Figure 4.25: Binding of different solHAs to CpKd cells. SolHAs were stained with anti-human-IgG-FITC, nuclei with DAPI. Lectin staining was performed using MAAII-Biotin + Streptavidin-Cy3 and SNA-FITC.
H17Fc binding to enzyme treated MDCKII cells
Immunofluorescence binding assay of H17 with and without NA pretreatment was com-plemented with flow cytometry. Therefore MDCKII cells were treated with NA as de-scribed previously and incubated with solHAs.
In a first experiment almost no difference in fluorescence intensity between NA treated and untreated cells was observed when incubated with H17Fc. H5Fc and H9Fc on the other hand showed reduction of fluorescent signals in NA treated cells compared to untreated cells (fig. 4.26).
Figure 4.26:Flow cytometry of solHA binding to MDCKII cells with and without pretreat-ment with NA fromC. perfringens.
As this indicated, that sialic acid might not be a receptor determinant for H17Fc cells, proteins were pretreated with different substrates to see which would result in a reduc-tion of binding of H17Fc and might therefore be a potential interacreduc-tion partner.
MDCKII cells were pretreated withβ-Galactosidase (cleaves terminalβ-galactose from gangliosides, glycoproteins and glycosaminoglycans), Phospholipidpase C (phosphatidyli-nositol specific, will release GIP anchored proteins), PNGase F (Peptide N-Glycosidase F; cleaves N-glycans) and Pronase (a mixture of proteases fromStreptomyces griseus which results in complete digestion of proteins) and again neuraminidase from Clostrid-ium perfringens. Incubation with these enzymes was performed in either the supplied buffer or in PBS and for 1 h at 37◦C. SolHAs were pretreated with heparin (250µg/ml) for 1 h at 4◦C. After that pretreated or untreated MDCKII cells were incubated with sol-HAs and untreated cells were treated with preincubated or untreated solsol-HAs for 1 h at 4◦C. After this, the normal staining protocol was performed and samples measured in the flow cytometre.
Fig. 4.27 shows the mean values of two experiments. Binding of H9Fc and H17Fc on untreated cells was set to 100%. Beta-galactosidase treatment reduced binding
of both subtypes by 20%. NA pretreatment reduced binding of both H9Fc (15%) and H17Fc (by 25%), phospholipase C did also reduce binding slightly for H9Fc (10%) and a little more for H17Fc (15%). PNGase F digestion of N-glycans reduced binding of both solHAs by 40%. Pronase digestion of cellular proteins reduced binding of both completely, but also drastically changed cell shape. Incubation of the HAs with heparin resulted also in a reduction of 60% for H9Fc and by 50% for H17Fc.
Preincubation of the solHAs with the glycosaminoglycans Chondroitin A and C was also performed once, but did not give a clear reduction of binding in any case. Only incubation with Chondroitin B resulted in loss in binding activity for H9Fc (60%) and less for H17Fc (only by 20%). All of these results should be interpreted with care as enzyme treatment of cells had often a great influence on cell morphology due to the buffers needed for certain enzymes or of the strength of the enzymes.
Figure 4.27: Enzymatic treatment of cells and solHA binding. MDCKII cells were prein-cubated with different enzymes to remove glycan structures on cell surfaces or solHAs were incubated with heparin to inhibit binding.
4.8 Search for potential interactions partners of the influenza HA
To look for cellular interaction partners, Calu-3 cells, that showed the strongest binding, were lysed and 110µgof cell lysates were separated by SDS PAGE and transferred to nitrocellulose membranes. After blocking at least over night, 800 pmol of soluble pro-teins were applied to the blots for 1 h at 4◦C under parafilm. Afterwards bound protein was detected by anti-human-PO and chemiluminescent substrate.
As the combination Calu-3 cells and H9Fc showed the strongest binding the blot was overlayed with H9Fc and FcATG as negative control. FcATG overlay gave some un-specific bands (between 53 kDa and 45 kDa). One clear un-specific band in the H9Fc overlay assay was detected at a size of approximately 125 kDa (fig. 4.28). This could be reproduced three times. A trial to find an interaction partner for H17Fc is shown in fig. 4.29. Here again the 125 kDa band could be reproduced in the H9Fc overlay. In an overlay with H5Fc, which was done on a separate piece of the same blot, no specific band was found as well as for the H17Fc. Here a high molecular band at the top of the blot can be seen. But overlay with H17Fc has not been repeated yet.
4.9 Sequence alignment of HA subtypes
To compare the receptor binding sites (RBS) and the key amino acids (position 190, 225, 226 and 228, H3 numbering) the amino acid sequences of the used HA sub-types were aligned. Sequence alignment was done with PRALINE multiple sequence alignment tool from the Centre for Integrative Bioinformatics, University Amsterdam (http://www.ibi.vu.nl/programs/pralinewww/, Simossis & Heringa [2005]).
Fig. 4.30 shows the alignment of all used subtypes. Herein the position 190 is at po-sition 207 and popo-sitions 225 to 228 corresponds to aa 242 to 245 as this alignment is
Figure 4.28:Overlay assay: Calu-3 cell lysates were separated in SDS-PAGE and blotted on nitrocellulose membrane. This was overlayed with 800 pmol of H9Fc and FcATG and bound solHAs were detected using anti-human-IgG-PO.
Figure 4.29:Overlay assay: Calu-3 cell lysates were separated in SDS-PAGE and blotted on nitrocellulose membrane. This was overlayed with 800 pmol of H9Fc, H5Fc, H17Fc and FcATG and bound solHAs were detected using anti-human-IgG-P
not in H3 numbering. Colors and numbers in the bottom row indicate degree of con-servation, the higher the number, the more conserved is the position among the HAs.
All avian HA subtypes have the same residues at these key aa positions: E190, G225, Q226 and G228. All of which are associated with α2,3-linkage specificity. The H1 sub-types are also shown in a separate alignment in fig. 4.31. H1_2009Fc and H1_1918Fc are identical in the RBS amino acid positions and only differ in position 227 (E or A) within this 220-loop. The amino acids D190, D225, Q226 and G228 all have been shown to confer binding to both linkage types. The porcine H1_WFc on the other hand differs in a 220-loop position of the RBS: it shows G225. This amino acid residue in combination with D190 was shown to confer binding to α2,6-linked sialic acids only.
Thus the only difference in those 4 aa positions between the three H1 subtypes is this D/G in position 225. Altogether they share 89% sequence identity of the full length HA.
The bat H17 subtype has completely different aa at these positions: Q190, A225, H226 and D228. These positions found here in this alignment are identical to those published
by Tong et al. [2012], therefore they are not due a mistake in the alignment.
Figure 4.30: Sequence alignment of the RBS of all used HA subtypes. Red boxes indicate aa positions 190, 225, 226 and 228 (H3 numbering) that are important factors in receptor specificity. Alignment was performed with PRALINE multiple sequence alignment tool from the Centre for Integrative Bioinformatics, University Amsterdam
Figure 4.31:Sequence alignment of the RBS of H1 subtypes. Red boxes indicate aa posi-tions 190, 225, 226 and 228 (H3 numbering) that are important factors in receptor specificity.
Alignment was performed with PRALINE multiple sequence alignment tool from the Centre for Integrative Bioinformatics, University Amsterdam
The influenza HA is the viral protein responsible for receptor recognition, binding and membrane fusion. Receptor determinants for influenza viruses are sialic acids on gly-cans of glycoproteins and glycolipids. The type of linkage between the terminal sialic acid and the neighboring galactose plays an important role for the receptor recogni-tion of influenza A viruses. Linkage in α2,3 conformation is typically detected by avian HA subtypes and α2,6-linkage specificity can be found mainly in human strains. This notion resulted from early studies investigating the sialic acid distribution of different tissues or by hemagglutination tests using erythrocytes that have been resialylated to contain only a single linkage type. To investigate sialic acid expression on cells and tis-sues two plant lectins are used: MAA, which detects α2,3-linked sialic acids and SNA, which binds to α2,6-linked ones. Binding of those lectins is not only dependent on the linkage type but is also affected by glycan density on the cell surface and the glycan chain following the galactose. The huge diversity of glycan structures on cell surfaces and the variety of hemagglutinin subtypes makes it unlikely that two plant lectins are sufficient to characterize receptor determinants of influenza viruses. To circumvent this problem the aim of this thesis was to generate soluble HAs and to use them for detection of receptor type sialic acids.
101
5.1 Expression of soluble HAs
Soluble proteins have been used for the investigation of receptor binding of other viruses before (Klein et al. [1994], Paul [2008]) and also for IAV (Ayora-Talavera et al. [2009], de Vries et al. [2011], Shelton et al. [2011]).
In this study mainly chimeric proteins with a human Fc-tag were used which has the advantage of simple purification via FPLC using protein A columns. Furthermore, the Fc-tag allows detection of the solHAs using anti-human IgG antibodies. Fc-tagged in-fluenza HAs were successfully used by others (Shelton et al. [2011], J.-l. Yang et al.
[2010]) that could show correct folding and biological activity of the soluble forms of influenza HA.
In this study, transfection of the HEK293T cells was achieved by calcium-phosphate precipitation. Transfection efficiency of this method can be up to 70% when buffers and cells are optimal (personal observation). Supernatants containing the soluble HAs were collected over several days. By FPLC purification, the proteins were concentrated and the result was a protein solution containing only the solHAs. Purified solHAs were subjected to SDS PAGE and Western blot. Fc-tagged solHAs appeared in WB as pre-sumptive dimers, which is supported by the Fc-tag but also monomeric and multimeric forms were observed. Computational calculation of the molecular weight (about 86 kDa per Fc-tagged monomer) and appearance of bands in Western blots indicated that the dimers are probably highly glycosylated (fig.4.1) as the molecular weight of detected bands in WB is higher than that of a presumptive dimer.
SolHAs carrying a trimerization domain and a his-tag showed a large band at a size above 200 kDa, with a calculated MW of the monomer of about 65 kDa, these are most likely trimers (fig.4.3).
5.2 Binding of solHAs to different cell lines
To investigate binding properties of solHAs and to compare them to classic lectin stain-ing a set of different cell lines was chosen.
H7T6his and H9T6his were the only solHAs connected to trimerization domains that were analysed for their binding abilities. His-tagged solHAs of other HA subtypes were cloned but not further characterized. H7T6his and H9T6his bound like their Fc-tagged counterparts to MDCKII, Calu-3 and HBE cells. Binding to NPTr, CLEC213 and pCKC was performed but due to strong unspecific binding of the anti-his antibody no conclu-sions could be drawn. Anti-his antibodies from different species were also tested but showed more or less strong background staining. This problem might be solved by preincubation of the antibody with control cells. His-tagged solHAs of H1_2009 and H1_1918 bound to MDCKII cells (Erdt [2012]) but have not yet been further tested.
Cloning and analysis of the remaining subtypes will be finished as soon as possible.
The Fc-tagged versions of solHAs were preferred, as they are easier to purify and detection by antibodies worked reliably. Using always the same antibodies allows to compare the binding efficiency. Thus, the Fc-tagged versions of solHA are preferred over the his-tagged counterparts for the analysis of the sialic acid binding activity of influenza viruses. H7Fc and H9Fc were both tested at different concentrations. H9Fc bound more efficiently to almost all tested cell lines (section 4.2.1). An amount of 50 pmol was sufficient for optimal binding; H7Fc on the other hand required 100 pmol to get satisfying results. It did not bind at all to human A549 and to porcine NPTr cells.
Those cells also showed more SNA staining, so express moreα2,6-linked sialic acids.
H9Fc bound well to both cell types, although the amino acid composition of the RBS is for "avian"-like receptors in α2,3-linkage. But glycan array analysis also showed bind-ing of H9Fc toα2,6-linked sialic acids. H7Fc on the other hand exhibited a much lower affinity forα2,6-linked glycans in the glycan array analysis (section 4.6). Low affinity for α2,6-linked sialic acids compared to α2,3-linked Sias was also shown by A. S.
Gam-baryan et al. [2012] for many H7 virus strains. For H9N2 strains it was shown, that binding to α2,6-linked sialic acid was increased when they possess a Q226L substi-tution (M. Matrosovich et al. [2001], Choi et al. [2004]) but infections of humans has also occurred with strains possessing the "avian"-glutamine (Butt et al. [2010]) which is also present in H9Fc (see fig.4.30). Binding of H9Fc to cells that are not recognized by H7Fc can result 1) from the different binding properties of the HAs: H9 can bind the α2,6-linked sialic acids on these cells whereas H7Fc can not and/or 2) those cells express special α2,3-linked sialic acids structures that can be bound by H9Fc but not by H7Fc and MAAII. Human Calu-3 cells display about the same amount of MAAII staining as A549 cells but nevertheless H7Fc bound to Calu-3 but not to A549 cells.
In the aforementioned publication of A. S. Gambaryan et al. [2012] it has been shown that most H7 HAs preferentially bind to sulfated receptor sequences that were also pre-ferred by H7Fc but not by MAAII in the glycan array analysis (section 4.6). Maybe such a specific structure (or others bound by H7Fc) is missing on A549 cells but present on Calu-3 cells.
H5Fc showed strong binding to most cell lines. It even bound to A549 and NPTr cells.
The H5 strain used in this thesis was isolated from turkey and possesses the same RBS key amino acids as H7 and H9. Binding to A549 was again weakest but the hu-man Calu-3 cells display a lot of H5Fc binding. Binding of H5N1 virus to the epithelium of the human lower respiratory tract has been shown by van Riel et al. [2006]. Ayora-Talavera et al. [2009] and A. Gambaryan et al. [2006] have shown that different H5 viruses do not bind to human receptor homologues and that a S227N mutation within the HA leads to moderate binding to this synthetic structures. Nevertheless Ayora-Talavera et al. [2009] showed binding to human tracheal epithelium. A glycan array of the H5Fc will reveal the receptor binding profile of this specific H5. Although others (e.g.
A. Gambaryan et al. [2006]) have reported a utilization of the same sulfated structure that H7 viruses prefer (A. S. Gambaryan et al. [2012]) and although key amino acids in
the RBS are identical between H7Fc and H5Fc both solHAs bound differentially to the cells analysed in this thesis: e.g. H5Fc bound to NPTr cells whereas H7Fc did not.
Interestingly, all avian HA subtypes bound not extremely well to pCKC cells, although they express both sialic acid linkage types according to lectin staining. Probably the
"right" glycans are not expressed abundantly on these cells. All three avian subtypes used in this thesis are from viruses known to be occasionally transmitted to humans. At least on permanent cell lines binding to cells of different species (dog, human, chicken and pig) is possible without a mutation of the RBS to a "humanized" sequence. These HAs already appear to possess specificity to receptors that are present in a variety of species indicating that expression of sialic acid linkage types is not as restricted to certain species as believed.
The three H1 subtypes differ in their RBS in one amino acid position (D225G) which is known to result inα2,6-linked sialic acids specificity. Both H1_2009Fc and H1_1918Fc should detect both sialic acid linkage types (aa D190 and D225). Other amino acids around the RBS are identical in H1_2009Fc and H1_WFc or in H1_1918Fc and H1_WFc.
The H1 HAs in this thesis also differ in positions 200 and 227 outside the RBS: the new pandemic H1_2009Fc carries a threonine or glutamic acid at these positions whereas H1_1918Fc and H1_WFc have alanine at both positions. This is associated with in-creased binding to α2,6-linked sialic acids in glycan array analysis of soluble H1 pro-teins by de Vries et al. [2011].
In general all H1 solHAs bound to all cell lines, though with different efficiencies. Again the weakest binding was observed with pCKC. The pandemic H1_2009Fc showed the strongest binding in relation to the other two H1 subtypes. Especially on NPTr cells only H1_2009Fc binding was judged as being "strong". H1_1918Fc bound weaker than the 2009 pandemic H1, only on Calu-3 cells binding was similar in intensity. The porcine H1 bound relatively weak to all cells (strongest again to Calu-3 cells). This could be due to the predicted restriction toα2,6-linked sialic acids of this strain (see section 4.9).
That H1_WFc still bound to CLEC213 cells, that showed no SNA staining at all, is ei-ther caused by the presence ofα2,6-linked sialic acids that are not detected by SNA or because receptor specificity is not absolute and alsoα2,3-linked sialic acids can serve as receptor. Especially H1_2009Fc did not appear to make a difference between cells showing exclusively MAAII or SNA staining (CLEC213 or NPTr respectively). Again two reasons are possible 1) lectin staining did not detect the receptor type sialic acid or 2) in this model affinity of H1_2009Fc to both sialic acid linkage types is equally strong.
Differences in receptor binding between H1_1918Fc/H1_WFc and H1_2009Fc may be the result of amino acid substitutions T200A/E227A. A further difference in the RBS of H1_WFc (D225G) may also be the reason for the differences in binding compared to H1_2009Fc and H1_1918Fc. These slight differences within the HA appear to have an impact on receptor binding that can be analysed using solHAs. Differences in re-ceptor binding between porcine H1 viruses and human isolates have also been shown by glycan array analysis (X. Chen & Varki [2011]) and further explain the differences observed here. Childs et al. [2009] could show that 2009 pandemic virus strains and classical swine H1 viruses both detectα2,3-linked sialic acids in their glycan array.
More recent studies focus on the conformation of the RBS rather than on the primary structure of the HA and the terminal sugars of glycan chains. Amino acids E190 and Q226 that are involved in α2,3-linkage specificity form a RBS that can interact with cone-like glycan topologies of α2,3-terminated glycans and short α2,6-linked ones.
The other possible glycan topology (umbrella like as it spans a wider region in the HA RBS) is mainly found in long α2,6-terminated glycans. Full adaptation to the human respiratory tract is accomplished by binding to sialoglycans in an umbrella-like topol-ogy for which (among others) position D225 is required (Chandrasekaran et al. [2008], Viswanathan et al. [2010]). HAs with the "avian" RBS probably are able to interact with bothα2,3- andα2,6-linked sialic acids as long as they are in a cone-like topology.
Binding of HAs to either sialic acid linkage type appears not to be exclusive, which is
supported by the glycan array analysis of H9Fc and H7Fc. Overall glycan topology and sugars beyond the galactose are involved in receptor binding and glycans detected by the two plant lectins are not the same as those, to which the HAs have the highest affinity.
5.2.1 Neuraminidase pretreatment
Hemagglutination of erythrocytes can be abolished when the sialic acids are removed by neuraminidase treatment. Correspondingly removal of sialic acids from cells de-creases virus binding and infection (e.g. Bohm [2010]).
To determine if the binding of the artificial solHAs is sialic acid-dependent, cells were pretreated with NA from Clostridium perfringens. Binding of H7Fc, H5Fc and H9Fc
To determine if the binding of the artificial solHAs is sialic acid-dependent, cells were pretreated with NA from Clostridium perfringens. Binding of H7Fc, H5Fc and H9Fc