Exosome-Driven Antigen Transfer for MHC Class II Presentation Facilitated by the Receptor Binding Activity of Influenza Hemagglutinin

Abstract

The mechanisms underlying MHC class I-restricted cross-presentation, the transfer of Ag from an infected cell to a professional APC, have been studied in great detail. Much less is known about the equivalent process for MHC class II-restricted presentation. After infection or transfection of class II-negative donor cells, we observed minimal transfer of a proteasome-dependent “class I-like” epitope within the influenza neuraminidase glycoprotein but potent transfer of a classical, H-2M–dependent epitope within the hemagglutinin (HA) glycoprotein. Additional experiments determined transfer to be exosome-mediated and substantially enhanced by the receptor binding activity of incorporated HA. Furthermore, a carrier effect was observed in that incorporated HA improved exosome-mediated transfer of a second membrane protein. This route of Ag presentation should be relevant to other enveloped viruses, may skew CD4⁺ responses toward exosome-incorporated glycoproteins, and points toward novel vaccine strategies.

CD4⁺ T lymphocytes are activated through the recognition of Ag-derived peptides complexed to MHC class II (MHC II) molecules. Conventionally, Ag is acquired through uptake of exogenous material in the form of whole or partial pathogens, soluble proteins, or cellular fragments from necrotic or apoptotic cells (1). Internalized Ags are then unfolded and digested within the endosomal network by resident reductases and proteases. Once sufficiently disordered, epitopes are loaded onto MHC II molecules in a late endosomal compartment with the assistance of the chaperone H-2M. The H-2–IE^d-restricted site 1 (S1) epitope aa 107–119 of the A/Puerto Rico/8/34 (PR8) influenza hemagglutinin (HA) is generated in this manner (2, 3). The conventional pathway for generation of MHC class I (MHC I)-restricted epitopes is distinctly different in requiring delivery of Ag to the cytosol, digestion by the proteasome, and transport of resultant peptides to the endoplasmic reticulum via TAP for loading onto nascent MHC I molecules.

The dichotomy of exogenous Ag for MHC II and endogenous Ag for MHC I (4) is becoming increasingly indistinct. For example, there are now numerous examples of MHC II-restricted peptides that are generated from endogenous sources of Ag (5–7), and several processing mechanisms have been reported including macroautophagy (8), chaperone-mediated autophagy (9), and a class I-like proteasome/TAP-dependent pathway (10). The NA79 epitope (aa 79–93), derived from the influenza neuraminidase (NA) and also H-2–IE^d restricted, is generated via this last pathway (10). A major exception to the conventional class I pathway is the phenomenon of cross-presentation in which Ag is transferred from infected cells to professional APCs, primarily dendritic cells (DCs). This route appears to be critical for naive T cell activation in cases where the professional APC cannot directly acquire Ag (11). Additionally, it allows for escape from immunosuppressive effects imposed by the pathogen in the infected cell (12). Conventionally, cross-presented material is delivered to the cytosol for proteasome/TAP-dependent processing (13, 14), but there are also reports in which cross-presented Ag is confined to a phagolysosomal compartment where it is processed and loaded onto resident MHC I molecules, a decidedly MHC II-like scenario (13, 15).

A priori, there is no reason why Ag transfer would be restricted to MHC I. Limited study of class II-restricted transfer may be due to the general view that class II processing is limited to the endosomal compartment after uptake of exogenous Ag. Thus, Ag transfer would not be much different from direct acquisition. However, given the aforementioned endogenous MHC II presentation pathways, there is ample reason for exploring cross-presentation within the class II system. Whereas several studies have suggested that class II-restricted Ag transfer is limited (16–22), others have described transfer of soluble proteins, apoptotic bodies, preprocessed peptides, and exosomes as being viable mechanisms (1, 23–30). With the goal of better understanding the parameters of class II-restricted Ag transfer, we tested the widely disparate S1 and NA79 epitopes in this setting. Given the relative efficiency of class I-restricted cross-presentation, our expectation was that NA79 would be transferred whereas the classically presented S1 epitope would not. Instead, results were the opposite, revealing a mechanism for efficient Ag transfer of classically presented epitopes.

Materials and Methods

Mice

All mice used were purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained by the Thomas Jefferson University Office of Laboratory Animal Services (Philadelphia, PA). All experimental protocols were preapproved by the Thomas Jefferson University Institutional Animal Care and Use Committee.

Synthetic peptides

Synthetic S1 (HA 107–119), NA79 (NA 79–93), and SIINFEKL (OVA 257–264) peptides were purchased from Invitrogen (Carlsbad, CA). Peptides were pulsed onto APCs at 10⁻²⁵ M.

Cell lines, transfection, and recombinant retroviral transduction

Cell lines described in this study were primary skin fibroblasts derived from C57B6 or H-2M^−/− C57B6, which were then transduced with appropriate retroviruses. All lines were maintained in DMEM supplemented with 10% FCS and 0.05 mM 2-ME.

The retrovirus construct used was MSCV–CMVor MSCV–CMV–IRES–GFP with pCL–Ampho or pCL–Eco as the helper plasmid (generous gifts from Dr. Jianke Zhang, Thomas Jefferson University). Retrovirus was produced as previously described (10). Briefly, 293T cells were transfected with the appropriate MSCV construct along with the helper plasmid. Constructs used include IE^dα, IE^dβ, human CIITA, HA [H1 and H1-Rv6 (31)], NA, and S1–Tac. Supernatant collected was then either used or stored at −80°C for later use. For transductions, supernatant plus polybrene (10 µg/ml) was overlaid onto cells for 24 h and replaced with complete medium. Once confluent, cells were expanded and sorted based on surface IE^d or IRES–GFP expression using a MoFlo (Dako Cytomation, Fort Collins, CO) cell sorter.

For all transfections of fibroblasts, Genejuice (Novagen, Madison, WI) was used according to the manufacturer’s instructions. Transfected cells were used in Ag presentation assays 48 h posttransfection.

Primary DCs

For experiments involving the use of primary DCs, splenocytes from CB6F1 mice (H-2^b×d) were enriched for DCs as described previously (32). Briefly, 1E7 B16 melanoma cells, stably transfected with flt3L (generous gift from Christopher Norbury), were injected s.c. at the left flank. After 10–14 d, the mice were sacrificed, and whole splenocytes were used for the negative isolation DC (Invitrogen, Carlsbad, CA).

T cell hybridomas

S1- and NA79-specific T cell hybridomas, which express β-galactosidase upon recognition of peptide-MHC class II complexes, and the SIINFEKL: K^b–specific T hybridoma (B3Z), which does so upon recognition of peptide-MHC class I complexes, have been described previously (10, 33). T cell hybridomas were maintained in RPMI 1640 plus 10% FCS and 0.05 mM 2-ME. Activation was measured by detection of fluorometric β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside as previously described (2).

Ag presentation assays

For experiments where cells were infected with influenza, the strain PR8, subtype H1N1, was used. Cells were infected with 50 hemagglutinating units (HAU) per million cells for 30 min in PBS with 0.1% BSA. Postinfection, cells were washed three times with serum-containing medium to remove any unbound virus.

In some experiments, we separated B6-HA donor cells from B6IE^d acceptor cells using Transwell inserts with 0.4-µm pores (Corning, Corning, NY).

For drug treatments of donor cells (all from Sigma-Aldrich, St. Louis, MO), working concentrations were PMA at 0.1 µg/ml, LPS at 1 µg/ml, chloroquine at 0.4 µM, and NH₄Cl at 200 µM. Cells were co-incubated with the compounds for 16 h followed by three washes with PBS before co-incubation with acceptor cells and T cell hybridomas. In some cases, supernatants were collected for exosome harvest. Purified exosomes were quantified by protein concentration analysis using a Nanodrop spectrophotometer (Thermo-Fisher, Waltham, MA) at an absorbance of 280 nm.

To assess receptor binding activity of HA, blocking Abs were used. H28E23 (anti-HA), CMI-1.1 (anti-S1), and NA21C1 (anti-NA) were purified from hybridoma supernatant using recombinant immobilized protein A beads (Thermo-Fisher).Purified and unconjugated anti-TfR was purchased from BD Pharmingen (Franklin Lakes, NJ). Working concentrations of all Abs were 100 ng/ml to 10 ng/ml. In some cases, B6IE^d acceptor cells were treated with 200 mU NA (Sigma) to remove terminal sialic acid residues 2 h before the addition of B6-HA donor cells. To evaluate the degree of anti-HA–mediated neutralization and NA-mediated de-sialation, treated B6IE^d cells were infected with PR8, and NA79 presentation was determined.

Exosome purification

Exosomes were purified from 4-d- or 16-h-old supernatants of appropriate cells according to standard protocol (34). Briefly, supernatant was first centrifuged at 10,000 × g for 30 min to clear cells and cell debris. Subsequently, it was ultracentrifuged at 100,000 × g to generate an exosome pellet, which was then washed with PBS and re-pelleted at 100,000 × g. Pellets were then floated on a continuous 5–65% sucrose gradient by ultra-centrifugation at 24,000 rpm in a Beckman-Coulter (Brea, CA) SW41TI rotor (~76,000 × g)overnight. The following day, the gradient fractions were analyzed for exosome content, and exosome-containing fractions were dialyzed against PBS prior to storage at 4°C.

Coupling of exosomes to latex beads

Exosomes were prepared for analysis by flow cytometry as previously described (35). Briefly, purified exosomes were adsorbed to 0.4-µm-diameter aldehyde/sulfate latex beads (Invitrogen) for 15 min at room temperature. Next, the bead/exosome mixture was diluted with 1 ml PBS and incubated another 2 h. After this incubation, beads were washed and stained for flow cytometric analysis using anti-HA.

Electron microscopy

For transmission electron microscopy analysis, purified exosomes were adsorbed onto carbon-coated grids (Ladd Research Laboratories, Williston, VT) for 3 min. After adsorption, the samples were fixed for 1 min with 1% glutaraldehyde. The grids were then washed with a drop of ddH₂O and transferred to a drop of 1% uranyl acetate for 1 min for negative staining. Excess uranyl acetate was absorbed with Whatman paper no. 1, and grids were dried on filter paper for 2–3 h. Grids were then visualized on an Tecnai (FEI, Hillsboro, OR) 12 transmission electron microscope at 80 kV.

Immunoblotting

We followed standard techniques for immunoblotting. Specifically, we used exosomal fractions generated by sucrose gradient separation. Dialyzed fractions were concentrated using Vivaspin 6 columns (Sartorious, Goettingen, Germany), stained using anti-S1 or anti-flotillin (BD Pharmingen), and analyzed using SDS-PAGE.

Statistics

All experiments were analyzed using Prism (GraphPad, La Jolla, CA) software. Statistical significance was calculated using a paired Student t test with one- and two-tailed distributions. Differences were determined to be significant if the p value was <0.05.

Results

Cells expressing influenza HA can transfer Ag for S1 presentation

As previously reported (5, 10), S1 is presented from both infectious PR8 (iPR8) and UV-inactivated PR8 by B6IE^d, B6 skin fibroblasts stably expressing IE^dα, β, and human CIITA (Fig. 1A). This reflected the ability of this epitope to be generated in the endosomal compartment. In contrast, NA79 is presented from only iPR8, reflecting the requirement for cytosolic delivery and proteasome-dependent processing, properties that we speculated would allow it to be cross-presented as an MHC I-restricted epitope.

FIGURE 1. MHC II presentation via Ag transfer.

A (top panel), B6IE^d cells were treated with the indicated amounts of iPR8 or UV-inactivated PR8 (UVPR8). As positive controls, B6IE^d cells were pulsed with 10⁻⁵ M synthetic S1 and NA peptides. Upon treatment, APCs were serially diluted, and equal number of S1 and NA79 T cell hybridomas were then added to assess presentation. A (bottom panel), B6IE^d cells were transduced with PR8 HA or NA (B6-HA-IE^d or B6-NA-IE^d). As positive controls, B6-IE^d were pulsed with S1 or NA79 peptides. S1 and NA79 T cell hybridomas were used to assess Ag presentation. B, Primary B6 skin fibroblasts (not expressing IE^d) were infected with 50 HAU iPR8 and used as donor cells in an Ag transfer assay. Prior to coculture with B6IE^d cells as recipients, donor cells were washed. After serial dilution of donors and acceptors, S1 and NA79 T cell hybridomas were added and used to read out Ag presentation. C, Primary B6 fibroblasts stably transduced with HA and NA (B6-HA and B6-NA) were cocultured overnight with B6IE^d cells or primary DCs and with S1 T cell hybridomas. D, 293T cells were transfected with genes encoding OVA, SIINFEKL-cytosolic-NP, or HA and incubated with B6IE^d. Presentation of NA79, S1, and SIINFEKL were assessed with the appropriate T cell hybridomas. All experiments are representative of three independent experiments. *p < 0.05. MUG, β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside.

Biosynthesized HA does not traffic efficiently to the endosomal compartment (36). Thus, because transduction of APCs with an NA-expressing retrovirus, just like infection with iPR8, should result in presentation of NA79, we expected that transduction with an HA-expressing retrovirus would result in minimal or absent presentation of S1. When this experiment was performed in B6IE^d cells, both epitopes were presented efficiently (Fig. 1A). This led us to consider the possibility that some form of Ag transfer is possible in both cases. To test this, we infected H-2–IE^d-negative B6 fibro-blasts with iPR8 and co-incubated these “donor cells” with “acceptor” B6IE^d cells or primary DCs extracted from H-2^b×d mice. Despite its class I-like nature, NA79 was poorly presented under these conditions, whereas the classically presented S1 epitope was potently presented (Fig. 1B). The remainder of the experiments reported in this study focused on the latter finding.

A caveat to using influenza-infected cells as donors is the possibility that nascent virions bud from the surface and infect neighboring acceptor cells, thereby giving the false impression of Ag transfer. This seemed unlikely because the NA79 epitope would then be presentable via the direct, endogenous route. Nevertheless, to rule this out, donor cells were infected, allowed sufficient time for de novo protein synthesis of viral proteins, and then UV-irradiated before coculture with acceptor cells (Supplemental Fig. 1). UV irradiation of donor cells had no impact on S1 presentation by the acceptor cells indicating that reinfection was not the basis for Ag transfer.

In addition, because transfer of apoptotic bodies has been reported to mediate class II-restricted Ag transfer (1), and influenza infection can lead to apoptosis (37), we considered the possibility that apoptosis might have facilitated the presentation of the S1 epitope and/or inhibited the presentation of the NA79 epitope (38, 39). To investigate this, we stably transduced donor B6 skin fibroblasts with the HA or NA genes, a procedure that produced no evidence of cell death (data not shown). Upon coculturing these donors with primary DCs or B6IE^d cells as recipients, we observed that S1 was efficiently presented (Fig. 1C), whereas NA79 was once again poorly presented (Supplemental Fig. 2). Thus, the disruptive effects of viral infection are not essential for S1 presentation via Ag transfer, nor can they explain the near lack of NA79 presentation.

We also considered the possibility that NA79 was not efficiently transferred because our system is unable to facilitate true MHC I cross-presentation. Thus, we transfected donor H-2K^b–negative 293T cells with genes encoding HA, NA, OVA, and a version of influenza nucleoprotein that contains the OVA_257–264 H-2K^b–restricted epitope. Upon coculture of these transfectants with B6IE^d cells, we observed robust cross-presentation of OVA_257–264 and presentation of S1 but not NA79 (Fig. 1D).

The transferred material is a soluble factor dependent upon processing by the acceptor cell

In exploring the basis for Ag transfer, we first eliminated the possibility that the transferred material was simply S1 peptide released from the donor cells and loaded onto acceptor cell IE^d at the plasma membrane as previously described (27–30). Ultracentrifugation of the medium from PR8-infected B6 cells cleared nearly all of the stimulatory activity, and the effect was complete for medium from the B6-HA transfectants (Fig. 2A). Furthermore, loading of synthetic peptide has been reported to be primarily H-2M independent (40), and, as shown in Fig. 2B, absence of H-2M in the acceptor but not the donor cells completely abrogates presentation. We also observed that fixation of acceptor cells by paraformaldehyde significantly impairs their ability to present S1 from transferred Ag but not synthetic peptide (data not shown).

FIGURE 2. Ag transfer is dependent on pelletable material and processing by the acceptor cell.

A, Supernatants from PR8-infected B6 cells (left panel) or B6-HA cells (right panel) were subjected to ultracentrifugation. The resuspended pellets and the supernatants were then added to B6IE^d cells and S1-T cells. Whole supernatant and cells were added as positive controls, and data are depicted with uninfected B6 subtracted. As an additional control, media were “spiked” with synthetic S1 peptide prior to ultracentrifugation and observed to be fully stimulatory (data not shown). After ultracentrifugation, supernatant from infected B6 cells was confirmed to be clear of virions using a flow cytometry-based infectivity assay (data not shown). B, Primary skin fibroblasts from B6 mice or H2M^−/− mice were transduced for stable expression of HA or NA. These cells were used as donors with B6IE^d or H2M^−/− B6IE^d as acceptor cells in an S1-specific presentation assay. All experiments are representative of three independent experiments. MUG, β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside.

Pulsing B6IE^d cells with conditioned media from B6-HA and B6-NA revealed that the transferred material is not cell associated and transmitted without direct contact between donor and acceptor (Supplemental Fig. 3A). To solidify this conclusion, B6IE^d acceptor cells were separated from B6-HA or B6-NA donor cells by an 0.4-µm membrane (Supplemental Fig. 3B). This had no impact on the potency of S1 presentation.

Ag is transferred in the context of secreted exosomes from the donor cells

Based on these results, we hypothesized that Ag transfer is mediated by exosomes secreted from donor cells and internalized by recipient cells, a mechanism previously described for tumor cell-derived Ags (41, 42). Accordingly, we purified exosomes from B6-HA supernatant over a sucrose gradient and pulsed fractions onto B6IE^d cells. As shown in Fig. 3A, S1 presentation was only observed in fractions whose density corresponded with that of typical exosomes (1.13–1.19 g/cm³) (43). Furthermore, immunoblotting of fractionated supernatants revealed co-migration with the exosomal marker flotillin (44) (Fig. 3A, lower panel). Finally, transmission electron microscopy (TEM) of the active fractions (Fig. 3A, inset) revealed structures typical of exosomes imaged by TEM (~100 nm). An exosome preparation from influenza-infected B6 donor cells was similarly stimulatory (Fig. 3B).

FIGURE 3. Ag is presented via secreted exosomes produced by donor cells.

A, Exosomes were purified from B6-HA supernatant by sucrose gradient fractionation. Fractions were analyzed for their sucrose density by refractometry. Dialyzed fractions were then pulsed onto B6IE^d cells and cocultured with S1-T cells. Fraction number is graphed against T cell activation (solid line) and sucrose density (dotted line). The accepted density range of exosomes (43) is highlighted. These fractions were also blotted for S1 and flotillin-1, an exosomal marker (44). Inset, Purified exosomes from B6-HA were negatively stained using uranyl acetate and imaged by TEM. Original magnification ×26,000. This image is representative of a typical field for three independent experiments. B, Exosomes were purified from supernatant of infected B6 cells that were infected with 50 HAU iPR8 and pulsed onto B6IE^d cells. S1 presentation was assessed with the S1-specific T hybridoma. C, B6-HA cells were treated with PMA, LPS, NH₄Cl, or chloroquine (CQ) for 16 h. Subsequently, exosomes were purified from condition supernatant, or extensively washed cells were co-incubated with B6IE^d and S1-T hybridomas. D, Exosomes purified from B6IE^d-HA cells were pulsed onto B6 cells and incubated with S1-T cells. All experiments are representative of three independent experiments. *p < 0.05. MUG, β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside.

To verify exosome-mediated transfer, donor cells were treated with PMA and LPS, enhancers of exocytosis (45,46),and cocultured with B6IE^d cells after extensive washing (Fig. 3C). In both cases, the amount of material recovered from the exosome fraction was increased (Supplemental Fig. 4), and enhanced presentation was observed (Fig. 3C). These compounds did not alter the amount of surface HA or cell viability (Supplemental Fig. 5), indicating that the effect is not due to increased Ag expression. The experiment was repeated with chloroquine and NH₄Cl, inhibitors of exocytosis (47, 48). Both compounds reduced exosome production (Supplemental Fig. 4) and similarly affected presentation of S1 (Fig. 3C) while having no effect on HA expression or viability (Supplemental Fig. 5).

Another reported transfer mechanism is the release of peptide–MHC II complex-containing exosomes by DCs (49), allowing for the acquisition of “prepresented” Ag. However, exosomes from B6IE^d-HA cells pulsed onto B6 cells did not result in S1 presentation (Fig. 3D), nor did they directly activate the S1-specific T hybridoma (data not shown). This form of Ag transfer may be limited to exosomes secreted from and taken up by professional APCs.

HA incorporated into exosomes is intact and in an outward-facing orientation

We next examined the possibility that the S1 epitope is transferred as part of soluble HA fragments within the exosome lumen, as would occur with cytosolic proteins (50). We first immunoblotted exosomal lysate with a monoclonal anti-S1 Ab and detected a single band at 64 kDa (Fig. 4A), the expected size of full-length HA. Because membrane proteins can be incorporated into exosomes in either orientation (50), we next determined the directionality of HA by flow cytometry. Latex beads adsorbed with purified exosomes from B6-HA cells demonstrated HA-specific staining (Fig. 4B). Together these results indicate that exosome-incorporated HA is full length with its extracellular domain facing outward.

FIGURE 4. S1 is transferred in the context of full-length, native HA on exosomes from the donor cells.

A , Exosomes were purified from B6-HA cells and subjected to Western blot analysis using anti-S1 Ab (top panel) or pulsed onto B6IE^d cells in the presence of S1-T hybridomas (bottom panel). B, Purified exosomes from B6-HA or B6-NA cells were adsorbed to aldehyde/sulfate latex beads and stained for HA or NA by flow cytometry. All experiments are representative of three independent experiments. MUG, β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside.

Sialic acid binding function of HA enhances transfer of exosomal Ag

HA mediates influenza cellular attachment by binding to terminal sialic acid residues at the plasma membrane. Thus, we speculated that incorporation of HA rendered the exosome capable of receptor-mediated attachment, increasing efficiency of Ag transfer. This was tested in two ways. First, a transfer assay was performed in the presence of an amount of neutralizing anti-HA Abs that was confirmed to inhibit virus infectivity (Supplemental Fig. 6A). Inhibition was substantial, though incomplete (Fig. 5A). A similar effect was observed when donor cells were replaced with purified exosomes (Fig. 5B). Next, acceptor cells were pretreated with NA to remove surface sialic acid residues, with efficacy of this procedure also assessed via infectivity assay (Supplemental Fig. 6B). NA-treated acceptor cells pulsed with exosomes from B6-HA or B6-NA cells presented S1 less efficiently than did untreated cells (Fig. 5C). As anticipated, this treatment was less effective than the neutralizing Ab because sialated glycoproteins and glycolipids will reappear at the plasma membrane during NA-free coculture with donor cells and T hybridomas.

FIGURE 5. Receptor binding activity of HA enhances exosome-mediated Ag transfer.

A, B6-HA cells were cocultured with B6IE^d cells and T cell hybridomas with or without purified anti-HA Ab (H28E23) or purified anti-TfR Ab. Peptide controls consisted of B6IE^d cells cocultured with B6-NA and S1-T hybridomas in addition to 10⁻⁵ M synthetic S1 peptide. B, Exosomes were purified from B6-HA and then pulsed onto B6IE^d cells and T cell hybridomas with or without equivalent amounts of anti-HA Ab, anti-NA Ab, or anti-S1 Ab. C, B6-HA cells were co-incubated with B6IE^d cells pretreated with NA and then co-incubated with S1-T cells. All experiments are representative of three independent experiments. *p < 0.05. MUG, β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside.

HA can enhance cross-presentation of a “bystander” exosome-associated Ag

Finally, with an eye toward practical application, we investigated whether HA enhances transfer and presentation of other Ags incorporated into the exosome. To this end, we generated stable B6 cells that express a modified version of the human IL-2 receptor α-chain (Tac), in which S1 was appended at the N terminus (S1–Tac) (Fig. 6A). First, cells transfected with the S1–Tac gene were tested as donors and observed to facilitate a low level of H-2M–dependent presentation (Fig. 6B). We then transfected stably expressing S1–Tac cells with a gene encoding the Rv6 variant of PR8 HA (H1-Rv6), which lacks the S1 epitope while retaining sialic acid binding activity (31). When HA-Rv6 was coexpressed with the S1-bearing Tac construct, presentation of S1 was markedly enhanced (Fig. 6C). As was the case with PR8 HA, HA-Rv6 expression alone did not enhance exocytosis (Supplemental Fig. 7). Taken together, our results demonstrate that the receptor binding activity of HA enhances exosome-mediated Ag transfer, not just of HA-associated epitopes but also of exosome-incorporated proteins in general.

FIGURE 6. HA can enhance presentation of another exosome-associated Ag.

A, Schematic of the S1–Tac construct. The S1 epitope was appended to the N terminus of wild-type (WT) Tac just after the signal sequence (SS). TM, transmembrane domain. B, B6-S1-Tac cells were co-incubated with either B6IE^d cells or H2M^−/− B6IE^d cells in addition to S1-T hybridomas. Peptide controls consist of acceptor cells pulsed with 10⁻⁵ M synthetic S1 pep-tide in the absence of donor cells. C, B6-S1-Tac cells were transduced with HA-Rv6 and incubated for 48 h. After incubation, these cells and empty vector-transduced cells were cocultured with B6IE^d cells and T cell hybridomas. All experiments are representative of three independent experiments. MUG, β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside.

Discussion

The concept of MHC II-restricted Ag transfer is ostensibly of limited interest because, as exemplified by the S1 epitope, classical MHC II processing entails internalization of exogenous Ag, unfolding and/or digestion in the endocytic compartment, and H-2M–dependent loading onto nascent class II molecules (51). In essence, there would be minimal difference between direct uptake of an Ag as part of the pathogen and uptake of the same Ag released from an infected cell. However, there is now convincing evidence for many alternative processing pathways including the loading of Ag onto “recycling” class II molecules in early endosomes (2, 52–54), delivery of Ag from cytosol to the endolysosomal compartment via autophagy (8, 9), and a cytosolic/proteasome-dependent pathway (10). Because this last route, which produces the NA79 epitope, is similar to conventional class I processing, we investigated whether a scheme similar to MHC I-restricted cross-presentation exists on the class II side. This was possible because NA79 is produced via only the cytosolic route (and not the classical route). Despite observing substantial direct presentation of NA79 from endogenous NA, presentation by Ag transfer was quite limited. However, we observed robust presentation of the classical S1 epitope driven by the incorporation of HA into exosomes. Because HA-bearing exosomes are similar to influenza virions in having the extracellular domain of HA facing outward, we investigated whether the receptor-binding function of HA enhances the efficiency of transfer. Blocking Abs and NA treatment of the acceptor cells confirmed that this was the case.

Exosomes originate as intralumenal vesicles in the multivesicular body (MVB), and protein incorporation is selective. One basis for localization to this compartment is monoubiquitination (55). Monoubiquitinated proteins interact with a heterooligomeric complex termed the endosomal sorting complex required for transport, which mediates targeting to the MVB. This is an unlikely basis for incorporation of HA, however, because it is not known to contain a ubiquitination motif. Another partitioning mechanism, exemplified by TfR, involves the rerouting of internalized proteins aggregated into lipid rafts (56, 57) that are preferentially sorted into MVBs rather than recycling compartments (55). Because HA is well known to aggregate into raft domains for the budding of new virions, we favor this mechanism as the basis for sorting into exosomes (58, 59). In addition, the high levels of expression associated with infection and the transfection conditions used might cause “overflow” into certain cellular compartments. This phenomenon has been described for protein disulfide isomerase; when superexpressed, it traffics beyond the endoplasmic reticulum into the endosomal compartments by saturating the relevant endoplasmic reticulum-retention machinery (60).

The near absence of NA79 presentation via Ag transfer is not explained by failure to localize into exosomes because NA is also incorporated into lipid rafts. Although NA levels are ~5-fold less than those of HA during a viral infection (61, 62), expression level does not appear to be the reason. NA is readily detected on the surface of exosomes from infected cells (data not shown), and titration of exosomes in S1 stimulation assays indicate that the amount of NA should not be limiting. Furthermore, marginal Ag transfer was also observed from NA- but not HA-transfected cells, despite comparable levels of expression (data not shown). We have speculated for some time that the NA79 epitope is destroyed immediately after virus or exosome internalization, precluding binding to class II. Such susceptibility has been noted for several other epitopes (63–65). In our own Ag system, presentation of the H-2–IE^d-restricted and HA-derived S3 epitope is considerably increased by addition of leupeptin, a serine and cysteine protease inhibitor (2, 66). We stress that proteolytic susceptibility is likely a property of the epitope itself and not the parent protein. Thus, other epitopes within NA may be presentable from exogenous sources. An intriguing possibility is that some MHC I epitopes are similarly susceptible to endosomal proteases such that they are presentable mainly or exclusively from endogenous sources. Effective responses to such epitopes might depend upon direct infection of the professional APCs. Indeed, endosomal susceptibility may be an explanation for MHC I epitopes that are inefficiently cross-presented, a property that has been correlated with subdominance, and low immunogenicity (67, 68)

We have refrained from describing exosome-mediated transfer of S1 as “cross-presentation.” Cross-presentation of MHC I-restricted epitopes is generally understood to involve transfer of biosynthesized Ag in complex form (apoptotic bodies, heat shock protein associated, etc.) from an infected cell to the cytosol of a DC for proteasome and TAP-dependent processing and presentation (11, 14, 69). In the case of S1 transfer, biosynthesis of HA is required for exosome incorporation, but cytosolic delivery is not. Additional studies may reveal MHC II-restricted Ag transfer that features both biosynthesis and cytosolic delivery and could rightfully be termed cross-presentation. In fact, this may be the case for the small amount of NA79 presentation by the recipient cell (Fig. 1). In preliminary experiments, anti-HA blocking Ab does not reduce presentation of NA (data not shown), suggesting a transfer mechanism that is not exosome mediated. Further, if transfer of NA79 is limited by susceptibility to endosomal proteases, less susceptible epitopes might be “cross-presented.” Further complicating the terminology is the existence of an alternative TAP-independent MHC I cross-presentation pathway that involves delivery of Ag to MHC I-containing phagolysosomes (11, 70, 71). In outline, this scheme is very similar to the Ag transfer mechanism that we have described for S1.

MHC I cross-presentation facilitates immune responses in at least two ways. First, it effectively separates Ag from pathogen-directed mechanisms that specifically interfere with processing and presentation within the infected cell. Although most examples of interference are in conjunction with MHC I (12, 72), several MHC II-directed strategies have now been reported (73–76). Thus, exosome-mediated Ag transfer could confer a similar benefit. Viruses are also capable of more global counterstrategies. For example, the NS1 protein of influenza is a powerful inhibitor of RIG-I (77), a cytosolic sensor that stimulates elaboration of type I IFNs, potent activators of the innate response (78). Transfer of Ag beyond this immunosuppressive environment may be instrumental in driving the CD4⁺ T cell response. A second function of cross-presentation is delivery of Ag from infected cells to professional APCs in cases where direct acquisition of Ag by professional APCs is inefficient due, for example, to tropism of the pathogen. This is also a potential benefit of exosome-mediated Ag transfer but with an important caveat. Like influenza, many enveloped viruses use lipid rafts for concentration of their membrane glycoproteins during the assembly process (79–81). Thus, the formation of exosomes with receptor binding activity may be widespread. However, whether this results in enhanced Ag transfer will depend upon receptor specificity. HA binds to terminal sialic acids, a common feature on essentially every cell type. Thus, transfer to DCs is straightforward. In contrast, other viral receptors are more selective. If DCs do not express the ligand, then exosome-mediated transfer may not be potentiated beyond basal levels.

The enhanced efficiency of receptor-mediated exosome transfer should ostensibly be conferred to every protein that is also incorporated into the exosome. Our experiment with the S1–Tac fusion construct coexpressed with S1-negative HA confirms this (Fig. 6C). As already noted, NA is also incorporated into the exosome, and epitopes other than NA79 could be effectively transferred. Of note, we have observed a decided skewing of the H-2^b–restricted influenza-specific CD4⁺ T cell response toward the membrane glycoproteins (G. Sinnathamby, J. Testa, M.K. Tewari, P. Cresswell, M. Maric, and L.C. Eisenlohr, submitted for publication). A prominent role of exosome-mediated transfer in the CD4⁺ response to influenza would be consistent with this skewing.

Historically, MHC II-restricted Ag transfer has not been studied in great detail. Considering the central role that CD4⁺ T cells play in most immune responses, the alternative processing pathways that have been elucidated, and results such as those reported in this study, the phenomenon deserves greater consideration. Not only could it provide important insight into the character of host responses to particular pathogens, but it may also lead to new platforms for vaccine development.

Abbreviations used in this paper

CQ

chloroquine

DC

dendritic cell

HA

hemagglutinin

HAU

hemagglutinating unit

iPR8

infectious PR8

MHC I

MHC class I

MHC II

MHC class II

MUG

β-galactosidase substrate methyl-umbelliferyl-β-d-galactoside

MVB

multivesicular body

NA

neuraminidase

NA79

neuraminidase aa 79–93 epitope

PR8

influenza A/Puerto Rico/8/34

Rv6

Rv6 variant of PR8 HA

S1

site 1

S1–Tac

S1 appended on the N terminus of Tac

SS

signal sequence

TEM

transmission electron microscopy

TM

transmembrane domain

UVPR8

UV-inactivated PR8

WT

wild-type