|Home | About | Journals | Submit | Contact Us | Français|
Dendritic cells (DCs) have the striking ability to cross-present exogenous antigens in association with MHC class I to CD8+ T cells. However, the intracellular pathways underlying cross-presentation remain ill-defined. Current models involve cytosolic proteolysis of antigens by the proteasome and TAP-dependent import into Endoplasmic Reticulum (ER) or phagosomal lumen. Here, we show that DCs express an ER-resident 47kDa immune-related GTPase, Irgm3. Irgm3 resides on ER and lipid body (LB) membranes where it binds the LB coat component ADRP. Genetic removal of either Irgm3 or ADRP leads to defects in LB formation in DCs and severely impairs cross-presentation of phagocytozed antigens to CD8+ but not antigen presentation to CD4+ T cells. We thus define a new role for LB organelles in regulating cross-presentation of exogenous antigens to CD8+ T lymphocytes in DCs.
DCs are highly specialized antigen presenting cells endowed with the ability to prime or tolerize antigen-specific naïve T cells (Steinman et al., 2003). CD8+ T cells recognize endogenous antigens (Ag) as short peptides associated to MHC I. Antigenic peptides arising from proteasomal degradation are translocated into the ER lumen by transporters associated with Ag processing (TAP1/2). Once in the ER, the peptides undergo amino terminal trimming by ERAP amino peptidases and are loaded on neo-synthesized MHC I molecules by a complex, which is composed of calreticulin, tapasin and ERp57 (Cresswell et al., 2005).
DCs can also present exogenous Ag in association with MHC I by a process called “cross-presentation” (Cresswell et al., 2005). Ag uptake by macropinocytosis (Norbury et al., 1997), receptor-mediated endocytosis (Bonifaz et al., 2002; Regnault et al., 1999) or phagocytosis (Kovacsovics-Bankowski and Rock, 1995) all initiate efficient cross-presentation.
Despite the importance of cross-presentation in tolerance and infectious immunity (Steinman et al., 2003; Steinman and Nussenzweig, 2002), the cellular pathways underlying this process are not fully understood. In most experimental systems, cross-presentation relies on cytosolic processing events since it can be blocked by pharmacological inhibition of the proteasome and requires functional TAP transporters (Kovacsovics-Bankowski and Rock, 1995). Indeed, there is considerable evidence that exogenous Ag can access the cytosol after phagocytosis (Kovacsovics-Bankowski and Rock, 1995), macropinocytosis (Norbury et al., 1997) or receptor-mediated endocytosis (Bonifaz et al., 2002; Rodriguez et al., 1999). The mechanisms underlying the delivery of Ag to the cytosol are ill-understood but may involve the ERAD pathway, which targets mis-folded, lumenal ER proteins to proteasomal degradation (Ackerman et al., 2006). Nevertheless, it remains unclear whether ERAD-mediated retro-translocation operates from the ER (after retrograde transport of endocytosed Ag) (Ackerman et al., 2005) or from the endocytic pathway (after recruitment of ER to endosomes and phagosomes) (Ackerman et al., 2003; Burgdorf et al., 2008; Guermonprez et al., 2003; Houde et al., 2003). In addition, a wealth of evidence supports the idea that cross-presentation is also favoured by the low proteolytic activity of the endocytic pathway in DCs (Hotta et al., 2006; Lennon-Dumenil et al., 2002; Savina et al., 2006). Finally, DEC-205 or mannose receptor-dependent retention of Ag in the early endocytic pathway favors cross-presentation (Bonifaz et al., 2002; Burgdorf et al., 2007; Burgdorf et al., 2008). However, the precise coordination of endocytic and cytosolic events leading to Ag proteolysis and MHC I loading during cross-presentation remains to be established.
Therefore, it is crucial that we begin to identify the molecular components of the transport pathways that control cross-presentation. GTPases are natural candidates, since they act as major regulators of intracellular traffic. Among those with dedicated immune function is a new family of 47kDa immunity-related GTPases (p47 GTPases) related to the dynamin superfamily and regulating vesicular trafficking (MacMicking, 2004; Shenoy et al., 2007). Recent genetic analysis has established that p47 GTPases exert pathogen-specific immunity to infectious agents occupying a vacuolar niche in both human and mouse settings (MacMicking, 2004; Shenoy et al., 2007). One p47 GTPase in particular, Irgm3 (IGTP), was found to reside extensively within the ER (Taylor et al., 1996; Taylor et al., 1997) and is critical for interferon (IFN)-dependent immunity to Toxoplasma gondii (Ling et al., 2006; Taylor et al., 2000). However, whether Irgm3 contributes to adaptive immunity remains unknown.
In this study, we show that Irgm3 controls cross-presentation in DCs. We find that Irgm3-deficient DCs exhibit a major impairment in their ability to cross-present phagocytic Ag to CD8+ but not to present Ag to CD4+ T cells. The role of Irgm3 in the cross-presentation pathway relies on intracellular events, since Irgm3-deficient DCs still phagocytose Ag particles and interact with T cells with the same efficiency as their WT counterparts. We demonstrate that Irgm3 localizes to ER membranes but also, unexpectedly, to lipid bodies (LBs). LBs (also called lipid droplets) are neutral lipid storage organelles composed of a central core of cholesteryl esters and triglycerides surrounded by a single layer of phospholipids which are thought to originate from ER membranes (Fujimoto et al., 2008). Irgm3 controls LB accumulation in DCs after activation both in vitro and in vivo. Complementation of Irgm3-deficient DCs by retroviral infection restores both LBs and efficient cross-presentation, which suggests that LBs have a direct effect on cross-presentation. In addition, we show that pharmacological inhibition of diacylglycerol acyltransferase decreases LB accumulation in DC and inhibits cross-presentation. ADRP belongs to the group of PAT-domain containing proteins that regulate LB biogenesis and dynamics (Brasaemle, 2007; Brasaemle et al., 1997; Chang et al., 2006; Fujimoto et al., 2008). We identified the LB coat protein ADRP as a partner of Irgm3: Irgm3 physically interacts and co-localizes with ADRP in LBs in DCs. This interaction is functionally relevant since accumulation of LB driven by Irgm3 requires ADRP. Finally, genetic inactivation of ADRP markedly inhibits both LB accumulation and cross-presentation of phagocytosed Ag by DCs.
Taken together, these results identify LBs as key organelles that regulate cross-presentation of phagocytosed Ag in DCs.
Previous studies have shown that the expression of Irgm3 is uniformly low in various cell types and further increased by IFN exposure (MacMicking et al., 2003; Taylor et al., 2000; Taylor et al., 1996). Thus we wanted to determine whether Irgm3 is expressed in DCs. Using RT-PCR and immunoblot (IB), we found constitutive expression of Irgm3 in monocyte derived DCs (mDCs, Figure 1A). IFN-γ induced a dramatic increase in Irgm3 expression in mDCs (Figure 1A). Significantly, the basal level of both Irgm3 mRNA and protein was 10 fold higher in untreated mDCs as compared to bone marrow-derived macrophages (Figure 1A, p<0.0001 and Figure 1B, p<0.005). Moreover, Irgm3 mRNA and protein was detected in both CD8+ and CD8− DCs purified from mouse spleen (Figure 1A and C). Finally, injection of the IFN-inducer polyI:C in mice led to an increase of Irgm3 protein (Figure 1C). In conclusion, Irgm3 is constitutively expressed by DCs and further increased upon activation.
We found that Irgm3 deficiency does not impair the differentiation of mDCs in GM-CSF dependent cultures, their cell surface phenotype or their ability to respond to LPS (Figure S1B).
To examine the contribution of Irgm3 in Ag presentation, mDCs were pulsed with ovalbumin (OVA) coated latex beads for 20 minutes, washed, chased for five hours and fixed before incubation with OVA-specific CD8+ OT-I transgenic T cells whose activation was monitored by CD69 upregulation. As shown in Figure 1D, Irgm3−/− mDCs were unable to efficiently cross-present OVA Ag after latex bead phagocytosis. Indeed, a 4-fold increase in OVA concentration on the beads was required to reach the same efficiency of cross-presentation as compared to control mDCs (Figure 1D). In line with the normal expression of MHC I and costimulatory molecules (Figure SD1), Irgm3−/− mDCs are not impaired in their ability to activate OT-I T cells after pulsing with OVA MHC I peptide (Figure 1E). Next we explored the possibility that Irgm3−/− mDCs were also defective for the MHC II Ag presentation pathway. We found that Irgm3−/− mDCs were equivalent as WT mDCs in their ability to activate OVA specific OT-II transgenic CD4+ T cell following OVA latex beads phagocytosis or OVA MHC II peptide loading (Figure 1F and G). These results indicate that Irgm3−/− mDCs have a selective defect for cross-presentation in vitro, while MHC II-restricted Ag presentation is unaffected.
To assess the contribution of Irgm3 to cross-presentation by lymphoid organs DCs in vivo, WT or Irgm3−/− mice were injected subcutaneously with OVA coupled to an anti-DEC205 (Bonifaz et al., 2002). As shown in Figure 1H, cross-presentation of targeted OVA to adoptively transferred OT-I T cells was slightly but significantly reduced in Irgm3−/− mice when compared with their WT counterparts (p<0.001). Ex vivo sorting of total lymphoid organ DCs loaded in vivo with anti-DEC-OVA conjugates (or high doses of soluble OVA) and in vitro co-culture with OT-I T cells further confirmed that Irgm3−/− deficient DCs display slightly reduced cross-presentation in situ (SD2, 3). The contribution of Irgm3 to normal cross-presentation in these systems; however, is smaller than it is in mDCs (see above). To examine the role of Irgm3 in phagocytic cross-presentation by activated DCs, we compared the ability of WT and Irgm3−/− mice to cross-prime endogenous CD8+ T cells after intravenous injection of apoptotic bodies loaded with OVA+polyI:C (Schulz et al., 2005). PolyI:C, a viral double-stranded RNA mimetic recognized by TLR-3, increases Irgm3 levels in splenic DCs (Figure 1C). As shown in Figure 1I, T cell cross-priming of endogenous CD8+ T cells was decreased in Irgm3−/− mice when compared with WT mice (p<0.005).
To assess the ability of Irgm3−/− mice to mount an endogenous CD8+ T cell response, we infected WT or Irgm3−/− mice with Listeria monocytogenes secreting OVA in the cytosol of infected cells. We found no significant difference in the activation of OVA-specific polyclonal CD8+ T cells in WT and Irgm3−/− mice (SD4). We conclude that the CD8+ T cell compartment is normal in Irgm3−/− mice, and that their DCs are competent to present OVA if large quantities of this Ag are delivered into the cytosol directly.
Taken together, these results identify Irgm3 as a positive regulator of cross-presentation in vitro and in vivo.
To begin dissecting the mechanism underlying the role of Irgm3 in cross-presentation, we analyzed phagosomal function in Irgm3−/− mice.
First we assessed the efficiency of Irgm3 deficient mDC in performing phagocytosis, using a stringent FACS-based assay quantifying only internalized beads (SD5). Control and Irgm3−/− deficient mDCs display the same efficiency of phagocytosis (Figure 2A). We conclude that Irgm3 is not involved in the phagocytosis of latex beads. Accordingly, Ag presentation by MHC II is not impaired in Irgm3 deficient mDCs (Figure 1F).
Since Irgm3 does not impact the efficiency of Ag uptake, we hypothesized that Irgm3 might modify phagosomal maturation.
Despite Irgm3 localization to ER (Taylor et al., 1997), we did not observe any difference in the transient association of calnexin to early phagosomes from control or Irgm3−/− mDCs (Ackerman et al., 2003; Gagnon et al., 2002; Guermonprez et al., 2003; Houde et al., 2003)(data not shown).
Next we tested whether Irgm3 controls the dynamics of phagolysosomal maturation. Control or Irgm3−/− mDCs were incubated with OVA-cross linked latex beads for 10 minutes, washed and their phagosomes were analyzed at various time points by FACS. We found that phagosomes from Irgm3−/− mDCs exhibit an accelerated acquisition of the phagolysosomal marker LAMP2, as well as an increased degradation of phagosomal OVA (Figure 2B and C). We therefore conclude that Irgm3 acts a negative regulator of phagosomal maturation.
To gain further insight into the intracellular mechanisms underlying Irgm3 involvement in cross-presentation, we first investigated the subcellular distribution of Irgm3 in IFN-γ treated mDCs by immunofluorescence (IF). As previously shown in other cell types (Taylor et al., 1997), a small percentage of Irgm3 colocalized with calnexin-positive ER, including perinuclear membrane (Figure 3A, Linescan 1). Much more pronounced staining; however, was noted on spherical cytosolic donut-shaped structures of large size (0.5–2µm), completely devoid of calnexin (Figure 3A, Linescan 2).
MHC I staining revealed a strong accumulation in the perinuclear area characteristic of Golgi apparatus, but did not colocalize with the Irgm3-bright vesicles (Figure 3B). Using mDCs from MHC II-GFP knockin mice (Boes et al., 2002); we did not find any colocalization between the Irgm3-bright vesicles and MHC II positive compartments (Figure 3C). Lastly, Irgm3-bright vesicles did not stain with the endo-lysosomal marker LAMP1 (Figure 3D) nor with endocytic compartments labeled with 10kD fluorescent dextran tracers that had been chased for 20 minutes and followed up to 240 minutes post-uptake (Figure 3E).
Thus Irgm3-bright vesicles are not ER, MHC I or II transport vesicles, nor intermediates of the endocytic network. Because of their morphology, we hypothesized that these vesicles were lipid bodies (LB) or lipid droplets. A striking feature of LBs is that they are comprised of a dense hydrophobic core of neutral lipids, surrounded by a hemi-membrane that do not support the insertion of transmembrane proteins (Fujimoto et al., 2008). The lipophilic dye Bodipy 493/503 (Bodipy) specifically stains the neutral lipids that form the core of the LB. Using this probe, we identified Irgm3-bright structures as LBs (Figure 4A). Usually, Irgm3 LBs are localized in the periphery of the cells and are sometimes close to or in direct contact with the perinuclear ER (Figure 4A). The cores of the LBs were stained with Bodipy while Irgm3 was concentrated at the periphery of these organelles (Figure 4B). Irgm3 localization to LBs increases with LBs numbers (SD6). Therefore, LBs represent a new intracellular localization for Irgm3, in addition to the previously described ER localization.
Since phagosomal maturation and phagocytic cross-presentation are dependent on Irgm3, we sought to determine whether Irgm3 could also be recruited to phagosomes. We did not detect any strong recruitment of Irgm3 in phagosomes from mDCs at any time points after phagocytosis (up to 240 minutes, Figure 4C). In contrast, we often found Irgm3/Bodipy positive LBs contacting phagosomes (20.9% at 60 min phagocytic pulse, Figure 4C, D). The percentage of phagosomal association decreased slightly after three hours of phagosomal maturation reaching 14.3% of the phagosomes (Figure 4D).
In summary, Irgm3 is mostly associated with LBs. This finding led us to investigate the possibility that Irgm3 controls LB accumulation in DCs.
To quantify LB accumulation under basal conditions, WT and Irgm3−/− mDCs were stained with Bodipy (Figure 5A). We found that the number of LBs per cell was significantly decreased in Irgm3−/− when compared to control mDCs (Figure 5A; median WT: 1 vs Irgm3−/−: 0; p<0.0001). Twice as many Irgm3−/− mDCs are devoid of any LBs as compared to controls (SD7).
Since IFN-γ increases Irgm3 expression (Figure 1), we tested the effect of IFN-γ treatment on LBs accumulation. We found a major increase in the median number of LBs in IFNγ-treated mDCs as compared to untreated WT mDCs (Figure 5A; median WT:1 vs WT+IFN-γ: 6). This increase is almost completely lost in the absence of Irgm3 (Figure 5A; median +IFN-γ WT: 6 vs Irgm3−/−:1; p<0.0001). Irgm3−/− mDCs were not able to accumulate more than 10 LBs per cell (SD7, middle panel). These findings demonstrate that Irgm3 is a major regulator of LB accumulation in IFN-γ–activated mDCs.
In order to determine whether LBs accumulate in the conditions of our cross-presentation assays, we counted LBs in mDCs exposed to OVA-coated latex beads. We found a significant increase in LBs in OVA-coated latex bead-exposed WT mDCs as compared to controls (Figure 5A, median WT:1 vs WT +OVA-beads: 5). OVA-beads dependent increase is almost completely absent in the case of Irgm3−/− mDCs (Figure 5A, median Irgm3−/−: 0 vs Irgm3−/− +OVA-beads: 1). In addition, Irgm3−/− mDCs exposed to beads accumulated only up to five LBs per cell, whereas many control mDCs harbored more than 15 LBs per cell (SD7, lower panel).
To obtain larger sample quantitation of LB accumulation, we analysed Bodipy-labeled DCs by FACS. This method measures the LB amount at the single cell level in a large number of cells. As shown in Figure 5B, the Bodipy signal was slightly higher in control mDCs when compared to Irgm3−/− mDCs. Significantly, the increase in LB accumulation in IFN-γ-induced or latex bead-pulsed WT mDCs was not observed in the absence of Irgm3 (Figure 5B).
In splenic DCs, we observed that CD8+ DCs have more LBs than their CD8− counterparts in WT mice (Figure 5C, 37% reduction, p<0.0001). This data is compatible with the idea that LBs may participate in cross-presentation, since CD8+ DCs are more efficient than their CD8− counterparts in cross-presentation of phagocytosed Ag (den Haan et al., 2000; Iyoda et al., 2002; Schnorrer et al., 2006). Irgm3−/− DCs exhibit a slight defect in LB accumulation both in CD8+ subset (p<0.05) and CD8− subset (p<0.005). Activation induced by in vivo injection of polyI:C induced efficient maturation both in control and Irgm3−/− mice (SD8), and also induced Irgm3 expression in control mice (Figure 1C). PolyI:C induced a net increase in the LB accumulation of CD8+ DCs (p<0.005) and, to a lesser extent, in the CD8− DCs (p<0.05)(Figure 5C). By contrast, Irgm3−/− DCs were unable to increase LB accumulation in response to polyI:C in both subsets. As a result, the difference between control and Irgm3−/− DCs increased after activation, leading to a 47% reduction in the CD8+ subset (p<0.05) and a 66% reduction in the CD8− subset (p<0.05) (Figure 5C).
Collectively, our results clearly demonstrate that Irgm3 plays a significant role in the establishment of the LB pool in DCs. Irgm3 induction promotes LB accumulation after DC stimulation (mDCs after IFN-γ, polyI:C in splenic DCs). Irgm3-dependent effects appear specific for immune stimulation since the induction of LBs by oleic acid feeding is not compromised in Irgm3−/− mDCs, which suggests that the basal LB producing machinery is not impaired in Irgm3−/− cells (SD10).
These results identify Irgm3 as an immune regulator of LB accumulation and suggest that LBs may facilitate cross-presentation.
The results presented so far show a role for Irgm3 in both cross-presentation and the control of LB accumulation in DCs. These findings led us to investigate the possibility that a connection exists between these two processes.
To distinguish whether Irgm3 impacts the cross-presentation by altering mDC development, we reintroduced Irgm3 in Irgm3−/−DCs. Retrovirally transduced Irgm3−/−mDCs expressed the protein which localized to LBs and increased LB levels (Figure 6A, B and C).
To test whether Irgm3 complementation would restore cross-presentation, we retrovirally transduced Irgm3−/− mDCs pulsed 30mn with OVA-coated beads. After six hours, cells were washed and incubated with OT-I T cells overnight. As shown in figure 6F, the cross-presentation defect of Irgm3−/− mDCs is partially corrected by the Irgm3 encoding virus, but not the control virus. We conclude that the effect of Irgm3 on cross-presentation is independent of any additional unknown effect that it might have on DC development.
To determine whether the LB accumulation in DC correlates with the efficiency of cross-presentation, we compared mDCs that were distinct in their LB accumulation but identical in Ag load. mDCs harboring one OVA-bead were FACS-sorted according to their “low” or “high” LB accumulation and co-cultured with OT-I T cell for 18 hrs (Figure 6E). We found far more efficient cross-presentation within the LB “high” population even though the two DC populations expressed similar levels of CD86 and activated OT-I T cells upon OVA MHC I peptide loading similarly (Figure 6F, SD9). We conclude that the total intracellular accumulation of LBs measured by Bodipy fluorescence correlates with cross-presentation by mDCs.
To further probe the requirement of LBs in cross-presentation, we designed a second approach aimed at reducing LB levels by inhibiting Acyl CoA:diacylglycerol acyltransferase (DGAT), which produces triglycerides stores. Xanthohumol has been identified as a DGAT inhibitor (Tabata et al., 1997). mDCs were fed with OVA-coated latex beads or pulsed with OVA MHC I peptide during 30mn, washed and then cultured for four hours with Xanthohumol at 85 µM, which resulted in a 30–50% inhibition of LB as measured by FACS (Figure 6G, SD11). Cells were then washed and fixed before overnight incubation with OT-I T cells. Xanthohumol inhibited cross-presentation of OVA-coated latex beads by 44% at the highest dose of beads and by 51% at the lowest dose of beads (p<0.001 in both cases) but left unaffected the presentation of the OVA peptide (Figure 6 H, I).
These experiments confirm that LBs are required for normal cross-presentation by mDCs in vitro.
The way in which Irgm3 exerts its effects on LB biogenesis remains unclear. However, recent yeast 2-hybrid analysis found that another p47 IRG, Irgd (IRG-47), interacts with the LB PAT protein ADRP (Yamaguchi et al., 2006). Loss of ADRP results in decreased LB formation in hepatocytes and macrophages (Brasaemle et al., 1997; Chang et al., 2006; Fujimoto et al., 2008; Paul et al., 2008). We found that ADRP is expressed in mDCs and colocalizes on the same LBs as Irgm3 (Figure 7A). To investigate the possibility that the two proteins interact directly, we performed pull-down assays using GST-Irgm3 as bait. ADRP was immunoprecipitated efficiently with GST-Irgm3 but not with GST-Irgm1, GST or glutathione beads alone (Figure 7B). Reciprocally, Myc-tagged ADRP immunoprecipitated Irgm3 and the positive control, Irgd, but not Irgm1 in mammalian cell extracts (Figure 7B). Notably, interaction between Irgm3 and ADRP did not require Irgm3 GTPase activity; a constitutive GDP-bound Irgm3 mutant (Irgm3S98N) still immunoprecipitated ADRP (SD12). These findings fit with earlier observations showing that Irgm3S98N remains associated to membranes, suggesting that GTPase activity is not required for steady-state localization (Taylor et al., 1997).
Having identified ADRP as a partner for Irgm3, we next asked whether genetic lesions in Adrp phenocopy those of Irgm3. In contrast to Irgm3−/−, Adrp−/− mDCs exhibited a strongly decreased level of LB even before activation (42 % of decrease in Bodipy MFI, p<0.005, Figure 6C) and a reduced induction of LBs upon oleate feeding (SD10). Since Irgm3 controls the up-regulation of LB accumulation upon IFN-γ exposure (Figure 5), we sought to analyze the LB accumulation of Adrp−/− cells after IFN-γ treatment. We found that IFN-γ treated Adrp−/− mDCs failed to upregulate LBs as their WT counterparts (Figure 6D, p<0.0001). This result was confirmed by LB counting in IF after Bodipy staining (Figure 7E): after IFN-γ treatment, the average number of LB was reduced from 6.4 LB/cell in WT versus 3.1/cell in Adrp−/− (p<0.0001). We conclude that LB accumulation after IFN-γ treatment is dependent on ADRP, which interacts with Irgm3.
Previous studies identified a greater than 10-fold enrichment of ADRP mRNA in CD8+ as compared to the CD8− spleen DC subset (Dudziak et al., 2007; Edwards et al., 2003). At the protein level, we found a two-fold enrichment of ADRP in the CD8+ DC subset when compared to the CD8− DC subset (IB, Figure 7F; p<0.05). Consistent with this observation, ADRP positive LBs were abundant in CD8+ DC; in addition, Irgm3 and ADRP colocalized on the same LBs (SD13).
We found that Adrp−/− splenic DCs are defective in LB accumulation, and even more so than Irgm3−/− DCs (Figure 7G). We conclude that ADRP is required for LB accumulation in splenic DCs.
To determine whether the loss of ADRP altered cross-presentation by splenic DCs, we compared OVA-bead cross-presentation assays with WT, Irgm3−/− and Adrp−/− splenic DCs. As shown in Figure 7H, OT-I T cells activation was decreased two-fold when they were activated by Adrp−/− or Irgm3−/− DCs, as compared to WT control. ADRP deficiency had no effect on cell surface phenotype, nor on phagocytic ability, nor on the ability to present OVA peptide to CD8+ T cells, nor on DC maturation (SD 14 and data not shown).
These experiments establish the convergence of Irgm3 and ADRP-containing LBs as a regulatory organelle for the cross-presentation pathway in DCs from lymphoid organs.
The present study identifies an IFN-inducible ER-resident GTPase, Irgm3, as a new component of the cross-presentation pathway. Irgm3-defective DCs fail to cross-present phagocytozed Ag efficiently, while interaction with T cells, phagocytosis and presentation to CD4+ T cells remain unaffected. Therefore, Irgm3 controls some cellular events required for efficient cross-presentation. In addition to its documented residence in the ER (Taylor et al., 1997), we show that Irgm3 associates with compartments distinct from the ER or endosome. Two characteristics define these vesicles as LBs: a) a dense core of neutral lipids stained by the LB-specific probe Bodipy, b) an outer envelope covered with the LB-specific coat protein ADRP. LBs are storage organelles for neutral lipids such as cholesteryl esters or triglycerides (Fujimoto et al., 2008). ADRP is sufficient to activate de novo formation of LBs in adipocytes as well as other cell types when over-expressed (Gao and Serrero, 1999). The removal of ADRP leads to hepatic and macrophage LB deficiency in vivo (Chang et al., 2006; Paul et al., 2008).
We found that Irgm3 controls the accumulation of LBs induced by cell activation following various stimuli (IFN-γ or polyI:C). This induction appears to involve ADRP that physically associates with Irgm3 during IFN-γ-induced upregulation of LBs. Irgm3 may participate in the neoformation of LBs by regulating the association of ADRP with the LB outer membrane. Alternatively, it may help stabilize a pre-existing interaction of ADRP with LBs since Irgm3 itself does not appear to bind neutral lipids directly (Tiwari et al., 2009). This latter interaction could prevent the recruitment of cytosolic lipases (Brasaemle, 2007).
We provide several lines of evidence that show that LBs are involved in cross presentation: First, activation of LB accumulation upon Irgm3 retroviral complementation increases cross-presentation in Irgm3−/− mDCs.
Second, depletion of LB by pharmacological inhibition of DGAT inhibits cross-presentation in WT mDCs.
Third, ADRP deficiency leads to decreased LB accumulation and decreased cross-presentation. ADRP is very specifically associated with LBs (Brasaemle et al., 1997). In this respect, the fact that both Irgm3 and ADRP are needed for cross-presentation—two proteins structurally and phylogenetically dissimilar to one another—is likely to be explained on the basis of their dual action at the level of LBs. This interpretation is further reinforced by these proteins almost exclusive co-localization to LBs and their ability to physically interact in vitro and in vivo.
In conclusion, three independent loss-of-function approaches that perturb LB homeostasis—two genetic, one pharmacologic—also disrupt cross-presentation in DCs. Moreover, complementation or gain-of-function of at least one of these components, Irgm3, overturns such defects. These data identify LBs as a positive regulator of the cross-presentation pathway in DCs.
How and where do LBs exert their effects on the cross-presentation pathway?
Irgm3 and Irga6, are both ER-associated p47 GTPases that have been proposed to disrupt Toxoplasma gondii parasitophorous vacuoles by triggering membrane vesiculation and fission (Ling et al., 2006; Martens et al., 2005). We did not observe any strong association to phagosomes, or any membranolytic activity of recombinant Irgm3 on synthetic membranes (SD15). This finding does not support the idea that Irgm3 promotes cross-presentation through a direct destabilization of phagosomal membranes enabling the delivery of phagosomal Ag to the cytosol.
It has been proposed that the process of LB neoformation may also promote the formation of transient lipidic pores at the level of ER membranes (Ploegh, 2007). This could be in relation to the established function of LBs as cytosolic sites for ubiquitinylation of ER-derived proteins (Fujimoto and Ohsaki, 2006). Further studies are required to determine whether these processes underlie the effect of Irgm3 and ADRP on cross-presentation of phagocytozed Ag.
Here we show that Irgm3 negatively effects phagosomal maturation both in terms of late marker acquisition (LAMP2) and proteolytic activity (OVA degradation). Thus, besides other mechanisms, the action of Irgm3 may participate in the establishment of endocytic compartment endowed with a low proteolytic activity favouring crosspresentation (Delamarre et al., 2005; Hotta et al., 2006; Lennon-Dumenil et al., 2002; Savina et al., 2006; Trombetta et al., 2003). The phagosome-LB contact sites that we have identified may support a regulatory function of LBs on phagolysosomal progression. Importantly, the negative effect of Irgm3 on phagolysosomal progression is in line with its lack of involvement in anti-mycobacterial innate immunity, which relies on the activation of phagolysosomal function (MacMicking et al., 2003).
LBs have already been described in DCs, including in the first description of these cells (Maroof et al., 2005; Steinman and Cohn, 1973). Our results have uncovered a completely new aspect of DC biology, related to lipid metabolism and its implication for Ag cross-presentation by MHC I. Further studies are required to identify the molecular pathways controlling the involvement of LBs in cross-presentation in DCs.
See supplementary methods for details.
Irgm3-deficient mice (Taylor et al., 2000), Adrp-deficient mice (Chang et al., 2006), OVA-specific CD8+ (OT-I) an CD4+ (OT-II) TCR transgenic mice, MHC II-GFP knockin mice (Boes et al., 2002) have been described previously.
Bone marrow-derived GM-CSF DCs (day 7–8) were used and termed as mDCs. For some experiments, Irgm3 expression was induced by 3–5 hours treatment with 250ng/ml IFN-γ. Spleen DCs (SDCs) were isolated from mouse spleens after collagenase digestion and purified using CD11c magnetic beads (Miltenyi) and further purified by FACS sorting. In vitro cross-presentation assays were adapted from procedures previously described (Guermonprez et al., 2003) and used CD69 up-regulation in OT-I T cells as a readout. For some experiments, mDCs were infected with pMx retrovirus harbouring Irgm3 cDNA (or control empty retrovirus) in the presence of polybrene (10µg/ml). DEC-OVA conjugates were prepared as previously described (Bonifaz et al., 2002). For in vivo cross-priming assay 30.107 splenic cells from β2m-deficient mice were loaded endocytically by pulsing with 20mg/ml OVA plus 2mg/ml polyI:C for 15mn prior γ irradiation and iv injection. Eight days later, spleens were harvested and IFN-γ production by endogenous CD8+ T cells was measured after peptide specific restimulation.
For flow cytometric analysis of phagosomes, 25.106 DCs were pulsed for 10 min at 37°C at a ratio of 1:3 with 3µm latex beads coupled covalently to OVA by glutaraldehyde cross linking. Then, cells were washed three times on a serum gradient at 150 G (4°C) before chase at 37°C. After hypoosmotic lysis and fixation, phagosomes were stained with anti-OVA or anti-LAMP2 antibodies and analyzed by FACS (Guermonprez et al., 2003).
After cell surface staining, cells were fixed with PFA 1%, and stained for LBs in PermWash solution (Becton Dickinson) with BOPIDY FL (Molecular probes) at 4µg/ml for FACS analysis and 10µg/ml for immunofluorescences.
Recombinant p47 GTPases expressing GST tAg were immobilized on gluthathione sepharose and incubated with HEK lysates of transiently expressed ADRP-myc for coimmunoprecipitation. In the reverse experiments, lysates of HEK cells transfected with Myc-ADRP and various EYFP-tagged p47 GTPases were immunoprecipitated with anti-Myc resin.
The authors thank G. Taylor, H. Shen, Z. Maciorowski, C. Guerin, I. Grandjean, A. Boissonnas, L. Saveanu, A. Savina, W. Faigle, D. Loew, B. Lombard, CL. Trubert, S. Davis for reagents, helping discussions or critical reading of the manuscript. LB was supported by ARC, Curie Institute and INSERM. LC was supported by the USA NIH Grant HL 51586. JDM was supported by NIH NIAID (R01 AI068041-01A1), BWF Award, Searle Foundation (05-F-114), CRI and W.W. Winchester Foundation. PG was supported by the CNRS, ANR and the Rockefeller University.
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.
The authors have no conflicting financial interests.