HIV-1 incorporation of host-cell–derived glycosphingolipid GM3 allows for capture by mature dendritic cells

Abstract

The interaction between HIV and dendritic cells (DCs) is an important early event in HIV-1 pathogenesis that leads to efficient viral dissemination. Here we demonstrate a HIV gp120-independent DC capture mechanism that uses virion-incorporated host-derived gangliosides with terminal α2–3-linked sialic acid linkages. Using exogenously enriched virus and artificial liposome particles, we demonstrate that both α2–3 gangliosides GM1 and GM3 are capable of mediating this interaction when present in the particle at high levels. In the absence of overexpression, GM3 is the primary ligand responsible for this capture mechanism, because siRNA depletion of GM3 but not GM1 from the producer cell and hence virions, resulted in a dramatic decrease in DC capture. Furthermore, HIV-1 capture by DCs was competitively inhibited by targeting virion-associated GM3, but was unchanged by targeting GM1. Finally, virions were derived from monocytoid THP-1 cells that constitutively display low levels of GM1 and GM3, or from THP-1 cells induced to express high surface levels of GM1 and GM3 upon stimulation with the TLR2/1 ligand Pam3CSK4. Compared with untreated THP-1 cells, virus produced from Pam3CSK4-stimulated THP-1 cells incorporated higher levels of GM3, but not GM1, and showed enhanced DC capture and trans-infection. Our results identify a unique HIV-1 DC attachment mechanism that is dependent on a host-cell–derived ligand, GM3, and is a unique example of pathogen mimicry of host-cell recognition pathways that drive virus capture and dissemination in vivo.

Of the estimated two and a half million new HIV infections that occur each year (1), approximately 85% are acquired at mucosal surfaces through sexual transmission (2). Dendritic cells are postulated to be one of the first cells to contact virus, and are believed to play a central role in establishing infection and subsequent viral dissemination (2). The majority of HIV/dendritic cell (DC) interactions result in a virus that remains associated with DCs in an infectious state (3). Captured virus particles can be transferred to HIV-1’s primary target cell, the CD4⁺ T cell (4) through formation of infectious synapses (5), a process known as DC-mediated trans-infection (3).

As the only viral protein present on the surface of the virion, HIV-1 envelope (Env) provides the primary mechanism for HIV-1 binding and entry. Although fusion of the virus particles to host-cell membranes require binding of HIV-1 Env to CD4 and coreceptor (CoR), several other Env attachment factors, such as dendritic cell-specific intercellular adhesion molecule-3-binding nonintegrin (3), syndecans (6), dendritic cell immunoreceptor (7), and galactosyl ceramide (8) are expressed by DCs and are believed to play key roles in binding and retention of virus particles without fusion. Despite the central importance of Env in mediating HIV-1 binding and fusion, there are on average only 7 to 14 irregularly spaced Env spikes on the surface of a virion (9). This low Env density results in a virion surface that is largely comprised of host-derived molecules and opens up the possibility that nonvirally encoded factors may also play important roles in virus interactions with target cells. Indeed, we have demonstrated that HIV-1 can bind to DCs in an Env-independent manner (10), and this mechanism of virus capture and trans-infection is greatly enhanced upon maturation of DCs (11).

HIV-1 is known to bud from specialized compartments of the cell membrane known as membrane rafts (12). Membrane rafts are defined as dynamic ordered assemblies with high levels of cholesterol, sphingomylein, glycosyphingolipids, and membrane proteins, such as tetraspanins and GPI-anchored proteins (13). Previous work in our laboratory has demonstrated a dramatic decrease in DC capture of HIV-1 that are derived from host cells treated with inhibitors to the glycosphingolipid (GSL) biosynthesis pathway (14, 15). As GSLs are a major component of membrane rafts, it suggests that glycoprotein independent capture of HIV-1 by DCs is mediated by a host-derived glycosphingolipid, which is incorporated into the virion as it buds from host-cell plasma membrane rafts. Here we demonstrate that GM3 is the host-derived GSL responsible for mediating the Env independent interaction between HIV-1 and mature DCs (mDCs).

Results

Ganglioside Depletion Results in Virions That Are Deficient for mDC Capture.

GSLs make up approximately 5% of the overall membrane lipid composition and are highly enriched in lipid rafts, and, interestingly, enriched in HIV-1 particles as they bud from the host cell (16, 17). Our previous studies had suggested a requirement for GSLs in capture of both HIV Gag-eGFP virus-like particles (VLPs) and infectious HIV-1 particles (14, 15). This requirement was observed with virus derived from HEK293T cells, monocyte-derived macrophages (MDMs), and peripheral blood mononuclear cells (PBMCs), wherein GSL depletion in the virus producer cell translated to virions with decreased mDC capture (Fig. S1). There are a large number of GSL variants, which can be further classified as gangliosides, asialogangliosides, and globosides (Fig. S2). Gangliosides have a terminal sialic acid (NeuNAc) residue, whereas none of the other GSL subsets are sialylated. To determine whether HIV-1 capture by mature DCs is dependent on sialylated GSLs, Gag-eGFP VLPs were treated at 37 °C for >4 h with a neuraminidase (NA) that removes α2–3-, α2–6-, and α2–8-linked NeuNAc from proteins and GSLs. NA-treated or untreated VLPs were probed for the presence of VLP-associated NeuNAc using an Alexa-594 conjugated wheat germ agglutinin (WGA). The level of NeuNAc detected on NA-treated VLPs was 2.8-fold lower than untreated VLPs (Fig. S3). The NA-treated VLPs show a marked deficit in capture, with a reduction of 58% relative to untreated VLPs (Fig. 1A). This finding suggests that the viral ligand responsible for glycoprotein independent mDC capture of VLPs is a ganglioside.

Fig. 1.

Gangliosides with α2–3 NeuNAc linkages are important for HIV-1 capture by mDCs. (A) Gag-eGFP VLPs were mock treated or treated with 0.5 units/μL α2–3, 2–6, 2–8 NA. (B) Gag-eGFP VLPs were derived from siRNA-treated HEK293T cells. NT, nontargeting; UGT8, galactosyl transferase; CERT, ceramide transfer protein; UGCG, glucosyltransferases, ST3, GM3 transferase. Capture of VLPs by mDCs was analyzed by FACS (A and B). Data are reported as percentage of eGFP⁺ mDCs normalized to NT-treated VLPs. (C and D) Ganglioside-deficient HIV_Lai was derived from HEK293T cells knocked down for NT, UGCG, or ST3. (C) Virions were labeled for p24^gag (green) and GM3 (red). Representative fields are shown and the average mean fluorescent intensity (MFI) of GM3 normalized to p24^gag ± SD is reported, *P < 0.001, one-way ANOVA with Dunnett’s multiple comparison. (D) Fold decrease of ganglioside-depleted HIV_Lai capture relative to NT-treated viruses by mDCs is reported. (E) Fold decrease in HIV_LaiΔEnv virus capture treated with 0.5 units/μL α2–3, 2–6, 2–8 NA or α2–3-specific NA relative to mock-treated viruses by mDCs is reported. All capture assays represent averaged data from a minimum of three donors, ±SEM, one-sample t test, *P < 0.05, **P < 0.01, ***P < 0.001.

To further verify that ganglioside reduction results in decreased virus capture, we used siRNAs to create ganglioside-deficient particles. HEK293T cells have a high level of endogenous gangliosides and provided an ideal system for targeted siRNA knockdowns to produce ganglioside-deficient VLPs. Because gangliosides are a subset of GSLs, the entire GSL family of sphingolipids was depleted from the cell by knockdown of glucosyl transferase (UGCG). This analysis was further refined by targeting GM3 synthase (ST3Gal5), which is responsible for all downstream ganglioside synthesis. As a comparison, the other two major sphingolipid families, sphingomyelin and sulfatides, were depleted by knockdown of ceramide transport protein (CERT) and galactosyl transferase (UGT8), respectively. Knockdown of each of the targeted GSL biosynthetic enzymes was verified by RT-PCR and depletion of the targeted sphingolipids from the cell membrane was assessed by FACS analysis for cell-surface–exposed GM1 (Fig. S4). Interestingly, a substantial decrease in VLP capture by mDCs was observed only with the virions produced from cells with knockdowns that impacted gangliosides (i.e., UGCG and ST3Gal5), but not upon knockdowns of sphingomyelin (CERT) or sulfatides (UGT8) (Fig. 1B).

To verify that the decreased mDC capture that we observed with ganglioside-depleted VLPs was recapitulated in the context of infectious virus, we performed a subset of the siRNA knockdowns described above and created replication competent HIV_Lai. The glycosyltransferases UGCG and ST3Gal5 were each targeted to produce ganglioside-depleted virus. Reduced incorporation of gangliosides in virus particles was verified by staining for the ganglioside GM3 and normalizing against the number of virions, determined by staining with an anti-p24^gag antibody (representative images and quantification shown in Fig. 1C, staining controls in Fig. S5). Because all other gangliosides are more complex derivations beyond GM3 (Fig. S2), the reduction of GM3 is a good indicator of generalized ganglioside depletion. In agreement with the VLP data, the ganglioside-depleted viruses were also significantly attenuated for capture by mDCs (>60% reduction) compared with HIV_Lai particles derived from nontargeting siRNA-transfected HEK293T cells (Fig. 1D). These results suggest that a virion-associated ganglioside plays a significant role in mDC capture of HIV-1 VLPs.

Removal of Virion-Associated α2–3-Linked Sialic Acid Results in Virus Deficient for mDC Capture.

Whereas all gangliosides possess a terminal NeuNAc, the number of residues and type of linkage varies. Relevant ganglioside linkages used in this study are indicated in Fig. S2. To further discern the type of ganglioside acting as a viral ligand, we treated virions with either an α2–3 NeuNAc-specific neuraminidase or a broadly acting α2–3, 2–6, 2–8 NeuNAc neuraminidase at 37 °C for >4 h. The HIV-1 glycoprotein is heavily glycosylated and contains several NeuNAc, which could bias these results. We therefore performed these assays with HIV_Lai lacking glycoprotein (HIV_LaiΔEnv). Virons treated with either of the two neuraminidases displayed a similar deficit in capture by mDCs (Fig. 1E), suggesting that of the many possible gangliosides that could be incorporated into the virion, the GSL(s) that possesses a terminal α2–3 NeuNAc linkage serves as the viral ligand necessary for capture by mDCs.

Enrichment of α2–3-Linked Gangliosides Results in Enhanced HIV-1 Capture by mDCs.

Because depletion of α2–3-linked gangliosides from the virion results in particles that are deficient for mDC capture, we predicted that enrichment of the virion with α2–3-linked gangliosides would result in enhanced mDC capture. Our analyses focused on GM3 and GM1, because both α2–3-linked gangliosides were previously reported to populate lipid rafts (13) and HIV-1 particles (12, 18). We used three independent strategies to test this hypothesis: exogenous addition of lipid, artificial liposomes, and ganglioside up-regulation during monocyte activation or differentiation into a macrophage.

Replication competent HIV_Lai were created from HeLa cells, which in contrast to HEK293T cells, have a low level of endogenous gangliosides. HeLa cells were transiently transfected with HIV_Lai-expressing plasmids and cultured in the presence of exogenous lipid, either GM3 or GM1, which integrated into the cell membrane before virus budding. Lipid enrichment of the cell was verified by FACS (Fig. 2A for GM3 and Fig. 2D for GM1). Direct staining of the virus particles verified that the lipid enrichment of the producer cell translated into a significant enrichment of the ganglioside in virus particles (Fig. 2B for GM3 and Fig. 2E for GM1). There was a significant enhancement in capture of both GM3- and GM1-enriched virus particles by mDCs compared with virus derived from untreated virus producer cells (Fig. 2 C and F, respectively).

Fig. 2.

Enrichment of HIV-1 particles with α2–3-linked gangliosides results in enhanced capture by mDCs. HIV_Lai particles were derived from HeLa cells treated with MeOH (mock), exogenous GM3 (+GM3) (A–C), or exogenous GM1 (+GM1) (D–F). (A and D) Enriched GSLs were detected on the cell membrane using (A) α-GM3 mAb or (D) CtxB and analyzed by FACS. Histograms for mock (gray line) and soluble (A) +GM3 or (D) +GM1 (black line) HeLa cells are shown. Gray fill shows secondary Ab-only control for A and unstained control for D. (B and E) GSL enrichment of virions was analyzed by staining for p24^gag (green) and (B) GM3 (red) or (E) GM1 (red). Representative fields are shown and the average MFI normalized to p24^gag ± SD is reported, **P < 0.001, Student’s t test. (C and F) Capture of HIV_Lai enriched with (C) GM3 or (F) GM1 by mDCs was analyzed by detecting p24^gag content in cell lysates by ELISA. Values represent the average from three donors and are reported as fold enhancement relative to mock-treated virus, ±SEM, *P < 0.05, **P < 0.001, one-sample t test.

Liposome Model of Virions Shows That the Inclusion of α2–3-Linked Gangliosides Results in Enhanced mDC Capture and Can Compete for Virion Binding.

The surface of an HIV-1 virion is complex, containing a myriad of host-derived sphingolipids, cholesterol, and membrane proteins in addition to the virally encoded glycoprotein (17). To further verify the role of GM3 and GM1 in mediating HIV-1 capture by mDC, we used a system wherein liposomes were constructed to represent a very basic and simplified model of a virus particle.

The lipid composition of the HIV-1 membrane has been previously analyzed quantitatively by mass spectrometry and shown to comprise 45.1% (molar percentage) cholesterol, 8.8% phosphatidyl choline (PC), 4.4% phosphatidylethanolamine (PE), 14.8% plasmalogen-PE (PI-PE), and 8.4% phosphatidylserine (PS) (16). We chose lipids with a simplified composition and comparable size to assemble liposomes that approximate the physicochemical properties of the virus membrane (Fig. 3A and Table S1). Liposomes were further given a fluorescent tag to enable ready detection by FACS analysis. These base-level liposomes comprising dipalmitoylphosphatidylcholine (DPPC), PS, and cholesterol are herein referred to as “blank” liposomes. We then created different versions of these liposomes by introducing an additional 1% of various phospholipids. In addition to the α2–3-linked gangliosides GM3 and GM1, we also created liposomes using the core phospholipid ceramide (Cer), galactosyl ceramide (Gal), to represent alternative phospholipid pathways, and tetrasialoganglioside GQ1b (GQ1b) to represent an α2–8-linked ganglioside with a complex branching structure. Mature DCs were challenged with equal amounts of liposomes and the level of capture was assayed by FACS analysis. Both the GM3 and GM1 liposomes were captured at a significantly enhanced level in comparison with blank liposomes or other derivatives (Fig. 3B).

Fig. 3.

Liposomes with α2–3-linked gangliosides are captured by mDCs and compete HIV-1 for binding. Lipid vesicles comprised 54% DPPC, 1% PS, 45% cholesterol, and a fluorescent tag (blank). Modified vesicles contained 1% of Cer, Gal, GM3 (α2–3 linked), GM1 (α2–3 linked), or GQ1b (α2–8 linked). (A) Representative transmission electron microscope negative stain of 1% GM3 liposomes at 25k magnification. (B) Liposomes were incubated with mDCs and analyzed for MFI by FACS. Data are normalized to blank liposomes and represent the average from at least three donors, ±SEM, *P < 0.001, one-sample t test. (C and D) Competition of virus (C) or VLP (D) capture by mDCs by increasing amounts of liposomes was analyzed by detecting p24^gag content in cell lysates by ELISA (C) or eGFP⁺ cells by FACS (D). Data shown represent the mean of three donors.

We next used the liposome panel to compete with virus capture by mDCs. On the basis of the capture profiles, we predicted that only the GM3- and GM1-enriched liposomes would competitively inhibit virus capture by mDCs. HIV-1 Env is known to bind GM3 (19); therefore, HIV_Lai devoid of glycoprotein (HIV_LaiΔEnv) and Gag-eGFP VLPs were used in these assays. Mature DCs were first treated with 0.1% sodium azide (NaN₃) to prevent membrane recycling and internalization of the liposome. Congruent with the capture data, both GM3- and GM1-enriched liposomes, but not Cer-, Gal-, or GQ1b-enriched liposomes, substantially inhibited mDC capture of both HIV_LaiΔEnv particles and VLPs (Fig. 3 C and D). These results suggest that mDCs encode recognition machinery that specifically binds α2–3-linked gangliosides such as GM3 and GM1 and that this mechanism is used to recognize HIV-1 particles even in the absence of Env.

Both GM1 and GM3 Are Up-Regulated upon Macrophage Differentiation, but only GM3 Shows Increased Virion Incorporation for Enhanced mDC Capture.

Although the previous assays demonstrated that modifications in the level of particle-associated GM3 or GM1 impacts the level of mDC capture, we sought to further verify this effect in a physiologically relevant cell type with naturally occurring differences in ganglioside levels. Whereas monocytes express a low level of membrane-associated GM3, this level is up-regulated either upon differentiation into macrophages (20) or upon immune activation (21). This dichotomy creates a tractable system wherein viruses are predicted to possess varying levels of GM3 when produced from differentially stimulated monocytes or macrophages. The monocytic cell line THP-1 can undergo macrophage differentiation upon stimulation with the TLR2/1 ligand Pam3CSK4 (21). We therefore used THP-1 cells that were either left untreated (GSL^lo) or Pam3CSK4 stimulated to induce macrophage differentiation (GSL^hi) before the production of virus particles.

As expected, Pam3CSK4 stimulated cells showed higher levels of both GM3 (Fig. 4A) and GM1 (Fig. 4B) than the untreated cells (Fig. 4 A and B). In agreement with data from exogenously enriched HIV_Lai particles (Fig. 2 C and F), viruses derived from Pam3CSK4-stimulated THP-1 cells (GSL^hi) were captured to a greater extent by mDCs than those derived from untreated THP-1 cells (GSL^lo) (Fig. 4C) and were transferred to T cells at an enhanced level (Fig. 4D). Furthermore, enhanced mDC capture of virus particles derived from Pam3CSK4-stimulated THP-1 cells was independent of the presence of exosomes in virus-containing supernatants (Fig. S6).

Fig. 4.

HIV-1 derived from Pam3CSK4-stimulated monocytoid cells have increased levels of GM3 and display enhanced capture by mDCs. (A and B) Cell surface expression of GSLs on THP1 cells was determined using (A) α-GM3 mAb or (B) CtxB and analyzed by FACS. Histograms for untreated (gray line) and Pam3CSK4-stimulated (black line) cells are shown. Gray fill shows secondary Ab-only control for A and unstained control for B. (C and D) HIV_Lai/YU2 derived from untreated or Pam3CSK4-treated THP-1 cells were incubated with mDCs and (C) analyzed by p24^gag ELISA for virus capture or (D) washed and cocultured with autologous CD4⁺ T cells. Cocultures were stained for cell surface expression of CD3 and intracellular p24^gag, and the number of dual-positive T cells was determined by FACS. Values were normalized to those observed with virus derived from untreated THP-1 cells and are reported as average fold increase from a minimum of four donors. ±SEM, *P < 0.05, **P < 0.01, one-sample t test. (E and F) HIV_Lai/YU2 derived from untreated (mock) or Pam3CSK4-stimulated (+Pam3CSK4) THP-1 cells were labeled for p24^gag and (E) GM3 or (F) GM1. A minimum of 10 fields were quantified for MFI of (E) GM3 or (F) GM1 normalized to that of p24^gag; average MFI ± SD is shown, *P < 0.0001, Student’s t test.

To verify that the observed cellular up-regulation in GM3 and GM1 translated into virions with increased levels of these gangliosides, virus particles were stained for GM3 or GM1. Surprisingly, although virions produced from GSL^hi THP-1 cells showed a higher level of GM3 than those produced from GSL^lo THP-1 cells (Fig. 4E), the level of GM1 was equivalent (Fig. 4F). This suggests that even though both GM3 and GM1 are up-regulated upon monocyte activation and both GM3 and GM1 are capable of mediating mDC capture (Figs. 2 and 3), only GM3 is incorporated into the virion at levels sufficient for this capture mechanism.

Blocking GM3-Dependent Interactions Results in a Decrease in mDC Capture of HIV-1 Particles, Whereas Neither Removal Nor Blocking of GM1 Shows Any Impact.

Our data show that, whereas both α2–3-linked gangliosides GM3 and GM1 are capable of mediating mDC capture when overexpressed on artificial liposome particles or on virions produced from cells exogenously enriched for lipids, only GM3 appears to be relevant in the context of native virus. To further address this point, we performed additional siRNA knockdowns in HEK293T cells, targeting the GM2 synthase (B4Galt), which is responsible for complex gangliosides and asialogangliosides (Fig. S2). The B4Galt knockdown acts downstream of GM3, leaving GM3 levels unchanged, although causing a decrease in GM1. As a comparison we also repeated the knockdown of the glucosyl transferase (UGCG), which impacts all glycosphingolipids, and GM3 synthase (ST3Gal5) to target all gangliosides. Whereas UGCG and ST3Gal5 knockdowns resulted in virions deficient for mDC capture, B4Galt knockdowns had no significant impact on the mDC capture of virions (Fig. 5A for HIV_Lai; Fig. 5B for Gag-eGFP).

Fig. 5.

Impairment of GM3-dependent interactions of HIV-1 particle results in decreased capture by mDCs. (A) HIV_Lai or (B) Gag-eGFP VLPs produced from siRNA transfected HEK293T cells were analyzed for mDC capture by (A) p24^gag ELISA or (B) % eGFP⁺ cells by FACS. NT, nontargeting; UGCG, glucosyltransferases; ST3, GM3 transferase; and B4, GM2 transferase. Data are normalized to NT-treated particles and reported as fold decrease in mDC capture. Values represent averaged data from at least two donors, ±SEM, *P < 0.05, one-sample t test. (C) HIV_LaiΔEnv or (D) Gag-eGFP VLPs were preincubated with increasing concentrations of α-GM3 Fab (solid gray lines) or CtxB (dashed black lines), before addition to mDCs. Isotype control Fab was tested at the highest concentration for each assay (open circles). Virion capture was quantified by (C) p24^gag ELISA or (D) % eGFP⁺ cells by FACS. All values were normalized to the (C) virus or (D) VLP-only condition. Average data from a minimum of two donors ±SEM are reported.

Because we were unable to detect a change in GM1 levels on virus produced from GSL^hi–THP-1 cells and knockdown of GM1 had no impact on mDC capture of the virions produced, we performed blocking experiments to further verify that GM3 has a significant role in mDC capture of HIV-1. Virus particles were preincubated with either cholera toxin B (CtxB) (to bind virion-associated GM1), or α-GM3 Fab (to bind virion-associated GM3). Both conditions were compared against a mock preincubation of media only, and an isotype control Fab was tested at the highest concentrations used for α-GM3 Fab. Whereas preincubation with increasing concentrations of CtxB had minimal impact on the ability of mDCs to capture HIV_LaiΔEnv particles (Fig. 5C, dotted line) or VLPs (Fig. 5D, dotted line), preincubation with increasing amounts of α-GM3 Fab competitively inhibited mDC capture of HIV_LaiΔEnv particles (Fig. 5C, solid line) and VLPs (Fig. 5D, solid line). The control Fab resulted in a modest decrease in capture of HIV_LAIΔEnv, although only α-GM3 Fab was statistically different from the mock condition. Of note, a higher concentration of Fab was required to block HIVΔEnv than Gag-eGFP VLP, likely as a consequence of the inherent differences in assembly and budding that exist between Gag-GFP VLPs and full-length virus (22) that could impact the relative amounts of GM3 incorporation.

These results demonstrate that, although GM1 is physically capable of mediating mDC capture when overexpressed, it is not present in virus at sufficient levels to play a substantial role in this process. Rather, virion-associated GM3 is the principal Env-independent ligand necessary for mDC-mediated HIV-1 capture and trans-infection.

Discussion

The results from this study demonstrate that mDCs can mediate HIV-1 capture through a particle-associated α2–3-linked sialic acid and that virion incorporation of GM3 mediates this interaction. The selective reduction of α2–3 NeuNAc from the virion results in a marked decrease in mDC capture (Fig. 1). In contrast, when virions are exogenously enriched for α2–3 NeuNAc (Fig. 2), or artificial liposomes are created that possess this residue (Fig. 3), capture by mDCs is dramatically enhanced. Differentiation of THP-1 monocytoids to macrophages up-regulates the expression of α2–3 NeuNAc gangliosides GM3 and GM1 (Fig. 4). Importantly, virions produced from these activated cells have increased levels of GM3, but not GM1, and demonstrate enhanced mDC capture and transfer (Fig. 4), which can be blocked by the addition of GM3-specific antibodies (Fig. 5).

Whereas previous studies have demonstrated that HIV-1 particles incorporate both α2–3 NeuNAc-linked GM1 and GM3 gangliosides (12, 18), a recent report suggests that assembly of HIV-1 Gag occurs at GM1-deficient lipid rafts (23). Our results support this recent finding and demonstrate that although GM1 can mediate mDC capture when added exogenously to virions (Fig. 2), in the absence of artificial lipid enhancement virus particles derived from activated monocytes are only enriched in GM3. Interestingly, virions are clearly capable of incorporating higher levels of GM1 when GM1 alone is overexpressed (Fig. 2). Therefore, the active exclusion of GM1 from virions derived from infected THP1 monocytes may reflect the heterogeneity of GM1- and GM3-containing plasma membrane rafts and utilization of GM3-containing rafts as the preferred virus assembly site. Hence, these results demonstrate that although mDCs can recognize both gangliosides GM3 and GM1, it is GM3 recognition that accounts for a significant portion of mDC-mediated capture of HIV-1.

GM3 is known to be present in the plasma membrane of macrophages and activated CD4⁺ T cells (20, 24), the primary targets of HIV-1 and therefore the primary cells from which de novo virions are produced in vivo. It has been well established that chronic immune activation is a hallmark of HIV-1 disease. Furthermore, HIV-1 patients exhibit a significant overexpression of GM3 (25) and also develop α-GM3 antibodies (26). It is very likely that viruses derived from activated monocytes or T cells in vivo would also result in enhanced incorporation of GM3 within progeny virions with concurrent enhancements in DC capture and dissemination in vivo.

The viral incorporation of GM3 may aid early transmission events of HIV-1 in at least two ways. First, the use of an Env-independent binding mechanism may provide a fitness advantage to the virus by limiting viral fusion and nonproductive cis-infection of DCs. Secondly, virus binding via GM3 may help to elucidate the way in which HIV-1 is able to transit through mDCs and retain infectivity. In this regard, HIV-1 particles share many properties with cell-derived exosomes (15, 27–29). Exosomes provide a mechanism of intercellular communication that can allow for the exchange of membrane proteins between cells or the delivery of signaling proteins, mRNA and miRNA (30). Intriguingly, exomsomes can also mediate cross-presentation by entering mDCs, limiting exposure to endosomal acidification (31), and ultimately presenting antigen to T cells in the absence of any fusion events. Because exosomes directly compete with HIV-1 for mDC capture (15) and both are enriched for GM3 (18, 32), we postulate that both types of particles bind to mDCs through GM3 and that this interaction leads to trans-dissemination (exosomes) or trans-infection (HIV). This trans-dissemination pathway, thus appears to be an intrinsic mechanism of mDCs that HIV has parasitized for facilitating its dissemination, while avoiding recognition as an agent that should be targeted for degradation.

Materials and Methods

Additional details are available in SI Materials and Methods.

Cells and Viruses.

Primary cells were cultured as previously described (14). THP-1 cells (ATCC; TIB-202) were cultured in RPMI, 10% (vol/vol) FBS, 1% (vol/vol) penicillin/streptomycin. Viruses were produced from HEK293T, HeLa, or THP-1 cells as described previously (14, 15, 29). Increased GM3 levels were obtained by Pam3CSK4 stimulation of THP-1 cells or exogenous enrichment of HeLa cells using solubilized GM3 or GM1 (Matreya). Decreased GSL levels were obtained through glucosyltransferase knockdowns in HEK293T cells. Sialic acid was removed from virions by treatment with α2–3, α2–6, α2–8 neuraminidase (NEB; P0230S) or α2–3 neuraminidase (NEB; P0728S).

Liposomes.

Liposomes 130 nM in diameter comprised 54% DPPC, 1% PS, and 45% cholesterol. Modifications included 1% Cer, GalCer, GM3, GM1, or GQ1b (Table S1) and fluorescently tagged with TopFluor cholesterol (Avanti Polar Lipids; 810255).

DC Capture and Transfer Assays.

Virus (40 ng p24^gag unless noted otherwise), VLPs (2 ng p24^gag), or liposomes (5 μL) were incubated for 1 h at 37 °C with mDCs and assayed for capture or transfer to autologous CD4⁺ T cells as previously described (14).

Competition Assays.

VLPs or HIV_LaiΔEnv were preincubated with increasing amounts of CtxB (Sigma; C9903) or α-GM3 DH2 Fab (derived from murine DH2 hybridomas (33), a kind gift from Sen-itiroh Hakomori, Pacific Northwest Diabetes Research Institute, Seattle, WA), or preincubated with increasing amounts of liposomes before performing mDC virus capture assays.