Assembly and Analysis of Conical Models for the HIV-1 Core

See allHide authors and affiliations

Science  01 Jan 1999:
Vol. 283, Issue 5398, pp. 80-83
DOI: 10.1126/science.283.5398.80


The genome of the human immunodeficiency virus (HIV) is packaged within an unusual conical core particle located at the center of the infectious virion. The core is composed of a complex of the NC (nucleocapsid) protein and genomic RNA, surrounded by a shell of the CA (capsid) protein. A method was developed for assembling cones in vitro using pure recombinant HIV-1 CA-NC fusion proteins and RNA templates. These synthetic cores are capped at both ends and appear similar in size and morphology to authentic viral cores. It is proposed that both viral and synthetic cores are organized on conical hexagonal lattices, which by Euler's theorem requires quantization of their cone angles. Electron microscopic analyses revealed that the cone angles of synthetic cores were indeed quantized into the five allowed angles. The viral core and most synthetic cones exhibited cone angles of approximately 19 degrees (the narrowest of the allowed angles). These observations suggest that the core of HIV is organized on the principles of a fullerene cone, in analogy to structures recently observed for elemental carbon.

HIV-1 assembly is initially driven by polymerization of the Gag polyprotein, which forms a spherical shell associated with the inner membrane of the budding particle. The three major regions of Gag all perform essential roles in viral assembly: the NH2-terminal MA (matrix) region binds the membrane, the central CA (capsid) region mediates important Gag-Gag interactions, and the COOH-terminal NC (nucleocapsid) region packages the viral RNA genome [reviewed in (1, 2)]. As the particle assembles, the viral protease cleaves Gag, producing discrete MA, CA, and NC proteins, which subsequently rearrange to form the mature, infectious viral particle. During maturation, MA remains associated with the inner viral membrane, while CA and NC condense about the viral RNA to form an unusual conical structure at the center of the virus (the “core”). The interior of the core is composed of an RNA/NC copolymer, and is surrounded by an outer shell composed of ∼1500 copies of CA. The conical core appears to be essential, because Gag mutations that disrupt proper core formation invariably inhibit viral infectivity (3). The core probably organizes the viral RNA genome (and its associated enzymes) for uncoating and replication in the new host cell, although these processes are not yet well understood.

Our initial goal was to develop a model system for studying viral core structure and assembly in vitro. The assembly properties of pure recombinant HIV-1 Gag protein fragments have been investigated in several laboratories (4–7). Pioneering work by Campbell and Vogt demonstrated that fragments of HIV-1 and Rous sarcoma virus (RSV) Gag proteins that encompass the CA and NC domains can assemble into long hollow cylinders in the presence of RNA (5). Building on this, we screened for conditions that would support the assembly of conical (rather than cylindrical) structures. Initially, we employed an HIV-1 RNA template that spanned sites required for genomic RNA packaging (Ψ) and dimerization (DLS), because some models for the viral core have suggested that the genomic RNA dimer dictates the cone morphology (8). The protein construct included both the CA and NC domains of HIV-1 Gag, because viral core morphology can be disrupted by mutations in either of these domains, or in the short spacer peptide that connects them (3). Finally, solution assembly conditions were varied, because cylinder formation is sensitive to protein and RNA concentrations, salt, and pH (4–7). Cone formation was assayed by transmission electron microscopy (TEM) of negatively stained samples.

A mixture of cones (Fig. 1A, arrows) and cylinders formed spontaneously upon incubation of a pure recombinant CA-NC fusion protein with a purified 1400-nucleotide (nt) HIV-1 RNA template in 500 mM NaCl (pH 8.0) (9). Cone:cylinder ratios as high as ∼2:3 were observed under these optimized conditions. Cones also formed under physiological conditions [that is, 150 mM NaCl (pH 7.2)], albeit at reduced efficiencies. The synthetic cones were capped at both ends, and many appeared strikingly similar to authentic HIV-1 cores (Fig. 1B). Cones formed in vitro varied between 100 and 300 nm in length. Viral cores are typically ∼100 nm long (8, 10); however, this can also vary considerably because HIV-1 virions range between ∼120 to 260 nm in diameter (11). These similarities suggest that the synthetic cores are accurate structural models for the viral core.

Figure 1

CA-NC/RNA complexes spontaneously assemble into cones in vitro. (A) TEM image of a representative field of negatively stained particles formed by the CA-NC protein on a 1400-nt HIV-1 RNA template. Conical structures are denoted by arrows. Scale bars in Figs. 1, 2, and 4 are 100 nm. (B) Selected thin-sectioned TEM images of an authentic HIV-1 virion grown in culture (bottom) and a synthetic CA-NC/RNA cone assembled in vitro (top). Electron microscope preparations of virions and synthetic cores were identical, and the two objects are shown at the same magnification.

To test whether specific HIV-1 RNA sequences or structures were essential for core formation, we also carried out assembly reactions using noncognate RNA templates (12). The 6400-nt RNA from tobacco mosaic virus (TMV) also supported cone formation (Fig. 2A), as did a 1400-nt fragment from theBacillus stearothermophilus 16S ribosomal RNA (13). Although the efficiency of cone assembly varied considerably in different reactions, these variations did not correlate with the RNA template used. Thus, our experiments seem to rule out models in which specific viral RNA sequences or structures (for example, the DIS structure) account for the asymmetry of the cone (8). We then examined assembly of the CA-NC protein in the absence of RNA to determine whether RNA was required at all for cone formation. The CA-NC protein alone could form cones (and cylinders), but only under very high ionic strength conditions (for example, 1 M NaCl) (Fig. 2B) (12). Therefore, RNA promotes CA-NC assembly, possibly by concentrating the CA-NC protein or by neutralizing charge, or both, but the polynucleotide chain itself does not play an indispensable structural role. Overall, our experiments demonstrate that the ability to form conical cores is an intrinsic property of the HIV-1 CA-NC complex, and it does not absolutely require an RNA template, the presence of any additional viral or cellular factors, or prior formation of an immature virion.

Figure 2

(A) Particles formed by CA-NC protein on a 6400-nt TMV RNA template. (B) Particles formed by the CA-NC protein in high salt concentration (1M NaCl) in the absence of RNA.

Although asymmetric, the synthetic cores often appeared highly regular, suggesting that they might be constructed from a regular underlying lattice. Our models for the lattice of the viral core were guided by the recent discovery that elemental carbon can also form cones of varying angles that are capped by different subsurfaces of buckminsterfullerene (14, 15). These carbon cones are composed of helices of hexamers arrayed on conical hexagonal lattices (derived from plane group symmetry p6) and are closed by pentameric “defects” at the narrow and wide ends of the cone (Fig. 3). From Euler's theorem, the cone angles (θ) of a conical hexagonal lattice will be quantized into the five allowed angles (Fig. 3, A and B) defined by Eq. 1:Embedded Image(1)Fullerene cones can be capped by the introduction of pentameric “defects” at the narrow and wide ends of the cone, with each pentamer surrounded by five hexamers (as in buckminsterfullerene). The cone angle determines the number of pentamers (P) required to close the narrow end of the cone (where P is an integer between 0 and 6), with narrower cone angles closed by larger numbers of pentamers.

Figure 3

(A) Quantization of the angles of cones derived from conical hexagonal (p6) lattices. A cone can be created by cutting along the vertical line at the top of the circle, making a second cut along another lattice line, removing the intervening slice, and resealing the two interfaces (with or without helical skew). Note, however, that if the second cut is made off a lattice line, the cone cannot be resealed without disruption of the hexagonal lattice. (B) The five allowable angles resulting from the conical constructs described in (A). Allowable cone angles are 112.9° (P = 1 cap), 83.6° (P = 2), 60° (P = 3), 38.9° (P = 4), and 19.2° (P = 5). (C) Stereoview of a capped 19.2° fullerene cone composed of 252 hexagons and 12 pentagons. Pentagons within the narrow (P = 5) and wide (P = 7) caps are shaded. Note that alternative arrangements of the pentagons will alter the shapes of the caps. (D) End-on view of aP = 5 cap configuration. The five pentagons at the apex of the cone are shaded. Figure parts (A) through (D) were adapted from (14) and drawn using Chem3D Ultra (Cambridge Software).

We propose that HIV-1 cores form cones using the same symmetry principles as carbon. In our model, the viral core assembles on a curved p6 lattice, with five pentameric defects arrayed in aP = 5 configuration at the narrow end (Fig. 3, C and D). Because a total of 12 pentameric defects are required to create a closed object from a hexagonal lattice, the wide end of the cone will be capped with an additional seven pentameric defects. The precise distribution of the pentamers and their distribution will determine the shapes of the caps. The driving force for closing the conical lattice could be saturation of all possible CA interfaces, as has also been observed for carbon cones (14).

To the best of our knowledge, accurate measurements of HIV-1 core angles have not been reported, but crude three-dimensional (3D) reconstructions of the core appear to us to be consistent with the narrow (19.2°) cone angle predicted for a P = 5,7 cone (8, 16). Cryo–electron microscopic images of our synthetic cores in vitreous ice similarly suggested the presence ofP = 5,7 cones (17). These analyses were complicated, however, because the synthetic cores could adopt a range of different orientations, and tilting a cone with respect to the horizontal widens the observed cone angle. Thus, the minimum observed cone angle for frozen cores was ∼19°, but a distribution of larger cone angles was also observed (18).

To test whether the cone angles of synthetic cores were quantized according to Eq. 1, oriented samples were prepared for TEM under dehydrating conditions, thereby flattening the cores onto an underlying Formvar/carbon grid (19). The measured angle (φ) for a cone flattened onto a horizontal surface is related to the original cone angle (θ) by Eq. 2.Embedded Image(2)Hence, if our model is correct, φ angles for the flattened synthetic cores should be quantized in increments of 30° (that is, a P = 5 configuration with a cone angle θ = 19.2° will flatten to give an observed cone angle φ = 30°; the other allowable cone angles will be 60°, 90°, 120°, and 150°). Measured φ angles did, in fact, fall into five groups that were integral multiples of 30° (Fig. 4). Narrow (P = 5,7) cones predominated, with approximately 90% of all of the cones examined exhibiting a ∼30° cone angle. This prevalence may reflect preferential nucleation of the narrow conical lattice or the reduced strain of the P = 5 cap configuration versusP = 2–4 configurations (14). Although rarer than the P = 5,7 cones, examples of all other predicted cone angles were also observed. The fact that the CA-NC/RNA complexes formed five (and only five) discrete cone angles, as well as flat sheets (that is, P = 0) and cylinders (that is,P = 6) (4), strongly implies that a hexagonal lattice symmetry underlies cone formation.

Figure 4

Cone angle distribution in negatively stained TEM images of synthetic cores. (A) Examples of synthetic cones with putative P = 1–5 caps (arrows). (B) Histogram showing the relative occurrence of different cone angles. Theoretical values for (flattened) cones withP = 1–5 cap configurations are denoted by arrows. Randomly selected, discrete cones or conical disclinations were measured from photographic images of fields of negatively stained CA-NC/TMV RNA particles (n = 160). Cones exhibiting obvious curvature or obscured by other particles were excluded from the measurements. Experimental cone angles and standard deviations for the five different groups were: 29 ± 4 (P = 5 cap), 59 ± 4 (P = 4), 94 ± 6 (P = 3), 123 ± 5 (P = 2), 148 ± 2 (P = 1).

Our model for the organization of the viral core requires that the CA protein shell (and possibly also the internal NC/RNA complex) adopt a curved p6 lattice. Lattices containing cages of hexameric rings have been proposed for various protein arrays thought to mimic immature retroviral Gag shells (although the precise arrangements of protein subunits differ between the models) (20). More recently, we have performed cryo–electron microscopic image reconstructions of hollow cylinders formed by the mature HIV-1 CA protein (4) and have shown that these cylinders are, in fact, composed of CA hexagons arrayed on helicalp6 lattices (21). Thus, these reconstructions provide direct experimental evidence that CA can spontaneously assemble into precisely the arrangement required by our model for the viral core.

Although the unusual conical shape of the HIV-1 core is characteristic of lentiviruses, other classes of retroviruses can have cylindrical cores (D-type) or spherical cores that can be centrically (C-type) or acentrically (B-type) positioned within the virion (22). Despite this morphological diversity, it is intuitively appealing to believe that these different shapes are created using similar underlying structural principles, particularly as retroviral Gag sequences share primary sequence similarities (23). An attractive aspect of our model is that the cores of all retroviruses could be composed of closed hexagonal lattices and differ simply in the spacing of their pentameric defects, thereby creating either spheres, cylinders, or cones.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: wes.sundquist{at}


View Abstract

Navigate This Article