Human and monkey immunodeficiency viruses are studded with ‘spikes’ that enable them to infect cells. Structural studies reveal that these spikes are tripod-like assemblies that cluster on the virus surface.
Newly produced HIV particles are in limbo between life and death, their fate determined by whether protruding structures — ‘spikes’ — on their surface make contact with receptors on the surface of white blood cells, in particular a subset known as CD4+ T cells. If a productive contact is made, a molecular sequence is triggered that results in the genetic material of the virus being injected into the target cell. Soon after, the viral genome begins to dictate events in the cell, leading to the production of many more virus particles, and full-blown infection gets under way. But if there is no productive contact between the viral spikes and target cells, then the spikes seem to lose their function and the HIV particles effectively die. The viral spike is thus central to HIV infection, making it a prime target for potential HIV vaccines, with the idea that blocking the spike–receptor contact using antibodies will prevent infection. On page 847 of this issue, Zhu et al.1 report a low-resolution (at about 3 nm) structure of the spikes and their distribution on the virus surface. There are some real surprises.
The HIV genome, and that of its monkey counterpart, simian immunodeficiency virus (SIV), is enveloped in a lipid membrane derived from an infected cell. Virally encoded spikes, often known as envelope spikes, are anchored in this membrane and protrude from the surface of the particle. They are composed of two interacting proteins: gp41, which spans the membrane, and gp120, which sits on gp41 at the surface of the virus particle. Both proteins are glycosylated (that is, have sugar groups attached); indeed, gp120 is one of the most heavily glycosylated proteins known, with about 50% of its molecular mass being carbohydrate. This sugar coating helps HIV to elude the immune system because the carbohydrate is derived from the infected cell and is poorly recognized by antibodies. But HIV also has other means of distracting the immune system, including loops of highly variable sequence that extend from the surface of gp120. These loops present an ever-changing target for antibodies and create a huge headache for vaccine designers.
Information about the envelope spike structure is limited, and it is not clear how many spikes there are on each virus or how they are distributed. The structure of the extracellular part of gp41 has been resolved, but only in a conformation adopted after the virus connects with the target cell — little is known about what gp41 looks like before this interaction2. The structures of isolated truncated gp120 ‘core’ molecules from HIV and SIV have also been solved3,4 and have been used to propose models for spikes4,5 in which gp120 and gp41 proteins assemble to form a trimer of heterodimers, that is (gp120–gp41)3. Indeed, electron tomography of fixed and stained preparations of HIV and SIV had revealed trimeric envelope spikes, but many other details were not visible6. Now Zhu et al.1 have studied unfixed, unstained, frozen hydrated HIV and SIV using cryoEM tomography, and their low-resolution structure provides valuable information (Fig. 1).
Diagrams tend to present HIV as fairly densely covered in evenly spaced spikes, an image probably based on early electron microscopy studies. Indeed, this is the picture Zhu et al. observed for an SIV strain that harbours a mutation enhancing the expression of spikes at the virus surface. These mutant particles displayed 73±25 spikes covering the entire surface. Normal HIV, however, had only 14±7 spikes, and these tended to cluster together.
Such clustering has several implications. First, it challenges a previous proposal that spikes are tethered in an ordered array of pores in the viral matrix underneath the envelope. Second, it suggests that the HIV surface may have yet another feature that evolved to avoid immune detection. Antibody responses to irregular displays of spikes are expected to be less than those elicited by dense, ordered arrays. Third, it is tempting to speculate that clusters of spikes might help the viruses to enter cells. Many researchers have argued that the interaction of multiple spikes with multiple target-cell receptors, namely CD4 and the chemokine receptors CCR5 or CXCR4, is needed to allow formation of a pore in the cell membrane and the transfer of genetic information from virus to target cell. Other researchers, however, have recently argued that just a single HIV spike may be sufficient to allow infection of a cell7.
A drawback of the Zhu et al. study, and of biophysical studies of HIV generally, should be noted. Under typical experimental conditions, most HIV particles are either not infectious or very slow to infect, and electron microscopy cannot distinguish between infectious and non-infectious particles. It could be that the only virus particles that do effectively infect are those with the highest number of spikes (up to 35 spikes per particle are described by Zhu et al.) and/or those particles with particular clustering patterns that are rarely seen.
Nonetheless, Zhu et al. provide some stunning images, particularly those of a tripod shape where a head composed mostly of three gp120 molecules balances on three gp41 legs, deduced by averaging data from a large number of SIV spikes. The image is not unlike depictions of menacing aliens that recur in films and books (for a selection try searching Google Image using ‘tripods’). The legs are well separated, in contrast to most representations so far. This may help to explain the observation that the external parts of gp41 near the membrane are accessible to neutralizing antibodies, and will encourage vaccine designers to target these regions. The gp120 molecules sit atop the legs, with a sugar-coated face upwards and the variable loops along the side of the spike, probably restricting antibody access to the crucial CD4-binding site.
What next? Independent corroboration of the tripod structure from different strains and differently treated HIV and SIV preparations is highly desirable. Atomic force microscopy studies have given a different view of the HIV envelope spike8, so determining which view is more likely in infectious particles is an important next step. Also, vaccine designers crave a high-resolution structure of an intact native HIV envelope trimer, and this is surely one of the most important unsolved structures in biomedicine.