Trichodysplasia spinulosa-Associated Polyomavirus Uses a Displaced Binding Site on VP1 to Engage Sialylated Glycolipids

Abstract

Trichodysplasia spinulosa-associated Polyomavirus (TSPyV) was isolated from a patient suffering from trichodysplasia spinulosa, a skin disease that can appear in severely immunocompromised patients. While TSPyV is one of the five members of the polyomavirus family that are directly linked to a human disease, details about molecular recognition events, the viral entry pathway, and intracellular trafficking events during TSPyV infection remain unknown. Here we have used a structure-function approach to shed light on the first steps of TSPyV infection. We established by cell binding and pseudovirus infection studies that TSPyV interacts with sialic acids during attachment and/or entry. Subsequently, we solved high-resolution X-ray structures of the major capsid protein VP1 of TSPyV in complex with three different glycans, the branched GM1 glycan, and the linear trisaccharides α2,3- and α2,6-sialyllactose. The terminal sialic acid of all three glycans is engaged in a unique binding site on TSPyV VP1, which is positioned about 18 Å from established sialic acid binding sites of other polyomaviruses. Structure-based mutagenesis of sialic acid-binding residues leads to reduction in cell attachment and pseudovirus infection, demonstrating the physiological relevance of the TSPyV VP1-glycan interaction. Furthermore, treatments of cells with inhibitors of N-, O-linked glycosylation, and glycosphingolipid synthesis suggest that glycolipids play an important role during TSPyV infection. Our findings elucidate the first molecular recognition events of cellular infection with TSPyV and demonstrate that receptor recognition by polyomaviruses is highly variable not only in interactions with sialic acid itself, but also in the location of the binding site.

Fig 1. Cell surface glycans featuring terminal sialic acids promote TSPyV attachment and pseudovirus infection.

(A) Binding of Alexa Fluor 488-conjugated TSPyV VP1 pentamers to mock (PBS) or with 1 U/ml Clostridium perfringens type V neuraminidase pre-treated cells was analysed by flow cytometry. 30,000 gated events were measured for each sample. The average from three independent experiments is shown, and error bars indicate the standard deviation. Data was standardized to signals of mock treated cells alone. S1 Fig shows non-standardized data of individual experimental replicates. (B) Attachment of Alexa Fluor 488-conjugated TSPsV to mock (PBS) or with 1 U/ml Clostridium perfringens type V neuraminidase pre-treated HEK293 and SVGA cells. The analysis was carried out by flow cytometry with 10,000 events measured for each sample. The average from three replicates is shown, and error bars indicate the standard deviation. (C-D) TSPyV pseudovirus (TSPsV) infections of HEK293 and SVGA cells (mock-treated cells and cells incubated with Clostridium perfringens neuraminidase for 30 min prior to PsV infection). BK Polyomavirus pseudovirus (BKPsV) was used as positive control for a neuraminidase-sensitive infection. The efficiency of the PsV infection was measured by transduction of the reporter plasmid phGluc coding for a secreted form of Gaussia luciferase 72 h post infection by detection of the secreted luciferase (measured in relative light units, RLUs using the BioLux Gaussia Luciferase Assay Kit). PsV experiments were done in quadruplicate and repeated three times and the average relative transduction is shown. Error bars indicate standard deviations and statistic analysis was performed using the two-tailed unpaired t test. The data was standardized to mock PsV transfections. The mock PsV control was generated by harvesting HEK293TT cells transfected with control plasmid instead of the capsid expression plasmids and then purifying according to the PsV purification protocol to measure background signal and non-specific transfer of luciferase to infected cells.

Fig 2. TSPyV VP1 specifically binds to terminal Neu5Ac.

(A) Crystal structure of the unassembled TSPyV VP1 pentamer in complex with the GM1 glycan. The TSPyV VP1 pentamer is shown in cartoon representation with one monomer depicted in purple. Glycan residues are shown in stick representations and glycan atoms are coloured according to the atom type, with oxygen in red, nitrogen in blue, and carbon in orange. (B) Close-up view of the Neu5Ac binding site. The simulated annealing omit difference map contoured at 3.0 σ is shown with a radius of 2.0 Å around glycan residues Neu5Ac-α2,3-Gal-β1,4GalNAc of the GM1 glycan. (C) Schematic representation of the GM1 glycan. Glycan rings built into electron density in panel B are highlighted in colours. (D) Interaction between TSPyV VP1 and the terminal Neu5Ac residue of the GM1 glycan in a binding site that is formed by the BC2-loop. Side chains and backbone atoms of VP1 interacting with the glycan are shown in stick representation, and atoms are coloured as in panel A. Water molecules are shown as spheres, hydrogen bonds between VP1 and Neu5Ac are drawn as black dashed lines, water-mediated contacts between VP1 and Neu5Ac are depicted in cyan, and intramolecular Neu5Ac hydrogen bonds are shown in orange.

Fig 3. Cell binding and PsV infection of TSPyV mutants.

(A) Cell binding analysis of TSPyV wild type (WT) and mutant VP1 pentamers to HEK293 and SVGA cells by flow cytometry. VP1 pentamers were covalently thio-labelled using Alexa Fluor 488 C5 maleimide. The histogram depicts the relative fluorescence signal of mutants compared to the binding signal of wild type pentamers. Data was measured in three independent experiments and was standardized for the background signal from cells alone. Error bars indicate standard deviations. (B) WT and mutant TSPsV binding (Alexa Fluor 488 amine-conjugated) to HEK293 and SVGA cells (mock-treated cells and cells incubated with Clostridium perfringens neuraminidase for 30 min prior to PsV infection) examined by flow cytometry. Relative fluorescence signals of mutants compared to the binding signal of WT PsV were plotted and error bars indicate standard deviations for three experimental replicates. Labeled PsV content was normalized by western blotting detection with purified PAB597, a hybridoma supernatant that produces a monoclonal antibody against JCPyV VP1, which cross-reacts with TSPyV VP1. The two-tailed unpaired t test was performed for binding of WT and mutant TSPsVs to mock-treated cells. (C) TSPsV infection of HEK293 and SVGA cells assayed 72 h post infection by detection of the secreted luciferase due to transduction of the reporter plasmid phGluc. The luciferase was quantified using a BioLux Gaussia Luciferase Assay Kit. The secreted luciferase catalyzes a photo-oxidation that emits light, which is measured in relative light units, RLUs. RLUs are given in logarithmic scale. Mock PsV infections were done with a control sample obtained according the PsV purification protocol from cells only transfected with phGluc and control plasmid measure background and nonspecific transfer of luciferase. PsV experiments were done in quadruplicate and repeated two times. Statistic analysis was performed using the two-tailed unpaired t test.

Fig 4. Unique location of the TSPyV Neu5Ac binding site.

The location of the Neu5Ac binding site on TSPyV VP1 is shifted compared to binding sites of polyomaviruses SV40, BKPyV, JCPyV, MCPy, LPyV, HPyV9 and MPyV, which all engage sialylated glycans by employing a Neu5Ac binding site in a conserved location on VP1. For clarity, only terminal Neu5Ac residues are shown in stick representations and are coloured according as assigned for each virus. TSPyV VP1-GM1 glycan (purple), SV40 VP1-GM1 glycan (blue; pdb 3BWR), BKPyV VP1-GD3 glycan (light green; pdb 4MJ0), JCPyV VP1-LSTc pentasaccharide (pink; pdb 3NXD), MCPyV VP1-GD1a glycan (cyan; pdb 4FMJ), LPyV-α2,3-sialyllactose (dark green; pdb 4MBY), HPyV9 VP1-α2,3-Neu5Gc-sialyllactose (brown; pdb 4POT), MPyV VP1-Neu5Ac-α2,3-Gal-β1,3-[α2,6-Neu5Ac]–GlcNAc-β1,3-Gal-β1,4-Glc (orange; pdb 1VPS).

Fig 5. Comparison of the Neu5Ac binding sites of HI- loop and BC2- loop regions.

(A, C) HI-loop regions of TSPyV VP1 and of BKPyV VP1 in complex with the GD3 glycan (pdb 4MJ0), are shown in the same orientation. The terminal Neu5Ac of GD3 is show in orange sticks. Residues that do not allow interaction with Neu5Ac in a manner similar to BKPyV are colored on the surface of VP1. Direct hydrogen bonds between BKPyV VP1 side chain residues and Neu5Ac are depicted as dashed lines. (B, D) BC2-loop regions of TSPyV VP1 and BKPyV VP1 are shown. Direct hydrogen bonds between side chain residues of TSPyV VP1 and the terminal Neu5Ac of the GM1 glycan and an intramolecular BKPyV VP1 hydrogen bond are shown as black dashed lines. The structures were aligned using their Cα atoms. One VP1 monomer is highlighted in purple for TSPyV and light green for BKPyV, respectively.

Fig 6. TSPyV binds to sialic acid containing glycolipids.

Cell binding analysis of thio-labelled Alexa Fluor 488 TSPyV VP1 pentamers to HEK293 cells treated with inhibitors interfering with the synthesis of glycosphingolipids, N. and O-linked glycoproteins. Cells were treated with (A, D) PPMP, (B, E) tunicamycin, (C, F) BenzylGalNAc, and the respective carrier controls. The binding of TSPyV pentamers was standardized to binding signals to mock (PBS) treated cells. 30,000 gated events were measured for each sample. Cell binding of cholera toxin B subunit and the lectins ConA and MALII was used to monitor the effect of the inhibitors on the cellular expression of glycans. Data was measured in three independent experiments. Error bars indicate standard deviations.

Fig 7. Glycolipids are important for TSPsV infection.

TSPsV infection of HEK293 pre-treated with PPMP or the carrier control for 6 days was assayed 72 h post infection by quantification of the secreted luciferase due to transduction of the reporter plasmid phGluc. BKPsV were used as a positive control for a ganglioside-dependent infection. Mock PsV infections were done with a control sample obtained according the PsV protocol from cells only transfected with phGluc and control plasmid to test for background signal of the luciferase assay for uninfected cells. PsV experiments were done in quadruplicate and repeated three times. The results from one representative quadruplicate experiment is shown here as an example. Statistic analysis was performed using the two-tailed unpaired t test.

Fig 8. Conserved binding site for terminal Neu5Ac build up by the BC2-loop.

Mapping amino acid differences between (A, C) TSPyV VP1 and OraPyV VP1 and (B, D) TSPyV, OraPyV and ApanPyV based on the TSPyV VP1-GM1 complex structure. Amino acid differences between the two or three closely related viruses are coloured according their conservations on the surface of TSPyV VP1. An amino acid sequence alignment was done with Clustal Omega. JalView [67] was also used to assign values for the respective conservation of the chemical acid amino acid properties from 10 (conserved; grey) to 0 (non-conserved; red). Values of 2 to 0 are colored in red. (C-D) Close-up views of the HI- and BC2-loop regions. The backbone of the BC2- and the ccw DE-loops are depicted in cartoon representations. Side chains of conserved key amino acids within the BC2-binding site are shown in sticks in panel C.