Research Article: The Achilles Heel of the Trojan Horse Model of HIV-1 trans-Infection

Date Published: June 27, 2008

Publisher: Public Library of Science

Author(s): Marielle Cavrois, Jason Neidleman, Warner C. Greene, B. Brett Finlay.


To ensure their survival, microbial pathogens have evolved diverse strategies to subvert host immune defenses. The human retrovirus HIV-1 has been proposed to hijack the natural endocytic function of dendritic cells (DCs) to infect interacting CD4 T cells in a process termed trans-infection. Although DCs can be directly infected by certain strains of HIV-1, productive infection of DCs is not required during trans-infection; instead, DCs capture and internalize infectious HIV-1 virions in vesicles for later transmission to CD4 T cells via vesicular exocytosis across the infectious synapse. This model of sequential endocytosis and exocytosis of intact HIV-1 virions has been dubbed the “Trojan horse” model of HIV-1 trans-infection. While this model gained rapid favor as a strong example of how a pathogen exploits the natural properties of its cellular host, our recent studies challenge this model by showing that the vast majority of virions transmitted in trans originate from the plasma membrane rather than from intracellular vesicles. This review traces the experimental lines of evidence that have contributed to what we view as the “rise and decline” of the Trojan horse model of HIV-1 trans-infection.

Partial Text

Dendritic cells (DCs) play a central role in initiating the adaptive immune response that counters pathogen infection. Immature DCs patrol the peripheral mucosal tissues, searching for unwanted intruders. Once a pathogen is sensed, captured, and internalized, DCs undergo a maturation process and migrate to the regional lymph nodes. Meanwhile, the internalized pathogens are processed into antigenic peptides, and co-stimulatory molecules are expressed on the cell surface, readying these professional antigen-presenting cells for effective T-cell stimulation [1]. To perform their key sentinel function, DCs express a repertoire of pathogen recognition receptors, including Toll-like receptors and C-type lectin receptors. Toll-like receptors relay pathogen alert signals to DCs through intracellular signaling pathways, culminating in both cellular maturation and cytokine production [2],[3]. C-type lectin receptors recognize specific carbohydrate structures on these pathogens and internalize them for degradation in lysosomal compartments, thus initiating the process of antigen presentation [4],[5].

DC-SIGN, the most-studied C-type lectin receptor that captures HIV-1 virions, is a calcium-dependent lectin that binds the HIV envelope with an affinity similar to that of CD4 [19]. The C-terminal domain of DC-SIGN interacts with unknown carbohydrate structures on gp120 [4],[22],[23],[24]. Expression of DC-SIGN in the lymphoblastoid cell line Raji (Raji-DC-SIGN), originally mistaken for the monocyte-type line THP-1, is sufficient to promote trans-infection of T cells [18],[25]. However, whether DC-SIGN is important in vivo remains controversial. Some reports indicate that trans-infection of T cells by DCs derived in vitro from monocytes (MDDCs) involves DC-SIGN [9],[18],[20],[26],[27], while other studies suggest the involvement of alternative C-type lectin receptors [16],[28],[29],[30]. DC-SIGN is expressed in vivo by some immature DCs located in mucosa within the lamina propria [31]. Nevertheless, further studies are needed to sort out the precise role of DC-SIGN versus other receptors in the capture of HIV-1 virions by these DCs.

As part of their normal function, DCs internalize pathogens and process proteins from these organisms into small antigenic peptides for subsequent presentation to CD4 T cells on MHC-II receptors. Typically, immature DCs display high levels of endocytic capacity while mature DCs are characterized by efficient antigen processing and presentation. Early in vitro and ex vivo studies supported the notion that DCs internalize structurally intact HIV-1 virions into large vacuolar structures [38],[39],[40]. At least some of these structures correspond to internal vesicles based on lack of co-staining with cationized ferritin [41] or ruthenium red [30], a small membrane-nonpermeable dye that binds to carbohydrates at the plasma membrane [42],[43]. Surprisingly, mature MDDCs harbor many more intact HIV-1 virions than immature MDDCs [41]. Numerous intact virions are found in mature MDDCs within large vesicles adjacent to the nucleus. In immature DCs, only a few virion-laden vesicles are detected, usually at the periphery of the DC. Virion internalization involves clathrin-dependent endocytosis based on visualization of virions within both clathrin-coated pits and internalized clathrin-coated vesicles [41],[44]. However, some virions remain detectable at the plasma membrane, notably within deep folds of the plasma membrane [30],[44] or between dendrites [45]. Surface-bound virions are also readily observed on LCs emigrating from vaginal epithelia [46]. In LCs isolated from skin biopsies and subsequently loaded with HIV-1, virions colocalize with Langerin at the cell surface and in Birbeck granules [9]. The latter are LC-specific cytoplasmic organelles likely involved in antigen processing [47].

Seeking to better understand the events involved in HIV-1 trans-infection, we tested the effects of soluble CD4 (sCD4). This agent selectively neutralizes gp120 on surface-bound virions while not altering internalized virions. To our surprise, sCD4 completely inhibited HIV-1 trans-infection [36], raising the possibility that the surface-bound virions, rather than internalized virions, represent the major source of virus for trans-infection. To exclude possible unappreciated effects of sCD4, we also inactivated surface-bound virions with pronase. Again, trans-infection was abrogated by the selective incapacitation of surface-bound HIV-1 virions in both MDDCs and CD34-derived LCs. These findings implicating surface virions in trans-infection (Figure 1) are also supported by two prior studies employing cell-impermeable inhibitors [53],[54], which prevent viral fusion by disrupting the viral envelopes. In the first study, amphibian-derived peptides effectively inhibited trans-infection when applied to HIV-1-loaded MDDCs and thoroughly washed out before incubation with target cells [53]. Van Coppernolle et al. raised for the first time the possibility that virions transferred from DCs originate from the cell surface. However, in an attempt to reconcile these findings with the strongly prevailing Trojan horse model, the authors proposed an alternative explanation, in which HIV-1 virions were targeted by small quantities of residual peptide remaining after the wash procedure at a time when vesicle contents are released into the synaptic cleft. In our studies evaluating possible “carryover” effects of sCD4, we find that sCD4 only inactivates virions accessible at the time of sCD4 treatment and not virions loaded after the application of sCD4 and its removal by washing [36]. In the other study, a novel topical microbicide candidate called SAMMA [55], which appears to interrupt HIV-1 gp120/gp41 fusion, also potently inhibited trans-infection of T cells by MDDCs [54] without gaining entry into cells. Together, these findings are consistent with our conclusion that trans-infection is primarily mediated by surface-bound rather than internalized HIV virions.

The Trojan horse metaphor became popular with the report of Steinman et al., which showed that the potency of DCs in stimulating trans-infection of CD4+ T cells [10]. This metaphor emphasized that DCs, a trusted party in the immune system, could actually carry an infectious agent to CD4+ T cells, leading to their destruction. The mechanism was refined later with the discovery of DC-SIGN as an HIV-1 gp120 binding protein [18]. The analogy to the Trojan horse became even more striking when Kwon et al. [20] suggested that HIV-1 virion internalization was required for trans-infection of T cells. The strongest support for this conclusion was the ability of both immature MDDCs and Raji-DC-SIGN cells loaded with R5-tropic HIV-1 virions to trans-infect T cells despite the inactivation of surface-bound virions with trypsin. While these experiments included appropriate controls indicating that the trypsin had efficiently inactivated surface-bound virions, the study did not exclude direct infection of immature MDDCs as the source of viral reporter expression. To take this analysis one step further, we performed similar experiments with reporter viruses encoding two distinct epifluorescent proteins, allowing clear distinction between signals deriving from DCs and T cells [36]. Again, we observed that inactivation of surface-bound HIV virions completely abrogated trans-infection by MDDCs. Direct infection of Raji-DC-SIGN cells could not readily explain the results obtained, as these cells are only weakly susceptible to HIV-1 infection [59]. However, two recent reports employing trypsin or pronase treatment of Raji-DC-SIGN cells suggest that trans-infection by these cells involves virions located at the cell surface [29],[30]. The initial study also concluded that Raji cells expressing DC-SIGN with a truncation of its cytoplasmic domain failed to trans-infect T cells. However, the large truncation in DC-SIGN might also have affected formation of the infectious synapse, which seems to require DC-SIGN or the effective migration of DC-SIGN-bound virions on the DC surface to the infectious synapse [27]. Of note, a subsequent study highlights how point mutations that compromise the internalization of DC-SIGN do not impair trans-infection [60].

The fact that virions are almost exclusively transmitted from the DC surface implies that virion internalization is chiefly a dead end for infectious virions. Several factors influencing the internalization of HIV-1 virions might affect their likelihood to trans-infect T cells. Such factors include the state of DC activation and maturation, the time elapsing between virion capture by DCs and contact with interacting T cells, and the nature of receptors that mediate binding of HIV-1 virions.

The finding that trans-infection of T cells by DCs involves primarily surface-bound virions argues that future research should be refocused on how HIV-1 hijacks the plasma membrane rather than the intracellular trafficking pathway as suggested by the original Trojan horse model. Unraveling how these virions are recruited to the infectious synapse is crucial. The presence of C-type lectin receptors in lipid rafts [70] suggests that HIV-1 virions likely reach the infectious synapse by “surfing” the surface of the plasma membrane of DCs on lipid rafts. Further studies to characterize the domain(s) of DC plasma membrane that serves as a source of infectious virions could reveal some similarities with the compartments in which HIV-1 buds in macrophages. Since that internalization seems to be mostly a dead end for infectious virions, elucidating how HIV-1 manages to remain at the cell surface poised for transfer in trans will be important. Finally, our new model suggests that in vivo transmission of virions captured by DCs to T cells is likely to be far more sensitive to attachment inhibitors and neutralizing antibodies than previously anticipated. Only time will tell whether this fact can be therapeutically exploited.