Selective suicide of cross-presenting CD8⁺ dendritic cells by cytochrome c injection shows functional heterogeneity within this subset

Cyt c Selectively Depletes DCs in Vivo and in Vitro.

To understand the role of cross-presentation in the intact animal, we designed a mouse model that allowed functional ablation of cross-presenting cells. Our strategy used the fact that DCs are highly macropinocytic cells (9) and that efficient cross-presentation involved a specialized machinery for endosome-to-cytosol transport. Release of endogenous cyt c from the mitrochondrion of higher eukaryotic organisms into the cytosol allows engagement of Apaf-1 and initiation of apoptosis within mammalian cells (7). In contrast, cyt c from lower eukaryotic organisms—e.g., yeast—has a low affinity for Apaf-1, does not trigger activation of caspases, and fails to initiate cell death (partly because of trimethylation of lysine-72) (10). Thus, we hypothesized that introduction of exogenous horse cyt c into mice would selectively induce apoptosis in cells that possess this unique capacity for cytosolic transfer, i.e., professional cross-presenting cells.

To test this hypothesis we injected C57BL/6 (B6) mice with either horse or yeast cyt c. After 16–24 h we compared the relative proportion and numbers of splenic CD11c⁺CD4⁻CD8⁺ (CD8⁺), CD11c⁺CD4⁺CD8⁻ (CD4⁺), CD11c⁺CD4⁻CD8⁻ (DN), and total CD11c⁺ cells by flow cytometry. As shown in Fig. 1, a 2- to 3-fold reduction in both the percentage (Fig. 1a) and number (Fig. 1b) of CD8⁺ DCs was consistently observed after administration of horse cyt c (P < 0.0001 compared with untreated, and P = 0.005 compared with yeast cyt c-treated). A small reduction was also seen in the CD4⁺ DC subset after horse cyt c treatment compared with no treatment; however, when compared with yeast cyt c treatment, no significant difference was found in absolute CD4⁺ DC numbers after horse cyt c treatment (P = 0.34). No reduction was observed in the CD4⁻CD8⁻ subset (P = 0.2).

To address whether other splenic populations were affected by cyt c treatment, we repeated the above experiments and stained splenocytes in parallel with CD11b and F4/80, CD19 and B220, or TCR-β and CD3 for analysis by flow cytometry. Horse cyt c treatment resulted in a 10–30% decrease in total DC numbers, but both horse and yeast cyt c had minimal cytotoxic effects on macrophages, B cells, and T cells (Fig. 1c). Larger cyt c doses of up to 15 mg did not enhance the observed reduction in CD8⁺ DCs (data not shown). Hence, all subsequent experiments were performed by using an i.v. dose of 5 mg.

We next sought to establish whether cyt c treatment caused DC apoptosis. Because cyt c requires the presence of Apaf-1 to initiate apoptosis, we investigated the impact of Apaf-1 deficiency. In addition, this would dispel the possibility that our cyt c preparation was acting through a contaminant. Because most Apaf-1^−/− mice die before birth, we made chimeras using mutant or WT fetal liver reconstitutions into bm1 host mice. These mice were chosen as hosts because cells from bm1 mice are incapable of directly presenting OVA_257–264 to OT-I CD8⁺ T cells (11). Six to 12 weeks after reconstitution, the proportion of DCs/lymphocytes in Apaf-1^−/−- or WT-reconstituted mice was similar to that of B6 mice (data not shown).

To determine the absolute numbers of DCs remaining in vitro after cyt c treatment, we cultured freshly isolated splenic DCs from WT or Apaf-1^−/− mice in the presence of 0–10 mg/ml cyt c. After 24 h we quantified the numbers from each DC subset by calibrating against a known quantity of beads. Analysis of the absolute cell numbers revealed a 3-fold reduction in CD8⁺ DCs from WT B6 mice (P = 0.0001), whereas the CD8⁻ subset (P = 0.2) and both CD8⁺ and CD8⁻ DCs of Apaf-1^−/− origin were not significantly affected (Fig. 1d).

Apoptosis of DCs from naïve B6 or Apaf-1^−/−-reconstituted mice was also measured by annexin V assays of purified DCs cultured with 0–10 mg/ml cyt c for 6 h. At the highest concentration used, cyt c induced apoptosis in almost 25% of splenic CD8⁺ DC from B6 mice, as detected by annexin V binding (Fig. 1e). In contrast, cyt c failed to induce early apoptosis in DCs of Apaf-1^−/− origin (Fig. 1e). The observed effect was dose-dependent (data not shown).

It should be noted that the preferential cyt c-induced suicide of CD8⁺ DC was not due to preferential initial uptake. Radio-iodinated cyt c was taken up equally by CD8⁺ DCs and CD8⁻ DCs (1,976 ± 676 vs. 1,720 ± 814 cpm) within 2 h; we used yeast cyt c to avoid any apoptogenic effects. Moreover, FITC-labeled horse cyt c bound equivalently to both DC subsets after 30 min at 37°C.

Cyt c Treatment Impairs Cross-Presentation in Vivo and in Vitro.

To verify whether the cyt c-induced suicide of CD8⁺ DCs was functionally significant, we determined whether cross-presentation of cellular Ag was preferentially impaired in vivo. We adoptively transferred 1 × 10⁶ CFSE-labeled OT-I cells into B6 mice on day −2, then injected mice with three daily doses of cyt c starting the day before i.v. injection of two forms of cellular Ag: either OVA-coated bm1 splenocytes (OCS) (Fig. 2a) or OVA expressed in a pig xenogeneic cell line, PK15-OVA (Fig. 2b). Three injections of cyt c were used because of its very short circulating half-life of 4 min (12) and because DCs have a rapid turnover rate. We then assessed cross-presentation 60 h after mice had received Ag by measuring OT-I proliferative responses in the spleen. Cross-presentation was profoundly inhibited in cyt c-treated mice; the overall number of expanded OT-I cells was up to 27-fold lower in cyt c-treated mice than that of untreated control B6 mice (Fig. 2a Right).

Next we investigated the effect on CD4⁺ proliferative responses. Adoptive transfer of equal numbers of CFSE-labeled OT-I and OT-II cells were injected into the same animal. When we compared MHC class II presentation to MHC class I (cross)-presentation, we observed that cyt c treatment preferentially abrogated OT-I proliferation but resulted in a modest reduction in OT-II proliferation (Fig. 2b). This effect was consistently seen when OVA transfectants (Fig. 2b) or OVA-coated cells (data not shown) were used as immunogens.

To test whether the suppression of CD8⁺ T cell proliferation in vivo was nonspecific, we performed similar experiments in CD11c-mOVA mice, which express membrane-bound OVA in DCs under the control of the CD11c promoter. Thus, OVA is constitutively synthesized and presented via the classical MHC class I pathway. When CFSE-labeled OT-I cells were adoptively transferred into these mice, equivalent proliferative responses were observed in mice treated with cyt c compared with mice that were not (data not shown).

To extend our in vivo observations, splenic CD8⁺, CD4⁺, and CD4⁻CD8⁻ DC subsets were cultured continuously in vitro in the presence of 2 mg/ml cyt c and OVA. Cyt c selectively abrogated the ability of CD8⁺ DC to cross-present OVA to OT-I cells [supporting information (SI) Fig. 8a] in a dose-dependent manner (data not shown). The inability of CD8⁺ DCs to stimulate OT-I proliferation in the presence of cyt c was not related to direct toxicity of cyt c on T cells in vitro, because all subsets of DCs were able to stimulate robust OT-I proliferative responses when incubated with OVA_257–264 peptide in vitro (SI Fig. 8b). Thus, our in vitro studies confirmed our in vivo cross-presentation results.

Cyt c-Mediated Inhibition of Cross-Presentation Is Apaf-1-Dependent.

To confirm the mechanistic specificity of the observed reduction in cross-presentation, we compared how exogenous cyt c affected the ability to stimulate naïve OT-I proliferation in Apaf-1^−/−- or WT-reconstituted mice. As expected, cross-presentation was inhibited in WT-reconstituted mice that received cyt c (Fig. 3a Upper). In contrast, cyt c treatment did not reduce the frequency (Fig. 3a Lower) and overall numbers (Fig. 3b) of expanded OT-I cells in Apaf-1^−/− mice. The level of cross-presentation in untreated Apaf-1^−/−-reconstituted mice was slightly lower than that observed in WT-reconstituted mice, but this was not attributable to lower hematopoietic reconstitution because the reconstitution efficiency of the blood and DC compartments was comparable in Apaf-1^−/− and WT mice. Regardless, the key issue is that there was no difference in the internal comparison between the presence or absence of cyt c within the Apaf-1^−/−-reconstituted mice. Thus, these results cogently indicate that exogenous cyt c resulted in apoptosis of cross-presenting cells through an Apaf-1-dependent pathway.

Cyt c Treatment Preferentially Reduces CTL Induction and Ag-Specific Proliferative Response.

Although we have shown that cyt c treatment inhibited cross-presentation of cell-associated Ag to OT-I cells in vivo, it was unclear whether this phenomenon was merely idiosyncratic of transgenic T cells. Thus, we designed a model whereby both T cells and DCs were of endogenous origin. We pretreated B6 mice with cyt c on three consecutive days, commencing a day before injection of OCS, and assessed in vivo CTL activity 7 d after T cell priming. There was almost 4-fold less OVA-specific CTL activity in cyt c-treated mice compared with untreated mice (Fig. 4a).

To assess endogenous anti-OVA proliferative responses (compared with OT-I responses), mice were pretreated with cyt c as described above. CFSE-labeled splenocytes were restimulated in vitro 7 d later with OVA. After 4 d we assessed proliferation of endogenous CD4⁺ and CD8⁺ T cells by CFSE dye dilution. A 2- to 3-fold impairment of CD8⁺ T cell expansion was observed in mice treated with cyt c with minimal nonspecific proliferation in unimmunized mice (Fig. 4 b and c). In contrast, there was minimal impairment of endogenous CD4⁺ T cell responses (Fig. 4d). These results strongly indicate that cyt c inhibited endogenous CD8⁺ T cell responses (both proliferative and killer responses) to cross-presentation in vivo.

Cyt c Impairs Immune Rejection of B Cell Lymphoma.

We also analyzed the effect of cyt c on the elimination of OVA-expressing Eμ-myc B cell lymphomas. Mice were either left untreated or given three doses of cyt c at the time of priming with OCS. After 7 d, 1 × 10⁶ Eμ-myc-OVA or Eμ-myc-GFP cells were inoculated i.v. into primed or naïve mice. Mice were left for 7 d to allow spontaneous lymphoma development. A robust Ag-specific tumor rejection response was detected in the spleens of mice that had been primed with Eμ-myc-OVA but not Eμ-myc-GFP (Fig. 5a). Cyt c-treated mice had 3-fold more tumor cells than untreated controls (Fig. 5 b and c), indicating that in vivo disruption of cross-presentation by cyt c impaired immunity to tumor Ag.

Cyt c-Resistant DCs Are Incapable of Cross-Presentation but Retain a Capacity for MHC Class II Presentation.

Despite a considerable remaining population of CD8⁺ DCs after cyt c treatment (Fig. 1a), cross-presentation was still markedly impaired. Therefore, it was important to determine the Ag presentation capacity of residual DCs. To this end, we treated B6 mice with cyt c for 24 h and identified a 50% decrease in the proportion of CD8⁺ DCs (see Fig. 1a). We cultured equal numbers of sorted CD8⁺ or CD8⁻ DCs from untreated or cyt c-treated mice with various concentrations of OVA protein and CFSE-labeled OT-I cells. CD8⁺ DCs from cyt c-treated mice were far less capable of stimulating OT-I proliferation in vitro than were CD8⁺ DCs from untreated mice (Fig. 6a). Indeed, the level of proliferation seen in cyt c-treated mice approximated that of CD8⁻ DCs from untreated mice, which are known for their poor cross-presentation capacity. Using CFSE-labeled OT-II cells, we showed that the remaining (cyt c-resistant) DCs were defective only in MHC class I but not MHC class II presentation ability because these DCs could present newly encountered OVA protein to OT-II cells equally well compared with DCs from untreated mice (Fig. 6b).

Cyt c Treatment Preferentially Reduces IL-12 Production and Toll-Like Receptor 3 (TLR3) Expression by CD8⁺ DCs.

Apart from their efficiency at cross-presentation, another functional specialization of CD8⁺ DCs is their propensity for IL-12 production in response to TLR engagement. Given that CD8⁺ DCs are the main targets of cyt c-mediated depletion, we compared the IL-12-producing capacity of residual cyt c-treated CD8⁺ DCs to their untreated counterparts. Twenty four hours after treatment with cyt c, remaining CD8⁺ or CD8⁻ DC populations from untreated or cyt c-treated B6 mice were stimulated with CpG and IFN-γ. The presence of IL-12p40 and IL-12p70 was assayed from cultured supernatants after 24 h. On a per-cell basis, CD8⁺ DCs from cyt c-treated mice produced nearly 2-fold less IL-12p40 (data not shown) and ≈4-fold less IL-12p70 than did CD8⁺ DCs from untreated mice (Fig. 7a). IL-12p70 was undetectable from CD8⁻ DCs stimulated with CpG and IFN-γ or from all unstimulated DCs (Fig. 7a).

CD8⁺ DCs have also been reported to express high levels of TLR3, and this has been linked to their cross-priming ability (13). When residual CD8⁺ DCs from cyt c-treated mice were quantitatively analyzed for their TLR3 expression by real-time PCR, we found a 1.5- to 2-fold reduction in the expression of this receptor, whereas very low levels of TLR3 were expressed on CD8⁻ DCs (Fig. 7b).

Thus, we concluded that cyt c reduced the proportion of CD8⁺ DCs that abundantly expressed IL-12 and TLR3.

Mice and Reagents.

C57BL/6 (B6), bm1 (11), OT-I (34), OT-II (35), and CD11c-mOVA (36) (from R. Steptoe, Centre for Immunology and Cancer Research, Brisbane, Australia) mice have been described. Reconstitution of lethally irradiated bm1 hosts with embryonic day 14.5 Apaf-1^−/− fetal liver cells was performed as described (37). OVA protein, cyt c from horse heart, and Saccharomyces cerevisiae were obtained from Sigma. Mice received 5 mg of horse or yeast cyt c i.v. dissolved in PBS.

Splenic DC Isolation and Analysis.

Splenic DCs isolated as previously described (15) were used for (i) high-speed sorting (MoFlo; Cytomation), (ii) immunomagnetic bead purification (AutoMACS; Miltenyi Biotec), or (iii) immunofluorescent labeling before analysis.

Immunofluorescent Staining.

Anti-CD11c, CD4, CD8, CD3, CD11b, CD19, Vα2, B220, F4/80, TCRβ, annexin V, and streptavidin were purchased from Pharmingenor Caltag. Propidium iodide was added at 1 μg/ml just before analysis.

Cell Lines.

The pig kidney cell line PK15 (American Type Culture Collection) was stably lipofected with OVA DNA (designated PK15-OVA) and maintained in the presence of 1.1 mg/ml geneticin (G418; GIBCO/BRL) and 4 μg/ml puromycin (Sigma). Stable Eμ-myc cell lines (38) bearing GFP (Eμ-myc-GFP) or GFP and OVA (Eμ-myc-OVA) were generated by retroviral transduction (39).

Proliferation and CTL Assays.

Usually OCS (36) were used as the immunogen. In vivo OT-I and OT-II proliferation assays, in vitro proliferation, and in vivo CTL assays were performed as described (36).

Endogenous Proliferative Responses.

A modified CFSE dilution method was used (40). B6 mice received three daily doses of cyt c starting the day before i.v. immunization with 2 × 10⁷ OCS. A week later, equivalent numbers of splenocytes were labeled with 5 μM CFSE (Molecular Probes) for 10 min at 37°C. CFSE-labeled cells were plated at 5 × 10⁶ per milliliter with OVA at 37°C for 4 d. Cells were enriched for viable lymphocytes by using density centrifugation. FACS analysis was performed (FACSAria; BD Biosciences).

RNA Extraction and Quantitative Real-Time PCR.

RNA extraction and quantitative real-time RT-PCR analysis were performed as described (41).

IL-12 ELISA.

ELISAs for murine IL-12 by Pharmingen were used on culture supernatants of DCs stimulated with 0.5 μM CpG1688 (Geneworks), 20 ng/ml IFN-γ (Pharmingen), and 100 ng/ml GM-CSF (PeproTech).

Statistical Analysis.

Student's t test was used to compare two sets of data.