FIELD OF THE INVENTION

The present invention relates to the use of di- or oligo- saccharide glycosides containing at least one terminal L- fucose unit for the preparation of pharmaceutical composi¬ tions for the treatment or prophylaxis in humans of condi¬ tions involving infection by Helicobacter pylori in the human gastric mucosa, as well as a method of treating such condi- tions using the di- or oligosaccharide glycosides.

BACKGROUND OF THE INVENTION

Helicobacter pylori is a microaerophilic spiral shaped orga¬ nism (originally assigned to the genus CampyloJbacter) which is found in the stomach and generally appears to have an exclusive habitat in the human gastrointestinal mucosa. It has been estimated that this bacterium infects the gastric mucosa of more than 60% of adult humans by the time they are 60 years old. Moreover, Helicobε cter pylori has been impli¬ cated as a contributing factor in a number of pathological conditions, including acute (type B) gastritis, gastric and duodenal ulcers, and gastric adenocarcinoma.

Tissue tropism of bacteria is partly governed by the ability of a bacterial strain to adjust to the local chemical envi¬ ronment in its specific habitat. In addition, adherence is a necessary prerequisite for colonization in order to prevent removal from the new habitat, e.g. through peristalsis in the gastrointestinal tract. In mammals, bacteria adhere to pro- teins or glycoconjugates (glycosphingolipids, glycoproteins) on or in the vicinity of epithelial cell surfaces (mucus) , and a number of specific bacterial adhesin-protein and adhe- sin-carbohydrate interactions have been described in the literature.

CONFIRMATION C0FY

With respect to Helicobacter pylori , studies in model systems such as mouse adrenal Y-l cells (see D. G. Evans, D. J. , Jr. Evans, and D. Y. Graham, (1989) Infect . I mun . 57, 2272-2278) have suggested that surface-associated flexible fibrillar structures that surround this bacterium function as adhesins or colonization factor antigens to mediate the binding of HelicoJbacter pylori to cellular sialic acid-containing glycoprotein receptors.

SUMMARY OF THE INVENTION

Through studies of interactions between Helicobacter pylori and human stomach mucosa epithelial cells, it has now sur¬ prisingly been found, firstly, that Helicobacter pylori adheres or binds only to certain differentiated epithelial cells in the glandular epithelium which is composed of gas- trie units. The gastric units in the glandular epithelium are lined with cells that have differentiated to perform various functions. Thus, the upper pit region of the gastric unit (gastric pit) is lined with mucus-producing surface epithe¬ lial cells, whereas the constricted mid-portion of the gas- trie unit (its isthmus) is composed of proliferating and non- proliferating immature cells, mucus neck cells and parietal cells, and the lower portion of the gastric unit (the gland) may contain intrinsic factor- and pepsinogen-producing chief cells, acid-producing parietal cells, mucous neck and gland cells, and a variety of enteroendocrine cell types. The present studies have shown that Helicobacter pylori binds solely to the mucus-producing surface epithelial cells in the upper pit region, i.e. the surface mucus cells.

Secondly, the studies have shown that the binding of Heli - coJacter pylori to the mucus-producing surface epithelial cells does not involve interaction between bacterial adhesins and sialic acid-containing receptors but rather involves interaction with carbohydrate structures containing terminal units of L-fucose (also known as 6-deoxy-L-galactose) .

Consequently, the present invention concerns in one aspect the use of a di- or oligosaccharide glycoside having at least one terminal L-fucose unit for the preparation of a pharma¬ ceutical composition for the treatment or prophylaxis in humans of conditions involving infection by Helicobacter pylori of the human gastric mucosa. Likewise, the invention concerns a method of treating and/or preventing diseases in humans caused by infection by Helicobacter pylori by admini¬ stering to a patient in need thereof an effective amount of a di- or oligosaccharide glycoside comprising at least one terminal L-fucose unit.

DETAILED DESCRIPTION OF THE INVENTION

In the present context, the term "terminal L-fucose unit" is intended to mean that the L-fucose unit or units in question, which is/are bound via the 1-position, is/are not itself/themselves glycosylated by other carbohydrate units or groups. This does not exclude, however, that a carbohydrate unit which is glycosylated by a L-fucose unit is itself glycosylated by other carbohydrate units or groups.

In the present context, the term "oligosaccharide" is intended to mean that the saccharide portion of the glycoside comprises three or more carbohydrate units, such as between three and ten carbohydrate units.

It is preferred that the di- or oligosaccharide glycoside having at least one terminal L-fucose unit used according to the invention is capable of binding to adhesins present on the surface of Helicobacter pylori .

At the present time, the most useful model system for de¬ termining the ability of the di- or oligosaccharide glycoside having at least one terminal L-fucose unit to inhibit the binding of cells of Helicobacter pylori to the gastric epithelium is an in vitro histological model involving the

use of histological sections of human gastric tissue con¬ taining gastric epithelium.

Consequently, a preferred embodiment of the use or the method of the invention is that in which the di- or oligosaccharide glycoside having a terminal L-fucose unit is one which is capable of inhibiting or substantially reducing the adhesion of cells of Helicobacter pylori to epithelial cells in a histological section of human gastric mucosa.

More preferably, the di- or oligosaccharide glycoside employed in the use or the method of the invention is one which, when cells of Helicobacter pylori are preincubated with the di- or oligosaccharide glycoside in a concentration of up to 500 μg/ml, is capable of inhibiting the adhesion of said bacterial cells to epithelial cells of a histological section of gastric human mucosa by at least 50% compared to the adhesion of corresponding non-preincubated bacterial cells.

In the presently preferred histological model system, the histological section of human gastric mucosa is prepared by fixing a sample of non-diseased gastric human mucosa tissue with formalin, embedding the sample in paraffin, providing an approximately 5 μm section of the embedded sample, placing the section on a glass slide, deparaffinizing the section by washing with xylene and isopropanol, and incubating the section with a buffer consisting of bovine serum albumin and a non-ionic polyoxyethylene sorbitan monolaurate surfactant (such as Tween-20) , preferably at a concentration of about 0.2% and 0.05%, respectively, in phosphate-buffered saline, cf. also the example.

For the purpose of determining the degree of adhesion of the Helicobacter pylori cells to the section of gastric mucosa, it is advantageous that the bacterial cells are labelled in some manner to enable detection of the bacterial cells at specific sites on the section. The labelling may conceivably

be performed according to any of the well-known methods for labelling live bacterial cells or bacterial cells located on a microscopic sample, such as staining by means of a dye (such as a dye specific to e.g. Gram-negative bacteria) followed by microscopy; labelling with a radioactive isotope or a compound containing such an isotope followed by micro- autoradiography; treatment (before or after the adhesion to the epithelium) with a labelled (enzyme-labelled, radiolabel- led or otherwise) antibody recognizing H. pylori , followed by detection or visualization of the adhered antibody by a secondary antibody enzyme conjugate or an ELISA-type detec¬ tion or by microautoradiography.

However, the presently preferred labelling principle is fluorescence-labelling, in particular labelling by means of fluorescein isothiocyanate. This may advantageously be carried out by treating a suspension of cells of Helicobacter pylori in buffer containing sodium chloride and sodium car¬ bonate, preferably at a concentration of about 0.15 M and 0.1 M, respectively, at about pH 9.0 with fluorescein isothiocya- nate at a concentration of about 0.1 mg/ml, incubating the bacterial cell suspension for 1 hour at room temperature and separating the bacterial cells by centrifugation followed by washing of the bacterial cells with phosphate buffered saline at about pH 7.5 containing a non-ionic polyoxyethylene sorbitan monolaurate surfactant, preferably at a concentra¬ tion of about 0.05%.

In order to determine the ability of a di- or oligosaccharide glycoside having at least one terminal L-fucose unit to inhibit the adhesion of Helicobacter pylori to human epithelial gastrointestinal cells, the bacteria are typically subjected to preincubation with the glycoside prior to bring¬ ing the bacteria into contact with the tissue section to allow them to adhere to the surface of epithelial cells. One useful way of performing the preincubation of the bacterial cells is by adding the di- or oligosaccharide glycoside in a concentration of up to 500 μg/ml to a suspension of the

bacterial cells in a buffer consisting of bovine serum albu¬ min and of a non-ionic polyoxyethylene sorbitan monolaurate surfactant, preferably at a concentration of about 0.2% and 0.05%, respectively, in phosphate-buffered saline at about pH 7.5 for a period of about 2 hours at room temperature, sepa¬ rating the bacterial cells by centrifugation, and washing the cells in the same buffer.

When cells of Helicobacter pylori (whether preincubated or not, and whether labelled or not) are to be tested for their ability of adhering to epithelial cells in the human gastrointestinal mucosa, this is typically performed by applying a dilute Helicobacter pylori cell suspension con¬ taining about 10 ⁶ -10 ⁸ bacterial cells per millilitre to the histological section, incubating the section with the bac- terial cell suspension for 1 hour at room temperature in a humidified chamber, washing the slide in phosphate buffered saline, and establishing the degree of adhesion to the epi¬ thelial cells of the section. In the embodiment where the bacteria are fluorescence-labelled, e.g. with fluorescine isothiocyanate, establishing the degree of adhesion is typi¬ cally and conveniently carried out by studying the section under a fluorescence microscope and assessing the degree of adhesion from the number of fluorescent bacteria attached to the epithelial surface cells in the section, cf. also the examples, in particular the figures explained therein.

The adhesion testing may be carried out using cells of the Helicobacter pylori strains NCTC 11637, NCTC 11638, WV229 or P466 (cf. the example below).

The conditions involving gastrointestinal infection by Heli - coJbacter pylori , the treatment or prophylaxis of which the pharmaceutical composition prepared in the use according to the invention as well as the treatment method according to the invention may be employed for, comprise for example chronic active (type B) gastritis, gastric ulcers, duodenal ulcers, gastric adenocarcinoma, and gastric lymphoma.

In a preferred embodiment, the oligosaccharide contains two terminal fucose units as defined above.

In a preferred embodiment, the di- or oligosaccharide glycoside to be employed is a glycoprotein. Interesting examples of compounds to be employed are human K -casein, human colostrum IgA, and bovine submaxillary gland mucin.

In a further preferred embodiment, the terminal tetra- saccharide at the non-reducing end of the oligosaccharide chain is Lewis b-tetrasaccharide, i.e. Fucαl-2Gal3l-3 (FuCQU-4)GlcNACj8l- .

The di- or oligosaccharide glycosides may be contained in any appropriate amount in a pharmaceutical composition, and are generally contained in an amount between 0.01% and 99%, or between 0.1% and 99% by weight of the total weight of the composition.

The pharmaceutical compositions containing the active in¬ gredients may suitably be formulated so that they provide doses within these ranges either as single dosage units or as multiple dosage units.

The active di- or oligosaccharide glycosides used according to the invention as well as compositions containing the glycosides may typically be administered to a human patient in amounts corresponding to from about l mg per day to about 50 g per day, preferably from about 10 mg per day to about 5 g per day, of the active glycoside ingredient.

The dosage of a di- or oligosaccharide glycoside depends in general on the choice of administration route, the particular disease to be treated and the severity of same, and also whether the disease is to be treated or prevented, as well as the age and weight of the person to be treated.

The pharmaceutical composition may preferably be administered by the oral route, by the parenteral route or by the rectal route.

The invention is further illustrated by the following non- limiting examples.

EXAMPLE 1

BIOLOGICAL EXPERIMENTS

MATERIALS AND METHODS

Purification of K -casein from human milk

Human milk was kept frozen at -20°C until used. The milk was thawed and skimmed milk obtained by centrifugation at 15,000 x g for 45 min. The fat layer at the top of the tube was carefully removed while the soluble fraction and the pellet were pooled. The pool was acidified with HC1 to pH 4.3 and CaCl ₂ was added to a concentration of 60 mM. Following incu¬ bation at room temperature for l h, the precipitated casein fraction was collected by centrifugation at 18,000 x g for 90 min.

The casein fraction was dissolved in 20 mM ethanolamine, 6 M urea, at pH 9.5. To remove any remaining fat, this aqueous fraction was extracted three times with 4 volumes of hexane. The aqueous phase was thereafter dialysed against distilled water and lyophilized. The lyophilized proteins were dis¬ solved in 50 mM imidazole-HCl, 0.5% SDS, 0.5% 2-mercaptoetha- nol at pH 7.0 and chromatographed on a Sephadex G-200 column (1.6 x 120 cm, Pharmacia-LKB, Uppsala, Sweden), equilibrated with 50 mM imidazole-HCl, 0.5% SDS, 0.5% 2-mercaptoethanol at pH 7.0 at a temperature of 37°C. K -Casein-containing frac¬ tions which were detected by Schiff staining were pooled, dialysed against 40% methanol in distilled water for 3 h and lyophilized to reduce the volume. The protein was redissolved

in 50 mM imidazole-HCl, 0.5% SDS, 0.5% 2-mercaptoethanol at pH 7.0 and a second time on the Sephadex G-200 column under the same conditions as above. Human K -casein-containing fractions completely free of β-casein were collected, dialysed against 40% methanol and lyophilized.

Purification of bovine mature milk and colostrum K -casein

Bovine mature milk and colostrum collected within 24 h of parturition were obtained from Agrisera AB (Tvar-iback, Sweden) . Both were defatted and the respective casein frac- tion was prepared as described above for the purification of human milk with the exception that no delipidation with hexane was necessary. The casein pellet was dissolved in 10 mM imidazole-HCl, 3.3 M urea, 10 mM 2-mercaptoethanol, at pH 7.0 and applied to a Q-Sepharose column (2.6 x 15 cm, Pharma- cia LKB, Sweden) . After washing, the column was eluted with a linear gradient of 0.1-0.5 M NaCl in the same buffer. Bovine K -casein-containing fractions were dialysed against water and lyophilized. The identification of bovine caseins was con¬ firmed by chymosin cleavage.

In situ adherence assay for H. pylori

Multiple non-diseased tissue samples of adult human esopha¬ gus, stomach, duodenum, colon, kidney, cervix, endometrium and midbrain were obtained from the surgical pathology and autopsy files of the Department of pathology at Washington University. The gastrointestinal tract of 250 g Sprague-

Dawley rats, and 6-12 week old FVB/N mice were removed after sacrifice by cervical dislocation and regionally dissected as described in Sweetser, D. A., Birkenmeier, E. H. , Hoppe, P. C, McKeel, D. W. , and Gordon, J. I., Genes Devel . 2 (1988), 1318-1332. Stomachs from adult dogs were obtained from the Department of Surgery, St. Louis University Medical Center. All tissues were fixed in 10% formalin or in a solution of picric acid/formaldehyde/glacial acetic acid (15:5:1; Bouin's fluid) and subsequently embedded in paraffin (see Sweetser,

D. A., Birkenmeier, E. H. , Hoppe, P. C. , McKeel, D. W. , and Gordon, J. I., Genes Devel . 2 (1988), 1318-1332). Sections with a thickness of 5 μm placed on glass slides were prepared and used for hematoxylin and eosin staining (to identify the cell types present in gastric units, and to verify that the tissue samples had no pathologic changes) and/or for sub¬ sequent adherence, histochemical and/or immunocytochemical assays.

Five previously characterized clinical isolates of H. pylori were used. NCTC 11637 and NCTC 11638 are reference strains isolated in 1982 from patients with active chronic gastritis (see Warren, J. R. , and Marshall, B., Lancet 1 (1983), 1273- 1275) . Strain WV229 was recovered from a patient with gastric ulcer (see Westblom, T. U. , Madan, E., Subik, M. A., Buriex, D. E., and Midkiff, B. R. , Scand. J. Gastroenterol . 27

(1992) , 249-252) , while P466 (obtained from The Johns Hopkins University School of Medicine, Baltimore) was obtained from a patient with acute gastritis. Strain M019 was isolated from an asymptomatic carrier (see Barthel, J. S., Westblom, T. U. , Havey, A. D., Gonzalez, F. and Everett, E. D., Arch. Intern . Med. 148 (1988), 1149-1151). Strains of H. pylori were grown at 37°C on Brucella agar supplemented with 10% bovine blood and 1% IsoVitalex (Becton Dickinson Microbiology Systems, Cockeyville, MD) under microaerophilic conditions (5% 0 ₂ , 10% C0 ₂ , 85% N ₂ ) and 98% humidity. Five days after inoculation, about 1 μl of bacteria were removed with a sterile loop from a plate and were resuspended by gentle pipetting in 1 ml of 0.15 M NaCl/0.1 M sodium carbonate with a pH of 9.0. 10 μl of a freshly prepared 10 mg/ml solution of fluorescein isothio- cyanate (FITC, Sigma Chemical Co.) in dimethylsulfoxide was added to the suspension which was then incubated for 1 h at room temperature in the dark. The bacterial cells were recovered by centrifugation at 3,000 x g for 5 min, and then resuspended in 1 ml of a phosphate buffered saline containing 0.2% bovine serum albumin and 0.05% polyoxyethylene sorbitan monolaurate (blocking buffer l, BB1) at pH 7.4 by gentle pipetting, and pelleted by centrifugation as above. The wash

procedure was repeated 3 times. The intensity of FITC- labelling of all bacterial strains was similar as judged by inspection of comparable numbers of organisms by fluorescence microscopy. Aliqouts of 100 μl were taken from the final suspensions with an optial density of 1.00 O.D./ml, measured by A _600nιn , and utilized immediately or stored at -20°C until use. No differences in binding patterns were observed between bacteria labelled and used fresh and bacteria that were frozen and thawed once before use.

Slide-mounted tissue sections were deparaffinized in xylene and isopropanol (xylene, 10 min; xylene, 3 min; isopropanol, 3 min; isopropanol, 3 min; and isopropanol 3 min) rinsed in water followed by 3 times in PBS and then incubated for 30-60 min in blocking buffer 1 (0.2% bovine serum albumin/0.05% polyoxyethylene sorbitan monolaurate in PBS) . The FITC- labelled bacterial suspension was diluted 20-fold in blocking buffer 1 and a 200 μl aliquot was placed on a slide-mounted tissue section and then incubated for 1 h at room temperature in a humidified chamber. The slides with the treated tissue sections were subsequently washed 4-6 times with PBS prior to inspection under a fluorescence microscope.

To analyze the ability of various glycoproteins or free oligosaccharides to inhibit binding, 200 μl aliquots of suspensions of FITC-labelled bacteria were preincubated for 2 h with bovine submaxillary mucin (Sigma, final concentra¬ tion of 500 μg/ml, dissolved in PBS pH 7), fetuin (Sigma, 100 μg/ml, dissolved in PBS pH 7) , asialofetuin (Sigma, 100 μg/ml, dissolved in PBS pH 7) , human sialyl-lactose (NeuAc- Q!2→6Galj8l→4Glc) (Sigma, 5 μg/ml, dissolved in PBS pH 7), bovine sialyl-lactose (NeuAcα.2→3Gal/3l-*4Glc) (Sigma, 5 μg/ml, dissolved in PBS pH 7) , purified human κ-casein (concentra¬ tions of 1000, 500, 250, 50, 10, and 1 μg/ml, dissolved in PBS pH 7) , or purified bovine mature or colostrum K -casein (concentration of 1,000 μg/ml, dissolved in PBS pH 7) for 2 h at room temperature, human serum IgA (Cappel Research Pro¬ ducts, 500 μg/ml, dissolved in PBS pH 7) , or human colostrum

secretory IgA (Cappel Research Products, 15 μg/ml, dissolved in PBS, pH 7) . The bacteria were washed once in blocking buffer 1 before the mixture was added to tissue sections.

Human K -casein or human colostrum secretory IgA was then washed with PBS using I _* . = 10,000 cutoff Centricon filters (Amicon) . Human K -casein and human colostrum secretory IgA were also incubated with 100 mU of bovine kidney α!-L-fucosi- dase or Vibrio colerae neuraminidase (Boehringer-Mannheim, Germany) for 2 h at 37°C prior to incubation with bacteria. K -Casein incubated in phosphate buffered saline at pH 7.4 served as control. Glycosidase treated samples and the con¬ trol were all incubated at 80°C for 20 min to inactivate the enzymes before the K -casein was incubated with bacteria. IgA incubated in PBS served as control. Glycosidase treated samples and the control were all incubated at 80°C for 20 min to inactivate the enzymes before the IgA was incubated with bacteria.

To further characterize the receptor active domain of human K -casein and of human colostrum secretory IgA, respectively, periodate oxidation was performed as described below.

Two other experiments were conducted to ascertain the nature of the bacterial receptor in tissue sections. Firstly, de- paraffinized sections were treated with 200 mU proteinase K which is a serine protease that is able to catalyse the hydrolysis of the peptide bond to the COOH-site of aliphatic and/or aromatic amino acids (see Ebeling et al. Eur. J. Biochem. 4 (1974) , 91-97) from Trichiratυm albus (Boehringer- Mannheim, Germany) prepared as a 1 mg/ml stock solution in PBS for 2 h at 37°C. They were subsequently washed three times in PBS and treated with blocking buffer 1, and a sus¬ pension of FITC-labelled H. pylori strain P466 or WV229 and was then overlaid as described above. Secondly, deparaffi- nized sections were treated with 10 mM sodium meta- periodate/50 M sodium acetate at a pH of 4.5 and for 10 min at 0°C to selectively cleave carbon atoms number 8 and 9 of

the unsubstituted side chain of terminal sialic acids (see Manzi, A. E., Dell, A., Azadi, P., and Varki, A., J. Biol . Chem. 265 (1990) , 8094-8107) , or with 10 mM sodium meta- periodate/50 mM sodium acetate at a pH of 4.5 for 1 h at room temperature to cleave carbon-carbon bonds between vicinal hydroxyl groups in most carbohydrates with a free hydroxyl group in the 3 position (thereby breaking up the ring struc¬ ture of the monosaccharide and thus destroying the carbohy¬ drate epitope while leaving peptides intact, see Woodward, M. P., Young, W. W Jr., and Bloodgood, R. A., J. Immunol . Meth. 78 (1985), 143-153. Control sections were incubated with 50 mM sodium acetate buffer alone. After two PBS washes, the sections were reduced by adding 50 mM sodium borohydride prepared in PBS with a pH of 7.6. After several additional washes with PBS, suspensions of FITC-labelled strains WV229, P466, or M019 were applied and the slides were processed as described above.

Control experiments were used to establish that the antige- nicity of proteins present in gastric epithelial cell line- ages was preserved under the "harsher" periodate oxidation at pH 4.5. Thus, periodate-treated sections of rat stomach were incubated with a rabbit polyclonal antisera (10 μg pro¬ tein/ml, 4°C, overnight) raised against intrinsic factor (IF) and the antigen-antibody complexes subsequently detected by Texas Red conjugated donkey anti-rabbit IgG (see Roth, K. A., Cohn, S. M. , Rubin, D. C, Trahair, J. F., Neutra, M. R. and gordon, J. I, Am. J. Physiol . (Gastrointest. Liver Physiol . ) 263 (1992), G186-G187, and Lee, E. Y., Seetharam, B., Alpers, D. H. and DeSchryver-Kecskemeti, K. , Gastroenterol. 97 (1989), 1171-1180 for the staining protocol used and details about the specificity of this sera) . The intensity of IF staining of chief cells in periodate-treated sections was similar to the intensity of staining of chief cells in sec¬ tions that had been incubated with buffer alone while the binding of the α.-L-fucose specific Ulex europaeus type l lectin to surface mucous cells was completely abolished under the same conditions.

Immunohistochemical studies

The cellular distribution of sialylated oligosaccharides was examined in deparaffinized sections of human stomach contain¬ ing zymogenic glands by incubation with (i) flourescein- , rhodamine- (LIST Biological Laboratories Inc.), or peroxidase- (Sigma) conjugated cholera toxin B subunit which recognizes internally positioned NeuAcα2→3 residues linked to galactose (see Holmgren, J., Lδnnroth, I., Mansson, J.-E., and Svennerholm, L, Proc. Natl . Acad. Sci . USA 72 (1975), 2520-2524, final concentration of 5 μg protein/ml, dissolved in blocking buffer 1), (ii) digoxigenin (DIG) -conjugated Sambucus nigra lectin (SNA) which recognizes NeuAc- c2→6Gal/GalNAc (see Knibbs, R. N. , Goldstein, I. J., Rat- cliffe, R. M. , and Shibuya, N. , J. Biol . Chem. 266 (1991), 83-88, 10 μg protein/ml, dissolved in blocking buffer 1) , and DIG-conjugated Maackia amurensis lectin (MAA) which reacts with NeuAcα.2-»3Galβl-»4GlcNAc (see Knibbs, R. N., Goldstein, I. J., Ratcliffe, R. M. , and Shibuya, N. , J. Biol . Chem. 266 (1991) , 83-88, 10 μg protein/ml, dissolved in blocking buffer 1) . DIG-conjugated lectins were detected with peroxidase- conjugated monoclonal mouse anti-DIG antibody (500 mU per- oxidase/ml, dissolved in blocking buffer 1) by washing the tissue three times in PBS, overlaying the secondary antibody on the tissue for 1 h at room temperature, followed by stan- dard H ₂ 0 ₂ /DAB detection.

Fucosylated blood group antigens were detected by first treating the sections with mouse monoclonal antibodies directed against the histo-blood group antigens A, B, and H (Dakopatts A/S, Glostrup, Denmark) (final concentration of 1 μg protein/ml, dissolved in blocking buffer 1), or Le ^a and Le ^b (Immucor, Norcross, GA) (final concentrations 10 μg protein/ml, dissolved in blocking buffer 1) . Antigen-antibody complexes were detected by visualizing them with a second overlay of FITC-, TRITC- or HRP-conjugated (fluorescein-, rhodamine- or peroxidase-conjugated) rabbit anti-mouse immu- noglobulins (Dakopatts A/S, Glostrup, Denmark) (final concen-

tration 30 μg protein/ml, dissolved in blocking buffer 1) and studying the sections under a fluorescence microscope. In addition, the distribution of fucosylated glycoconjugates was assessed in a similar manner by treating and visualizing sections with FITC-conjugated Ulex europaenus type 1 lectin (UEAl, 5 μg protein/ml, dissolved in blocking buffer 1) which recognizes and binds to cn-L-fucose groups.

RESULTS AND DISCUSSION

Analysis of cell lineage-specific binding of H. pylori using an in situ adherence assay

Five clinical isolates of Helicobacter pylori were labelled with fluorescein isothiocyanate followed by 5 days of growth on rich agar under microaerophilic conditions and then over¬ laid on sections of formalin-fixed human stomach. Strains NCTC 11637, NCTC 11638, WV229, and P466 which all were recovered from patients with dyspeptic syndrome bound to surface mucous cells situated in the upper pit and luminal surface (see Figure 2A,B) . Remarkably, mucous neck cells located in the upper portions of the glandular segment of the gastric units were negative indicating that these bacteria were able to distinguish between two differentiated mucus- producing cell lineages present in the stomach (see Ota, H. , Katsuyama, T., Ishii, K., Nakayama, J., Shiozawa, T. , and Tsukahare, Y., Histochem. J. 23 (1991), 22-28). No binding to parietal or chief cells was noted in zymogenic glands. Strain M019, the only isolate tested which was recovered from an asymptomatic "healthy" carrier, did not bind at a detectable level to any human gastric epithelial cell lineages (see Figure 2C) .

Strains WV229 and P466 did not adhere to the squamous epithe¬ lium of human esophagus but did adhere under the reaction conditions employed to esophageal submucosal glands and their ducts, and to duodenal villus-associated enterocytes (see Figure 2D) . A weak binding to enterocytes situated in the

colonic homolog of small intestinal villi - the surface epi¬ thelial cell cuff which surrounds each crypt orifice (see Figure 2E) was also observed. The intensity of staining and hence the density of adherent organisms was considerably less down through the cephalocaudal axis of the intestine than it was in the stomach epithelium (gradient down through the intestine) . Control experiments indicated that Helicobacter pylori P466 and WV229 did not bind to any epithelial cell populations represented in proximal or distal nephrons, in the cervix, or in the endometrium. Surveys of the central nervous system including the midbrain also failed to produce a signal above background.

The in si tu adherence assay was used to ascertain whether the cell lineage-specific and Helicobacter pylori strain-specific patterns of binding observed in the human gut occurred in other mammalian species. Strains WV229, P466 and M019 grown and labelled with FITC as described above, were incubated with sections of rat and mouse gastrointestinal tracts as well as with stomach from dogs. Strain M019 did not bind in detectable amount to tissue sections prepared from any of these three species. As in the human stomach, strains WV229 and P466 bound to the gastric pit region in the rat but not to any other differentiated epithelial cell population located in the zymogenic or pure mucous zones (see Figure 2F) . Unlike in similar human preparations, the stratified squamous epithelium of the rat esophagus and forestomach produced positive results in the in si tu adherence assay, while the small intestinal epithelium was negative with the exception of Brunner's glands. The forestomach of the mouse showed high-density binding with adherent bacteria, while the remainder of the stomach, i.e. the zymogenic, mucoparietal and pure mucous zones (see Lee, E. R., Trasler, J., Dwivedi, S., and leBlond, C. P., Am. J. Anat. 164 (1982), 187-207), and the epithelial and mesenchymal components of the intes- tine did not yield a signal with FITC-labelled strains WV229 and P466. Finally, no binding of either of the two strains was observed in the dog stomach.

Biochemical studies of the interaction between Helicobacter pylori and gastric epithelial cell lineages

To characterize the nature of the interactions between Heli ¬ cobacter pylori and the surface mucous cells, the ability of a series of compounds to inhibit binding of FITC-labelled strains P466 and WV229 to sections of human stomach and proximal small intestine was tested.

When calf serum fetuin, asialofetuin, soluble sialyl-lactose prepared from bovine milk (in which 85% of the sialic acid is linked to galactose via an α.2→3 linkage and 15% is linked via an α.2→6 linkage, i.e. NeuAcα.2→3/6Galβl-»4Glc) or human milk sialyl-lactose (15% NeuAcα.2→3; 85% NeuAcα2-»6) were preincuba¬ ted with the above-mentioned two Helicobacter pylori strains, no subsequent reduction in adherence was noted as judged by the in situ assay. These findings, i.e. sialic acid indepen¬ dent binding, are in agreement with the work of Fauchere, J.- L., and Blaser, M. J. in MicroJb. Pathog. 9 (1990), 427-439.

However, preincubation of the bacteria with 0.5% bovine submaxillary gland mucin which is a rich source of both fucosylated and sialylated carbohydrates (see Savage, A. V. , D'Arcy, S. M. T. , and Donoghue, C. M. , Biochem. J. 279 (1991), 95-103 and 33), completely inhibited binding of bacteria to surface mucous cells (see Figure 1A,B) and to duodenal, villus-associated, epithelial cells. Human κ-casein and human colostrum secretory IgA were also potent inhibitors: 250 μg/ml of human K -casein (see Figure 3A) and 15 μg/ml of human colostrum secretory IgA, respectively, fully inhibited adherence of Helicobacter pylori .

Human K -casein oligosaccharides are fucosylated via an Q!l-»4- linkage to N-acetylglucosamine and differs in this respect from bovine K -casein. Human colostrum secretory IgA carries a highly varied set of N- and O-linked oligosaccharides.

Thus, it is therefore interesting that only the human - - casein and human colostrum secretory IgA were able to prevent the cell-lineage specific attachment of Helicobacter pylori to human gastric surface mucous cells. Several observations suggested that this inhibition of Helicobacter pylori binding by human colostrum secretory IgA was not mediated by a bacte¬ rial Fc-receptor or by a classical immunoglobulin recognition effect, involving the hypovariable of the Fab fragments. Thus, it was observed that (i) inhibition was abolished when the human K -casein or human colostrum secretory IgA was pretreated with meta-periodate, (ii) preincubation at 85°C for 30 min in PBS did not affect the inhibitory activity of the two glycoproteins, and (iii) treatment of human K -casein (see Figure 3C) or human colostrum secretory IgA with α-L- fucosidase markedly reduced inhibitory activity while neuraminidase treatment had a lesser effect on the inhibitory activity of human K -casein and no detectable effect on the inhibitory activity of human colostrum secretory IgA.

In comparison, K -casein from bovine mature milk and colostrum (see Figure 3B) did not inhibit binding of Helicobacter pylori at concentrations as high as 1000 μg/ml, and human serum IgA did not inhibit binding at concentrations as high as 100 μg/ml. Pretreatment of sections of human stomach with proteinase K also produced a marked decrease in the binding of the two Helicobacter pylori strains P466 and WV229.

Taken together, these results strongly indicate that binding is mediated by a fucosylated rather than a sialylated glycoprotein receptor expressed on surface mucous cells.

The indication that binding of Helicobacter pylori adhesin(s) to the surface mucous cell population of human stomach does not depend upon sialic acid epitopes in a cellular receptor is further supported by the results of two additional experi¬ ments.

Firstly, the distribution of sialic acid-containing complex carbohydrates in the human gastric mucosa does not correlate with the cellular patterns of adherence observed in the in si tu assay. Determination of this distribution was carried out in the experiments using Maackia amurensis agglutinin (MAA) which is specific for NeuAcα.2→3Galj3l→4GlcNAc epitopes (on sialylated N/O-glycans and glycosphingolipids) (see Knibbs, R. N., Goldstein, I. J., Ratcliffe, R. M. , and Shi¬ buya, N., J. Biol . Chem. 266 (1991), 83-88), and Sambucus nigra agglutinin which recognizes NeuAcα.2→6Gal/GalNAc struc¬ tures (see Shibuya, N. , Goldstein, I. J., Broekaert, W. F., Nsimba-Lubaki, M. , Peeters, B., and Peumans, W. J., J. Biol . Chem. 262 (1987), 1596-1602). It turned out that binding of the MAA and SNA lectins was confined to the submucosal com- partment of the human stomach. Thus, they did not react with members of any gastric epithelial cell lineage (see Figure 1C,D) .

Similarly, the cholera toxin B subunit recognizes internally linked sialic acid in glycoproteins and glycosphingolipids, specifically GalNAcβl→4 (NeuAcα2→3)Gal- β-structures (see

Holmgren, J., Lόnnroth, I., Mansson, J.-E., and Svennerholm, L, Proc. Natl . Acad. Sci . USA 72 (1975), 2520-2524). It turned out that this lectin did not bind to surface mucous cells but rather that binding was confined to mucous neck cells located in the upper glandular domains of gastric units (see Figure IE) . Control experiments with dog stomach demon¬ strated (i) that MAA binds to surface mucous cells, (ii) that SNA binds to surface mucous and parietal cells, and (iii) that cholera toxin B subunit does not bind to any gastric epithelial cell lineage in this species. This suggests that these lectins can be used to define fundamental differences in the differentiation program of the surface mucous cell lineage between dogs and humans - differences which may account for their distinct abilities to bind Helicobacter pylori .

Secondly, when tissue sections were incubated with 10 mM meta-periodate/sodium acetate at pH 5.5 and at 0°C, selective cleavage of carbons 8 and 9 of the unsubstituted side chain of terminal sialic acids can be achieved (see Manzi, A. E., Dell, A., Azadi, P. and Varki, A., J. Biol . Chem. 265 (1990), 8094-8107), e.g. as was evident by the loss of SNA binding to surface mucous cells in the dog stomach. Using these periodate oxidation conditions, there was no reduction in adherence of any of the binding strains to human stomach or small intestinal epithelial cell populations as compared to control sections that had been treated with sodium acetate buffer alone.

Other observations support a role for fucosylated epitopes in the glycoprotein(s) that mediate binding of Helicobacter pylori to surface mucous epithelial cells in the human sto¬ mach. Bacterial binding was reduced in sections of human o stomach containing zymogenic glands that had been preincu¬ bated with monoclonal antibodies specific for either of the three blood group antigens (e.g. see Figure 1F-H) . Ulex europaeus type 1 agglutinin (UEAl) which is specific for α.-L- fucose also bound to the same cells which contained a receptor for Helicobacter pylori adhesin(s) (see Figure II) . The "mild" meta-periodate oxidation reaction conditions employed above had no effect on UEAl binding (see Figure 1J) . This treatment also had no effect on Helicobacter pylori binding (see Figure 1K,L) . Harsher cleavage conditions (i.e. reducing the pH to 4.5, increasing the incubation time to 1 h; and raising the incubation temperature to 20°C) were required to ablate UEAl binding and the binding of the monoclonal blood group H antibody to human surface mucous cells. These conditions also resulted in loss of cholera toxin B subunit binding and loss of adherence of Helicobacter pylori strains WV229 and P466.

The above described multilabel immunohistochemical studies indicated that receptor sites for FITC-labelled strains P466 and WV229 are co-expressed in members of this epithelial cell

lineage together with the fucosylated histo-blood group antigens H, B and Le ^b (e.g. see Figure 1F-H) .

Together, these results strongly imply that the binding of Helicobacter pylori to UEAl-positive, fucosylated blood group antigen-positive, surface mucous cells in the human stomach is not dependent on terminal non-substituted sialic acid residues. Furthermore, the distribution of MAA and SNA lectin and cholera toxin B subunit binding sites plus the failure of fetuin and soluble sialyl-lactose to inhibit binding argue strongly against involvement of the α.2→3 linked sialic acid residues that has been invoked in previous studies (see Evans, D. G., Evans, D. J. Jr., and Graham, D. Y., -Infect. I mun . 57 (1989), 2272-2278, Evans, D. G., Evans, D. J. Jr., Moulds, J. J., and Graham, D. Y., Infect. Immun. 56 (1988), 2896-2906, and Saitoh, T., Natomi, H. , Zhao, W. , Okuzumi, K., Sugano, K. , Iwamort, M. , and Nagai, Y., FEBS Lett. 282 (1991) , 385-387) ; studies which employed cells that are not known targets for the organisms in vivo (see Evans, D. G., Evans, D. J. Jr., and Graham, D. Y., Infect. Immun . 57 (1989), 2272-2278, and Evans, D. G. , Evans, D. J. Jr.,

Moulds, J. J. , and Graham, D. Y. , Infect. Immun. 56 (1988), 2896-2906) .

Figure 1 shows the characterization of a putative Helico¬ bacter pylori adhesin receptor as a glycoprotein that lacks sialyl-lactose.

Fig. IA and IB: Sections of human stomach containing gastric units with zymogenic glands were incubated with FITC-labelled Helicobacter pylori strain WV229 (IA) or with bacteria that had been treated with a 0.5% solution of bovine submaxillary gland mucin (IB) . The mucin prepa¬ ration produces marked reductions in binding.

Fig. 1C and ID: Section of human stomach containing members of the surface mucous, mucous neck, parietal and chief epithelial cell lineages were incubated with

digoxigenin-labelled Maackia amurensis agglutinin (MAA, 1C) or Sambucus nigra agglutinin (SNA, ID) . DIG-conju¬ gated lectins were detected with a peroxidase-conjugated monoclonal mouse anti-DIG antibody preparation. None of the epithelial cell lineages produces a signal above background when incubated with these lectins that recog¬ nize sialic acid-containing carbohydrate epitopes. Con¬ trol experiments employing the monoclonal antibody alone produced no staining (data not shown) .

Fig. IE: Adjacent section of human stomach incubated with rhodamine-conjugated cholera toxin B subunit. Mucus- producing cells located in the isthmus or upper portion of the glandular domain of gastric units react with this toxin which recognizes internally linked sialic acids in glycoproteins and glycosphingolipids. The subunit does not bind to surface mucous cells located in the pit.

Fig. 1F-1H: Sections of human stomach were incubated with FITC-labelled Helicobacter pylori strain P466 (IF) and a mouse monoclonal antibody directed against the fucosy- lated blood group antigen H (visualized with rhodamine- conjugated rabbit anti-mouse IgG in Fig. 1G) . Fig. 1H is a double exposure showing that surface mucous cells co- express the bacterial adhesin receptor and the blood group antigen. A marked reduction in adherence of bacteria to surface mucous cells was noted (IF) when compared to sections not treated with this monoclonal antibody (e.g. Fig. IA) . Similar reductions in binding were obtained using mouse monoclonal antibodies directed against fucosylated blood group antigens B and Le ^b (data not shown) . Non-immune mouse IgG failed to produce this effect (data not shown) .

Fig. II: A section of human stomach was incubated with FITC-conjugated ϋlex europaeus agglutinin (UEAl) which recognized α-L-fucose. The lectin binds to surface mucous cells.

Fig. 1J: Pretreatment of a section of human stomach with sodium meta-periodate, pH 5.5 at 0°C for 10 min produces no appreciable reduction in binding of UEAl.

Fig. IK and IL: The binding of Helicobacter pylori to pit mucous cells (IK) was also unaffected by the sodium meta- periodate pretreatment (IL) .

Figure 2 shows the in situ assay for binding of Helicobacter pylori to gut epithelial cell lineages.

Fig. 2A: Section of human stomach stained with hematox- ylin and eosin showing the uppermost portion of gastric units in the zymogenic zone. Surface mucous (M) and parietal (P) cells are indicated.

Fig. 2B and 2C: Sections of stomach of an adult human were incubated with FITC-labelled Helicobacter pylori strains WV229 (2B) and M019 (2C) after the bacteria had been grown on blood agar for 5 days under microaerophilic conditions. Strain WV229, which was recovered from a patient with gastric ulcer disease, is associated with surface mucous cells located in the upper pit of gastric unit and their associated luminal surfaces. Strain M019, which was isolated from a healthy carrier, is not bound to any cell lineage.

Fig. 2D and 2E: Incubation of strain WV229 with section of duodenum (2D) and colon (2E) from an adult human. The FITC-labelled bacterial preparation used for staining sections of stomach also stains villus-associated epi¬ thelial cells (including enterocytes) and very weakly colonocytes located in the upper portion of colonic crypts and their surface epithelial cuffs. Less differen- tiated, proliferating and non-proliferating, cells located in small and large intestinal crypts did not contain adherent bacteria. Cell populations located in the mesenchymal compartment were also negative. Incuba-

tion of section of human stomach, intestine and colon with strains NCTC 11637, NCTC 11638 and P466 produced results comparable to those shown in Fig. 2B, 2D and 2E (data not shown) .

Fig. 2F: A section from the zymogenic zone of stomach from an adult Sprague-Dawley rat was incubated with FITC- labelled strain WV299 grown for 5 days on blood agar under microaerophilic conditions. Helicobacter pylori associated with rat surface mucous cells but not with cell lineages associated with the isthmus or glandular domains of rat gastric units.

Figure 3 shows the evaluation of the inhibitory activity of various K -caseins on the attachment of Helicobacter pylori to the human gastric mucosa in situ.

Fig. 3A: Helicobacter pylori strain P466 preincubated with human K -casein (250 μg/ml) prior to overlay on section of human stomach.

Fig. 3B: Helicobacter pylori strain P466 preincubated with bovine colostrum K -casein (1000 μg/ml) prior to overlay on section of human stomach.

Fig. 3C: Pretreatment of human κ-casein with oj-L-fuco- sidase prior to overlay on section of human stomach. Overlays were done on adjacent sections from the same paraffin embedded tissue.

EXAMPLE 2

SDS-PAGE of glycoproteins

The labelling of the bacteria was performed as described in Example 1 and the bacterial strains used were the H. pylori strain P466. The in si tu adherence assay for H. pylori and

the immunohistochemical studies were carried out by the procedures described in Example 1.

SDS-PAGE of human colostrum samples, goat and cow milk, ammonium sulphate precipitated fractions of human colostrum and neoglycoconjugates was performed with 4-20% or 12% pre¬ cast gels (Bio-Rad, Melville, NY.). Proteins were denatured for 3 min at 100°C in SDS-2-mercaptoethanol. Gels were stained with Coomassie brilliant blue and the molecular weight standards were obtained from Sigma, St. Louis, MO.

Iππ-runoblot analysis

Proteins which were separated on SDS-PAGEs were transferred by a semidry Multiphore Nova Blot system onto nitrocellulose for 1.5 hour. Proteins were stained with Ponceau S solution

(Sigma, St. Louis, MO) and blots were then incubated for 1-3 o hours in blocking buffer 2 (BB2) consisting of 1% Tris buffered saline (TBS) , 1% Blocking reagent (Boehringer Mann¬ heim, Indianapolis, IN.), ImM MnCl ₂ , ImM MgCl ₂ , ImM CaCl ₂ , 0.05% polyoxyethylene sorbitan monolaurate, 0.1% NaN ₃ . The following blood group-specific monoclonal antibodies were used: Anti-Le ^a (CBM-LAl) and anti-Le ^b (CBM-LBl) from Immucor, Norcross, GA; anti-Le ^x (630/7H1) , anti-Le ^y (672/7E3) , anti-Le ^b ' ^y (64/4D8) , and anti-H-2 (92FR A2) from Accurat Chemical & Scientific Corporation, Westbury, NY.; anti-type 1 chain (LNT) (K21) , anti-H-1 (17-206), and anti-Le ^b (T218) from Signet Laboratories, Dedham, MA. The following lectins were used: Biotin-labelled ϋlex europaeus type 1, Lotus tetragronoloJus, and Anguilla anguilla recognizing the H-anti- gen (Sigma, St. Louis, MO); digoxigenin (DIG) -labelled Aleu- ria aurantia recognizing branched Fuco!l-6GlcNAc in glycocon- jugates (Boehringer Mannheim) . Blots were incubated with antibodies or lectins in concentrations of 5-10 μg/ml in BB2 for 6 hours followed by 7 washes in TBS. Monoclonal anti¬ bodies were detected by alkaline phosphatase (AP) -conjugated goat anti-mouse antibodies (Boehringer Mannheim) . Digoxigenin or biotin labelled lectins were detected with AP-conjugated

sheep anti-DIG or anti-biotin Fab fragments (Boehringer Mann¬ heim) , respectively. Blots were washed 5x in TBS, once in 0.1M Tris-HCl pH 9.5, 0.1M NaCl, 5 mM MgCl ₂ (APBIII) and developed with BCIP/NBT.

Biotin labelling of lectins

Anguilla anguilla lectin was biotin labelled using a bio- tin-NHS kit from Calbiochem Immunochemicals, La Jolla, CA. The degree of biotinylation (n=7) was determined by the shift in molecular weight on SDS-PAGE.

Size fractiona ion of human secretory IgA-preparation

500 μg of human secretory IgA from pooled human colostrum (Cappel Laboratories, West Chester, PA.) was applied to a Superose 6HR 10/30 eluting with 0.2 ml/min in PBS, pH 6.8. Samples of 400 μl were collected and 20 μl aliquots of every UV-absorbing fraction were run on SDS-PAGE. The fractions of the major peak and the minor peak corresponding to secretory IgA and 120-150 kDa glycoproteins, respectively, were pooled separately and assayed for bacterial adherence inhibition properties (see below) in the in si tu adherence assay.

PNGase F digestion of human secretory IgA-preparation

50 μg of secretory IgA was incubated in 100 μl 0.1M phosphate buffer, pH 8.0, 10 mM j8-MSH, 20 mM EDTA and 6 units of PNGase F (Boehringer Mannheim) (native conditions) or 50 μg of secretory IgA was first incubated in 0.2% SDS, 10 mM jS-MSH for 3 min at 100°C and then diluted with 0.1M phosphate buffer pH, 8.0 to 100 μl (denatured conditions). n-Octylglu- coside (0.5%) was added to the denatured sample to reduce the interference of the PNGase F enzyme with SDS. Finally 6 units of PNGase F were added. Incubations were carried out at 37°C and samples were removed after 0 hour, 1.5 hour, 4 hours, 7 hours, and 18 hours and analyzed on SDS-PAGE and immunoblots.

Preparation of human colostrum and milk samples

Human colostrum samples were from Children's Hospital, St. Louis, MO. Colostrum, goat and cow milk were delipidated and cellular debris was removed by centrifugation at 20 krpm in a Sorvall SS-34 rotor for 30 min before the material was used for inhibition experiments or analysis on SDS-PAGE.

Glycoconjugate inhibition studies

The ability of glycoconjugates to inhibit the bacterial adherence to human stomach in si tu was analyzed by preincu- bation of 200 μl of labelled bacterial suspension in BBl for 1.5 hour at room temperature with the glycoconjugates listed below. The bacteria were washed once, resuspended in 200 μl of BBl, and added to the sections. Adherence was compared to sections incubated with non-inhibited bacteria. The following glycoconjugates were used where (n) represents the number of carbohydrate chains chemically attached to the neoglycopro- tein as determined by the total carbohydrate anthron assay: Secretory IgA, lacto-N-tetraose (LNT) -human serum albumin (HSA) (n=26) , lacto-N-neotetraose (LNnT) HSA (n=26) , lacto- N-fucopentaose I (H-l)-HSA (n=35) , lacto-N-fucopentaose II (Lea) -HSA (n=30) , lacto-N-fucopentaose III (Le ^x ) -HSA (n=30) , lacto-N-difucohexaose I (Le ^b ) -HSA (n=32) , Le ^y -tetrasaccharide (Le ^y ) -HSA (n=26) , lacto-N-neofucopentaose I (H-2) -bovine serum albumin (BSA) (n=29) , GlcNAq8l-4(Fucαl-6)GlcNAc/3-BSA (all from IsoSep AB, Tullinge, Sweden) . The Hl-HSA was also obtained from Accurate Chemical & Scientific Corporation. All oligosaccharides for neoglycoconjugate preparations were purified by HPLC and structurally identified and characte¬ rized and shown by the manufacturer to be more than 95% pure with NMR-spectroscopy. The identity of the neoglycoproteins was verified by immunoblots using monoclonal antibody K21 recognizing LNT (non-fucosylated precursor chain) , anti-H-1 (17-206) (the H-antigen on a lactoseries type 1 chain) , anti-H-2 (92FR A2) (the H-antigen on a lactoseries type 2 chain) , anti-Le ^a (CBM-LAl) (monofucosylated type 1 chain) ,

(anti-Le ^b CBM-LBl) (difucosylated type 1 chain) , anti-Le ^b (T218) (a clone with less cross-reactivity in our hands) , anti-Le ^x (630/7H1) (monofucosylated type 2 chain) , anti-Le ^y (672/7E3) (difucosylated type 2 chain), anti-Le ^b ' ^y (64/4D8) (recognizing difucosylated type 1 and 2 chains) and the H- antigen specific lectins Ulex europaeus type 1 and .Anguilla anguilla.

Oligosaccharide inhibition studies

The ability of glycoconjugates to inhibit bacterial adherence in si tu was analyzed by preincubating labelled bacteria for 3.5 hours at room temperature with the following oligosaccha¬ rides diluted in BBl: Lacto-N-fucopentaose I (H-l) , lacto-N- fucopentaose II (Le ^a ) , lacto-N-difucohexaose I (Le ^b ) , lacto- N-fucopentaose III (Le ^x ) , lacto-difucotetraose, 2'-fucosyl- lactose, 3-fucosyllactose (all from IsoSep AB) , H-disacchari- de (Accurate Chemical & Scientific Corporation) , Fuco-l-4-GlcNAcβO-TMSE (Symbicom AB, Umei, Sweden) , and L-fu¬ cose (Sigma, St. Louis, MO) .

Antibody inhibitions of bacterial binding in the in situ adherence assay

Two-fold serial dilutions of anti-Le ^b and anti-H-2 monoclonal antibodies were applied to separate sections of human gastric mucosa in BBl and incubated at room temperature overnight. Sections were washed 5x in PBS and bound antibodies were detected with FITC-conjugated rabbit antimouse immunoglo- bulins diluted 1:100 in BBl and incubated for 1 hour at room temperature. Antibody dilutions resulting in similar fluore¬ scent intensity were chosen for the bacterial inhibition expe¬ riments (1:2000 and 1:5000, respectively) . Sections were preincubated with the monoclonal antibodies as above, follow¬ ed by bacterial overlay as described above.

Bacterial Western blot overlay analysis

SDS-PAGE was performed as described above, with 20 μg each of unfractionated human colostrum, goat milk, cow milk or 1 μg each of the different neoglycoconjugates. After transfer to nitrocellulose, the filters were incubated with BB2 over¬ night, and washed 2x in TBS. Bacterial suspension (0.1 OD ₆₀₀ ) was added (0.2 ml/cm2 in BB2) and incubated overnight in room temperature in rolling bottles. Filters were washed 6x5 min in TBS and rabbit antiserum was added 1:100 and incubated for 1 hour. Blots were washed 5x in TBS. Bacterial bindings were then incubated with either AP-conjugated antibody diluted l:2000x for l hour, washed 5x in BB2, washed once in APBIII and developed with BCIP/NBT or incubated with gold-labelled antibody, washed 5x in BB2, washed 4x in ddH ₂ 0, and developed with silver enhancement IntenSE BL (Amersham, Arlington Heights, IL.) .

Glycolipid preparations

Immunostainings and bacterial EPTLC overlay analyses

Monoclonal antibodies anti-Le ^a (CBM-LAl) and anti-Le ^b (CBM-LBl) were used to detect the presence of the correspon¬ ding antigens in the glycolipids with secondary antibodies and development as described in Materials and Methods in the section entitled Immunoblot Analysis.

RESULTS AND DISCUSSION

Analysis of fucosylated blood group antigens in human secre¬ tory IgA and serum IgA by lectins and monoclonal antibodies

Immunoblots using the H-antigen recognizing lectins Ulex europaeus type 1, and Anguilla anguilla revealed the presence of H-antigens in the heavy and light chains of both secretory IgA and serum IgA. Aleuria aurantia recognizing branched

Fuco!l-6GlcNAc in glycoconjugates also identified common fuco-

sylated structures. These observations taken together indi¬ cate that terminal Q.-1-2 linked fucose presented in the H-antigen or α-1.6 linked fucose does not by itself consti¬ tute the H. pylori receptor. Monoclonal antibodies against the Lewis blood group antigens Le ^a , Le ^b , Le and Le ^y recog¬ nized the presence of Le ^x and Le ^y in the heavy chains of both IgA types while Le ^b was restricted to the secretory component of secretory IgA. Le ^a could not be found in either of the two IgA types but was detected in a 120-150 kD glycoprotein co- purified with the secretory IgA. This glycoprotein in addi¬ tion presents the Le ^b antigen of equal intensity to the secretory component corresponding to a 10-fold higher Le ^b /- protein ratio.

Size fractionation of secretory IgA preparation

Human secretory IgA was applied to a FPLC-Superose 6 column o and the dominating secretory IgA component was separated from the minor 120-150 kD glycoproteins. The peaks were identified by SDS-PAGE and pooled separately. The purified fractions were then analyzed for H. pylori inhibitory properties. The two components did not mediate adherence inhibitory activity related to their respective protein content but rather reflected the Le ^b content, since 15 μg/ml of secretory IgA was needed for an efficient bacterial inhibition as described previously (PNAS) compared to 2 μg/ml of the 120-150 kD protein.

Analysis of the distribution of the Le ^b antigen among N- and 0-linked carbohydrate chains

The unfractionated secretory IgA preparation was digested by the N-chain releasing enzyme PNGase F in denatured and native conditions. During native conditions the digestion was very limited and only a minor part of the carbohydrate content was removed, while after SDS-denaturation the oligosaccharide chains of the secretory component were readily released as visualized by a decrease of molecular weight in SDS-PAGE and

the absence of Le ^b antigens as detected by immunoblots. Neither the Le ^a nor the Le ^b content of the 120-150 kD glyco¬ protein was, however, affected by the PNGase F treatment indicating the presentation of these blood group antigens in 0-linked carbohydrate chains on this protein.

Inhibition of H. pylori adherence by human colostrum samples

Human delipidated colostrum samples were screened for Lewis blood group activity using Le ^a and Le ^" monoclonal antibodies on SDS-PAGE transferred immunoblots. One Le ^a" Le ^b+ individual (A) expressing low levels of Le ^a and high levels of Le ^b colo¬ strum glycoproteins was found. In addition, one Le ^a+ Le ^b" individual (B) with high Le ^a levels and undetectable Le ^b level was identified. Both these individuals expressed the H-antigen and Fucα.l-6GlcNAc as recognized by Ulex europaeus type 1 and Aleuria aurantia lectins, respectively. Titration of the inhibitory levels of colostrum protein showed that 10 μg/ml of protein of the Le ^a" Le ^b+ colostrum proteins totally abolished bacterial adherence (Fig) . In contrast, 100 μg/ml of the Le ^a+ Le ^b" colostrum proteins resulted in only a minor reduction in bacterial binding.

Inhibition of H. pylori adherence by goat and cow milk antigens

No Le or Le ^b antigen activities could be detected in delipi¬ dated goat and cow milk. Neither of these non-human milk samples exhibited inhibitory activities in concentrations of 100 μg/ml.

Analysis of adherence inhibitory properties of fucosylated neoglycoproteins

The following neoglycoproteins having natural HPLC-purified or synthesized carbohydrate chains characterized by NMR- spectroscopy chemically attached to serum albumin were ob¬ tained: Lacto-N-tetraose (LNT) -HSA, lacto-N-neotetraose

(LNnT) -HSA, lacto-N-fucopentaose I (H-l)-HSA, lacto-N-fuco¬ pentaose II (Le ^a ) -HSA, lacto-N-fucopentaose III (Le ^x ) -HSA, lacto-N-difucohexaose I (Le ^b ) -HSA, Le ^y -tetrasaccharide (Le ^y ) -HSA, lacto-N-neofucopentaose I (H-2)-BSA, GlcNAc/3l-4 (Fucα.l-6)GlcNAc/3-BSA. The identity of the neoglyco¬ proteins were verified by immunoblots using monoclonal anti¬ bodies. The results confirmed a high degree of purity of the oligosaccharide chains of the neoglycoproteins since all antibodies except the anti-Le ^b monoclonal antibodies selec- tively stained the corresponding neoglycoconjugate. The anti-Le ^b (CBM-LBl) clone cross-reacted with the H-l-HSA and to a lesser extend with the Le ^y -HSA. The anti-Le ^b (T218) clone which recognizes the Le ^b -HSA with identical sensitivity did not cross-react with the Le ^y -HSA but recognized the H-l-HSA weakly. Consequently, this indicates that the reacti¬ vity with the Le ^y -HSA is a true cross-reactivity of the CBM-LBl clone. This was also confirmed by information from IsoSep AB since the Le ^y -oligosaccharide was chemically syn¬ thesized and therefore could possibly not be contaminated by Le ^b . The remaining reactivity with the H-l-HSA conjugate could nevertheless be interpreted as a low level of contami¬ nation (less than 5%) of Le ^b -oligosaccharide.

For the analysis of adherence inhibitory properties of the fucosylated neoglycoproteins, bacteria were preincubated as described above with 20 μg/ml of neoglycoconjugates, and the inhibitory activity was compared to 20 μg/ml of the previous¬ ly described secretory IgA. In addition to secretory IgA, the Le ^b neoglycoprotein was the only glycoconjugate that could totally eliminate H. pylori binding at this concentration although a reduction in binding could be observed with the H-1-neoglycoconjugate. Nevertheless, this could be inter¬ preted as a reduction due to low level of Le ^b contamination of the H-l-neoglycoconjugate as discussed previously. Increa¬ sing the concentrations of the glycoconjugates to 100 μg/ml did not result in any inhibition by H-2-BSA, Le ^a -HSA,

Le -HSA, Le ^y -HSA or GlcNAcJl4(Fucα-1-6)GlcNAc/3-BSA (data not shown) . The levels of Leb-glycoconjugate and secretory IgA

which resulted in almost complete reduction of bacterial binding was estimated to 5 μg/ml and 20 μg/ml, respectively, although partial inhibitory effects could be demonstrated at concentrations below 1 μg/ml for the former.

Analysis of adherence inhibitory properties of fucosylated oligosaccharides

For the analyses of bacterial inhibition of the fucosylated oligosaccharides, bacteria were preincubated with series of 0.25 mM, 2.5 mM and 25 mM concentrations of the H-l, Le ^a , Le ^b , Le ^x , and Le ^y oligosaccharides. The H-l, Le ^b , and Le ^y oligosaccharides reduced bacterial binding at a concentration of 2.5 mM and almost eliminated binding at a concentration of 25 mM. Le ^a and Le ^x oligosaccharides did not exhibit binding inhibition at a concentration of 25 mM. The H-disaccharide and 2' -fucosyllactose reduced bacterial binding by 50-75% at a concentration of 20 mM. The Fuco.l-4GlcNAc-?0-TMSE and 3-fu¬ cosyllactose preparations reduced bacterial binding at a concentration of 50 M. The monosaccharide α-L-fucose was inactive even at a concentration of 175 mM.

Western blot overlays of H. pylori to milk proteins

Human colostrum, goat milk, and cow milk were separated on SDS-PAGE and transferred on to nitrocellulose and the filters were incubated with bacteria. Binding to the immobilized glycoproteins was detected to a 90 kD protein in case of human colostrum while the goat and cow milk samples did not show any binding.

Western blot overlays of H. pylori to neoglycoproteins

The series of neoglycoproteins were run on SDS-PAGE and transferred on to nitrocellulose. The filters were incubated with bacteria and binding to the immobilized glycoproteins was detected with gold-conjugated antibodies. The gold-con¬ jugated antibody signal was amplified by silver enhancement.

Specific binding of H. pylori to the Le ^b -HSA and weak binding to the H-l-HSA could be demonstrated.

HPTLC overlays of H. pylori to glycolipids

The panel of glycolipids was chromatographed on HPTLC-plates, incubated with bacteria and binding to the immobilized glyco¬ lipids was detected with AP-conjugated antibodies. No binding of H. pylori could be detected to any glycolipid even after extensive development of the chromatogram. Le ^a and Le ^b antigens were detected in the respective glycolipids by the corresponding monoclonal antibodies.

To do: overlays of mabs and bacteria to all 5 colostrum samples for comparison and to purified secretory IgA and the 50% and 80% AmSo4 fractions (8 samples + Le ^b conjugate and a mw-reference) .

Attachment of Helicobacter pylori to human gastric epithelium using an in situ adherence assay was shown in Example 1 to be inhibited by human colostrum secretory IgA, a molecule carry¬ ing a highly variable set of N- and 0-linked oligosaccharides while serum IgA was devoid of such inhibitory properties. This inhibitory activity of secretory IgA could be markedly reduced by o;-L-fucosidase treatment of the secretory IgA. The efforts to delineate the fucosidase sensitive receptor struc¬ ture for H. pylori have focused on the distribution of fucose residues in the secretory IgA molecule. Immunoblots with monoclonal antibodies and lectins revealed that both IgA types have fucosylated carbohydrate structures such as the H- antigen, branched FUCQ;1-6G1CNAC, and the Le ^x - and Le ^y -blood group antigens in common. These observations indicate that terminal o--l,2-linked fucose as presented in the H-antigen, branched c-1,6-linked fucose, or blood group antigens on type-2 chains (containing Galβl,3GlcNAc) do not by themselves constitute the H. pylori receptor. The Le ^b -blood group anti¬ gen was found to be a carbohydrate structure restricted to the secretory component of secretory IgA. Free secretory com-

ponent has previously been described to present the Le ^x -anti¬ gen and GlcNAc3l-4 (Fucc-1-6)GlcNAc3 (Mizoguchi/Kobata, 1982) and unusual fucosylated carbohydrate structures such as Fucαl-3Fuc and Galβl4 (Fuco!l-6)GlcNAc (Purkayastha/Lamm, 1979) The Le ^a -blood group antigen, however, could not be found in either of the two IgA types, but was detected in a 120-150 kD glycoprotein co-purified with the secretory IgA.

Size-fractionation chromatography of the secretory IgA prepa¬ ration separated the secretory IgA from the 120-150 kD pro- teins. This high molecular weight glycoprotein in addition presents the Le ^b -antigen of equal intensity to the secretory component, corresponding to a 10-fold higher Le ^b to protein ratio.

Subsequent in si tu adherence assays of the separated protein fractions revealed the 120-150 kD protein as an efficient inhibitor of H. pylori binding with an inhibitory titre reflecting the Le ^b -antigen content.

This observation indicated the Le ^b -antigen as the only fuco¬ sylated carbohydrate structure unique for both the secretory IgA molecule, as well as for fractions with H. pylori adhe¬ rence inhibitory activity.

The carbohydrate chains of the secretory component have been reported to be exclusively N-linked. Subsequent analysis of the distribution of the Le ^b antigens of the secretory IgA and the 120-150 kD proteins revealed that these antigens can be presented on both N- and 0-linked carbohydrate chains, since only the Le ^b -presenting carbohydrate chains of the secretory component were released by PNGase F even after SDS-denatura- tion of the glycoprotein for maximum processivity of the glycosidase. This indicates that the Le ^b -blood group antigen can be presented on a multitude of complex carbohydrate chains in glycoproteins, in addition to glycolipids, and variations in receptor presentation can probably affect their

efficiencies as functional receptors in the human gastric epithelium (Gl, PNAS) .

As a natural source of Le ^b -positive and Le ^b -negative glyco¬ proteins, human colostrum samples were analyzed for bacterial adherence inhibitory properties. Colostrum with glycoproteins rich in Le ^a -antigen but devoid of Le ^b -antigen, as demon¬ strated by immunoblots, poorly inhibited bacterial binding while colostrum rich in Le ^b -antigen and low in Le ^a -antigen was a powerful inhibitor of bacterial adherence in the in situ adherence assay. For comparison, pooled goat milk and cow milk were both negative for bacterial inhibition in the in situ adherence assay, correlating to their lack of Le ^a - and Le ^b -antigens, as demonstrated by immunoblots.

In order to analyze the fine-detailed inhibitory properties of glycoproteins carrying defined fucosylated structures, a library of neoglycoproteins was analyzed for their inhibitory properties in the in situ adherence assay. The neoglycoprote- in conjugate exclusively carrying the Le ^b -blood group antigen was found to be the only structure which, at a concentration of 1 μg/ml, could interfere with bacterial adherence, while the Le ^a -, Le ^x - and Le ^y -neoglycoproteins at a concentration of 100 μg/ml did not display any inhibitory activities. This indicates the presence of further fucose residues in addition to terminal fucose presented on a type 1 chain as important properties of a receptor analogue for H. pylori . There could also be an alternative explanation, where the receptor speci¬ ficity for H. pylori would be the intact difucosylated saccharide chain, but since the Le ^y - (difucosylated type 2 chain) is synthesized as a tetrasaccharide, the GlcNAcβl-3Galβl-4Glc sequence is missing compared to an intact type 2 chain, which might affect the binding proper¬ ties in the H. pylori interaction. This extremely high degree of specificity could be dictated by the presentation of the oligosaccharides in the glycoconjugates and in order to investigate this, a panel of free fucosylated oligosacchari¬ des were analyzed for adherence inhibitory properties in the

in si tu adherence assay. Interestingly, H. pylori accepts both the monofucosylated H-l- and the difucosylated Le ^b - and Le ^y - oligosaccharides as efficient inhibitors of bacterial adherence in concentrations of 20 mM. The relaxation in receptor specificity for free oligosaccharides compared to the glycoconjugates indicates that due to steric freedom the importance of the branched fucose residue is diminished. This observation is in analogy to the behaviour of lectins, where the sugar specificities are often less discriminate for free oligosaccharides compared to glycoconjugates. The type 2 chain derived 2-fucosylamine and the Fucαl-2Gal (H) -disac- charide were also efficient, although in slightly higher concentrations. This might point to the preference of type 1 chains compared to type 2 chains and/or further indicate that the length of the carbohydrate chain could be an important parameter as discussed above. The branched fucose residue in the difucosylated Le ^y - might also contribute to the receptor- adhesin interaction somewhat, since the 2fucosyllactose seems slightly less receptor active. The Fucαl-4GlcNAcSθ-TMSE oligosaccharide chain surprisingly inhibits at somewhat higher concentrations, emphasizing the importance of a ter¬ minal fucose residue, although the specific linkage might not be crucial and subsequently the monofucosylated 3-fucosyl¬ lactose (short Le ^x ) with a Fuco;l-3-terminal reduces bacterial binding at high concentrations. This indicates that in free oligosaccharides, the terminal Fuco;l-2Gal-linkage preferable on an full-length type 1 chain is the most efficient struc¬ ture of a receptor analogue for H. pylori , but this configu¬ ration can be minimized to the H-disaccharide while still retaining inhibitory activity.

Since the Le ^b -antigen can be presented in a multitude of glycoconjugates of more or less complex structure, it was of great interest to see whether monoclonal antibodies against the Le ^b -epitope could inhibit bacterial adherence to the human gastric epithelium. As an appropriate negative control, monoclonal antibodies recognizing the H-antigen on type 2 chains were used. By adding similar quantities of monoclonal

antibodies to the sections of gastric epithelium, as detected by FITC-conjugated anti-mouse antibodies, a substantial reduc¬ tion in binding could be seen when the section had been pre¬ incubated with the Le ^b -monoclonal antibody in comparison to the H-2 monoclonal antibodies before bacterial overlay. This indicates that the Le ^b - ntigen could also be the H. pylori receptor structure of the host target tissue in addition to the receptor structure of natural soluble receptor analogues or clearance factors, such as colostrum. The interaction of H. pylori with soluble Le ^b -containing glycoproteins could nevertheless be skewed when it comes to immobilized receptors presented on cell surfaces. In order to investigate the fine-detailed specificity of the interaction with immobilized glycoreceptors, two solid phase assay systems were used: bacterial overlay to neoglycoproteins separated on Western blots and bacterial overlays to glycolipids separated on HPTLC-plates. The bacterial Western blot overlays demon¬ strated a specific interaction with the Le ^b -neoglycoprotein and in addition some weak binding to the H-antigen on type 1 chains (LNF1) .

REFERENCES

Barthel, J. S., Westblom, T. U. , Havey, A. D., Gonzalez, F. and Everett, E. D., Arch . Intern. Med. 148 (1988), 1149-1151

Ebeling et al. Bur. J. Biochem. 4 (1974), 91-97

Evans, D. G., Evans, D. J. Jr., and Graham, D. Y. , Jnfect. Immun . 57 (1989), 2272-2278

Evans, D. G., Evans, D. J. Jr., Moulds, J. J., and Graham, D. Y., Infect. Immun. 56 (1988), 2896-2906

Fauchere, J.-L., and Blaser, M. J., Microb. Pathog. 9 (1990), 427-439

Holmgren, J., Lδnnroth, I., Mdnsson, J.-E., and Svennerholm, L, Proc . Natl . Acad. Sci . USA 72 (1975), 2520-2524

Knibbs, R. N. , Goldstein, I. J., Ratcliffe, R. M. , and Shi¬ buya, N., J. Biol . Chem. 266 (1991), 83-88

Lee, E. R. , Trasler, J., Dwivedi, S., and leBlond, C. P., Am. J. Anat. 164 (1982) , 187-207

Lee, E. Y., Seetharam, B., Alpers, D. H. and DeSchryver- Kecskemeti, K., Gastroenterol. 97 (1989), 1171-1180

Manzi, A. E., Dell, A., Azadi, P., and Varki, A., J. Biol . Chem. 265 (1990), 8094-8107

Mizoguchi, A., Mizuochi, T. , and Kobata, A., J. Biol . Chem. 257(16) (1982), 9612-9621

Ota, H., Katsuyama, T. , Ishii, K. , Nakayama, J. , Shiozawa, T., and Tsukahare, Y., Histochem. J. 23 (1991), 22-28

Purkayastha, S., Rao, C.V.N. , and Lamm, M.E., J. Biol . Chem. 254(14) (1979), 6583-6587

Roth, K. A., Cohn, S. M. , Rubin, D. C, Trahair, J. F., Neutra, M. R. and gordon, J. I, Am. J. Physiol . (Gastroin- test. Liver Physiol . ) 263 (1992), G186-G187

Saitoh, T. , Natomi, H., Zhao, W. , Okuzumi, K., Sugano, K., Iwamort, M. , and Nagai, Y., FEBS Lett . 282 (1991), 385-387

Savage, A. V., D'Arcy, S. M. T. , and Donoghue, C. M. , Bio¬ chem. J. 279 (1991), 95-103

Shibuya, N. , Goldstein, I. J., Broekaert, W. F., Nsimba-

Lubaki, M. , Peeters, B., and Peumans, W. J., J. Biol . Chem. 262 (1987), 1596-1602

Sweetser. D. A., Birkenmeier, E. H. , Hoppe, P. C, McKeel, D. W. , and Gordon, J. I., Genes Devel . 2 (1988), 1318-1332

Warren, J. R. , and Marshall, B., Lancet 1 (1983), 1273-1275 Westblom, T. U. , Madan, E., Subik, M. A., Buriex, D. E., and Midkiff, B. R. , Scand. J. Gastroenterol . 27 (1992), 249-252

Woodward, M. P., Young, W. W Jr., and Bloodgood, R. A., J. Immunol . Meth . 78 (1985), 143-153