Biochemical and Biologic Characterization of Exosomes and Microvesicles as Facilitators of HIV-1 Infection in Macrophages

Abstract

Exosomes and microvesicles are cell membranous sacs originating from multivesicular bodies and plasma membranes that facilitate long-distance intercellular communications. Lipidomic, proteomic and cell biologic approaches uncovered processes by which the human immunodeficiency virus type-1 (HIV-1) can use exosomes and MV to facilitate its dissemination. Macrophage MV and exosomes were isolated by immunoaffinity and sucrose cushion centrifugation and characterized by morphologic, biochemical and molecular assays. HIV-1 was “entrapped” in exosome aggregates. Robust HIV-1 replication followed infection with exosome-enhanced fractions isolated from infected cell supernatants. MV and exosomes facilitated viral infection that was affected by a range of cell surface receptors and adhesion proteins. HIV-1 readily completed its life cycle in human monocyte-derived macrophages but not in CD4 negative cells. The data support a significant role for exosomes as facilitators of viral infection.

Introduction

Despite more than two decades of intense research, the realization of a protective vaccine for the human immunodeficiency virus (HIV) remains elusive (1, 2). This failure highlights the need for a better understanding of host-pathogen interactions and a better understanding of the mechanisms surrounding viral immune evasion. One such mechanism rests in the abilities of HIV to exploit the host’s cell own intercellular communication machinery to evade immune surveillance. This can occur through cell-to-cell conduits or tunneling nanotubes serving as cytoplasmic tunnels between cells. Such direct communications facilitate physical transport of macromolecules and cell constituents. Each such conduit and viral immune synapse serve as facilitators of intercellular viral spread among immune cells for microbial dissemination while shielding virus from humoral immunity (3). Nonetheless, even more opportunities are available for the cell to affect viral spread (3, 4). One mechanism is “Trojan exosomes.” The Trojan exosome provides retroviruses the ability to take advantage of the cell-encoded intercellular vesicle traffic (5, 6). If such pathway(s) are operative for viral dissemination, exosomes could facilitate packing and transport of HIV-1 constituents within intraluminal vesicles (ILV) independent of infection by mature progeny viral particles (7). Alternatively, exosomes may facilitate infection of progeny virus through their unique repertoire of cell surface receptors or by engaging progeny virus. To this end, we demonstrate herein that microvesicles (MV) and exosomes facilitate virus infection. HIV-1 exploits the surface properties of the exosomes to speed infection of progeny virus and in so doing camouflages the virus from immune surveillance. Most importantly, the process could facilitate viral access into cells of the innate and adaptive immune system and as such explain, in part, HIV-1 resistance to some neutralizing antibodies. We demonstrate that by surrounding itself with exosomes, HIV-1 can accelerate its infection and dissemination. This route of infection has implications to better understand how the virus can gain entry into cells. Such results offer new pathways for the control of viral replication and its dissemination in the infected host.

Materials and Methods

Antibodies and reagents

Human and mouse Abs to HIV-1 proteins including those to Gag (Gag^Pr55, p24, p17, matrix), gp41, gp120, gp160 and the TZM-bl reporter cell line were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH (Bethesda, MD). CD4^−ve Hela cells were purchased from ATCC (Manassas, VA). Microbeads conjugated to HLA-DR Ab were purchased from Miltenyi Biotec (Bergisch Gladbach, Germany). Rabbit CHMP4a, Tsg101, goat Rab11 Ab and early endosome antigen 1 (EEA1) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Chicken anti-rabbit Ab- and goat anti-mouse Ab-conjugated Alexa 488, 594, 647, donkey anti-goat Ab-Alexa 488, -galactosidase (β-gal) staining kit, 1,1′dioctadecyl-3, 3,3′, 3′-tetramethylindodicarbocyanine perchlorate (DiD), 3,3′dioctadecyloxacarbocyanine perchlorate (DiO), ProLong Gold anti-fading solution with DAPI, Dulbecco’s modified medium (DMEM) with Glutamax® all purchased from Invitrogen Life Technologies (Carlsbad, CA). Exoquick-TC™ was purchased from System Biosciences (Mountain View, CA). Rabbit anti-talin and anti-vinculin Abs, indomethacin and dynasore were purchased from Sigma-Aldrich (St. Louis, MO).

HIV-1 infection of monocyte-derived macrophages (MDM)

Experiments with human PBMC were performed in full compliance with the ethical guidelines of the National Institutes of Health and the University of Nebraska Medical Center. Monocytes from HIV-1, -2 and hepatitis seronegative human donors were obtained by leukophoresis, and purified by countercurrent centrifugal elutriation. Cells were differentiated and infected as previously described (8, 9). MDM were infected with the HIV-1_ADA strain at a multiplicity of infection of 0.1 infectious viral particles/cell.

Isolation of MV and exosomes

Uninfected and HIV-1 infected MDM used for MV and exosome harvesting were maintained in DMEM with heat inactivated human serum (HS) depleted of plasma MV and exosomes. HS was vacuum-filtered through a 0.22 μm porous membrane and then MV and exosomes were sedimented by ultracentrifugation at 100,000 × g for 3h. Supernatents were used to supplement the DMEM [10% (v/v)]. Culture fluids from HIV-1 infected MDM were collected on days 3–7 following viral infection and cleared of cellular debris by centrifugation at 500 × g for 30 min at 4° C. Supernatants were filtered through a 0.45 and 0.22 μm porous membrane and centrifuged at 100,000 × g for 1 h on a 20% sucrose weight/volume (w/v) cushion. Protein A/G paramagnetic beads conjugated to HLA-DR, CHMP4a, Tsg101, myosin II, Rab11, vinculin and talin Abs were added for 24 h at 4° C. MV and exosome populations were washed three times in PBS and isolated by magnetic separation. Reverse transcriptase (RT) activity on culture fluids performed after inoculation of cells with virus-infected supernatants, MV and exosomes, was measured as described (9). HIV-1 viral RNA were determined on fractions and measured by a COBAS Amplicor v1.5 system (Roche, Nutley, NJ). HIV-1 infection was alternatively assessed by immunocytochemical staining for HIV-1p24. Infected MDM culture fluids were collected, on alternate days, when RT was 100-fold background and > 20% of total cells expressed HIV-1p24 for exosome study. Fluids were cleared of cellular debris by centrifugation at 3000 × g for 15 min at 4° C. The fluids were filtered through 0.22 μm filter unit and Exoquick-TC™ added in 1:5 ratio vortexed gently and stored at 4°C. After centrifugation at 1500 × g for 30 min the exosome pellet was collected (10). This method was used for three components of the data collection. First, to obtain cryo-electron microscopy (cryo-EM) micrographs. Second, for cross validation of the infectivity assays in macrophages and Hela cells. Third, for Western blot assays to assess the relative quantities of viral core (HIV-1p24) proteins in the virion and exosome purified fractions and in the crude infected culture fluids.

Immunocytochemistry and confocal imaging of the vesicular fraction

The MV and exosome pellet generated as described above was resuspended in PBS containing 5% BSA and applied to poly-D-lysine- and fibronectin-coated LabTek® chamber slides (BD Biosciences, San Jose, CA) for 24 h at 4° C to allow binding of MV and exosomes to the slide surface. Immunostaining and confocal imaging of MV/exosomes and both MDM (1:1 uninfected and infected mixed cultures) and Hela cells were performed as previously described (3, 11, 12).

Endocytic uptake of MV and exosomes

The secretome fraction was pelleted as described above and depleted of HIV-1 using protein A/G paramagnetic microbeads conjugated to HIV-1 Env (gp120/gp160) Ab for 24 h at 4° C and subsequent magnetic separation. Supernatants were labeled with 2 μM DiD for 20 min at 37° C and isolated by centrifugation at 100,000 × g on a 20% sucrose cushion for 1 h at room temperature. The fluorescently-labeled fraction was added to MDM previously pre-treated with endocytic inhibitors (200 μM indomethacin, dynasore or a combination of both) for 3 h. Uptake of MV and exosomes were analyzed by flow cytometry (3).

EM assays

MV and exosomes to be examined by scanning and transmission EM (SEM and TEM) were pelleted as described above and vacuum-trapped on a 0.22 μm filter membrane. The membrane was cut and fixed in 2% glutaraldehyde and 2% paraformaldehyde in 0.1M Sorensen’s phosphate buffer for 3 h at room temperature. Samples were critical point dried, mounted on specimen stubs, sputter-coated with 40 nm of gold/palladium and observed with a FEI Quanta 200 SEM (Fei Company, Hillsboro, OR) operated at 25 KV. For examination by TEM, immune isolated MV and exosomes were fixed in 2% glutaraldehyde and 2% paraformaldehyde in 0.1M Sorensen’s phosphate buffer for 3 h. MV and exosomes were resuspended in warm agar and transferred onto cover slips. Following exposure to 1% osmium tetroxide for 30 min, samples were washed three times and dehydrated at different ethanol concentrations (50%, 70%, 90%, 95%, and 100%) for 10 min at each ethanol concentration. Coverslips were removed, and areas of MV and exosomes were selected for ultrathin sectioning. Sections were stained with uranyl acetate and lead citrate and observed in a Philips 410LS TEM operated at 80kV (Philips, Inc., Madison, WI). The exosome and virion morphology was evaluated by cryo-electron microscopy (cryo-EM) (13). A 3μl aliquot of the exosome-enriched preparations was applied to a glow-discharged C-flat holey carbon grid (Ted Pella, Redding, CA) and used for plunge freezing in liquid ethane (14) with a FEI Vitrobot MarkIII system or a home-made manual plunging device. The frozen grids were transferred to a FEI TF30 field emission gun 300kV transmission electron microscope (FEI Company, Hillsboro, OR) at liquid nitrogen temperature. Images were recorded at 59,000x nominal magnification using a Gatan 4 k by 4 k CCD camera (Gatan, Inc., Pleasanton, CA).

Lipid extraction and characterization

Lipid extraction and purification from MV and exosomes followed by mass spectrometry analysis was performed as previously described (15).

MV and exosome proteomes by liquid chromatography mass spectrometry (LC-MS)

For protein identification by LC-MS, protein precipitated from the immunoisolated MV or exosomes was loaded at 20 μg per lane onto a 4–12% gradient gel and separated by 1D SDS-PAGE using a NOVEX system (Invitrogen Life Technologies, Carlsbad, CA). Protein bands were detected by Coomassie staining and excised into 2- to 4-mm fragments. Gel Pieces were destained in 200 μl of 50% acetonitrile and 50 mM NH₄HCO₃ at room temperature for 1 h and then dried. For in-gel tryptic digestion, gel pieces were saturated with 2 μl of 0.5% trypsin (Promega, Madison, WI) and 50 μl of 10 mM NH₄HCO₃ and incubated for 12–16h at 37°C. Peptides were extracted with 0.1% trifluoroacetic acid and 60% acetonitrile. Peptide extraction and purification were performed with C₁₈ ZipTip^® (Millipore, Billerica, MA) and Proprep Protein Digestion and Mass Spec Preparation Systems (Genomic Solutions, Ann Arbor, MI). Extracted peptides were fractionated on microcapillary RP-C₁₈ column (New Objectives, Woburn, MA) and sequenced using ESI-LC-MS/MS system (ProteomeX System with a LTQ-Orbitrap mass spectrometer, Thermo Scientific, West Palm beach, FL) in a nanospray configuration. The acquired spectra were searched against the NCBI.fasta protein database narrowed to a subset of human proteins using Sequest search engine (BioWorks 3.1SR software from Thermo Scientific). The TurboSEQUEST search parameters were set as follows: Threshold Dta generation at 10000, Precursor Mass Tolerance for the Dta Generation at 1.4, Dta Search, Peptide Tolerance at 1.5 and Fragment Ions Tolerance at 1.00. Charge state was set on “Auto.” Database nr.fasta was retrieved from ftp.ncbi.nih.gov and used to create “in-house” an indexed human.fasta.idx (keywords: Homo sapiens, human, primate). Proteins with two or more unique peptide sequences (p< 0.05) were considered highly confident. Functional and subcellular classification of the identified proteins was performed using the UniProt protein database (www.uniprot.org) and IPA Ingenuity systems pathway (http://www.ingenuity.com/).

Cytokine arrays and protease degradation tests

Cytokine content of immunoisolated MV and exosomes was performed using Human Cytokine Ab Arrays 2.0 (Panomics, Fremont, CA) per manufacturer’s instructions. Exosomes were enriched by immunoisolation with protein A/G beads conjugated to Ab as per described above. Exosomes bound to beads were washed three times in PBS under a magnetic field and stored for 48h in PBS at 4 °C for a complete removal of any residual extra-exosomal cytokines. A portion of the exosomes was disrupted with a mild detergent and evaluated for presence of cytokines. Another portion of the intact exosomes was exposed to protienase K for 3h at 37 °C, washed with PBS by a series of ultracentrifugations and then disrupted with a mild reagent for cytokine array analysis. To test functional activity of the proteinase K and shielding effect of exosomal membranes, immunoisolated exosomes were disrupted prior to proteinase K exposure and cytokines detected by the Panomics array.

Time-lapse confocal imaging

MDM were labeled with 2 μM DiD and DiO per manufacturer’s instructions. The dye was removed and excess was quenched by washing three times with DMEM supplemented with 10% heat deactivated human serum at 37° C and centrifuged at 500 × g for 3 min. Cells were seeded on 35-mm glass-bottom plates (MatTek, Ashland, MA) and imaged after 15 min in 5 s time lapses at 37 °C and 5% CO₂. Movies were acquired using a 100X oil immersion objective in a Nikon TE2000-U utilizing a Swept Field Confocal Microscope (Nikon, Melville, NY) and a 63X oil lens with a 0.7 numerical aperture in a LSM 510 confocal microscope (Zeiss, Peabody, MA). Brightness and contrast were adjusted to improve visibility of particles released into the extracellular space. For quantitation of fluorescence overlap and membrane mixing, each frame was analyzed with ImageJ software, utilizing JACoP plugins (http://rsb.info.nih.gov/ij/plugins/track/jacop.html) to calculate Pearson’s co-localization coefficients.

Infectivity of MV and exosomes

Hela (CD4^−ve) and TZM-bl cells were cultured in high glucose DMEM supplemented with 10% FBS, 100 units of penicillin and 100 μg/ml of streptomycin. Cells were allowed to reach 70% confluence then detached by 25 mM trypsin/EDTA for 5 min at 37° C. Cells were then allowed to recover by culturing in 4 well poly-D-lysine LabTech chamber slides for 2 days (40% confluence). HLA-DR⁺ MV, exosomes, or a pool of isotype control antibodies bound to protein A/G paramagnetic beads were reconstituted in supplemented DMEM (20 μl of slurry/500 μl of medium/well) and added to TZM-bl for 24 h at 37° C. To test for true infection, TZM-bl was exposed to 200 nM efavirenz nanoformulations for 24 h prior to incubation with the MV or exosome populations. Untreated TZM-bl and those exposed to HIV-1 were used as controls. Cells were maintained in culture medium for an additional 48 h prior to β-gal detection using β-gal staining kit (Invitrogen, Grand Island, NY) per manufacturer’s instructions. Bright field images were acquired using a Nikon Eclipse TE300 microscope (Nikon).

Western blot assays

Samples treated with lysis buffer containing protease inhibitors were added to the loading buffer under reducing conditions followed by heating for 5 min at 95°C. Aliquots of those samples were added to a 4–12% Bis-Tris gel, electrophoresed, and proteins were transferred to a nitrocellulose membrane and detected using mouse monoclonal antibodies directed to (Abs p24 and gp41). Secondary antibody was AP124P (a goat anti-mouse Ab). The films were visualized using a Konica Minolta SRX-101A X-ray processor. New blots were used for each antibody used in assay.

Statistical analyses

MV and exosome size (diameter) were measured manually using LSM-Browser software (Zeiss). Scatter plots and bar graphs were generated using Prism (GraphPad Software Inc., La Jolla, CA). Two-tailed Student’s t-tests were used in study and the error bars were shown as “s.e.m”.

Results

MV and exosomes are increased in HIV-1-infected MDM

We first compared MV- and exosome-mediated intercellular transfers between uninfected and HIV-1-infected MDM. Equal ratios of cells were labeled with lipophilic dyes DiO (green, infected) and DiD (red, uninfected) and both cell types were incubated in teflon flasks for 180 min. (Fig. 1A). Release of cell components from the infected MDM followed a burst-like pattern (Movie S1). Transfers to uninfected cells occurred within 5 min of release and was completed within 16 min when infected and uninfected MDM were mixed at equal ratios; Fig. 1B and C, Movie S2). MV and exosomes are round and vesicular (Fig. 1D) and distinct from one-another and HIV-1 in size, structure and electron density. MV and exosome membranes show a high affinity for the surface of the target cells, and their adhesion to the plasma membrane was resistant to washes (Fig. 1E, F). These vesicular structures displayed a range of adhesion surface markers including the human leukocyte antigen (HLA-DR on MV), myosin II, Rab11, charged MVB protein (CHMP4a)/tumor susceptibility gene 101 (Tsg101) and vinculin/talin (exosomes; Fig. 1G–K). CHMP4a (ESCRT-III family member) and Tsg101 are resident proteins involved in biogenesis and extrusion of ILV (30–100 nm), which upon extracellular release were exosomes (16).

Figure 1. MDM communicate via secreted vesicular compartments.

A, Flow cytometry analysis of infected and uninfected MDM labeled with fluorescent green (DiO) and red (DiD) lipophilic dyes respectively shows increased membrane exchange among HIV-1 infected MDM at 0 and 6 min in suspension cultures. B, Time-lapse confocal images showing vesicle release (encircled structures) from infected MDM (DiO-labeled green) and their fusion with uninfected MDM (DiD-labeled, red; traced cell) within 4–6 min of the burst (scale bar, 10 μm) and completed at 16 min. C, Dynamics of lipid mixing (% lipid fusion on a single cell) and fluorophore overlap (% of cells changing color in the mixed population) between vesicular membranes with plasma membrane of the uninfected target cells. D, SEM image of vesicles released from infected MDM (scale bar, 1 μm). E, F, SEM images of MDM seeded on a filter membrane prior and post-migration towards the enriched vesicles in a Boyden chemotaxis chamber (scale bar, 1 μm; expanded boxes 10 μm). G–K, Vesicles released in the extracellular space (red arrows) by infected MDM contain various surface markers (scale bar, 10 μm; inset box, 5 μm).

Biochemical and ultrastructural characterization of MV and exosome populations

Due to similar centrifugation densities for virions, MV and exosomes, surface markers were targeted for immunoisolation from the infected MDM secretome. TEM revealed that the virus-infected MDM secretome consisted of MV, exosomes and mature progeny HIV-1 (Fig. 2A). Sub-fractionation of the secretome with protein A/G paramagnetic beads conjugated to HLA-DR, CHMP4a/Tsg101, myosin II, Rab11 and vinculin/talin Ab demonstrated that the HLA-DR⁺ fraction were MV averaging 300 nm in diameter and mature HIV-1 virions were ~125 nm; Fig. 2B–G. CHMP4a/Tsg101⁺, myosin II⁺, Rab11⁺ and vinculin/talin⁺ vesicle populations were exosomes (~60 nm in diameter) based on reported size and MVB markers (4, 17). Homogenous exosomes between 40–130 nm were seen as membrane vesicles after Exoquick-TC™ density purification as seen by cryo-EM (Fig. 2H). An individual exosome had two membrane leaflets averaging 80 nm in diameter (Fig. 2I).

Figure 2. MDM secretome contains distinct populations of MV and exosomes.

A, TEM images of the non-fractionated HIV-1-infected MDM secretome. Image demonstrates mature HIV-1 and vesicular compartments of various dimensions and electron densities. B, Immunoisolation of MV using HLA-DR-paramagnetic beads. C–G, TEM images of exosomes attached to the surface of a protein A/G paramagnetic bead (traced structure, ~15 μm in diameter) conjugated to antibodies to CHMP4a/Tsg101, myosin II, Rab11, and vinculin/talin (scale bar, 1 μm; inset box, 100 nm). H, Groups of exosomes that show an average size of 80 nm diameter (scale bar, 50 nm). and I, an maganified view of an individual exosome with clarified morphologic features.

We next evaluated relationships between HLA-DR, CHMP4a/Tsg101⁺, myosin II⁺ and HIV-1 constituents in the MV and exosomes. Clear co-localization was observed by confocal microscopic analyses between each of the cellular proteins and HIV-1Gag (Fig. 3A–C). White scale lines are 4.3, 5.0 and 6.0 μm (A, B and C), respectively. However, the possibility remained that the viral proteins were in fact encased in the MV and exosomes themselves or aggregated between progeny virions. To begin to address this, exosomes and virions were isolated from HIV-1 infected MDM by Exoquick-TC™ and analyzed by cryo-EM. The exosomes showed distinct morphology with size ranging, from 40–130 nm, diameter (Fig. 3D). Virions were readily differentiated from exosomes based on size (average 140nm in diameter) (18) and the presence of heavy electron densities within a membrane envelope (Fig. 3E). Evaluation of mixed exosomes and virions in infected cell culture fluids showed viral particles associated with a group of exosomes (Fig.3F, where “V” is a virus particle and “E” are exosomes) and entrapment of virus inside membrane coats with exosomes (data not shown).

Figure 3. Co-localization and morphologic evaluation of exosomes and virions.

A–C, MV and exosomes are associated with HLA-DR, CHMP4b-Tsg101, Myo II and HIV-1 Gag. Scale bar is 500 nM. D, Cryo-EM image of aggregated MDM exosomes with size ranges from 40–130 nm diameter observed after gradient centrifugation. Scale bar is 50nm. E, Cryo-EM images of the gradient-purified virions with an average diameter of 140 nm. F, Cryo-EM image of infected MDM culture fluids showing a virus particle (labeled “V”) associated with a group of exosomes (labeled “E”) that were membrane encased.

MV and exosomes share unique lipid and protein fingerprints with the endocytic compartments of their origin (4, 19). In particular, both HIV and exosomes utilize similar endosome-like domains for their maturation as was demonstrated during T cell infections (18). However, unlike HIV-1, the lipid content of exosomes and MV have not been previously identified (20). Therefore, their lipidomes were characterized to determine whether exosome lipid profiles were distinct from HIV-1 (Fig. 4A). Quantitative mass spectrometry analyses demonstrated that exosomes populations display showed unique lipid signatures compared to MV and HIV-1 (21). The MV fraction, which contained HIV-1 virions and vesicles, displayed a high content of HIV-1 lipids and others not characteristic for viral membranes. More specifically, lipids that enrich viral membranes including glycerophosphoserines [eg. phosphatidylserine (PS)], sphingomyelins (SM) and dihydrosphingomyelin (DHSM) were easily identified in the MV fraction; however, they were neither present nor were identified at background levels in the exosome populations (Fig. 4A). Interestingly, phosphatidylethanolamine-ceramide (PE-Cer), a sphingolipid that triggers budding of ILV in MVB, was identified only in the immune isolated exosome population using MVB markers (CHMP4a/Tsg101⁺) (22). MV and exosome also displayed distinct complex proteomes (Fig. 4B, Supplement Table S1). The protein identifications include accession numbers for UniProt (accessible at http://www.uniprot.org/) and postulated subcellular locations (accessible at http://locate.imb.uq.edu.au and http://www.uniprot.org/) as plasma membrane (PM); secreted (S); endoplasmic reticulum (ER); ribosomes (R); cytoskeleton (CSK); cytosol (C); mitochondria (M); endocytic vesicles (EV); extracellular matrix (ECM); nucleus (N); and not available (NA) which include proteins with no postulated localization. These attest to the complex nature and composition of the exosomes. Although mature virus was observed by TEM in the immune isolated exosomes, these were highly enriched for viral proteins (7–17% of their overall proteomes). These populations also displayed a high content of endocytic markers (13–25%), immune response (12–50%) and adhesion/cytoskeletal factors (21–35%; Fig. 4B, Supplement Table S1). Consistent with their role in the generation of a long-distance immune response, the identified MV and exosome proteins function primarily as regulators of cell-free antigen-presentation, T cell activation, chemotaxis and cellular migration (Supplement Fig. 1).

Figure 4. Characterization of the MV and exosome, lipidomes, proteomes and cytokines.

MV (HLA-DR) and exosomes (CHMP4a, Tsg101, Myo II, Rab 11, and Vinc/Tal) were immunoaffinity enriched. A, Identification and characterization of MV and exosome lipidomes. Exosomes and MV were immune isolated from pooled culture fluids of four independent cultures. B, Characterization of the proteomes from unfractionated MV and exosomes (Uninfected) and infected MV and exosome proteomes classified by function. Note abundance of immune response factors, adhesion and viral proteins (refer to supplemental table S1 for detailed protein content). C, MV and exosomes carry distinct chemotactic factors. Densitometry analyses of cytokines identified in uninfected MDM culture fluids (Uni) and immunoisolated MV and exosomes from HIV-1-infected MDM. D, MV and exosomes induce secretion of pro-inflammatory and migratory cytokines. Cytokines were detected in culture fluids of uninfected MDM after 24 h post-exposure to beads, uninfected unfractionated MV and exosomes (Uni) or immunoisolated MV and exosomes from HIV-1-infected MDM.

MV and exosomes carry chemotactic factors and induce pro-inflammatory and migratory proteins

MV and exosome-mediated intercellular communication and immune cell recruitment may be enhanced by chemotactic factors that they carry. The notion that cytokines may be released within “membranous bubbles” has not been previously reported. Establishment of an MV/exosome-mediated cytokine gradient would explain how immune cells “contact” these compartments during cell-free antigen-presentation or viral spread. Cytokine arrays demonstrated that immune isolated MV and exosomes carry IL-3, IL-4, IL-8, IL-17, leptin, and tumor necrosis factor alpha (TNF-α; Fig. 4C). Cytokines in these vesicles remain stable for days and are resistant to proteolytic degradation (Supplement Fig. 2). In addition to cell recruitment, these vesicles can also induce cytokine release, that includes IL-4, IL-6, IL-8, and MIP-1ß, by target cells (Fig. 4D). The intensity of cytokine response from the target cells is unique to each vesicle population. Interestingly, these pro-inflammatory cytokines have also been shown to influence the viral life cycle and enhance HIV-1 infection (23, 24).

MV and exosomes engage cells lacking HIV-1 receptors

The unique make up of exosome lipid membranes, presence of adhesion factors, endocytic and antigen-presenting molecules on their surface and use of chemotactic factors to recruit their targets may influence the dynamics of MV- and exosome-target cell fusion. These factors may also be exploited by HIV-1 to facilitate cell entry. Consistent with the exosome roles in antigen-presentation and viral spread, their lipids including phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylinositol (PI) and lysophosphatidylcholine (LysoPC) can activate naïve T cells and are elevated during HIV-1 infection (25). The presence of antigen presenting receptors and adhesion factors such as HLA-DRA1, HLA-DRB1, CD14, CD44R5, CD74, CD89b (Fc receptor), galectin-3, vitronectin, fibronectin, complement proteins, immunoglobulins, and regulators of endocytic recycling such as transferrin (Tfn), vinculin, talin on the surface of MV and exosomes may explain how these compartments facilitate cell attachment (Supplement Table S1). Presence of galectin-3 and fibronectin can enhance HIV-1 entry and infection up to 20 fold (26–28). Interestingly, carboxylate-modified fluorescent latex beads (80–500 nm in diameter) coated with Tfn are sequestered at the same rate as those coated with HIV-1 gp120 (Fig. 5A). Clathrin-dependent endocytic entry of DiD-labeled vesicles and their association with Tfn⁺ compartments raised the possibility of either alternate CD4-independent entry or facilitation of infection (Fig. 5B, C). Indeed, fluorescently labeled MV and exosomes were readily identified in the early endosomes of Hela cells lacking CD4 (Fig. 5D, E). The question remained, however, is this sufficient to establish infection or to facilitate it through more classic CD4 and co-receptor routes?

Figure 5. Vesicles enter target cells in a CD4-independent clathrin-mediated endocytosis.

A, Surface proteins on MV and exosomes may improve entry onto the target cells. Flow cytometry analyses of MDM exposed to fluorescent latex beads (80–300 nm) coated with adhesion and endocytic trafficking factors (complement, fibronectin, galectin, transferrin (Tfn) and recombinant HIV-1 gp120. Analyses shows Tfn⁺ beads are sequestered at similar levels to those coated with HIV-1 gp120. B, Vesicles enter MDM via clathrin-coated pits. The secretome of HIV-1 infected MDM was immune depleted of HIV-1 and labeled with DiD lypophilic dye. Uninfected untreated MDM, pre-treated with 200 μM indomethacin (disruptor of lipid rafts), dynasore (inhibitor of clathrin-coated pit formation) or a combination of both were analyzed by flow cytometry for uptake of fluorescently-labeled MV/exosomes. C, In target cells, vesicles are trafficked in Tfn⁺ recycling compartments. MDM were exposed to DiO-labeled (green) vesicles and Tfn conjugated to Alexa Fluor 647 (red). Confocal images show parallel sorting of vesicles with the Tfn receptor (scale bar 10 μm, inset box 5 μm). D, Confocal images of Hela cells immunostained for CD4 (green) and filamentous actin (rhodamine-phalloidine, red). Images demonstrate absence of CD4 in Hela cells. E, Confocal images of Hela cells exposed to the DiD-labeled MV/exosome fraction (red) and immunostained for early endosomes (endosome antigen 1, green). Arrows indicate sequestration of MV/exosomes into Hela early endocytic compartments (scale bars, 10 μm).

Exosome, viral constituents and mature progeny virus

The “facilitating” hypothesis postulates the existence of exosomes carrying viral constituents (5, 6). Ultrastructural imaging, lipidomic and proteomic analyses demonstrate that exosomes are distinct from mature HIV-1 in size (60 nm versus 125 nm diameters), morphology (high electron density throughout the sack versus presence of a central capsid) and possess unique functional fingerprints. An association of HIV-1 viral RNA was demonstrated in MV and different exosome populations entrapping virus (Fig. 6A). The MV fraction where HIV-1 was also detected had abundant viral RNA (> 10⁶ copies/μg of Ab used for immunoisolation). Exosome aggregates also displayed high viral RNA ranging from 45,000 copies/μg of Ab (equivalent to the genome of 22,500 mature virions) identified in CHMP4a/Tsg101 fraction to 515,000 copies/μg of Ab (~ 257,000 mature HIV-1) detected in vinculin/talin exosomes. Interestingly, beads conjugated to the isotype Ab pool displayed background level to no viral RNA excluding the possibility of non-specific HIV-1 binding to paramagnetic beads (Fig. 6A). Enriched MV and exosomes recovered from infected cell fluids also displayed robust RT activities, indicating the plausibility of presence of functional polymerases in these compartments (Fig. 6B). In addition HIV-1 p24/capsid was identified at higher levels in MV and vinculin/talin exosomes compared to the other exosome populations (Fig. 6C–E). Thus, to investigate the exosome virulence for HIV-1 we next tested immunoisolated fractions for infectivity in TZM-bl cells. The reporter cell line contains an integrated copy of the β-gal gene under control of the HIV-1 LTR. β-Gal detection (blue) in TZM-bl exposed to MV and exosomes encased virus supported the fact that these compartments carry viral constituents capable of viral genome integration and expression (Fig. 6F). Significant suppression (P0.0001) of β-gal expression following exposure to efavirenz, a non-nucleoside reverse transcriptase inhibitor confirmed infections by MV and exosome cell fractions (Fig. 6F).

Figure 6. Virions and exosomes and mixtures of both are infectious.

A, B, MV and exosomes contain high HIV-1 RNA copies and RT activity. C–E, Unprocessed forms of Gag (P55, P41 and P24) were identified in exosomes. Arbitrary unit values were generated by densitometry analysis of HIV-1 Gag immunostained western blots. F, TZM-bl reporter cells are infected (ß-gal expression, blue) by MV and exosome containing virion fractions. Arrows indicate beads (bound to exosomes) internalized by the TZM-bl. Viral infection initiated by exosomes in TZM-bl is susceptible and reduced by the non-nucleoside reverse transcriptase inhibitor, efavirenz used at 100 nm (**P < 0.0001; error bars ± s.e.m, n= 500 cells; Scale bar, 100 μm). The exosome fractions contain viral proteins and infectious virus. G, Evaluation of progeny virus production (RT activity) of exosomes, virus and control culture supernatant fluids following infection of MDM at equivalent protein concentrations. RT activity was evaluated every three days after MDM infection. Inset picture Western blots of 20x, 1x exosomes (E), virus (V) and control infected culture supernatant mixtures (MX).

It is, nonetheless, possible that exosome-MV still adhering on protein A/G paramagnetic beads could, themselves, facilitate cell-vesicle contact. This would reflect cell-virus interactions favored by centrifugation. Thus, validation tests were obtained using different isolation procedures that those used for the immunoisolated fractions in the TZM-bl cell infectivity assays. Exosomes (E) and virus (V) were purified by Exoquick-TC™ and gradient centrifugation, respectively. Crude cell fluid mixtures (MX, control culture supernatant fluids) served as controls. Each of the fractions was first tested for levels of viral proteins by Western blot tests. This showed readily identifiable HIV-1p24 in exosomes but at markedly reduced concentrations than were seen in MX and in virus fractions (Fig. 6G). HIV-1p24 protein within exosomes was in or around 50-fold less (see 1x and 20x concentrations) than what was detected with equivalent proteins loaded on gel and observed for purified progeny virus. These data, taken together, suggested that residual virus could be entrapped in clusteres of exosomes beyond what was contained inside the particles. Importantly, viral infectivity assays were performed after total protein content normalization for the exosomes, virus and supernatant mixtures and MDM exposures. The results showed equivalent level of RT activity in MDM culture fluids (measures of progeny virus production) for each of the infecting fractions (Fig. 6G). Interestingly, such high level viral infection by exosomes was seen despite the fact that this component contained 2% of core viral proteins as demonstrated by the Western blot. These data, raise the alternative possibility that exosomes facilitate infection in MDM through capturing viral particles in clusters and facilitating the engagement of virus-cell membrane or “other” receptor interactions.

The differences between facilitation and direct infection were next tested based on the use of the known viral receptors. Indeed, if such fractions were infectious and independent of progeny virus they would gain entry into cell by means independent of CD4. Thus, to evaluate the pathway of exosome infection, (CD4^−ve) Hela cells were used as indicators to test infectivity of virions, exosomes and exosome virus mixtures after protein normalization of each fraction. Assay for RT activity of culture fluids and HIV-1p24 cell staining showed no evidence of viral infection in any of the CD4^−ve Hela cells exposed to each of the fraction (Supplement Fig. 3). These results provide additional evidence that exosomes are not capable of CD4 independent infection. The data, taken together, best support the role of exosome or MV in viral infection.

Discussion

We demonstrate that MV and exosomes clearly facilitate the transfer of virus and viral constituents from infected macrophages to neighboring cells. This is significant as both are common in cells appearing as membranous vesicles at sites of viral assembly. They are released after fusion between MV and the plasma membrane and are readily secreted by a range of different cell types associated with endosomal vesicles. They bud inward using endosomal membranes, and their composition reflect their function. As they promote intercellular communications and, with bridging conduits, regulate immune responses and speed viral dissemination. A significant body of work has also focused on exosomes as carriers of tumor antigens in antigen presentation to T cells, in microbial infection, in immune surveillance and in neurodegeneration and neural development (5, 29–34). Following microbial infection, constituents of Mycobacterium tuberculosis and Toxoplasma gondii are found within exosomes and engage the innate and adaptive immune systems (35–39). Notably, studies of exosome biogenesis and function may lead to improved biomarkers and vaccine candidates for a range of disorders with currently limited therapeutic options.

It is also well accepted that retroviruses use endocytic pathways to coordinate their assembly and release. In this manner, they may exploit specific cellular processes for infection spread. This has led to the hypothesis that retroviruses, notably HIV-1, are aggregated within exosomes and are more easily transmitted from cell to cell. The overarching idea is that HIV-1 and exosomes have similar biochemical properties (40). Also viruses and exosomes have similarities in their biogenesis and composition, both either exploit the same cellular machineries for assembly and inter- and intra-cellular trafficking (5, 19, 41). Is exosome facilitated infection new for HIV-1? We think not. Nearly 25 years ago, we showed that HIV-1 could accumulate in intracellular vacuoles of infected macrophages (8, 42). Others demonstrated that such intracellular compartments are endosomes (43, 44). Virus can enter the cell through clathrin-mediated endocytosis then traffic from early to recycling and then late endosomes (3). Interestingly, a very similar method is used for antiretroviral nanoparticles (11). However, endosomes may take on MVB features. The importance of this cannot be overstated as MVB ultimately can fuse with lysosomes or with the cell surface and become exocytosed vesicles or ‘exosomes’ and emerge at sites of viral assembly (40, 45). Here, the isolated exosomes contain entrapped virus and facilitated infection. Mature virus was observed by EM in aggregated exosomes although their composition was 7–17% viral constitutes/total proteome. HIV-1 certainly takes advantage of exosomes to facilitate its infection and even help it escape immune surveillance and resist humoral antiretroviral responses. This is supported by vigorous proteomic, cell biologic, and virologic evidence provided in this report.

The association between virus and exosomes was also clear as 13–25% of its proteome contained endocytic markers with lesser compositions of immune response and adhesion/cytoskeletal factors. Although the aggregated exosomes contained less viral particles, infection was notably equivalent after protein normalizations. It is likely that multivesicular bodies and exosome entry into cells is influenced by a range of exosome and microvesicle receptors and surface adhesion proteins as shown in this report. Infection by aggregated exosomes, despite significantly reduced levels of infectious virus confirms the existence of exosome-facilitated infection. In any event, more work is required to elucidate how exosomes facilitate HIV-1 infection. Notably, future works are required in order to circumvent such infection pathways in order to design effective immunization strategies.

Abbreviations used in this article

β-gal

β-galactosidase

CHMP4a

charged multivesicular body protein 4a

DHSM

dihydrosphingomyelin

DiD

1,1′dioctadecyl-3, 3,3′, 3′-tetramethylindodicarbocyanine perchlorate

DiO

3,3′dioctadecyloxacarbocyanine perchlorate

ER

endoplasmic reticulum

ILV

intraluminal vesicles

MDM

monocyte-derived macrophages

MV

microvesicles

MVB

multivesicular bodies

PS

phosphatidylserine

SEM

scanning electron microscopy

SM

sphingomyelins

TEM

transmission electron microscopy

Tsg101

tumor susceptibility gene 101

RT

Reverse transcriptase

PBMC

Peripheral blood mononuclear cells