Steady-state cross-presentation of OVA is mannose receptor-dependent but inhibitable by collagen fragments

Cross-presentation allows dendritic cells (DCs) to load antigenic peptides on MHC-I molecules, which is important for activating cytotoxic T cells (CTL) that combat tumors and viruses. We recently showed that, under steady-state conditions, antigens internalized by DCs through the Mannose Receptor (MR) are efficiently cross-presented (1). The MR conveyed antigen into early endosomes, where peptide was loaded on MHC-I (2, 3).

Recently, Segura et al. (4) reported that MR-mediated endocytosis is involved in cross-presentation only under inflammatory and not steady-state conditions (4), which, at first sight, contradicted our findings (1). They concluded that the MR operates only in bone marrow- or monocyte-derived DCs but not in CD8⁺ DCs, which are thought to be most relevant for in vivo cross-presentation in mice.

Certainly, many immunologic phenomena may be explained by differential involvement of DC subsets, but in this case, two alternative explanations more likely explain the findings of Segura et al. (4).

First, Segura et al. (4) reported that splenic CD8⁺ DCs do not express the MR and thus, cannot use it for ovalbumin (OVA) uptake. This statement was reached using splenic DCs isolated by collagenase digestion, which were subsequently exposed to OVA. However, collagen fragments are MR ligands (5). The massive amounts of such fragments, which are inevitably generated during collagenase digestion, bind the MR and cause its internalization. Segura et al. (4) stained for MR expression on the DC surface and thus, failed to detect intracellular MR. We had previously used intracellular staining and detected the MR within splenic DCs (2). Splenic DCs did endocytose OVA through the MR when we used a collagenase-free DC isolation technique (Fig. 1). Moreover, MR-mediated endocytosis by bone marrow-derived DCs was abolished when lysates from collagenase-digested spleens were added, confirming their ability to block the MR (Fig. 1).

Fig. 1.

Collagenase digestion products block MR-mediated antigen uptake. (Upper) Splenic DCs from wild-type or MR-deficient mice were isolated by using collagenase digestion for 30 minutes at 37 °C or mashing spleen fragments without collagenase followed by a CD11c-based magnetic separation (Miltenyi Biotec). Subsequently, cells were plated and incubated with 1 μg/mL fluorochrome-labeled OVA for 20 minutes at 37 °C. (Lower) Bone marrow–derived DCs (BM-DCs) were generated as described before (1) and preincubated with a 10-fold dilution of spleen lysates after collagenase digestion. After 15 minutes, 1 μg/mL fluorochrome-labeled OVA was added for another 20 minutes. Internalization of OVA was monitored by flow cytometry.

Second, antigen dose is a major determinant of immunity. Previously, we had injected mice with 0.1 mg OVA and showed that the absence of the MR reduced cross-presentation severely, albeit not completely (1). This confirmed its dominant role in cross-presentation of OVA and indicated the existence of other uptake mechanisms. Segura et al. (4) injected 2 mg, which is 20-fold higher than our dose. Under these conditions, cross-presentation was MR-independent, also in our experimental setup (Fig. 2). However, less unphysiologic doses depended on the MR (Fig. 2). Moreover, commercially available OVA preparations usually contain contaminating fragments, which may be internalized by other mechanisms or may bind directly as peptides to surface MHC-I without intracellular processing. To avoid this problem, we had always removed OVA fragments by gel filtration. Presentation of unpurified OVA contaminated with such fragments is MR-independent (Fig. 2).

Fig. 2.

Importance of antigen dose and purity for investigating MR-dependent cross-presentation. For the in vivo experiments, wild-type or MR-deficient mice were injected with 1.5 × 10⁶ carboxyfluorescein succinimidyl ester (CFSE)-labeled OVA-specific CD8+ T cells (OT-I). After 16 hours, mice were challenged with the indicated amounts of crude or gel-filtrated OVA. After another 48 hours, proliferation of OT-I cells was analyzed by flow cytometry. Graphs depict mean division index ± SEM. For the ex vivo experiments, splenic DCs were isolated without using collagenase; 10⁵ DCs were incubated with the indicated amounts of OVA and 1.5 ± 10⁵ OT-I cells. IL-2 secretion in the supernatant was determined by ELISA after 24 hours. Graphs depict mean IL-2 secretion ± SEM. Gel filtration of OVA solutions was performed using PD10 columns (GE Healthcare).

In summary, studying the role of the MR receptor in cross-priming requires (i) DC isolation by techniques that do not employ collagenase digestion, (ii) clean OVA free of contaminating fragments, and (iii) avoidance of antigen in excessive doses. We conclude that not only inflammatory but also steady-state cross-presentation of OVA is partially MR-dependent.