Immature dendritic cell-derived exosomes can mediate HIV-1 trans infection

Abstract

Immature dendritic cells (DCs) capture HIV type 1 (HIV-1) and can transmit captured virus particles to T cells. In this report, we show that HIV-1 particles captured by DCs can be transmitted to T cells by exocytosis without de novo infection. Captured HIV-1 particles were rapidly endocytosed to tetraspan protein (CD9, CD63)-positive endocytic compartments that were reminiscent of multivesicular endosomal bodies. Furthermore, some of the endocytosed virus particles were constitutively released into the extracellular milieu in association with HLA-DR1⁺, CD1b⁺, CD9⁺, and CD63⁺ vesicles (exosomes) and could initiate productive infections of CD4⁺ target cells. Surprisingly, the exocytosed vesicle-associated HIV-1 particles from DCs were 10-fold more infectious on a perparticle basis than cell-free virus particles. These studies describe a previously undescribed mechanism of DC-mediated HIV-1 transmission and suggest that virus particle trafficking to multivesicular endosomal bodies and subsequent exocytosis can provide HIV-1 particles captured by DCs an avenue for immune escape.

Immature dendritic cell (DC) subsets resident in peripheral mucosal tissues are presumably the first cells targeted by HIV type 1 (HIV-1) (1, 2). In vitro data have demonstrated that DCs can bind HIV for extended periods of time (3–5), and that bound HIV-1 particles can be subsequently transmitted to either quiescent or activated CD4⁺ T cells to induce productive infection (3). Furthermore, establishment of productive infection in CD4⁺ T cells is greatly enhanced in the setting of DC-CD4⁺ T cell coculture system, and virus particles captured initially by DCs have a greater infectivity potential than cell-free virus stocks (6). It has also been suggested that transfer of virus particles from DCs to T cells can occur in the form of an “infectious synapse” (7). But the sequence of events that mediates the transfer of captured virus particle from DCs to CD4⁺ T cells remains unclear.

The fate of HIV-1 particles after capture by DCs depends on the type of interaction between the virus envelope glycoprotein (env) and DC-specific attachment factor(s). Although DCs express CD4 and CCR5 at low levels, productive infection of DCs occurs infrequently (3, 5, 8–12). Furthermore, DCs do not express CXCR4 and have proven refractory to infection with CXCR4-tropic isolates of HIV-1 (13, 14). Alternatively, capture of virus particles by HIV-attachment factors, such as C-type lectin receptors expressed on varied DC subsets (15), results in endocytosis of virus particles. In this report, we describe a molecular pathway for virus particles endocytosed by DCs. As opposed to lysosomal-dependent antigen presentation pathways, a significant fraction of the endocytosed HIV-1 particles were targeted for exocytosis and released into the extracellular milieu. Exocytosed HIV-1 particles in DC supernatants were associated with vesicles that were ≈100 nm in diameter. Furthermore, these HIV-1-bearing vesicles also expressed HLA-DR1, CD1b, and the tetraspan proteins, CD9 and CD63, a proteomic profile suggestive of an multivesicular endosomal body (MVB) origin; hence, they were termed exosomes (16). Finally, we demonstrate that exosome-associated HIV-1 particles can initiate productive infections in CD4⁺ T cells and have a higher infectivity potential than cell-free virus particles. Our findings suggest HIV-1 trafficking through DCs subverts normal antigen-presentation pathways and can result in productive infection of interacting CD4⁺ lymphocytes.

Results

Constitutive Release of Captured Virus Particles into Cell-Free Supernatants. HIV-1 particles captured by DCs can be transmitted to T cells by a mechanism that remains unclear. Moreover, there are conflicting data on the requirement of virus replication in DCs before HIV-1 transfer to T cells (7, 8, 10). To better understand the mechanism of DC-mediated trans HIV-1 infection, we exposed DCs to CXCR4-(Lai) or CCR5-(NL4–3/Ba-L) tropic virus isolates at 37°C and 4°C and determined the fate of virus particles captured by DCs subsequent to trypsin treatment. DCs have high endocytic capacity, and endocytosed antigens are usually targeted for the antigen-presentation pathway (17). As expected, both CXCR4 (Lai)- (Fig. 1A) and CCR5 (NL4–3/Ba-L)-tropic (Fig. 1B) virus particles were rapidly endocytosed and protected from trypsin digestion when virus binding was allowed to proceed at 37°C. In contrast, HIV binding at 4°C prevented endocytosis of bound particles, and hence all bound particles remained sensitive to trypsin digestion (Fig. 1 A and B).

Fig. 1.

HIV-1 particles captured by DCs can escape degradation and are released into the extracellular milieu. DCs were exposed to HIV/Lai (A) or NL4 –3/Ba-L (B) virus particles at 37°C or 4°C for 2 h, washed, and incubated with trypsin. Cells were lysed immediately and the p24^gag content of the lysates determined by an ELISA. The data are reported as relative virus binding to DCs after trypsin treatment to that observed in the absence of trypsin and are the mean ± standard deviations of three independent experiments. DCs exposed to HIV/Lai (C) or NL4 –3/Ba-L (D) virus particles at 37°C were untreated or incubated with trypsin before culture for a period of 24 h. Cells and cell-free supernatants were harvested periodically and the p24^gag content determined by ELISA. The data reported are percent p24^gag left associated with cells or present in supernatants over time and are the mean of three independent cultures.

We next performed a kinetic analysis to test the stability of captured virus particles over time in DCs. Cells exposed to either CXCR4- (Lai) (Fig. 1C) or CCR5-tropic (NL4–3/Ba-L) (Fig. 1D) viruses were placed in culture and harvested periodically over a 24-h time period to determine the amount of p24^gag left associated with cells. Although 75% of the cell-associated p24^gag fraction (both CXCR4- and CCR5-tropic viruses) was degraded in DCs over the 24-h time period, some of the p24^gag protein remained associated with cells for at least 24 h after virus exposure. Surprisingly, over the same time period, there was a time-dependent increase in the release of p24^gag into the cell-free supernatants (Fig. 1 C and D), indicative of a constitutive release. To determine whether the source of p24^gag present in cell supernatants is endocytosed virus particles, HIV-exposed DCs were treated with trypsin and then placed in culture. Trypsin treatment of HIV-exposed DCs that resulted in a 50-fold reduction in surface expression of DC-specific intercellular adhesion molecule-3-binding nonintegrin (DC-SIGN) and CD11c (data not shown) had no impact on the amount of p24^gag present in the cell-free supernatants (Fig. 1 C and D), suggesting that endocytosed virus particles were the source of supernatant-associated p24^gag fraction. It should be noted that there is no de novo production of virus particles in DCs over the 24-h time period [discussed in Fig. 2, and previously published results (6)], rather the presence of virus capsids in supernatants represents endocytosed virus particles that were constitutively released after capture. (See Supporting Text, which is published as supporting information on the PNAS web site.)

Fig. 2.

Exocytosed HIV-1 capsids present in DC-derived supernatants are infectious. DCs were exposed to Lai-luc or SFV-luc virus particles (moi = 0.001), washed, and placed in culture by themselves or cocultured with T cells. Supernatants, harvested 24 h after virus exposure from DC-alone cultures, were incubated with Jurkat T cells. Cells were harvested 3 days postinitiation of culture, lysed, and lysates used for assaying luciferase activity. The data reported are from infections of three independent cultures ± standard deviations and are representative of four independent experiments.

HIV-1 Particles Released into Cell-Free Supernatants Are Infectious. Because the pulse–chase analysis detected HIV p24^gag protein in the supernatants, we next wanted to determine whether HIV-1 capsids released into the extracellular milieu after capture by DCs are infectious. We used luciferase-expressing HIV-reporter viruses, pseudotyped with either Lai or Semliki forest virus (SFV) env and hence capable of a single round of infection, to initiate infections. DCs infected with Lai-luc and cultured alone displayed low levels of luciferase activity, similar to mock infections (Fig. 2), suggesting that CXCR4-tropic HIV-1 replication in DCs is attenuated, in agreement with previously published results (6). Interestingly, cell-free supernatants harvested from DCs 24 h after exposure to Lai env pseudotyped HIV/luc reporter virus [Lai-luc; multiplicity of infection (moi) = 0.001] contained infectious virus particles (Fig. 2). Although the efficiency of trans infection in T cells mediated by DC-derived supernatants was lower than that mediated by virus-exposed DCs (DC-T cell cocultures), there was productive infection of Jurkat T cells with 24-h cell-free supernatants from Lai-luc-exposed DCs (Fig. 2). Furthermore, DCs infected with Lai-luc and placed in culture for 24 h retained an immature phenotype, as determined by FACS analysis (Fig. 6, which is published as supporting information on the PNAS web site). Because SFV infects cells via endocytic uptake in clathrin-coated vesicles and low pH-dependent fusion within the endosome (18), infection with SFV-env pseudotyped HIV reporter virus would result in productive infection of DCs. Indeed, high levels of luciferase activity were observed in SFV-luc-infected DCs that were cultured alone (Fig. 2). As a consequence, cell-free supernatants from SFV-luc-exposed DCs did not contain infectious particles and hence did not result in productive trans infection of T cells (Fig. 2). We conclude from these studies that HIV-1 particles endocytosed by DCs can be constitutively released into the extracellular milieu without establishing productive infection in DCs, and that virus particles that have trafficked through DCs remain infectious for T cells.

Microscopy Analysis for HIV-1 Particles in DCs. To gain insight into the trafficking patterns of HIV-1 particles in DCs, we performed thin-section electron microscopy on DCs acutely infected with Lai for 2 h at 37°C. Virus particles in DCs were present at the surface and in endosomal compartments near the plasma membrane (Fig. 3A). Interestingly, some of these endosomal compartments that harbored virus particles also contained multiple intraluminal vesicles [designated MVBs (19); Fig. 3B]. Furthermore, some virus particles in MVBs were found attached to the limiting membrane of the endosome (Fig. 3B). To further analyze the compartment in DCs that harbored HIV-1 particles after virus capture, we labeled infected cells with markers that define MVBs. DCs were infected with Lai for 3 h at 37°C, washed, and placed in culture for 1 h before adherence and fixation on glass slides. We observed two localization patterns for HIV-1 p24^gag protein in DCs: at the periphery underneath the plasma membrane and a deeper perinuclear localization (Fig. 3 D and G). Interestingly, there was some colocalization of internalized HIV-1 particles with the tetraspan proteins CD9 at the cell periphery (Fig. 3E) and CD63 in perinuclear compartments (Fig. 3H). In contrast, localization of internalized p24^gag protein in virus-exposed DCs was clearly distinct from nuclear pore complex proteins (Fig. 3K) (20). These results are in good agreement with recently published studies that have also demonstrated colocalization of DC-captured HIV-1 particles within tetraspan protein-positive endocytic compartments that are suggestive of MVBs (21).

Fig. 3.

Microscopy analysis for HIV-1 particles in DCs suggests localization of captured HIV-1 particles in MVB-like compartments. DCs exposed to Lai for 3 h were fixed either for thin-section electron microscopy (A and B) or for immunofluorescence microscopy (C–K). (A) An overview of an infected immature DC. Note the presence of virus particles at the surface (white arrowhead), in macropinosomes (black arrow), and in endosomal compartments (black arrowhead). B shows the presence of a mature HIV-1 particle found tethered to the limiting membrane of a MVB (arrow). Cells were coimmunostained for CD9 (C), CD63 (F), or 414 (I) and p24^gag (D, G, and J) and counterstained with DAPI. Merged images (red, CD9, CD63 or 414; green, p24^gag) are shown in E, H, and K. (Scale bar, 10 μM in C–K.)

Thus, localization of virus particles within MVB-like compartments after HIV-1 capture by DCs and the fact that DC supernatants contained infectious virus particles within 24-h after virus exposure without de novo infection (Fig. 2) led us to test the hypothesis that the mechanism of virus transfer mediated by DCs could involve exocytosis of vesicles (exosomes) bearing captured HIV-1 particles. Exocytosis has also been demonstrated as a mechanism for virus particle release from productively infected macrophages, especially for those viral capsids that assemble at internal endosomal membranes of the MVB (22–24). Alternatively, in the case of DCs, HIV capture and endoyctosis to MVBs could also result in an association of infectious particles with exocytic vesicles that express MHC Class II and tetraspan proteins on the vesicle surface and could be released to the extracellular environment upon fusion of MVB with the plasma membrane.

Released HIV-1 Particles from DCs Are Associated with Exosomes. To determine whether HIV-1 particles released from DCs are associated with exosomes, 24-h supernatants from HIV-exposed DCs were subjected to differential sucrose gradient ultracentrifugation and the proteomic composition of the membranes collected from the gradient fractions analyzed by Western blotting for MHC Class II (Fig. 4A) and for p24^gag protein by ELISA (Fig. 4B). Most of the MHC Class II molecules were found in fractions with a sucrose density of 1.12–1.15 g/ml (Fig. 4A), similar to the previously reported density profile for DC-derived exosomes in sucrose gradients (1.135 g/ml) (25). More importantly, HIV-1 p24^gag expression overlapped HLA-DR1α expression within these fractions (1.12–1.15 g/ml) (Fig. 4B). We could not detect any HLA-DR1α expression in HIV/Lai cell-free virus particles or in HEK293T cells used for the generation of virus stocks, whereas robust HLA-DR1α expression could be detected in DC (cell) lysate. Furthermore, we could detect electron-lucent vesicular structures in pelleted HLA-DR1α⁺ p24^gag+ fractions from the sucrose gradient (density 1.12–1.15 g/ml) by electron microscopy (data not shown).

Fig. 4.

Infectious HIV-1 particles in DC supernatants are associated with exosomes. (A) Vesicle-enriched fractions obtained by differential centrifugation and flotation on linear sucrose gradients were analyzed by Western blotting for the presence of MHC Class II. Densities of the different gradient fractions are indicated at the bottom. (B) The p24^gag content of the sucrose density gradient fractions was determined by an ELISA and plotted against the density of the fractions. (C) HLA-DR1⁺ vesicle fractions isolated from 24-h supernatants of HIV-luc-exposed DCs were incubated with T cells. Luciferase activity of lysed extracts was determined 3 days after infection. The data reported are the mean of three independent cultures ± standard deviations and are representative of three independent experiments. (D) DCs exposed to Lai-luc virus were left untreated or treated with trypsin for 10 min at 37°C before return to culture. Twenty-four-hour supernatants harvested from these cultures were incubated with magnetic beads conjugated to either normal mouse sera, HLA-DR1, CD1b, CD9, CD1d, or CD14, and the positively eluted fractions incubated with MAGI-CCR5 cells. The number of blue foci (indicative of the number of infectious virus particles associated with eluted fractions) was determined 2 days after infection. The data reported are the mean of three independent cultures ± standard errors of mean and are representative of two independent experiments. The absence of a histogram indicates that no infectivity was detected.

To directly test for vesicle association of infectious HIV-1 particles, we harvested cell-free supernatants from DCs previously exposed to pseudotyped HIV-luc reporter viruses for 24 h. Because exosomes originate from MVB, a defining feature of exosomes is the presence of MHC Class II molecules and tetraspan proteins on the vesicle surface (16). Hence, vesicles present in cell-free supernatants were isolated by an immunoisolation strategy initially using HLA-DR1-conjugated magnetic beads. Incubation of Jurkat T cells with immunoisolated HLA-DR1⁺ fractions from supernatants harvested from Lai-luc-exposed DCs resulted in productive infection of Jurkat T cells (Fig. 4C). In contrast, HLA-DR1⁺ immunoisolated fractions from supernatants of SFV-luc-exposed DCs (that results in productive infection of DCs) did not contain any infectious virus particles (Fig. 4C). Note that HIV-luc reporter viruses, generated from transient transfections of HEK293T cells, do not contain the HLA-DR1 antigen (Fig. 4A), and hence immunoprecipitation of virus infectivity from DC supernatants with anti-HLA-DR1 antibody is suggestive of a virus particle association with DC-derived vesicles. To determine whether the immunoisolation strategy using HLA-DR1-conjugated beads immunoprecipitated vesicles, DC-derived supernatants were incubated with HLA-DR1 or Protein G-conjugated magnetic beads that were 4.5 μM in diameter for ease of visualization. Thin-section transmission electron microscopy revealed large numbers of vesicles coating the surface of HLA-DR1-conjugated beads that could be clearly visualized under higher magnification (Fig. 7, which is published as supporting information on the PNAS web site). No vesicles were visualized with Protein G-coated magnetic beads (data not shown). There was some heterogeneity in vesicle size, but most were of uniform size, ≈100 nm in diameter (Fig. 7), reminiscent of MVB-derived exosomes (16, 26). Furthermore, the HLA-DR1-conjugated 4.5-μM-sized beads could immunoprecipitate HIV-1 infectivity from DC supernatants (Fig. 7).

To test for the presence of additional exosomal markers on the surface of isolated vesicles, cell-free supernatants from DCs previously exposed to Lai env-pseudotyped HIV-luc reporter viruses were incubated with magnetic beads conjugated to either HLA-DR1, CD9, CD1b, CD1d, CD14, or normal mouse sera. Incubation of MAGI-CCR5 cells with HLA-DR1⁺, CD1b⁺, or CD9⁺ immunoisolated fractions resulted in robust levels of virus infection (Fig. 4D). Treatment of DCs with trypsin after virus exposure had a negligible effect on the amount of virus particles precipitated by HLA-DR1, CD1b, or CD9-conjugated magnetic beads 24 h postinfection from cell-free supernatants, suggesting that endocytosed virus particles were the main contributors to the exocytosed virus fraction. In contrast, immunoisolation with magnetic beads conjugated to either mouse Ig or nonexosomal markers CD14 and CD1d (27) resulted in negligible precipitation of virus infectivity (Fig. 4D). Furthermore, immunoisolated HLA-DR1⁺ vesicle fraction derived from DC supernatants also contained CD63 tetraspan protein (Fig. 8, which is published as supporting information on the PNAS web site). Coexpression of HLA-DR1 and CD63 in the same vesicle fraction provides additional evidence for an exosomal origin for virus particle-bearing vesicles.

Infectivity Potential of Exosome-Associated HIV-1 Particles. It is possible that virus particles released into the extracellular milieu could have a diverse endosomal origin. To determine the relative amount of particles in DC supernatants associated with HLA-DR1⁺ exocytic fraction, we compared the infectivity potential of whole supernatants with that mediated by CD14- or HLA-DR1-immunoprecipitated fraction on MAGI-CCR5 cells. We could immunoprecipitate most of the infectivity associated with the 24-h supernatants from Lai-luc-exposed DCs by HLA-DR1⁺-conjugated magnetic beads (Fig. 5A), suggesting that HLA-DR1⁺ vesicles account for most of the infectious virus particles present in the cell-free supernatants of virus-exposed DCs. Furthermore, incubation of HLA-DR1⁺ supernatant fraction with an anti-gp120 neutralizing antibody, 2G12 (28) completely inhibited virus infection of MAGI-CCR5 cells (Fig. 5B), suggesting that virus particles are exposed on the surface of exocytic vesicles. Hence, vesicle-mediated HIV-1 trans infection still depends on HIV env–host cell receptor interaction. We compared the infectivity potential of Lai-luc-containing exosome fractions from DCs with that of Lai-luc virus supernatants generated from transient transfections of HEK293T cells. Focus-forming infectivity assays on MAGI-CCR5 cells revealed that HLA-DR1⁺ vesicle fractions isolated from 24-h supernatants from DCs contained 10-fold more infectious particles per nanogram of p24^gag than Lai-luc virus preparations generated from transient transfections of HEK293T cells (Fig. 5C). These studies implicate exosomes derived from DCs as an endosomal compartment that plays a significant role in mediating HIV-1 trans infection.

Fig. 5.

Exosome-associated HIV is more infectious on a per-particle basis. (A) Unfractionated (whole supernatant), HLA-DR1⁺, or CD14⁺ fraction from 24-h supernatants harvested from Lai-luc exposed DCs was incubated with MAGI-CCR5 cells, and the number of blue foci counted 2 days after infection. The data are reported as the relative number of infectious particles present in HLA-DR1⁺ or CD14⁺ fractions to that present in unfractionated (whole) supernatants and are the means of five independent experiments ± standard deviation. (B) CD14⁺ or HLA-DR1⁺ immunoisolated fractions from DC supernatants were left untreated or incubated with either anti-gp120 mAb (2G12, 10 μg/ml) or normal mouse serum before initiation of infection of MAGI-CCR5 cells. The data reported are the average of two independent experiments, each performed in triplicate, ± standard error of mean. (C) Cell-free virus derived from transient transfections of HEK293T cells or HLA-DR1⁺ fractions derived from 24-h supernatants of virus-exposed DCs was incubated with MAGI-CCR5 cells. The number of blue foci was determined 2 days after infection and the data reported as the number of infectious particles per ng of p24^gag (mean ± standard deviation of five independent experiments performed with three independently generated virus stocks). (D) A proposed model for DC-derived exosome-mediated HIV-1 trans infection.

Discussion

The results described in this report indicate that DC-mediated HIV-1 trans infection can be mediated by exocytosis of captured HIV-1 particles. Based on these data, we propose the following model, depicted in Fig. 5D, to account for exosome-mediated transfer of HIV-1 particles captured by DCs. After endocytosis, captured HIV-1 particles are targeted to MVB in DCs. Sorting of MVB cargoes into luminal vesicles via inward budding at the limiting membrane of the MVB could result in exposure of HIV and exosomal antigens (HLA-DR and tetraspan proteins) at the vesicle surface. Thus, an endocytosed virus particle present in the lumen of an endosome could be expressed on the surface of an HLA-DR1⁺ luminal vesicle (exosome) within a MVB. Alternatively, virus particles could acquire exosomal antigens while trafficking through the MVB through unknown mechanisms without physically associating with an intraluminal vesicle. Although some of the MVB-localized virus fraction is targeted to the lysosome and degraded, fusion of MVB with the plasma membrane could result in the release of captured virus particles along with exosomes and hence escape from lysosomal degradation pathways.

Prior studies with monocyte-derived DCs have suggested that cell-to-cell contact is necessary for stimulating efficient CD4⁺ T cell infection (6, 29), and that HIV-1 transfer from DCs to T cells does not depend on productive infection of DCs. We demonstrate that DC-derived supernatants contain exosome-associated HIV-1 particles early after virus capture, and that extracellular localization of HIV is independent of a productive infection of DCs. Although exosome-associated HIV was clearly infectious, efficiency of viral replication in T cells mediated by exosomes was lower than that achieved by DC–T cell cocultures (Fig. 2). It is possible that, as previously suggested, the DC–T cell milieu is more optimal for HIV replication than T cells alone (3, 5, 6). We speculate that exosomes released from HIV-exposed DCs could be captured and presented to T cells by uninfected DCs in culture (30, 31), and that exosomes presented in the context of DCs (as opposed to exosomes alone) are a much more efficacious mechanism of HIV trans infection. Furthermore, instead of the constitutive release of exosomes observed in the absence of T cells, interactions of T cells with DCs could result in maturation of DCs and trigger directed release of MHC Class II compartments containing endocytosed virus particles toward T cell targets, thus increasing the efficiency of trans T cell infection (32, 33).

The efficiency of exosome-mediated HIV-1 trans infection could also be impacted by the presence of multiple activation and adhesion molecules, including MHC Class II and CD86 as well as members of the tetraspanin protein family, such as CD9, CD63, and CD81 (16), in the exosomal membrane. The presence of activation and adhesion proteins in DC-derived exosomes may play an indirect role, either via enhancing adhesion of virus particle to the target T cell surface or via activating T cells (30, 31), thus increasing the probability of the establishment of productive infection. Alternatively, trafficking of virus particles through MVB could induce conformational changes in the virus envelope, such that the fusion capability of the HIV env glycoprotein is greatly enhanced. Regardless of the pathway of infectivity enhancement, initiation of productive infection by exosome-associated HIV-1 could conceivably increase the effective target population for virus in vivo. Hence, inhibition of exocytic mechanisms could provide an additional cellular target for antiretroviral therapy.

Virus particle association with exosomes was observed only with HIV env-pseudotyped virus particles, because pseudotyping with SFV env resulted in productive infection of DCs. This implies that DC-specific HIV attachment factors could play a role in targeting captured particles to MVB. It remains to be determined whether DC-SIGN (4) or other members of the C-type lectin receptors family expressed on mucosal DCs (15) can target captured HIV-1 particles to MVB.

Materials and Methods

Cells and Viruses. Primary monocytes were isolated from peripheral blood mononuclear cells of healthy donors by CD14-conjugated magnetic beads (Miltenyi Biotech, Auburn, CA), as described (34). CD14⁺ monocytes (3 × 10⁶ cells/ml) were cultured in RPMI/10% FBS in the presence of granulocyte–macrophage colony-stimulating factor (2,000 units/ml, Leukine, Immunex, Seattle) and rIL-4 (30 ng/ml, PeproTech, Rock Hill, NJ) for 6–8 days, at the end of which cells acquired a DC phenotype. HEK293T, Jurkat-CCR5 (T cell line) stably expressing CCR5, and MAGI-CCR5 cell line, a HeLa cell clone expressing CD4, CXCR4, CCR5, and HIV-LTR-β gal, have been described (35). HIV-1 molecular clones, Lai (CXCR4-tropic) and NL4–3/Ba-L (CCR5-tropic), have been described (6). Viruses capable of a single round of replication were derived by pseudotyping a luciferase-expressing reporter virus, HIV/Δnef/Δenv/luc⁺ [HIV-luc, luciferase gene inserted into the nef ORF and does not express env (36)] with either a CXCR4-tropic (Lai) (36) or a SFV (37) envelope glycoprotein. Virus stocks were generated by calcium phosphate-mediated transfections of HEK293T cells (6). All viruses were titered on MAGI-CCR5 cells, and an ELISA (Coulter) was used to determine the p24^gag content.

Virus Infections. DCs (5 × 10⁵ cells) were incubated with virus particles (10 ng of p24^gag) for 2 h at 37°C. Cells were washed four times with PBS to remove unbound virus and cultured in 96-well U-bottom plates (10⁵ cells/well). In some experiments, virus-exposed cells were incubated with trypsin (0.25%) for 10 min at 37°C to degrade any cell-surface-exposed virus particles, washed twice in serum containing media to inactivate trypsin, and cultured as above. Cells and cell-free supernatants were harvested periodically over a 24-h period and lysed in PBS/1% Triton X-100. An ELISA was used to determine the amount of HIV-1 p24^gag protein present in the lysates. For virus transfer assays, DCs (1 × 10⁵ cells) were exposed for 3 h at 37°C to pseudotyped luciferase-expressing lentiviruses (moi = 0.001), washed four times with complete media to remove unbound virus particles, and then cocultured with Jurkat/CCR5 cells (1 × 10⁵) in a 96-well U-bottom plate. Alternatively, virus-exposed DCs were cultured alone for 24 h. At the end of 24 h, cell-free supernatants from DCs were harvested by serial centrifugation steps (twice at 300 × g, 5 min; 800 × g, 5 min; 2,000 × g, 10 min) and used for the infection of Jurkat-CCR5 T cells. Cultures were lysed 2 days after initiation of infection of T cells, and luciferase activity of the lysates was determined as a measure of virus replication.

Electron Microscopy. Cells (3 × 10⁶), either uninfected or infected with Lai (300 ng of p24^gag) for 2 h at 37°C, were washed three times with cold PBS, fixed overnight in 4% paraformaldehyde/0.1% glutaraldehyde in 200 mM Hepes·KOH (pH 7.4) at 4°C, and processed for electron microscopy. Images were captured on a JEOL 1010 electron microscope.

Immunofluorescence. DCs, uninfected or infected with Lai virus particles (moi = 1) for 3 h at 37°C, were washed and allowed to adhere to poly(l-lysine)-coated glass chamber slides for 1 h at 37°C and fixed with 4% paraformaldehyde for 10 min at room temperature. Cells were rinsed with PBS, permeabilized with 0.1% Triton X-100, and incubated with mouse mAbs against CD9 (Clone M-L13, Becton Dickinson), CD63 (Clone H5C6, Becton Dickinson), and nuclear pore complex proteins (mAb 414, Covance, Richmond, CA) (20). Staining was visualized with affinity-purified Texas red-conjugated goat anti-mouse secondary antibody (Jackson Immunoresearch). Virus particles were visualized by staining with FITC-conjugated anti-p24^gag mAb (clone KC-57, Coulter) and ALEXA488-conjugated rabbit anti-FITC polyclonal antibody (Molecular Probes). Images of stained cells were observed and collected with an Axiovert 200M microscope (Zeiss) fitted with an Apotome for optical sectioning. Using appropriate filters and excitation wavelengths for each fluorophore, a Z series of 1.0-μm sections was collected for each sample. Images were analyzed by using axiovision (Zeiss) software.

Immunoisolation of Exosomes. DCs (3 × 10⁶ cells) were pulsed with HIV-luc pseudotyped with either Lai or SFV env (moi = 0.001) for 3 h at 37°C. Cells were washed four times with complete medium to remove unbound virus particles and resuspended in medium containing granulocyte–macrophage colony-stimulated factor and IL-4 (1.5 × 10⁶ cells/ml). In some experiments, virus-exposed cells were treated with trypsin before initiation of culture, as described above. Cultures were harvested 24 h later and cell-free supernatants isolated by serial centrifugations, as described above. Cell-free supernatants were then used for direct infection of Jurkat-CCR5 T cells and/or MAGI-CCR5 cells or used for immunoisolation of exosomes with a magnetic bead separation strategy. Briefly, cleared cell-free supernatants were incubated with mouse anti-human HLA-DR1 or CD14-conjugated magnetic beads (1:10 bead to supernatant volume ratio; Miltenyi Biotec) for 1 h at 37°C with intermittent agitation. In some instances, goat anti-mouse IgG-coated magnetic beads (Miltenyi Biotec) were precomplexed with mouse anti-human CD9 (Clone M-L13, Becton Dickinson), anti-human CD1b (Clone 3.1), anti-human CD1d (Clone 42.1.1), or normal mouse serum at 4°C for 20 min before incubation with DC supernatants. Magnetic bead-bound fractions were isolated from the rest of the supernatants by selective retention on equilibrated μMACS magnetic columns (Miltenyi Biotec). Columns were washed four times with PBS/2% FBS, and positive fractions were eluted with complete RPMI medium after detachment of the column. In some instances, eluted magnetic bead fractions were incubated with anti-gp120 mAb 2G12 (10 μg/ml) (28) or normal mouse serum for 30 min at room temperature before initiation of infections of Jurkat/CCR5 and/or MAGI-CCR5 cells.

Sucrose Gradient Centrifugation. Exosomes from HIV/Lai-exposed or uninfected DC supernatants (5 ml of culture medium from 2 × 10⁷ cells) were isolated by flotation on linear sucrose gradients (26). Supernatants were harvested by serial centrifugation (twice for 5 min at 300 × g, once at 800 × g for 5 min, once for 10 min at 2,000 × g, once for 30 min at 10,000 × g, and once for 60 min at 70,000 × g using a SW28 rotor). The 70,000 × g pellet was resuspended in 5 ml of 0.25 M sucrose/20 mM Hepes/NaOH, pH 7.2, and floated on a linear sucrose gradient (2.0–0.25 M sucrose/20 mM Hepes/NaOH, pH 7.2) in a SW28 tube for 15 h at 100,000 × g. Two-milliliter fractions were collected from the top of the tube, and the density of the fractions was determined by using an ABBE-type refractometer (Milton Ray, Rochester, NY). Membrane and proteins in each of the fractions were pelleted by diluting with 3 ml of PBS and centrifugation for 60 min at 200,000 × g by using an SW50.1 rotor (Beckman). The pellets were solubilized in PBS/1% Triton X-100. Samples from each fraction were removed for p24^gag ELISA, whereas the rest of the sample was further solubilized in nonreducing SDS/PAGE loading buffer and analyzed by SDS/PAGE and Western blotting with anti-HLA-DRα (Clone MEM-36, BioVendor, Candler, NC).