Adeno-associated virus 2: classic T = 1 icosahedral design. The parvoviruses are very small animal viruses, with particles of ~25 nm in diameter that encase single-stranded DNA genomes of <5 kb. These small, naked capsids are built from 60 copies of a single subunit organized according to T = 1 icosahedral symmetry. The protein that forms the subunits of adenovirus-associated virus type 2, a member of the dependovirus subgroup of parvoviruses (Appendix, Fig. 19), contains a core domain commonly found in viral capsid proteins (the β-barrel jelly roll; see next section), in which β-strands are connected by loops (Fig. 4.11A). Interactions among neighboring subunits are mediated by these loops. The prominent projections near the threefold axes of rotational symmetry (Fig. 4.11B), which have been implicated in receptor binding, are formed by extensive interdigitation among the loops from adjacent subunits. Adenovirus-associated virus vectors have proved valuable in human gene therapy (Volume II, Chapter 9), in part because variations in the sequences of these loops confer differences in tissue tropism.
Poliovirus: a T = 3 structure. As their name implies, the picornaviruses are among the smallest of animal viruses. In contrast to the T = 1 parvoviruses, the ~30-nm-diameter poliovirus particle is composed of 60 copies of a multimeric structural unit. It contains a (+) strand RNA genome of ~7.5 kb and its covalently attached 5′-terminal protein, VPg (Appendix, Fig. 21). Our understanding of the architecture of the Picornaviridae took a quantum leap in 1985 with the determination of high-resolution structures of human rhinovirus 14 and poliovirus.
The heteromeric structural unit of the poliovirus capsid contains one copy each of VP1, VP2, VP3, and VP4. The VP4 protein is synthesized as an N-terminal extension of VP2 and restricted to the inner surface of the particle. The poliovirus capsid is built from asymmetric units that contain one copy of each of three different proteins (VP1, VP2, and VP3), and is therefore described as a pseudo T = 3 structure (Fig. 4.12A). Although these three proteins are not related in amino acid sequence, all contain a central β-sheet structure termed a β-barrel jelly roll. The arrangement of β-strands in these β-barrel proteins is illustrated schematically in Fig. 4.12B, for comparison with the actual structures of VP1, VP2, and VP3. As can be seen in the schematic, two antiparallel β-sheets form a wedge-shaped structure. The protein backbones in β-barrel domains of VP1, VP2, and VP3 are folded in the same way; that is, they possess the same topology, and the differences among these proteins are restricted largely to the loops that connect β-strands and to the N- and C-terminal segments that extend from the central β-barrel domains.
The β-barrel jelly roll conformation of these picornaviral proteins is also seen in the core domains of capsid proteins of a number of plant, insect, and vertebrate (+) strand RNA viruses, such as tomato bushy stunt virus and Nodamura virus. This structural conservation was entirely unanticipated. Even more remarkably, this relationship is not restricted to small RNA viruses: the major capsid proteins of the DNA-containing parvoviruses and polyomaviruses also contain a β-barrel domain, and two such domains form the major capsid proteins of larger DNA-containing viruses, such as adenoviruses and mimiviruses. It is well established that the three-dimensional structures of cellular proteins have been highly conserved during evolution, even though there may be very little amino acid sequence identity. For example, all globins possess a common topology based on a particular arrangement of eight α-helices, even though their amino acid sequences are different. One interpretation of the common occurrence of the β-barrel jelly roll domain in viral capsid proteins is that seemingly unrelated modern viruses (e.g., picornaviruses and parvoviruses) share some portion of their evolutionary history. It is also possible that this domain topology represents one of a limited number commensurate with packing of proteins to form a sphere, and therefore an example of convergent evolution. The structural (and other) properties of viruses with double-stranded DNA genomes provide compelling support for the first hypothesis (Box 4.5).
Figure 4.12 Packing and structures of poliovirus proteins. (A) The packing of the 60 VP1-VP2-VP3 structural units, represented by wedge-shaped blocks corresponding to their β-barrel domains. Note that the structural unit (outlined in black) contributes to two adjacent faces of an icosahedron rather than corresponding to a facet. When virus particles are assembled, VP4 is covalently joined to the N terminus of VP2, from which it is later cleaved. It is located on the inner surface of the capsid shell (see Fig. 4.13A). (B) The topology of the polypeptide chain in a β-barrel jelly roll is shown at the top left. The β-strands, indicated by arrows, form two antiparallel sheets juxtaposed in a wedge-like structure. One of the β-sheets comprises one wall of the wedge, while the second, sharply twisted β-sheet forms both the second wall and the floor. The two α-helices (purple cylinders) that surround the open end of the wedge are also conserved in location and orientation in these proteins. As shown, the VP1, VP2, and VP3 proteins each contain a central β-barrel jelly roll domain. However, the loops that connect the β-strands in this domain of the three proteins vary considerably in length and conformation, particularly at the top of the β-barrel, which, as represented here, corresponds to the outer surface of the capsid. The N- and C-terminal segments of the protein also vary in length and structure. The very long N-terminal extension of VP3 has been truncated in this representation. The structures of VP1, VP2, and VP3 are from PDB ID: 1HXS.
The overall similarity in shape of the β-barrel domains of poliovirus VP1, VP2, and VP3 facilitates both their interaction with one another to form the 60 structural units of the capsid and the packing of these units. How well these interactions are tailored to form a protective shell is illustrated by the model of the capsid shown in Fig. 4.13: the extensive interactions among the β-barrel domains of adjacent proteins form a dense, rigid protein shell around a central cavity in which the genome resides. The packing of the β-barrel domains is reinforced by a network of protein-protein contacts on the inside of the capsid, which are particularly extensive about the fivefold axes (Fig. 4.13C). The interaction of five VP1 molecules, which is unique to the fivefold axes, results in a prominent protrusion extending to about 25 Å from the capsid shell (Fig. 4.13A). The protrusion appears as a steep-walled plateau encircled by a valley or cleft. In the capsids of many picornaviruses, these depressions, which may contain the receptor-binding sites, are so deep that they have been termed canyons.
One of several important lessons learned from high-resolution analysis of picornavirus capsids is that their design does not conform strictly to the principle of quasiequivalence. For example, despite the topological identity and geometric similarity of the jelly roll domains of the proteins that form the capsid shell, the subunits do not engage in quasiequivalent bonding: interactions among VP1 molecules around the fivefold axes are neither chemically nor structurally equivalent to those in which VP2 or VP3 engage.