Mature dendritic cells use endocytic receptors to capture and present antigens

Expression and Internalization of Endocytic Receptors by Mature DCs.

We first confirmed the surface expression of DEC-205 and FcγRII/III in mature DCs. We found that FcγRII/III was down-regulated slightly in both BMDCs and splenic DCs (Fig. S1) as previously shown (13, 14). In contrast, DEC-205 was highly up-regulated in mature BMDCs, although its expression on CD8α⁺ DCs did not change on maturation (Fig. S1) (15). We also detected low levels of DEC-205 on CD8α⁻ DCs, which, as with BMDCs, increased on maturation (15) (Fig. S1).

Next, we asked if FcγRs and DEC-205 were internalized in mature BMDCs. We assessed uptake of ICs, which are the physiological ligands for FcγRs. Because no physiological ligands have yet been identified for DEC-205, we used anti-DEC-205 monoclonal antibodies, which lead to the presentation of covalently linked antigens after internalization by DCs (16). ICs and anti-DEC-205 antibodies initially bound to the cell surface at 4 °C were rapidly internalized at 37 °C by both immature and mature DCs, with >60% of the ICs and anti-DEC-205 disappearing from the cell surface within 10 min, although receptor internalization by mature DCs appeared slightly less efficient (Fig. 1 A and B). Most ICs reached H-2M⁺ late endosomal/lysosomal compartments in mature DCs 2 h after internalization (Fig. 1C, Upper). In contrast, anti-DEC-205 antibodies partially localized to the lysosomal compartment after only 15 min but exhibited a diffuse pattern after 2 h, suggesting their degradation in, or recycling from, the lysosomal compartment (Fig. 1C, Lower). Recycling would be in agreement with previous analysis of DEC-205 dynamics in transfected cells (17). Immunoelectron microscopy confirmed these results, with ~50% of IC-gold and anti-DEC-205 antibody-gold detected in LAMP-2⁺ lysosomal compartments 60 min after internalization (Fig. 1C and Fig. S2A). Thus, mature DCs can efficiently internalize and deliver receptor-bound antigen to late endocytic compartments.

Phagocytosis is a physiologically important route of entry that permits DCs to present pathogen-derived antigens. It is also one that has been shown to be down-regulated during maturation, attributable, in part, to a decrease in active GTP-loaded Cdc42 (7, 18). Salmonella typhimurium can overcome this block by injecting a guanine nucleotide exchanger for Cdc42 (7, 18), suggesting that phagocytosis might also be triggered in mature DCs by coupling the phagocytic ligand to an appropriate signaling receptor. Because cross-linking of surface FcγR is known to induce phagocytosis in macrophages (19), we examined whether FcγR cross-linking by IgG could trigger phagocytosis in mature DCs. As expected, both BSA and IgG-coated latex beads were taken up by immature DCs and targeted to H-2M⁺ lysosomes, although uptake of FcγR-bound beads was more efficient (Fig. 2 A and B). In mature DCs, far fewer BSA-coated beads were phagocytosed, confirming that nonspecific constitutive phagocytosis is down-regulated by DC maturation (20, 21). In contrast, mature DCs phagocytosed IgG-coated beads nearly as efficiently as immature DCs (Fig. 2 A and B). These data indicate that although mature DCs down-regulate nonspecific forms of endocytosis (both macropinocytosis and phagocytosis), receptor-mediated endocytosis and phagocytosis remain efficient routes of uptake.

Mature DCs Efficiently Present Antigens to T Cells in Vitro.

Although the down-regulation of endocytosis accompanying maturation is thought to be a major barrier to the presentation of newly acquired antigens on MHCII (21–23), downstream events may also contribute. First, MHCII synthesis is reduced in mature DCs (4, 5, 24, 25). Second, as DCs mature, MHCII is translocated from lysosomes to the plasma membrane, which greatly decreases acceptor MHCII for peptides generated following antigen uptake (26, 27). However, the ability of mature DCs to process and present newly internalized antigens has not been directly addressed.

To answer this question, BMDCs were matured with LPS (or left untreated) and sorted into mature and immature populations by two criteria: expression of the costimulatory molecule CD86 and macropinocytotic activity. Cells were then pulsed with soluble or receptor-targeted ovalbumin (OVA) and cultured with OT.2 CD4⁺ T cells specific for MHCII-OVA complexes. As expected, soluble OVA was efficiently presented to T cells by immature DCs but not by mature DCs, reflecting the down-regulation of macropinocytosis (Fig. 3 A and B, Middle). When exposed to receptor-targeted OVA (OVA/anti-OVA IC or anti-DEC-205 antibody-OVA), however, mature DCs were even more efficient at activating T cells than immature DCs (Fig. 3 A and B, Left).

The poor presentation of DEC-205-targeted antigen by immature cells most likely reflected the low levels of DEC-205 on these cells (Fig. S1). However, it could not be the explanation for FcγR-targeted IC, because FcγRII/III surface expression decreases during BMDC maturation (Fig. S1). It is also unlikely that the increased ability of mature DCs to stimulate T cells results from their increased costimulatory capacity, because we provided T cells with artificial costimulation by coating assay plates with anti-CD28 antibodies.

Still, because the mature DCs also presented OVA peptide, which did not require processing, with somewhat higher efficiency (Fig. 3 A and B, Right), it was possible that the enhanced ability of mature DCs to stimulate T cells could have masked an inferior antigen processing ability. We thus directly measured MHCII-peptide complex generation using the YAe monoclonal antibody, which recognizes MHCII-Eα-peptide (amino acids 52–68) complexes. We compared the ability of mature and immature DCs to present aggregates of Eα protein (taken up nonspecifically by macropinocytosis) or Eα ICs (taken up by FcγR). Immature DCs formed MHCII-peptide complexes from untargeted Eα protein significantly better than the mature DCs, reflecting their different capacities for macropinocytosis (Fig. 3C). However, mature DCs were at least as good as immature DCs at generating MHCII-peptide complexes from Eα ICs (Fig. 3C). Similar results were obtained following the phagocytosis of opsonized Escherichia coli expressing the Eα protein (Fig. S3). DCs matured by other TLR ligands showed similar abilities to present receptor-targeted antigens on MHCII (Fig. S4), suggesting that this is a general property of mature DCs. Thus, although mature DCs lose the ability to present untargeted antigens, their ability to present antigens on MHCII following receptor-mediated uptake by clathrin-coated pits or phagocytosis remains intact.

Because OVA has been reported to bind the mannose receptor (MR) (28), we tested the contribution of this receptor to OVA uptake by immature DCs under our experimental conditions. We found that when MR was blocked by an excess of mannan, uptake was reduced by <20% (Fig. S5). This confirms that immature BMDCs internalize OVA by a combination of receptor-mediated and fluid-phase macropinocytosis, as previously observed (7). MR is markedly down-regulated during maturation (29), however, precluding receptor-mediated OVA uptake in mature DCs. The Eα protein was purified from E. coli, and is therefore not glycosylated. Thus, lectins could not contribute to the uptake of this antigen.

Previous studies have shown that mature DCs efficiently present endogenous antigen but lose the ability to cross-present exogenous antigen on MHCI (9, 21). This has been attributed to a lack of antigen internalization (21) or to poor translocation of internalized antigen to the cytosol (9). In contrast, we found that mature DCs cross-presented receptor-targeted antigen more efficiently than immature DCs (Fig. 3D). Although the mechanism of cross-presentation on MHCI is still poorly understood, the increased levels of MHCI synthesis in mature DCs may explain the enhanced cross-presentation under conditions in which antigen uptake is not limiting (30). Our studies differ from previous work by making use of receptor-mediated rather than nonspecific endocytosis (21) and by using a more sensitive T-cell assay that more clearly revealed the cross-presentation that was indeed observed earlier (9).

Mature DCs Load MHCII in the Lysosomal Compartment.

How MHCII presentation of newly acquired antigens by mature DCs occurs is more difficult to rationalize. The dramatic reorganization of MHCII on maturation suggests that few, if any, acceptor MHCII for antigenic peptides remain in the endocytic compartments of mature DCs (4, 26, 27, 31, 32) (Fig. S6, Upper). Because high levels of surface MHCII might artifactually mask a reduced internal pool of MHCII, we imaged cells after blocking surface MHCII with unlabeled antibody. Indeed, under these conditions, a lysosomal pool of MHCII in mature DCs could be detected (Fig. S6, Lower). Electron microscopy confirmed that ICs internalized via the FcγR accumulated in MHCII⁺ lysosomes over time (Fig. S2B). To determine if peptide loading can occur in the MHCII-diminished lysosomal compartment of mature DCs, we incubated mature and immature DCs with Eα ICs, and used the YAe antibody to visualize the distribution of the newly formed MHCII-peptide complexes. By 90 min, both mature and immature DCs had the majority of MHCII-peptide complexes localized to the H-2M⁺ late endosomal/lysosomal compartment (Fig. 4A). These complexes were translocated to the plasma membrane with further incubation (Fig. 4A), as found previously for immature DCs then induced to mature (33). This result indicates that like immature DCs, mature DCs load antigenic peptides on MHCII in the lysosomal compartment and then make the MHCII-peptide complexes available on the plasma membrane for T-cell recognition.

It has been suggested that antigenic peptides can be loaded both on newly synthesized or recycling MHCII (2). We first examined the rate of MHCII synthesis in immature and mature DCs by metabolic labeling and autoradiography. Like many other studies, we found significant but incomplete down-regulation of MHCII synthesis after maturation (4, 5, 24, 25). MHCII synthesis was decreased by ~90% in mature BMDCs and by 65% in mature splenic DCs when compared with immature cells (Fig. S7).

To determine the contribution of the residual de novo synthesis of MHCII to peptide loading, we treated DCs with cycloheximide (CHX) to arrest protein synthesis 30 min before antigen pulse. As expected, exposure to CHX led to a near-complete blockade of protein synthesis in both mature and immature DCs; the cells remained viable because the protein synthesis block was reversed on CHX removal (Fig. S8). CHX treatment significantly impaired but did not completely abrogate the formation of YAe-reactive MHCII-peptide complexes in immature DCs, demonstrating the contribution of both newly and previously synthesized MHCII to peptide loading (Fig. 4B). However, CHX had little effect in fully mature DCs, indicating that previously synthesized (possibly recycled) MHCII plays a major role in the presentation of newly captured antigens (Fig. 4B).

Altered Access to Antigens Explains Reduced Antigen Presentation by Mature Splenic DCs in Vivo.

We next tested whether capture of antigens by mature DCs can occur in vivo. After induction of systemic DC maturation by LPS injection, mice were immunized i.v. with soluble OVA or OVA ICs. T cell activation was assessed by measuring CD69 up-regulation of adoptively transferred CD4⁺ OT.2 T cells. In control mice injected with either antigen formulation (containing LPS as an adjuvant), robust splenic T cell activation was observed as expected. However, in contrast to our in vitro data but consistent with recent results using soluble antigen (21, 24), pretreatment of mice with LPS substantially inhibited activation of T cells even after injection of FcγR-targeted OVA ICs (Fig. 5A).

Although surprising, this result did not reflect an inherent inability of mature splenic DCs to capture and present antigens. As found for BMDCs matured in vitro, CD86-high splenic DCs isolated from LPS-treated mice presented both anti-DEC-205-OVA and OVA ICs in vitro (Fig. 5B). The in vivo-matured splenic DCs even presented soluble OVA (Fig. 5B, Upper), suggesting that they retained the capacity to internalize soluble OVA either after binding to lectins or through fluid-phase endocytosis, at least when assayed in vitro. This persistence of fluid endocytosis may reflect the fact that splenic DCs activated in vivo are conventionally harvested relatively soon after LPS administration (after approximately 9 h) (24). In addition, splenic DCs down-regulate macropinocytosis with slower kinetics than BMDCs and exhibit residual fluid-phase uptake for at least 24 h (6). In either event, the failure of splenic DCs to present antigen delivered by receptor-mediated or fluid-phase endocytosis after LPS treatment in vivo was unlikely to be attributable to the loss of endocytic capacity.

Instead, the reduced presentation of antigen by mature DCs in vivo may reflect reduced access of injected antigen to DCs in the spleen. We tested this hypothesis by directly assessing the ability of i.v.-injected fluorescently labeled anti-DEC-205 antibodies to bind mature splenic DCs in situ. Compared with immature splenic DCs, mature splenic DCs in LPS-treated mice showed greatly reduced labeling with the antibody (Fig. 5C). Because expression of DEC-205 was unchanged or increased during maturation in CD8α⁺ and CD8α⁻ splenic DCs (Fig. S1), this result further suggests that systemic inflammation limited the access of splenic DCs to antigens.

Why does LPS reduce the access of splenic DCs to i.v.-injected antigen? One possibility is that LPS (and other microbial stimuli) causes migration of DCs away from the marginal and medullary zones, possibly reducing their ability to compete for incoming antigen (34, 35). We also noted that injection of even low concentrations of LPS changed the overall appearance of the spleen, which became darker and less pliable. As a result, we next asked if there were alterations in splenic blood flow that might interfere with entry of antigen or even DC precursors into the spleen. We measured blood flow to the spleen using microbubble ultrasound perfusion imaging (36, 37) after injecting mice with LPS or PBS as control and observed, on average, a 31% reduction in blood flow in the LPS-treated mice (Fig. S9 A and B). This alteration might also contribute to the failure of splenic DCs to access and capture i.v.-injected antigen in our study as well as in previous studies (21, 24).

Although i.v.-injected antigens poorly access the spleen, we examined whether antigens could access DCs in lymph nodes (LNs) under conditions of systemic administration of TLR agonists. The i.v. injection of LPS led to full maturation of LN DCs, as indicated by the uniformly enhanced expression of CD86 (Fig. S10).

We next measured the formation of MHCII-peptide complexes directly in vivo. We again used Eα as (nonglycoslylated) antigen delivered either as soluble protein (which can be internalized only by nonselective macropinocytosis) or FcγR-targeted IgG IC. When injected into control mice, as expected, both antigen formulations generated detectable amounts of surface MHCII-Eα-peptide complex, even if coinjected with a small quantity of LPS to trigger maturation at the time of antigen encounter (Fig. 5D, Left). However, when injected into mice whose DCs had been matured by prior injection of LPS, only the Eα IC, which is internalized via FcγR, generated MHCII-peptide complexes (Fig. 5D, Right). Therefore, mature LN DCs can capture, process, and present receptor-targeted antigens in vivo.

Mice.

B6/C57J, OT.1, and OT.2 mice were obtained from Jackson ImmunoResearch Laboratories. Animal care and experiments were conducted according to the guidelines of Yale University and Genentech, Inc.

Cells.

BMDCs were prepared as described (4). Day 5 BMDCs were stimulated with LPS (50 ng/mL; Sigma) for 20 h or left untreated. DCs were isolated from spleen and LN with blendzyme 2 (Roche) and DNaseI (Invitrogen) and purified with a CD11c isolation kit (Miltenyi).

Reagents.

The antibodies used were as follows: CD86 (GL1), CD8α (53-6.7), CD45R (RA3-6B2), CD11c (HL3), FcγII/III (2.4G2), CD69 (¹H.2F3), MHCII (KH74 and TIB-120), anti-CD3 (17A2), and anti-CD11c (N418), streptavidine-phycoerythrin (all from BD Biosciences); anti-6× His, anti-E. coli-FITC and anti-DEC-205 (all from Serotec); YAe (eBiosciences); anti-LAMP-2 (4); anti-H-2M (4); anti-OVA (30); anti-DEC-205-OVA; and isotype control-OVA (R. Steinman, The Rockefeller University, New York, NY) (16). Secondary antibodies and OVA-Alexa647 were from Molecular Probes. E. coli expressing 6× His-tagged Eα protein was a gift from Ruslan Medzhitov (Yale University). ICs were generated as previously described (10).

Endocytosis Assays.

To quantify internalization, anti-DEC-205 antibodies or isotype control (5 μg/mL) or ICs (10 μg/mL) were surface bound at 4 °C for 20 min, washed, and then allowed to internalize by incubation at 37 °C or 4 °C as a control. Remaining surface antibodies were detected with anti-rat or anti-rabbit antibody, respectively. The percentage of surface antibodies was determined over time [(mean surface fluorescence at 37 °C − mean surface isotype fluorescence at 4 °C)/(mean surface fluorescence at 4 °C − mean surface isotype fluorescence at 4 °C) × 100]. One-micrometer Fluoresbrite yellow-green carboxylated latex microspheres (Polysciences, Inc.) were coated overnight in 1 mg/mL mouse IgG (Jackson ImmunoResearch Laboratories) or BSA (Sigma) before incubation with DCs at 9.1 × 10⁷ beads/mL.

In Vitro Antigen Presentation Assays.

DCs were incubated for 30 min with antigen and LPS, washed, and cultured with 1 × 10⁵ CD4⁺ or CD8⁺ T cells purified by negative selection (Miltenyi) from OT.2 or OT.1 mice, respectively. When indicated, the 96-well plates were coated with anti-CD28 antibody (10 μg/mL; BD Biosciences). T-cell response was monitored by measuring either IL-2 accumulation in supernatant at 24 h by ELISA or by division of carboxyfluorescein succinimidylester-labeled T cells.

Immunofluorescence Confocal Microscopy.

Cells were stained with anti-H-2M, anti-MHCII, or YAe-biotin antibodies as previously described (4). Imaging was performed with a Zeiss LSM 510 or Leica SP5 confocal microscope.

Electron Microscopy.

Cells were processed as previously described (4). Sections were labeled with anti-MHCII or anti-LAMP-2 antibody followed by rabbit anti-rat antibody and 5 nm protein A-gold. They were examined with a Tecnai 12 Biotwin electron microscope (FEI).

Flow Cytometry.

Data were collected on a BD Canto-II, FACSCalibur, or Aria (all from BD Biosciences), and analyzed with FlowJo software (Tree Star).