|Home | About | Journals | Submit | Contact Us | Français|
Effective major histocompatibility complex-II (MHC-II) antigen presentation from phagocytosed particles requires phagosome-intrinsic toll-like receptor (TLR) signaling, but the molecular mechanisms underlying TLR delivery to phagosomes and how signaling regulates antigen presentation are incompletely understood. We show a requirement in dendritic cells (DCs) for adaptor protein-3 (AP-3) in efficient TLR recruitment to phagosomes and MHC-II presentation of antigens internalized by phagocytosis but not receptor-mediated endocytosis. DCs from AP-3-deficient pearl mice elicited impaired CD4+ T cell activation and Th1 effector function to particulate antigen in vitro and to recombinant Listeria monocytogenes infection in vivo. Whereas phagolysosome maturation and peptide:MHC-II complex assembly proceeded normally in pearl DCs, peptide:MHC-II export to the cell surface was impeded. This correlated with reduced TLR4 recruitment and proinflammatory signaling from phagosomes by particulate TLR ligands. We propose that AP-3-dependent TLR delivery from endosomes to phagosomes and subsequent signaling mobilize peptide:MHC-II export from intracellular stores.
Professional phagocytes ingest and process particulate matter for distinct purposes. During infection, most phagocytes primarily function to eliminate pathogens, but dendritic cells (DCs) balance pathogen destruction with antigen (Ag) preservation for presentation to Ag-specific T cells. This balance requires both selective recruitment of factors that modify the nascent phagosome environment (Savina and Amigorena, 2007) and selective activation of maturation programs via innate immune receptors such as Toll-like receptors (TLRs) (Blander and Medzhitov, 2007). While general features of phagosome maturation have been intensively studied (Flannagan et al., 2009; Kinchen and Ravichandran, 2008), the molecular mechanisms underlying selective factor recruitment to phagosomes in DCs and their integration with TLR signaling have been largely unexplored.
Several factors required for organelle maturation have been identified by studying genetic disorders of organelle biogenesis. For example, protein complexes required for the biogenesis of lysosome-related organelles (LROs) –cell type-specific organelles that derive from the endocytic pathway (Raposo et al., 2007) –are disrupted in patients with Hermansky-Pudlak syndrome (HPS), a group of genetic diseases characterized by oculocutaneous albinism, excessive bleeding and other variable symptoms (Huizing et al., 2008; Wei, 2006). The coat protein complex adaptor protein-3 (AP-3), for which the β3A subunit is genetically inactivated in HPS type 2 and its mouse model, pearl (Dell’Angelica et al., 1999; Feng et al., 1999), sorts integral membrane cargoes primarily from early endosomes towards lysosomes, LROs or related compartments (Dell’Angelica, 2009); hence, LROs such as melanosomes and platelet dense granules are malformed in AP-3-deficient patients and animals (Huizing et al., 2008; Wei, 2006). HPS type 2 patients also suffer from recurrent bacterial and viral infections. This immunodeficiency has been correlated to defective polarization of lytic granules in cytotoxic T cells (Clark et al., 2003) and natural killer cells (Fontana et al., 2006) and impaired CD1-dependent lipid Ag presentation (Briken et al., 2002; Sugita et al., 2002). Neutropenia is also observed in most patients (Huizing et al., 2002; Jung et al., 2006; Shotelersuk et al., 2000), but cannot solely account for recurrent bacterial infections that persist even after neutrophil counts are restored by granulocyte-colony stimulating factor treatment (Enders et al., 2006; Wenham et al., 2010). Nevertheless, the hemophagocytic lymphohistiocytosis observed in some patients implies a defect in phagocyte function and suggests that AP-3 might control other phagocyte processes.
AP-3 has recently been linked to TLR signaling in a DC subset. In plasmacytoid DCs (pDCs), AP-3 is required to stimulate type I interferon production by mobilizing activated TLR7 and TRL9 from endosomes to a specialized LRO (Blasius et al., 2010; Sasai et al., 2010). This pathway primarily influences antiviral immunity, but it raises the possibility that AP-3 might more generally regulate TLR trafficking in DCs – such as TLR4 delivery from endosomes to maturing phagosomes (Husebye et al., 2010).
Here, we tested whether phagosome maturation and consequent Ag presentation in DCs is altered in pearl mice, which lack functional AP-3 complexes, and in an unrelated mouse HPS model. We show that pearl DCs are impaired in both major histocompatibility complex (MHC) class II (MHC-II) presentation of phagocytosed Ag and proinflammatory TLR signaling from phagosomes. Our data support a requirement for AP-3 in phagosome-intrinsic TLR signaling and consequent delivery of peptide-loaded MHC-II to the plasma membrane, and suggest that recurrent bacterial infections in HPS type 2 patients might in part reflect a defect in TLR-dependent MHC-II presentation by DCs and consequent proinflammatory responses.
Fusion of nascent phagosomes in DCs with an “inhibitory LRO”, which harbors the gp91phox subunit of the phagocyte NADPH oxidase (NOX2), enables cross-presentation of phagocytosed Ag by MHC class I (MHC-I)(Jancic et al., 2007; Mantegazza et al., 2008; Savina et al., 2006). Because other LROs are malformed in HPS, we tested for impaired cross-presentation in DCs from the AP-3-deficient pearl mouse model of HPS type 2 and the pallid mouse model of HPS type 9, which lacks a separate protein complex called biogenesis of lysosome related organelle complex (BLOC)-1 (Cullinane et al., 2011; Wei, 2006). As shown previously (Sasai et al., 2010), wild-type C57BL/6 (WT), pearl and pallid bone marrow cells differentiated comparably to DCs in response to GM-CSF, as assessed by surface expression of CD11c, CD11b, MHC-I, MHC-II and CD86 and by soluble LPS-induced up-regulation of MHC-I, MHC-II and CD86 (Figure S3A, B). Likewise, WT and pearl mice had similarly composed splenic populations of CD11chiB220lo conventional DCs, their CD11chiCD8hi and CD11chiCD8lo DC subsets, and CD11cintB220hi pDCs (and of monocytes, macrophages, B cells and NK cells; Figure S3D-H). Thus, neither BLOC-1 nor AP-3 is required for DC differentiation in vitro or in vivo.
To test whether AP-3 or BLOC-1 is required to cross-present the model Ag ovalbumin (OVA), bone marrow-derived DCs (BMDCs) were pulsed with increasing doses of either OVA257-264 peptide, soluble OVA or particulate OVA conjugated to polystyrene beads (OVA-beads), and then cocultured with OVA257-264-specific, H-2Kb-restricted OT-I T cells. The dose-response curve to induce OT-I CD69 expression or IL-2 production was identical for WT, pearl and pallid BMDCs with each source of antigen (Figures 1, S1A-C). Similar results were obtained using splenic CD11chi DCs isolated from WT and pearl mice (Figure S2A–C). In contrast, cross-presentation of both soluble and particulate OVA by gp91phox-deficient (Cybb−/−) BMDCs was impaired (Figures 1C–F, S1) as expected (Savina et al., 2006). Thus, neither AP-3 nor BLOC-1 is required for Ag cross-presentation.
To test if AP-3 is required for MHC-II presentation, BMDCs or splenic DCs from WT, pearl or pallid mice were pulsed with increasing concentrations of OVA323-339 peptide or soluble or particulate OVA, and then cocultured with OVA323-339-specific, I-Ab-restricted OT-II T cells. The dose-response curves of OT-II CD69 expression or IL-2 production to OVA peptide or soluble OVA (Figures 2A–D; S1D, E; S2D, E) presented by WT, pearl and pallid DCs were identical, supporting reports (Caplan et al., 2000; Sevilla et al., 2001) that AP-3 was dispensable for MHC-II trafficking and showing that neither AP-3 nor BLOC-1 was required for processing and presentation of Ag internalized by receptor-mediated endocytosis. By contrast, OT-II activation was significantly impaired after coculture with pearl DCs pulsed with OVA-beads relative to similarly pulsed WT or pallid BMDCs (Figures 2E–G, S1F, S2F). This did not reflect a defect in phagocytosis, since WT, pearl and pallid BMDCs ingested fluorophore-conjugated beads to a similar extent (Figure S3C). Thus, whereas BLOC-1 is dispensable, AP-3 is required for effective MHC-II presentation of OVA only when captured by phagocytosis.
To determine whether the impaired T cell activation correlated with altered downstream OT-II T cell effector function, we evaluated production of the Th1 cytokine IFNγ and the Th2 cytokine IL-4 in response to Ag-pulsed WT or pearl BMDCs. The fraction of cytokine-producing OT-II cells was identical after coculture with peptide- or soluble OVA-pulsed WT or pearl BMDCs (Figure 3A–D). However, whereas WT and pearlBMDCs pulsed with OVA-beads elicited a similar IL-4 response, IFNγ production was significantly impaired in response to pearl BMDCs pulsed with OVA-beads at all concentrations (Figure 3E–G). Thus, following Ag uptake by phagocytosis, pearl BMDCs elicit a T cell response that is selectively defective in IFN-γ secretion.
To determine if AP-3 in conventional DCs is required in vivo for efficient T cell responses to phagocytosed bacteria, we exploited a recombinant OVA-secreting Listeria monocytogenes strain (Foulds et al., 2002) (referred to here as rLM-OVA) in the Δhly mutant, which lacks listeriolysin and is thus restricted to phagosomes. BMDCs from WT or pearl mice were equally susceptible to infection and ingested similar numbers of CFSE-labeled rLM-OVA (Figure S4A, B). Nevertheless, whereas pearl BMDCs that were either infected with rLM-OVA or exposed to soluble OVA stimulated MHC-I-restricted OT-I T cells as effectively as wild-type BMDCs (Figure S4C), rLM-OVA-infected pearl BMDCs induced less OT-II T cell activation and subsequent IFNγ production (Figure S4D–G). Thus, MHC-II presentation of phagocytosed Ag on both indigestible latex beads and bacteria is impaired in pearl DCs.
To test if Ag presentation by pearl DCs was impaired in vivo, we first assessed the ability of splenic DCs isolated from immunized WT or pearl mice to stimulate OT-I or OT-II cells in vitro. One day after immunization with OVA, splenic DCs from WT or pearl mice stimulated CD69 expression on OT-II cells equally well (Figure 4A, C), confirming that pearl DCs effectively processed and presented soluble Ag in vivo. By contrast, splenic DCs from pearl mice one or three days after infection with rLM-OVA stimulated OT-II cells less efficiently than WT DCs (Figure 4A–C), despite eliciting a similar OT-I cell response (Figure 4B, D; note the low background stimulation by DCs immunized with the parental Δhly LM mutant, referred to here as rLM). This confirms that MHC-II presentation of phagocytosed Ag in vivo is specifically impaired in pearl splenic DCs. To probe T cell activation during an infection, we examined the response of adoptively transferred OT-II T cells upon infection of WT or pearl mice with rLM-OVA or rLM as a negative control. OT-II cells harvested from spleens (Figure 4E, F) or lymph nodes (not shown) of WT mice were robustly activated (expressed CD69) 24h after infection with rLM-OVA or soluble OVA but not with rLM. By contrast, OT-II activation at this time point in pearl mice infected with rLM-OVA was dramatically impaired and barely above background. Moreover, whereas a substantial fraction of OT-II cells harvested from spleen or lymph node of WT mice 8d post-infection with rLM-OVA produced IFNγ and IL-4 upon restimulation in vitro, cells harvested from pearl mice produced only IL-4 at levels above background (Figure 4G, H). Thus, MHC-II presentation of phagocytosed antigen and Th1 polarization are severely impaired in pearl mice. Although the data do not exclude an influence of other AP-3-expressing cell types on IFN-γ secretion, the similarity between in vivo and in vitro T cell stimulatory responses suggests that this reflects reduced Ag presentation and/or proinflammatory cytokine production (see below) by pearl DCs.
To understand how AP-3 regulates MHC-II presentation of phagocytosed Ags, we quantified surface peptide:MHC-II complexes using Yae, an antibody that recognizes an Eα-derived peptide complexed with I-Ab (Rudensky et al., 1991). BMDCs from WT and pearl mice were pulsed with soluble or bead-conjugated Eα–GFP-His6 fusion proteins (Pape et al., 2007), and surface Eα:I-Ab complexes were detected by flow cytometry using Yae (Figure 5A–F). Following exposure to either Ag source, the percentage of WT and pearl BMDCs that expressed surface Eα:I-Ab complexes rose linearly over time. However, whereas an identical percentage of WT and pearl BMDCs expressed surface Eα:I-Ab complexes after exposure to soluble Eα fusion protein (Figure 5A, B), the percentage of surface Yae-positive pearl BMDCs was significantly reduced at all time points relative to WT BMDCs following exposure to particulate fusion protein (Figure 5E, F). These data were supported by analyses of OT-II restimulation using DCs that were fixed 0–6h after OVA-bead phagocytosis (Figure S5A and data not shown), and indicate that the reduced presentation of phagocytosed Ags by pearl DCs reflects reduced surface expression of peptide:MHC-II complexes.
To test whether the reduced presentation of phagocytosed Ag by pearl BMDCs reflected defective Ag processing, we analyzed phagosomal degradative capacity. The pH of phagosomes harboring fluor-conjugated polystyrene beads was initially higher in pearl than in WT BMDCs and remained slightly more alkaline over time (Figure S5C), even after LPS stimulation (not shown). However, a similar pattern was observed in pallid BMDCs (Figure S5C), indicating that the pH change alone did not impair Ag processing. Moreover, the rates of active cathepsins B/L and S recruitment to phagosomes (Figure S5D–F) and of OVA degradation on OVA-beads were identical in WT and pearl BMDCs (Figure S5G; whereas OVA in Tlr4−/− BMDC phagosomes was degraded more slowly, as expected) (Blander and Medzhitov, 2004). Based on these observations, pearl BMDCs efficiently degrade Ag. Consistently, intracellular EαI-Ab complexes formed effectively in pearl BMDCs after phagocytosis of particulate Eα fusion protein. However, they were retained intracellularly up to 18h later (Figure 5E). By immunofluorescence microscopy, the Eα I-Ab complexes accumulated in both WT and pearl BMDCs shortly after phagocytosis largely in punctate structures throughout the cytoplasm (Figure 5F). These punctate structures did not overlap with markers of lysosomes or early or late endosomes (not shown). However, whereas in WT cells the EαI-Ab complexes were translocated to the plasma membrane by 4–18h, they were retained within the punctate structures in pearl BMDCs (Figure 5F). This suggests that AP-3 is required to optimize peptide:MHC-II complex transport from phagosomes or post-phagosomal compartments to the plasma membrane.
Since AP-3 targets cargoes to late endosomal compartments and is dispensable for MHC-II presentation of non-phagocytosed Ag, it does not likely function directly in peptide:MHC-II transport to the cell surface. Because our OVA- and Eα-GFP bead preparations likely contain the TLR4 ligand LPS, and rLM stimulates other TLRs, we reasoned that TLR signaling – a known intrinsic requirement for phagosome maturation and Ag processing (Blander and Medzhitov, 2007) - might also be required for effective surface delivery of phagosome-derived peptide:MHC-II complexes. TLR4 is recruited to phagosomes from recycling endosomes (Husebye et al., 2010), from which AP-3 might function (Peden et al., 2004); we therefore tested whether TLR4 recruitment to and activation on phagosomes was dysregulated in pearl DCs. WT, pearl or pallid BMDCs were pulsed with OVA-beads and chased for various times, phagosomes were partially purified, and phagosomal TLR4 was quantified by flow cytometry analysis. Whereas no labeling was detected on phagosomes from Tlr4−/− BMDCs, TLR4 was progressively recruited to phagosomes from 0–4h in WT, pearl and pallid DCs; however, the recruitment to pearl phagosomes was specifically and significantly reduced by 30–40% (Figure 6A, B). This result was confirmed by immunoblotting using phagosomes that were magnetically isolated from homogenates of cells that had ingested OVA-coated magnetic beads (Figure 6C). Furthermore, a small pool of TLR4 was detected by immunofluorescence confocal microscopy surrounding a subset of phagosomes 3h after phagocytosis of OVA-beads in wild-type DCs, but less frequently in pearl DCs (Figure 6G). By contrast to TLR4, both the lysosomal membrane protein LAMP2 and the MHC-II molecule I-Ab were recruited to phagosomes with similar kinetics and efficiency in both WT and pearl BMDCs (Figure S6A–D). Thus, TLR4 undergoes AP-3-dependent recruitment from an intracellular pool to late stage phagosomes in DCs. AP-3 itself was detected on early phagosomes in wild-type DCs but was less abundant on late phagosomes (Figures 6E, F, and S6E).
Ligand-stimulated TLR4 recruits two different sets of signaling adaptor proteins, TIRAP/MyD88 and TRAM/TRIF (Kawai and Akira, 2011). TIRAP and MyD88 stimulate an early proinflammatory response via MAP kinases and NFkB and are thought to signal mainly from the plasma membrane (Kagan et al., 2008), although MyD88 is also recruited to E. coli-containing phagosomes (Husebye et al., 2010). TRAM and TRIF stimulate type-I interferon production via IRF3 and a late-phase proinflammatory response from intracellular sites. By flow cytometry and immunoblot analyses of phagosomes purified over time, both MyD88 and TRAM were progressively recruited to phagosomes in WT BMDCs (Figure 6D, H, I). While TRAM was recruited to WT and pearl phagosomes at similar levels, MyD88 recruitment was reduced by approximately 50% in pearl phagosomes (Figure 6D, H, I). Thus, optimal recruitment of the proinflammatory TLR adaptor to phagosomes requires AP-3.
We next tested whether the impaired TLR4 and MyD88 recruitment in pearl BMDCs affected signaling from phagosomes. BMDCs derived from WT, pallid and pearl mice were incubated with LPS-containing preparations of OVA or ligands for TLR2 (peptidoglycan), TLR3 (poly (I:C)), TLR4 (LPS), TLR5 (flagellin), TLR7/8 (R848), or TLR9 (CpG) either in soluble or particulate (coated on polystyrene beads) form, and proinflammatory signaling was measured by stimulation of IL-6 or IL-12- and IL23-p40 secretion into culture supernatants. As previously shown in pDCs and BMDCs (Sasai et al., 2010), soluble TLR ligands elicited secretion of comparable levels of IL-6 and p40 from WT, pallid and pearl BMDCs (Figures 7A, S7A). By contrast, the same ligands in particulate form elicited significantly (40–50%) reduced cytokine secretion from pearl BMDCs compared to their WT or pallid counterparts (Figures 7A, S7A). Consistently, by qPCR array analysis, IL6, IL12p35 and IL12p40 mRNAs were induced to similar levels in WT and pearl BMDCs in response to soluble LPS but induction was 50% reduced in pearl BMDCs in response to LPS-beads (Figure 7B). By contrast, both soluble and particulate LPS induced IFNβ1 mRNA to comparable levels in both WT and pearl BMDCs. These data are consistent with the reduced recruitment of MyD88 to phagosomes in pearl BMDCs and the unimpaired recruitment of TRAM (Figure 6D, H, I). Thus, whereas both AP-3 and BLOC-1 are dispensable for cytokine signaling from the plasma membrane or endosomes, AP-3 is specifically required for efficient proinflammatory TLR signaling from phagosomes.
To further probe AP-3-dependent TLR4 signaling from phagosomes, we examined downstream effectors of MyD88 and TRIF pathways. Phosphorylation of MAP kinase p38 is stimulated with different kinetics by MyD88 or TRIF signaling, and IRF3 phosphorylation is entirely TRIF dependent (Kawai and Akira, 2011). As expected based on the cytokine secretion profiles, p38 and IRF3 were both phosphorylated with similar kinetics in WT and pearl BMDCs upon stimulation with soluble LPS (Figure 7C left panel, D). IRF3 phosphorylation also followed similar kinetics after phagocytosis of LPS-beads in WT and pearl BMDCs (Figure 7C right panel, D; note that the kinetics differed in response to soluble LPS), consistent with the similar IFNβ1 mRNA induction in both cell types. By contrast, whereas both WT and pearl BMDCs responded to LPS beads with robust p38 phosphorylation, the phospho-p38 signal was sustained longer in WT than in pearl BMDCs (Figure 7C right panel), remaining two-fold over background at 2 and 3h after phagocytosis only in WT BMDCs (Figure 7D left panel). These data suggest that whereas initial MyD88-dependent TLR4 signaling does not require AP-3, a second phase of MyD88-dependent signaling from phagosomes requires AP-3 and is necessary for optimal induction of proinflammatory responses. We propose that this second phase of signaling also stimulates MHC-II presentation of phagocytosed Ag.
TLR signaling from phagosomes stimulates effective processing of phagocytosed Ag and ultimate presentation to CD4+ T cells (Blander and Medzhitov, 2007). Our data show that AP-3 plays a critical role in this process by facilitating the delivery of TLR4, likely other TLRs, and downstream signaling components to maturing phagosomes. We show that this recruitment is necessary for optimal proinflammatory signaling and correlates with export of peptide-loaded MHC-II complexes to the plasma membrane. Thus, in response specifically to phagocytosed antigens, AP-3-deficient pearl DCs are markedly impaired in Ag presentation and subsequent stimulation of effective Th1 responses in vitro and in vivo. Defective Ag presentation is not due to reduced Ag processing, but rather to peptide:MHC-II complex retention within intracellular stores. Our data reveal an important role of AP-3 in intracellular trafficking to and from phagosomes, a requirement for phagosomal TLR signaling in regulating both proinflammatory responses and mobilization of peptide:MHC-II complexes from phagosomes, and novel insights into the recurrent bacterial infections noted in AP-3-deficient HPS type 2 patients.
Nascent phagosomes in DCs fuse with a NOX2-bearing inhibitory LRO to promote alkalinization and consequent optimal Ag cross-presentation (Jancic et al., 2007; Mantegazza et al., 2008; Savina et al., 2006). We predicted that DC inhibitory LROs, like other LROs, would be malformed in mouse HPS models, consequently impairing cross-presentation. Surprisingly, cross-presentation was equally efficient in WT, pearl and pallid DCs regardless of the mode of Ag internalization. This might indicate that neither AP-3 nor BLOC-1 is required for inhibitory LRO formation. Alternatively, NOX2 might be missorted to the plasma membrane in the absence of AP-3 or BLOC-1, as observed for other LRO cargoes (Clark et al., 2003; Dell’Angelica et al., 1999; Di Pietro et al., 2006), and incorporated directly into newly formed phagosomes, circumventing the need for an LRO in cross-presentation. This might explain the increased pH of nascent phagosomes in pearl and pallid DCs.
Our results show that AP-3 plays a heretofore unappreciated role in MHC-II-dependent presentation of phagocytosed Ag in DCs. Relative to wild-type DCs, pearl DCs in vitro or in vivo elicited delayed and less potent stimulation of Ag-specific CD4+ T cells only when Ag was ingested by phagocytosis. The diminution in T cell priming reflected decreased surface expression of peptide:MHC-II complexes. Upon restimulation, CD4+ T cells primed by particulate Ag-pulsed pearl DCs were less apt to secrete IFNγ, consistent with a higher threshold of Ag stimulation for effective Th1 responses (Constant et al., 1995) and with the reduced TLR-stimulated IL-12 release that we show here. IFNγ responses by Ag-specific T cells in vivo were profoundly reduced in pearl mice in response to infection with recombinant L. monocytogenes, suggesting that the defects observed in vitro underestimate the functional deficiencies in anti-bacterial responses in vivo. A weakened early response by poorly activated DCs can blunt subsequent innate and adaptive responses as in Toxoplasma gondii infection (Hou et al., 2011), and thus likely in part explain the recurrent bacterial infections in AP-3-deficient HPS type 2 patients.
A requirement for AP-3 in optimal MHC-II Ag presentation was only apparent when Ag was internalized by phagocytosis. Pearl DCs were as effective as WT DCs at presenting soluble OVA or Eα, supporting earlier reports that AP-3 is dispensable for MHC-II trafficking and peptide loading (Caplan et al., 2000; Sevilla et al., 2001). Whereas a phagocytosis-specific Ag presentation defect is also observed in Atg5−/− DCs (Lee et al., 2010), the mechanism appears to be distinct; Atg5−/− DCs are defective in phagolysosome formation (Lee et al., 2010), whereas the normal kinetics of OVA degradation and phagosome recruitment of lysosomal contents in pearl DCs suggest that phagolysosome formation is unperturbed. Moreover, Eα:I-Ab peptide:MHC-II complexes formed with similar efficiency in pearl and WT DCs. Thus AP-3 is required not for Ag processing, but rather for efficient plasma membrane delivery of peptide:MHC-II complexes that derive from phagosomes. Surprisingly, these peptide:MHC-II complexes accumulate in punctate post-phagosomal compartments that resemble those observed in DCs infected by Leishmania major (Muraille et al., 2010) and do not harbor markers of early or late endosomes or lysosomes. We speculate that they represent intermediates in the redistribution of MHC-II into tubular structures that ultimately fuse with the cell surface (Boes et al., 2002; Chow et al., 2002).
How does AP-3 regulate this process? We suggest that defective surface delivery of peptide:MHC-II complexes in pearl DCs is secondary to impaired phagosomal TLR recruitment and phagosome-intrinsic TLR inflammatory signaling. The latter is a known requirement for Ag presentation (Blander and Medzhitov, 2006) and TLR signaling is required for MHC-II redistribution to the plasma membrane in DCs (Boes et al., 2002; Chow et al., 2002). Whereas soluble TLR ligands elicited WT proinflammatory responses from pearl DCs, particulate TLR ligands elicited diminished responses; the magnitude of the reduction in the proinflammatory response was similar to that of presentation of phagocytosed Ag. This implies that TLR-dependent signaling in DCs is more tightly regulated from phagosomes than from the plasma membrane or endosomes, consistent with the need to present foreign Ags and not self Ags on phagocytosed apoptotic cells (Blander and Medzhitov, 2007). The observed defects correlated with impaired recruitment of TLR4 and MyD88 from intracellular stores, as in Rab11-depleted macrophages (Husebye et al., 2010). While we focused on TLR4, phagosomal delivery of other TLRs such as TLR2 – previously documented in macrophages (Underhill et al., 1999) – likely also requires AP-3, since: (a) inflammatory signaling in pearl BMDCs by ligands of TLR2, 3, 5 and 7 was also impaired only when coated on beads; (b) presentation of OVA from rLM, which likely stimulates TLR2 and/or TLR5 but not TLR4 (Flo et al., 2000; Hayashi et al., 2001), was also impaired in pearl DCs; (c) MyD88 recruitment to phagosomes in pearl BMDCs is more strikingly reduced than TLR4; and (d) preliminary proteomics analysis showed fewer TLR2- and TLR7-derived peptides on phagosomes from pearl than WT BMDCs 2h after uptake of OVA-coated beads (unpublished results). The requirement for AP-3 in phagosomal inflammatory TLR signaling in conventional DCs is reminiscent of its requirement in pDCs for translocation of endosomal TLR7 and TLR9 to an LRO to promote IRF7- and TRAF3-dependent signaling (Blasius et al., 2010; Sasai et al., 2010). However, the mechanisms are clearly distinct since in conventional DCs AP-3 is not required for TLR4 stimulation of the TRAM-TRIF pathway toward type-I interferon production, but rather is necessary for optimal MyD88 recruitment to phagosomes, with subsequent sustained p38 phosphorylation and TLR-dependent proinflammatory cytokine production. Importantly, our data provide evidence that a second wave of MyD88-dependent signaling from DC phagosomes is necessary for optimal proinflammatory responses to phagocytosed particles, and likely for downstream MHC-II Ag presentation. Given that phagosomal degradative capacity and acquisition of lysosomal markers proceeded normally in pearl BMDCs but not in Tlr4−/− BMDCs (Blander and Medzhitov, 2004) (and our results), it is tempting to speculate that the first phase of MyD88 signaling from the plasma membrane, which is maintained in the absence of AP-3, would be sufficient to ensure phagosome maturation and MHC-II peptide loading.
How AP-3 regulates TLR4-MyD88 delivery to phagosomes is not clear, but it is unlikely to bind directly to the TLR4 cytoplasmic domain. Although potential consensus tyrosine-based AP-3-binding sorting signal sequences (Ohno et al., 1998) are present in most TLR cytoplasmic domains, they are in regions that are buried in the crystal structures of the TLR1 and TLR2 TIR domains and thus likely unavailable for binding (Xu et al., 2000); consistently, we could not detect an interaction between the TLR4 TIR domain and the μ chains of AP-3 by yeast two-hybrid assay (data not shown). AP-3 might directly engage a TLR-associated inflammatory adaptor when in complex with TLR4, a novel binding interface on TLR4 or an as yet unidentified trafficking adaptor. Alternatively, it might deliver a recruitment factor required for fusion of TLR4-containing transport intermediates from endosomes. Our data suggest that AP-3 might also play a secondary role in phagosome maturation beyond TLR recruitment, since AP-3 itself accumulates on early phagosomes and trails off later when TLR4, MHC-II and lysosomal markers appear. This suggests that AP-3 might function not only in cargo trafficking to phagosomes but also from phagosomes. One potential destination for this trafficking is the vesicular compartment in which peptide:MHC-II complexes accumulate and from which they appear to be exported. It is intriguing to speculate that AP-3 facilitates transport of peptide:MHC-II complexes, activated TLR signaling complexes, and/or other regulatory components from phagosomes to these compartments and/or the subsequent formation of plasma membrane-bound tubules from them. How AP-3-dependent inflammatory signaling from phagosomes supports these late events in Ag presentation provides a new venue for future research.
Sex- and age-matched mice between 6 and 12 weeks of age were used in all experiments. Mice were housed and bred under pathogen-free conditions in accordance with Institutional Animal Care and Use Committee-approved protocols at the University of Pennsylvania School of Veterinary Medicine Animal Facility or at the Children’s Hospital of Philadelphia. See Supplemental Experimental Procedures for details on mouse strains. BMDCs were cultured for 7–9d in granulocyte-macrophage colony stimulating factor-supplemented medium, splenic DCs were isolated by both negative and positive selection, and OT-I and OT-II T cells were isolated by positive selection as detailed in Supplemental Experimental Procedures. See Supplemental Experimental Procedures for antibodies and other reagents and additional procedure details.
DCs were exposed to OVA, OVA:BSA-coated 3μm latex beads (Polysciences, Warrington, PA), or OVA-specific MHC-I or MHC-II peptides for 15–30 min at 37 °C, then washed in PBS and chased in complete medium at 37 °C. DCs were then either fixed with 0.5% paraformaldehyde (PFA) in PBS for 10 min (for OT-I stimulation) or not (for OT-II), washed with PBS and cultured with 5–10×104 CD8+ OT-I T cells or CD4+ OT-II T cells. T cell activation was monitored 18h later as CD69 expression by flow cytometry (FACSCalibur, BD Biosciences, San Diego, CA) and IL-2 secretion in coculture supernatants by ELISA (BD Biosciences). At day 3 of coculture, IFNγ and IL-4 production was measured by intracellular flow cytometry after a 5h restimulation with PMA plus ionomycin or anti-CD3 and -CD28 antibodies with brefeldin A (Bougneres et al., 2009). For Ag presentation kinetics after OVA-bead phagocytosis, BMDCs were fixed with 0.5% PFA after different chase times and cocultured for 18h with prestimulated OT-II cells (Fitch et al., 2006). For presentation of Eα52-68 peptide on I-Ab, BMDCs were incubated with 0.5 mg soluble EαGFP or EαGFP-coated (5 mg/ml) beads for 30 min, washed with PBS and chased. Cells were then fixed with 3% PFA in PBS, stained with biotinylated YAe (clone eBioY-Ae, eBioscience, San Diego, CA) and APC-streptavidin (Invitrogen), and analyzed by flow cytometry. Intracellular Eα52-68:I-Ab complexes were detected by flow cytometry using biotinylated Yae after saturating the cell surface with unlabeled antibody (Lee et al., 2010). YAe labeling was quantified on CD11c+ cells that had taken up one bead as gated on a forward scatter vs. side scatter plot (Lee et al., 2010).
BMDCs were fixed with 3% formaldehyde in PBS, permeabilized with Permwash (BD), and labeled with primary antibodies and Alexafluor- or Dylight-conjugated secondary antibodies (Calvo et al., 1999). Cells were analyzed by fluorescence microscopy using a Zeiss LSM 710 confocal microscope and Zeiss ZEN 2010 software, or a DM IRBE microscope (Leica Microsystems, Wetzlar, Germany) and an Orca digital camera (Hamamatsu, Bridgewater, NJ) with OpenLab software (Improvision, Waltham, MA). For the latter, images obtained from consecutive z-planes were processed using subtractive volume deconvolution.
OVA-secreting or non-secreting Listeria monocytogenes Δhly2 strain (rLM-OVA, rLM) were grown as described (Foulds et al., 2002). WT and pearl mice (CD45.2+) were injected intravenously (IV) with 50,000 CFU of rLM-OVA or rLM or OVA in 200μl of PBS. In some mice, splenic DCs were isolated 24h or 3d later and cocultured with OT-I or OT-II cells for analysis of T cell CD69 induction. In other mice, 3×106 CD45.1+, CFSE-labeled OT-II cells were transferred IV 24h before immunization. Spleen and lymph nodes were harvested either 18h post-infection to measure CD69 expression on OT-II cells by flow cytometry, or 8d post-infection and T cells assayed for IL-4 and IFN-γ by intracellular flow cytometry after restimulation in vitro (Bougneres et al., 2009).
BMDCs were incubated for 15 min with OVA-coated latex beads or 3 μm magnetic beads (Dynabeads M-280 streptavidin, Invitrogen, NY) and then chased. Magnetic and non-magnetic phagosomes were purified after different chase times using a magnet or differential centrifugation, respectively, as described (Guermonprez et al., 2003; Mantegazza et al., 2008). Purified OVA-bead phagosomes were fixed, permeabilized, stained with antibodies to AP-3, TLR-4, MyD88, TRAM, I-Ab, LAMP2 or negative controls, and analyzed concurrently by flow cytometry, gating on the OVA-positive bead population (Savina et al., 2010). Protein extracts from purified magnetic phagosomes were analyzed by immunoblotting.
To generate TLR ligand-coated beads, TLR ligands were incubated overnight in PBS with 3μm latex beads and extensively washed with endotoxin-tested PBS (Invitrogen). BMDCs were incubated in complete medium supplemented as indicated with soluble TLR ligand or TLR ligand-coated beads for 3h as described (Sepulveda et al., 2009). Cytokine concentration in culture supernatants was measured by ELISA (ELISA Ready-SET-Go! Mouse interleukin-6, mouse interleukin-23 (p19-p40) and mouse interleukins-12 and -23 (total p40), eBioscience).
RNA was isolated using RNeasy (Qiagen, CA, USA) from WT and pearl BMDCs after 2h incubation with or without LPS (50 ng/ml) or LPS-coated beads. 1 μg RNA was reverse transcribed to cDNA using RT2 First Strand (Qiagen). Real-time PCR was performed in a 7300 ABI PCR System (Applied Biosystems, CA, USA) using RT2 SYBR Green/ROX PCR mastermix and the Mouse Antibacterial Response RT2 Profiler PCR array (96-well format, SABiosciences, Qiagen). Data were normalized to the average of five housekeeping genes. Relative levels of IL6, IL12p35, IL12p40 and IFNβ1 mRNA were calculated with the ΔΔCt method (Caino et al., 2011) and represented as mRNA fold-induction compared to unstimulated cells.
For TLR signaling, WT and pearl BMDCs were stimulated with LPS (50 ng/ml) or LPS-coated beads, chased as indicated, and lysed as described (Proietti et al., 2005). Proteins were fractionated by 10 or 12% SDS-PAGE and analyzed by immunoblotting using alkaline-phosphatase-conjugated secondary antibodies, enhanced chemiluminescence substrate (Roche), and phosphorimaging (Berson et al., 2001). Densitometric analysis on the bands was performed using NIH Image J software normalizing to total protein levels.
Statistical significance was determined by the unpaired Student’s t test and analysis of variance. *p < 0.05, **p < 0.01, and ***p < 0.001.
We thank R. Steinman, M. Pepper Pew, M. Jenkins, A. Rudensky, J. Liboon, S. Chatterjee and A. Fisher, S. Ross, S. Akira, P. Oliver, L. Eisenlohr, M. Chou and C. López for generous gifts of reagents, A. Stout, P. Zhang, S. Leach, Anand Sitaram and C. López Haber for technical assistance, and the Amigorena laboratory for experimental protocols. This work was supported by National Institutes of Health grants R01 EY015625 (to MSM), R21 AI092398 (to ARM and MSM), R21 AI079724 (to HS), R01 AI081884 and R01 AI064705 (to AI), a University Research Foundation grant from the University of Pennsylvania (to MSM), Donna and Richard Appel and the HPS Network (to SG) and a research grant from the United States Veterans Administration (to TML).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.