Caister Academic Press

Papillomavirus-like Particles and Their Applications in MolecularVirology, Human Serology and Vaccines

Recommended reading:   Climate Change and Microbial Ecology | Polymerase Chain Reaction | SUMOylation and Ubiquitination
A review of Papillomavirus-like Particles.

Papillomavirus-like Particles and Their Applications in MolecularVirology, Human Serology and Vaccines

(adapted from Roden and Viscidi in Papillomavirus)
Papillomavirus-like Particles and Their Applications in MolecularVirology, Human Serology and Vaccines: Early studies of Papillomavirus (PV) structure were dependent upon virions purified from warts because the virus failed to replicate in conventional tissue culture. These studies revealed a 55-60 nm diameter non-enveloped icosahedral structures containing an approximately 8 Kb double stranded supercoiled DNA genome packaged as a nucleohistone complex. Biochemical studies indicated that the capsid surrounds a nucleohistone core and is composed of the major and minor capsid proteins, designated L1 and L2 respectively. The capsid has T=7 icosahedral symmetry and is composed of 72 pentameric capsomers that have a right-hand skew. Computer generated reconstructions from cryo-electron micrographs indicate that the capsomers at the 12 vertices of the virion contact five other capsomers (pentavalent), whereas the remaining sixty capsomers contact six other capsomers (hexavalent). Thus PV virions resemble SV40 and polyoma virions. However, in contrast to both SV40 and polyoma, no high resolution X-ray crystallographic structure has been determined for an intact PV virion despite several attempts. This failure reflects the difficulties in obtaining native virus from warts or tissue culture in sufficient quantity and quality for crystallization studies.

In the early 1990s several groups demonstrated that the major capsid protein L1 derived from diverse PV genotypes can assemble alone into virus-like particles (VLPs) that morphologically resemble native virions when expressed in mammalian, insect, yeast or bacterial cells. Thus, these L1 VLPs provided an opportunity to better define the contribution of L1 to the capsid structure. Unlike virions, the VLPs can easily be generated in quantities sufficient for crystallization necessary for a high resolution structural model for PV virions based on X-ray diffraction analysis. Since VLPs are produced in systems amenable to genetic manipulation, a number of virus related processes can now be explored by mutagenesis of the capsid proteins, including virion assembly and structure-function relationships.

Cryo-electron micrograph images of virions can be reconstructed into three dimensional electron density maps. This approach has been used successfully both for the polyomaviruses and papillomavirius native virions, including HPV1, BPV1 and CRPV. While this approach verifies the conserved T=7d structure of papillomavirus, it is not clear how each capsid protein contributes to the virion structure. To address the contribution of L1 and L2 to the structure of the capsid, Hagensee et al attempted to reconstruct L1 only and L1/L2 VLPs generated by recombinant Vaccinia virus-driven overexpression. A three dimensional reconstruction from cryo-electron micrographs revealed HPV1 L1 VLP comprise 72 pentamers of L1 centered on the vertices of a T=7 icosahedral lattice. This structure appeared identical at a resolution of 35Angstrom to the capsid structure of HPV1 virions, thus implying that L1 VLPs closely replicate native capsid structure. Unfortunately, the analysis failed to distinguish L1 only VLPs from those containing L2, suggesting that higher resolution analysis was necessary to identify L2 or that its occupancy was limited. However the relative heterogeneity of VLP assembly compared to native virions complicates attaining much higher resolution images. Although we achieved a resolution of 9Angstrom for an image reconstruction of native Bovine Papillomavirus type 1 purified from cow warts, we were only able to generate reconstructions of BPV1 L1 only or L1/L2 VLPs to a greater than approximately 25 Angstrom (unpublished data). At higher resolutions, the image reconstruction BPV1 virions revealed a blob of density at the center of the pentavalent capsomers at the capsid vertices which was absent from the hexavalent capsomers. Notably the T=7 structure of papillomavirus contains 60 hexavalent and only 12 pentavalent capsomers. Furthermore estimates from Coomassie-stained SDS-PAGE analyses of purified BPV1 virions suggest a ratio of 30 L1 molecules for each L2 molecule. Since there are 360 L1 molecules per virion we hypothesized that each virion contains only 12 L2 molecules and that the blob of density at the center of the 12 pentavalent capsomers represented a portion of the L2 protein. This remains unproven, and surprising since the minor capsid proteins VP2 and/or VP3 are in the center of all capsomers in polyomaviruses.

While full length HPV16 L1 assembles into relatively regular 72 pentamer T=7 VLPs at low pH, a deletion mutant lacking the N-terminal ten amino acids assembled in to a 12-pentamer T=1 capsid of 318Angstrom diameter under all conditions tested. The X-ray crystal structure of T=1 HPV16 L1 VLPs demonstrates that residues 20-282 of L1 form a classic 'jelly roll' ß sandwich, similar to that described for polyoma and SV40 VP1. Thus despite minimal sequence similarity between polyomavirus VP1 and L1, their architecture is closely related. Elaborate L1 interstrand loops protrude from the surface of the capsomer structure. Comparison of the L1 sequences amongst different HPV genotypes reveals highly variable stretches that are interspersed with regions of sequence conservation. The regions of low sequence conservation correspond to the loops protruding from the capsid surface. The apparent lack of tissue specificity of the primary papillomavirus cell surface receptor suggests that these changes are not to determine tropism. However, several studies have identified residues that contribute to the structure of conformationally-dependent L1 neutralizing epitopes using VLPs or capsomers derived from mutant L1 proteins. The locations of sequence differences in L1 both map to the surface of the atomic model and are all close to the residues identified as contributing to the conformationally-dependent L1 neutralizing epitopes.

The C-terminal arm of HPV16 L1 is significantly alpha helical and three helices (h2-4) mediate contact with adjacent L1 monomers. This projection is anchored back to the central jelly roll structure via a short strand, which adds to the C edge of the CHEF sheet, and a final helix (h5). This h5 helix tucks into the axial cavity of the pentamer via a hydrophobic interaction with the base of the BIDG sheet. The final 31 residues project towards the lumen but are disordered. The backbones of the five L1 monomers associate intimately with adjacent molecules to form the tightly packed donut shape of the pentamer. The tightest contacts within the pentamer occur at the higher radii and the L1 monomers splay outwards towards the floor of the coat. Thus the L1 pentamer exhibits a conical hollow that progressively opens towards the interior. This hollow is only ~14 Angstrom in diameter at its narrowest. Contact is mediated by the G strand at the inner edge of the BIDG sheet which interacts with the CHEF sheet of the adjacent L1 monomer, before returning. The HI loop also intertwines between the FG and EF loops of the adjacent L1 monomer, thus generating the five points of the pentamer's stellate cap.

Clearly the atomic model of small VLPs has greatly informed our understanding of papillomavirus structure. However, there are limitations to the small VLPs as a model of virion structure including the absence of interactions with L2 and the nucleohistone core. Furthermore, the T=1 structure of the small VLP suggests that the interpentamer contacts may not be representative of those in native virions. In the small VLP, the interpentamer contact occurs as an HPV16 L1 C-terminal subdomain protrudes into the adjacent pentamer then returns to the central channel of the pentamer from which it originated. By contrast, the C-terminal arms of polyomavirus VP1 tether by directly invading the adjacent pentamer. Furthermore, while T=1 small VLPs contain only pentamers co-ordinated with five adjacent pentamers (i.e. pentavalent capsomers), the native T=7 virion and VLP of both polyomaviruses and papillomaviruses comprise 60 pentavalent and 12 hexavalent capsomers. Thus the 'arm invasion' model is more consistent with the true structure rather than the simple contact between L1 capsomers in the small VLPs. Indeed, biochemical and mutagenesis studies of T=7 VLPs also support the 'arm invasion' model.

Comparison of the apparent molecular weight of L1 on reducing and non-reducing SDS-PAGE indicates that disulphide bonding mediates the formation of L1 trimers. Point mutation of each individual cysteine in HPV33 L1 revealed that two residues, cysteines 176 and 427 form a disulphide bridge and that this bond is critical to particle assembly. Likewise, point mutation of HPV11 cysetine 424 to glycine resulted in the production of capsomers but not VLPs. Indeed, HPV VLPs can be quantitatively disassembled and re-assembled by addition and removal of reducing agents under the appropriate ionic conditions. The presence of calcium ions was not required for VLP assembly, although it is necessary for in vitro reassembly of BPV1 virions. Furthermore, proteolytic digestion and mutagenesis studies suggest that these cysteines tether the C terminal arm of one L1 subunit to an outer loop of an adjacent L1 subunit. Li et al demonstrated that trypsin cleavage of HPV11 L1 at R 415 removed the C-terminal 86 residues and prevented assembly beyond capsomers. However, the C-terminal 24 amino acids of BPV1 L1 were dispensable for the formation of VLPs. Additional evidence for this model was obtained by LC-MS/MS analysis of the tryptic fragments derived from L1 dimers separated on a non-reducing SDS-PAGE gel. This study revealed the corresponding peptides in HPV18 L1 linked via a disulphide bond between cysteine 175 and cysteine 429, indicating that this is conserved between three distinct evolutionary clades of papillomaviruses, represented by HPV11, HPV18 and HPV33. To generate a model for the structure of T=7d papillomavirus VLP, Modis et al fitted the X-ray crystal structure into the high resolution image reconstruction generated from native BPV1 virions. The two structures demonstrated an excellent fit. The only region of the HPV16 L1 small VLP atomic coordinates that failed to match the BPV1 map contained residues 402-446, which forms the intercapsomer contacts in the small VLP. Therefore, these residues were rebuilt to fit in a new conformation into unoccupied volume within the BPV1 map and constrained by the requirement of cysteine 428 in the C-terminal arm forming a disulphide bond with cysteine 175 in the EF loop of the adjacent L1, consistent with the 'invading arm' model. Confirmation of this model structure awaits an X-ray crystallographic analysis of native virions. Despite such a high resolution structure, the residues responsible for binding to cell surface receptor molecules are not known. Furthermore, studies using monoclonal antibodies to probe structure indicate that the capsid may undergo a significant conformational change upon interaction with the cell surface. This change may relate to the closed (compact) and open (swollen) structures described for CRPV virions. It is unclear how the X ray crystallographic structure of small VLPs relate to these additional structural forms of the papillomavirus capsid.

The study of VLPs has also informed our understanding of L1 assembly in vivo. While purified L1 can self-assemble in vitro, it is likely that chaperones guide this process inside the cell. Expression of HPV11 L1 in insect cells via recombinant baculovirus resulted in the accumulation of VLPs in both the cytoplasm and nucleus, although predominantly in the latter. Indeed L1 lacking a nuclear localization signal is able to assemble into VLPs but fails to bind DNA. Studies of the polyomavirus major and minor capsid proteins, VP1 and VP2/3, indicate that they associate in the cytoplasm and enter the nucleus as a complex. However, L2 is expressed prior to L1 in papillomas and accumulates in the nucleus prior to L1 when co-expressed in recombinant Vaccinia virus infected cells. Further, nuclear translocation of L2 but not L1 is blocked by treatment with proteosome inhibitor. Once in the nucleus, L2 trafficks to the subnuclear domain termed ND-10 or PML-oncogenic domain (POD). Although L1 alone exhibits a diffuse nuclear pattern by indirect immunofluoresence, co-expression with L2 causes their co-localization in ND-10. Accumulation of L1 in ND-10 occurs after L2-induced exit of Sp100 from ND-10.

L2 binds to L1 capsomers but not whole VLP, suggesting that L2 co-assembles with L1 rather than being inserted into a pre-formed capsid. Indeed, L2 is required for efficient genome encapsidation, suggesting that the capsid assembles around the histone-bound genome, rather than by injection of the genome into the capsid via a portal vertex. L2 does not require disulphide bridges to interact with L1 in a VLP. Discrete salt sensitive binding and salt insensitive L1-binding was observed upon disassembly of HPV33 L1/L2 VLPs. Indeed two distinct L1 binding domains were described for BPV1 L2; a C-terminal L1 binding domain (residues 384-460) that interacts with L1 capsomers in vitro, and a central region (residues 129-246) that fails to interact with capsomers. The interaction between L2 residues 129-246 and BPV1 L1 were described on the basis of co-immunoprecipitation and co-localization studies. An L1 binding domain was described between residues 396-439 of HPV11 L2, consistent with the C-terminal L1 binding domain in residues 384-460 of BPV1 L2. This interaction, which occurs in vitro and is relatively salt insensitive, enhances in vitro assembly of VLPs. Previous studies have commented upon enhanced or more regular in vivo assembly of VLPs in the presence of L2. Indeed, fourfold higher hemagglutinating activity was described for BPV1 L1/L2 versus L1 only VLP preparations. An early study reported that HPV 16 L1 was unable to self-assemble in the absence of L2. However, the L1 of the reference strain used in these experiments was later shown to be an assembly-deficient mutant. Most HPV 16 L1 proteins, and all that efficiently self-assemble, contain an aspartate at position 202 rather than the histidine found in the L1 of the reference strain. It is interesting to note that the presence of L2 provided for significant assembly of even this defective mutant of L1. Therefore, care must be taken in comparison between the mechanisms of in vivo virion assembly and the 'five-around-one' in vitro L1 VLP assembly mechanism, since L2 and chaperones are likely to profoundly influence this process.

L2-specific antibody neutralizes infection in vitro and is protective in vivo demonstrating that a small portion of L2 is exposed on the surface of the capsid. Several studies have attempted to identify the nature of these neutralizing epitopes using L1/L2 VLP to better define the topology of L2. Neutralization data suggest that HPV16 L2 residues 108-120 and 69-81 are epitopes displayed on the surface of VLPs and virions. Clearly our knowledge of L2's topology in the capsid is limited but perhaps the L1 capsomer-L2 complex or pseudovirions may be suitable for X-ray crystallographic studies. Unlike structures of VLPs or capsomers, analysis of pseudovirion or true virion preparations would also clarify the interaction between the capsid and the nucleohistone core. Studies with purified capsid proteins or VLPs indicate that the C-terminal positively charged tail of L1 that includes a nuclear localization signal is also critical for binding to and packaging DNA. Similar sequences on both termini of L2 may also play a role in encapsidation of the viral genome as well as infection. While suggestive, the studies with VLPs and purified capsid proteins await detailed virion mutagenesis and structural studies for confirmation.

In summary, VLPs have been important reagents for studies of capsid structure and assembly. Unlike virions, the VLPs can easily be generated in quantities sufficient for crystallization and have provided a high resolution structural model for PV virions based on X-ray diffraction analysis. While this is a great achievement, certain aspects of the structure of papillomavirus such as inter capsomer bonding and the structures of L2 and the nucleohistone core await high resolution structural and site-directed genetic analysis of virions.

An authoritative reference on Papillomavirus is provided by the new book Papillomavirus.

Further reading