was demonstrated (21, 22). Two recent studies have demonstrated that the gangliosidessialyl-neolactohexosylceramide NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (26) and sialyl-dimeric-Lewis x NeuAcα3Galβ4(Fucα3)GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer (19) are high-affinity H. pylori receptors. The H. pylori adhesin (SabA) that mediates binding to sialyl-Le^x, and the gene encoding it (sabA), were also identified in the second study.

The primary target tissue of H. pylori, i.e., the human gastric epithelium, has a very low content of sialic acid (18). However, H. pylori also interacts with sialylated glycoconjugates of human neutrophil granulocytes (22), and several of the identified H. pylori-binding gangliosides are also present in human neutrophils (32, 33). The aim of the present study was to further investigate the structural requirements for binding of H. pylori to complex gangliosides. The human neutrophil gangliosides are a very complex mixture, and isolation of pure ganglioside species is difficult to achieve (32, 33). Therefore, gangliosides were instead isolated from a range of different sources in which the occurrence of complex gangliosides has been described (e.g., human erythrocytes, bovine erythrocytes, rabbit thymus, human meconium, and human cancer tissues [34]) and were characterized by mass spectrometry and proton nuclear magnetic resonance (NMR). Binding of representative sialic acid binding H. pylori strains, and mutant H. pylori strains with knockout of the sialic acid binding adhesin SabA and the neutrophil-activating protein HPNAP, to the gangliosides was assessed by means of the chromatogram binding assay.

MATERIALS AND METHODS

Ganglioside preparations.

For the isolation of gangliosides, a number of tissues previously reported to contain complex gangliosides, e.g., human erythrocytes, bovine erythrocytes, rabbit thymus, human meconium, and human cancers (34), were collected. Isolation of total acid glycosphingolipid fractions was done as described previously (12). Briefly, the tissues were lyophilized, followed by extraction in two steps with chloroform-methanol (2:1 and 9:1 [vol/vol]) in a Soxhlet apparatus. The material obtained was pooled and subjected to mild alkaline hydrolysis and dialysis, followed by separation on a silicic acid column. Acid and nonacid glycosphingolipids were separated on a DEAE column.

The acid glycosphingolipid fractions were separated by DEAE-Sepharose chromatography, followed by repeated silicic acid chromatography, and final separation was achieved using high-performance liquid chromatography on a Kromasil 5 silica column 250 mm long, with an inner diameter of 10 mm and a particle size of 5 μm (Phenomenex, Torrance, Calif.), using linear gradients of chloroform-methanol-water (60:35:8 to 40:40:12 or 65:25:4 to 40:40:12 [vol/vol]) over 180 min, with a flow rate of 2 ml/min. The 2-ml fractions collected were analyzed by thin-layer chromatography and anisaldehyde staining (see below), and the H. pylori-binding activity was assessed using the chromatogram binding assay (see below). The fractions were pooled according to mobility on thin-layer chromatograms and their H. pylori-binding activities.

Reference glycosphingolipids.

Reference glycosphingolipids were isolated and characterized at the Institute of Medical Biochemistry, Göteborg University, Göteborg, Sweden. Structural characterization was performed using proton NMR (15), mass spectrometry (28), and degradation studies (31, 40). Sialyl-Le^x hexaglycosylceramide was purchased from ARC, Edmonton, Canada.

Bacterial strains, growth conditions, and labeling.

H. pylori strain CCUG 17874 was obtained from the Culture Collection of the University of Göteborg (CCUG). Strain J99 and the construction of the sabA(JHP662) mutant (designated the J99/SabA⁻ mutant) were described previously (19). The construction of the J99/NAP⁻ mutant will be described elsewhere (M. Hurtig, unpublished data).

Bacteria were grown on brucella medium (Difco Laboratories, Irvine, Calif.) containing 10% fetal calf serum (Harlan Sera-Lab, Loughborough, United Kingdom) inactivated at 56°C and BBL IsoVitale X Enrichment (Becton Dickinson Microbiology Systems, Franklin Lakes, N.J.). The mutant strain J99/SabA⁻ was cultured on the same medium supplemented with chloramphenicol (20 μg/ml). Bacteria were radiolabeled by the addition of 50 μCi of [³⁵S]methionine (Amersham Pharmacia Biotech, Little Chalfont, United Kingdom) diluted in 0.5 ml of phosphate-buffered saline (PBS), pH 7.3, to the culture plates. After incubation for 12 to 72 h at 37°C under microaerophilic conditions, the bacteria were harvested and centrifuged three times in PBS.

Alternatively, colonies were inoculated (10⁵ CFU/ml) in Ham's F12 medium (Invitrogen Corp., Carlsbad, Calif.) supplemented with 10% heat-inactivated fetal calf serum and 50 μCi of [³⁵S]methionine. The culture bottles were incubated with shaking under microaerophilic conditions at 37°C for 24 h. Bacterial cells were harvested by centrifugation and washed three times with PBS.

In both cases, the bacteria were finally resuspended in PBS containing 2% (wt/vol) bovine serum albumin (PBS-BSA) to ∼10⁸ CFU/ml. Both labeling procedures resulted in suspensions with specific activities of ∼1 cpm per 100 H. pylori organisms.

Thin-layer chromatography.

Total acid glycosphingolipid fractions (40 μg) or pure gangliosides (0.0002 to 4 μg) were separated on aluminum-backed silica gel 60 high-performance thin-layer chromatography plates (Merck, Darmstadt, Germany) using chloroform-methanol-water (60:35:8 [vol/vol/vol]) or chloroform- methanol- 0.25% KCl in water (50:40:10 [vol/vol/vol]) as a solvent system. Chemical detection of glycosphingolipids on thin-layer chromatograms was carried out using anisaldehyde (39) or resorcinol (35) reagent.

Chromatogram binding assay.

The chromatogram binding assay was essentially carried out as described previously (13). Dried thin-layer chromatograms with separated glycosphingolipids were treated in 0.5% polyisobutylmethacrylate (wt/vol) (Aldrich Chemical Company Inc., Milwaukee, Wis.) in diethylether- n-hexane (1:5 [vol/vol]) for 1 min and then air dried. To reduce nonspecific binding, the plates were incubated in PBS-BSA containing 0.1% (wt/vol) NaN₃ and 0.1% (vol/vol) Tween 20 at room temperature for 2 h. The plates were then incubated for 2 h at room temperature with ³⁵S-labeled H. pylori diluted in PBS-BSA. Following the final wash and drying, autoradiography was carried out overnight using Biomax film (Eastman Kodak Company, Rochester, N.Y.).

Derivatization of gangliosides.

Gangliosides were permethylated using sodium hydroxide and methyl iodide in dimethyl sulfoxide as described previously (16). Reduction of permethylated samples was carried out using LiAlH₄ in diethylether (9).

FAB and EI mass spectrometry.

Negative-ion fast atom bombardment (FAB) and electron ionization (EI) mass spectra were obtained on an SX 102A mass spectrometer (JEOL, Toyko, Japan). Negative-ion FAB mass spectra of native gangliosides were obtained using Xe atom bombardment (6 eV), an acceleration voltage of −8 kV, and triethanolamine as a matrix. EI spectra of derivatized glycosphingolipids were obtained with an ionization voltage of 70 eV, an ionization current of 300 μA, and an acceleration voltage of 8 kV. The temperature was raised from 150 to 410°C at a rate of 10°C/min. For the collection of both FAB and EI spectra, a resolution of 1,000 was used.

Proton NMR spectroscopy.

1H NMR spectra were acquired on Varian 500- and 600-MHz spectrometers at 30°C. The samples were dissolved in dimethyl sulfoxide-D₂O (98:2 [vol/vol]) after deuterium exchange.

RESULTS

Ganglioside preparations.

To be able to dissect the ganglioside binding preferences of H. pylori, a ganglioside library was compiled (summarized in Table 1). Each ganglioside was characterized by mass spectrometry and proton NMR. The procedure is illustrated by the following description of the isolation and characterization of one H. pylori-binding ganglioside of human erythrocytes (Table 1, no. 19).

TABLE 1.

Ganglioside library and results of H. pylori binding

No.	Trivial name	Structure	Bindinga		Source
			CCUG 17874, J99, and J99/NAP−	J99/SabA−
1	NeuAc-GM3	NeuAcα3Galβ4Glcβ1Cer	−	−	Human meconium
2	NeuAcα3SPG	NeuAcα3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	+	−	Human erythrocytes
3	NeuAcα6SPG	NeuAcα6Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Human meconium
4	NeuGcα3SPG	NeuGcα3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Rabbit thymus
5	NeuAc-DPG	NeuAcα8NeuAcα3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Human kidney
6	NeuAcα3-Lea	NeuAcα3Galβ3(Fucα4)GlcNAcβ3Galβ4Glcβ1Cer	−/+b	−	Human gallbladder cancer
7	NeuAcα3-Lex	NeuAcα3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer	+	−	Commercial
8	NeuAcα3-nLc6	NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	+++	−	Human hepatoma
9	NeuGcα3-nLc6	NeuGcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Rabbit thymus
10	NeuAcα3-nLc8	NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	+++	−	Human erythrocytes
11	NeuGcα3-nLc8	NeuGcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Rabbit thymus
12	VIM-2	NeuAcα3Galβ4GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer	+++	−	Human colon cancer
13	NeuAcα3-dimer-Lex	NeuAcα3Galβ4(Fucα3)GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer	+++	−	Human gallbladder cancer
14		Galβ4GlcNAcβ6(NeuAcα6Galβ4GlcNAcβ3)Galβ4Glcβ1Cer	−	−	Bovine buttermilk
15		Galβ4GlcNAcβ6(NeuAcα6Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Human meconium
16	NeuAc-G-10	NeuAcα3Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer	+++	−	Human erythrocytes
17	NeuGc-G-10	NeuGcα3Galβ4GlcNAcβ6(NeuGcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Bovine erythrocytes
18		Galα3Galβ4GlcNAcβ6(NeuGcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer	−	−	Bovine erythrocytes
19	G9-B	Galα3(Fucα2)Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer	+++	−	Human erythrocytes

^aBinding is defined as follows: +++, binding when <0.5 μg of the glycosphingolipid was applied on the thin-layer chromatogram; +, occasional binding at 0.5 μg; −, no binding even at 4 μg.

^bBinding of the J99 strain to NeuAcα3-Le^a was occasionally observed (detection limit, 0.5 μg), while no binding of the CCUG 17874 strain to this compound was obtained.

Total acid glycosphingolipids were isolated from 500 liters of pooled outdated blood group B erythrocytes by standard procedures (12), yielding 5.5 g. A subfraction of 390 mg was separated on a 700-ml DEAE-Sepharose column eluted with a linear gradient using 2,100 ml of ammonium acetate in methanol (0.05 to 0.45 M). Each 10-ml fraction collected was analyzed by thin-layer chromatography using the resorcinol reagent. The fractions were pooled according to the mobilities of the major compounds. Pooling of fractions 62 to 67 yielded 15.4 mg, and the fraction obtained had a major compound migrating in the sialyl-neolactotetraosylceramide region. However, when it was tested for H. pylori-binding activity using the chromatogram binding assay, a slow-migrating binding-active compound was detected. The 15.4 mg of acid glycosphingolipids was further separated by high-performance liquid chromatography using a linear gradient of chloroform-methanol-water (60:35:8 to 40:40:12 [vol/vol/vol]). The H. pylori-binding compound eluted in fractions 39 to 56, which yielded 0.9 mg after pooling.

Negative-ion FAB mass spectrometry of H. pylori-binding ganglioside from human erythrocytes.

The negative-ion FAB mass spectrum of the H. pylori-binding ganglioside (not shown) had molecular ions at m/z 2639, 2667, and 2683, indicating a ganglioside with one NeuAc, one Fuc, three HexNAc, and six Hex and d18:1-22:0, d18:1-24:0 and d18:1-h24:0 ceramides. A series of fragment ions, obtained by sequential loss of terminal carbohydrate units from the ion at m/z 2667, were observed at m/z 2505, 2360, 2197, and 1994, demonstrating a Hex-Fuc-Hex-HexNAc sequence. Further sequence ions were observed at m/z 1703, 1541, 1338, 1176, 973, 811, and 648, suggesting a NeuAc-Hex-HexNAc-Hex-HexNAc-Hex-Hex sequence with a d18:1-24:0 ceramide.

Since an internal NeuAc is unlikely, the result from negative-ion FAB mass spectrometry thus suggested a branched undecaosylceramide with a Hex-Fuc-Hex-HexNAc-(NeuAc-Hex-HexNAc)Hex-HexNAc-Hex-Hex sequence and a d18:1-24:0 ceramide.

EI mass spectrometry of permethylated H. pylori-binding ganglioside from human erythrocytes.

EI mass spectrometry of the permethylated H. pylori-binding ganglioside (Fig. 1) confirmed the presence of two branches, since the ions at m/z 825 and 793 (825 − 32) are characteristic of a terminal NeuAc-Hex-HexNAc sequence, while the ions at m/z 842 and 810 (842 − 32) demonstrated a terminal (Hex-Fuc-HexHexNAc) sequence. In addition, the ions at m/z 376 and 344 (376 − 32) indicated a terminal NeuAc, while the ions at m/z 219 and 187 (219 − 32) demonstrated a terminal Hex. A terminal Fuc was indicated by the ions at m/z 189 and m/z 157 (189 − 32). A type 2 carbohydrate chain (Hexβ4HexNAc) was indicated by the ion at m/z 182 (10, 11). The ion at m/z 660 was derived from a d18:0-24:0 ceramide.

FIG. 1.

EI mass spectrum of the permethylated H. pylori-binding ganglioside from human erythrocytes. Above the spectrum is a simplified formula for interpretation, representing the species with sphingosine and nonhydroxy 24:0 fatty acid. The analysis was done as described in Materials and Methods. The spectrum was recorded at 380°C. The peak at m/z 354 is due to a contaminant.

EI mass spectrometry of permethylated and LiAlH₄-reduced H. pylori-binding ganglioside from human erythrocytes.

EI mass spectrometry of permethylated and reduced glycosphingolipids gives dominant immonium ions, representing the complete carbohydrate chain and the fatty acid, along with rearrangement ions obtained by sequential loss of terminal carbohydrate units from the immonium ions (9). In the spectrum of the permethylated and reduced ganglioside from human erythrocytes (Fig. 2), the ion observed at m/z 2501 was obtained by loss of a terminal NeuAc from the immonium ion of a Hex-Fuc-Hex-HexNAc-(NeuAc-Hex-HexNAc)HexHexNAc-Hex-Hex sequence with nonhydroxy 24:0 fatty acid (2835 − 334; see the interpretation formula in Fig. 2). Further rearrangement ions were found at m/z 2297 (2835 − 538) and at m/z 2066 (2835 − 769), confirming a terminal NeuAc-Hex-HexNAc sequence.

FIG. 2.

EI mass spectrum of the permethylated and reduced H. pylori-binding ganglioside from human erythrocytes. Above the spectrum is a simplified formula for interpretation, representing the species with sphingosine and nonhydroxy 24:0 fatty acid. The analysis was done as described in Materials and Methods. The spectrum was recorded at 340°C.

Thereafter, rearrangement ions obtained by the loss of a terminal Hex-(Fuc)-Hex were found at m/z 1700 (2297 − 597) and at m/z 1466 (2066 − 597), followed by a rearrangement ion obtained by the loss of a terminal Hex-(Fuc)-Hex-HexNAc at m/z 1238 (2066 − 828). Further rearrangement ions were found at m/z 1049, 818, and 614.

Again, two terminal carbohydrate sequence ions were obtained. The ion at m/z 769 confirmed the terminal NeuAc-Hex-HexNAc sequence, while the ion at m/z 828 confirmed a terminal Hex-(Fuc)-Hex-HexNAc sequence. Sphingosine with a nonhydroxy fatty acid was indicated by the ion at m/z 632.

Thus, by mass spectrometry of the ganglioside from human erythrocytes, a Hex-(Fuc)-Hex-HexNAc-(NeuAc-HexHexNAc)Hex-HexNAc-Hex-Hex sequence with a d18:1-24:0 ceramide was established.

Proton NMR of H. pylori-binding ganglioside from human erythrocytes.

The anomeric region of the 600-MHz proton NMR spectrum of the human erythrocyte ganglioside (Fig. 3A) is in agreement with a branched structure having the sugar composition indicated by mass spectrometry. In this anomeric region, two α signals are evident at 5.15 and 4.97 ppm, which by comparison with earlier data on blood group B-active glycosphingolipids from human erythrocytes (4) can be ascribed to Fucα2 and Galα3 of a blood group B determinant. Irrespective of whether this determinant is located on the three- or six-linked branch of the structure, the anomeric signal of a Fucα2- and Galα3-substituted Galβ4 is expected at 4.39 ppm, as is also observed. Furthermore, the presence of a GlcNAcβ anomeric signal at 4.36 ppm, in addition to the lack of anomeric signals in the range 4.4 to 4.5 ppm, shows that the B determinant is situated on the six-linked branch (4). Therefore, the sialic acid must be located on the three-linked branch, which is confirmed by the presence of two overlapping GlcNAcβ signals, centered around 4.65 ppm, belonging to the GlcNAc residues of a neolactotetra core and three branch, respectively. The NeuAc residue is α3 linked, as seen by the presence of the H3eq resonance at 2.75 ppm (Fig. 3B), and is attached to a Galβ4 residue, as evidenced by the doublet at 4.19 ppm (8). The remaining anomeric signals stemming from the core are the branching Galβ4 residue at 4.30 ppm, the second Galβ4 residue at 4.26 ppm, and the Glcβ1 residue at 4.16 ppm, in accordance with earlier data (4, 8). Additional resonances consistent with the assignments made above are found at 4.11 and 1.06 ppm (Fucα2 H5 and H6, respectively), as well as four methyl resonances from the N-acetamido moieties of the NeuAc residue (1.88 ppm) and the GlcNAc residues (1.84, 1.82, and 1.81 ppm). Overall, the spectral features are very similar to the corresponding spectrum of the blood group A-containing structure previously published (9).

FIG. 3.

Proton NMR spectrum from 4 to 5.2 ppm (A) and from 2.58 to 2.98 ppm (B) at 600 MHz of the H. pylori-binding ganglioside from human erythrocytes (30°C). The sample was dissolved in dimethyl sulfoxide-D₂O (98:2 [vol/vol]) after deuterium exchange. The broad peak(s) centered around 4.8 ppm (indicated by an asterisk) represents a contaminant of unknown origin.

Thus, by mass spectrometry and proton NMR of the H. pylori-binding ganglioside, a Galα3(Fucα2)Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer, i.e., a branched undecaglycosylceramide carrying a blood group B determinant on the β6-linked branch and a terminal sialic acid on the β3-linked branch, was identified. This ganglioside was previously characterized by Kannagi et al. and designated ganglioside G9-B (8).

Binding of H. pylori to ganglioside library.

The H. pylori strains CCUG 17874 and J99, used in the chromatogram binding experiments, are both sialic acid binding (19). In addition, two mutant strains were used, strain J99/SabA⁻, with knockout of the sialic acid binding adhesin SabA (19), and strain J99/NAP⁻, with knockout of the NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ-binding neutrophil-activating protein of H. pylori, HPNAP (36).

(i) Binding and nonbinding gangliosides.

The results from binding of the H. pylori strains to the isolated gangliosides are shown in Fig. 4 to 8 and summarized in Table 1. Gangliosides were classified as nonbinding when no binding was obtained although 4 μg of the compound was on the thin-layer plates. As shown in Fig. 4, the sialic acid binding wild-type strains CCUG 17874 and J99 recognized the NeuAc-terminated gangliosides NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAc-neolactohexaocylceramide) (lane 1), NeuAcα3Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAc-G-10 ganglioside) (lane 3), and Galα3(Fucα2)Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (G9-B ganglioside) (lane 6) but not the corresponding NeuGc-terminated isostructures (lanes 2, 4, and 5). Other gangliosides recognized were NeuAcα3(Galβ4GlcNAcβ3)₃Galβ4Glcβ1Cer (NeuAc-neolactooctaaocylceramide) (Fig. 6, lanes 1 to 7), NeuAcα3Galβ4GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer (VIM-2 ganglioside) (Fig. 8, lanes 6 to 10), and NeuAcα3Galβ4(Fucα3)GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer (sialyl-dimeric-Le^x ganglioside) (Fig. 8, lanes 1 to 5).

FIG. 4.

Binding of H. pylori to NeuAc- and NeuGc-terminated gangliosides. Shown are chemical detection by anisaldehyde (A) and autoradiograms obtained by binding of the ³⁵S-labeled H. pylori strains CCUG 17874 (B) and J99 (C). The gangliosides were separated on aluminum-backed silica gel plates, using chloroform- methanol- 0.25% KCl in water (50:40:10 [vol/vol/vol]) as a solvent system, and the binding assay was performed as described in Materials and Methods. Lanes 1, NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactohexaocylceramide) of human hepatoma (2 μg); lanes 2, NeuGcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuGcα3-neolactohexaocylceramide) of rabbit thymus (2 μg); lanes 3, NeuAcα3Galβ4GlcNAcβ6 (NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAc-G-10 ganglioside) of human erythrocytes (2 μg); lanes 4, NeuGcα3Galβ4GlcNAcβ6 (NeuGcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuGc-G-10 ganglioside) of bovine erythrocytes (2 μg); lanes 5, Galα3Galβ4GlcNAcβ6 (NeuGcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer of bovine erythrocytes (2 μg); lanes 6, Galα3(Fucα2)Galβ4GlcNAcβ6 (NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (G9-B ganglioside) of human erythrocytes (2 μg). Autoradiography was performed for 12 h.

FIG. 8.

Binding of H. pylori to serial dilutions of gangliosides. Shown is an autoradiogram obtained by binding H. pylori strain CCUG 17874 using the chromatogram binding assay. Lanes 1 to 5, serial dilutions (4 to 100 pmol) of NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactohexaocylceramide) (NeuAcα3-nLc6) and NeuAcα3Galβ4(Fucα3)GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer (NeuAc-dimeric-Le^x ganglioside) (NeuAcα3-dimer-Lex); lanes 6 to 10, serial dilution (4 to 100 pmol) of NeuAcα3Galβ4GlcNAcβ3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer (VIM-2 ganglioside). The binding assay was done as described in Materials and Methods. The results from one representative experiment out of three are shown.

FIG. 6.

Binding of H. pylori to serial dilutions of gangliosides. Shown is an autoradiogram obtained by binding H. pylori strain CCUG 17874 using the chromatogram binding assay. Lanes 1 to 7, serial dilutions (1 to 100 pmol) of NeuAcα3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactotetraosylceramide) (NeuAcα3-nLc4), NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactohexaocylceramide) (NeuAcα3-nLc6), and NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactooctaocylceramide) (NeuAcα3-nLc8); lane 8, NeuAcα3Galβ3(Fucα4)GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-Le^a hexaglycosylceramide) (NeuAcα3-Lea), 1 mmol. The binding assay was done as described in Materials and Methods. The results from one representative experiment out of three are shown.

Occasional binding to NeuAcα3-neolactotetraosylceramide (Table 1, no. 2, and Fig. 6, lanes 1 to 7) was also detected, while NeuAcα6-neolactotetraosylceramide (Table 1, no. 3) was nonbinding, in line with previous reports (6, 20). The NeuAcα6-carrying gangliosides Galβ4GlcNAcβ6(NeuAcα6Galβ4GlcNAcβ3)Galβ4Glcβ1Cer (Table 1, no. 14) and Galβ4GlcNAcβ6(NeuAcα6Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (Table 1, no. 15) were also nonbinding. No binding to NeuGcα3-neolactotetraosylceramide (Table 1, no. 4) or disialyl-neolactotetraosylceramide (Table 1, no. 5) was obtained. Occasional binding to the sialyl-Le^x hexaglycosylceramide (Table 1, no. 7) was observed. The sialyl-Le^a hexaglycosylceramide (Table 1, no. 6) was not recognized by the CCUG 17874 strain, while the J99 wild-type strain occasionally bound to this compound. In all other respects, the ganglioside binding pattern obtained with the J99 wild-type strain was identical to the pattern observed with the CCUG 17874 strain (Fig. 4 and 5; summarized in Table 1). However, the J99/SabA⁻ strain failed to bind to any of the gangliosides recognized by the parent strain (Fig. 5E; summarized in Table 1).

FIG. 5.

Comparison of binding of H. pylori strains CCUG 17874, J99, J99/NAP⁻, and J99/SabA⁻. (A) Chemical detection by anisaldehyde. (B to E) Autoradiograms obtained by binding of ³⁵S-labeled H. pylori strains CCUG 17874 (B), J99 (C), J99/NAP⁻ (D), and J99/SabA⁻ (E). The gangliosides were separated on aluminum-backed silica gel plates, using chloroform- methanol- 0.25% KCl in water (50:40:10 [vol/vol/vol]) as a solvent system, and the binding assays were performed as described in Materials and Methods. Lanes 1, gangliosides of human neutrophil granulocytes (20 μg); lanes 2, NeuGcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuGcα3-neolactohexaocylceramide) of rabbit thymus (2 μg); lanes 3, NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactohexaocylceramide) of human hepatoma (1 μg); lanes 4, NeuAcα3Galβ4GlcNAcβ6 (NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAc-G-10 ganglioside) of human erythrocytes (1 μg); lanes 5, Galα3(Fucα2)Galβ4GlcNAcβ6 (NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (G9-B ganglioside) of human erythrocytes (1 μg); lanes 6, Galα3(Fucα2)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (B6 type 2 hexaglycosylceramide) of human erythrocytes (4 μg); lanes 7, Galβ3GalNAcβ4Galβ4Glcβ1Cer (gangliotetraosylceramide) of mouse feces (4 μg). Autoradiography was performed for 12 to 24 h.

(ii) Comparison of relative binding affinities.

Binding of radiolabeled H. pylori to glycosphingolipids diluted in microtiter wells was initially attempted in order to appreciate the relative binding affinities for the various binding-active gangliosides. However, the results obtained were not reproducible. Therefore, binding assays using dilutions of gangliosides on thin-layer plates were utilized. In initial experiments, we found that binding to NeuAcα3-neolactohexaosylceramide and the NeuAc-dimeric-Le^x ganglioside was saturated at ∼100 pmol, and therefore, lower concentration ranges (1 to 100 pmol) were utilized. The results presented in Fig. 6 to 8 are representative of a large number of binding assays. A general observation is that this type of experiment allows comparison only between compounds applied on the same chromatogram. Although the level of binding varied somewhat between different batches of radiolabeled bacteria, the same relationships between the binding-active gangliosides were repeatedly obtained.

(a) Effect of carbohydrate chain length.

Binding of H. pylori strain CCUG 17874 to dilution series of NeuAcα3-neolactotetraosylceramide (Table 1, no. 2), NeuAcα3-neolactohexaosylceramide (Table 1, no. 8), and NeuAcα3-neolactooctaosylceramide (Table 1, no. 10) demonstrated a clear preference for NeuAcα3-neolactooctaosylceramide (Fig. 6).

(b) Effect of branching.

To evaluate the effect of branching of the carbohydrate chain, the levels of binding of H. pylori strain CCUG 17874 to NeuAcα3-neolactohexaosylceramide (Table 1, no. 8), the NeuAc-G-10 ganglioside (Table 1, no. 16), and the G9-B ganglioside (Table 1, no. 19) were compared. As shown in Fig. 7, the blood group B type 2 epitope on the β6-linked branch of the G9-B ganglioside impaired binding compared to the linear NeuAcα3-neolactohexaosylceramide. On the other hand, the NeuAc-G-10 ganglioside was the preferred ligand, which indicates that the NeuAcα3Galβ4GlcNAc sequence on the β6-linked branch in this case increased the binding affinity.

FIG. 7.

Binding of H. pylori to serial dilutions of gangliosides. Shown is an autoradiogram obtained by binding H. pylori strain CCUG 17874 using a chromatogram binding assay. Lanes 1 to 5, serial dilutions (3 to 100 pmol) of NeuAcα3Galβ4(Fucα3)GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-Le^x hexaglycosylceramide) (NeuAcα3-Lex), NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAcα3-neolactohexaocylceramide) (NeuAcα3-nLc6), NeuAcα3Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (NeuAc-G-10 ganglioside) (NeuAc-G-10); lanes 6 to 10, serial dilution (3 to 100 pmol) of Galα3(Fucα2)Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer (G9B ganglioside). The binding assay was done as described in Materials and Methods. The results from one representative experiment out of three are shown.

(c) Effects of fucose residues.

To investigate the effects of fucose branches, the relative levels of binding of H. pylori strain CCUG 17874 to NeuAcα3-neolactohexaosylceramide (Table 1, no. 8), the VIM-2 ganglioside (Table 1, no. 12), and the NeuAcα3-dimeric-Le^x ganglioside (Table 1, no. 13) were compared. As shown in Fig. 8, the bacteria bound with higher affinity to the VIM-2 and the NeuAcα3-dimeric-Le^x gangliosides than to NeuAcα3-neolactohexaosylceramide.

DISCUSSION

Recognition of sialic acid-containing glycoconjugates by certain H. pylori strains has been repeatedly demonstrated (5, 6, 7, 19-22, 26, 27). In the present study, a library of gangliosides was collected and used for dissection of H. pylori-binding preferences utilizing representative sialic acid-recognizing H. pylori strains and mutant strains with knockout of putative ganglioside binding proteins.

Two ganglioside binding proteins of H. pylori have been identified, the SabA adhesin and the NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ-binding neutrophil-activating protein HPNAP (19, 36). HPNAP is a major immunogen of H. pylori (29) and is to some extent associated with the bacterial cell surface (2). Since whole bacterial cells also bind to NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ-terminated glycosphingolipids, it is tempting to speculate that this interaction is due to surface-associated HPNAP. However, after knockout of the gene coding for HPNAP, the bacteria still recognized glycosphingolipids with terminal NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ. The binding of NeuAcα3Galβ4GlcNAcβ3Galβ4GlcNAcβ-terminated gangliosides, as well as interaction with all other gangliosides, however, was lost after knockout of the gene for the SabA adhesin, demonstrating that ganglioside recognition of H. pylori bacterial cells is mediated solely by the SabA adhesin.

An occasional binding of the J99 strain to sialyl-Le^a hexaglycosylceramide was observed when 0.5 μg was applied on the thin-layer plate. However, the CCUG 17874 strain did not bind to this ganglioside. Binding of the J99 strain to both sialyl-Le^a and sialyl-Le^x neoglycoproteins has also been demonstrated (19). This indicates that the SabA carbohydrate binding sites of the J99 and the CCUG 17874 strains are not identical. However, in all other respects, the CCUG 17874 strain and the J99 strain bound to gangliosides in identical manners, and both strains recognized N-acetyllactosamine-based gangliosides with terminal NeuAcα3, but not NeuAcα6, in line with previous reports (6, 20). Furthermore, gangliosides with terminal NeuGcα3 or NeuAcα8NeuAcα3 were not recognized.

The factors that affected binding affinity were identified as (i) the length of the N-acetyllactosamine carbohydrate chain, (ii) the branches of the carbohydrate chain, and (iii) fucose substitution of the N-acetyllactosamine core chain.

N-Acetyllactosamine core length.

Preferential binding of H. pylori to NeuAcα3-neolactooctaosylceramide over NeuAcα3-neolactohexaosylceramide and NeuAcα3-neolactotetraosylceramide was observed. This effect is most likely due to improved accessibility of the carbohydrate head group when presented on a longer core chain.

Divalency.

Cooperative binding may account for the increased affinity for the NeuAc-G-10 ganglioside, with two NeuAcα3Galβ4GlcNAcβ branches, relative to the linear NeuAcα3-neolactohexaosylceramide. This is in agreement with the report of Simon et al. (30) demonstrating that multivalent albumin conjugates of sialyl-lactose (NeuAcα3Galβ4Glc) inhibited the adherence of H. pylori to epithelial monolayers more effectively than monovalent sialyl-lactose.

The lower binding affinity to the G9-B ganglioside than to NeuAcα3-neolactohexaosylceramide shows that the blood group B determinant on the β6-linked branch interfered with the binding process. Still, there is no absolute hindrance, since the detection level for the G9-B ganglioside was ∼100 pmol. This suggests that the H. pylori-binding determinants are mainly exposed on the β3 axis of the Galα3(Fucα2)Galβ4GlcNAcβ6(NeuAcα3Galβ4GlcNAcβ3)Galβ4GlcNAcβ3Galβ4Glcβ1Cer structure.

Fucose branches on the N-acetyllactosamine core.

The higher binding affinity for the VIM-2 ganglioside than forNeuAcα-neolactohexaosylceramide suggests that the α-linked Fuc at the innermost GlcNAc contributes to the high-affinity binding of NeuAcα-dimeric-Le^x. This fucose residue may either interact with the carbohydrate binding site of the SabA adhesin or affect the conformation of the ganglioside,providing optimal presentation of the head group. Resolution of this issue must, however, await the expression and crystallizationof the SabA adhesin.

Simon et al. have demonstrated that sialyl-lactose inhibits the binding of fresh H. pylori isolates to gastrointestinal epithelial cells and even promotes the detachment of bound bacteria (30). Furthermore, oral administration of sialyl-lactose to H. pylori-infected rhesus monkeys had a curative effect in two out of six monkeys, and one monkey was transiently cleared of the infection (24). The structural features required for high-affinity H. pylori ganglioside binding outlined above (i.e., repetitive N-acetyllactosamine units, fucose branches, and di- or multivalency) may be utilized for the construction of more efficient inhibitors of H. pylori adherence.

The expression of the SabA adhesin is, unlike that of the Le^b-binding BabA adhesin, subject to phase variation (19). Varying numbers of bacteria expressing the SabA adhesin within the bacterial-cell population used in the binding assays may account for the difficulties in determining an absolute affinity of binding for a given ganglioside. However, in repeated binding assays, the same relationships between the binding-active gangliosides were observed.

The biological significance of these findings requires further study. Although the sialic acid content of the normal human gastric epithelium is very low (18), gastric inflammation leads to an upregulation of the expression of sialic acid-containing glycoconjugates (20). Our present hypothesis is thus that the initial attachment of H. pylori is achieved through binding to receptors present in the normal gastric epithelium, e.g., the Le^b antigen and lactotetraosylceramide. The ensuing inflammation leads to enhanced expression of sialyltransferases in the gastric mucosa, ultimately providing novel binding sites for H. pylori SabA adhesin and thereby contributing to the chronicity of the infection. In addition, it was recently demonstrated that the nonopsonic H. pylori-induced activation of human neutrophils occurs by lectinophagocytosis, i.e., recognition of sialylated glycoconjugates on the neutrophil cell surface by a bacterial adhesin leads to phagocytosis and an oxidative burst with the production of reactive oxygen metabolites (37). Thus, the sialic acid binding capacity of H. pylori may have a dual role. On one hand, it mediates adhesion of bacteria to the epithelium in the already diseased stomach, and on the other, it leads to the activation of neutrophils to an oxidative burst with the production of reactive oxygen metabolites and the release of biologically active enzymes, giving rise to further tissue damage.