Final Report Summary - GLYCAN CARTOGRAPHY (Mapping of evolution and adaptation of the influenza virus hemagglutinin protein by glycan array cartography)
Influenza viruses belong to the family of Orthomyxoviridae and cause acute viral disease of the respiratory tract. Influenza virus enters the cell via receptor-mediated endocytosis and subsequent fusion with the cell membrane. Influenza virus hemagglutinin (HA) is responsible for these binding and fusion events. HA binds to sialic acid (SA) receptors on the host cell-surface.
These SA-receptors vary in structures and are species and tissue specific. The preference for a certain SA receptor is influenza virus specific; human and avian influenza viruses prefer (SA)-a2-6-Gal and (SA)-a2-3-Gal terminated structures respectively. Studies on the interactions of glycan binding proteins such as HA have advanced significantly since the development of glycan array technologies. Using this new technology it has been shown that the receptor specificity is more complex than previously recognised and is not solely dependent on the type of glycan sialic-acid linkage, but also involves glycan modifications, such as fucosylation, sulphation, etc.
There is a huge potential in the use of glycan arrays in the understanding of receptor specificity, but a major drawback for this technique is the opacity of the array data, making it difficult to interpret the data and compare data of different arrays. We proposed to set up glycan array cartography, a novel method for the analysis of glycan array data, in clear accessible maps. This provides a spatial layout of assay components (virus strains and SA-receptors), allowing the precise measurement of distances and directions amongst components. To generate 'glycan array maps' MDS algorithms of existing cartography software were adapted to analyse glycan array data. Because viruses are tested against multiple SA-receptors, and SA-receptors tested against multiple viruses, many measurements can be used to determine the position of the virus and SA-receptor points in a glycan array map. The glycan cartography software was tested and optimised using influenza A virus glycan array data publicly available at http://www.functionalglycomics.org(si apre in una nuova finestra).
We have generated glycan array maps for a glycan array dataset of influenza A/H5N1 strains, where the researchers tested recombinant HA, whole virus, different virus concentrations and mutations that change receptor specificity from avian to human-like (Stevens et al J. Mol Bio. 2008). The glycan array maps show clustering of viruses and and SA-receptors. Viruses with human-like receptor specificity revealed clear clustered within 2,6 linked SA-receptors, and Viruses with avian-like receptor specificity clustered with 2,3 linked SA-receptors. These glycan array maps were tested extensively by various statistical methods. Glycan array maps made for various subsets show that relative positions of the viruses remain very similar. Dimensionality testing of the maps revealed that two-dimensional maps can represent the available glycan array data, although a higher dimensionality might be required when using larger datasets. Treating the glycan array data as linear (quantitative) or binary (on/off data) resulted in remarkably similar glycan array maps.
Next a large set of recombinant H5N1 viruses with single, double and triple mutations in the receptor-binding site was tested in glycan arrays. Although the array data revealed clear differences in the receptor specificity of the different H5N1 viruses, however these data were not consistent between separate experiments and the overall signal intensity of the array was low. Although glycan cartography has the potential to reduce variation of the assay, the glycan maps produced with these data were not consistent. To reduce the variation and increase the signal intensity, different incubation protocols were used, none of which resulted in an improvement in quality of the data.
As an alternate approach to determine which SA-receptors are important for binding of the influenza HA we set up a natural glycan array based on SA-receptors isolated from natural recourses, like ferret respiratory tissue. The isolated SA-receptors were coupled with the fluorescent label anthranilic acid followed by fractionation. The labeled SA-receptors were printed on an epoxide-activated chip. Preliminary data has shown that influenza viruses bind to the SA-receptors printed on the array, but that the current concentration of SA-receptors on the array is too low to get a clear binding pattern. Therefore we are currently in the process to increase the concentration of SA-receptors on the array. Because these glycans are obtained from natural sources we can subsequently compare them to virus histochemistry studies. Glycan array data produced with these natural glycan arrays can be further used to make glycan array maps. It is expected that these maps based on natural glycan arrays can be compared to viral genetic, structural, and functional datasets. The use of natural glycan arrays will give us new insights in the SA-receptor variation in the respiratory tract, and about the relevance of the SA-receptors, present on the synthetic glycan arrays, for influenza binding. Glycan array cartography has the potential to revolutionise the interpretation of glycan array data, and hugely increase its utility, by facilitating visualisation of the data, potentially increase its resolution, and allowing quantification of differences in binding patterns between viruses.
These SA-receptors vary in structures and are species and tissue specific. The preference for a certain SA receptor is influenza virus specific; human and avian influenza viruses prefer (SA)-a2-6-Gal and (SA)-a2-3-Gal terminated structures respectively. Studies on the interactions of glycan binding proteins such as HA have advanced significantly since the development of glycan array technologies. Using this new technology it has been shown that the receptor specificity is more complex than previously recognised and is not solely dependent on the type of glycan sialic-acid linkage, but also involves glycan modifications, such as fucosylation, sulphation, etc.
There is a huge potential in the use of glycan arrays in the understanding of receptor specificity, but a major drawback for this technique is the opacity of the array data, making it difficult to interpret the data and compare data of different arrays. We proposed to set up glycan array cartography, a novel method for the analysis of glycan array data, in clear accessible maps. This provides a spatial layout of assay components (virus strains and SA-receptors), allowing the precise measurement of distances and directions amongst components. To generate 'glycan array maps' MDS algorithms of existing cartography software were adapted to analyse glycan array data. Because viruses are tested against multiple SA-receptors, and SA-receptors tested against multiple viruses, many measurements can be used to determine the position of the virus and SA-receptor points in a glycan array map. The glycan cartography software was tested and optimised using influenza A virus glycan array data publicly available at http://www.functionalglycomics.org(si apre in una nuova finestra).
We have generated glycan array maps for a glycan array dataset of influenza A/H5N1 strains, where the researchers tested recombinant HA, whole virus, different virus concentrations and mutations that change receptor specificity from avian to human-like (Stevens et al J. Mol Bio. 2008). The glycan array maps show clustering of viruses and and SA-receptors. Viruses with human-like receptor specificity revealed clear clustered within 2,6 linked SA-receptors, and Viruses with avian-like receptor specificity clustered with 2,3 linked SA-receptors. These glycan array maps were tested extensively by various statistical methods. Glycan array maps made for various subsets show that relative positions of the viruses remain very similar. Dimensionality testing of the maps revealed that two-dimensional maps can represent the available glycan array data, although a higher dimensionality might be required when using larger datasets. Treating the glycan array data as linear (quantitative) or binary (on/off data) resulted in remarkably similar glycan array maps.
Next a large set of recombinant H5N1 viruses with single, double and triple mutations in the receptor-binding site was tested in glycan arrays. Although the array data revealed clear differences in the receptor specificity of the different H5N1 viruses, however these data were not consistent between separate experiments and the overall signal intensity of the array was low. Although glycan cartography has the potential to reduce variation of the assay, the glycan maps produced with these data were not consistent. To reduce the variation and increase the signal intensity, different incubation protocols were used, none of which resulted in an improvement in quality of the data.
As an alternate approach to determine which SA-receptors are important for binding of the influenza HA we set up a natural glycan array based on SA-receptors isolated from natural recourses, like ferret respiratory tissue. The isolated SA-receptors were coupled with the fluorescent label anthranilic acid followed by fractionation. The labeled SA-receptors were printed on an epoxide-activated chip. Preliminary data has shown that influenza viruses bind to the SA-receptors printed on the array, but that the current concentration of SA-receptors on the array is too low to get a clear binding pattern. Therefore we are currently in the process to increase the concentration of SA-receptors on the array. Because these glycans are obtained from natural sources we can subsequently compare them to virus histochemistry studies. Glycan array data produced with these natural glycan arrays can be further used to make glycan array maps. It is expected that these maps based on natural glycan arrays can be compared to viral genetic, structural, and functional datasets. The use of natural glycan arrays will give us new insights in the SA-receptor variation in the respiratory tract, and about the relevance of the SA-receptors, present on the synthetic glycan arrays, for influenza binding. Glycan array cartography has the potential to revolutionise the interpretation of glycan array data, and hugely increase its utility, by facilitating visualisation of the data, potentially increase its resolution, and allowing quantification of differences in binding patterns between viruses.