Heavy metal protease takes a tiki torch to HIV assembly

A treatment-resistant reservoir exists in people infected with human immunodeficiency virus (HIV) who receive antiretroviral therapy (ART). Even after many years of ART, viremia predictably recurs within a few weeks if therapy is stopped. Pioneering studies demonstrated that resting CD4⁺ T cells isolated from people on ART harbor a ‘latent’ form of the virus. Such resting T cells do not release virus in their quiescent state, but if activated, they produce virus capable of spreading infection¹. Initially, the ‘latent’ reservoir was thought to be transcriptionally and translationally silent. However, it is now clear that a fraction of the reservoir is transcriptionally active² and, possibly, translationally active³. So, why do such resting CD4⁺ T cells not release HIV virions? An article by Liang et al. in this issue of Nature Immunology provides a potential answer; the authors propose a mechanism whereby an infected resting CD4⁺ T cell can express HIV proteins without release of the virus⁴. This mechanism, which involves the degradation of viral proteins by the metalloprotease TRABD2A before virions can be produced, suggests that the HIV reservoir could be transcriptionally and translationally active while not producing infectious viral particles.

The HIV replication cycle begins when the viral envelope glycoprotein (Env) binds to the receptor CD4 on the surface of a target cell. Binding of Env to CD4 is followed by engagement of a co-receptor, which leads to fusion of the virion and target cell. This membrane-fusion event delivers the viral capsid into the cytoplasm of the target cell. Then, the dimeric viral RNA genome is reverse-transcribed into a single double-stranded DNA copy as it translocates to and through the nuclear pore; the newly synthesized viral DNA then becomes integrated into the host cell genome. The integrated HIV provirus is then transcribed by the host-cell machinery to generate viral RNAs. Those viral RNAs are exported from the nucleus to the cytoplasm and are translated into viral proteins or are packaged into the next generation of viral particles during the process of virion assembly (Fig. 1a).

Fig. 1 |. HIV-1 replication is blocked at multiple stages in resting CD4⁺ T cells.

a, Major steps in the HIV-1 replication cycle, including attachment and entry of the virus, reverse transcription, import into the nucleus, integration of viral DNA, expression of viral genes, export of RNA, particle assembly, incorporation of Env, and particle budding and maturation. The steps in red font have been shown before to occur inefficiently in resting CD4⁺ T cells relative to that in activated CD4⁺ T cells (as discussed in the text). CXCR4 and CCR5, co-receptors; Pol, HIV polymerase; PIC, pre-integration complex. b, Proposed model for the degradation of Gag at the plasma membrane by the Tiki-family metalloprotease TRABD2A (by Liang et al.⁴). In this model, TRABD2A-induced degradation of Gag prevents the assembly of viral particles in resting CD4⁺ T cells and their release from those cells.

Liang et al. describe TRABD2A, a metalloprotease expressed on the plasma membrane of resting CD4⁺ T cells that degrades the precursor of the HIV type 1 (HIV-1) polyprotein Gag (group-associated antigen), which thus restricts the assembly of particles in and their release from nonactivated T cells⁴. The Gag polyprotein is the main HIV-1 structural protein that drives assembly of the virus. After being synthesized, Gag translocates to the host-cell plasma membrane, where it binds the phospholipid PI(4,5)P₂ (phosphatidylinositol-(4,5)-bisphosphate) in cholesterol-enriched plasma membrane microdomains often referred to as ‘lipid rafts’. At the plasma membrane, Gag undergoes extensive multimerization, promoted by Gag – Gag, Gag – membrane and Gag – RNA interactions; this leads to assembly of the immature viral particle, which contains over 3,000 molecules of Gag. The assembling Gag protein recruits components of the cellular ESCRT (endosomal sorting complex required for transport) machinery, which mediates release of the virus by catalyzing the fission reaction that severs the membranous stalk joining viral and cellular membranes. As the viral particle is released, the viral protease cleaves Gag at various sites, which triggers virion maturation⁵.

The genome of HIV-1, like the genomes of other retroviruses, is relatively small with limited coding capacity. HIV-1 has evolved to take advantage of host-cell machinery at nearly every step in its replication cycle. However, the host cell is not a passive vessel that welcomes the viral intruder; instead, in an attempt to combat viral invasion, the mammalian cell genome encodes various inhibitory factors (often referred to as ‘restriction factors’) that throw roadblocks into the path of replicating viruses. Several of these restriction factors that target HIV-1 are quite well characterized; for example, TRIM5α, Mx2, SAMHD1 and APOBEC3G target early steps in the viral replication cycle, and tetherin blocks the release of particles from the infected cell⁶. The newly proposed virus-inhibitory factor TRABD2A seems to attack the virus near the end of its replication cycle.

While activated CD4⁺ T cells can be readily infected with HIV-1 in vitro, a variety of barriers in resting CD4⁺ T cells have been reported that limit spread of the virus (Fig. 1). These include inefficient reverse transcription⁷ owing to low dNTP pools⁸, inefficient HIV gene expression⁹ and nuclear retention of viral RNAs¹⁰. However, the ultimate effect of these barriers is temporal. For example, after 3 days of infection, the amount of integrated DNA in resting T cells approaches that in activated T cells¹¹. Despite such barriers, roughly 10% of the resting T cells that contain integrated HIV DNA after in vitro infection also have detectable expression of HIV Gag as well as of other viral proteins¹². Curiously, these Gag-positive cells do not release virions efficiently¹¹. The study by Liang et al.⁴ reveals yet another barrier to spreading infection in resting T cells that could explain why these infected resting cells do not efficiently produce virus particles.

Liang et al. show that resting T cells have high expression of TRABD2A that is rapidly downregulated after T cell activation⁴. TRABD2A is a member of the Tiki family of metalloproteases that utilize Mn²⁺ for catalysis. The prototype of this family of enzymes, Tiki, has been reported to act extracellularly to cleave the Wnt signaling protein involved in development and cancer¹³. Liang et al. show that in resting CD4⁺ T cells that are infected with HIV-1, TRABD2A degrades the Gag proteins that are associated with the plasma membrane⁴. When TRABD2A is overexpressed in 293T human embryonic kidney cells, activation of TRABD2A with Mn²⁺ stimulates degradation of Gag, whereas inhibition of the enzyme with Co²⁺, Ni²⁺ or a metal chelator blocks degradation of Gag (Fig. 1b). Disruption of Gag’s binding to the plasma membrane prevents TRABD2A-induced degradation of Gag, which suggests that only membrane-bound Gag is susceptible to TRABD2A’s activity. Knockdown of TRABD2A in resting T cells results in a modest but significant increase in release of the virus, albeit not to the same extent as that induced by the activation of T cells. TRABD2A shows some specificity: it has no effect on expression of the transmembrane Env glycoproteins; it is active against Gag proteins of HIV-1 and the closely related viruses HIV-2 and simian immunodeficiency virus but is inactive against a more distantly related retrovirus (murine leukemia virus) and against hepatitis B virus. Liang et al. show that antibodies to TRABD2A block the degradation of Gag, although the mechanism for this is uncertain, as the antibodies do not penetrate the cell in which the Gag-degrading activity occurs⁴. Nonetheless, several of the monoclonal antibodies to TRABD2A enhance release of the virus without changing the amount of TRABD2A detected by immunoblot analysis.

The results described above represent a potentially important advance in the field of HIV latency, as they may explain the expression of viral proteins by resting T cells without the production of viral particles. In vivo relevance is suggested by the authors’ observation that HIV-infected people with greater viral loads have lower expression of TRABD2A⁴. Those findings are also consistent with in vivo studies showing a lack of correlation between the level of cell-associated viral RNA and the amount of viral release¹⁴. Because T cells in lymph nodes are more activated than those circulating in the blood¹⁵, it seems likely that lymph node T cells have lower expression of TRABD2A and thus release more HIV virions.

The intriguing data of Liang et al.⁴ raise various questions for future studies. The active site of the metalloprotease Tiki is the extracellular surface of the plasma membrane, yet Gag is localized on the inner leaflet. Likewise, it is unclear how antibodies that bind TRABD2A on the outside of the cell would block TRABD2A-induced degradation of Gag on the inside of the cell. Thus, it will be important to define the topology of TRABD2A relative to the plasma membrane. It will also be of interest to determine the levels of TRABD2A in T cells isolated from lymphoid tissues and in T cells after exposure to cytokines and chemokines such as IL-4 and IL-7 that are present at higher levels in vivo than in the in vitro studies. Defining the sequence specificity of Gag cleavage by TRABD2A is also of interest, since the Tiki protease cleaves its substrate at a single, specific site, while no cleavage products of Gag are identified in the study by Liang et al.⁴.

At one time, persistence of the HIV reservoir was thought to be related to its transcriptional and translational silence. However, evidence is accumulating that the reservoir can be expressed at least transcriptionally², and now there is a potential mechanism for translational ‘leaky latency’. The protease TRABD2A provides one potential mechanism by which a cell could express HIV proteins without releasing new virions. That in turn provides a mechanism by which a ‘latently’ infected cell might nevertheless be subject to immunological pressure. While it remains unclear what fraction of the intact HIV proviruses in the reservoir is transcribed and translated over time, that fraction should be susceptible to clearance by the immune system.