Antigen-presenting cells (APCs) are expected to present peptides from endocytosed proteins via major histocompatibility complex (MHC) class II (MHCII) molecules to T cells. However, a large proportion of peptides purified from MHCII molecules are derived from cytosolic self-proteins making the pathway of cytosolic peptide loading onto MHCII of critical relevance in the regulation of immune self-tolerance. We show that peptides derived from cytoplasmic proteins either introduced or expressed in the cytoplasm are first detectable as MHCII-peptide complexes in LAMP-1+ lysosomes, prior to their delivery to the cell surface. These peptide-MHC complexes are formed in a variety of APCs, including peritoneal macrophages, dendritic cells, and B cells, and are able to activate T cells. This process requires invariant chain (Ii)-dependent sorting of MHCII to the lysosome and the activity of the molecular chaperone H-2M. This pathway is independent of the ER resident peptide transporter complex TAP and does not take place by cross-presentation from neighbouring cells. In conjunction with our earlier results showing that these peptides are derived by cytosolic processing via the proteasome, these observations provide evidence for a ubiquitous route for peptide transport into the lysosome for the efficient presentation of endogenous and cytoplasmic proteins to CD4 T cells.
The location of antigens in distinct intracellular compartments of antigen-presenting cells (APCs) influences their proteolytic processing as well as access to major histocompatibility complex (MHC) molecules and presentation to T cells. MHCI and MHCII molecules are directed to different intracellular compartments for sampling peptide antigens (Neefjes et al., 1990; Peters et al., 1991). It is probable that pathways by which MHC molecules acquire peptides will have a significant impact on generation of peptide diversity that will be ultimately sensed by the T cell receptor. Thus, understanding the pathways involved in peptide generation and loading is an extremely important issue from an immunological perspective. This is especially relevant for the dissection of the mechanism of generation of an immune response against foreign proteins and development of tolerance against self proteins.
Peptides generated in the cytosol by proteasomal degradation are generally thought to be presented on MHCI. These peptides are first transported across the endoplasmic reticulum (ER) membrane by the transporter associated with presentation (TAP), for loading on to MHCI in the ER lumen (Rock and Goldberg, 1999; York and Rock, 1996). Peptide-MHCI interaction is expected to take place during MHCI assembly, and is required for MHCI export out of the ER (Cresswell et al., 1999). In contrast, the generation and loading of peptides on MHCII molecules which present to the other class of T cells, namely the CD4 T cells, is believed to take place in endosomes and involves peptides derived from proteins and other antigens delivered from the outside via endocytic mechanisms. The apparent dichotomy between the sources of peptides for loading onto MHCI and MHCII is reinforced because newly synthesized MHCII molecules are protected from peptide loading in the ER and Golgi by the invariant chain (Ii) (Cresswell, 1994; Cresswell, 1996) which also sorts associated MHCII molecules from the Golgi to endolysosomal compartments (Bakke and Dobberstein, 1990).
Immunocytochemistry, subcellular fractionation and immunoelectron microscopy approaches have identified MIIC or CIIVs as the canonical loading compartments for newly synthesized MHC class II (Neefjes, 1999; Peters et al., 1991). These are sites where MHCII molecules accumulate, and it is often argued that MIICs (or CIIVs) are a collection of late endocytic compartments that contain the necessary proteins for efficient peptide loading of MHCII molecules (Geuze, 1998; Neefjes, 1999). It is in the acidic milieu of these endosomal compartments that Ii-peptide exchange is facilitated by the endosomally localized peptide exchange catalyst H-2M in mice or HLA-DM in humans (Martin et al., 1996; Miyazaki et al., 1996). Peptide-MHCII complexes may also be generated when cell surface MHCII molecules recycle to early endosomes (Pathak and Blum, 2000; Pinet and Long, 1998).
A breakdown of the MHCI and MHCII peptide-compartment sampling dichotomy (MHCI, intracellular versus MHCII, extracellular) is apparent from the finding that a large proportion of peptides extracted from purified MHCII appear to be from endogenous sources (Chicz et al., 1993; Rudensky et al., 1991a), and many of them are derived from self-cytosolic proteins (Dongre et al., 2001). This suggests that this mode of sampling pathway is likely to be important in establishing self-tolerance in CD4 T cells, a breakdown in which can lead to autoimmune disease.
We have previously shown that cytosolically delivered proteins as well as an endogenous trans-membrane protein, IEα, are efficiently presented in a MHCII-restricted manner to CD4 T cells (Mukherjee et al., 2001); other examples of this type of presentation have also been reported (Bonifaz et al., 1999; Lich et al., 2000). The exact nature of the peptide-MHCII complexes differ from those generated via endosomal processing routes in their ability to be presented to specific subsets of T-cell clones (Barlow et al., 1998). Data from our laboratories has showed that intracellular proteins are likely to be processed cytosolically by the proteasome for the generation of peptides to be presented on MHCII (Mukherjee et al., 2001).
A specific peptide-MHCII complex, Eα52-68-I-Ab, derived from a peptide fragment of amino acids 52-68 of transmembrane protein I-Eα (Eα52-68), is detectable in subcellular fractions containing MIIC-like compartments of a B cell line (Rudensky et al., 1994). These compartments, which also contain HLA-DM and lysosome-associated-membrane protein 1 (LAMP-1), are multivesicular and multilamellar. Disruption of endosomal acidification using lysosomotropic agents such as chloroquine, or Golgi transport via monensin, blocks the presentation of cytosolic, secreted or ER resident forms of ovalbumin (OVA) and HEL proteins transiently expressed in APCs (Bonifaz et al., 1999; Malnati et al., 1992; Michalek et al., 1992). Together, these data indicate a role for endosomes and biosynthetic transport in the formation of peptide-MHCII complexes derived from endogenous proteins, similar to the requirement for exogenous proteins. However, a disruption of lysosomal pH affects several lysosomal functions such as lysosomal proteases, peptide loading on MHCII and intracellular transport (Clague et al., 1994; Mellman et al., 1986), making the effects of such treatments on antigen presentation difficult to interpret. Furthermore, intracellular organelles where peptides from endogenous cytosolic proteins are loaded on MHCII remain unidentified.
We have examined the pathway of MHCII-restricted presentation of cytosolically introduced protein, constitutive endogenous trans-membrane protein and cytoplasmically expressed protein. All these proteins contain the peptide sequence Eα52-68. We have localized the subcellular compartment where the peptide-MHC complex (Eα52-68-I-Ab) forms in a variety of APCs, including bone marrow derived dendritic cells (BMDCs), macrophages or peritoneal exudate cells (PECs) and a macrophage cell line, BMC-2. Our data are consistent with MHCII-restricted presentation of cytosolic and endogenous proteins originating with peptide transport directly into late endosomal compartments colocalized with LAMP-1 and cathepsin-D, for loading onto MHCII molecules.
Materials and Methods
The mouse strains and genotypes used - C3H/HeJ (H-2k), C57BL/6 (H-2b), B10.A(5R) (H-2i5), H-2Mα-/-, Ii-/- (from the Jackson Laboratory, Bar Harbor, USA) and Tap1-/- (gift from Dr L. van Kaer, Vanderbilt University, Nashville, USA) - were bred in the small animal facility of the National Institute of Immunology (New Delhi, India) and the animal facility of the National Centre for Biological Sciences (Bangalore, India), and used for experiments when 6-12 weeks old. All experiments were done with the approval of the Institutional Animal Ethics Committees.
A recombinant fusion protein (GST-Eα52-68-myc) consisting of GST, aa residues 52-68 of the mouse I-Eα sequence (Eα52-68) and the oligopeptide sequence of the c-myc protein recognized by the 9E10 monoclonal antibody was made in Escherichia coli from the plasmid pGEX-2T with IPTG-inducible expression and was affinity purified over glutathione-Sepharose beads (data not shown). Ovalbumin (Sigma, St Louis, USA) was maleylated with maleic anhydride (Sigma) at alkaline pH as previously described (Mukherjee et al., 2001), dialysed against PBS and the degree of maleylation was estimated as the loss of free ϵ-amino groups measured by trinitrobenzenesulfonic acid. Maleylated proteins were used only when maleylation was >90%. All proteins were dialysed against PBS to remove small degradation products and this was additionally confirmed for the lack of presentation to T cells by fixed APCs (data not shown) before use. The Eα52-68-GFP recombinant DNA was made by synthesizing an oligonucleotide containing a start codon and sequence encoding for amino acids (aa) 52-68 of the I-Eα sequence, with flanking BglII and BamHI restriction endonuclease sites. This was cloned in frame with eGFP in eGFP-N3 vector (Clontech-BD Biosciences, Palo Alto, CA, USA), and confirmed by sequencing.
Cell lines and primary cell cultures
The MHCI (H-2Kb)-restricted OVA-specific T cell transfectant line B3 was a gift from Drs K. Hogquist and M. Bevan (University of Washington, Seattle, USA). The MHCII (I-Ab)-restricted OVA-specific T cell hybridoma 13.8 was generated from OVA-immunized C57BL/6 mice and characterized (data not shown). The I-Ab-restricted T cell hybridoma specific for an I-Eα peptide (aa 52-68; Eα52-68), 1H3.1, and the I-Ab-expressing macrophage cell line, BMC-2 were gifts from Dr C. A. Janeway, Yale University, New Haven, USA. All macrophage cells unless otherwise mentioned were plastic-adherent peritoneal resident macrophages or PECs. Non-adherent mouse bone marrow cells were cultured with either monocyte colony stimulating factor (M-CSF; 30% L929 fibroblast-conditioned medium as M-CSF source) or recombinant granulocyte-monocyte colony stimulating factor (GM-CSF; Pharmingen, San Diego, CA, USA), respectively, for 9 days, with growth factor replenishment at 3, 5 and 7 days. Tightly adherent cells were excluded from the GM-CSF-containing cultures on day 7. Live cells growing out on day 9 were used as a source of bone marrow-derived DCs (DC).
Monoclonal antibodies; Y-Ae (anti-Eα52-68-I-Ab), Y3P (anti-I-Ab) were detected using a goat anti-mouse IgG-Fc-specific polyclonal antibody (Jackson ImmunoResearch, USA) conjugated with Alexa-488 or Alexa 568 fluorophores (Molecular Probes). Polyclonal antibodies; rabbit anti-TGN38 (gift from Dr K. Stanley (Heart Research Institute, Sydney, Australia), rabbit anti-ERp57 (Dr U. Tatu, IISc Bangalore, India), rabbit anti-mannose-6-phospate receptor (MPR-300, from Dr Peter Schu, Georg-August-University, Goettingen, Germany) were detected with a polyclonal donkey anti-rabbit IgG (Jackson) conjugated with Alexa-568 flurophore. Rat anti-mouse LAMP-1 (BD-Pharmingen, USA) was used as a primary antibody labelled with Alexa-568 fluorophore.
Transfections and cytoplasmic loadings
Cells were transfected using Fugene 6™ (Roche, Indianapolis, IN) according to the manufacturer's protocol, with 6-10 μg of plasmid DNA for 2×106 cells and stained for Y-Ae antibodies. Apart from indicated cell lines, primary macrophages and DCs grown from mouse bone marrow were also transfected using similar protocols.
To assess for the involvement of inhibitors of PI 3-kinase in the generation of peptide-MHC complexes, BMC-2 cells were transfected with Eα52-68-GFP construct and 2 hours post-transfection were treated with titrating doses of PI 3-kinase inhibitor, LY294002 (1 μM-50 μM), or autophagy inhibitor, 3-methyladenine (1 mM-10 mM). 20 hours post-transfection, cells were stained with Y-Ae to determine surface levels of Eα52-68-IAb complexes.
Proteins were delivered into the cytosol using osmotic lysis of pinosomes as described previously (Mukherjee et al., 2001). Briefly, APCs were incubated in hypertonic serum-free Dulbecco's modified Eagle's medium (DMEM) with 0.5 M sucrose, 10% polyethylene glycol 800, and 10 mM Hepes containing antigenic protein for 10 minutes at 37°C, followed by washing and incubation in isotonic serum-free DMEM for 5 minutes. Cells were again washed and chased further in serum containing DMEM before being fixed with paraformaldehyde. APCs were chased for 15 minutes (after the 10 minute pulse), for detection of intracellular peptide-MHC complexes and chased for 1 hour when used to stimulate T cells. Post-fixation, APCs were washed, counted and used in T cell stimulation assays. Exogenous pulses of protein were done similarly, except in isotonic DMEM, thus, resembling a fluid phase uptake protocol.
Proteins were also delivered into the cytosol using the Chariot™ transfection reagent according to the manufacturer's instructions. Briefly, 1 mg of GST-Eα52-68-myc was incubated with 10 μl of Chariot™ reagent for 30 minutes to allow complex generation. This mixture was then overlayered on 1-2×106 cells that were kept on ice. The cells were incubated on ice for a further 15 minutes before the complexes were washed off with PBS and the cells were shifted to 37°C. Loading of streptavidin conjugated to FITC (SA-FITC) was done in the same way except that the concentration was 50 μg/ml.
Imaging and image processing
PECs were adhered on glass coverslip bottom dishes, immediately after removing from the animal. After antigen pulsing, cells were fixed with 2% paraformaldehyde (15 minutes), permeabilized with 0.1% saponin (20 minutes) and stained with antibodies. Confocal imaging was carried out on a Bio-Rad MRC-1024 confocal microscope (Bio-Rad Microsciences, UK) equipped with factory set dichroics and a Krypton-Argon laser. Images were processed using MetaMorph software (Universal Imaging, PA, USA) and corrected for cross-talk only in the images in Fig. 4 and Fig. 7 (E and F). For this purpose single Alexa-488 labelled cells were scanned at the same settings of laser power and gain as the sample to be imaged, and images of green fluorescence and spillover in the red channel were collected; under these conditions there was no spillover fluorescence of the fluorophore imaged in the red channel into the green. The percentage/fraction spillover was calculated by dividing the pixel value for the red image by that for the green and multiplying by 100. This was of the order 3-10%. This fraction of the signal in the green channel was subtracted from the image in the red-channel and the corrected images were used for further analyses. Cross-talk correction was not required for the other images since under the imaging conditions used there was no detectable cross-talk. Images were acquired on a Nikon TE 300 inverted microscope, 60×, 1.4 NA objective lens, using the Bio-Rad acquisition software as described in detail elsewhere (Sabharanjak et al., 2002). Wide-field fluorescence microscopy was done using similar optics and images collected via a charge-coupled device camera (Princeton Instruments, Princeton, NJ, USA) using MetaMorph acquisition software (Universal Imaging, West Chester, PA, USA). All images were finally processed in Adobe Photoshop software for output.
BMC-2 cells were transfected with eGFP or Eα52-68-GFP using Fugene6 according to manufacturer's instructions, and were stained with Y-Ae antibody on ice and analysed on a BD-LSR flow cytometer (Becton Dickinson).
Antigen presentation assays
Macrophages in suspension were either exogenously pulsed or cytosolically loaded with various antigens. They were allowed to process antigens at 37°C for 1 hour and then fixed with 1% paraformaldehyde (Sigma) for 1 minute, washed and used as APCs in T cell stimulation assays. To assess bystander presentation, H-2k macrophage APCs (Ag+) were first pulsed with antigen as above. After rigorous washing, these APCs were subsequently incubated with an equal number of unpulsed H-2b APCs (-Ag) prior to fixation and subsequent use as mixed APCs in T cell stimulation assays.
T cell lines (1-10×104 cells/well) were stimulated with titrated concentrations of antigen-pulsed APCs in triplicate cultures in 200 μl of DMEM with 10% FCS, antibiotics, L-glutamine and 0.5 mM 2-ME in 96-well flat-bottom plates (Falcon, Franklin Lakes, USA) as indicated. Culture supernatants were collected 24-36 hours later and used to estimate the IL-2 induced by stimulating the IL-2-dependent cell line CTLL-2 (1×104 cells/well) with them, incubating for 24 hours, and pulsing the plates with 0.5 μCi per well of [3H]thymidine (NEN Life Science, Boston, USA) for 12-16 hours to measure the proliferative responses. Plates were harvested onto glass-fibre filters for scintillation counting (Betaplate; Wallac, Turku, Finland). Data are shown as proliferation (mean±s.e.m.) observed in triplicate CTLL-2 cultures. IL-2 secreted in the culture supernatant was also estimated by an IL-2-specific cytokine ELISA (R&D Systems, Minneapolis, USA), and the data are plotted as Absorbance Units at 450 nm or 490 nm. Alternately, stimulation of the T cell line 13.8, which carries a transfected construct capable of expressing β-galactosidase under the consensus IL-2-promoter, was also assayed by detecting the induction of expression of intracellular β-galactosidase in them by an enzyme detection assay on cell lysates, data for which are plotted as absorbance units at 570 nm.
Cytoplasmically expressed proteins are capable of MHCII-restricted presentation of the same epitope as endogenously expressed proteins or proteins loaded via osmotic lysis of pinosomes
Previously (Mukherjee et al., 2001), we detected peptide-MHC complexes derived from proteins introduced into the cytoplasm of diverse APC via osmotic lysis of pinosomes using peptide-MHCII-specific T cell lines, 1H3.1 [for the Eα52-68-I-Ab complex (Murphy et al., 1992)] and 13.8 [a MHCII (I-Ab)-restricted OVA-specific T cell hybridoma]. In the work presented here we have used a monoclonal antibody, Y-Ae, directed against a specific MHC-peptide complex, Eα52-68-I-Ab complex (Rudensky et al., 1991b), to study the formation of MHC-peptide complexes in intracellular compartments. Both Y-Ae and 1H3.1 T cells are specific for the Eα52-68-I-Ab complex, a major self peptide-MHCII complex in cells expressing I-E and I-Ab (Rudensky et al., 1991a). In addition, since we had used osmotic lysis of pinosome protocol to deliver proteins into the cytoplasm, we wished to rule out the possibility that these peptide-MHC complexes detected thus far are from locations other than the cytosol. For example, these peptide-MHCII complexes could arise by incomplete release of proteins into the cytoplasm (during osmotic lysis of pinosomes) or by endosomal degradation of the transmembrane proteins (in cells expressing I-Ab and I-Eα). For this purpose we expressed Eα52-68eGFP, a fusion protein containing the Eα52-68 fragment of IEα in the cytoplasm of I-Ab expressing APCs, the BMC-2 cell line. Y-Ae staining was detected by flow cytometry specifically in cells that expressed Eα52-68-eGFP but not in cells expressing only eGFP (Fig. 1A). This is similar to the situation in which the Y-Ae+ complexes are detected on primary macrophage cells pulsed either cytosolically by osmotic lysis of pinosomes or exogenously via the fluid phase with GST-Eα52-68-myc fusion protein (Fig. 1B). The complexes generated in BMC-2 cells derived by transfected Eα52-68-eGFP were capable of eliciting a T cell response from 1H3.1 T cells (Fig. 1C), similar to that obtained from GST-Eα52-68-myc fusion protein delivered via the fluid phase (Fig. 1C) or cytoplasmically delivered via osmotic lysis of pinosomes (Mukherjee et al., 2001). Thus, expression of recombinant Eα52-68 containing fusion proteins in the cytosol of APCs specifically leads to presentation-competent peptide-MHCII molecules, confirming that peptides derived from cytosolically located proteins are also processed and presented on MHCII.
MHCII-restricted presentation of cytosolic antigen is not mediated through extra-cellular peptide regurgitation
The detection of peptides derived from endogenously expressed proteins or cytoplasmically introduced proteins suggest two possibilities for loading onto MHCII. Peptides from the cytoplasm may be regurgitated onto bystander APCs, alternatively peptides may be loaded onto autonomous MHCII by hitherto uncharacterized mechanisms. To distinguish between these possibilities we investigated whether peptides from cytosolic proteins are delivered into the extra-cellular medium, and subsequently bind MHCII either at the cell surface or in a conventional endosomal compartment. Such a pathway has been demonstrated in some instances involving MHCI-mediated presentation (Harding and Song, 1994).
To examine the role of bystander presentation (or cross-presentation), macrophages from mice of MHCII haplotype H-2k were cytosolically loaded with native OVA via osmotic lysis of pinosomes as previously described (Moore et al., 1988; Mukherjee et al., 2001). They were then co-incubated with bystander macrophages from mice with a different MHCII haplotype (H-2b) and the response of the OVA-specific H-2b-restricted 13.8 T cell line was tested (Fig. 2). No T cell stimulation was observed on bystander APCs [H-2b APCs (-Ag) incubated along with H-2k APCs (+Ag)], in contrast to the result when H-2b APCs were themselves loaded cytosolically with OVA (compare circles with squares in Fig. 2). These results show that regurgitation (bystander presentation or cross-presentation) is unlikely to account for the processing of cytosolic proteins for MHCII-restricted presentation via bystander APCs.
Peptide-MHC complexes from cytosolically located proteins as well as an endogenous trans-membrane anchored protein accumulate in a late lysosomal compartment
We next used the monoclonal antibody Y-Ae to investigate the intracellular localization of Eα52-68 peptide-I-Ab complexes. When the H-2b macrophage cell line BMC-2 transiently expresses Eα52-68-eGFP fusion protein, we find that Y-Ae+ complexes colocalized with LAMP-1, a lysosomal marker (Fig. 3A; Y-Ae staining was not detected in cells expressing eGFP alone; Fig. 3B). In macrophages from H-2b C57Bl/6 mice that were loaded cytosolically via osmotic lysis of pinosomes (Fig. 3C) or exogenously via fluid phase uptake (Fig. 3D) with the fusion protein GST-Eα52-68-myc, Y-Ae+ complexes were colocalized with intracellular compartments containing LAMP-1.
In APCs from B10.A(5R) mice, the Eα52-68 peptide is derived from endogenously expressed trans-membrane anchored I-Eα protein and is loaded onto I-Ab molecules (Rudensky et al., 1991b). The localization of endogenous Y-Ae-+ peptide-MHCII complexes in macrophages from B10.A(5R) mice was also found to be predominantly perinuclear, colocalized with LAMP-1 (Fig. 4A) and a lysosomal lumenal protease cathepsin-D (Fig. 4C). These molecules have been earlier shown to be marking MHCII peptide-loading compartments in LB27.4, a B-lymphoblastoid cell line (Geuze, 1998; Rudensky et al., 1994). Late endosomes and lysosomes are distinguished by the presence of the mannose-6-phosphate receptor, MPR (Griffiths et al., 1988). MPR traffics from the Golgi to the late endosomes, delivering mannosylated lysosomal lumenal enzymes, and is rapidly recycled back to the Golgi (Hunziker and Geuze, 1996; Kornfeld, 1992). The steady state distribution of the MPR is predominantly towards late endosomes and hence it serves as a late endosomal marker. Endogenously generated Y-Ae+ complexes were not colocalized with MPR; neither antibodies raised against MPR-300 (Fig. 4B) or MPR-46 (data not shown) exhibited significant colocalization with the Y-Ae+ structures. These data locate the main site of accumulation of the Eα52-68-peptide MHCII complex to the late-lysosomal system in these APCs.
We next examined the localization of Y-Ae+ complexes in bone marrow-derived I-Ab+ dendritic cells either endogenously expressing I-Eα (from B10.A(5R) mice) or derived from cytoplasmically introduced GST-Eα52-68-myc. In both cases intracellular Y-Ae reactivity was detected mainly in LAMP-1-positive structures (Fig. 5).
These data show that peptide-MHCII complexes derived from three different sources, cytosolic (expressed or introduced via osmotic lysis of pinosomes), constitutively expressed-transmembrane proteins (I-Eα), or delivered via the fluid phase (or receptor mediated pathway using maleylated proteins; data not shown), are localized in a large perinuclear compartment that colocalize extensively with LAMP-1 in a variety of APCs, including dendritic cells.
Rapid formation of peptide-MHCII complexes from cytosolic proteins in macrophages
To investigate the site of formation of peptide-MHCII complexes derived from cytosolic proteins, we used a pulse chase protocol. To deliver a short pulse of GST-Eα52-68-myc fusion protein into the cytosol of APCs we used the osmotic lysis of pinosomes protocol (Fig. 6B) or the Chariot™ method (Fig. 6C,D), and for exploiting the exogenous route we used the fluid phase pathway. C57B1/6 macrophages were thus loaded with GST-Eα52-68 fusion protein and allowed to process the antigen for various times (the chase; Fig. 6A,B), fixed and stained with the Y-Ae antibody to detect formation of intracellular peptide-MHCII complexes. We found that in all cases, Y-Ae+ peptide-MHCII complexes were formed rapidly, being detectable within 5-10 minutes of the start of the chase (Fig. 6A-C); the predominant distribution was in a large perinuclear compartment colocalized with LAMP-1. In the case of exogenously delivered protein, Y-Ae+ complexes were seen in some smaller peripheral compartments prior to their detection in the perinuclear LAMP-1+ compartment at extremely early times in the chase (Fig. 6A). However, cytosolically derived Y-Ae complexes were always detected only in large perinuclear compartments (Fig. 6B,C). With subsequent times of chase (20 minutes), the distribution of the exogenously and cytosolically derived complexes was indistinguishable, with only a perinuclear localization seen. Further chase (60 minutes) resulted in the emptying of the internal compartments, concomitant with the appearance of the peptide-MHC complexes on the cell surface. Thus, peptides from exogenous as well as cytosolically delivered protein are rapidly processed and loaded on MHCII within a similar time scale inside the macrophage APC, and trafficked to the cell surface in identical fashion. The peptide-MHC complexes generated in such a pulse-chase paradigm are also competent to be presented efficiently to MHCII-restricted T cells [Fig. 1C (see Mukherjee et al., 2001)].
To preclude any complication arising from the method of osmotic lysis of pinosomes used to deliver proteins into the cytosol of APCs, we examined whether this methodology was likely to disrupt endosomal trafficking in the APC. For this purpose we preceded a 10-minute pulse of exogenously delivered GST-Eα52-68-myc fusion protein to C57Bl/6 APCs with an osmotic lysis procedure marked by FITC-conjugated dextran, F-Dex (Supplemental text, http://jcs.biologists.org/supplemental/). Uptake of Cy3-mBSA (used to monitor the efficiency of endocytosis after the osmotic lysis protocol) was unaffected by the osmotic lysis protocol. The formation of YAe+ complexes from exogenously delivered GST-Eα52-68-myc fusion protein was also unaffected; these were detected in a similar perinuclear location at identical times of chase inside macrophages in treated and untreated cells (see supplemental data Fig. S1, http://jcs.biologists.org/supplemental/). Thus, the osmotic lysis method does not appear to have any effect on overall endosomal trafficking and antigen presentation properties of the APC.
The use of the Chariot™ reagent to deliver proteins into the cytoplasm of cells as a short pulse (Fig. 6C; see also supplemental data Fig. S2) further validates the results of the osmotic lysis protocol (Fig. 6B). The time course of generation of these complexes is remarkably similar to the osmotic stress-induced pinosome lysis protocol (Fig. 6D), confirming the validity of the different means used to deliver proteins into the cytoplasm of cells. Together these results show that peptides from cytoplasmic proteins are rapidly loaded onto MHCII in LAMP-1 compartments prior to their delivery to the cell surface.
Role of invariant chain in MHCII-restricted presentation of cytosolic protein
A major factor involved in determining the availability of MHCII molecules for peptide loading is the invariant chain (Ii). Invariant-chain-bound MHCII molecules intersect the endocytic pathway. The Ii chain is degraded by endosomally resident proteases, the CLIP (Class-II-associated invariant-chain peptide) is replaced and MHCII molecules bind peptides present in the endosomal lumen. Ii is thought to prevent newly synthesized MHCII molecules from being stably occupied by endogenous peptides present in the endoplasmic reticulum lumen (Cresswell, 1996). Ii also affects the intracellular transport and distribution of MHCII molecules by changing the efficiency of their egress from the ER and delivery to endosomal compartments (Bertolino and Rabourdin-Combe, 1996). Thus, in Ii-deficient mouse APCs there is a decrease in the MHCII-mediated presentation of many but not all exogenous antigens (Viville et al., 1993).
We therefore examined the role of Ii in presentation of cytosolic proteins on MHCII, using macrophage APCs from Ii-deficient or wild-type H-2b mice. For both exogenous and cytosolic pathways, the presentation of native OVA was lower in Ii-deficient mice, than in wild-type mice, as was the presentation of exogenous maleyl-OVA (Fig. 7A,B). The presentation of cytosolically delivered OVA on MHCI was unaffected in Ii-deficient APCs (Fig. 7C), confirming that the lowered efficiency of presentation of OVA peptides in Ii-deficient APCs was specific to MHCII-restricted T cells. To further examine whether Y-Ae-+ peptide-MHC complexes are detected in Ii-containing compartments we examined the colocalization of endogenous peptide-MHCII complexes with P4H5, an antibody that detects the luminal 99-116 aa of the Ii (Mehringer et al., 1991), and with In1, an antibody that recognizes the N-terminal cytosolic region of Ii (Koch et al., 1982). In neither case was there any significant colocalization (see supplemental data Fig. S3, http://jcs.biologists.org/ supplemental/). Consistent with this observation is the finding that Y-Ae-reactivity was not detected along with a resident ER chaperone, ERp57, a molecule also encoded by the MHC locus and important for conformational maturation of MHCI molecules (Hughes and Cresswell, 1998), nor colocalized with a marker of the Golgi (see supplemental data Fig. S3,). These data suggest that peptide-MHCII complexes from cytosolic sources are not formed in the secretory/biosynthetic pathway. Thus, Ii requirement is likely to be reflective of the impaired efficiency of MHCII to be delivered to the endosomal system.
The peptide exchange catalyst H-2M is required for the presentation of cytosolic proteins on MHCII
We next asked whether cytoplasmically generated peptides require H-2M, a molecular chaperone (Morris et al., 1994) to facilitate peptide loading onto MHCII. This chaperone is usually most effective in loading peptides in lysosomes because of its acidic pH optimum (Denzin and Cresswell, 1995; Sanderson et al., 1996; Vogt and Kropshofer, 1999). Peptides from many exogenously delivered proteins require the peptide exchange catalyst H-2M for their efficient loading and presentation on MHCII (Martin et al., 1996). The H-2M molecule is an MHC-like molecule, present in the lumen of endosomal compartments, where it catalyses the exchange of the CLIP peptide with an antigenic peptide in the groove of MHCII molecules (Sanderson et al., 1994). In the absence of H-2M, the major peptide species found on cell surface MHCII is the Ii-derived CLIP peptide, instead of the usual diverse repertoire (Miyazaki et al., 1996). Mice deficient for H-2M do not load exogenous Eα52-68 peptide on MHCII (Kovats et al., 1998).
We therefore asked if H-2M is important for presentation of peptides from cytosolic sources. Macrophages from H-2Mα-/- mice were used for peptide presentation and loading by MHCII molecules by delivering GST-Eα52-68-myc either via the cytosolic or exogenous pathways. Cytosolically delivered GST-Eα52-68-myc fusion protein (as well as OVA, data not shown), is not presented to T cells by H-2Mα-/- APCs (Fig. 8A,B). Similarly, Eα52-68-I-Ab complexes were also not detected inside these APCs by the Y-Ae antibody upon exogenous as well as cytosolically administered GST-Eα52-68-myc (Fig. 8C,D; see also supplemental data Fig. S4, http://jcs.biologists.org/supplemental/). This shows that after accessing endosomal compartments, peptides derived from both the cytosolic and exogenous protein sources depend on the lysosomally active peptide exchange catalyst H-2M for loading on MHCII.
To further confirm the role of peptide-loading chaperones in generation of peptide-MHCII complexes from cytoplasmic proteins we compared the generation of the Y-Ae+ peptide-MHCII complex in stably IAb-transfected wild-type and HLADM (the human homologue of H2-M)-deficient B cells. We find that transient expression of cytoplasmic Eα52-68 eGFP (but not eGFP alone) in wild-type B cells resulted in enhanced levels of the Y-Ae+ complexes whereas in B cells deficient for HLA-DM, the low levels of endogenous Y-Ae-staining were not enhanced by expression of cytoplasmic Eα52-68 eGFP (Fig. 8E,F). It should be noted that these B cell lines constitutively express DRα (human homologue of mouse IEα MHC class II molecules) thus generating detectable levels of Y-Ae+ complexes from the processing of endogenous DRα protein. Predictably the extent of Y-Ae+ complexes in the mutant cells is less than that in wild-type cells, consistent with a role for HLA-DM in endogenous Y-Ae+ complex generation as well [compare thin lines in Fig. 8E,F (Rudensky et al., 1994)]. Together, these results confirm that the peptide-loading chaperone in late endosomes is involved in the generation of the peptide-MHCII complex from cytoplasmically expressed proteins.
To ascertain that in H-2Mα-/- mouse cells, MHCII delivery to the lysosomal compartment or Ii processing is normal, we immunolocalized the CLIP peptide-MHCII complex (Fig. S4, http://jcs.biologists.org/supplemental/). Predictably we find that the CLIP compartment was completely colocalized with Y3P, an antibody that detects I-Ab-molecules (Janeway et al., 1984) in both H-2Mα-/- and control cells. However, Y-Ae reactivity was not detected in H-2Mα-/- cells when GST-Eα52-68-myc was delivered into the cytosol or provided exogenously via the fluid phase. This strongly suggests that in the absence of H-2M activity in the endosomal system, the YAe+ complex is not efficiently generated from cytoplasmically localized proteins.
The TAP transporter is not required for peptides from endogenous proteins to be presented on MHCII; the route of peptide loading on MHCII involves direct transport into an endo-lysosomal compartment
Peptides from endogenous proteins are processed in the cytoplasm and access the ER lumen via the TAP transporter, where they bind MHCI molecules (York and Rock, 1996). The TAP transporter belongs to the ABC family of transporters and functions to efficiently transport peptides for loading on MHCI molecules. We hence examined the possibility that the cytosolic peptides that load on MHCII also gain access to the membrane bound organelles via the TAP transporter.
We used macrophages from Tap1-/- mice (Van Kaer et al., 1992) which are deficient for TAP1. These macrophages are unable to present cytosolic OVA to MHCI-restricted B3 T cells (Fig. 9A). They also showed lower levels of surface MHCI (data not shown), because empty MHCI molecules do not exit the ER efficiently or last long on the cell surface (Ljunggren et al., 1990). However, Tap1-/- APCs presented cytosolically delivered proteins, GST-Eα52-68-myc and OVA (data not shown) on MHCII to the respective T cells at unaltered efficiencies compared to wild-type cells (compare Fig. 9B,C). Peptide-MHCII complexes from cytosolically delivered GST-Eα52-68-myc are also found localized to LAMP-1+ endosomal compartments in macrophages from Tap1-/- mice (Fig. 9D). In addition, the extent of generation of Y-Ae+ complexes from transiently transfected Eα52-68-eGFP in bone marrow-derived macrophages from Tap1-/- mice appeared no different from those formed in the wild-type mice (Fig. 9E). These data show that, unlike MHCI-restricted presentation, MHCII-restricted presentation of cytosolic antigens is Tap independent.
Cytoplasmic loading onto MHCII is insensitive to potent inhibitors of macroautophagy
To examine whether macroautophagy is a plausible mechanism for cytoplasmically expressed proteins to gain access to the lumenal peptide binding site of MHCII, BMC-2 cells, transfected with Eα52-68-eGFP were treated with inhibitors of PI 3-kinase activity, wortmannin (WT), LY294002 or 3-methyladenine (3-MA) (Stromhaug and Klionsky, 2001). None of the inhibitors reduced the generation of Y-Ae+- peptide-MHCII complexes (Fig. 10). However, inhibition of PI 3-kinase did induce a vesiculation of the LAMP-1 staining pattern (data not shown), indicative of inhibition of PI 3-kinase activity inside the cell. These results imply the lack of involvement of PI 3-kinase dependent macroautophagy in this process of cytoplasm to lysosome transport.
We present evidence here that a novel peptide transport mechanism is likely to be involved in the translocation of peptides derived from cytosolic and endogenous transmembrane anchored proteins. We have shown earlier that MHCII-mediated presentation of proteins introduced into the cytosol by osmotic lysis of pinosomes depends on cytosolic proteolytic mechanisms such as the proteasome (Mukherjee et al., 2001). We have examined the route and localisation of peptide transport and loading onto MHCII molecules.
It is possible that degraded peptides from dead or dying cells may be loaded on surface or recycling MHCII molecules on neighbouring cells. Using bystander APC assays, we show that this is not the mode of peptide loading on MHCII for cytosolically delivered proteins. It is hence possible that loading occurs in the ER lumen after peptide transport from the cytosol using mechanisms related to those used by MHCI. However, we find that peptide-MHCII complexes from endogenous proteins are not formed in the ER or the Golgi complex. In addition, the TAP transporter is not required for loading of peptides from cytosolic proteins on MHCII.
Peptide derived from proteins delivered into or expressed in the cytosol, and those derived from trans-membrane anchored proteins are loaded on MHCII in a large perinuclear compartment of the APC. This compartment contains markers of lysosomes, LAMP-1 and cathepsin-D, but does not contain a marker of the late endosomes (MPR), confirming the lysosomal nature of the compartment where peptide loading on MHCII occurs in these cells.
A major facilitator of peptide loading on MHCII in the endo-lysosomal compartment is the H-2M molecule. Previous reports (Kovats et al., 1998) as well as our data here show that the endogenous pathway of MHCII-mediated presentation of cytosolic proteins is dependent on H-2M, similar to the classic pathway of loading peptides derived from exogenous antigens. Peptides from cytosolic proteins required the peptide-loading chaperone H-2M or HLA-DM as seen in experiments using APCs from mice lacking H-2M or in the human B cell line deficient in HLA-DM. Together with the almost immediate immunolocalization of the peptide-MHCII complex in the lysosome after a short cytosolic pulse, these data show that while the peptides from endogenous proteins are generated in the cytosol of APCs, they are rapidly and possibly directly transported into lysosomal compartments followed by a H-2M-dependent loading onto MHCII. Our data are thus consistent with a model of cytoplasmic antigen processing coupled to rapid transport into the endo-lysosomal compartment, and suggest a novel mechanism of peptide transport for the efficient presentation of endogenous cellular proteins to CD4 T cells.
How then do peptides generated from cytosolic sources enter vesicular compartments and reach the endo-lysosomal region for binding MHCII? Peptides generated via proteasomal degradation in the cytosol will have to cross the lysosomal membrane barrier. We discuss two possibilities. One involves transport via peptide transporters and the other via multiple mechanisms that come under a umbrella term `autophagy' (Cuervo, 2004; Klionsky and Ohsumi, 1999).
While our data shows that TAP, the ER resident peptide transporter from the ABC family is not involved in this peptide transport process, TAP is a member of the transmembrane ABC family of transporters, and other members of this family may be involved. The ABC transporter ABC-1 has been claimed to affect the transport of the leaderless protein interleukin-1β from the cytosol to the lysosome in monocyte cells (Andrei et al., 1999; Hamon et al., 1997). ABC-B9, another promising candidate transporter, phylogenetically related to TAP, is localized to LAMP-1-containing lysosomes, although its function is as yet unknown (Zhang et al., 2000). Other as yet unidentified ABC transporters may also be present on lysosomal membranes. Molecular tools would be helpful in addressing whether the function of any of the ABC transporters is necessary in translocation of proteins or peptides into lysosomes.
The MHCII peptide loading compartments exhibit a multivesicular or multilamellar ultrastructure in many cells. Another mechanism that can result in inclusion of cytoplasmic proteins is the process of autophagy, of which there are many forms (Cuervo, 2004). The lack of an effect of the selective autophagy inhibitor, 3-methyladenine, and inhibitors of PI 3-kinase on the generation of Y-Ae+ complexes, argues against any role for macroautophagy. An alternative autophagic process, described as `chaperone mediated autophagy' (Cuervo and Dice, 1996; Cuervo et al., 2003) is up-regulated under stress, serum withdrawal, and involves the selective uptake of cytosolic proteins into lysosomes. Uptake and processing of OVA and Eα52-68-GFP (both of which lack KFERQ-like motifs) via this pathway, argues against this mechanism. This leaves the constitutive (but poorly characterized) micropinocytic pathway wherein cytoplasmic proteins are included in multivesicular lysosomes, which could be degraded and presented by MHCII molecules (Klionsky and Ohsumi, 1999). Additional, as yet unidentified mechanisms may also exist that are either specific to antigen presenting cells, or ubiquitous to many cell types, and result in the import of proteins and peptides directly into lysosomes.
This work was supported in part by grants from the Departments of Biotechnology and Science and Technology, Government of India (A.G., S.R., V.B.) and by intra-mural funds from NCBS (S.M.). The National Institute of Immunology is supported by the Department of Biotechnology, Government of India. A.D. was supported by a PhD fellowship from TIFR and a Kanwal Rekhi Fellowship from the TIFR Endowment Fund. S.M. is a Senior Research Fellow of the Wellcome Trust (UK), and thanks V. Sriram for critically reading the manuscript, and K. Belur and F. F. Bosphorus for inspiration.
Supplemental data available online
- Accepted April 22, 2004.
- © The Company of Biologists Limited 2004