An N-terminal extension to the hepatitis B virus core protein forms a poorly ordered trimeric spike in assembled virus-like particles

Abstract

Virus-like particles composed of the core antigen of hepatitis B virus (HBcAg) have been shown to be an effective platform for the display of foreign epitopes in vaccine development. Heterologous sequences have been successfully inserted at both amino and carboxy termini as well as internally at the major immunodominant epitope. We used cryogenic electron microscopy (CryoEM) and three-dimensional image reconstruction to investigate the structure of VLPs assembled from an N-terminal extended HBcAg that contained a polyhistidine tag. The insert was seen to form a trimeric spike on the capsid surface that was poorly resolved, most likely owing to it being flexible. We hypothesise that the capacity of N-terminal inserts to form trimers may have application in the development of multivalent vaccines to trimeric antigens. Our analysis also highlights the value of tools for local resolution assessment in studies of partially disordered macromolecular assemblies by cryoEM.

Keywords: Cryo-electron microscopy; Hepatitis B virus; Local resolution; Three-dimensional reconstruction; Vaccine; Virus-like particle.

Fig.1

3D reconstructions of His-β-L HBcAg virus-like particles. (A) Cryomicroscopy of His-β-L HBcAg VLPs revealed two sizes 30 nm T = 3 and 34 nm T = 4. (B) The T = 3 structure was generated using 6048 particle images and achieved a resolution of 10 Å. (C) The T = 4 VLP was reconstructed using 2040 particles and achieved a resolution of 12 Å. Resolution assessment was by the FSC^0.5 criterion (D). The general architecture of both capsids is similar to the wild type HBcAg, as demonstrated by comparison with a solvent exclusion surface representation of the T = 4 capsid calculated from the X-ray structure PDB ID 1QGT (E). His-β-L HBcAg particles however were seen to bear an extra spike density hypothesised to comprise a trimer of the extended N-terminus, located at sites of local threefold symmetry in the T = 3 VLP (magenta) and at both icosahedral (green) and local (magenta) threefold symmetry axes in the T = 4 particle. Wall-eyed stereo pair images are presented (B, C and E). Cross-sections through the T = 3 (F) and T = 4 (G) reconstructions show that the density in the N-terminal spikes is blurred compared to that of the HBcAg component (magenta and green arrows).

Fig.2

ResMap analysis of His-β-L HBcAg VLP reconstructions. ResMap was used to evaluate the local resolution of both T = 3 (A) and T = 4 (B) reconstructions. Isosurfaced representations of the reconstructions were coloured accordingly revealing that the HBcAg component was solved at higher resolution than the N-terminal extension. In the case of the T = 3 particle the core density achieved resolution approaching 5 Å, ranging to 8 Å in places, while the inserted polypeptide was seen to be poorer than 19 Å resolution. The T = 4 reconstruction was calculated from fewer particles and achieved lower resolution ranging between 8 and 10 Å in the HBcAg region and 11–>19 Å in the N-terminal extension. Interestingly both ResMap analyses gave resolution assessments that included small patches of poor-resolution throughout the structure, in particular within the dimeric spike of HBcAg, these are apparent in clipped isosurface representations (C and D).

Fig.3

Blocres local resolution analysis of His-β-L HBcAg VLP reconstructions. Local FSC resolution analysis was performed using the BSOFT routine Blocres. The T = 3 (A) and T = 4 (B) reconstructions are presented in stereo pair view and coloured according to local resolution. For both structures, the N-terminal inserted region, has a lower resolution than the native HBcAg capsid. This finding is consistent with the ResMap analysis and suggests that the inserted region is more flexible than the core capsid structure. Unlike the ResMap analyses however the HBcAg dimeric spikes are assessed as being higher-resolution although the distal portions of these four-helix bundles are lower resolution than the shell of the capsid (C and D).

Fig.4

Reconstructions of the sharpened T = 3 His-β-L HBcAg VLP shows extensive secondary structure. (A) The T = 3 VLP structure is shown sharpened to 8 Å resolution and at an arbitrary isosurface threshold, set to highlight secondary structure elements. (B) The crystal structure of each HBcAg monomer from the T = 4 capsid structure (PDB ID 1QGT) was docked to each quasiequivalent position within the T = 3 asymmetric unit. Interestingly chain A (mauve), chain B (green) and chain C (magenta) from the T = 4 structure each gave the best fit when docked into those respective positions in the T = 3 reconstruction. (C) Expanding the symmetry of the asymmetric unit highlights the good fit of the HBcAg crystallographic coordinates into the cryoEM 3D reconstruction. (D) A close up view of a single AB dimer docked into the 3D reconstruction shows that the four-helix bundle is very well resolved. (E and F) Cross-sections through the reconstructed density show that following sharpening the N-terminal spike density is still somewhat blurred. Density measurements show it to be ∼45 percent of the peak intensity value for the map. This suggests that although flexible the feature is most likely constrained by trimerisation.

Fig.5

Homology modelling of His-β-L HBcAg. To evaluate the structure of the N-terminal extension of His-β-L HBcAg the sequence was homology modelled in MOE against the known structure of HBcAg with outgap modelling to predict the structure of the extension. The models revealed a largely unstructured polypeptide with regions of α-helix close to the N-terminus. The best model is shown (rainbow – blue = N terminus, red = C terminus) as a stereo pair viewed from the top (A) and side (B) and as a dimer with the HBcAg structure (grey) for comparison. In this model the N-terminus is oriented towards the capsid interior. Most likely this is not correct as this region includes the polyhistidine tract that was employed in purifying the particles by immobilised metal affinity chromatography. Moreover docking the model into our sharpened T = 3 reconstruction (thresholded to enclose the correct molecular volume) reveals that the model does not fit into the reconstructed density and there is substantial collision with symmetry related protomers (C and D). Three dimers are shown arranged about the N-terminal spike feature at a local threefold symmetry axis. Each monomer that contributes an N-terminal extension to the spike feature is shown in the rainbow colour scheme, while the symmetry related monomers are shown as the HBcAg structure in grey.