Bile-salt-stimulated lipase and mucins from milk of 'secretor' mothers inhibit the binding of Norwalk virus capsids to their carbohydrate ligands

Abstract

Breast-feeding-associated protection against calicivirus diarrhoea is associated with the presence of high levels of 2-linked oligosaccharides in mother's milk, and human calicivirus strains including the NV (Norwalk virus) use gut 2-linked fucosylated glycans as receptors, suggesting the presence of decoy receptors in milk. Our aim was to analyse the ability of human milk to inhibit the attachment of rNV VLPs (recombinant NV-like particles) to their carbohydrate ligands and to characterize potential inhibitors found in milk. Milk from women with the secretor phenotype was strongly inhibitory, unlike milk from women that are non-secretors, which is devoid of 2-linked fucosylated structures. At least two fractions in human milk acted as inhibitors for the NV capsid attachment. The first fraction corresponded to BSSL (bile-salt-stimulated lipase) and the second to associated mucins MUC1 and MUC4. These proteins present tandem repeat O-glycosylated sequences that should act as decoy receptors for the NV, depending on the combined mother/child secretor status.

Figure 1. FUT2-dependent molecules from human milk inhibit rNV VLP attachment to H type 1 histo-blood-group antigen

(A) Milk samples from women genotyped for the FUT2 locus were coated on to ELISA plates at a dilution of 1:1000 and the presence of H type 1-reactive molecules was determined by reactivity of the anti-H type 1/Le^b mAb, LM137. A total of five women were considered as non-secretors (2, 4, 5, 6 and 8) and the remaining as secretors. SE=active FUT2 alelle; se=inactive FUT2 allele. (B) Inhibition of rNV VLP attachment to polyacrylamide-conjugated H type 1 by milk samples from secretor and non-secretor women diluted to 1:10. (C) Examples of the inhibitory potency of milk samples, from 2 secretors (●, ▲) and a non-secretor (□), on rNV VLP attachment to saliva from an O type secretor individual. The inhibition assay was performed as described in the Materials and methods section. The percentage of inhibition is shown as a function of the reciprocal milk dilution.

Figure 2. Binding of rNV VLPs to human gastroduodenal tissue and inhibition by human milk

Tissue sections from the gastroduodenal junction of a secretor (A, C and D) and a non-secretor (B) were incubated with rNV VLPs and the binding was detected as described in the Materials and methods section. In (C, D) VLPs were co-incubated with milk samples from secretor and non-secretor women respectively, at a dilution of 1:10. All pictures taken are from the pyloric area. A complete absence of VLP binding was observed in the presence of milk from a secretor, as in the tissue from a non-secretor.

Figure 3. Inhibition of rNV VLP attachment to saliva from a secretor individual by milk fractions from a secretor donor

A fresh secretor milk-sample was centrifuged at 1500gfor 20 min to separate the cream from the milk. The skimmed milk was dialysed to remove free oligosaccharides and the cream was washed by centrifugation in PBS to remove soluble proteins. Washed cream was resuspended in PBS and sonicated to disrupt fat globules. The human fat globule membranes (HFGM) were pelleted by centrifugation at 100000 g for 1 h and separated from the fat layer. Each fraction corresponding to the original milk dilution at 1:10 was co-incubated with rNV VLPs in the inhibition assay as described above.

Figure 4. Detection of human milk rNV VLP ligands by Western blotting

(A) Skimmed milk samples were run on SDS/PAGE (5% gel under reducing conditions and probed with rNV VLPs; lane 1, secretor sample number 3; lane 2, non-secretor sample number 4. (B) Skimmed milk from a secretor (sample A) was run on an SDS/PAGE 5–10% gradient gel under reducing conditions and probed either with the anti-H type 1/Le^b mAb, LM137 (lane 1) or with rNV VLPs (lane 2). Positions of the molecular mass markers are shown by arrows.

Figure 5. Purification of BSSL from the milk of secretor woman A

(A) Total proteins were loaded on to an immobilized anti-BSDL pAbL64 (polyclonal antibody L64) column equilibrated with PBS and incubated overnight at 4 °C. Unbound fractions were eluted and the column was washed with 0.02 M Tris/HCl and 0.5 M NaCl, pH 7.4 until zero absorbance. Bound fractions were eluted with 0.1 M glycine, pH 2.5. Elution of material was recorded by the absorbance at 280 nm (■). Enzyme activity (◇) was determined with 4-nitrophenyl hexanoate as substrate. (B) Coomassie Blue stained SDS/PAGE gels and (C, D) Western blotting of fractions eluted from the immobilized pAbL64 column, (lane 1, 1 μg of starting material), (lane 2, 1 μg of unbound material), (lane 3, 1 μg of bound material). Immunodetection was performed with mAb pAbL64 (C) or rNV VLPs (D).

Figure 6. Binding of rNV VLPs to purified BSDL and BSSL

(A) Purified glycoproteins (1, 5 or 10 μg) from secretor donors (BSDL from the pancreatic juice of an O type secretor patient and BSSL from mother number 9) or from a non-secretor milk donor (mother number 4) were coated on to ELISA plates and the binding of rNV VLPs was tested. (B) Inhibition of rNV VLP attachment to the saliva of a secretor individual by purified BSSL from secretor A.

Figure 7. Binding of rNV VLPs to human milk mucins

Skimmed milk samples from non-secretor (lane 1) or secretor (lane 2) donors and the mucin fraction from milk of a secretor donor prepared as described in the Materials and methods section (lane 3) were run on SDS/PAGE (5%) gel with a 4% stacking gel. After transfer, membranes were probed with either rNV VLPs, an anti-MUC1 mAb or an anti-MUC4 mAb. The part of the gels shown is above the 200 kDa marker and the arrow indicates the limit of the stacking gel.