Glycans on influenza hemagglutinin affect receptor binding and immune response

Abstract

Recent cases of avian influenza H5N1 and the swine-origin 2009 H1N1 have caused a great concern that a global disaster like the 1918 influenza pandemic may occur again. Viral transmission begins with a critical interaction between hemagglutinin (HA) glycoprotein, which is on the viral coat of influenza, and sialic acid (SA) containing glycans, which are on the host cell surface. To elucidate the role of HA glycosylation in this important interaction, various defined HA glycoforms were prepared, and their binding affinity and specificity were studied by using a synthetic SA microarray. Truncation of the N-glycan structures on HA increased SA binding affinities while decreasing specificity toward disparate SA ligands. The contribution of each monosaccharide and sulfate group within SA ligand structures to HA binding energy was quantitatively dissected. It was found that the sulfate group adds nearly 100-fold (2.04 kcal/mol) in binding energy to fully glycosylated HA, and so does the biantennary glycan to the monoglycosylated HA glycoform. Antibodies raised against HA protein bearing only a single N-linked GlcNAc at each glycosylation site showed better binding affinity and neutralization activity against influenza subtypes than the fully glycosylated HAs elicited. Thus, removal of structurally nonessential glycans on viral surface glycoproteins may be a very effective and general approach for vaccine design against influenza and other human viruses.

Fig. 1.

Schematic overviews and circular dichroism spectra of HAs with different glycosylations. (A) Four variants of HA proteins with different glycosylations: HA_fg, HA [a consensus sequence (ref. 21) expressed in HEK293E cells with the typical complex type N-glycans]; HA_ds, NA-treated HA resulting in removal of sialic acids from HA_fg; HA_hm, HA expressed in GnTI⁻ HEK293S cells with the high-mannose-type N-glycans; and HA_mg, Endo H-treated HA with GlcNAc only at its N-glycosylation sites. Circular dichroism spectra of HA_fg, HA_ds, HA_hm, and HA_mg demonstrate that the secondary structures of the four HA proteins with different glycosylations are similar. (B) Structure representation of HA_fg, HA_ds, HA_hm, and HA_mg with different N-glycans attached at their N-glycosylation sites. The protein structures are created with Protein Data Bank ID code 2FK0 (Viet04 HA), colored in gray, and the N-linked glycans are displayed in green. All N-glycans are modeled by GlyProt (39), and the graphics are generated by PyMOL (www.pymol.org).

Fig. 2.

Glycan microarray analysis of HA with different glycosylations. (A) Glycan microarray profiling of HA variants HA_fg, HA_ds, HA_hm, and HA_mg are shown. The related linkages of glycans were grouped by color, predominantly 17 α2,3 sialosides (yellow) or 7 α2,6 sialosides (blue). The structures of glycans on the array are indicated in Fig. 3. (B) Association constants of HA variants HA_fg, HA_ds, HA_hm, and HA_mg are shown with values of K_A,surf of HA variants in response to α2,3 sialosides 1–15.

Fig. 3.

The binding energy contributions from sugars or modifications of HA–glycan interactions in response to HAs with different glycosylations. These values were obtained by subtraction of ΔG values of the indicated reference glycan (highlighted in red boxes; see Table S3). (A) Glycans 2, 3, 6, and 8–10 possess the same backbone of the disaccharide glycan 1 but only differ in the third sugar from the nonreducing end. The values of ΔΔG are calculated to demonstrate the binding energy difference by changing the third sugars. (B) Glycans 10–12 and 15 possess the same backbone of the disaccharide glycan 8 but differ either by elongating the sugar structure linearly or by adding a branched sugar. (C) Glycans 4 and 5 possess the same backbone of the trisaccharide glycan 3 but differ either by the branched fucose or the sulfate group on the third position from the nonreducing end. (D) Glycans 6 and 7 differ in the sulfate group on the third position from the nonreducing end of glycan 7. (E) Glycans 13 and 14 are α2,3 biantennary sialosides but differ in the change of the internal sugar. (F) Glycans 16 and 17 and α2,6 sialosides (nos. 21–27) show little or no binding to HA.

Fig. 4.

Comparison of HA_fg and HA_mg as vaccine. (A) The bindings between antisera from HA_fg and HA_mg, and various HAs are analyzed by using ELISA. In comparison with HA_fg antiserum, HA_mg antiserum shows better binding to H5 (Vietnam 1194/2004 and CHA5). In addition, the HA_mg antiserum also binds to H1 (California 07/2009 and WSN). (B) Microneutralization of H5N1 (NIBRG-14) virus with HA_fg and HA_mg antisera. In comparison with HA_fg antiserum, HA_mg antiserum shows better neutralizing activity against influenza virus infection to MDCK cells (P < 0.0001). (C) Vaccine protection against lethal-dose challenge of H5N1 virus. BALB/c mice were immunized with two injections of the HA protein vaccine HA_fg, HA_mg, and control PBS. The immunized mice were intranasally challenged with a lethal dose of H5N1 (NIBRG-14) virus. After challenge, the survival was recorded for 14 days.