Calreticulin recognizes misfolded HLA-A2 heavy chains

Abstract

Our studies investigated functional interactions between calreticulin, an endoplasmic reticulum chaperone, and major histocompatibility complex (MHC) class I molecules. Using in vitro thermal aggregation assays, we established that calreticulin can inhibit heat-induced aggregation of soluble, peptide-deficient HLA-A2 purified from supernatants of insect cells. The presence of HLA-A2-specific peptides also inhibits heat-induced aggregation. Inhibition of heat-induced aggregation of peptide-deficient HLA-A2 by calreticulin correlates with a rescue of the HLA-A2 heavy chain from precipitation, by forming high-molecular-weight complexes with calreticulin. Complex formation between HLA-A2 heavy chains and calreticulin occurs at 50°C but not 37°C, suggesting polypeptide-based interactions between the HLA-A2 heavy chain and calreticulin. Once complexes are formed, the addition of peptide is not sufficient to trigger efficient assembly of heavy chain/β2m/peptide complexes. Using a fluorescent peptide-based binding assay, we show that calreticulin does not enhance peptide binding by HLA-A2 at 37°C. We also show that calreticulin itself is converted to oligomeric species on exposure to 37°C or higher temperatures, and that oligomeric forms of calreticulin are active in inhibiting thermal aggregation of peptide-deficient HLA-A2. Taken together, these results suggest that calreticulin functions in the recognition of misfolded MHC class I heavy chains in the endoplasmic reticulum. However, in the absence of other endoplasmic reticulum components, calreticulin by itself does not enhance the assembly of misfolded MHC class I heavy chains with β2m and peptides.

MHC class I molecules are heterodimers consisting of a transmembrane glycoprotein (heavy chain), a small soluble protein [β2-microglobulin (β2m)], and a short peptide. This complex of proteins is found on the surface of nearly every mammalian cell and serves as the ligand for CD8 T cells. Assembly of major histocompatibility complex (MHC) class I-peptide complexes occurs in the endoplasmic reticulum (ER). Before peptide binding in the ER, class I heterodimers and class I heavy chains are unstable, and are found associated with the ER chaperones calreticulin and calnexin (reviewed in refs. 1 and 2). Calreticulin has been found to be associated with MHC class I heavy chains and β2m (3, 4), and been implicated in MHC heterodimer assembly with peptides (reviewed in ref. 1). Calnexin has been found to be associated with free human MHC class I heavy chains and has been implicated in the initial folding of class I heavy chains (reviewed in ref. 2). Calnexin and calreticulin are structurally related, and contain lectin domains that bind to glycoproteins that have partially trimmed, monoglucosylated core glycans with the structure Glc₁Man₉GlcNAc₂ (Glc, glucose; Man, mannose; GlcNAc, N-acetylglucosamine) (5). It was originally believed that recognition and binding of monoglucosylated oligosaccharides present on glycoproteins were necessary and sufficient for calnexin and calreticulin to promote folding of the glycoproteins (see, for example, ref. 6). However, more recent experiments have indicated that both calnexin and calreticulin directly discriminate between protein conformational states, and that both proteins can function in vitro as chaperones for glycosylated, as well as nonglycosylated, substrates (7, 8).

Based on previous immunoprecipitation analyses from detergent lysates of mammalian cells, MHC class I heavy chains and β2m that are in complex with calreticulin have been suggested to be simultaneously associated with ERp57, a thiol disulfide isomerase, the transporter associated with antigen processing (TAP), and tapasin (refs. 3 and 4, and other studies summarized in ref. 1). Together, these proteins have been referred to as the MHC class I peptide loading complex (1). Individually or in combination, these proteins have been suggested to perform the general chaperone functions of promoting folding and assembly of MHC class I/peptide complexes, and inhibiting cell-surface trafficking of incompletely assembled MHC class I components. In the present studies, we investigated functional interactions between purified calreticulin and a peptide-deficient form of the human MHC class I molecule, HLA-A2. Previous analyses of functional interactions between calreticulin and MHC class I proteins have been carried out in Drosophila and Sf9 insect cells, and in these systems, calreticulin has been shown to enhance expression of MHC class I heavy chains and heterodimers (9, 10). Our studies indicate direct binding between calreticulin and HLA-A2 heavy chains at elevated temperatures. Although it has been suggested (3), the occurrence of direct binding between calreticulin and MHC class I has previously been difficult to unambiguously demonstrate by coimmunoprecipitation analyses from detergent lysates of cells, owing to the presence of several proteins that coprecipitate with MHC class I molecules and calreticulin. We show here that the calreticulin/HLA-A2 interaction inhibits heavy chain aggregation, and also that HLA-A2-specific peptides can inhibit aggregation. However, the calreticulin–heavy chain interaction does not directly facilitate subsequent assembly of heavy chain/β2m/peptide complexes. Furthermore, calreticulin does not affect HLA-A2/peptide assembly at 37°C. These studies provide insights into the nature and possible functions of calreticulin interactions with MHC class I molecules within cells.

Methods

Peptides and Proteins.

Peptides were obtained synthetically and labeled with FITC as described (11). Calreticulin was purified as described (12). A baculovirus construct encoding a soluble histidine-tagged version of the HLA-A2 heavy chain and chimp β2m (which is identical to the human β2m sequence at the amino acid level) was constructed after subcloning the corresponding DNA sequences into the BamH1 (heavy chain) and BglII (β2m) sites of the transfer vector pAcUW31 (PharMingen). High-Five insect cells (≈9 × 10⁸) were infected with the HLA-A2 baculovirus at a multiplicity of infection of 30. After 72 h, culture supernatants were collected and brought to a neutral pH. Soluble HLA-A2/β2m molecules were purified from the culture supernatant by using a W6/32 affinity column as described (13).

Thermal Aggregation Assays.

Thermal aggregation was monitored by measuring the light scattering at 360 nm by using a Spex Fluorolog (Spex Industries, Edison, NJ) with a 150- or 450-W xenon lamp, exciting through a single monochromator (Model 1681, 220 nm focal length) with emission monitored through a double monochromator (Model 1682, 220 nm focal length), with a cell holder attached to a circulating water bath set at the desired temperature. Measurements were recorded every 1 min. Aggregation of HLA-A2 (4–6 μM) in aggregation assay buffer [10 mM Tris/150 mM NaCl/5 mM CaCl₂ (pH 7.2)] was induced by incubation at 45-50°C (the extent of aggregation varied between protein preparations; thus, some preparations required a temperature of 50°C to visualize aggregation) in the presence or absence of indicated amounts of calreticulin, IgG, or peptides.

Peptide Binding Assays.

Proteins and peptide were incubated under the indicated conditions with the fluorescent peptide LLDC_FITCPTAAV in aggregation assay buffer, followed by centrifugation and native-polyacrylamide gel electrophoresis (7.5–8% gels). HLA-A2/LLDC_FITCPTAAV complexes were visualized by fluorescence imaging of the gels by using a FluorImager SI Fluorescence Scanner (Molecular Dynamics). The scanner uses an argon-ion laser (488 nm) for gel illumination. The emitted light is detected by a photomultiplier tube and the signal converted into a digital image. Fluorescent bands were quantified and background subtracted using imagequant software.

Coimmunoprecipitation Analyses to Detect Complex Formation.

Calreticulin, HLA-A2, or HLA-A2+calreticulin (12 μM each, in a total volume of 60 μl) were heated in aggregation assay buffer at 50°C or in indicated experiments at 37°C for 1 h. The samples were diluted with aggregation assay buffer to 500 μl, and incubated with anti-his ascites (2 μl per immunoprecipitation) or control antibody for 1 h at 4°C, followed by centrifugation. The supernatants were incubated overnight at 4°C with washed G protein beads. The beads were washed three times with aggregation assay buffer containing 0.5% Triton X-100, resuspended in SDS/PAGE buffer, and separated by SDS/PAGE (15% gels). Immunoprecipitated proteins were visualized by Coomassie blue staining.

Results

HLA-A2 Purified from Insect Cells Rapidly Associates with Fluorescent HLA-A2-Specific Peptides.

Insect cells lack essential components of the MHC class I assembly pathway, such as transporter associated with antigen processing (TAP), and class I molecules expressed in these cells are expected to be deficient in peptides. HLA-A2 is an unusual MHC class I molecule in that it binds to signal peptides within the ER lumen (14). Previous experiments have shown the presence of HLA-A2-specific peptides within the ER of Sf9 insect cells (10), suggesting that the soluble A2 expressed in insect cells may be at least partially peptide-occupied. To assess the peptide occupancy of the soluble HLA-A2, we developed a fluorescence-based peptide binding assay. The HLA-A2-specific peptide LLDVPTAAV (14) was modified by replacing the valine position 4 with a cysteine (LLDCPTAAV), and the modified peptide labeled using iodoacetamido-fluorescein, and purified by HPLC (LLDC_FITCPTAAV). It has previously been shown that LLDCPTAAV, when immobilized to a Biacore biosensor chip via the cysteine residue, bound HLA-A2 with high affinity, indicating that the residue at position 4 is likely to be surface-accessible when HLA-A2-bound (15). HLA-A2 was incubated at 4°C or 37°C with a 10-fold molar excess of LLDC_FITCPTAAV or buffer for 1 h, or overnight, followed by native gel electrophoresis. In the absence of added peptide, three major bands were observed on native-PAGE and Coomassie staining of HLA-A2, most likely because of heterogeneous peptide occupancy, as discussed below (Fig. 1A, lanes 1 and 3, bands 1–3). When preincubated with peptide, protein corresponding to some of the bands was converted to a faster-migrating species (Fig. 1A, lanes 2, 4, and 5). Fluorescence imaging of the gel by using a fluorescence scanner indicated the presence of a fluorescent band at a similar migration position as the fastest migrating HLA-A2 band seen in the presence of peptides (Fig. 1B, lanes 2, 4, and 5). The fluorescence intensity of the band increased after overnight incubation with the fluorescent peptide compared with the 1-h incubation (Fig. 1B, lane 4 compared with lane 5). Furthermore, more peptide bound to HLA-A2 after the 37°C incubation compared with the 4°C incubation. These observations are consistent with the possibility that the HLA-A2 purified from insect cells is a heterogeneous mixture of empty protein, protein occupied with low affinity peptides, and protein occupied with high-affinity insect-cell-derived peptides. At 4°C, empty HLA-A2 is likely to bind rapidly to LLDC_FITCPTAAV, whereas exchange of LLDC_FITCPTAAV for endogenous peptides is likely to proceed at a much slower rate. At 37°C, dissociation of low-affinity peptides is likely to be enhanced, allowing for better binding of LLDC_FITCPTAAV. The HLA-A2 that does not bind peptide even after incubation for 1 h at 37°C is likely to be occupied with high-affinity peptides. Circular dichroism-based assays have previously been used to demonstrate the reduced thermal stability of peptide-deficient MHC class I molecules compared with peptide-filled MHC class I molecules (see, for example, ref. 16). Using similar assays, we could demonstrate reduced thermal stability of the insect-cell-purified HLA-A2 compared with the same protein that was preincubated with LLDCPTAAV (data not shown).

Figure 1.

HLA-A2 purified from insect cells rapidly associates with A2-specific peptides. (A and B) Peptide binding by HLA-A2 was assessed by native-PAGE followed by fluorescence imaging of gels (B) to visualize HLA-A2/peptide complexes, or by Coomassie staining of gels to visualize total proteins (A). HLA-A2 (12 μM) was incubated for 1 h (lanes 1–4) or overnight (lane 5) in the absence (lanes 1 and 3) or presence (lanes 2, 4, and 5) of the fluorescent peptide LLDC_FITCPTAAV (120 μM) at 37°C (lanes 1 and 2) or 4°C (lanes 3–5). Calreticulin (12 μM) incubated overnight at 4°C without (lane 6) or with 120 μM LLDC_FITCPTAAV (lane 7) were also analyzed, as was a control protein (IgG 12 μM) incubated without (lane 8) or with 120 μM LLDC_FITCPTAAV (lane 9). The incubation mixtures were in a total volume of 15 μl, of which 12 μl were loaded in each lane of the native gel.

By native PAGE, calreticulin migrates predominantly as a single discrete band (Fig. 1A, lanes 6 and 7). Significant fluorescence signals were not observed on incubation of either calreticulin or the control protein (IgG) with LLDC_FITCPTAAV (Fig. 1B, lanes 7 and 9), although a faint signal was visualized with calreticulin in some protein preparations. These observations indicate that under the conditions of the analyses, HLA-A2, but not calreticulin or IgG, form high-affinity complexes with LLDC_FITCPTAAV.

Calreticulin and Peptides Inhibit Thermal Aggregation of HLA-A2 Purified from Insect Cells.

To investigate functional interactions between calreticulin and the peptide-deficient HLA-A2, we measured heat-induced aggregation of the HLA-A2 in the presence or absence of calreticulin or a control protein (IgG). In these assays, light scattering was monitored for 30 min after placing the samples in a cuvette maintained at 45–50°C with a circulating water bath. HLA-A2 aggregation was observed under the conditions of the experiments (Fig. 2A). Calreticulin efficiently inhibited HLA-A2 aggregation even when present in substoichiometric amounts (HLA-A2:calreticulin, 1:0.5 molar ratio), whereas the control protein (IgG) did not inhibit aggregation even when present in stoichiometric excess relative to HLA-A2 (Fig. 2A). Subsequent to the thermal aggregation assays, aliquots of the protein mixtures were analyzed by SDS/PAGE, to exclude the possibility that decreased aggregation was due to proteolysis of HLA-A2 by contaminating proteases present in either protein preparation. In these analyses, no decrease in total HLA-A2 content were observed (data not shown). These results suggested that calreticulin recognized a conformation of HLA-A2 that was induced at the higher temperatures, which resulted in an inhibition of HLA-A2 aggregation.

Figure 2.

Calreticulin and peptides inhibit thermal aggregation of soluble peptide-deficient HLA-A2. The indicated samples were diluted into aggregation assay buffer and heated to 45–50°C while measuring light scattering at 360 nm. The HLA-A2 concentrations were ≈5 μM in all experiments. Results shown are representative of three (A and B) or two (C) independent experiments. (A) The effect of calreticulin or IgG (at the indicated stoichiometric ratios) on HLA-A2 aggregation was assessed by mixing the proteins with HLA-A2 immediately before the start of the assay. (B) The effects of the indicated peptides on aggregation were assessed by preincubating HLA-A2 with 50 μM specific peptide (LLDCPTAAV) or control peptide (RRYQKSTEL) at room temperature for 1 h. Results similar to that with LLDCPTAAV were obtained on preincubation of HLA-A2 with other HLA-A2-specific peptides (GICGFVFTL and GILGCVFTL). (C) Fifty-micromolar peptides were added immediately before inducing thermal aggregation.

Thermal aggregation assays were also performed in the absence or in the presence of added exogenous peptides. Peptide-deficient HLA-A2 heterodimers were incubated for 1 h with HLA-A2-specific or nonspecific peptides (at a peptide: HLA-A2 molar ratio of 10:1). Subsequently, thermal aggregation of these HLA-A2/peptide mixtures were determined. The results shown in Fig. 2B indicated that, like calreticulin, HLA-A2-specific peptides also efficiently prevented HLA-A2 thermal aggregation. The HLA-B27-specific peptide RRYQKSTEL did not inhibit thermal aggregation of HLA-A2 molecules, but instead appeared to enhance aggregation. LLDCPTAAV also inhibited aggregation even when added immediately before the start of the aggregation assay (Fig. 2C), indicating that under the conditions of the experiments, the aggregation kinetics were slower than peptide-binding kinetics. Taken together, these experiments demonstrated that both calreticulin and peptides inhibited thermal aggregation of HLA-A2, and also indicated that the aggregating species was peptide-deficient HLA-A2.

Calreticulin Rescues the HLA-A2 Heavy Chains from Precipitation by Forming High-Molecular-Weight Complexes with HLA-A2.

To further investigate the nature of interactions between calreticulin and HLA-A2, samples containing HLA-A2 alone, or HLA-A2 and calreticulin, were incubated at 45–50°C for 1 h. Samples were centrifuged, separated into pellet and supernatant fractions, and analyzed by SDS/PAGE. In the presence of calreticulin compared with when calreticulin was absent, less HLA-A2 heavy chain was recovered in the pellet, and more HLA-A2 heavy chain was recovered in the supernatant (Figs. 3A).

Figure 3.

Calreticulin rescues the HLA-A2 heavy chain from precipitation and the rescued heavy chain co-migrates with calreticulin as a high-molecular-weight species. (A) Calreticulin (15 μM), HLA-A2 (15 μM), or HLA-A2 and calreticulin (15 μM each) were incubated in aggregation assay buffer at 45°C for 1 h. Protein mixtures were centrifuged at 15,000 × g for 15 min and separated into supernatant (S) or pellet (P) fractions. Pellets were resuspended in SDS/PAGE sample buffer in a final volume equal to that of the supernatants. Twenty microliters of supernatant or pellet fractions were separated by SDS/PAGE and visualized by Coomassie blue staining. (B) Two hundred microliters of supernatants from A were analyzed by gel filtration chromatography on a Superose 6 column. Chromatographic profiles were monitored by measuring the absorbance at 280 nm. (C) Peaks I-III from the HLA-A2+calreticulin chromatogram indicated in B were analyzed by SDS/PAGE and visualized by Coomassie blue staining. Peak I corresponds to complexes containing both calreticulin and HLA-A2 heavy chains. Peak II corresponds to the migration position of calreticulin monomers. Peak III corresponds to the migration position of intact HLA-A2 heterodimers. Peak IV contains free β2m and is not shown in the gels. The results showing co-migration of calreticulin and HLA-A2 heavy chains are representative of three independent experiments.

We analyzed, by gel filtration chromatography on a Superose 6 column (Pharmacia), supernatants recovered after the thermal aggregation analyses shown in Fig. 3A. Gel-filtration analyses indicated that the presence of calreticulin did not significantly enhance HLA-A2 heterodimer (peak III) recovery on cooling of the samples; rather, heavy chains rescued from precipitation by calreticulin co-migrated with calreticulin as high-molecular-weight species (Fig. 3 B and C, peak I). Significantly, no β2m was recovered in peak I. Although these observations suggested the possibility that the co-migrating heavy chain and calreticulin were in a complex, such an interpretation was complicated by the observation that calreticulin itself was oligomerized on incubation at the higher temperatures (Fig. 3B Top, and see below). Coimmunoprecipitation analyses were carried out to further verify complex formation between HLA-A2 and CRT (Fig. 4). Samples containing HLA-A2 alone, calreticulin alone, or HLA-A2 and calreticulin were incubated at 50°C for 1 h. The mixtures were immunoprecipitated with an antibody directed against the HLA-A2 histidine tag, followed by SDS/PAGE analyses of coimmunoprecipitating proteins. These analyses indicated that calreticulin was immunoprecipitated with the anti-his antibody when it is preincubated with HLA-A2 at 50°C (Fig. 4, lane 5). However, complex formation was not observed when the proteins were preincubated at 37°C (Fig. 4, lane 7), suggesting that complex formation required temperature-induced structural changes in HLA-A2, calreticulin, or both proteins. The presence of LLDCPTAAV subsequent to thermal aggregation did not result in dissociation of the complexes formed between calreticulin and HLA-A2 (Fig. 4, lane 6).

Figure 4.

Complex formation between HLA-A2 and calreticulin assessed by coimmunoprecipitation analyses. HLA-A2 (12 μM), calreticulin (12 μM), or HLA-A2+calreticulin (12 μM each) were incubated at 50 or 37°C for 1 h as indicated, followed by immunoprecipitation with the anti-his antibody that recognizes the C-terminal histidine tag present on HLA-A2, or a control antibody (directed against the irrelevant AU5 epitope tag). In one experiment (lane 6), peptide (120 μM LLDCPTAAV) was added subsequent to the 50°C treatment, during the immunoprecipitation. Immunoprecipitated proteins were separated by SDS/PAGE and visualized by Coomassie blue staining. Results shown are representative of several independent experiments.

In the Absence of Other Components, Calreticulin Does Not Enhance the Extent of HLA-A2 Assembly with Peptide.

We measured HLA-A2 assembly with LLDC_FITCPTAAV at 37°C in the presence or absence of calreticulin by using the native gel-based assay. The binding analyses (Fig. 5 A and B) yielded apparent K_D values for the HLA-A2/LLDC_FITCPTAAV peptide interaction that are likely to be larger than the true K_D values, because some of the HLA-A2 is occupied with slow-exchanging high-affinity peptides. Analyses carried out under conditions of peptide excess (Fig. 5A) or HLA-A2 excess (Fig. 5B) yielded slightly different apparent K_D values. However, over the concentration ranges tested, the presence of calreticulin had no effect on the resulting fluorescence signal, indicating that calreticulin did not affect peptide binding or peptide exchange. These results are consistent with the expectation that the soluble heavy chain concentrations are similar in the presence or absence of calreticulin, because significant aggregation of heavy chain did not occur at 37°C (data not shown). We next investigated HLA-A2-LLDC_FITCPTAAV assembly after first inducing thermal aggregation in the presence or absence of calreticulin. Samples containing HLA-A2 alone or HLA-A2 and calreticulin were incubated at 47°C for 30 min followed by an additional incubation at 37°C with a 10-fold molar excess of LLDC_FITCPTAAV. Samples were centrifuged, and separated into pellet and supernatant fractions. The supernatant fractions were analyzed by both SDS/PAGE (Fig. 5C) and native-PAGE (Fig. 5D), and the pellet fractions analyzed by SDS/PAGE (Fig. 5C). The total amount of HLA-A2 heavy chains in the supernatant and pellet fractions were quantified from the SDS/PAGE gels, by SYPRO-Orange staining and fluorescence-imaging (Fig. 5C). HLA-A2-LLDC_FITCPTAAV complexes were quantified by fluorescence imaging of native gels (Fig. 5D).

Figure 5.

Calreticulin does not enhance peptide binding by HLA-A2. (A and B) HLA-A2 in the presence or absence of calreticulin (at a 1:1 ratio, and the indicated concentrations) was incubated at 37°C with LLDC_FITCPTAAV (120 μM) (A) or 80 nM (B). HLA-A2-LLDC_FITCPTAAV complexes were quantified following their separation by native-PAGE and fluorescence-imaging of gels. Data were analyzed using the prism software package (GraphPad, San Diego). The derived apparent K_D values for experiments in A were 6.8 ± 1.8 μM and 9.0 ± 2.μM and for experiments in B were 2.6 ± 1.3 and 1.5 ± 0.2 μM (in the absence and presence of calreticulin, respectively), and are based on two independent analyses each in duplicate. (C and D) HLA-A2 in the presence or absence of calreticulin (12 μM each protein) was incubated 47°C for 30 min. Following this incubation, peptide LLDC_FITCPTAAV (120 μM) was added and the samples further incubated at 37°C for 1 h. Samples (40 μl total) were centrifuged, separated into supernatant and pellet fraction, and the pellets resuspended in 40 μl of buffer. Twenty microliters of supernatants or pellets were separated by SDS/PAGE, stained with SYPRO-Orange, and HLA-A2 heavy chain quantified using a Fluorimager and imagequant analyses (C). Six or 12 μl of supernatants were separated by native-PAGE and HLA-A2-peptide complexes quantified as described above (D). Results shown in C and D are representative of several independent experiments.

The results from the SDS/PAGE analysis (Fig. 5C) resemble those shown in Fig. 3A. In the presence of calreticulin, less HLA-A2 heavy chain is recovered in the pellet fraction, and more heavy chain is recovered in the supernatant. Fluorescence images corresponding to HLA-A2-LLDC_FITCPTAAV complexes provided a measure of the extent of HLA-A2 assembly with peptide after thermal aggregation of the protein in the presence or absence of calreticulin. Soluble heavy chains present in the supernatant after thermal aggregation in the presence of calreticulin include heterodimeric HLA-A2 and protein that is in complex with calreticulin (Figs. 3 and 4). Free β2m (released from heavy chains that aggregate and associate with calreticulin; peak IV, Fig. 3B) is also present in the supernatant. If peptides could trigger dissociation of calreticulin–HLA-A2 complexes, enhanced assembly of LLDC_FITCPTAAV with heavy chains and β2m might be expected in samples that contained calreticulin compared with those that did not. Consistent with the result that the presence of LLDCPTAAV subsequent to thermal aggregation did not result in dissociation of complexes formed between calreticulin and HLA-A2 heavy chains (Fig. 4), an enhancement in the extent of assembled HLA-A2–LLDC_FITCPTAAV complexes was not observable in the presence of calreticulin (Fig. 5D). In fact, a small but reproducible reduction was observed (Fig. 5D). Thus, the increase in soluble heavy chain content in the supernatant in the presence of calreticulin does not result in an increase in assembly of HLA-A2 with peptide, correlating with the observation that peptides are not sufficient to trigger dissociation of calreticulin-HLA-A2 heavy chain complexes (Fig. 4).

Temperature-Dependent Conversion of Calreticulin to Active Oligomers.

Gel filtration (Fig. 3B) and native gel electrophoresis of calreticulin indicated the presence of oligomeric forms subsequent to protein exposure to 37°C or higher temperatures. To further investigate this phenomenon, purified calreticulin was incubated at 4, 37, or 45°C for 1 h and analyzed by gel filtration chromatography. Fig. 6A shows that in contrast to calreticulin that had been stored at 4°C, which eluted as a single species, calreticulin incubated at 37°C or higher temperatures eluted as one or more additional higher-molecular-weight species. To investigate the possibility that conversion of the calreticulin monomeric form to a higher-molecular-weight species at 45°C was the result of aggregation and inactivation of the protein, we analyzed whether the higher-molecular-weight species of calreticulin could inhibit thermal aggregation of HLA-A2. Oligomeric calreticulin was purified by a preparative-scale gel filtration chromatography on a Superose 6 column, after heating calreticulin to 45°C for 1 h. Protein corresponding to the higher-molecular-weight fractions were pooled, concentrated, and dialyzed into aggregation assay buffer. When analyzed by native-PAGE, >90% of the total protein was high-molecular-weight (HMW, Fig. 6B), and the proportion of monomers did not increase on prolonged storage (up to 3 weeks). Thermal aggregation assays of HLA-A2 were performed using oligomeric calreticulin (HMW) or calreticulin that had been stored at 4°C. As shown in Fig. 6C, oligomeric species of calreticulin are still able to inhibit precipitation of HLA-A2, and similar results were obtained using the spectrometric thermal aggregation assays (data not shown). At a 1:1 ratio of HLA-A2:calreticulin, calreticulin oligomers appear to be slightly less active than monomers in inhibiting thermal aggregation of HLA-A2 (Fig. 6C Lower, lane 6 compared with lane 7). However, calreticulin oligomers at a 1:1 ratio (HLA-A2:calreticulin) are significantly more active than monomers at a 1:0.5 ratio (HLA-A2:calreticulin) (Fig. 6C Lower, lane 7 compared with lane 4), demonstrating that the oligomeric calreticulin is not a heat-aggregated, inactive species.

Figure 6.

At physiological or higher temperatures, calreticulin forms oligomeric species, which are capable of inhibiting thermal aggregation of HLA-A2. Calreticulin (250 μg, 1.25 mg/ml) was incubated at 4, 37, or 45°C for 1 h and then centrifuged at 15,000 × g for 15 min. (A) Supernatants were analyzed by gel filtration chromatography. (B) Native-PAGE of calreticulin monomers and purified oligomers (HMW). (C) Inhibition of thermal aggregation of HLA-A2 by monomeric calreticulin, oligomeric calreticulin (HMW), or human IgG were compared using precipitation-based assays. HLA-A2 (12 μM) was incubated at 50°C for 1 h in the presence of calreticulin, oligomeric calreticulin (in the same microgram amount as the monomers), IgG, or in the absence of added proteins in the indicated molar ratios. Samples were centrifuged, separated into supernatant and pellet fractions. Pellets were resuspended in a volume equal to that of supernatants. Supernatants (S) and pellets (P) were analyzed by SDS/PAGE as indicated. These data are representative of three independent analyses.

Discussion

We used insect cell-expressed HLA-A2 as a source of soluble peptide-deficient MHC class I molecules for investigations of functional interactions with calreticulin. On incubation at 45–50°C, aggregation of the HLA-A2 could be visualized using a light scattering assay. Aggregation was inhibited by the presence of calreticulin, but not IgG, indicating a functional interaction between calreticulin and HLA-A2 (Fig. 2A). Preincubation with HLA-A2-specific peptides also inhibited thermal aggregation of HLA-A2 (Fig. 2B). These observations suggested that the aggregating species was either empty HLA-A2 or HLA-A2 bound to low affinity peptides. Thermal aggregation of HLA-A2 was also inhibited if peptides were added immediately before the start of the aggregation assay (Fig. 2C), indicating that peptides are able to rapidly interact with HLA-A2 molecules as the temperature is elevated to 45°C. This observation, taken together with the observation of enhanced binding of LLDC_FITCPTAAV to HLA-A2 at 37°C compared with 4°C (Fig. 1B) suggests that the rate-limiting step in peptide binding to the purified HLA-A2 might involve the dissociation of prebound suboptimal peptides. The question of whether peptides in general can inhibit aggregation of MHC class I molecules is likely to depend on the aggregation kinetics of the particular MHC molecule relative to the binding kinetics of the particular peptide–MHC complex.

Under the conditions described here, the HLA-A2 did not aggregate significantly at 37°C, and thus the observation of functional effects of calreticulin on HLA-A2 aggregation required higher temperatures. Within the environment of the ER membranes, it is possible to achieve very high local concentrations of proteins; thus, significant misfolding of newly synthesized (unfolded and partially folded) HLA-A2 heavy chains might occur in vivo even at 37°C. The HLA-A2 heavy chains used in our studies were also fully oxidized. It is likely that oxidative intermediates of heavy chains or heterodimers would display reduced stability at 37°C compared with the protein used in this study. The observations that ERp57 and calreticulin are both found associated with the MHC class I peptide loading complex (1) suggests that oxidative intermediates of the heavy chain could be substrates for calreticulin and ERp57 (1).

In mouse cells, mutation of MHC class I glycosylation motifs were found to destabilize class I-calreticulin interactions, arguing for a role for class I oligosaccharides in class I-calreticulin interaction (17). Experiments using the inhibitor castanospermine have also indicated a role for oligosaccharide-based interactions in the formation and dissociation of class I-calreticulin complexes from detergent lysates of human and murine cells (3, 4). In the present studies, we observe stable complexes between calreticulin and HLA-A2 as assessed by gel filtration chromatography and coimmunoprecipitation analyses (Figs. 3 and 4). However, we do not believe that the HLA-A2/calreticulin complexes we observe are oligosaccharide-dependent, because HLA-A2 purified from insect cell supernatants is not expected to contain monoglucosylated oligosaccharides that are recognized by calreticulin (monoglucosylated oligosaccharide structures are transient intermediates during ER-trafficking of proteins). Consistent with this expectation, we do not observe stable complexes between calreticulin and HLA-A2 when the two proteins are mixed at 37°C or lower temperatures. Rather, complex formation between calreticulin and HLA-A2 in our experiments required conditions that induced protein unfolding, results that suggest the importance of polypeptide-based binding in the interactions described here. Furthermore, the results shown here of the effects of calreticulin on HLA-A2 aggregation have been extendable to a nonclassical MHC class I molecule, HLA-E, expressed, purified, and reconstituted from Escherichia coli in the absence of peptide (L.M., S.M.R., and M.R., unpublished observations). Because proteins purified from E. coli are not glycosylated, these observations emphasized the role of polypeptide-based binding in functional interactions between calreticulin and MHC class I proteins. Taking these observations together, we suggest that both polypeptide- and oligosaccharide-based interactions are likely to contribute to binding between calreticulin and MHC class I heavy chains in vivo. Demonstration of this possibility awaits a better understanding of factors that can regulate calreticulin–substrate interactions in vivo.

The addition of peptide does not appear to be sufficient to dissociate preformed HLA-A2/calreticulin complexes, and to trigger assembly of heavy chain/β2m/peptide complexes (Figs. 4 and 5D). As with oligosaccharide-based interactions between calreticulin and substrate proteins, which cannot be dissociated if deglucosylation is inhibited (see, for example, ref. 18), dissociation of polypeptide-based interactions between calreticulin and substrate proteins might require accessory ER-resident factors, or other conditions not explored in our assays. Other classical chaperone systems have been shown to involve on-and-off cycles that depend on a set of co-chaperones and in some cases, ATP hydrolysis (see, for example, refs. 19 and 20). Similarly, it is likely that other ER co-chaperones exist that can refold calreticulin-associated heavy chains, and ultimately enhance the formation of class I-peptide complexes. The alternative model, whereby calreticulin-bound MHC class I heavy chains are destined for degradation, does not gain much support based on the existing literature on the effects of calnexin and calreticulin on the folding of various substrate proteins (see, for example, ref. 18), but remains a formal possibility.

A recent report by Bouvier et al. describes that, on heating, calreticulin (in the presence of calcium ions) undergoes a two-state structural transition with a T_m of about 45°C (21). This was suggested to correspond to an unfolding transition, which was surprising in light of our present observations, and previous reports that calreticulin could suppress thermal aggregation of substrate proteins at 45°C and higher temperatures. The interpretation of the CD-based thermal stability analyses might be complicated by the occurrence of calreticulin self-association at higher temperatures (Fig. 6). Our observations suggest that exposure to higher temperatures might alter the accessibility of oligomerization scaffolds within calreticulin, which is likely in turn to also be related to the recognition of misfolded segments in substrate polypeptides. Because oligomerization of calreticulin is induced even at 37°C, we suggest that oligomeric forms of calreticulin exist under physiological conditions. The acquisition of chaperone activity that is coupled to self-oligomerization has previously been described for other chaperones [for example, Hsp90 (22)], and a similar model might be applicable to calreticulin function.

We suggest that the unfolded polypeptide segments that are exposed on HLA heavy chains on incubation at higher temperature (Fig. 2) resemble unfolded segments present on newly synthesized heavy chains before the completion of their folding and assembly in the ER. By its ability to recognize misfolded polypeptide segments of incompletely and inappropriately assembled MHC class I heavy chains, calreticulin could play a role in the quality control of MHC class I molecule assembly in the ER. The overall mechanisms involved in the interactions between calreticulin and MHC class I polypeptides (Figs. 3 and 4) are likely to be similar to those used by calnexin for recognition of MHC class I heavy chains, and also similar to mechanisms used by both calreticulin and calnexin for recognition of other misfolded protein substrates (7, 8). However, it is possible that calreticulin and calnexin differ in their sequence specificities for polypeptides substrates. Such a difference could explain the observed preferential interaction of free MHC class I heavy chains with calnexin rather than calreticulin (reviewed in ref. 2). Alternatively, the observed differences in class I recognition by calnexin vs. calreticulin might arise because of the different topological environments of calnexin (membrane-associated) and calreticulin (soluble) in the ER, as previously suggested (9). The observations described in Fig. 3B indicate that calreticulin oligomers bind to the heavy chain of misfolded HLA-A2. The requirement for β2m to observe class I-calreticulin interaction in coimmunoprecipitation experiments from cell lysates (3, 4) might primarily arise because β2m association is required to trigger release of heavy chains from calnexin (23), rather than a requirement for β2m for the class I-calreticulin interaction per se.

Calreticulin oligomers, if existing in vivo, could recruit into a single complex various partially folded intermediates of MHC class I molecules, and MHC class I molecules that are in transient association with loading complex components. It will be of interest to examine direct polypeptide-based binding between calreticulin and tapasin, between calreticulin and the transporter associated with antigen processing (TAP) proteins, and between calreticulin and MHC class I molecules in the presence of tapasin and the TAP proteins. These studies could provide additional insights into mechanisms underlying the cooperative nature of the binding interactions within the loading complex (1).

Abbreviations

MHC

major histocompatibility complex

ER

endoplasmic reticulum

Note Added in Proof.

A related manuscript has recently been published (24) in which loading of MHC class I molecules with peptides was shown to be suboptimal in calreticulin-deficient cells, consistent with an ER quality control function for calreticulin in the MHC class I assembly pathway.