Cutting Edge: CD91-Independent Cross-Presentation of GRP94(gp96)-Associated Peptides

Cells and tissue culture

C57BL/6 mice (Charles River Breeding Laboratories, Wilmington, MA) were used to prepare thioglycolate-elicited peritoneal macrophages. Macrophages were harvested 4–5 days postinjection and enriched by adherence selection.

Protein purification and labeling

GRP94 was purified by the method of Wearsch and Nicchitta (20). Pseudomonas exotoxin was obtained from Sigma-Aldrich (St. Louis, MO). Texas Red-, biotin-, and fluorescein-succinymidyl esters (Molecular Probes, Eugene, OR) were used to label proteins according to manufacturer’s protocols. Abs against K^b-OVA complex (25-D1.16) and CD91 were the kind gifts of Dr. J. Yewdell (National Institutes of Health, Bethesda, MD) and Dr. S. Argraves (University of South Carolina, Charleston, SC), respectively.

α₂M was purified as previously described (21). Purified α₂M was converted to the CD91-binding, thiol ester-cleaved derivative (α₂M*) by incubation with 0.2 M NH₄HCO₃. α₂M* was then labeled with Alexa Fluor (Molecular Probes). RAP was purified as previously described (22). Dr. J. Herz (University of Texas Southwestern Medical Center, Dallas, TX) kindly provided pGEX-RAP expression vectors.

Cell surface binding and uptake

Receptor-mediated uptake reactions were performed as described previously (23). For confocal microscopy analysis, cells were fixed in 4% paraformaldehyde and mounted under 10% PBS, 90% glycerol, and 1 mg/ml phenylenediamine.

Biotin-GRP94 cell surface receptor binding interactions were analyzed by incubating 10⁶ cells/assay with increasing concentrations of biotin-GRP94 for 30 min on ice. Cells were then washed and resuspended in SDS-PAGE sample buffer, and extracts were prepared for SDS-PAGE. Following transfer to nitrocellulose membranes, biotin-GRP94 levels were determined by ECL, following avidin-HRP detection of biotin-GRP94. Quantification of cell surface-bound biotin-GRP94 was determined against a standard curve prepared from serial dilutions of biotin-GRP94.

Adherent macrophages were incubated at 4°C in RPMI 1640, 0.5% BSA for 30 min with the following ligands: fluorescein-GRP94 and Alexa-α₂M* in the presence or absence of RAP. For competition assays final concentrations used were 25 nM for GRP94 and α₂M* and 2500 nM for RAP (100-fold molar excess). After the 30-min binding period, macrophages were washed and then fixed in 4% paraformaldehyde/PBS. Following fixation, the cells were rinsed and unreacted paraformaldehyde was quenched with 0.05 M NH₄Cl. Coverslips were then mounted onto slides in mounting medium. All images were obtained on a Zeiss LSM 410 laser scanning confocal microscope (Thornwood, NY) using Zeiss LSM version 3.95 software. All image size and contrast adjustments were performed with PhotoShop (version 4) software (Adobe Software, Palo Alto, CA).

For study of CD91 binding and RAP competition by FACS, adherent macrophages were incubated for 30 min at 4°C with fluorescein-GRP94 or α₂M*-AF488, in the presence or absence of RAP. Ligand and RAP concentrations were as indicated above. Following incubation, cells were rinsed and fixed in 1% paraformaldehyde. Cells were analyzed for fluorescence by FACS and analysis was performed using CellQuest (BD Biosciences, San Jose, CA).

Chaperone-based re-presentation

Peptide re-presentation assays were performed with GRP94 complexed with SIINFEKL peptide (by heat shock, 15 min at 50°C) (24, 25) and subsequently isolated from free peptide by Sephadex G-75 size exclusion chromatography (Sigma-Aldrich, St. Louis, MO). This preparation does not contain free peptide (26). To assay GRP94/peptide re-presentation, GRP94/peptide complexes and, where indicated, α₂M* were incubated with elicited primary peritoneal macrophages and, following a 3-h incubation at 37°C, subsequently stained with 25-D1.16 Ab (27). The cells were then fixed in 2% paraformaldehyde for FACS analysis.

Trafficking itineraries of GRP94 and CD91

CD91 has been identified as the unique receptor responsible for directing GRP94 into MHC class I re-presentation pathways of APCs (12, 13, 14). CD91 is a member of the LDL family of lipoprotein receptors, which bind and internalize an array of ligands, the majority of which are targeted to lysosomes (28). Because little is known regarding the CD91-dependent trafficking of GRP94, the trafficking itineraries of CD91, GRP94, and known CD91 ligands were examined.

To evaluate the subcellular trafficking of CD91 ligands, the trafficking pattern of GRP94 was compared with that of Pseudomonas exotoxin, an obligate CD91 ligand (29). In these experiments, fluor-labeled GRP94 and Pseudomonas exotoxin were bound to macrophage cell surface receptors, the cells were washed, and the staining pattern was analyzed following warming to 37°C. As is evident in Fig. 1⇓A, receptor-internalized GRP94 (red channel) and Pseudomonas exotoxin (green channel) were trafficked to distinct subcellular compartments. The lack of costaining suggested that GRP94 and CD91 were rapidly segregated upon internalization to yield distinct trafficking itineraries. To examine this hypothesis, we determined whether, at early time points, internalized GRP94 and CD91 colocalized. Surprisingly, and as shown in Fig. 1⇓B, GRP94 (red channel) taken up by receptor-mediated endocytosis did not colocalize with internalized CD91 (green channel). This is in direct contrast to the colocalization observed of receptor-internalized GRP94 and IgG (Fig. 1⇓C), which traffic to a FcR⁺rab5a⁻ endosomal compartment (26). In interpreting these data with respect to a role for CD91 in chaperone-mediated cross-presentation, it is important to note that a functional role for CD91 in this process has been proposed not on the basis of direct trafficking studies but rather from the observed inhibition of peptide re-presentation by CD91-specific Abs and by α₂M competition (12). To reconcile these differences, a direct analysis of CD91 function in GRP94 cell surface binding and uptake was performed.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Divergent trafficking pathways for GRP94, CD91, and Pseudomonas exotoxin. Texas Red (TR)-GRP94 and fluorescein (Fl)-labeled Pseudomonas exotoxin (PE) (A), fluorescein-labeled Abs against CD91 (B), or fluorescein-labeled IgG (C) were bound to primary peritoneal macrophage cell surface receptors on ice, the cells were washed, and receptor-mediated uptake was initiated by warming the cells to 37°C. After a 30-min (A) or 7-min (B and C) trafficking period, cells were processed for confocal microscopy. Colocalization is observed in the two-channel merge as yellow.

Analysis of LDL receptor-related protein family activity in receptor-mediated uptake of GRP94

CD91 is expressed on a diverse array of cell types, including, but not limited to, APCs (15). Previously, we reported that HepG2, a CD91-positive human hepatoma cell line, did not display GRP94 cell surface binding (23, 30). To examine whether there was a positive correlation between CD91 expression and cell surface binding of GRP94, these two parameters were evaluated in an additional CD91-positive cell line, Chinese hamster ovary (CHO), and in RAW264.7 macrophages, a cell line that is CD91 positive and that binds GRP94 (12, 23, 31, 32). As depicted in Fig. 2⇓A, both CHO and RAW264.7 cells express nearly identical levels of CD91, findings consistent with previous reports (14, 31, 32). However, though both cell types were CD91 positive, only RAW264.7 cells display appreciable cell surface binding of GRP94; thus, as previously concluded with regard to HepG2 cells, a positive correlation between CD91 expression and GRP94 cell surface binding could not be demonstrated.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

CD91-independent cell surface binding of GRP94 to elicited macrophages. A, CHO and RAW264.7 macrophages were incubated with fluorescein-GRP94 on ice, the unbound GRP94 was removed, and the cells were subsequently decorated with anti-CD91 mAb (rhodamine channel). Bound GRP94 and CD91 levels were determined by FACS. B, Fluorescein-GRP94 binding to the cell surface of RAW264.7 macrophages was conducted in the presence of BSA, BSA plus α₂M* (400 μg/ml), or DMEM plus 10% bovine serum, followed by FACS analysis. Inset, Scatchard analysis of GRP94 binding to macrophages. GRP94 exhibits a K_d of ∼2 × 10⁻⁷ M to macrophages, occupying ∼10⁶ high-affinity sites per cell. C, The effects of the CD91 ligand α₂M* on the receptor-mediated internalization of GRP94 was examined in RAW264.7 macrophages. Fluorescein-GRP94 was allowed to bind to RAW264.7 macrophages in the presence of 200 μg/ml α₂M*, the free ligand was removed, and the cells were subsequently warmed to 37°C for 7 min to promote internalization.

Subsequently, the capacity of the CD91 ligand, α₂M*, to compete for GRP94 cell surface binding was examined. Previously, Binder et al. (12) reported that α₂M efficiently inhibits re-presentation of GRP94-associated peptides, as assayed using peptide-dependent stimulation of T cells as a surrogate for receptor function in GRP94 peptide uptake. This observation is surprising, as CD91 ligands generally do not cross-compete for binding (28). However, a second high-affinity α₂M* binding site has been demonstrated on the surface of APCs (33); thus, it was considered that GRP94 cell surface binding was conferred by this related activity. Therefore, we performed competition experiments with highly purified α₂M*, the CD91 binding form of α₂M, or serum, which contains ∼2 mg/ml α₂M (34). As shown in Fig. 2⇑B, BSA, BSA supplemented with α₂M*, and DMEM plus 10% serum had no effect on the binding of GRP94 to thioglycolate-elicited macrophages. Indeed, GRP94 was observed to bind to and be internalized by elicited macrophages in the presence of 200 μg/ml α₂M* (Fig. 2⇑C). High-affinity binding of GRP94 to macrophage cell surfaces has previously been shown to be specific (8, 13) and saturable (Fig. 2⇑B). Because α₂M* constitutes ∼1% of total α₂M (35),the α₂M* levels (200 μg/ml) used in these experiments are equivalent to a total α₂M concentration of 20 mg/ml, a concentration in vast excess to that necessary for competition at a shared site(s).

Re-presentation of GRP94-associated peptides via a CD91-independent pathway

RAP is a 39-kDa protein that acts as a universal antagonist to all known CD91 ligands, including α₂M* (16, 17, 19). If CD91 functions in the internalization of GRP94, and as α₂M has been reported to block the interaction of GRP94 with CD91, RAP would be predicted to antagonize the binding of GRP94 to APCs. Therefore, we tested the effects of RAP on cell surface binding of GRP94, using an established ligand, α₂M*, as a positive control. Fig. 3⇓B indicates that RAP, at a 100-fold molar excess, did not inhibit GRP94 binding to elicited macrophages. Importantly, under identical assay conditions, α₂M* binding was efficiently blocked by RAP (Fig. 3⇓C). The inhibition of GRP94 binding by RAP was not due to GRP94 displaying a higher affinity for APCs than α₂M*, as GRP94 exhibits an overall K_d of ∼2 × 10⁻⁷ M to elicited macrophages (Fig. 2⇑B, inset), as compared with a low nanomolar K_d for α₂M* binding to CD91 (36). The effects of RAP on cell surface binding of fluorescein-labeled GRP94 and Texas Red-labeled α₂M* was also examined by confocal microscopy (Fig. 3⇓, D and E). Consistent with the FACS data, RAP was without effect on GRP94 binding (green channel), whereas α₂M* binding (red channel) was efficiently blocked (Fig. 3⇓, compare D, minus RAP, with E, plus RAP). Finally, we tested whether α₂M* directly blocked the re-presentation of GRP94-associated peptides. In these experiments, it was observed that SIINFEKL peptide, complexed with GRP94, was re-presented both in the presence of α₂M-containing serum and in the presence of serum supplemented with 100 μg/ml α₂M* (Fig. 3⇓F), a level previously reported to abolish GRP94-dependent cross-priming of T cells (12).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

GRP94 binding to macrophages is not inhibited by the CD91 antagonist RAP, nor is peptide re-presentation suppressed by α₂M*. A, Elicited peritoneal macrophages were assayed by FACS analysis for their ability to bind 25 nM GRP94 (B) or α₂M* (C) in absence or presence of a 100-fold molar excess of RAP. As illustrated in D and E, RAP competition of α₂M* but not GRP94 was confirmed by confocal microscopy. These experiments were conducted identically to those illustrated in A–C. F, Elicited peritoneal macrophages were incubated in the presence of 50 μg/ml GRP94-OVA complex in the absence (GRP94/OVA), or the presence of 100 μg/ml α₂M* (GRP94/OVA/α₂M) for 3 h at 37°C. Peptide re-presentation was assessed by subsequent staining with the K^b-OVA-specific Ab 25-D1.16 and FACS analysis. Also shown is 25-D1.16 staining of macrophages incubated in the absence of GRP94-OVA.

In noting the marked time interval differences in assays of peptide re-presentation on cell surface class I molecules (≤3 h) and GRP94-mediated T cell activation (>20 h) (12), it should be considered that the reported inhibition of T cell activation by CD91 Ab and α₂M may reflect physiological responses of the cells to prolonged culture with these reagents. For example, ligand-bound CD91 elicits pronounced cell activation; thus, we suggest that CD91 Ab and α₂M may disrupt the ability of APC to activate cognate T cells (37, 38, 39).

In summary, the included data indicate that macrophage CD91 displays biochemical properties consistent with its role as a α₂M* receptor. However, the lack of a positive correlation between CD91 expression and GRP94 cell surface binding, the absence of colocalization between GRP94 and CD91 in the early trafficking itinerary of the two proteins, the inability of α₂M* to inhibit the binding or internalization of GRP94 or the re-presentation of its associated peptides, and the observation that RAP, a biological antagonist for CD91 receptor ligands, is without effect on GRP94 binding argue strongly against a role for CD91 in the receptor-mediated internalization of GRP94. From these data, it is equally evident that APCs bear cell surface receptors that are capable of directing GRP94 into the class I Ag re-presentation pathway.