Distinct Pathways of Antigen Uptake and Intracellular Routing in CD4 and CD8 T Cell Activation

See allHide authors and affiliations

Science  27 Apr 2007:
Vol. 316, Issue 5824, pp. 612-616
DOI: 10.1126/science.1137971


The mechanisms that allow antigen-presenting cells (APCs) to selectively present extracellular antigen to CD8+ effector T cells (cross-presentation) or to CD4+ T helper cells are not fully resolved. We demonstrated that APCs use distinct endocytosis mechanisms to simultaneously introduce soluble antigen into separate intracellular compartments, which were dedicated to presentation to CD8+ or CD4+ T cells. Specifically, the mannose receptor supplied an early endosomal compartment distinct from lysosomes, which was committed to cross-presentation. These findings imply that antigen does not require intracellular diversion to access the cross-presentation pathway, because it can enter the pathway already during endocytosis.

Adaptive immunity requires activation of T lymphocytes by dendritic cells (DCs), which present antigen bound by major histocompatibility complex (MHC) molecules (1). Antigens that are synthesized intracellularly (for example, those of viral or tumor origin) are presented by MHC I molecules and activate cytotoxic CD8+ T cells. In contrast, extracellular antigens are presented by MHC II molecules to activate CD4+ T helper cells (1, 2). A further mechanism termed cross-presentation permits some forms of extracellular antigen to also stimulate CD8+ T cells via the MHC I pathway (35). This is required for immunity against viruses that do not infect APCs directly or against tumor antigens that are not endogenously expressed by DCs (510). The mechanisms that divert endocytosed antigen from the classical MHC II–restricted presentation pathway to that facilitating cross-presentation are controversial (1016). Before substantial experimental evidence for cross-presentation had become available, the assumption that extracellular antigen was presented exclusively to CD4+ T cells had implied that the different mechanisms of endocytosis supplied only MHC II–restricted antigen presentation. Thus, the possibility that a differential influence of these uptake mechanisms on antigen presentation might exist, in particular on cross-presentation, remained to be clarified.

We recently demonstrated that mannose receptor (MR)–mediated endocytosis of the model antigen soluble ovalbumin (OVA) enabled its cross-presentation to CD8+ T cells (17). We therefore investigated whether this was specific for the class I pathway or whether the MR also targets extracellular antigen for MHC II–restricted presentation. DCs from MR-deficient (18) and wild-type control mice were allowed to endocytose OVA (19), and the response of OVA-specific CD4+ T cells (OT-II cells) was examined. This revealed a complete independence from the MR for all OVA concentrations tested (Fig. 1A). In contrast, the response of OVA-specific CD8+ T cells (OT-I cells) was absolutely dependent on the MR, regardless of the antigen concentration (Fig. 1A). Intrinsic differences in DCs resulting from MR deficiency, such as altered costimulatory molecule expression or proportions of DC subtypes, were not observed (17). Thus, MR-mediated endocytosis was essential for cross-presentation but dispensable for MHC II–restricted presentation of OVA.

Fig. 1.

DCs use only MR-mediated endocytosis to obtain antigen for CD8+ T cell activation and pinocytosis for CD4+ T cell activation. (A) MR-deficient DCs cannot cross-present OVA to OT-I cells, whereas OT-II cell activation is unaffected. IL-2, interleukin-2. (B) The MR mediates the uptake of large amounts of OVA by a subpopulation of DCs, resulting in an ∼100-fold fluorescence increase. All DCs took up small amounts of OVA by DMA-inhibitable pinocytosis, causing a four-to-sixfold fluorescence increase. Gray areas denote DCs cultured without antigen. Mean fluorescence intensity values ± SD are shown. Asterisks refer to given P values. (C) Blocking pinocytosis does not alter cross-presentation to OT-I cells, whereas OT-II cell activation is abolished. (D) Pinocytosis of LY is unaffected by MR-mediated OVA uptake. (E) MR DCs, obtained by cell sorting, present OVA to OT-II cells more efficiently than MR+ DCs. Error bars in (A), (C), and (E) indicate 1 SD.

To elucidate the pathway by which antigen was endocytosed for the activation of MHC II–restricted OT-II cells, we studied the in vitro uptake of OVA labeled with the fluorochrome alexa647 (OVA647) by DCs. Although a distinct DC subset took up large amounts of OVA647 via the MR, a smaller fluorescence shift of about four- to fivefold was noted, which was MR-independent and appeared to encompass all DCs (Fig. 1B). This shift could be blocked by dimethylamiloride (DMA), an inhibitor of pinocytosis that leaves receptor-mediated endocytosis intact (Fig. 1B). DMA treatment of DCs during OVA uptake abrogated the activation of OT-II cells in a dose-dependent and MR-independent manner (Fig. 1C), indicating that only pinocytosed, but not MR-endocytosed, OVA was used to activate CD4+ T cells. The response of OT-I cells was unaffected by DMA (Fig. 1C), excluding toxic effects.

Pinocytosis was active both in MR+ DCs that endocytosed a large dose of OVA and in MR DCs that did not, because both DC subsets were able to take up lucifer yellow (LY) (Fig. 1D), a pinocytosis marker (20). Despite slightly smaller pinocytotic activity (Fig. 1D) and severely reduced total uptake of OVA (Fig. 1B), MR DCs were superior at activating OT-II cells as compared to MR+ DCs (Fig. 1E). This indicated that MR DCs processed pinocytosed antigen more efficiently than MR+ DCs, which is consistent with the recent finding that antigen processing for MHC II–restricted presentation depended on intrinsic properties of particular DC subtypes (21). In summary, all DCs constitutively pinocytosed small amounts of OVA, which was used specifically for presentation to CD4+ T cells, while MR+ DCs could simultaneously internalize large amounts of OVA exclusively for cross-presentation.

Cross-presentation requires a high antigen dose (7). Thus, its dependency on the MR may simply be due to the large amount of antigen endocytosed by this receptor as compared to the small amount internalized by pinocytosis (Fig. 1B). This possibility was addressed with the use of another type of APC capable of cross-presentation: bone marrow–derived macrophages (MΦs) (22). As with DCs, cross-presentation of OVA by these APCs was entirely MR-dependent (Fig. 2A). In contrast to DCs, however, MΦs also internalized large amounts of OVA in the absence of the MR (Fig. 2B). This additional uptake could be blocked with polyinosinic acid (polyI) (Fig. 2B), a specific inhibitor of scavenger receptors (SRs) (23). The inability of MR-deficient MΦs to activate OT-I cells, despite having internalized large amounts of OVA via SRs, indicated that the mechanism of endocytosis, rather than levels of antigen, was a requisite for cross-presentation. Moreover, SR-mediated OVA uptake was not responsible for OT-I cell activation, because MR-mediated endocytosis was sufficient (Fig. 2C). In further experiments using polyI and DMA blockade, we demonstrated that MΦs used pinocytosed OVA, as well as SR-endocytosed OVA, only for presentation to OT-II cells (Fig. 2, A and C).

Fig. 2.

MΦs use only MR-endocytosed antigen for CD8+ T cell activation, whereas pinocytosed and SR-endocytosed antigens were used only for CD4+ T cell activation. (A) MR-deficient MΦs cannot cross-present OVA to OT-I cells, whereas OT-II cell activation is unaffected. (B) Endocytosis via MR and polyI-inhibitable SR is responsible for high-dose OVA uptake by MΦs. Gray areas denote controls without antigen. Mean fluorescence intensity values ± SD are shown. Asterisks refer to given P values. (C) Simultaneous blockade of the SR with polyI and of pinocytosis with DMA abolishes OT-II cell activation, whereas cross-presentation to OT-I cells is unaffected. Error bars in (A) and (C) indicate 1 SD.

To investigate why pinocytosed, SR- and MR-endocytosed antigen were presented differently, we monitored their intracellular routing by fluorescence microscopy in immature APCs, which are active in endocytosis (Fig. 3 and fig. S1). Within DCs, endocytosed OVA647 strictly colocalized with the MR (Fig. 3A), whereas the small quantities taken up by pinocytosis (Fig. 1B) were insufficient for microscopic visualization. Therefore, pinocytosed antigen was monitored with the use of the surrogate marker LY amplified by antibody staining. This approach revealed that LY and OVA did not colocalize (Fig. 3A and table S1), indicating that pinocytosis and the MR supplied distinct intracellular compartments. MR-endocytosed OVA was localized in organelles expressing the early endosomal marker Rab5 and the early endosomal antigen 1 (EEA1). Consistent with the previous finding that the cytoplasmic domain of the MR did not target antigen toward lysosomes (24), MR-endocytosed OVA was excluded from late endosomes or lysosomes revealed by Rab7 or lysosomal-associated membrane protein 1 (LAMP-1) staining or by the fluorescent dye lysotracker that accumulates in the acidic lysosomes (Fig. 3B and table S1). In contrast, pinocytosed antigen was transported exclusively toward lysosomes lacking the early endosomal markers (Fig. 3B). This separation was stable because, even for 6 hours after antigen uptake, MR-endocytosed OVA remained confined to EEA1+ endosomes and was excluded from lysosomes, whereas the opposite was observed for pinocytosed material (fig. S2).

Fig. 3.

MR-mediated endocytosis conveys antigen into an endosomal compartment distinct from that supplied by SRs and by pinocytosis. (A) Immature DCs were allowed to endocytose OVA647 and/or LY for 20 min, were chased for another 20 min, and were fixed and stained for intracellular MR expression or for pinocytosed LY (19). MR-endocytosed OVA is excluded from the organelles that received LY. (B) Pinocytosed LY is localized in lysosomes identified by lysotracker but not in organelles expressing EEA1, Rab5, or Rab7. MR-endocytosed OVA is located only in early endosomes containing Rab5 and EEA1 but not Rab7, LAMP-1, or lysotracker. (C) Pinocytosed LY colocalizes with MHC II but not with MHC I. MR-endocytosed OVA does not colocalize with MHC II but partially colocalizes with MHC I. (D) In DCs, cross-presented OVA, intracellularly revealed by the 25-D1.16 antibody, was located exclusively in MR+ EEA1+ endosomes but colocalized neither with MHC II nor with pinocytosed LY. Cross-presented OVA partially colocalized with unprocessed OVA. (E) In MΦs, cross-presented OVA was also found only in MR+ EEA1+ endosomes but not in MHC II+ lysosomes. (F) Cross-presented OVA is confined to MR+ endosomes also in CD11c+ splenic DCs that had been isolated ex vivo. (G) MR-deficient MΦs internalized high-dose OVA, which was not delivered into EEA1+ endosomes but instead colocalized with pinocytosed LY. (H) Visualization of simultaneous uptake of SR- and MR-endocytosed antigen into separate compartments of the same MΦ (19). Cell nuclei were revealed with 4',6'-diamidino-2-phenylindole staining in blue. Lysotracker staining was performed with unfixed cells, and nuclei of unfixed cells were visualized with Hoechst 33342 dye. Separate channel images and controls are available in figs. S1 and S3. Statistical analysis is provided in table S1. (I) Proposed model of antigen uptake, routing, and presentation.

Subcellular antigen localization correlated with the selectivity of its presentation, because pinocytosed LY colocalized specifically with lysosomal MHC II, whereas MR-endocytosed OVA colocalized exclusively with MHC I (Fig. 3C). Only MR+ early endosomes, and not MHC II+ lysosomes, contained cross-presented OVA (Fig. 3D and table S1), which was visualized by means of the monoclonal antibody 25-D1.16 (19) that specifically recognizes the OVA peptide SIINFEKL (Ser-Ile-Ile-Asn-Phe-Glu-Lys-Leu) when bound to the MHC I molecule Kb (25) and that has been previously used for immunofluorescence microscopy (11, 26). Specificity of 25-D1.16 staining for cross-presented OVA was confirmed by the lack of such staining in the absence of antigen or of the MR and after blockade of the proteasome (fig. S3), which is required for cross-presentation (4, 7, 8, 1012). The confinement of cross-presented OVA to a distinct class of endosomes and its exclusion from lysosomes were also observed in MΦs (Fig. 3E and table S1) and in ex vivo–isolated splenic DCs (Fig. 3F). This suggests that cross-presentation of OVA in these experiments depended on antigen location in an early endosomal compartment. This interpretation was tested with the use of MR-deficient MΦs to determine the subcellular location of SR-endocytosed OVA. Such OVA was not cross-presented (Fig. 2A and fig. S3), despite uptake at high amounts (Fig. 2B) that even sufficed for direct intracellular visualization (Fig. 3G). If antigen location in these early endosomes was required for cross-presentation, then such organelles should not contain SR-endocytosed OVA. Indeed, this appeared to be the case (Fig. 3G and table S1), because SR-endocytosed OVA did not colocalize with EEA1 (Fig. 3G) or MR-endocytosed antigen (Fig. 3H). Instead, it colocalized with pinocytosed LY in lysosomes over an extended period (Fig. 3G and figs. S4 and S5). These findings provide further evidence that high-dose antigen intended for MHC II–restricted presentation and for cross-presentation are routed through different organelles in MΦs. The ability of the MR to sequester OVA away from lysosomal degradation may also explain why bone marrow–derived MΦs were particularly efficient at cross-presentation in our experiments (Fig. 2), although their lysosomal compartment has been shown to degrade antigen far more efficiently than that of DCs (27).

The experiments examining T cell responses in Figs. 1 and 2 did not permit us to quantitatively and directly compare the capabilities of DCs and MΦs to cross-present. However, some comparisons could be made by directly examining cross-presented antigen on the APC surface. When viable DCs and MΦs were stained with the 25-D1.16 antibody, flow cytometry revealed a small fluorescence shift (fig. S6), which has been reported to indicate the presence of cross-presented OVA on the cell surface (8). This shift was confined to MR+ DCs and MR+ MΦs and was absent from APCs lacking the MR (fig. S6). Thus, only those APCs carrying cross-presented OVA in endosomes also displayed it on their cell surface, which is consistent with the observed restriction of OT-I cell activation to these APCs (Figs. 1A and 2A). Quantitative analysis showed that, on a per cell basis, MR+ DCs were superior to MΦs at displaying cross-presented OVA on the surface (fig. S6).

These results support a model of antigen presentation, in which the MR introduces OVA specifically into a stable early endosomal compartment for subsequent cross-presentation. Simultaneously, pinocytosis and, in MΦs, SR-mediated endocytosis conveyed OVA to lysosomes for MHC II–restricted presentation (Fig. 3I). Our results indicate that mechanisms of antigen uptake can dictate the intracellular destination compartment and thus the presentation of antigen to CD4+ or CD8+ T cells.

Several possible implications arise from this model. First, receptor dependence of cross-presentation may allow APCs to restrict this pathway to stages of maturation that express suitable receptor(s) or to distinct classes of antigen. Second, receptor dependence may be a complementary mechanism to differences in intracellular processing pathways (21) that restrict cross-presentation to particular APC subpopulations. Consistent with previous reports showing that murine CD8α+ DCs had dedicated cross-presenting functions (9, 21, 28, 29), we observed that only CD8α+ DCs expressed the MR (fig. S7), suggesting that only these DCs could internalize OVA into endosomes dedicated to cross-presentation. DEC-205 may be a further receptor linked to such endosomes, because antigen targeted to this molecule on CD8α+ DCs by means of a specific antibody was cross-presented (21). However, DEC-205 differs from the MR by its ability to also target antigen toward MHC II–restricted presentation (21, 24). Consistent with this capability, a major part of DEC-205–internalized antigen has been reported to be localized within lysosomes (24). Future studies may determine whether DEC-205 can target its cargo both to the lysosomal compartment and to the MR+ early endosomal compartment described here. Third, the constitutive activity of pinocytosis seen in all DC subsets may ensure that cross-presenting DCs can acquire antigen for induction of cognate CD4+ T cell help. Such help requires that the same DC stimulates both specific CD4+ and CD8+ T cells, which is essential for effective CD8+ T cell responses (30, 31). Constitutive pinocytosis avoids situations in which cross-presenting DCs cannot simultaneously stimulate specific CD4+ T cells because of a lack of receptors capable of targeting the MHC II presentation pathway. A fourth implication pertains to the cell biology of cross-presentation. Current mechanistic models assume that antigen must be rescued from lysosomal degradation by intracellular diversion toward less acidic compartments (4, 712, 15, 32). Our finding that soluble antigen can already enter the cross-presentation pathway during endocytosis demonstrates that intracellular antigen “crossing” may not be necessary, but this of course does not rule out an additional role of intracellular antigen-sorting mechanisms. Fifth, the Rab5+ EEA1+ stable early endosomes that we identified as committed to cross-presentation resemble organelles recently described by Lakadamyali et al.(33), who proposed that endocytosed cargo was not indiscriminately delivered to a common endosomal pool. Instead, the cargo was introduced either into dynamic, fast-maturing endosomes carrying cargoes toward rapid lysosomal degradation or into Rab5+ EEA1+ “static” early endosomes, whose biological role in antigen presentation was not addressed in that study. Further studies that characterize the cell biology of a possible early endosomal compartment committed to cross-presentation are required to elucidate their relation to previously described static early endosomes (33). Finally, the lack of intracellular staining for cross-presented OVA after proteasome blockade suggested that antigen had to be exported to the cytoplasmic proteasome for peptide generation, reminiscent of the recently described export of cell-associated antigen from phagosomes and subsequent reimport of peptides for MHC I loading within phagosomes (8, 11, 12). It is possible that similar antigen-relocation mechanisms operate in the stable early endosomes committed to cross-presentation of soluble antigen. It has to be emphasized, however, that cross-presentation of cell-associated antigen mechanistically differs from that of soluble antigen (8, 1012, 17). Thus, a role of uptake mechanisms and stable early endosomes in classical cross-priming (3) remains to be shown. We suggest that targeting antigen to the MR may represent an avenue for improving vaccines aimed at inducing CD8+ T cell–mediated immunity against viruses or tumors.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Table S1


References and Notes

View Abstract

Navigate This Article