Human milk oligosaccharide 2′-fucosyllactose guards norovirus histo-blood group antigen co-factor binding site

LETTER

Histo-blood group antigens (HBGAs) are critical polymorphic glycan co-factors required for most norovirus infections (1, 2). However, human milk oligosaccharide (HMO) 2’-fucosyllactose (2’-FL), a component of human breast milk and made up of related glycan saccharides, also binds to norovirus at the HBGA pocket as we have previously discovered (3, 4). We have also shown that 2’-FL weakly inhibits norovirus virus-like particles (VLPs) from attaching to HBGAs for genetically distinct noroviruses GI.1, GII.10, and GII.17 (3, 4).

In this study, we continued our analysis of 2’-FL inhibition potential using consumer-grade 2’-FL powder and chewable tablets. We confirmed that predominant GII.4 norovirus VLPs bound to human A-type saliva (a HBGA source) using a direct ELISA (Fig. 1A) (3, 4). We then showed that both 2’-FL powder and crushed 2’-FL tablets inhibited the GII.4 VLP from binding to A-type saliva with half maximal inhibitory concentration (IC₅₀) of 7.5 mM and 36 mM, respectively (Fig. 1B). This result was comparable to our previous 2’-FL inhibition studies with GI.1, GII.10, and GII.17 VLPs with IC₅₀ values ranging between 11-38 mM (3, 4).

Fig 1.

HMO 2′-FL inhibition was examined using a surrogate HBGA blocking assay. (A) The absorbance was measured at optical density 450 nm (OD₄₅₀). The untreated GII.4 VLPs bound to human A-type saliva in a dose-dependent manner using a direct ELISA. (B) The ELISA inhibition data show 25 μg/mL of GII.4 VLPs mixed with serially diluted 2′-FL powder (light blue bar, ASIN: B08J8DTMGK, Layer Origin, USA) and serially diluted 2′-FL crushed tablet (dark blue bar, ASIN: B09RF982C6, Momstamin, Korea) binding to A-type saliva. The OD₄₅₀ value of the untreated VLPs (at 25 μg/mL) was set as the reference value corresponding to 100% binding, and the percentage of inhibition was calculated as 1 − (treated VLP mean OD₄₅₀/untreated VLP mean OD₄₅₀) × 100. Triplicate experiments were performed with six repeats per 2′-FL concentration (representative shown). The standard deviation is shown as error bars, and the IC₅₀ cutoff is indicated by a dashed line.

To confirm the ELISA inhibition, we co-crystallized GII.4 and GII.10 P domains with the 2′-FL powder and/or 2′-FL tablets and determined and solved the X-ray crystal structures as previously described (Table 1) (4, 5). The 2′-FL is a trisaccharide composed of an α-L-fucose-(1, 2)-β-D-galactose-(1-4)-α-D-glucose. The electron density was well defined for all three saccharides in all three structures (Fig. 2A). The 2′-FL formed a network of hydrophilic and hydrophobic interactions at the dimeric interface (Fig. 2B and C), which were comparable to previous 2′-FL complex structures we solved for GII.10 and GII.17 P domains (3, 4).

TABLE 1.

Data collection and refinement statistics for GII.4-Syd 2′-FL and GII.10 2′-FL complex crystal structures^a

	GII.4-Syd 2′-FL (powder)	GII.10 2′-FL (powder)	GII.10 2′-FL (tablet)
Data collection
Space group	C121	P1211	P1211
Cell dimensions
a, b, c (Å)	98.62, 55.76, 63.61	67.99, 78.70, 70.77	65.41, 78.93, 70.07
α, β, γ (°)	90.00, 120.14, 90.00	90.00, 102.83, 90.00	90.00, 101.13, 90.00
Resolution (Å)	47.02–1.54 (1.56–1.54)	44.63–1.47 (1.49–1.47)	43.55–1.41 (1.43–1.41)
Ramerge	0.060 (0.841)	0.059 (1.150)	0.054 (1.097)
<I/σ(I)>	11.1 (1.1)	9.2 (0.7)	9.0 (0.7)
CC1/2	0.999 (0.601)	0.999 (0.395)	0.998 (0.506)
Completeness	95.3 (86.6)	99.0 (84.1)	99.4 (93.7)
Redundancy	4.8 (4.6)	4.5 (4.3)	4.5 (4.2)
Refinement
Resolution (Å)	41.69–1.54 (1.60–1.54)	39.35–1.47 (1.52–1.47)	43.55–1.41 (1.46–1.41)
No. unique reflections	42,167 (4,114)	122,514 (11,652)	133,809 (13,194)
Rfactor/Rbfree	15.5/18.3 (25.6/30.4)	16.7/19.1 (28.8/30.5)	17.1/19.5 (34.7/35.4)
No. atoms	2,766	5,681	5,444
Protein	2,373	4,792	4,763
Water	340	688	597
Ligand	59	353	154
B-factors (Å2)	19.73	24.43	25.88
Protein	18.42	23.06	24.80
Water	27.97	32.85	33.20
Ligand	25.28	28.26	35.18
RMS bond length (Å)	0.009	0.009	0.009
RMS bond angle (°)	1.06	1.00	1.01
Ramachandran plot statisticsc
Residues	308	625	623
Most favored region	97.71	97.25	97.24
Allowed region	2.29	2.75	2.76
Disallowed region	0.00	0.00	0.00
Clashscore	1.90	4.97	1.99
PDB ID	8Y5V	8Y6C	8Y6D

^a R_merge = [∑_h∑_i|I_h − I_hi|/∑_h∑_iI_hi], where I_h is the mean of I_hi observations of reflection h. Numbers in parentheses represent the highest resolution shell.

^b R_factor and R_free = ∑||F_obs| − |F_calc|| / ∑|F_obs| × 100 for 95% of recorded data (R_factor) or 5% data (R_free).

^c Determined using MolProbity.

Fig 2.

Structures of 2′-FL complexed with GII.4 and GII.10 P domains. (A) Representative simulated annealing difference omit map for the saccharides for GII.4-Syd P domain 2′-FL (powder) complex, GII.10 P domain 2′-FL (powder) complex, and GII.10 P domain 2′-FL (tablet) complex, and labeled fucose (Fuc), galactose (Gal), or glucose (Glc). The omit map (orange) is contoured between 2.5–2.0σ. The 2′-FL is a trisaccharide composed of an α-L-fucose-(1, 2)-β-D-galactose-(1-4)-α-D-glucose. (B) The X-ray crystal structures of GII.4 and GII.10 P domains bound to 2′-FL are shown as cartoons (note: the X-ray structure of GII.4 P domain 2′-FL complex is shown as a dimer). Three 2′-FL molecules from the powder bound per GII.10 P domain dimer, two at the typical HBGA binding sites and one at a non-specific contact indicated with an “*”. One 2′-FL molecule from 2′-FL tablet bound per GII.10 P domain dimer. (C) Close-up of 2′-FL binding interactions with the GII.4 and GII.10 P domains.

Our structural analysis shows that several GII P domain residues (GII.4 numbering N344, R345, G443, T444, and D374) that regularly interact with the fucose moiety of HBGAs also bind the fucose moiety of 2′-FL in a similar manner, whereas the galactose and glucose moieties of 2′-FL interactions with P domain residues are weaker, comparable to previous complex structures (3–7). The structural association of 2′-FL and different HBGA types in complex with GII.4 and GII.10 P domains shows that the fucose moieties were anchored at the same position (Fig. 3A). For GII.4, the bound 2′-FL mimicked the first three saccharides of H2, Lewis-Y, Lewis-X, and Lewis-B, whereas the second and third saccharides of A-type and B-type HBGAs kink up and turn away from this orientation (Fig. 3B). A similar arrangement was observed for GII.10; except that in the bound H2 complex, the second and third saccharides deviated from this orientation (5, 6).

Fig 3.

2′-FL and HBGAs bind at a regular pocket on the GII P domain and share similar orientations. (A) Surface representation and closeup (right side) of the GII.4 P domain dimer (chains A and B, light gray and dark gray, respectively) superimposed with 2′-FL (cyan sticks) and different HBGA types (colored accordingly): A-type (PDB ID: 4WZT, pink), B-type (PDB ID: 4OP7, gray), H2-type (PDB ID: 4WZK, brown), Lewis-Y (PDB ID: 4WZE, magenta), Lewis-X (PDB ID: 4X0C, purple), and Lewis-B (PDB ID: 4OPO, lime). (B) The P domain is removed to show the orientation of 2′-FL with respect to HBGA H2-type, Lewis-Y, Lewis-X, and Lewis-B, A-type, and B-type.

In summary, these new findings show that 2′-FL can inhibit three GII genotypes, including the dominant GII.4 genotype. Interestingly, the FDA and EU consider 2′-FL to be safe, and it was reported that infants and young children could consume 2′-FL at concentrations up to 1.2 g/L/day (8), but its application in a clinical setting remains unclear.

DATA AVAILABILITY

Coordinates and structure factors for all X-ray crystallography structures solved in this study have been deposited to the Protein Data Bank (PDB) under the following accession codes: for GII.4-Syd 2′-FL-powder, 8Y5V; for GII.10 2′-FL-powder, 8Y6C; and for GII.10 2′-FL-tablet, 8Y6D.