Tissue samples and immunohistochemical analysis

Strips of 1 × 10 cm including gastric and duodenal mucosa were resected from 13 kidney donors 5 minutes after death. These human samples were obtained before the law 88-1138 (December 20, 1988) concerning resection of human tissues after death for scientific investigations. Phenotypes of the tissue donors were: ALe(a+b−) 1 individual, BLe(a+b−) 1 individual, OLe(a+b−) 3 individuals, ALe(a−b−) 2 individuals, OLe(a−b−) 1 individual, ALe(a−b+) 2 individuals, BLe(a−b+) 1 individual, ABLe(a−b+) 1 individual, OLe(a−b+) 1 individual. The 3 Le(a−b−) donors were secretors of H in the gastroduodenal junction. Tissue samples were fixed in 95% ethanol for 48 hours, coiled into Swiss rolls, and paraffin embedded. Sections (5 μm) were rehydrated in graded ethanol and washed in phosphate-buffered saline (PBS). Endogenous peroxidase was inhibited by using methanol/H₂O₂ 0.3% for 20 minutes. Sections were then washed in PBS for 5 minutes and covered with PBS/bovine serum albumin (BSA) 1% for 20 minutes at room temperature in a humid atmosphere. After washing in PBS, sections were covered with either the primary antibodies or rNV VLPs diluted in PBS/BSA 1% and left at 4°C overnight. Antibodies were used at 1 μg/mL and VLPs at 5 μg/mL. Sections were then rinsed 3 times with PBS and incubated with biotinylated anti-mouse immunoglobulin (Ig)G (Vector Labs, Burlingame, CA) or rabbit anti-rNV VLPs diluted 1/1000, respectively, for 60 minutes at room temperature. After washing in PBS, the sections were covered with either peroxidase-conjugated avidin (Vector Labs) or with peroxidase-conjugated anti-rabbit IgG (Sigma) for 45 minutes, washed with PBS, and reactions were revealed with 3-amino-9-ethylcarbazol. Counterstaining was performed with Mayer's hemalun. Periodate treatment was performed immediately after the endogenous peroxidase quenching step by incubating sections with 10 mmol/L sodium periodate in 50 mmol/L, pH 5.0, sodium acetate buffer for 30 minutes at room temperature, followed by a 10-minute incubation with 1% glycine. Fucosidase treatment was performed on some sections by incubation at 37°C with 4 mU fucosidase II in 50 mmol/L, pH 5.0, sodium phosphate buffer for a total of 18 hours with a renewal after 6 hours. Controls were performed by using fucosidase, heat-inactivated by prior boiling in a water bath for 10 minutes. Inhibition of rNV VLPs attachment with synthetic oligosaccharides was performed by preincubating the VLPs with the oligosaccharides at a final concentration of 1.8 mmol/L in 1% BSA/PBS for 1 hour at room temperature. Thereafter, the VLPs/oligosaccharide mixtures were incubated overnight at 4°C on tissue-coded sections as described earlier. To assay the ability of anticarbohydrate mAbs to compete with rNV VLPs attachment, coded tissue sections were first incubated with mAbs at 10 μg/mL for 1 hour at room temperature. rNV VLPs were then added and incubated as described earlier. Their binding was detected by using a rabbit anti-rNV VLPs antiserum at 1/1000 in 1% BSA/PBS, followed after washings by peroxidase-labeled anti-rabbit IgG (Sigma). Intensity of staining was semiquantitatively estimated by 3 persons; 2 of them did not know the code of sections.

Binding of rNV VLPs to saliva samples

Saliva samples were boiled for 10 minutes immediately after collection and centrifuged for 5 minutes at 13,000g. The clear supernatant was collected and stored frozen until use. The secretor/nonsecretor phenotypes of the donors were determined by a classic inhibition of agglutination technique. Salivas from 4 unambiguous secretors and 4 unambiguous nonsecretors were chosen for further study. Their status was confirmed by genotyping of the individuals for the FUT2 gene polymorphism in accordance with a published method.29

Attachment of rNV VLPs to saliva was tested by enzyme-linked immunosorbent assay. For this, NUNC (Raskilde, Denmark) immunoplates were coated with saliva samples serially diluted in 100 mmol/L, pH 9.6, carbonate/bicarbonate buffer by an overnight incubation at 37°C in a wet atmosphere. After a blocking step in 3% BSA/PBS for 30 minutes at 37°C, rNV VLPs at 1 μg/mL in 1% BSA/PBS were added and incubated for 1 hour at 37°C. Binding was detected by using the rabbit anti-rNV VLPs antiserum at 1/1000 in 1% BSA, followed by alkaline phosphatase–conjugated anti-rabbit IgG (Sigma), both reagents incubated for 1 hour at 37°C. Reactions were developed by using p-nitrophenyl phosphate (Sigma) as substrate and optical densities read at 405 nm. Between each step, plates were washed 5 times with PBS containing 0.05% Tween 20. The inhibitory potency of synthetic oligosaccharides was tested by preincubating the rNV VLPs with oligosaccharides at 1.8 mmol/L for 1 hour at 37°C before addition to the coated wells.

Cell lines and transfections

The rat FTB complementary DNA (cDNA; Genbank accession number: AF131238) was excised from the pUC18 vector (Pharmacia-Amersham, Les Ulis, France) and inserted into the eukaryotic expression vector pBK-CMV (Stratagene, La Jolla, CA) as previously described.30 This last vector was deleted from the lacZ promoter by digestion with Spe1 and Nhe1 to obtain higher expression of the inserted cDNA, as recommended by the supplier. The rat A enzyme cDNA was cloned by rapid amplification cDNA end polymerase chain reaction (RACE-PCR) into the pUC18 vector from a BDIX rat stomach cDNA library. Primers were deduced from the sequence of a rat exon homologous to the human A gene exon 6. This DNA fragment had been isolated after screening a rat genomic DNA library with a probe derived from the cDNA of the human A gene. The complete coding sequence of the rat A gene (Genbank accession number: AF264018) was inserted into the pDR2 eukaryotic expression vector (Clontech, Palo Alto, CA) deleted of the sequences lying between the EcoRV and Cla I sites.

CHO cells and TS/A cells, Chinese hamster ovary carcinoma, and mouse mammary carcinoma cell lines, respectively, are devoid of α1,2fucosyltransferase activity and therefore of ABH antigens. They were first transfected by using Lipofectamin with the FTB cDNA in the pBK-cytomegalovirus expression vector (Gibco, Paisley, UK) according to the manufacturer's instructions. Twenty-four hours later, fresh medium was added, and 48 hours later, selective medium containing 0.6 mg/mL G418 (Gibco) was added. Transfected cells expressing α1,2-linked fucose residues were selected by flow cytometry using FITC-labeled UEA-I, after cloning by limiting dilutions as previously described.30 For each cell line, a strongly expressing clone was selected and transfected a second time by using the same procedure with the rat A enzyme cDNA in the pDR2 vector. Forty-eight hours later, cells were cultured in selective medium containing 0.6 mg/mL hygromycin. After cloning by limiting dilutions, strongly A antigen–expressing transfectants were selected. Control transfected cells were prepared by transfection with the empty vectors. These stable transfectants were cultured in RPMI 1640, 10% fetal-calf serum, 2 mmol/L L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco) supplemented with 0.25 mg/mL G418 and 0.2 mg/mL hygromycin in the case of double transfectants. Medium for CHO cells culture contained in addition free nucleotides (10 μg/mL).

Caco-2 cells were cultured in Dulbecco's modified Eagle medium supplemented with 20% fetal-calf serum, 5% minimum essential medium, 100 U/mL penicillin, and 100 μg/mL streptomycin. For confocal microscopy studies, cells were seeded at a density of 10,000 cells/cm² on either Transwell filters (NUNC) containing pores of 0.4 μm or on glass slides. Apical and basal media were replaced each day from day 3. Differentiated cells were obtained late after confluency at 21 days.

All cells were maintained at 37°C in a 5% CO₂ humid atmosphere, passaged after dispersal with 0.025% trypsin in 0.02% ethylenediaminetetraacetic acid, and routinely checked for mycoplasma contamination by Hoescht 33258 (Sigma) labeling.

Flow cytometry analysis

The binding of NV capsids to transfected cells was tested by incubating 2 × 10⁵ viable cells with rNV VLPs at 10 μg/mL in PBS containing 0.1% gelatin for 1 hour at 4°C. After 3 washes with the same buffer, a second 45-minute incubation was performed with mAb 9C3 at 1 μg/mL. After washings, a third incubation was performed with FITC-labeled anti-mouse immunoglobulins (Sigma) under the same conditions. The presence of histo-blood group antigens at the cell surface was detected by the same method by using primary antibodies (BG-4, 19-OLE, MBr1, 3-3A) at 1 μg/mL and FITC-labeled anti-mouse immunoglobulins as secondary reagent. After final washings in the same buffer, fluorescence analysis was performed on a FACScan (Becton-Dickinson, Rungis, France) by using the CELLQuest program.

rNV VLPs recognize the H types 1 and 3 antigens

Earlier studies have shown that the expression of ABH and Lewis antigens in the pyloric and duodenal surface epithelia is dependent on the secretor status, whereas that in the pyloric or Brünner's glands is not. In addition, histo-blood group antigens based on type 1 precursor, such as Lewis a or Lewis b, are found exclusively at the level of the surface epithelia, whereas antigens based on type 2 precursor, such as Lewis Y and H type 2, are preferentially found at the glandular level.27 By using a specific mAb, we now show that H type 1 epitopes are present at the level of the surface epithelia from secretor individuals (Figure 1E) irrespective of the ABO and Lewis status and undetectable on digestive epithelial cells from nonsecretors (not shown). In contrast, H type 2 epitopes were readily detected both on the surface epithelia and the glands from secretors as well as on the glands from nonsecretors (Figure 1F). The expression of A and H types 3 and 4 epitopes in the normal gut has previously been reported.31, 32 They could not be detected at the cell surface, but were evidenced at the Golgi level. This was confirmed in the present study (data not shown). Thus, the binding of rNV VLPs to the gastroduodenal junction correlates with the presence of the H type 1 antigen and not with that of H type 2 or H types 3 and 4 antigens.

To confirm that rNV VLP attachment was mediated by recognition of a histo-blood group H-related structure and to define which one, competition experiments were performed with either type-specific anti-H mAbs or free oligosaccharides (Figure 2).

Open in a separate window

Fig. 2

Inhibition of rNV VLPs binding to duodenal tissue sections from secretor individuals by (A–D) anti-H mAbs or by (E–H) synthetic oligosaccharides. Sections were first incubated with the anti-H mAbs and then with VLPs. Binding of the VLPs was detected as described in the Materials and Methods section. (A) No preincubated antibody, (B) preincubated anti-H type 1 BG-4, (C) preincubated anti-H type 2 19-OLE, and (D) preincubated anti-H types 3 and 4 MBr1. Inhibitions with oligosaccharides were performed by preincubating the VLPs with the following trisaccharides: (E) Lewis a, (F) H type 1, (G) H type 2, and (H) H type 3.

Near-complete inhibition of rNV VLPs binding was obtained with the anti-H type 1 mAb as well as with the H type 1 trisaccharide (Figure 2B and F). Both the anti-H types 3 and 4 mAb and the H type 3 trisaccharide gave a partial inhibition (Figure 2D and H). Some degree of inhibition was also obtained with the anti-H type 2 and H type 2 trisaccharide (Figure 2C and G). In contrast, the Lewis a and Lewis b oligosaccharides did not inhibit (Figure 2E and data not shown). The H type 4 trisaccharide was not tested in these inhibition experiments. However, it is unlikely that rNV VLPs attach to the gut epithelial cells via H type 4 because the glycolipids carrying this antigenic determinant were not represented among glycolipids of the small intestine.33

ABH and Lewis antigens are present in saliva depending on the secretor status of the individuals. To get a more quantitative estimate of the qualitative results obtained on tissue sections, saliva samples from individuals phenotyped and genotyped as either secretors or nonsecretors were coated onto enzyme-linked immunosorbent assay plates and the ability of rNV VLPs to bind was assessed. In agreement with the histochemistry experiments, strong binding occurred on saliva from 4 of 4 secretors but not from 4 of 4 nonsecretors (Figure 3A).

Open in a separate window

Fig. 3

(A) Binding of rNV VLPs to serially diluted saliva samples from either secretor (n = 4) or nonsecretor individuals (n = 4) coated to an enzyme-linked immunosorbent assay plate. Three of the secretors were homozygotes, and 1 was heterozygote. Binding was detected by using the polyclonal rabbit anti-NV VLPs as described in the Materials and Methods section. The optical densities at 405 nm are plotted against the log₂ of reciprocal dilutions of saliva. (B) Inhibition of rNV VLPs binding to saliva from a secretor individual by synthetic oligosaccharides. VLPs were preincubated with oligosaccharides at a final concentration of 1.8 mmol/L before incubation on saliva coated at a dilution of 1/4000.

In addition, this binding was clearly inhibited by the H type 1 and H type 3 trisaccharides, whereas the closely related H type 2, type 1 precursor, or Lewis a oligosaccharides showed little or no inhibition (Figure 3B). These inhibitions required concentrations of free oligosaccharides about 4 orders of magnitude higher than the concentration of capsid protein. In addition, a preincubation step between the oligosaccharides and the rNV VLPs was required to obtain inhibition of the rNV VLPs attachment to saliva. These observations suggest that the affinity of the free H type 1 or H type 3 trisaccharides for the recombinant capsid protein is quite low.

To determine if the presence of fucose residues in α1,2 linkage on cellular glycoconjugates was sufficient to mediate rNV VLPs attachment, histo-blood group H determinants were expressed on 2 different cell lines by transfection of the rat FTB α1,2fucosyltransferase cDNA. FTB is the rat homologue of human FUT2 and we recently showed that, like its human counterpart, it catalyzes the transfer of a fucose on either precursor, yet with a clear preference for types 1, 3, and 4 over type 2.34 Neither parental CHO cells nor TS/A cells have detectable α1,2fucosyltransferase activity and, consequently, they are devoid of A, B, or H histo-blood group antigens. However, after transfection with the FTB cDNA, both cell lines express H types 2 and 3 and 4 determinants, but not H type 1 (Figure 4).

Open in a separate window

Fig. 4

Cytofluorimetric analysis of the expression of A and H histo-blood group antigens and of the rNV VLPs attachment to glycosyltransferases-transfected CHO and TS/A cells. Both cell types were transfected with empty vectors (controls), with the rat FTB α1,2fucosyltransferase cDNA (FTB transfectants), or with both the rat FTB α1,2fucosyltransferase and the rat A enzyme cDNAs (FTB + A transfectants). Cells were incubated with either the anti-H type 1 BG-4, the anti-H type 2 19 OLE, the anti-H types 3 and 4 MBr1, the anti-A 3-3A, or with rNV VLPs. The log of fluorescence intensities in arbitrary units is plotted against the cell number.

The lack of H type 1 epitopes on these transfectants is likely caused by an absence of type 1 precursor, as in the case of CHO cells.35 When assayed by flow cytometry, binding of rNV VLPs to these H antigen–expressing CHO and TS/A transfectants was readily detected. In contrast, no binding was evidenced on mock transfected cells (Figure 4). Addition of either an N-acetylgalactosamine or a galactose residue onto H structures generates A or B antigens, respectively, and this is known to hinder recognition by anti-H reagents.36 To get further insight into the specificity of NV capsids, both CHO and TS/A cells were doubly transfected with the FTB and rat A enzyme cDNAs. As expected, the double transfectants expressed the A antigen whereas their recognition by either the anti-H type 2 or the anti-H types 3 and 4 mAbs was severely diminished. The binding of rNV VLPs to these double transfectants also dropped, indicating that the presence of the A histo-blood group N-acetylgalactosamine hinders recognition of the subjacent H antigens by these viral capsids. In addition, this weak attachment of rNV VLPs to A-positive CHO and TS/A cells confirms specificity of the binding to strongly H-positive cells.

The authors thank Dr. J. Bara for the collection of tissues and generous gift of antibodies, Dr. M. Colnaghi for her kind gift of mAb MBr1, Dr. N. Bovin for providing the synthetic oligosaccharides, and Dr. D. Bouhours for the sequence of exon 6 from the rat A gene.