of V. Reddy, The Scripps Research Institute. (B) Views of the outer (left) and inner (right) surfaces indicating the locations of the minor capsid proteins IX, IIIa, V, VI, and VIII (colored as in Fig. 4.15A) with respect to hexons (gray) and penton base (magenta). Data from Yu Y et al. 2017. Sci Adv 3:e1602670.
Removal of the outermost protein layer, a process thought to occur during entry into a host cell, yields an inner core structure, comprising one shell (orthoreoviruses) or two (rotaviruses and members of the genus Orbivirus, such as bluetongue virus). These subviral particles also contain the genome and virion enzymes and synthesize viral mRNAs in vitro under appropriate conditions. High-resolution structures have been obtained for bluetongue virus and human reovirus cores, some of the largest viral assemblies that have been examined by X-ray crystallography. Their thin inner layer contains 120 copies of a single protein (termed VP3 in bluetongue virus). These proteins are not related in their primary sequences, but they nevertheless have similar topological features and the same plate-like shape. Moreover, in both cases, the dimeric proteins occupy one of two different environments, and to do so, they adopt one of two distinct conformational states, indicated as green and red in Fig. 4.17C (right). Because of this arrangement, the green and red dimers are not quasiequivalent, and virtually all contacts in which the two monomer conformations engage are very different. However, these differences allow the formation of VP3 assemblies with either five- or threefold rotational symmetry and hence of an icosahedral shell. This VP3 shell of bluetongue virus abuts directly on the inner surface of the middle layer, which comprises trimers of a single protein organized into a classical T = 13 lattice (Fig. 4.17C, left). A large number of different (nonequivalent) contacts between these trimers and VP3 weld the two layers together and hence stabilize both. These properties of reoviruses illustrate that a classic quasiequivalent structure is not the only solution to the problem of building large viral particles: viral proteins that interact with each other and with other proteins in multiple ways can provide an effective alternative. The organization of the two protein shells described above appears to be conserved in most viruses with double-stranded RNA genomes. However, it is not yet known whether symmetry mismatch is also a feature of other large viruses that contain multiple protein layers.
Figure 4.17 Structures of members of the Reoviridae. The organization of mammalian reovirus (A) and rotavirus (B) particles is shown schematically to indicate the locations of proteins, deduced from the protein composition of intact particles and of subviral particles that can be readily isolated from them. dsRNA, double-stranded RNA. (C) X-ray crystal structure of the core of bluetongue virus, a member of the Orbivirus genus of the Reoviridae, showing the core particle and the inner scaffold. Trimers of VP7 (VP6 in rotaviruses; panel B) project radially from the outer layer of the core particle (left). Each icosahedral asymmetric unit, two of which are indicated by the white lines, contains 13 copies of VP7 arranged as five trimers colored red, orange, green, yellow, and blue, respectively. This layer is organized with classical T = 13 icosahedral symmetry. As shown on the right, the inner layer is built from VP3 dimers that occupy one of two completely different structural environments, colored green and red. Green monomers span the icosahedral twofold axes and interact in rings of five around the icosahedral fivefold axes in a T = 2 structure. In contrast, red monomers are organized as triangular “plugs” around the threefold axes. Differences in the interactions among monomers at different positions allow close packing to form the closed shell. As might be anticipated, VP7 trimers in pentameric or hexameric arrays in the outer layer make different contacts with the two classes of VP3 monomer in the inner layer. Nevertheless, each type of interaction is extensive, and in total, these contacts compensate for the symmetry mismatch between the two layers of the core. The details of these contacts suggest that the inner shell both defines the size of the virus particle and provides a template for assembly of the outer T = 13 structure. From Grimes JM et al. 1998. Nature 395:470–478, with permission. Courtesy of D.I. Stuart, University of Oxford.
Figure 4.18 Asymmetric capsids of retroviruses. (A) Variation in the morphology of retroviruses shown schematically. Although all retrovirus particles are assembled from the same components (see the text), the cores are primarily spherical, cylindrical, or conical in the case of gammaretroviruses (e.g., Moloney murine leukemia viruses), betaretroviruses (e.g., Mason-Pfizer monkey virus), and lentiviruses (e.g., human immunodeficiency virus type 1), respectively. (B) Cryoelectron tomographic slice of human immunodeficiency virus type 1 showing the conical core and the glycoprotein spikes projecting from the surface of the particle. © Jun Liu, Yale University School of Medicine, with permission. Courtesy of H. Winkler, Florida State University.
EXPERIMENTS
A fullerene cone model of the human immunodeficiency virus type 1 capsid
Diverse lines of evidence support a fullerene cone model of this capsid based on principles that underlie the formation of icosahedral and helical structures.
(A) A purified human immunodeficiency virus type 1 protein comprising the capsid linked to the nucleocapsid proteins, CA-NC self-assembles into cylinders and cones when incubated with a segment of the viral RNA genome in vitro. The cones assembled in vitro are capped at both ends, and many appear very similar in dimensions and morphology to cores isolated from viral particles (compare the two panels, shown at the same scale, as indicated by the bars). From Ganser BK et al. 1999. Science 283:80–83, with permission. Courtesy of W. Sundquist, University of Utah. (B) The very regular appearance of the synthetic CA-NC cones suggested that, despite their asymmetry, they are constructed from a regular, underlying lattice analogous to the lattices that describe structures with icosahedral symmetry discussed in Box 4.3. In fact, the human immunodeficiency virus type 1 cores can be modeled using the geometric principles that describe cones formed from carbon. Such elemental carbon cones comprise helices of hexamers closed at each end by caps of buck-minsterfullerene, which are structures that contain pentamers surrounded by hexamers. As in structures with icosahedral symmetry, the positions of pentamers determine the geometry of cones. However, in cones, pentamers are present only in the terminal caps. The human immunodeficiency virus type 1 cones formed in vitro and isolated from mature virions can be modeled as a fullerene cone assembling on a curved hexagonal lattice with five pentamers (red) at the narrow end of the cone, as shown in the expanded view. The wide end would be closed by an additional 7 pentamers (because 12 pentamers are required to form a closed structure from a hexagonal lattice). (C) The fullerene cone model was subsequently confirmed and refined by cryo-EM of helical tubes of CA at higher resolution, molecular dynamics simulations, and cryo-EM of cores purified from and within virus particles. Shown is an example of computational slices of perfect fullerene cones observed within virus particles, with cryoelectron tomographic models superimposed. The C-terminal domains of CA molecules are shown in gray, the N-terminal domains of CA pentamers in blue, and those of CA hexamers colored according to the quality of their alignment, from red (low) to green (high). From Mattei S. 2017. Science 354:1434–1437, with permission. Courtesy of J. Briggs, European Molecular Biology Laboratory, Heidelberg, Germany.
