In the model presented earlier, the mature virion (see Fig. 3B) is composed of “shells” of individual Gag proteins: MA, which forms the outer shell, lies just underneath the lipid membrane and makes contact with it via the amino-terminal myristylated and positively charged segment. The myristate moieity is presumably buried in the lipid bilayer, but there is no direct evidence for this supposition. The balls that represent the protein are packed closely and, as a consequence, in places form hexagonal arrays, but the model does not portray any overall symmetry in virion structure since this has not been established. The number of balls approximates the number of MA molecules in a virion. Jutting through the membrane is the TM component of Env, with its internal domain contacting MA in an unspecified way. The external portion of TM is bound to the SU component of Env. The segment of SU that interacts with TM is rendered arbitrarily as two small spheres. The SU/TM complex is shown as a dimer to indicate an oligomeric status (Chapter 3).

Figure 3

Models for retroviral structure. (Models courtesy of Lou Henderson.) (Top) The immature HIV-1 virion. The Gag and Gag-Pol proteins are shown with different colors to suggest the domains corresponding to the mature proteins formed from (more...)

Farther inside the MA layer is a shell of CA protein that forms the capsid, defined as the outer layer of the core. The shape of the core is conical in HIV, but cores from other retroviruses may have different shapes. The small blue balls represent the p6 protein that forms the carboxyl terminus of HIV-1 Gag, and which is present only in lentiviruses. The location of p6 in the virion is uncertain and is shown here arbitrarily on the outside of the CA shell. In the center of the core is the complex of NC protein and RNA. Associated with this complex are smaller numbers of RT and IN molecules. Both of these proteins bind to nucleic acids and thus may be expected to be in contact with the ribonucleoprotein complex, but direct evidence for this interaction in virions is lacking. It is also not known if the complex has any kind of higher-order structure, as suggested by the drawing. The third enzyme in virions, PR, is shown as two balls because as a mature protein it is a dimer. The location of PR in virions is not known, but since PR must have had access to all the cleavage sites on the Gag and Gag-Pol proteins before maturation, some PR molecules are shown inside the core and some outside. Finally, the model shows several small peptides (short bars) that are derived from the spacer regions between CA and NC, between NC and p6, and amino-terminal to PR. Not drawn in the model are tRNAs or other small host RNAs. Some host-cell proteins are shown incorporated in the viral membrane, as demonstrated for the MHC class I proteins for HIV-1.

The immature virion (see Fig. 3A) is composed of exactly the same polypeptide sequences. However, before processing has occurred, its structure is quite different. Only three types of proteins make up the particle, in addition to the lipid envelope and the RNA. Env is the same as in the mature virion, although its contacts with Gag underneath the membrane may be different. Gag is shown as an elongated tooth-like object. The MA domain with its myristate abuts the membrane, consistent both with its ability to be crosslinked to lipid in immature MLV and with the expectation that the myristate would interact with the lipid environment of the membrane. The NC domain at the other end of the molecule is rendered as a bicuspate shape, reflecting the presence of two Cys-His motifs and two basic RNA-binding sites. The RNA in the center of the virion is not drawn in a distinct way. Presumably, it is bound to many of the NC domains, but perhaps not to all of them. CA is drawn not as a ball but as an elongated connector between MA and NC to suggest possible conformational changes that might occur upon proteolytic processing. The p6 domain is shown arbitrarily as bending back to contact CA. Gag-Pol is rendered as a dimer. Although there is no direct evidence for the multimeric status of Gag-Pol in immature virions, as mature proteins, IN, RT, and PR are dimeric. The Gag-Pol protein that is their precursor is likely to be dimeric as well. In particular, the PR domains must dimerize to initiate proteolytic processing.

As Bolognesi et al. (1978) predicted many years ago, the organization of protein domains in Gag is a key feature that sets up the structure of the virion: The shells of protein from outside to inside match the amino- to carboxy-terminal order of MA, CA, and NC and probably also define the relationship of RT and IN. In the model, each shell comprises a single protein, and thus all protein-protein interactions are between like partners. This organization is consistent with early crosslinking results which detected only homotypic dimers and higher multimers. In the model for immature virions, the primary interactions again involve only like domains. The sparse available crosslinking data also are consistent with this view. The proteolytic cleavages responsible for maturation sever the domains from each other, leading to condensation of the core, still leaving each protein to interact with its like neighbors, but possibly altering their relationships somewhat.

As discussed in the first part of this chapter, the actual structure of the retroviral particle has been difficult to decipher, in large part due to its lability. The models shown make a number of predictions, most of which have not been addressed adequately. One is that each shell of protein is composed only of a single Gag protein. This prediction has never been confirmed biochemically except by crosslinking, a technique for which negative results are difficult to interpret. It may be, for example, that the capsid shell contains not only CA, but also some NC or other proteins. Similarly, it may be that some MA molecules are not in the shell underneath the membrane. This would be consistent with the nuclear targeting signal on HIV-1 MA and the finding that this protein is part of the HIV-1 preintegration complex. Recent findings also indicate that the analogous VSV M protein has more than one location in the virion. Another prediction is that the Pol domains are located together with the RNA in the central portion of the virion. The model as well as thin-section electron micrographs of immature particles might be taken to suggest that the concentration of protein in the central portion of the virion is low. However, such an inference may not be warranted. The model is drawn with few Gag-Pol molecules for clarity, and the amount of heavy metal deposited in the electron microscopy technique is not directly proportional to protein content, and hence may be misleading. In fact, if protein distribution were uniform, the 5% of protein mass due to Pol would occupy a sphere that is about three-eighths the diameter of the whole virion, not very different from the electronlucent region seen for immature viruses by electron microscopy. Cryo-EM studies should be able to yield more definitive information on this and other aspects of retroviral structure.