Principles of Virus Structural Organization
Viruses, the molecular nanomachines infecting hosts ranging from prokaryotes to eukaryotes, come in different sizes, shapes, and symmetries. Questions such as what principles govern their structural organization, what factors guide their assembly, how these viruses integrate multifarious functions into one unique structure have enamored researchers for years. In the last five decades, following Caspar and Klug’s elegant conceptualization of how viruses are constructed, high-resolution structural studies using X-ray crystallography and more recently cryo-EM techniques have provided a wealth of information on structures of a variety of viruses. These studies have significantly furthered our understanding of the principles that underlie structural organization in viruses. Such an understanding has practical impact in providing a rational basis for the design and development of antiviral strategies. In this chapter, we review principles underlying capsid formation in a variety of viruses, emphasizing the recent developments along with some historical perspective.
KeywordsCapsid Protein Tobacco Mosaic Virus Scaffolding Protein Triangular Facet Icosahedral Symmetry
Viruses are metastable macromolecular assemblies composed of the viral genome enclosed within a proteinaceous capsid. They come in variety of sizes, shapes, and forms. Some are large, and some are small; some are spherical, and some are rod-like; some have lipid envelopes. Many of these viruses exhibit exquisitely symmetric organization. Irrespective of their shape and size, the underlying theme in all these viruses is that the virus structure is designed to contain and protect the viral genome and deliver it to a specific host cell for subsequent replication of the virus. Viruses are also distinguished based on the type of the genome that they contain: single-stranded or double-stranded RNA or DNA. The viral genome, in addition to encoding the proteins that constitute the capsid, also encodes other proteins referred to as nonstructural proteins, so called because they are not part of the final capsid’s organization. These nonstructural proteins are essential for viral replication inside the host cell. In some viruses, particularly of bacterial origin, viral genome encodes a protein called scaffolding protein that may not be part of the mature capsid but may be a critical factor in facilitating the capsid assembly.
Often, the size of the virus is proportional to the size of the genome. However, the viral genome contributes far less to the total mass of the virion than the capsid proteins. It was this observation that prompted Watson and Crick to suggest that the capsid has to be formed by the association of multiple copies of the capsid protein(s) (Crick and Watson 1956, 1957). Such an assembly with repeating subunits then greatly reduces the amount of genetic information required. In some viruses, the capsid formation involves a single gene product, whereas in other viruses which are more complex, it involves multiple gene products. Such an assembly involving repeating subunits raises several interesting questions. How do these subunits interact with one another with high fidelity and specificity to form the capsid architecture? This question becomes even more interesting in complex viruses in which the capsid formation involves multiple gene products. Are there any specific structural properties that these proteins should have for the capsid formation? How is capsid assembly directed and controlled? How is the genome encapsidated? In addition to containing and protecting the genome, the capsid architecture must also be conducive for interactions with the host cell for entry; how is this process coordinated? Given that capsid has to disassemble to make the genome available for replication, what are the cues for disassembly? How does the capsid organization respond to and evade the antiviral response mounted by the host?
In the last half century, structural studies on a variety of viruses have provided a wealth of information regarding some of the questions listed above. In addition to providing insight into the fundamental principles underlying various aspects of capsid assembly, more importantly, such studies have had practical impact in providing a rational basis for the design and development of antiviral strategies. Several excellent reviews on virus structures and principles underlying capsid formation have been published periodically over years (Klug and Caspar 1960; Caspar and Klug 1962; Rossmann and Johnson 1989; Johnson and Speir 1997; Harrison 2007); we will emphasize here the recent developments along with the some historical perspective.
3.2 Structural Techniques
Two principal techniques used in the structural studies on viruses are electron microscopy and X-ray crystallography. Contributions from other elegant studies using a variety of biochemical and biophysical techniques and theoretical modeling have been crucial in providing a more complete understanding of the capsid construction and assembly pathways. Electron microscopy of negatively stained virus specimens provided the first glimpse of viruses and led to early classification of viruses based on shape and form (Green et al. 1956; Brenner and Horne 1959; Horne and Wildy 1962, 1979; Wildy and Horne 1963). Even today, this technique is used as a diagnostic tool in identifying clinical virus samples. Subsequently, the discovery that EM images of virus particles, which are essentially projection images, can be used to reconstruct the three-dimensional structure of the virus using computer image analysis protocols (Crowther et al. 1970) paved the way for spectacular advances in specimen preparation (Knapek and Dubochet 1980; Dubochet et al. 1988), electron imaging, and computer image reconstructions. In the last two decades, this exciting new technology called three-dimensional cryo-electron microscopy (cryo-EM) has revolutionized the structure analysis of a variety of viruses (Baker et al. 2010; Crowther 2010; Grigorieff and Harrison 2011).
Much of our understanding of subunit interactions in a viral capsid at the atomic level has come from X-ray crystallographic structure of spherical viruses. Beginning with the structures of three small spherical plant viruses in early 1980s (Harrison et al. 1978; Abad-Zapatero et al. 1980; Liljas et al. 1982), over the last three decades, X-ray crystallography has been successfully applied to study a variety of larger and more complex spherical viruses including human viruses (Hogle et al. 1985; Rossmann et al. 1985; Liddington et al. 1991; Grimes et al. 1998; Reinisch et al. 2000; Wikoff et al. 2000; Reddy et al. 2010). The closely related technique of X-ray fiber diffraction has been used to study viruses that have helical symmetry (Namba and Stubbs 1986; Namba et al. 1989). In recent years, cryo-EM technique has allowed visualization of a variety of spherical viruses at subnanometer (Bottcher et al. 1997; Conway et al. 1997; Jiang et al. 2003; Zhang et al. 2003; Saban et al. 2006; Li et al. 2009) to near-atomic resolutions (Yu et al. 2008a, b; Baker et al. 2010; Liu et al. 2010; Wolf et al. 2010; Settembre et al. 2011, see also Chap. 4). For some viruses that are not amenable for high-resolution structural analysis by these techniques, complementarity between cryo-EM and X-ray crystallography has been exploited in deriving the pseudoatomic models of the capsid (Grimes et al. 1997; Mathieu et al. 2001; Zhang et al. 2002, 2007; Settembre et al. 2011). In these studies, when the virus capsid could not be crystallized, but a lower resolution structure could be determined by cryo-EM, as this technique does not require the specimen in a crystalline form, independently determined X-ray crystallographic structures of the capsid components are fitted into lower-resolution cryo-EM map of the capsid. Such a hybrid technique has been most useful in studying capsid–receptor, capsid–antibody interactions and in studying capsid-associated structural dynamics (Rossmann et al. 1994; Ilag et al. 1995; Smith et al. 1996; Stewart et al. 1997; Belnap et al. 2000; Conway et al. 2001; Martin et al. 2001; Nason et al. 2001; Dormitzer et al. 2004; Gan et al. 2006). Structure determination of spherical viruses either by X-ray crystallography or cryo-EM techniques relies implicitly on the symmetry of the capsid. As a result, the structural organization of the encapsidated genome is amenable to these structural techniques only when the genome follows the capsid symmetry. However, in recent years, there are several examples in which the entire genome or a significant portion of it is observed to follow the capsid symmetry and visualized in the structural analysis (Chen et al. 1989; Namba et al. 1989; Fisher and Johnson 1993; Larson et al. 1993). A detailed discussion on the genome organization in viruses is provided in a review by Prasad and Prevelige (2003). In this review, we mainly focus on the capsid organization. In addition to the X-ray crystallographic and cryo-EM structural techniques, other diffraction techniques such as neutron diffraction (Bentley et al. 1987), low-angle X-ray scattering (Tsuruta et al. 1998), and spectroscopic techniques (Tuma and Thomas 1997; Benevides et al. 2002) have been useful in understanding the capsid organization in viruses.
3.3 Capsid Organization in Spherical Viruses
3.3.1 Cubic Symmetry
3.3.2 Icosahedral Capsid Organization
3.3.3 Triangulation Numbers
The concept of triangulation and how it allows for more than 60 subunits with consequent increase in the size of the icosahedron can be illustrated using a hexagonal lattice with H- and K-axes crossing at 60° angle (Fig. 3.4a). By arbitrarily choosing a lattice point as the origin (0, 0) and considering it as the position of a fivefold vertex of an icosahedron, the position (H, K) of the neighboring fivefold vertex that is closest to the origin signifies the T number of that icosahedron. The equilateral triangle with the length of each side equal to the distance between the origin and the position (H, K) corresponds to one of the 20 triangular faces of such an icosahedron with each of its corners representing a fivefold vertex. The icosahedron with no triangulation can be described as having a triangulation number of 1 (T = 1) in which the closest fivefold vertex is positioned at (H = 1, K = 0). By defining the equilateral triangle representing each face in such a T = 1 icosahedron as a unit triangle, in a T = 4 icosahedron, for example, with closest fivefold vertex positioned at (H = 2, K = 0), the equilateral triangle describing each facet consists of four unit triangles (Fig. 3.4a). That means, in each triangular facet of the T = 4 icosahedron, 12 subunits can be placed, in contrast to three subunits in the facet of the T = 1 icosahedron. With 12 subunits in each of the 20 facets, a T = 4 icosahedron then will accommodate 240 (60T) subunits in contrast to 60 subunits in the T = 1 icosahedron (Fig. 3.4b). Also, as can be seen, the size of the T = 4 icosahedron compared to the size of the T = 1 icosahedron has proportionally increased. It should be pointed out here that subdividing the triangular facet resulting in icosahedra with T > 1 need not necessarily increase the size and that increased size is only with the assumption that molecular mass of the subunit remains approximately the same. Generally, however, spherical viruses with T > 1 tend to be of larger size.
3.3.4 Classes of Icosahedra
3.3.5 Icosahedral Asymmetric Unit and Quasi-Equivalent Subunits
An important concept that emerges from the triangulation is the icosahedral asymmetric unit. The threefold rotational symmetry axis passing through the center of the triangular icosahedral facet divides the subunits in the facet into three symmetrically equivalent sets. Each set of the subunits is defined as one icosahedral asymmetric unit, and application of the 5-3-2 rotational symmetry to this asymmetric unit produces all the 60T subunits in the icosahedron. In a T = 1 icosahedron, the asymmetric unit consists of one subunit, whereas in a T = 4 icosahedron, for example, the asymmetric unit consists of four subunits (Fig. 3.4b). As the interacting environment between these subunits in the icosahedron with T > 1 cannot be strictly equivalent, they are termed as “quasi-equivalent” subunits (black, blue, red, and green commas in Fig. 3.4b). Caspar and Klug proposed that these quasi-equivalent subunits in the icosahedral shell retain similar bonding interactions with minor distortions in their intersubunit interactions in order to adapt to the nonsymmetry-related environments.
3.3.6 Pentavalent and Hexavalent Positions
Another important consequence of the “quasi-equivalence” theory developed by Caspar and Klug is that triangulation necessarily results in the generation of hexavalent (sixfold) locations in addition to pentavalent (fivefold) positions on the icosahedral lattice (Fig. 3.4b). Irrespective of the T number, all icosahedra necessarily have 12 pentavalent positions. In the icosahedra with T > 1, because of the triangulation, 10(T − 1) hexavalent positions are generated. As a result, in icosahedra with T > 1, the 60T subunits could cluster into 12 pentamers around the fivefold vertices and 10(T − 1) hexamers at the hexavalent lattice points. Caspar and Klug defined these subunit clusters as morphological units. An icosahedron with a specific T number will then have 10(T + 2) morphological units [i.e., 10(T − 1) hexamers + 12 pentamers]. Although the arrangement of 60T subunits into rings of 5 and 6 is a geometrical necessity, clustering of these subunits into pentamers and hexamers is not; clustering into 20T trimers, 30T dimers, or 60T monomers is possible (Fig. 3.4b). Casper and Klug argued that if a particular oligomeric state, for example, hexamers, if they are particularly stable, they might be preformed, but when they are assembled into the shell, these conceptually planar units would have to be transformed into convex pentamers, resulting from the removal of a subunit, to occupy the “domed” pentavalent positions in the icosahedral structure. They suggested that such transformation would only require minor alterations in the dihedral angles between the subunits but essentially maintaining similar intersubunit contacts.
3.4 High-Resolution Structures of Spherical Viruses and Quasi-Equivalence Theory
Two basic tenets of the quasi-equivalence theory as discussed above are that (1) the icosahedral lattice with the possibility of triangulation presents the most efficient geometrical design for close packing of identical subunits in a spherical shell and (2) that the structural organization involves quasi-equivalent interactions requiring minimal distortions in the subunit bonding. Since the proposal of this theory in 1962, in the last three decades, high-resolution structures of several spherical viruses of different sizes have been determined. These structures clearly established that the capsid organization in spherical viruses follows icosahedral symmetry and that in viruses with subunits greater than 60, capsid organization is based on a triangulated icosahedral lattice as suggested from the quasi-equivalence theory. For example, several ssRNA plant viruses and some human viruses such as noroviruses, with capsid composed of 180 copies of the capsid protein, exhibit the expected T = 3 icosahedral organization with rings of five and six subunits. In dsDNA bacterial viruses, such as P22 (Jiang et al. 2003), HK97 (Wikoff et al. 2000), with 420 copies of the capsid protein, the capsid organization is based on the expected T = 7 icosahedral lattice with 60 hexamers and 12 pentamers. However, there are major surprises as well. The capsid organization in papovaviruses (papilloma- and polyomaviruses) represents a stunning departure from the quasi-equivalence theory (Liddington et al. 1991; Wolf et al. 2010). In these dsDNA spherical viruses, the capsid consists of 360 copies of the major capsid protein VP1. Such a number cannot be accommodated on a triangulated icosahedral lattice because T = 6 (360/60) is forbidden as it does not adhere H 2 + HK + K 2 rule (see above). Initially, although in conflict with the biochemical analysis which indicated 360 copies of the capsid protein in these virus, based on EM images of negatively stained specimens and computer reconstruction, an icosahedral structure with 420 copies organized as 60 hexamers and 12 pentamers on a T = 7 icosahedral lattice, as expected from the quasi-equivalence theory, was proposed (Klug and Finch 1968). However, subsequent high-resolution structures of these viruses unambiguously revealed that the capsid indeed consists of 360 subunits and that these subunits, instead of pentamer–hexamer clustering, are organized as 72 pentamers at 60 hexavalent and 12 pentavalent location on a triangulated T = 7d icosahedral lattice. (Liddington et al. 1991; Wolf et al. 2010) Other examples include adenovirus, a dsDNA virus, in which 240 trimers of the major capsid protein (hexons) occupy hexavalent positions on a T = 25 (pseudo) icosahedral lattice (Roberts et al. 1986; Liu et al. 2010; Reddy et al. 2010) and a unique icosahedral organization with 120 (forbidden T = 2) subunits which is a recurring theme in dsRNA viruses including fungal L-A virus (Naitow et al. 2002), partitivirus (Ochoa et al. 2008), and inner shells of bluetongue virus (Grimes et al. 1998), rotavirus (Lawton et al. 1997; Chen et al. 2006; Settembre et al. 2011), and reovirus (Reinisch et al. 2000).
3.4.1 Conformational Switching
Although the high-resolution structures of spherical viruses show general concordance with the quasi-equivalent theory in terms of subunit arrangement on triangulated icosahedral lattices, they exposed limitations of the theory particularly in regard to the concept of quasi-equivalent interactions that bond the subunits in the context of icosahedral lattice. These structures showed that formation of the icosahedral capsid can be governed by nonequivalent interactions involving internally located conformationally “flexible” arms of the capsid protein subunits. Except for these conformationally flexible arms, which function as molecular switches to allow the subunits to adapt to quasi-equivalent environments of the triangulated icosahedral lattice, the majority of the capsid protein remains structurally invariant.
3.4.2 Triangulated Lattice and Subunit Packing
One of the main tenets of the quasi-equivalence theory that triangulated icosahedral lattices allow for efficient close packing of subunits is clearly evident in the high-resolution structures of spherical viruses including the papovaviruses. However, there are distinct variations in how triangulated lattices confer optimal subunit packing. The all-pentamer papovavirus structure represents a unique variation in which subunit packing involves a triangulated lattice despite the mismatch between the molecular symmetry and the lattice coordination. Except for the 12 pentamers at the fivefold positions of the T = 7 lattice, the location of other pentamers, at the six-coordinated positions, is not consistent with their molecular symmetry. The advantage, however, is clearly evident considering that a triangulated lattice intrinsically provides locations for efficient hexagonal close packing of the subunits. The papovavirus structures elegantly demonstrate how the six-coordinated positions in the T = 7 lattice allow for a favorable close packing of the roughly cylindrical pentamers between the fivefold positions requiring only three different types of interpentamer contacts (Fig. 3.8d).
In the T = 3 structures with 180 identical subunits, as discussed previously, the subunit packing is as per the quasi-equivalent theory. The trapezoidal-shaped β-barrel domains of the capsid protein pack closely into rings of 6 at the hexavalent lattice points and rings of 5 around the pentavalent positions. Structures of picornaviruses (Hogle et al. 1985; Rossmann et al. 1985; Luo et al. 1987; Acharya et al. 1989) present an interesting variation depicting how a T = 3 lattice allows similar packing but with nonidentical subunits. In picornaviruses, the icosahedral asymmetric unit consists of chemically nonidentical VP1, VP2, and VP3. Despite not having any sequence homology with capsid proteins of T = 3 viruses or with themselves, VP1, VP2, and VP3 exhibit the same canonical β-barrel fold. The arrangement of 180 β-barrels, from 60 copies of each of the VP1, VP2, and VP3, is strikingly similar to that observed in the T = 3 icosahedral lattices formed by 180 identical copies of a capsid protein. In the picornavirus structures, VP1, VP3, and VP2 occupy the same positions as chemically equivalent A, B, and C, respectively, in the T = 3 icosahedral viruses. Because VP1, VP2, and VP3 are chemically nonidentical, the picornavirus structures are described as a pseudo T = 3 lattice. The pseudo T = 3 organization in the comovirus provides yet another interesting variation with two chemically distinct polypeptide chains, the small (S) protein with one β-barrel domain and the large (L) protein with two β-barrel domains (Lin et al. 1999). The packing of the β-barrels from these two proteins follows T = 3 lattice and is similar to that observed in the T = 3 icosahedral structures. The β-barrel of the S protein occupies the position corresponding to A in the T = 3 lattice, whereas the two β-barrels of the L protein occupy positions corresponding to B and C.
3.4.3 Capsid Assembly
In an attempt to answer what dictates appropriate conformational switching in the formation of icosahedral shells, Berger et al. (1994) has proposed a local rule-based theory suggesting that protein subunits make use of local information to guide the capsid assembly and that the choice for a particular interaction is dictated by its immediate neighbors. In the T = 3 icosahedral viruses, dimer of the capsid protein is thought to be the building block for the assembly. In solution, these dimers are perhaps in a dynamic equilibrium between the “bent” and “flat” conformations, and during the assembly, these dimers adopt appropriate conformational states. Two different assembly pathways have been proposed for the T = 3 icosahedral viruses. In the case of sobamoviruses, such as southern bean mosaic virus (SBMV), it is suggested that assembly involves the formation of pentamers of dimers as an intermediate step followed by the association of other dimers resulting in the formation of a T = 3 shell (Erickson et al. 1985). Such a pathway is consistent with the observation that deletion of NTA in the SBMV results in a T = 1 structure formed by the association of 12 pentamers of dimers. A similar pathway involving pentamers of dimers as an assembly intermediate is also proposed for noroviruses (Prasad et al. 1999), which is supported by recent mass spectroscopic analysis of recombinant Norwalk virus particles (Shoemaker et al. 2010). In the case of tombus viruses, such as TBSV, it is suggested that the assembly intermediate involves trimers of dimers, consistent with the observation that the ordered NTAs of the C subunits form a stable internal structure at the icosahedral threefold axes of the T = 3 shell (Sorger et al. 1986). These viruses have a stretch of basic residues at the N terminus (R arm) of the capsid protein that can interact with the RNA. The intermediate assembly unit, either pentamers of dimers or trimers of dimers, is thought to be nucleated by interaction with the packaging signal in the genomic RNA (Sorger et al. 1986; Harrison 2007). In the case of noroviruses, the capsid protein, lacking the basic R arm, itself has all the determinants for the formation of the T = 3 structure because the recombinant capsid protein of noroviruses readily assembles into T = 3 structures (Prasad et al. 1999). The role of RNA in directing the assembly pathway is readily apparent in the structure for flock house virus (a nodavirus) (Fisher and Johnson 1993). For picornaviruses, the assembly pathway is somewhat better characterized. The 5S structural unit consisting of one copy of VP0, VP3, and VP1 and 14S pentameric caps of VP1–VP2–VP3 are known to be the intermediates. In the case of papovaviruses, in which preformed stable pentamers are the building blocks, the correct assembly of the pentamers onto a T = 7 lattice is likely dictated by interactions with the viral minichromosome. Initial interactions between the pentamer and the DNA, involving the N-terminal arms of the subunits, may serve as a nucleation center for further stepwise addition of individual pentamers or a cluster of pentamers, consisting of one pentavalent pentamer surrounded by five other pentamers (1+5 cluster), to form the T=7 capsid structure with the encapsidated genome (Stehle et al. 1996; Mukherjee et al. 2007). During this process, the curvature is appropriately modulated by local alterations in the “bonding” between the pentamers involving the CTAs.
As can be seen from the above discussion, the interplay between the global restraints, which allow optimal packing of the subunits to form a closed shell of proper size to accommodate the genome, and local conformational variability, which allows, necessary flexibility in the intersubunit contacts for modulating the curvature, dictates structural realization in icosahedral viruses.
3.4.4 Scaffolds, Glue Proteins, and Cores
In more complex icosahedral viruses such as dsDNA bacteriophages, herpes simplex viruses, and adenovirus, the establishment of the icosahedral lattice involves several other factors like scaffolding proteins, accessory proteins, maturation-dependent proteolysis, and even larger-scale conformational changes in their major capsid proteins than is observed in simpler icosahedral viruses.
In contrast to assembly based on the encapsidation of the nucleic acid concurrently with the formation of the capsid shell by stepwise addition of smaller assembly intermediates as discussed in the previous section, in some of these viruses, capsid assembly is accomplished by the formation of a complete capsid shell, followed by the insertion of the nucleic acid. Such a mechanism avoids some of the problems associated with the former assembly mechanism, such as the requirement that both the capsid proteins and the nucleic acid be brought to a common assembly point and properly staged for assembly, and the necessity of using a nonicosahedral component (the RNA) to help build an icosahedral capsid. However, this mechanism introduces other problems that must be solved. First, since most dsDNA phages and viruses packaged by this mechanism have a single genome segment, it is absolutely required that only one copy of the nucleic acid be inserted into each capsid shell. This is accomplished by having a DNA packaging machine, called the portal, located at only one of the 12 vertices of the capsid (see also Chap. 22). Second, the capsid shell must be of the correct size to fit the nucleic acid genome, and it must be empty of cellular proteins, which could interfere with DNA insertion. This is accomplished by assembling the capsid around a set of virally encoded scaffolding proteins. Thus, the shell is guaranteed to be the right size and T number, and the scaffolding proteins are removed either during DNA insertion or by proteolysis after capsid shell assembly. Third, in the cases studied so far, the capsid proteins of the procapsid have a different conformation that they do in the mature virion. This could serve as a signal that the DNA has been packaged, and is ready for the final steps of virus maturation, like the addition of the phage tail. Finally, the DNA insertion is accompanied by an increase in internal pressure generated by the repulsion of the charged nucleic acid. This pressure may be as high as 40 atm and has been suggested to facilitate or power DNA release. Most dsDNA bacteriophages have mechanisms, such as glue proteins in epsilon 15 or chemical crosslinking of the capsid shell proteins as in HK97, to strengthen the capsid shell against this pressure. So how do the capsid proteins, the scaffolding proteins, the glue, and the portal interact in this system of virus assembly?
220.127.116.11 “True” T = 7 Capsid Organization
18.104.22.168 Scaffolding Protein
To ensure accurate formation of the capsid shell, these viruses use an elaborate mechanism involving a scaffold which is either provided by a separate virally encoded protein as in the case of P22 (Fane and Prevelige 2003, see also Chap. 14) or by a part of the capsid protein itself as in the case of HK97 (Huang et al. 2011). In P22, the scaffolding protein exits signaling for the maturation of the capsid, and then, it is recycled for the further rounds procapsid assembly. In the case HK97, the reordering of the N-terminal region of the capsid protein is suggested to provide a trigger for the maturation (Huang et al. 2011).
Recent cryo-EM reconstructions of P22 procapsid have provided some insights into the possible role of scaffolding proteins in directing the T = 7 procapsid assembly (Parent et al. 2010a, b; Chen et al. 2011). In the reconstructions of the P22 procapsid, imposing icosahedral symmetry, the major portion of the scaffolding protein density is not clearly observed, suggesting that the majority of the scaffolding protein does not conform to the icosahedral symmetry of the capsid. However, the portion that interacts with the inside surface of the capsid shell is visible (Fig. 3.9d). These interactions are observed to be stronger with the hexavalent subunits than with the pentavalent subunits, suggesting that scaffolding protein may promote or stabilize the hexamer of the capsid protein relative to the pentamer. It is clear that the scaffolding proteins influence the icosahedral lattice because P22 mutants with altered or missing scaffolding proteins make a significant number of T = 4 capsid shells with half the number of hexavalently disposed units, i.e., 30 versus 60 (Thuman-Commike et al. 1998). Thus, the scaffolding protein may influence the formation of the hexameric and pentameric clustering of the capsid protein subunits by altering the ratio of hexamers to pentamers during the assembly process.
22.214.171.124 The Portal
There is no evidence that the capsid proteins of P22 oligomerize into either hexamers or pentamers before being incorporated into the capsid shell. Thus, the shell appears to be built monomer by monomer, without the participation of preformed pentamers or hexamers. It is likely that the portal protein together with scaffolding protein guides the assembly of capsid protein into the correct shell. Visualization of the portal complex, because it is located at only one of the 12 fivefold vertices, requires reconstructions of the capsid structure without imposing icosahedral symmetry as was first done in the case of phi29 (Tao et al. 1998) and subsequently with several other dsDNA bacteriophages including P22 (Jiang et al. 2006; Chen et al. 2011). In the cryo-EM reconstructions of P22 procapsid, without imposing the icosahedral symmetry, the portal protein complex is clearly seen and sits as a dodecamer at one of the fivefold vertices (Chen et al. 2011) (Fig. 3.9e, f). Such a reconstruction shows that the portal interacts more directly with scaffolding protein than with the capsid shell in the procapsid consistent with the notion that scaffolding protein recruits the portal protein complex (Fane and Prevelige 2003). This suggests that the portal protein is a key participant in the capsid assembly (Chen et al. 2011). If the level of portal protein is kept well below that of the scaffolding or capsid proteins, the portal protein may nucleate the assembly of a capsid shell by its interactions with scaffolding and capsid shell proteins. This would in turn assure that there is only one portal per capsid, which is a requirement for proper DNA packaging. Capsids assembled in the absence of portal proteins are aberrant in their size, form, and/or symmetry.
126.96.36.199 Maturation and Expansion
During maturation of P22, the capsid protein undergoes a major conformational change, making the capsid shell thinner and the hexon interactions more symmetric. The shell becomes more angular from its round shape in the procapsid, is thinner by about 40 Å, and expands in diameter by about 100 Å (Prasad et al. 1993; Chen et al. 2011). The process of DNA packaging and concomitant release of the scaffolding proteins, through the small holes at the hexavalent positions of the procapsid, may trigger the conformational changes in the subunits and subunit contacts resulting in a dramatically expanded capsid. Similar expansion accompanied by large-scale conformational change upon maturation is also observed in HK97 and ε15 (Jiang et al. 2006; Wikoff et al. 2006). In the HK97, maturation is accompanied by the formation of an intersubunit isopeptide linkage that cross-links the entire capsid conferring stability to the capsid to withstand the pressure created by DNA packaging. In P22, which does not form a cross-linked capsid like HK97, strong protein–protein interactions among its capsid shell subunits appear to render sufficient structural rigidity in the mature virion (Parent et al. 2010a, b; Chen et al. 2011).
188.8.131.52 Glue Proteins
The nucleocapsid of herpesvirus, a dsDNA virus of eukaryotic origin with a distinctly different structural organization, nevertheless shares several characteristics with dsDNA bacteriophages (Heymann et al. 2003). It exhibits a T = 16 icosahedral structure (∼1,250 Å in diameter) with pentamer and hexamer clustering of the major capsid protein VP5. 960 copies of VP5 forms 12 pentamers and 150 (10T − 1) hexamers as expected from the quasi-equivalent symmetry (Schrag et al. 1989; Zhou et al. 1994, 2000). In addition to VP5, capsid also consists of what are known as triplex proteins, VP19c and VP23 (in a stoichiometric ratio of 1:2), that are positioned at the local and strict threefold axes of the T = 16 lattice. These triplexes function as glue holding together the surrounding VP5 capsomeres. Similar to dsDNA bacteriophages, capsid assembly is assisted by scaffolding protein which is proteolytically degraded prior to the packaging of the genomic DNA (Nicholson et al. 1994; Rixon et al. 1996). Recombinant baculovirus expression of VP5 and VP19c alone, i.e., in the absence of VP23 and the scaffolding protein, results in smaller-sized particles with T = 7 symmetry suggesting the role of VP23 and the scaffolding protein in ensuring the correct capsid assembly (Saad et al. 1999). Another interesting similarity with dsDNA bacteriophages is that VP5 exhibits the HK97-like fold (Baker et al. 2003; Bowman et al. 2003). In addition to these components, as revealed by the electron cryo-tomographic reconstructions, the nucleocapsid has a unique fivefold vertex with a 12-fold symmetric portal for inserting the genomic dsDNA (Cardone et al. 2007; Chang et al. 2007; Deng et al. 2007). The nucleocapsid capsid also exhibits maturation-dependent morphological alterations (Heymann et al. 2003). These observations clearly indicate that DNA packaging mechanism in herpes virus resembles that in the tailed dsDNA bacteriophages.
One common theme in these dsRNA viruses is the assembly of the outer capsid layer(s), typically based on T = 13ℓ icosahedral symmetry, and is assembled on the innermost layer that surrounds the dsRNA genome (see also Chap. 17). This innermost layer exhibits unusual icosahedral organization consisting of 120 subunits (forbidden “T = 2”) arranged as 60 asymmetric dimers on a T = 1 icosahedral lattice (Lawton et al. 1997; Grimes et al. 1998; Reinisch et al. 2000; Chen et al. 2006) (Fig. 3.11e). Such an organization appears to be highly conserved in all the dsRNA viruses including dsRNA viruses of bacterial and fungal origins (Naitow et al. 2002; Ochoa et al. 2008), and the available structural data also indicate that the polypeptide folds of the proteins that form this layer are similar, despite lacking any noticeable sequence similarity.
It is possible that this unique organization of the core capsid layer in these viruses has evolved to serve a dual purpose: to properly position the transcription enzyme complex and organize the genome to facilitate endogenous transcription. In many of these viruses, the virally encoded enzymes required for the transcription (RNA-dependent RNA polymerase) and capping intimately associate with this layer at the fivefold positions (Prasad et al. 1996; Zhang et al. 2003; Nason et al. 2004). One model is that pentamers of dimers of the subunits in the core layer associate along with the transcription enzyme complex, and 12 of these pentamers further associate to form the core layer with 120 subunits, which then serves as a platform for the assembly of the outer layers perhaps by sequential addition oligomeric (typically trimers) of capsid protein that constitutes the outer layer(s).
3.5 Helical Viruses
In addition to icosahedral symmetry that is commonly found in the spherical viruses, another symmetry that is prevalent among viruses is the helical symmetry. Many rod-shaped viruses such as viruses belonging to family Tobamoviruses (Namba and Stubbs 1986, see also Chap. 28), potyviruses (Kendall et al. 2008), rhabdoviruses (Ge et al. 2010), and nucleocapsids of several animal viruses such as Sendai viruses (Egelman et al. 1989) exhibit this kind of symmetry. In contrast to icosahedral symmetry which involves only rotational symmetry, helical symmetry involves both rotational and translational components, which when combined, give a screw axis. Repeated application of rotation to a motif followed by translation along an axis gives rise to a structure with helical symmetry with a defined pitch (P). Helical structures are typically characterized by the radial location of the subunit with respect to the helix axis; the rotation per subunit, which is equivalent to the number of subunits per turn (n) in the helix; and the axial rise per subunit (h).
The interactions between the TMV capsid protein and the genomic RNA in general are nonspecific involving the basic residues of the protein and the phosphate groups of the RNA. One exception, however, is the anomalous repulsive interaction between the carboxylate group of Asp116 and a phosphate group of the RNA (Namba et al. 1989; Ge and Hong Zhou 2011) (Fig. 3.12b). Considering that TMV, and in general any viral assembly, has to assemble and disassemble during its infectious cycle, it is suggested that such a repulsive interaction, which may be required to maintain an energy balance, confers metastable nature to the TMV and functions as a trigger for driving viral disassembly (Namba et al. 1989; Culver et al. 1995; Stubbs 1999).
The assembly process in TMV has been studied extensively, and it serves as the best-characterized example of cocondensation of capsid protein and genomic RNA. In this process, the viral RNA interacts with a 20S aggregate, a two-turn helix of the capsid protein, and the assembly proceeds by addition of 20S aggregates through a highly cooperative process pulling the 5′ end of the RNA through the central hole of the growing TMV rod (Caspar and Namba 1990; Butler 1999; Klug 1999). It is suggested that a disorder-to-order transition of a loop in the capsid protein, analogous to conformational switching in the icosahedral viruses, that is induced by the binding of RNA may play an important role in this process (Namba et al. 1989; Culver et al. 1995). In the X-ray structure of the 20S aggregate, this particular loop is disordered, whereas in the TMV structure, it is ordered. The TMV assembly is initiated by the specific recognition of the sequence AAGAAGUCG in the viral RNA by the TMV capsid protein (Zimmern 1976; Butler 1999). The high-resolution structure of the fully assembled TMV has provided some insights in to how the capsid protein recognizes this sequence (Namba et al. 1989). Although the three RNA-binding sites in each subunit can accommodate any base, one of the binding sites is particularly suitable for G and allows favorable hydrogen bond interactions. In the initiation sequence, every third nucleotide being G thus may provide a strong discrimination for the higher affinity binding of the packaging signal over the rest of the sequence in which the XXG motif does not occur in phase with a statistically significant frequency.
3.6 Enveloped Viruses
During their morphogenesis, several viruses acquire a lipid envelope derived from a cellular organelle which remains part of their capsid organization. In some of the viruses such as influenza viruses, herpes viruses, coronaviruses, bunya viruses, and HIV, the lipid envelop is externally located and is studded by various viral proteins, whereas in others such as alphaviruses and flaviviruses, the lipid envelop is internally located underneath the outer proteinaceous capsid layer. With the exception of alphaviruses and flaviviruses, many of the enveloped viruses lack highly symmetric organization and are less amenable for high-resolution structural analysis. However, structures of the individual protein components of these enveloped viruses, e.g., hemagglutinin (Wilson et al. 1981; Bullough et al. 1994), neuraminidase (Varghese et al. 1983; Xu et al. 2008), and M2 protein of influenza virus (Acharya et al. 2010); glycoprotein of mouse hepatitis virus (Xu et al. 2004); gp120 and gp41 of retroviruses (Chan et al. 1997; Chen et al. 2005); envelop protein of a flavivirus (Rey et al. 1995; Rey 2003; Li et al. 2008); and most recently, E1/E2 complex of alphaviruses, have been well characterized (Li et al. 2010; Voss et al. 2010). These structures have provided significant insight into the mechanism of how the lipid bilayers in these viruses fuse with the membrane of the target cell (Rey 2006; Lamb and Jardetzky 2007; Harrison 2008). In addition, the highly symmetric parts of enveloped viruses such as the nucleocapsid of herpes virus that exhibit T = 16 icosahedral organization (as discussed above) and the core of hepatitis B virus, which can form either a T = 3 or a T = 4 icosahedral structure, have been well characterized (Crowther et al. 1994; Conway et al. 1997; Wynne et al. 1999).
The quasi-equivalence theory proposed almost half a century ago has been remarkably useful in describing icosahedral virus structure. The generality of its predictions arises from the simplicity afforded by the assumption that groups of subunits packing in a plane are energetically favored to have six neighbors (i.e., are hexagonally close-packed) and that curvature can be generated by introducing nodes (vertices) with only five such neighbors. However, such generality and adaptability can come at the cost of specificity. The propensity for subunits to form quasi-equivalent interactions leading to a wrong (usually smaller and simpler) final capsid structure indicates that more information is sometimes required. So we observe that in some viruses with T = 7 symmetry, such as P22, scaffolding proteins are used either as a template core around which a proper-sized capsid can form, or, more subtly, to control the relative proportion of subunits which adopt hexameric interactions to those which form pentamers. This idea of altering the stability of one conformational form over another is also exemplified in the RNA viruses, where the RNA plays a role in stabilizing capsid proteins in the proper ratio of conformations. In the members of Reoviridae, the inner core layer forms a simple T = 1 “permanent scaffold” from 60 dimers, upon which the T = 13 outer layers are built. In adenovirus and herpesvirus, the role of scaffolding is augmented by the incorporation of accessory proteins in the capsid shell, which also appear to function to influence the curvature, and therefore the triangulation number of the capsid shell. Finally, the tendency for hexagonal close packing to be a controlling influence in virus assembly appears to be so robust that in the case of polyomavirus, where pentamers seem to be the only oligomeric species present, pentamers come to occupy the hexavalent positions. The exceptions to the quasi-equivalence theory prove the rule.
We acknowledge the support from NIH grants R37AI36040 (BVVP), PO1AI057788 (BVVP), P41RR02250 (MFS) and a grant from Robert Welch Foundation (Q1279) to BVVP. We are grateful to Donghua Chen, Corey Hryc, Liya Hu, Wah Chiu, and Hong Zhou for the help in the figures.
- Crick FHC, Watson JD (1957) Virus structure: general principles. In: Wolstenholme GEW, Millar ECP (eds) CIBA foundation symposium on the nature of viruses. Little Brown and Co, BostonGoogle Scholar
- Crowther, RA (2010) From envelops to atoms: The remarkable progress of biological electron microscopy. Advances in Protein Chemistry and Structural Biology Vol 81 “Recent Advances in Electron cryomicroscopy” Part A, (eds, Ludtke S, and Prasad, BVV), pp 2–33.Google Scholar
- Harrison SE (2007) Principles of virus structure. Lippincott Williams and Wilkins, Philadelphia, PAGoogle Scholar
- Zhang R, Hryc CF, Cong Y, Liu X, Jakana J, Gorchkov R, et al (2011) Cryo-EM structure of an enveloped alphavirus venezuelan equine encephalitis virus. submitted for publication. EMBO J 30:3854–3863Google Scholar