Immunological Synapse Formation Induces Mitochondrial Clustering and Mitophagy in Dendritic Cells

Immune responses are initiated in the lymph nodes by a specific interaction of T cells and dendritic cells (DCs). During these interactions, a specialized cell–cell junction called the immunological synapse (IS) is formed in the zone of contact between a DC and T cell (1–4). The DC and the T cell part of the IS include specific molecular components, including surface proteins and associated cytoplasmic cytoskeletal and signaling molecules, which may result in different functionality in these two regions (5, 6). Most studies have focused on the IS of T cells, implying that the information available on the IS of DCs is relatively sparse (5–10). Autophagy is a homeostatic process involved in the degradation and recycling of proteins and organelles (11, 12). It involves the formation of double-membrane vesicles called autophagosomes, which enclose part of the cytoplasm and, subsequently, fuse with lysosomes, where their cytoplasmic protein content is digested by proteases. Autophagy is controlled by autophagy-related (Atg) proteins, which drive the formation of autophagosomes (11, 12). In this regard, autophagosome formation requires the conjugation of Atg12 to Atg5 and also phosphatidylethanolamine to the microtubule-associated protein 1 L chain 3 (LC3-I) to form the autophagosome-associated LC3-phosphatidylethanolamine (also called LC3-II) (13). LC3-II is a bona fide marker of autophagy because it binds specifically to the autophagosome membrane (11, 12). The degradation of the intracellular components of the autophagosome allows the recycling of cellular components. Excess of nutrients and availability of growth factors lead to the inhibition of autophagy because under these conditions the autophagy inhibitor mechanistic target of rapamycin complex 1 (mTORC1) is activated (14).

Mitochondria are an important source of reactive oxygen species (ROS) in cells (15, 16). These ROS can cause depolarization and damage of these organelles, leading to high levels of oxidative stress and cell death (17). Mitophagy is a form of autophagy that controls the selective clearance of damaged mitochondria (18). Its regulation involves molecular components that drive the general autophagy machinery and, in addition, mitophagy-specific elements, including the kinase PTEN-induced putative kinase 1 (PINK1) and the E3-ubiquitin ligase Parkin (19, 20). Upon depolarization and damage of the mitochondria, PINK1 becomes stabilized on the outer membrane of these organelles, leading to the recruitment of Parkin that tags mitochondria with ubiquitin and leads to their degradation by the lysosomes (19, 20). In this article, we show that IS formation promotes local mitochondrial clustering, depolarization of the mitochondria, increase in ROS levels, and mTORC1-independent autophagy/mitophagy, which can remove depolarized/damaged mitochondria. The data gathered provide novel information on the molecular mechanisms operating in the DC side of the IS.

CFSE, JC-1, CellROX Green, Alexa Fluor 568 goat anti-rabbit IgG, Alexa Fluor 647 goat anti–guinea pig IgG, and MitoTracker Red were from Molecular Probes (Leiden, the Netherlands). Dihydroethidium (DHE), carbonyl cyanide m-chlorophenyl hydrazone (CCCP), polyornithine, Hoechst 33342, BSA, LPS, Pepstatine A, E-64d, and S3QEL2 obtained were from Sigma-Aldrich (St. Louis, MO). DCF (2′,7′-dihydrodichlorofluorescein diacetate) was obtained from Invitrogen (Carlsbad, CA). The OVA_323–339 peptide (ISQAVHAAHAEINEAGR) and the citrullinated human vimentin_59–78 peptide (GVYAT(Cit)SSAV(Cit)L(Cit)SSVPGVR) were obtained from GenScript, and the human fibrinogen α-chain_79–91 peptide (QDFTNRNKLKNS) was from Cayman Chemical (Ann Arbor, MI). The anti-LC3B and the anti-p–dynamin-related protein 1 (Drp-1) (Ser616) Abs were from Cell Signaling Technology (Beverly, MA); the anti-Atg5 Ab was from Abnova (Taipei, Taiwan); and the anti-Atg12 (C-6), the anti-Parkin (PRK8), the anti–translocase of the outer membrane of mitochondria 20 (TOM20), and the anti-p70 S6K (Thr421/Ser424) Abs were from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-Beclin1 and anti PINK1 Abs were from Novus Biologicals (Cambridge, U.K.). The phosphatidylethanolamine anti-mouse I-A/I-E Ab and anti-mouse CD11c Abs were from BioLegend (San Diego, CA). The phosphatidylethanolamine-conjugated rat anti-mouse Vα2T Cell receptor mAb was from BD Pharmingen.

C57BL/6 wild-type (WT) and OTII mice (21) (male animals of 8 wk) were maintained in the animal facility at the Centro de Investigaciones Biológicas/Consejo Superior de Investigaciones Científicas. GFP-LC3 transgenic mice on C57BL/6 background (22) were provided by Noboru Mizushima (Tokyo Medical and Dental University, Japan). Mice were treated according to Animal Care Committee guidelines of the CIB.

Murine splenic DCs and CD4 were purified as indicated before (8, 23). In some experiments CD4 OTII T cells were labeled for 30 min at 37°C with the fluorescent cell tracker CFSE (5 μM) in 0.1% BSA in PBS. In the experiments of localization of the mitochondria, the DCs were labeled for 30 min at 37°C with MitoTracker Red (200 nM) in RPMI 1640 medium. IS formation was induced as indicated before (8, 23). Briefly, first the DCs were pulsed in RPMI 1640 medium for 30 min with the specific Ag OVA_323−339 peptide or, in the case of the controls, with either of two nonspecific Ags, namely, human citrullinated vimentin peptide or human citrullinated fibrinogen. In some control experiments we also used DCs not pulsed with peptides. Subsequently, unloaded DCs or peptide-loaded DCs and the CD4 OTII T cells (ratio 1 DC: 5 CD4 T cells) were dissolved in complete medium (10% FCS in RPMI 1640 medium). The cells were spun (50 × g, 5 min) in a conical tube and then maintained in the incubator in complete medium to foster IS formation for 15 min (unless otherwise is indicated), as indicated in the legends to the figures.

The Leica Application Software Advance Fluoresce Lite was used to measure both, in the MitoTracker-labeled and in the anti-TOM20–stained DCs that formed IS with CD4 T cells, the fluorescence ratio between two regions of similar areas, one close and another distal to the IS of the DC. With the values of the fluorescence ratio of all the DCs—MitoTracker labeled or anti-TOM20 stained—examined, the mean fluorescent ratio ± SD of the DCs forming Ag-specific or non–Ag-specific IS was determined.

The DCs were pulsed with OVA_323−339 peptide for 30 min in RPMI 1640 medium. Subsequently, the cells were transferred to 10% FCS in RPMI 1640 medium and loaded with JC-1 (final concentration 10 μM) at 37°C, for 10 min. The DCs were washed twice with medium, mixed with CD4 T cells and centrifuged (800 × g) for 5 min to allow synapsis formation. The DCs were left in the medium for an additional 30 min. The cells were centrifuged, suspended in cold PBS, and analyzed by flow cytometry. The gating of single or conjugated JC-1–labeled DCs was based on forward and side scatter light characteristics.

ROS accumulation in live DCs was measured using DHE and CellROX Green, using the fluorescence microscope, or with DCF, using flow cytometry. DHE or CellROX (data not shown) were added to the OVA-loaded DCs (at a final concentration of 50 μM) simultaneously with the CFSE-labeled CD4 T cells (see above IS formation). Subsequently, the DCs were allowed to form IS with the CD4 OTII T cells as indicated above. After 30 min the conjugates were fixed with 4% formaldehyde at room temperature, washed in PBS, and plated on dishes coated with polyornithine (20 μg/ml) for at least 30 min. The cells stained with DHE were permeabilized with cold methanol, stained with Hoechst 33342 (5 μg/ml) in PBS 0.1% BSA, and mounted in fluorescent mounting medium. Cells stained with CellROX were directly mounted with fluorescent medium. The fluorescence intensity of the DCs stained with DHE or CellROX was quantified with the Leica Confocal Software.

To analyze ROS levels in DCs using cytometry, these cells were labeled for 25 min at 37°C with DCF (50 μM) in serum-free RPMI 1640 medium prior to the induction of IS with unlabeled CD4 T cells. Subsequently, the levels of ROS of the DCs in the conjugates were analyzed by flow cytometry at time 0 and 15 min after IS formation. For the analysis of the DCF-positive DCs, we gated on a region that includes DC-CD4 T cell conjugates, which largely excludes T cells (Supplemental Fig. 2).

Flow cytometry analysis of autophagy of GFP-LC3-II–expressing DCs was performed as described before (24). This technique is based on the GFP quenching in an acidic environment after autophagosome–lysosome fusion. The reduction in the fluorescence emission of LC3-GFP correlates with the degree of autophagy induction (24). The cells were then analyzed on Coulter Epics XL cytofluorometer using a CXP Analysis software (Beckman Coulter).

Cells were fixed in 4% glutaraldehyde in phosphate buffer (pH 7.4) for 30 min at 4°C, postfixed in 1% (w/v) osmium tetroxide in 3% (w/v) potassium ferrocyanide in PBS, dehydrated with a graded series of ethanol, and embedded in LX-112 resin. Ultrathin sections were doubly stained with uranyl acetate and lead citrate and observed under a JEOL JEM-1230 electron microscope.

Immunofluorescences were performed as described (8, 25, 26). After the fixation with 4% paraformaldehyde in PBS (10 min), the cells were permeabilized at room temperature (10 min) with 0.2% Triton X-100. Before staining with specific Abs, the cells were first treated with anti-mouse CD16/32 to block Ab FcR binding, and with 1% BSA (15 min) to block unspecific binding. Subsequently, cells were stained with suitable primary or secondary Abs as indicated before (8, 25). In the experiments in which GFP-LC3 was analyzed, the cells were plated at 37°C for 30 min onto coverslips coated with polyornithine (20 μg/ml). In some experiments the DCs were pretreated with the protease inhibitors E64d and pepstatin A, both at 10 μg/ml, to inhibit lysosomal proteases. Before fluorescence microscopy analysis, the DCs expressing GFP-LC3-II were subjected to a brief wash with PBS including saponin (0.05%) to remove the soluble GFP-LC3 and ensure that only autophagosome-associated GFP-LC3 was analyzed (26). In all cases, immunostained samples and experiments with LC3-GFP–expressing DCs, prior to mounting, the samples were extensively washed with PBS and distilled water. Subsequently, coverslips were mounted in fluorescent mounting medium, and representative fields were photographed through oil immersion lens. Laser-scanning confocal microscopy was performed with argon and helium/neon laser beams and attached to an Ultra-spectral Leica TCS-SP2-AOBS inverted epifluorescence microscope using oil immersion objectives. Image analysis was performed using Adobe Photoshop 7.0 (Adobe System).

DCs (500 × 10³ cells) were solubilized in SDS-PAGE sample buffer and analyzed by Western blotting as indicated before (8, 25).

The graphs shown represent the mean ± SD of at least three different experiments. Each experiment was performed using three to six animals whose purified splenic DCs or CD4 T cells were pooled together. In the fluorescent microscopy analyses, we analyzed 50–100 IS-forming DCs or single DCs in each experiment. Significance of differences between results was assessed using the Student’ unpaired t test. The p values < 0.05 were considered significant.

We analyzed whether IS formation induces redistribution of mitochondria in DCs. DCs pulsed with OVA peptides were allowed to form IS with the OVA peptide–specific OTII CD4 T cells. Electronic microscopy (EM) analysis showed that in the IS-forming DCs, mitochondria tend to cluster close to the IS region (Fig. 1A). The analysis by confocal microscopy of IS formed between OVA-loaded DCs labeled with MitoTracker, a cell-permeant mitochondria-specific fluorescent dye, and OTII CD4 T cells showed that mitochondria consistently localized to the IS region of DCs (Fig. 1B, upper). In another set of experiments, IS formation was induced with OVA-loaded DCs and OTII CD4 T and, subsequently, the mitochondria were immunostained using Abs against the mitochondrial marker TOM20. The analysis of the conjugates by confocal microscopy corroborated that mitochondria tend to cluster in the vicinity of the IS region of the DCs (Fig. 1C, upper). We performed control experiments in which OTII CD4 T cells were allowed to interact either with DCs loaded with human citrullinated vimentin peptides (Fig. 1B, lower, and Fig. 1C, lower) or with human citrullinated fibrinogen–loaded DCs, or with unpulsed DCs; in addition, we also analyzed WT CD4 T cells interacting with DCs loaded with OVA (Supplemental Fig. 1). Consistent with prior reports (27, 28), we observed Ag-nonspecific IS formation among control DCs (unloaded, citrullinated vimentin, or fibrinogen loaded) and OTII T cells or between pulsed DCs and WT CD4 T cells, although in these controls, a significantly lower number of the interacting DCs displayed mitochondria in the DC-OTII T cell contact region compared with the OVA-loaded DCs interacting with OTII T cells. These results indicate that largely Ag-specific contacts promote mitochondria clustering in the region of contact between DCs and T cells.

As electrons that leak from the mitochondrial respiratory chain can generate ROS that can depolarize/damage the mitochondria (17, 29), we tested whether, in parallel to the observed mitochondrial clustering induced by IS formation, there was an increase in the number of depolarized/damaged mitochondria in the DC. OVA-DCs were labeled with JC-1, a cationic dye whose fluorescence emission shifts from red to green as the mitochondrial membrane potential decreases (30). These DCs were allowed to form IS with OTII CD4 T cells, and the conjugates were analyzed by flow cytometry (see in Supplemental Fig. 2 the strategy used to analyze the DCs in the DC-CD4 T cell conjugates). Control OVA-DCs were stimulated with the protonophore CCCP, which depolarizes mitochondria. The IS-forming DCs displayed almost a 40% reduction in the red/green JC-1 fluorescence ratio (Fig. 2A). This loss in the mitochondrial membrane potential was similar to that induced by CCCP (Fig. 2A). To analyze whether the depolarization/damage of mitochondria in DCs could correlate with an increase in the ROS generated by these organelles (17), OVA-DCs were allowed to form IS with OTII CD4 T cells in the presence of DHE, a membrane-permeable agent that emits red fluorescence upon being oxidized by superoxide radicals (31, 32). As shown in Fig. 2B (upper), DCs that formed IS displayed an increased DHE fluorescence when compared with single DCs, indicating that IS formation enhances ROS levels. Interestingly, ROS generation was observed only when CD4 OTII T cells formed Ag-specific IS with OVA peptide DCs but not with unloaded DCs (Fig. 2B, lower), suggesting that only Ag-specific IS formation is able to induce significant ROS increase in DCs. Similar results were obtained when ROS production was analyzed with CellROX, another membrane-permeable agent that emits green fluorescence upon oxidation by ROS (data not shown).

To confirm that the ROS observed in the prior experiments were generated in the mitochondria, DCs were treated with mitoquinone (MitoQ), an agent that selectively blocks mitochondrial oxidative damage (33, 34). OVA-DCs pretreated either with MitoQ or control compound triphenylphosphonium (TPP) (34) were allowed to form IS with OTII T cells in the presence of DHE (Fig. 3A). Analysis of DHE fluorescence showed that the increase in ROS upon IS formation was abrogated in the DCs pretreated with MitoQ, but not with TPP (Fig. 3A), indicating that mitochondria were the most important source of ROS upon IS formation. To confirm that the ROS observed upon IS formation were generated in the mitochondria of the DCs, these cells were treated with S3QEL2, a novel pharmacological agent that selectively inhibits ROS production in these organelles (35). DCs, treated or not with S3QEL2, were labeled with DCF and then allowed to form IS with OTII CD4 T cells. Subsequently, the increase in the levels of ROS in the DCs that formed IS with CD4 T cells were analyzed by flow cytometry (2Materials and Methods). As shown in Fig. 3B, when the DCs that formed IS had been previously treated with S3QEL2, they generated lower levels of ROS compared with the control DCs (compare control DC-CD4 [t = 15 min] and S3QEL2 DC-CD4 [t = 15 min]). As shown in Fig. 3, the inhibitory effects exerted by MitoQ on the ROS produced by IS forming DCs were stronger compared with the effects of S3QEL2, pointing out differences between these two mitochondrial ROS inhibitors. Of note, NS differences were observed when the levels of ROS were compared in the IS-forming DCs treated with MitoQ or with S3QEL2 (t = 15 min) and the DC controls (t = 0) (compare S3QEL2 DC-CD4 [t = 15 min] and control DC-CD4 [t = 0 min] and TPP DC and control MitoQ DC-CD4]. Importantly, the treatment of the DCs with the inhibitors did not affect the number of IS formed between the DCs and CD4 T cells (data not shown). In summary, together the results indicate that inhibition of mitochondrial ROS reduce the ROS levels observed in the DCs after IS formation to almost basal levels, suggesting that upon IS formation and concomitant with observed clustering of the mitochondria in the IS, there is an increase in mitochondrial ROS in the DCs.

It has been shown that damaged mitochondria can be selectively cleared by mitophagy (19, 20). As mitophagy is a type of autophagy, we analyzed if autophagy was taking place locally in the IS region of DCs. For this purpose, we induced IS formation between OVA-pulsed DCs obtained from GFP-LC3 C57/BL6 transgenic mice (13) and OTII CD4 T cells. The conjugates were plated onto polyornithine-coated dishes, and unconjugated and IS-forming DCs were analyzed for the presence of the autophagosome marker LC3-II by confocal microscopy. Autophagy experiments were carried out in the presence of the lysosomal protease inhibitors (E64d and pepstatin A) to ensure that the measured changes in autophagy were due to an increase in autophagic flux. A significant increase in the GFP-LC3-II puncta in the vicinity of the IS region of DCs was observed, indicating that autophagy takes place in this region (Fig. 4A). Notably, IS-forming DCs displayed an increase in autophagy that was 3.5-fold higher than that observed when unconjugated DCs were placed in glucose-free medium, a condition that upregulates autophagy (Supplemental Fig. 3). Autophagy induction required Ag-specific interactions because it was not observed when OTII CD4 T cells were allowed to interact with DCs not loaded with OVA peptide (data not shown). As shown in Fig. 4B, we corroborated that IS formation induces autophagy in DCs using a cytometric technique that correlates the reduction in the fluorescence of the LC3-GFP–expressing DCs with the increase in autophagy (24). Immunoblot analysis performed with extracts obtained from OVA peptide–pulsed GFP-LC3 DCs that formed IS with OTII CD4 T cells further confirmed that IS formation leads to an increase of the autophagy marker GFP-LC3-II in DCs (Fig. 4C). Before, we showed that IS formation leads to the activation of the kinase Akt in DCs (8), which suggests that the Akt target mTORC1 is active in the IS-forming DCs. As mTORC1 inhibits autophagy in many cells (14), the observed IS-induced increase in autophagy was contradictory. To confirm that mTORC1 was active in DCs upon IS formation, we induced IS formation between DCs and OTII CD4 T cells and immunostained the cells with an Ab against an active form of S6 kinase, which is a direct target of mTORC1. As shown in Fig. 4D, mTORC1 was more active in the IS-forming DCs compared with the unconjugated DCs, suggesting that IS formation induces autophagy in these cells even when mTORC1 is active.

If autophagy is increased in DC upon IS formation, it is expected that autophagy markers were upregulated. We analyzed whether key regulators of autophagy (9, 10) were upregulated in the vicinity of the IS of DC. Immunofluorescence analysis showed that Beclin1 (Fig. 5A), Atg5 (Fig. 5B), and Atg12 (Fig. 5C) clustered in this region. We also analyzed whether mitophagy could be also upregulated in the IS region of DCs, suggesting that this process could be involved in the removal of depolarized/damaged mitochondria. OVA-DCs’ mitochondria were labeled with MitoTracker, and then these cells were allowed to form IS with OTII CD4 T cells. The conjugates obtained were then stained with Abs against the mitophagy regulators PINK1 or Parkin and analyzed by confocal microscopy. Both PINK1 (Fig. 6A) and Parkin (Fig. 6B) stained part of the MitoTracker-labeled mitochondria in the IS region of DCs, suggesting that mitophagy was active in this region. Mitochondrial fission/fusion is an important mechanism for controlling mitochondrial quality (36). Upon fission, daughter depolarized mitochondria can be cleared out by mitophagy, whereas polarized daughter mitochondria are preserved (36). As mitochondria fission precedes mitophagy (36), we tested whether a phosphorylated form of the regulator Drp-1 that promotes fission (p-Drp1-S616) (37) localizes in the IS region of DCs. Phosphorylated Drp1 colocalized with mitochondria to the IS of DCs, suggesting active mitochondrial fission in this region (Fig. 6C). Finally, EM analysis of conjugates showed autophagosomes and isolation membranes surrounding the mitochondria in the vicinity of the IS region of DCs (Fig. 7), further suggesting the autophagy/mitophagy was taking place in this region (Fig. 8).

The IS is a cell–cell contact region formed during T cell activation at the zone of contact between T cells and DCs. The IS is in fact a bipartite structure that includes specific molecular components at the T cell and DC sides. Each side represents a different superstructure resulting in different functionality and, arguably, complementary effects on T cell activation (5, 6). Most studies on the role of the IS in T cell activation have focused on the T cell side (1–4), and much less information is available on the DC side of this structure. In this regard, emerging data suggest that the IS of DCs may also play important roles during T cell activation (5–10). Previously, it has been shown that this region regulates the survival of DCs and the polarized delivery of cytokines to the opposite T cell (8, 9). In this article, it is shown that IS formation leads in DCs to the translocation of mitochondria to the vicinity of the IS region, partial depolarization of these organelles, increase in mitochondrial ROS, and local upregulation of autophagy and mitophagy. In Fig. 8, we present a model suggesting how these processes could be taking place at the DC side of the IS.

With respect to the translocation of mitochondria to the IS region of DCs, it has been shown that these organelles also accumulate at the T cell side of the IS (38, 39). In T cells it has been suggested that mitochondria clustered in the vicinity of the IS regulate calcium levels, pointing out a novel mechanism whereby IS formation could mediate T cell activation (38). It is also possible that mitochondria in the IS region of both the T cell and DC could be also a local store of ATP necessary to maintain cytoskeletal actin dynamics and tension required to keep integrin-mediated DC–T cell interactions (40). Future studies are required to determine the precise role(s) performed by mitochondria in the DC side of the IS.

Regarding the increase in ROS levels observed in DCs upon IS formation, we speculate that as mitochondria are an important source of ROS (15, 16), the proximity among these organelles in the synaptic region could contribute to their mutual damage, favoring their depolarization and the subsequent increase in ROS (Fig. 8). Consistent with this scenery, we observe that IS formation leads to the depolarization of the mitochondria in DCs. The activation of autophagy/mitophagy upon IS formation could be a homeostatic mechanism involved in the clearance of depolarized/damaged mitochondria in this region, preventing further ROS production and the damage of the DCs, which may lead to impairment in T cell activation (Fig. 8). In this line, previously it has been suggested that the increase in mitochondrial ROS rapidly induces mitophagy (29). Supporting that autophagy/mitophagy could play homeostatic roles at the IS of DCs in vivo, studies suggest that T cell activation is impaired in mice with autophagy-deficient DCs (41). The maintenance of the homeostasis of the IS in DCs could represent an additional function for autophagy/mitophagy in the immune system, added to other functions already described for this homeostatic mechanism in cells, including pathogen removal and Ag processing (12, 42, 43). Predictably, autophagy/mitophagy would be similarly activated to remove faulty mitochondria in other settings where these organelles are also found clustered, including the IS and the uropod of T cells (38, 39, 44, 45).

Interestingly, IS-induced autophagy in DCs takes place even when the autophagy-inhibitor mTORC1 is active, implying mechanisms that circumvent the inhibitory effects of this kinase on autophagy (14, 46). However, mTORC1-independent autophagy regulation has also been described in other leukocytes and contexts, suggesting that this process can be regulated through mTORC1-dependent and -independent pathways (47, 48). Recently, another group indicated that IS formation induces autophagy in DCs (49); however, the authors observed that autophagosomes locate near the membrane in the periphery of DCs, but not selectively in the IS region. They also suggest that autophagy promotes IS disruption, which we did not observe in our experiments. The reasons for these discrepancies are not clear; however, the type of leukocytes used in the experiments may explain the different results. Mentioned authors analyzed IS formed between murine OVA-loaded bone marrow–derived DCs and OTII splenocytes (49), which have been shown to be an extremely diverse population of cells (50), and between human monocyte-derived DCs and allogeneic T cells. We used purified OTII CD4 T cells and OVA-loaded CD11c⁺ splenic DCs, which are bona fide DCs differentiated in the animals. We also think that the suggestion that IS-induced autophagy, which takes place early upon IS formation, causes IS disruption is not consistent with in vivo observations showing that T cell activation by DCs requires a long period of IS formation (51, 52). In summary, our data provide insight on the molecular processes that take place in the IS of DCs. The results obtained indicate that the IS of DCs is a complex structure, where homeostatic mechanisms important to maintain the integrity of the DCs and T cell activation may be very efficient.

We acknowledge C. Ardavín for the OTII mice, N. Mizushima for the GFP-LC3 transgenic mice, P. Lastres for help with the cytometer, Fernando Escolar and Begoña Pou for help with the EM, and Carolina Gómez Moreira for critical reading of the manuscript.

The authors have no financial conflicts of interest.