The assembly of MHC Ia molecules in the endoplasmic reticulum requires the presence of peptide ligands and β2m and is facilitated by chaperones in an ordered sequence of molecular interactions. A crucial step in this process is the interaction of the class I α-chain/β2m dimer with TAP, which is believed to ensure effective peptide loading of the empty class I molecule. We have previously demonstrated impaired intracellular transport of the class Ib molecule HLA-E in mouse myeloma cells cotransfected with the genes for HLA-E and human β2m, which is most likely attributable to inefficient intracellular peptide loading of the HLA-E molecule. We therefore analyzed the ability of HLA-E in the transfectant cell line to bind synthetic peptides by means of their ability to enhance cell surface expression of HLA-E. Peptide binding was confirmed by testing the effect on the thermostability of soluble empty HLA-E/human β2m dimers. Two viral peptides binding to HLA-E were thus identified, for which the exact positioning of the N terminus appeared critical for binding, whereas the contribution of the length of the C terminus seemed to be minor, allowing peptides as short as seven amino acids and up to 16 amino acids to exhibit considerable binding activity. Furthermore, we demonstrate that HLA-E interacts with TAP and that this interaction can be prolonged by the proteasome inhibitor N-acetyl-l-leucyl-l-leucyl-l-norleucinal, which reduces the intracellular peptide pool. The presented data indicate that HLA-E is capable of presenting peptide ligands similar to the repertoire of HLA class Ia molecules.

Major Histocompatibility Complex class Ia molecules present endogenously derived Ags to CTL (1). This function enables the recognition of virally infected or transformed cells as well as the selection of the T cell repertoire (2). Stable cell surface expression requires the assembly of a trimolecular complex composed of MHC class I heavy chain, β2m, and peptide within the endoplasmic recticulum (ER)4 (3). The peptide component, 8 to 10 amino acids in length, is usually generated in the cytosol by the action of proteasomes and is transported into the ER by a heterodimeric TAP (4, 5, 6, 7). A transient association of the nascent heavy chain with TAP is believed to enhance the efficiency of peptide loading (8, 9). In addition other chaperones (Ig binding protein (BIP), calreticulin, calnexin, and tapasin) were shown to be involved in MHC class I assembly, although their significance with regard to class I assembly has not yet been fully addressed (10, 11, 12).

Using the similarity of amino acid sequence to class Ia molecules, a second group of class Ib molecules can be defined that contains genes encoded within and outside the MHC and is heterogeneous with regard to its size in different species as well as to the tissue expression pattern and the degree of polymorphism of its members. The function of these class Ib molecules remains largely unknown, although a role for these molecules in Ag presentation is most likely (for review, see 13 . In the mouse the majority of these molecules are encoded in the Q, T, and M subregions of the MHC (14, 15). They comprise the Qa-2 molecules that bind nonameric peptides (16), the Qa-1 Ag that is capable of interacting with mycobacterial heat shock protein 65 (17), and a Glu50-Tyr50 copolymer (18), as well as the H-2 M3 Ag that selectively presents bacterial peptides with unique chemical properties (19, 20). In contrast, the thymus leukemia Ag is recognized by T cells expressing γδ TCRs in the absence of detectable bound peptides (21, 22). In man, class Ib molecules comprise the low polymorphic HLA-E, -F, -G, and CD1 Ags in addition to the highly polymorphic MIC-A and -B molecules (23). Of these, HLA-G was shown to present nonameric peptides, whereas CD1b has been implicated in the recognition of mycobacterial lipid and lipoglycan compounds by CD4CD8αβ+ T cells (24). HLA-E is the only class Ib gene that is transcribed in all human tissues and cell lines (25, 26), and we have previously reported that the HLA-E Ag is weakly expressed at the cell surface upon cotransfection with the gene for human β2m (hβ2m) into the mouse myeloma cell line P3X63Ag8.653 (X63) (26). In these transfectants the intracellular transport of the HLA-E molecule was impaired, most likely due to inefficient peptide loading within the ER (27). In this report we address the question of whether HLA-E binds peptides, in particular exclusively special peptides, possibly present only in low abundance within the cell. Moreover, we wanted to know whether the peptide binding groove of HLA-E is only accessible for exactly processed peptides that might reduce its efficient loading. Finally, we investigated whether a disturbed interaction of HLA-E with TAP might reduce its chance of encountering appropriate peptides.

Abs used for immunofluorescence and immunoprecipitation were mAb B9.12.1 (IgG2a) (28), which is specific for HLA class I complexed with β2m and which was used as a supernatant of the hybridoma; mAb W6/32 (IgG2a, American Type Culture Collection, Rockville, MD), which is specific for HLA class I complexed with β2m and was applied purified; mAb A1.4 (IgG2b; United Biomedical, Hauppauge, NY), which is specific for HLA class I heavy chains and was used as hybridoma supernatant; mAb 27-11-13s (IgG2a), which is specific for H-2Dd and Db (provided by F. Kievits, Amsterdam, The Netherlands) as ascites; and two antisera, D90, which is specific for rat TAP1 and cross-reactive with mouse TAP1, and 116/3, an antiserum specific for rat TAP2 and cross-reactive with mouse TAP2 (both provided by Dr. Jonathan Howard, Cologne, Germany). All the Abs were raised in mice, with the exception of D90 and 116/3, which are rabbit antisera. As secondary reagents, dichlorotriazinyl aminofluorescein-conjugated goat anti-mouse IgG F(ab′)2 (Jackson ImmunoResearch Laboratories, West Grove, PA) were used for immunofluorescence, and alkaline phosphatase-labeled goat anti-mouse Ig or biotin-labeled goat anti-rabbit Ig (both from Tropix, Bedford, MA) were applied for immunodetection of Western blots.

Peptides were either obtained RP-HPLC purified from A. McMichael (Oxford, U.K.), G. Pape (Munich, Germany), S. Modrow (Regensburg, Germany), B. Schoel (Ulm, Germany), P. Robinson (London, U.K.), H. Zweerink (Rahway, NJ), and S. Shawar (Houston, TX) or were synthesized using a 9050 PepSynthesizer (Milligen, Burlington, MA) and F-moc coupling technology as described previously (29). The latter were further purified by reverse phase chromatography on a PepRPC 15 μm HR 16/10 column (Pharmacia, Piscataway, NJ) using a Pharmacia fast protein liquid chromatography system. The peptides E-1L6 and E-2F6 were custom synthesized and RP-HPLC purified by Affiniti Research Products (Nottingham, U.K.).

The transfection of P3X63Ag8.653 (X63) with human β2m (X63M), human β2m and HLA-E*01033 (X63EM), or human β2m and HLA-B*2705 (X63BM) has been described previously (26). The RMA-S cell line was obtained from Dr. Marika Pla (Institut National de la Santé et de la Recherche Médicale (INSERM), Hôpital St. Louis, Paris, France). Cells were cotransfected by standard electroporation procedures (Gene Transfector 300, BTX, San Diego, CA) with 15 μg of a BamHI-linearized 15-kb BamHI/SalI subclone of the human β2m-gene in pUC19 (28) together with 5 μg of a SalI-linearized derivative of the COS-203 vector (30), in which the EBV sequences and COS site had been deleted by Eca digestion and which contained either no insert or a HindIII/BglII fragment of cosmid cd3.14 (26) encoding the HLA-E*01033 allele. Cells were maintained in 5% CO2 at 37°C in RPMI 1640 medium containing 2 mM l-glutamine (BioWhittaker, Walkersville, MD) and supplemented with 10% FCS (Serva, Heidelberg, Germany), 1 mM sodium pyruvate, 50 U/ml penicillin, and 50 μg/ml streptomycin (all from BioWhittaker). To select and grow the transfectant clones of the RMA-S and X63 cells, 0.4 mg/ml hygromycin B (Boehringer Mannheim, Indianapolis, IN) and 1 mg/ml G418 (Life Technologies, Gaithersburg, MD), respectively, were added to the culture medium.

For expression of soluble HLA-E molecules in Drosophila melanogaster cells, a truncated cDNA that encodes the signal peptide and α1-α3 domains of the HLA-E*0101 allele was generated by PCR and cloned into the pRMHa-3 expression vector (31). This construct was cotransfected with a hβ2m cDNA cloned into pRMHa-3 and the pUCshsneo vector (32) as a selection marker. Transfection of SC2 D. melanogaster tissue culture cells and selection of transformants have been described previously (33). Transfectants were maintained in Schneider medium (Life Technologies) supplemented with 10% FCS and 0.5 mg/ml G418 at room temperature. For the induction of soluble HLA-E/hβ2m dimer expression, cells were transferred into serum-free Insect Xpress medium (BioWhittaker) at a density of >107 cells/ml, and 1 mM CuSO4 was added for 48 h.

The supernatant of transfected SC2 D. melanogaster cells that were induced to express soluble HLA-E/hβ2m dimers was diluted 1/20 with PBS and either supplemented with Nonidet P-40 to 1% (w/v) or used directly. To 1 ml of this dilution either peptides (50 and 100 μM) or purified hβ2m (Serologic Reagents, East Grinstead, U.K.; 2 μM) were added. The solution was incubated overnight at 4°C and then heated to 32°C for 1 h. Then, 5 μg of mAb W6/32 was added for 40 min at 4°C, followed by a 20-min incubation at 4°C in the presence of 20 μl of swollen protein A-Sepharose CL-4B (Pharmacia). The immunoprecipitates were recovered by centrifugation, washed five times with 0.1% Nonidet P-40 in PBS, separated by 10% SDS-PAGE, and transferred onto nitrocellulose membranes (Schleicher and Schuell, Keene, NH) using standard Western blotting techniques. HLA-E heavy chains were visualized using the Western-Light (Tropix) chemiluminescent detection system according to the manufacturer’s instructions. In brief, blocked membranes were incubated for 1 h at room temperature in 12 ml of a 1/20 dilution of A1.4 hybridoma supernatant in 0.1% (v/v) Tween-20/PBS (TPBS), washed twice for 5 min each time at room temperature in TPBS, and then incubated in 14 ml of a 1/14,000 dilution of the secondary anti-mouse antiserum conjugated with alkaline phosphatase in TPBS for 45 min at room temperature. After washing three times for 10 min each time at room temperature in TPBS, membranes were processed as recommended using CSPD (Tropix, Bedford, MA) as the chemiluminescent substrate.

X63EM, X63BM, and X63M transfectant cells (7 × 107) were lysed for 30 min at 4°C in 0.6 ml of lysis buffer (1% (w/v) digitonin (Sigma), 0.1 mM PMSF, and 1% (v/v) Trasylol (Bayer, Pittsburgh, PA; equivalent to 14 μg/ml aprotinin) in PBS). After removal of cellular debris by centrifugation, 200 μl of B9.12.1 supernatant was added, followed by incubation for 40 min at 4°C. After addition of 20 μl of swollen protein A-Sepharose CL-4B (Pharmacia) and incubation for 40 min at 4°C, immunoprecipitates were recovered by centrifugation and washed five times with 0.1% digitonin in PBS. Two aliquots, each equivalent to 3 × 107 cells, and one aliquot equivalent to 1 × 107 cells were separated in different lanes by 10% SDS-PAGE and blotted onto nitrocellulose membranes (Schleicher and Schuell) by standard Western blotting techniques. Membranes with the transfer of 3 × 107 cell equivalent immunoprecipitates were incubated for 1 h at room temperature in 5 ml of TPBS with either the TAP1-specific antiserum D90 (1/1000) or the TAP2-specific antiserum 116/3 (1/2000). After washing twice for 5 min each time at room temperature with TPBS, the membranes were processed using Western-LightPlus (Tropix) chemiluminescent detection according to the manufacturer’s instructions. The resulting immune complexes were labeled with a biotinylated secondary Ab and streptavidin alkaline phosphatase and were visualized using CSPD as the chemiluminescent substrate. HLA class I heavy chains were visualized on membranes with the immunoprecipitates of 1 × 107 cells/lane using A1.4 as first Ab and the Western-Light (Tropix) chemiluminescent detection system as described above.

X63EM and X63BM transfectants were kept at 1.2 × 106/ml in methionine- and cysteine-free DMEM (BioWhittaker) for 1 h at 37°C and 5% CO2. During this period, X63EM transfectants were either treated with LLnL (Sigma) at 250 μM or left untreated. Thereafter, Trans35S-label (ICN, Costa Mesa, CA) was added to the cells at 100 μCi/ml. After 20 min of labeling, a 100-fold molar excess of cysteine and methionine was added; an aliquot of 1.2 × 106 cells was removed, washed, and placed on ice; and the remaining 1.2 × 106 cells were chased at 37°C for 90 min. The cells were lysed for 30 min at 4°C in 500 μl of PBS supplemented with 1% digitonin, 1 mM PMSF, and 1% (v/v) Trasylol. After removal of debris by centrifugation, lysates were precleared overnight at 4°C by adding 4 μl of a purified IgG2a mAb (ICN) and 20 μl of protein A-Sepharose. Precleared lysates were immunoprecipitated by the addition of 2 μl of the 116/3 antiserum followed by incubation for 40 min at 4°C and then addition of 30 μl of protein A-Sepharose followed by incubation for 20 min at 4°C. Immunoprecipitates were recovered by centrifugation, washed five times with 0.1% digitonin/PBS, resuspended in 30 μl of 2% SDS and 0.5 mM PMSF in PBS, and boiled for 5 min. After addition of 600 μl of PBS containing 1% Nonidet P-40, 1 mM PMSF, and 1% (v/v) Trasylol (Bayer), the solution was kept for 30 min at room temperature and for another 30 min at 4°C before the addition of 150 μl of A1.4 hybridoma supernatant and incubation for 45 min at 4°C. After an additional incubation for 20 min at 4°C with 20 μl of protein A-Sepharose, the immunoprecipitates were recovered by centrifugation and washed five times with 0.1% Nonidet P-40/PBS before analysis by 10% SDS-PAGE. Gels were treated with Amplify (Amersham, Arlington Heights, IL), dried, and fluorographed at −80°C.

X63 transfectant cells (1–2 × 105) were incubated in a 96-well microtiter plate well with peptides at 25, 50, or 100 μM for 6 h at 37°C in culture medium before immunofluorescent staining. Alternatively, subconfluent transfectants were incubated in culture medium for 4 h at room temperature. After two washes with RPMI 1640, 1 to 2 × 105 cells were incubated in a 96-well microtiter plate well with peptides at 100 μM for 1 h at room temperature in RPMI 1640. After incubation at 37°C for 1 h, cells were stained for flow cytometric analysis with mAbs B9.12.1 and A1.4 as described previously (27).

We have previously shown that acid-extracted and RP-HPLC-purified material derived from mouse myeloma X63 cells cotransfected with hβ2m and HLA-E (X63EM) was able to enhance the cell surface expression of HLA-E on these cells (27). Although this enhancement was most likely caused by peptides present in the extracted material, we were unable to determine either an HLA-E-specific peptide motif or a distinct length of peptides bound to HLA-E by pool sequencing peptide material eluted from HLA-E molecules purified from the X63EM transfectant cell line (unpublished data). We therefore tested 79 different synthetic peptides from various organisms (viral and bacterial proteins, human heat shock proteins, and HLA-A2 signal peptides, some partly modified at their N termini by formylation and acetylation; table available on request) to assess their abilities to enhance the HLA-E cell surface expression of X63EM transfectants after incubation of the cells with the peptides at 25 μM for 6 h at 37°C. The peptides used ranged from 7 to 24 amino acids in length and had partly overlapping sequences. Of these, two viral peptides, a 13 mer (BZLF-139-51, SQAPLPCVLWPVL) derived from the BZLF-1 protein of EBV and a 12 mer (InflM57–68K62, KGILGKVFTLTV) derived from the matrix protein of influenza A virus, in which wild-type F62 had been substituted for K, enhanced HLA-E cell surface expression on the X63EM transfectants, as detected with mAbs B9.12.1 (Fig. 1) and A1.4 (data not shown). None of the peptides enhanced staining with these mAbs of the negative control X63 cells expressing either HLA-B27 and hβ2m (X63BM) or hβ2m alone (X63M), even at 100 μM (data not shown). The effect of the peptides added at 25 μM on HLA-E cell surface expression was considerably less than that of exogenous hβ2m added at 12 μg/ml (∼1 μM), which served as a positive control and which we considered to produce the maximum increase in HLA-E cell surface expression achievable in this system. Moreover, BZLF-139–51 enhanced HLA-E cell surface expression slightly more than did InflM57–68K62. We therefore performed a titration experiment using increasing peptide concentrations. As shown in Figure 2, increasing peptide concentrations resulted in the same enhancement of HLA class I staining as that obtained upon stabilization with hβ2m. The slight differences in enhancement maxima were within the range of experimental variation. In this experiment, no increase in HLA-E cell surface expression with 25 μM InflM57–68K62 could be observed. BZLF-139–51 enhanced the HLA-E cell surface expression at lower concentrations than InflM57–68K62. This enhancement leveled off at 50 μM, whereas the maximum effect of InflM57–68K62 was reached only at 100 μM. No effect was observed when peptides were tested on X63BM or X63M cells (data not shown).

FIGURE 1.

Enhancement of cell surface expression of HLA-E class I molecules by peptides. X63EM cells were cultured for 6 h at 37°C in the presence or the absence of the peptides indicated at 25 μM or purified hβ2m at 12 μg/ml and subsequently stained with mAb B9.12.1, which is specific for HLA class I complexed with hβ2m. In each diagram the staining profiles obtained with or without the addition of either hβ2m or the peptide indicated are compared.

FIGURE 1.

Enhancement of cell surface expression of HLA-E class I molecules by peptides. X63EM cells were cultured for 6 h at 37°C in the presence or the absence of the peptides indicated at 25 μM or purified hβ2m at 12 μg/ml and subsequently stained with mAb B9.12.1, which is specific for HLA class I complexed with hβ2m. In each diagram the staining profiles obtained with or without the addition of either hβ2m or the peptide indicated are compared.

Close modal
FIGURE 2.

Titration of peptide-induced enhancement of HLA-E cell surface expression. X63EM cells were cultured for 6 h at 37°C in either the absence or the presence of the peptides at the concentrations indicated. As a control, cells were cultured in the presence of 12 μg/ml hβ2m. Cells were then stained for HLA class I with mAb B9.12.1. In the first three diagrams the staining profiles obtained with or without the addition of the three different peptides, 4/5 Gag, BZLF-139–51, and InflM57–68K62, at concentrations of 25, 50, and 100 μM are compared. In the fourth diagram the staining profile obtained after adding hβ2m at 12 μg/ml is compared with that of untreated cells.

FIGURE 2.

Titration of peptide-induced enhancement of HLA-E cell surface expression. X63EM cells were cultured for 6 h at 37°C in either the absence or the presence of the peptides at the concentrations indicated. As a control, cells were cultured in the presence of 12 μg/ml hβ2m. Cells were then stained for HLA class I with mAb B9.12.1. In the first three diagrams the staining profiles obtained with or without the addition of the three different peptides, 4/5 Gag, BZLF-139–51, and InflM57–68K62, at concentrations of 25, 50, and 100 μM are compared. In the fourth diagram the staining profile obtained after adding hβ2m at 12 μg/ml is compared with that of untreated cells.

Close modal

From these results we concluded that BZLF-139–51 binds with higher affinity to HLA-E than does InflM57–68K62. As HLA-E apparently is inefficiently loaded with peptides in the ER (27), we wanted to determine whether a structural basis, attributable to peptides binding to HLA-E, might account for this phenomenon. To reduce the possibility of extracellular peptide processing that might have taken place in the previous experiment, as described in other systems (34, 35), we set up an FCS-free assay with shorter incubation periods. Following preincubation of the X63EM cells for 4 h at room temperature, peptides were added at 100 μM for 1 h in medium without FCS, whereafter the cells were kept at 37°C for another hour before staining with mAbs B9.12.1 (Fig. 3) and A1.4 (data not shown). We thus tested whether the peptides were able to sustain the enhancement of HLA-E cell surface expression obtained during the preincubation of cells at room temperature. Peptides capable of stabilizing HLA-E cell surface expression were considered to bind to HLA-E. As negative controls, the same protocol was performed on X63M and X63BM cells, which did not result in enhanced B9.12.1 or A1.4 staining of either cell line with any of the peptides used (data not shown). Figure 3 shows exemplary results. A carboxyl-terminal truncation of the initially identified BZLF-139–51 peptide, resulting in the nonameric BZLF-139–47 peptide, stabilized HLA-E at least as well as BZLF-139–51 itself. In contrast, InflM57–68K62 could not maintain the enhanced HLA-E cell surface expression caused by the preincubation (data not shown), presumably due to the absence of extracellular processing, whereas a decamer (InflM59–68K62) and a nonamer (InflM59–67K62) obtained by N-terminal truncation of InflM57–68K62 could stabilize HLA-E. A further single amino acid truncation of the InflM59–68K62 peptide at the N terminus, giving InflM60–68K62, abrogated its stabilizing effect on HLA-E. The results of the assay are summarized qualitatively in Table I. The exact positioning of the peptides’ N termini appears to be critical for their binding to HLA-E. Extension or truncation of the N terminus by a single amino acid absolutely abrogated peptide binding to HLA-E in the temperature shift assay, although some of the peptides (BZLF-132–46, BZLF-134–46, and InflM57–68K62) containing an extended N terminus enhanced HLA-E cell surface expression when X63EM cells were incubated with them for 6 h at 37°C in the presence of FCS. Thus, it is possible that appropriate N termini of these peptides were generated during the incubation by extracellular processing. In contrast, in the thermal shift assay, C-terminal length variations of the peptides were tolerated without considerable alteration to the peptides’ stabilizing effect on HLA-E (Fig. 4), although maximum stabilization was achieved with the 9 mer BZLF-139–47. Only when the 7 mer, BZLF-139–45, was truncated to give the 6 mer, BZLF-139–44, was the stabilizing ability of the peptide almost undetectable (although it was still visible in experiments in which the maximum enhancement of HLA-E cell surface expression reached higher levels than those shown in Fig. 4). The only internal substitution abrogating the peptide’s effect on HLA-E that we identified to date exchanged P at amino acid position 4 of BZLF-139–47 for a Q. Since this position tolerates amino acids as different as K and F in the InflM peptide without affecting binding activity to HLA-E, it seems unlikely that position 4 is a dominant anchor for HLA-E binding peptides. Substitution of small hydrophobic amino acids at the C terminus of the peptides by K did not affect peptide binding. In the case of BZLF-140–47 and InflM59–68, these C-terminal substitutions with the hydrophilic K rendered the peptides soluble in water, which were otherwise only soluble in DMSO, and thus increased their effective concentrations in the assay. Since InflM59–68K68 stabilizes HLA-E, the lack of detectable binding of InflM59–68 is most likely attributable to its insolubility in water. In contrast, BZLF-140–47 does not bind to HLA-E, even when its solubility in water is increased by C-terminal K substitution.

FIGURE 3.

Influence of peptides on the thermostability of HLA-E molecules at the cell surface. X63EM cells were preincubated at room temperature for 4 h. Cells were then cultured at room temperature for 1 h in the presence or the absence of either peptides (as indicated) at 100 μM or hβ2m at 12 μg/ml, in medium without FCS. Cells were subsequently incubated at 37°C for another hour and then stained with mAb B9.12.1. An equal aliquot of cells in medium without FCS was kept at room temperature. The diagram on the left compares staining profiles obtained when cells were shifted to 37°C with (RT→37°C + 12 μg/ml hβ2m) or without (RT→37°C) the addition of hβ2m, with those resulting when cells were kept at room temperature (RT). In the other diagrams, three B9.12.1 staining profiles are compared: that of cells shifted to 37°C without addition of peptides or hβ2m (RT→37°C), that of cells shifted to 37°C in the presence of peptide at 100 μM (RT→37°C + 100 μM peptide), and that of cells following the addition of hβ2m (RT→37°C + 12 μg/ml hβ2m). In the second row, results obtained with two different peptides (BZLF-139–51 and BZLF-139–47) derived from BZLF-1 are shown. In the third row, results obtained with three different peptides (InflM59–68K62, InflM59–67K62, and InflM60–68K62) present in the InflM protein are given. In the case of the InflM peptides, F at position 62 of the protein is substituted for K to allow for solubility in physiologic buffers. The staining of the isotype control is shown as a dotted line in each diagram.

FIGURE 3.

Influence of peptides on the thermostability of HLA-E molecules at the cell surface. X63EM cells were preincubated at room temperature for 4 h. Cells were then cultured at room temperature for 1 h in the presence or the absence of either peptides (as indicated) at 100 μM or hβ2m at 12 μg/ml, in medium without FCS. Cells were subsequently incubated at 37°C for another hour and then stained with mAb B9.12.1. An equal aliquot of cells in medium without FCS was kept at room temperature. The diagram on the left compares staining profiles obtained when cells were shifted to 37°C with (RT→37°C + 12 μg/ml hβ2m) or without (RT→37°C) the addition of hβ2m, with those resulting when cells were kept at room temperature (RT). In the other diagrams, three B9.12.1 staining profiles are compared: that of cells shifted to 37°C without addition of peptides or hβ2m (RT→37°C), that of cells shifted to 37°C in the presence of peptide at 100 μM (RT→37°C + 100 μM peptide), and that of cells following the addition of hβ2m (RT→37°C + 12 μg/ml hβ2m). In the second row, results obtained with two different peptides (BZLF-139–51 and BZLF-139–47) derived from BZLF-1 are shown. In the third row, results obtained with three different peptides (InflM59–68K62, InflM59–67K62, and InflM60–68K62) present in the InflM protein are given. In the case of the InflM peptides, F at position 62 of the protein is substituted for K to allow for solubility in physiologic buffers. The staining of the isotype control is shown as a dotted line in each diagram.

Close modal
Table I.

Summary of the peptide binding results obtained by the thermoshift assay with X63EM cellsa

PeptideSequenceLengthBinding to HLA-ESoluble in
BZLF-132–46 YQDLGGPSQAPLPCV 15 mer − H2
BZLF-138–46 PSQAPLPCV 9 mer − H2
BZLF-140–47 QAPLPCVL 8 mer − DMSO 
BZLF-140–47K47 QAPLPCVK 8 mer − H2
BZLF-139–51 SQAPLPCVLWPVL 13 mer H2
BZLF-139–48 SQAPLPCVLW 10 mer H2
BZLF-139–47Q42 SQAQLPCVL 9 mer − H2
BZLF-139–47K47 SQAPLPCVK 9 mer H2
BZLF-139–47K46 SQAPLPCKL 9 mer H2
BZLF-139–47 SQAPLPCVL 9 mer H2
BZLF-139–45 SQAPLPC 7 mer H2
BZLF-139–44 SQAPLP 6 mer − H2
BZLF-139–46 SQAPLPCV 8 mer H2
InflM59–66K62 ILGKVFTL 8 mer H2
InflM59–67K62 ILGKVFTLT 9 mer H2
InflM59–68K62 ILGKVFTLTV 10 mer H2
InflM59–68K62K68 ILGKVFTLTK 10 mer H2
InflM59–68K68 ILGFVFTLTK 10 mer H2
InflM59–68 ILGFVFTLTV 10 mer − DMSO 
InflM60–68K62 LGKVFTLTV 9 mer − H2 
InflM58–67K62 GILGKVFTLT 10 mer − H2
InflM58–66K62 GILGKVFTL 9 mer − H2
InflM57–66K62 KGILGKVFTL 10 mer − H2
InflM57–68K62 KGILGKVFTLTV 12 mer − H2
PeptideSequenceLengthBinding to HLA-ESoluble in
BZLF-132–46 YQDLGGPSQAPLPCV 15 mer − H2
BZLF-138–46 PSQAPLPCV 9 mer − H2
BZLF-140–47 QAPLPCVL 8 mer − DMSO 
BZLF-140–47K47 QAPLPCVK 8 mer − H2
BZLF-139–51 SQAPLPCVLWPVL 13 mer H2
BZLF-139–48 SQAPLPCVLW 10 mer H2
BZLF-139–47Q42 SQAQLPCVL 9 mer − H2
BZLF-139–47K47 SQAPLPCVK 9 mer H2
BZLF-139–47K46 SQAPLPCKL 9 mer H2
BZLF-139–47 SQAPLPCVL 9 mer H2
BZLF-139–45 SQAPLPC 7 mer H2
BZLF-139–44 SQAPLP 6 mer − H2
BZLF-139–46 SQAPLPCV 8 mer H2
InflM59–66K62 ILGKVFTL 8 mer H2
InflM59–67K62 ILGKVFTLT 9 mer H2
InflM59–68K62 ILGKVFTLTV 10 mer H2
InflM59–68K62K68 ILGKVFTLTK 10 mer H2
InflM59–68K68 ILGFVFTLTK 10 mer H2
InflM59–68 ILGFVFTLTV 10 mer − DMSO 
InflM60–68K62 LGKVFTLTV 9 mer − H2 
InflM58–67K62 GILGKVFTLT 10 mer − H2
InflM58–66K62 GILGKVFTL 9 mer − H2
InflM57–66K62 KGILGKVFTL 10 mer − H2
InflM57–68K62 KGILGKVFTLTV 12 mer − H2
a

Binding to HLA-E is indicated by +, whereas − corresponds to no detectable binding activity. Substitutions of the proteins’ wild type sequences are given in bold type. The alignment of the shortest InflM peptide, tested for binding to HLA-E with a BZLF-1 peptide of identical length, is boxed.

FIGURE 4.

Influence of the carboxyl-terminal peptide sequence on binding to HLA-E. A thermoshift assay, outlined in Figure 3, was used to test the binding abilities of six BZLF-1 peptides with identical N termini but different lengths. The black and gray bars show the mean channel of immunofluorescence, as detected by staining for HLA class I with mAb B9.12.1 in flow cytometric analysis of X63EM cells, shifted after preincubation at room temperature to 37°C for 1 h in the presence of the peptides indicated. Each peptide analysis was performed in triplicate, represented by individual bars. The hatched bars are the control experiments, i.e., immunofluorescence of X63EM cells kept at room temperature and then shifted to 37°C for 1 h in the absence of peptides (37°C) and immunofluorescence of X63EM cells kept at room temperature and then shifted to 37°C for 1 h in the presence of 12 μg/ml hβ2m. The sequences and lengths of the peptides are given on the right of the diagram. Their names and positions within the BZLF-1 protein are given on the left.

FIGURE 4.

Influence of the carboxyl-terminal peptide sequence on binding to HLA-E. A thermoshift assay, outlined in Figure 3, was used to test the binding abilities of six BZLF-1 peptides with identical N termini but different lengths. The black and gray bars show the mean channel of immunofluorescence, as detected by staining for HLA class I with mAb B9.12.1 in flow cytometric analysis of X63EM cells, shifted after preincubation at room temperature to 37°C for 1 h in the presence of the peptides indicated. Each peptide analysis was performed in triplicate, represented by individual bars. The hatched bars are the control experiments, i.e., immunofluorescence of X63EM cells kept at room temperature and then shifted to 37°C for 1 h in the absence of peptides (37°C) and immunofluorescence of X63EM cells kept at room temperature and then shifted to 37°C for 1 h in the presence of 12 μg/ml hβ2m. The sequences and lengths of the peptides are given on the right of the diagram. Their names and positions within the BZLF-1 protein are given on the left.

Close modal

From the temperature shift assay it appeared that the exact positioning of the peptides’ N termini was critical for their binding to HLA-E. To test whether this might play a role in limiting the availability of HLA-E peptide ligands within the ER or, rather, reflect restraints imposed by the HLA-E conformation present on the X63EM cell surface, we applied another thermal shift assay using soluble empty HLA-E/hβ2m dimers expressed in D. melanogaster cells as previously described (33). We confirmed that the HLA-E*01032 allele expressed in these cells encoded the same polypeptide as the HLA-E*01033 allele transfected into the X63EM cells. The HLA-E/hβ2m dimers were present in the supernatant of induced SC2 transformants and remained stable at 4°C in the diluted supernatant even in the presence of 1% Nonidet P-40, as detectable by immunoprecipitation with mAb W6/32 and subsequent immunoblotting (Fig. 5, lane 9). When 1 ml of this diluted supernatant was heated to 32°C for 1 h, the amount of HLA-E/hβ2m dimers detectable by W6/32 immunoprecipitation and immunoblotting was greatly diminished (Fig. 5, lane 1). In the presence of 2 μM hβ2m, the immunoprecipitated amounts of HLA-E/hβ2m dimers equaled those in the positive control experiment (Fig. 5, lane 8). The two peptides, InflM57–68K62 and BZLF-139–51, were able to stabilize the HLA-E/hβ2m dimers in a similar manner, although less effectively than hβ2m, as can be deduced from the weaker intensity of the HLA-Esol α-chain-specific bands of the immunoblot (Fig. 5, lanes 2–5). The 13 mer gag47–59 peptide that was negative with regard to enhancement of HLA-E cell surface expression of X63EM cells in the previous experiments did not increase the thermostability of the HLA-E/hβ2m complexes (Fig. 5, lanes 6 and 7). There was no apparent difference between the abilities of InflM57–68K62 and BZLF-139–51 to stabilize HLA-E/hβ2m dimers in this assay, although only the latter was positive in the thermal shift assay using X63EM cells. Thus, in the presence of Nonidet P-40, the conformation of HLA-E does not select for an optimal trimming of the N terminus of peptides for binding.

FIGURE 5.

Influence of peptides on the thermostability of soluble empty HLA-E/hβ2m dimers. The supernatant of D. melanogaster SC2 transfectants expressing soluble empty HLA-E/hβ2m dimers was diluted 1/20 in PBS and supplemented with 1% (w/v) Nonidet P-40. One-milliliter aliquots of these dilutions were supplemented with either peptides or hβ2m at the concentrations indicated or with an appropriate volume of PBS, and then incubated overnight at 4°C followed by heating of the reactions to 32°C for 1 h. After subsequent immunoprecipitation of HLA-E/hβ2m dimers with mAb W6/32, immunoprecipitates were separated by 10% SDS-PAGE and blotted onto nitrocellulose membranes. HLA class I α-chains were visualized using mAb A1.4 and the Western-Light chemiluminescent detection system (Tropix). In a control experiment the temperature shift was omitted, and the reaction was kept at 4°C for the same time period. The positions of HLA-E α-chains and W6/3 light chains, which appear as doublets, are indicated. In lane 3 the signal intensity of the HLA-E α-chain must be corrected for the reduced amount of precipitate loaded, as can be estimated from the weaker signal for W6/32 light chains.

FIGURE 5.

Influence of peptides on the thermostability of soluble empty HLA-E/hβ2m dimers. The supernatant of D. melanogaster SC2 transfectants expressing soluble empty HLA-E/hβ2m dimers was diluted 1/20 in PBS and supplemented with 1% (w/v) Nonidet P-40. One-milliliter aliquots of these dilutions were supplemented with either peptides or hβ2m at the concentrations indicated or with an appropriate volume of PBS, and then incubated overnight at 4°C followed by heating of the reactions to 32°C for 1 h. After subsequent immunoprecipitation of HLA-E/hβ2m dimers with mAb W6/32, immunoprecipitates were separated by 10% SDS-PAGE and blotted onto nitrocellulose membranes. HLA class I α-chains were visualized using mAb A1.4 and the Western-Light chemiluminescent detection system (Tropix). In a control experiment the temperature shift was omitted, and the reaction was kept at 4°C for the same time period. The positions of HLA-E α-chains and W6/3 light chains, which appear as doublets, are indicated. In lane 3 the signal intensity of the HLA-E α-chain must be corrected for the reduced amount of precipitate loaded, as can be estimated from the weaker signal for W6/32 light chains.

Close modal

Since HLA-E seems to bind peptides much like other MHC class Ia molecules without detectable constraints regarding peptide sequence or length, we addressed the question of whether HLA-E associates with TAP in X63EM cells. Such an interaction of MHC class Ia molecules with TAP has been implicated in effective loading of these molecules with peptides, a prerequisite for further trafficking of the molecules to the cell surface (8, 9). Furthermore, this interaction seems to be the final step in a series of chaperoning events that depend upon each other and are required for the cell surface expression of MHC class Ia molecules (12). Following immunoprecipitation of HLA class I molecules from digitonin lysates of X63EM and X63BM cells with B9.12.1, associated TAP1 and TAP2 molecules were detected by immunoblotting (Fig. 6). Although detection with A1.4 showed a far lower abundance of HLA α-chains in X63EM cells than in X63BM cells (Fig. 6, upper panel), TAP1 and TAP2 were only detectable in association with HLA class I molecules in X63EM cells (Fig. 6, middle and lower panels). Only after prolonged exposure were TAP1 and TAP2 molecules associated with HLA-B27 detectable (not shown), whereas the lane of X63M cells remained negative for TAP1 or TAP2 signals. Therefore, under steady state conditions HLA-E is complexed with TAP to a much greater extent than HLA-B27. TAP1 and TAP2 are both required to form a functional peptide transporter and are both associated with HLA-E. Unfortunately, the experimental design did not allow us to test for possible disturbances in the stoichiometry of the complexes, which might affect the function of the peptide transporter itself.

FIGURE 6.

Association of HLA-E with TAP. After preparation of digitonin lysates of X63EM, X63M, and X63BM cells, HLA class I molecules were immunoprecipitated with mAb B9.12.1. Aliquots of the immunoprecipitates were separated on different 10% SDS-gels and blotted onto nitrocellulose membranes. Mouse TAP1 was detected with antiserum D90 (middle panel). For the detection of mouse TAP2, antiserum 116/3 was used (bottom panel), and mAb A1.4 was used to visualize HLA class I α-chains (top panel). Binding of D90 and 116/3 was detected using the Western-Light Plus chemiluminescent detection system (Tropix). For the visualization of A1.4, the binding Western-Light chemiluminescent detection system was used. In each panel the protein of interest is indicated. In the top panel the B9.12.1 heavy chain is also visualized, whereas in the middle and bottom panels the B9.12.1 light chains are detected as doublets. The intensity of these B9.12.1 chains indicates the amount of immunoprecipitate loaded. Hence, in the top panel less immunoprecipitate of the X63M cells was loaded, whereas in the middle and bottom panels equal amounts of immunoprecipitates of the three cells were loaded. In lane 1 of the middle panel a second fainter band is visible above TAP1. The nature of this upper band is unknown.

FIGURE 6.

Association of HLA-E with TAP. After preparation of digitonin lysates of X63EM, X63M, and X63BM cells, HLA class I molecules were immunoprecipitated with mAb B9.12.1. Aliquots of the immunoprecipitates were separated on different 10% SDS-gels and blotted onto nitrocellulose membranes. Mouse TAP1 was detected with antiserum D90 (middle panel). For the detection of mouse TAP2, antiserum 116/3 was used (bottom panel), and mAb A1.4 was used to visualize HLA class I α-chains (top panel). Binding of D90 and 116/3 was detected using the Western-Light Plus chemiluminescent detection system (Tropix). For the visualization of A1.4, the binding Western-Light chemiluminescent detection system was used. In each panel the protein of interest is indicated. In the top panel the B9.12.1 heavy chain is also visualized, whereas in the middle and bottom panels the B9.12.1 light chains are detected as doublets. The intensity of these B9.12.1 chains indicates the amount of immunoprecipitate loaded. Hence, in the top panel less immunoprecipitate of the X63M cells was loaded, whereas in the middle and bottom panels equal amounts of immunoprecipitates of the three cells were loaded. In lane 1 of the middle panel a second fainter band is visible above TAP1. The nature of this upper band is unknown.

Close modal

The association of MHC class I molecules with TAP is lost when peptides bind (8, 9), while inhibition of proteasome activity by LLnL, which decreases the pool of peptides available for class I molecules, prolongs their association with TAP (36). We therefore performed pulse-chase experiments to investigate the time course of the association of HLA-E with TAP. After a 20-min metabolic pulse label with Trans35S-label, X63EM and X63BM cells were chased for the times indicated (Fig. 7). Immediately after the labeling, HLA-B27 Ag associated with TAP was detected; this association was lost entirely after a 90-min chase. HLA-E α-chains, coprecipitating with TAP, gave a very faint signal at time zero. However, this did not reflect a weak association of HLA-E with TAP, but was due to a weak labeling efficiency of HLA-E in the X63EM cells (27). After a 90-min chase, no TAP-associated HLA-E α-chains were detectable. The large differences in the intensities of TAP-associated HLA α-chains between X63EM and X63BM cells did not allow us to make assumptions about differences in the association kinetics of the class I molecules with TAP. In particular, it was not possible to demonstrate a prolonged association of HLA-E with TAP that might explain the differences in TAP association between HLA-E and HLA-B27 in the steady state. To determine whether TAP association of HLA-E is dependent on peptides and thus whether HLA-E is loaded with peptides in the ER, we performed a parallel experiment in which proteasomes were inhibited by the addition of 250 μM LLnL (Fig. 7, lanes 1 and 2). In the presence of this inhibitor, the absolute amount of HLA-E associated with TAP increased, and the signals from TAP-associated HLA-E α-chains showed identical intensities after 0- and 90-min chase, the latter pointing to a peptide dependency for HLA-E’s TAP association and indirectly to the peptide loading of HLA-E in the ER.

FIGURE 7.

Inhibition of HLA-E dissociation from TAP by LLnL. X63EM cells were preincubated for 1 h in the presence or the absence of 250 μM LLnL. These cells as well as X63BM cells were metabolically pulsed with Trans35S-label (ICN) in the presence or the absence of 250 μM LLnL and either lysed immediately or chased for 90 min as indicated. From the digitonin lysates, TAP was precipitated with the TAP2-specific antiserum 116/3. After dissociation of the precipitates by boiling in the presence of 2% SDS, HLA class I molecules were reprecipitated with mAb A1.4 and analyzed by 10% SDS-PAGE.

FIGURE 7.

Inhibition of HLA-E dissociation from TAP by LLnL. X63EM cells were preincubated for 1 h in the presence or the absence of 250 μM LLnL. These cells as well as X63BM cells were metabolically pulsed with Trans35S-label (ICN) in the presence or the absence of 250 μM LLnL and either lysed immediately or chased for 90 min as indicated. From the digitonin lysates, TAP was precipitated with the TAP2-specific antiserum 116/3. After dissociation of the precipitates by boiling in the presence of 2% SDS, HLA class I molecules were reprecipitated with mAb A1.4 and analyzed by 10% SDS-PAGE.

Close modal

To assess the importance of TAP for the cell surface expression of HLA-E, we generated transfectants in the TAP2-deficient mouse cell line RMA-S. In Figure 8 we compare the class I cell surface expression pattern of transfectant clones with the cell lines RMA-S and X63, which were shown to contain equal amounts of HLA-E and hβ2m transcripts by Northern blot analysis. The RMA-S cells expressing hβ2m and HLA-E (RMA-SEM) stained slightly less brightly for the HLA-E Ag than control X63EM cells (histogram shaded in gray). Addition of either exogenous hβ2m (thin lines) or peptide (BZLF-139–47) enhanced cell surface expression of HLA-E on both cells to a similar extent. This increase is low compared with the stabilization of H-2Db expressed by RMA-S cells by adding a Db-specific NP2 peptide (ASNENMETM). This result shows that HLA-E molecules can be expressed on the cell surface in a TAP-independent manner.

FIGURE 8.

Influence of BZLF-139–47 and hβ2m on HLA-E cell surface expression of transfected X63 and RMA-S cells. X63EM, RMA-SEM, and RMA-SM cells were cultured for 6 h at 37°C in the presence or the absence of 100 μM BZLF-139–47 (specific for HLA-E) or NP2 (specific for H-2Db) or with purified 12 μg/ml hβ2m and subsequently stained with the mAbs B9.12.1 (HLA-E) or 27-11-13s (H-2Db and H-2Dd). In each diagram the specific staining of untreated cells (shaded in gray) is compared with that of cells incubated with peptides (thick lines) or hβ2m (thin lines) and with the isotype control staining (dotted lines).

FIGURE 8.

Influence of BZLF-139–47 and hβ2m on HLA-E cell surface expression of transfected X63 and RMA-S cells. X63EM, RMA-SEM, and RMA-SM cells were cultured for 6 h at 37°C in the presence or the absence of 100 μM BZLF-139–47 (specific for HLA-E) or NP2 (specific for H-2Db) or with purified 12 μg/ml hβ2m and subsequently stained with the mAbs B9.12.1 (HLA-E) or 27-11-13s (H-2Db and H-2Dd). In each diagram the specific staining of untreated cells (shaded in gray) is compared with that of cells incubated with peptides (thick lines) or hβ2m (thin lines) and with the isotype control staining (dotted lines).

Close modal

We have previously reported the impaired intracellular transport of HLA-E in mouse X63 cells cotransfected with the genes for HLA-E and hβ2m (27). In this study we identified peptides capable of binding to HLA-E and investigated the distinct peptide features that are required for binding, such as length and sequence. Moreover, we addressed the questions of whether HLA-E interacts with the cellular peptide loading machinery, namely TAP, and whether HLA-E presents endogenous peptides.

Of 79 partially overlapping synthetic peptides, two viral peptides were demonstrated to bind to HLA-E. These peptides increased the HLA-E cell surface expression of X63EM cells at 37°C as well as stabilized the enhancement of HLA-E cell surface expression at 37°C that was generated by incubating X63EM cells at room temperature (Figs. 1 and 3). Furthermore, these modified peptides increased the thermostability of soluble HLA-E/hβ2m dimers (Fig. 5). For screening we also included peptides with N-terminal modifications, in particular N-formylation, which is a major structural requirement for binding to H-2 M3 (37). Since the low availability of these ligands may limit the expression of H-2 M3 (38), we wanted to test whether a similar mechanism operates on the expression of HLA-E. None of these modified peptides showed any detectable binding activity to HLA-E. As the N-formylated peptides display a low solubility in water, it is possible that their effective concentrations in the assay were too low to detect binding. However, this seems unlikely, since in a similar assay Vyas et al. (38) could demonstrate that f-Bla-z at concentrations as low as 5 μM significantly increased the cell surface expression of M3a-Ld chimeric molecules. Moreover, the binding of the N terminus of N-formylated peptides to H-2 M3 is coordinated by residues in and around the B pocket of the class I molecule, since its A pocket is occluded by L and F replacing the consensus T-167 and W-171, respectively (39). In HLA-E the latter positions are identical with the consensus. Consequently, it does not meet the structural requirements to bind formylated peptides.

We also tested peptides from heat shock proteins, since, for Qa-1, induction of cell surface expression by either heat shock or incubation with a tryptic digest of Mycobacterium bovis hsp65 has been reported (17). Thus, at least some nonpolymorphic MHC class I molecules are conceptually appreciated as potential components of a first-line immune defense (40). Of 17 peptides representing different hsp65 sequences, none had an influence on HLA-E cell surface expression or had a tryptic digest of purified mycobacterial hsp60 (data not shown). Moreover, heat shock treatment of the X63EM cells did not alter the level of expression of HLA-E at the cell surface, although heat shock has been shown to up-regulate Qa-1 expression (17).

With regard to the peptide binding characteristics of HLA-E, it appears that an exact positioning of the peptide’s N terminus is crucial for a stable interaction with HLA-E. This feature was observed when peptides were tested for the ability to stabilize an increase in HLA-E cell surface expression of X63EM cells, obtained by preincubation of the cells at room temperature, but was undetectable when peptides were tested for their effects on the thermostability of soluble HLA-E/hβ2m dimers in the presence of Nonidet P-40. Peptides with extended N termini also stabilized HLA-E on X63EM cells at 37°C in the presence of FCS. One possible explanation might be that longer peptides are cleaved by proteases present in FCS to produce an N terminus suitable for binding, although how such a processing mechanism could generate the exact peptide length in solution is unclear. Therefore, it is not possible to judge whether the conformation of the peptide binding cleft of HLA-E present at the cell surface of X63EM cells at room temperature is a particularly closed one that can only be accessed by peptides with an appropriate N terminus or whether the addition of Nonidet P-40 to empty HLA-E/hβ2m dimers as well as culturing X63EM cells at a physiologic temperature induce an open conformation of the peptide binding groove of HLA-E that is accessible for peptides that protrude at either end. The second explanation is supported by the finding that addition of Nonidet P-40 to soluble empty HLA-E/hβ2m dimers in solution reduces their thermostability (data not shown). Furthermore, it has been shown that class I peptide dissociation kinetics vary according to the presence or the absence of detergent in the assay (33). In contrast to other MHC class I molecules for which the stability of the complex has been shown to depend on the peptides’ N and C termini (41), in our assay systems the C terminus of the peptide is of only minor importance for the binding to HLA-E. Only when the peptide is as short as six amino acids is its binding activity considerably diminished. The possibility that the stabilizing effect of longer peptides on HLA-E at the cell surface of X63EM cells is due to the presence of shortened peptides generated by proteolysis is unlikely, since binding of BZLF-1 peptide activity gradually declines with further C-terminal truncations of the 9 mer in a serum-free assay, and its 13 mer stabilizes HLA-E more effectively than the 10 mer (Fig. 4). Matsumura et al. (33) also report for H-2Kb that the exact positioning of the N-terminal amino acid of a peptide contributes more to the binding affinity than does correct positioning of the C-terminal amino acid of the peptide. Computer modelling of HLA-E predicts an enlarged F pocket, resulting from the replacement of both T-143 and W-147 by S residues (42). It should therefore be possible for a peptide to extend the F pocket of HLA-E. For HLA-A2, the stability of a class I-peptide complex with the peptide’s C terminus extending the binding cleft has been shown (43).

Recently, it has been demonstrated that HLA-E can bind nonamer peptides contained in the leader sequences of MHC class I molecules by testing the ability of peptides to increase the thermoresistance of W6/32-precipitable endogenous HLA-E Ag present in lysates of the B-LCL .221 (44). By genetic and 1D-IEF typing we found that this cell line expresses the HLA-E*0101 allele, which differs from the alleles (E*01031 and E*01033) we used in our study only at amino acid position 107 (R in E*0101 instead of G in E*01031 and E*01033). Braud et al. (44) identified positions 2 and 9 as primary anchors occupied by L or M and L, respectively, and position 7 as an auxiliary anchor. The minimum sequences of the different peptides identified as binding to HLA-E in this study and in the report by Braud et al. are aligned in Table II. In our experiments position 9 did not appear to be important, since there was no considerable difference in binding activity between the 9 mer BZLF-139–47 and the 7 mer BZLF-139–45 (Fig. 4) or between the 9 mer InflM59–67K62 and the 8 mer InflM59–66K62. Moreover, in the InflM peptide, position 9 is occupied by T, and in the BZLF-1 peptide, this position can be substituted by a K without affecting binding to HLA-E, although the latter substitution in an HLA-A2 leader sequence-derived peptide abrogated binding. Position 2, on the other hand, is occupied in the InflM peptide by L, one of the anchor residues defined by Braud et al. (44), although a polar Q is present at this position in the BZLF-1 peptide. Concerning the proposed auxiliary anchor at position 7, the BZLF-1 and InflM peptides share polar residues, whereas this position is occupied by a hydrophobic L in the leader sequence-derived peptides. The dramatic reduction in binding activity observed after truncation of position 7 (Fig. 4), on the other hand, might indicate the importance of this position. In contrast to Braud et al. (44), we found that position 4 shows some degree of side chain selectivity (K, F, and P are tolerated, whereas Q is not). Apparently, the structural requirements for peptide binding to HLA-E cannot be answered conclusively. For example, our screening included four peptides that contain XL/MX6L but did not stabilize HLA-E. This might be partly due to possible dominant negative effects on binding by amino acid residues at auxiliary anchor positions of the peptide (45). It is also conceivable that the different HLA-E ligands vary with regard to the anchor positions. For example, H2-Ld, which preferentially complexes nonameric peptides through anchors at positions 2 and 9, also binds naturally processed 8 mers by the interaction with amino acid 7 instead of 2 as an anchor (46, 47). The differences between our results and those obtained by Braud et al. (44) may also reflect the influence of the amino acid exchange at position 107 in the HLA-E allele (HLA-E*0101) they investigated on peptide ligand specificity. This explanation seems unlikely, since this substitution is located on the loop connecting the first and the second β strand of the α2 domain and thus is outside the binding groove.

Table II.

Peptides binding to HLA-Ea

PeptideSequenceBinding to HLA-ELength
BZLF-139–47 SQAPLPCVL 9 mer 
BZLF-139–47K46 SQAPLPCVK 9 mer 
BZLF-139–45 SQAPLPC 7 mer 
InflM59–66K62 ILGKVFTL 8 mer 
InflM59–68K68 ILGFVFTLTK 10 mer 
HLA-A23–11 VMAPRTL⃨VL 9 mer 
HLA-B83–11 VMAPRTV⃨LL 9 mer 
H-2D/L3–11 AMAPRTL⃨LL 9 mer 
HLA-G3–11 VMAPRTL⃨FL 9 mer 
HLA-A23–11L2 VLAPRTV⃨LL 9 mer 
HLA-A22–11K11 AVMAPRTLVK − 10 mer 
HLA-A23–11K11 VMAPRTLVK − 9 mer 
BZLF-139–47Q42 SQAQLPCVL − 9 mer 
PeptideSequenceBinding to HLA-ELength
BZLF-139–47 SQAPLPCVL 9 mer 
BZLF-139–47K46 SQAPLPCVK 9 mer 
BZLF-139–45 SQAPLPC 7 mer 
InflM59–66K62 ILGKVFTL 8 mer 
InflM59–68K68 ILGFVFTLTK 10 mer 
HLA-A23–11 VMAPRTL⃨VL 9 mer 
HLA-B83–11 VMAPRTV⃨LL 9 mer 
H-2D/L3–11 AMAPRTL⃨LL 9 mer 
HLA-G3–11 VMAPRTL⃨FL 9 mer 
HLA-A23–11L2 VLAPRTV⃨LL 9 mer 
HLA-A22–11K11 AVMAPRTLVK − 10 mer 
HLA-A23–11K11 VMAPRTLVK − 9 mer 
BZLF-139–47Q42 SQAQLPCVL − 9 mer 
a

Peptides identified as binding to HLA-E in this study are boxed. Only the sequences of the shortest peptides capable of binding to HLA-E as well as of those with important substitutions (bold faced) with regard to possible anchor residues are shown. Primary and secondary anchor residues defined by Braud et al. (45) in the leader sequence-derived peptides are underlined by solid and dotted lines, respectively. Binding to HLA-E is indicated by +, whereas − corresponds to no detectable binding activity.

It is therefore clear that HLA-E does not require any special structural features for peptides to bind. Consequently, we looked at whether the peptide loading of HLA-E might be impaired within the cell. For MHC class Ia molecules efficient peptide loading largely depends on the presence of functional TAP. Since the two molecules transiently interact, it is tempting to speculate that this interaction ensures the effective loading of peptides onto class I molecules. This idea is supported by the finding that the association of MHC class I molecules with TAP is disrupted when peptides binding to the class I molecule are added (8, 9). Moreover, proteasome inhibition prolongs and enhances the interaction of MHC class I molecules with TAP (36). Thus, the observed strong association of HLA-E with mouse TAP1 and TAP2 in the mouse myeloma transfectants under steady state conditions indicates that although HLA-E is connected to the cellular peptide loading machinery, its function with regard to HLA-E is inefficient. In contrast, pulse-chase experiments performed on X63BM cells revealed an initial high amount of HLA-B27 in association with mouse TAP that entirely dissociated within the 90-min chase in agreement with the unhampered transport of HLA-B27 to the cell surface in the X63 transfectant (Fig. 7) (27). Thus, under steady state conditions, with the majority of HLA-B27 being stably expressed at the cell surface, an association of HLA-B27 and TAP is barely detectable. Since the course of transient association of mouse TAP and HLA-B27 in X63BM cells is similar to the TAP/HLA class I interaction in human cells, we considered the transporter to be functional in our system with regard to peptide loading. The possibility that the lack of HLA-E cell surface expression in the mouse cell line might be explained by a too high affinity of HLA-E for mouse TAP preventing its dissociation even in the presence of appropriate peptide ligands seems unlikely as the HLA-E Ag is not efficiently expressed on the cell surface of the human cell lines 721.221 and K562 (44) (our unpublished observations). A high degree of association between human TAP and HLA-E has also been described in 721.221 by Braud et al. (44). However, some peptide transfer onto HLA-E does occur, as can be estimated from the release of HLA-E from TAP after 90-min chase, which can be inhibited by the addition of LLnL. These data also indicate that some TAP-dependent peptide loading of HLA-E occurs in much the same way as that reported for MHC class Ia molecules, although interaction of MHC class I molecules with TAP does not seem to be an absolute requirement for efficient peptide loading (48). On RMA-S cells cotransfected with HLA-E and hβ2m the HLA-E Ag is detected at the cell surface at similar levels as on X63EM cells (Fig. 8), indicating either that TAP is not required for the assembly of cell membrane-expressed HLA-E or that a lack of ER retention of HLA-E by TAP is responsible for the HLA-E cell surface expression observed on transfected RMA-S cells. With regard to the HLA-E peptide ligands derived from MHC class I signal sequences, a TAP-independent loading onto HLA-E is possible, but the identical peptides bind to Qa-1 in a TAP-dependent manner (49). Finally, it is conceivable that the strong association of HLA-E with TAP in X63EM cells reflects the inability of mouse TAP to supply the appropriate peptides. Indeed, mouse TAP has been shown to display a higher peptide selectivity than human TAP (50). At least with regard to HLA-B27, these differences are apparently not important, since sequencing of HLA-B27-bound peptides isolated from X63BM and human cells gave identical results (51).

In a model proposed by Sadasivan et al. (12), interaction with TAP is the last step in a series of chaperoning events leading to fully assembled class I complexes that are effectively transported to the cell surface. From these results (and since HLA-E interacts with TAP) we conclude that HLA-E molecules undergo unimpaired preceding assembly events, although the amount of HLA-E molecules metabolically labeled within 20 min in X63EM cells is much less than that of HLA-B27 in X63BM cells despite both transfectants expressing equal amounts of HLA class I mRNA (27). The observed increased detection of labeled HLA-E molecules in the presence of LLnL points to an early degradation of misfolded HLA-E molecules by a proteasome-dependent mechanism under normal circumstances, as has been reported for HLA class I molecules in mutant cells expressing either no functional TAP or β2m (52).

Although our data imply that HLA-E can present endogenous peptides, this Ag presentation either is not very effective or is highly specific. We have no evidence that HLA-E displays a high degree of peptide selectivity; of the tested 79 peptides, two could be shown to bind to HLA-E. It is therefore likely that other factors reduce HLA-E’s ability to present peptides. In contrast to HLA-B27, HLA-E strictly depends on the presence of hβ2m for its expression in mouse X63 cells (our unpublished observations). It is thus conceivable that high affinity peptides have to compensate for the low affinity of β2m in the formation of a trimeric HLA-E complex that is transported to the cell surface. Interestingly, pulse-chase experiments demonstrated that a large quantity of the HLA-E molecules expressed in X63EM cells is retained within the ER without substantial degradation even after 4 h (27). Therefore, additional molecular mechanisms, such as the association with calnexin or calreticulin, might retain the HLA-E molecule, indicating that it plays a physiologic function within the ER, namely in the Ag presentation by MHC class Ia molecules. In this respect it might be of significance that both BZLF-1 and InflM viral peptides overlap epitopes presented by HLA-A2. Thus, HLA-E possibly operates in an ER resident peptide loading and/or trimming process.

We demonstrated that two peptides present in mature proteins of different viruses bind HLA-E in vitro. In another report leader peptides derived from sequences of MHC class I molecules were described as ligands for HLA-E (44). The intracellular processing of the HLA-E Ag is impaired, and cell surface expression is low due to insufficient peptide supply, much like H-2Ld and Qa-1. Therefore, the HLA-E Ag might have a role in vivo as a restriction element for viral peptides similar to H-2Ld and/or might function similarly to Qa-1. It should now be possible to test whether HLA-E is a restriction element for cytotoxic, virus-specific T cells. The preference of HLA-E to bind hydrophobic and signal sequence-derived peptides supports the assumption that HLA-E might be involved in the control of the T cell response by presenting TCR-derived peptides to CTL as has been implicated for Qa-1 (53).

As cell surface expression of HLA-E seems to be regulated by the cell’s class Ia expression pattern and the supply of appropriate ligands, the HLA-E Ag presented on the cell surface per se might be recognized by effector cells in a possibly peptide-independent manner. Thus, HLA-E might regulate NK cell responses. In the placenta at the feto-maternal interface, the expression of fetus-derived invariant HLA-G may prevent the attack of cytotrophoblast cells by maternally derived NK cells. Since cell surface expression of HLA-E has been reported on amnion cells (54), it is possible that HLA-E fulfills a related function.

We thank Dr. F. A. Lemonnier for supplying the B9.12.1 hybridoma as well as the human β2m gene, Dr. F. Kievits for supplying the 27-11-13s ascites, Dr. K. Dornmair for providing the RP-FPLC equipment and his technical assistance, Dr. M. Pla for supplying the RMA-S cell line, as well as A. Brunner for technical assistance. Drs. A. McMichael, G. Pape, P. Robinson, B. Schoel, S. Shawar, and H. Zweerink are acknowledged for providing peptides. We thank Drs. J. Kellermann and F. Lottspeich for peptide sequencing.

1

This work was supported by the Deutsche Forschungsgemeinschaft (SFB 217).

4

Abbreviations used in this paper: ER, endoplasmic reticulum; hβ2m, human β2-microglobulin; LLnL, N-acetyl-l-leucyl-l-leucyl-l-norleucinal; RP-HPLC, reverse phased HPLC; B-LCL, B lymphoblastoid cell line; hsp, heat shock protein.

1
Townsend, A. R., F. M. Gotch, J. Davey.
1985
. Cytotoxic T cells recognize fragments of the influenza nucleoprotein.
Cell
42
:
457
2
Germain, R..
1994
. MHC-dependent antigen processing and peptide presentation: providing ligands for T lymphocyte activation.
Cell
76
:
287
3
Bijlmkers, M., H. Ploegh.
1993
. Putting together an MHC class I molecule.
Curr. Opin. Immunol.
5
:
21
4
Falk, K., O. Rötzschke, S. Stevanovic, G. Jung, H.-G. Rammensee.
1991
. Allele-specific motifs revealed by sequencing of self-peptides eluted from MHC molecules.
Nature
351
:
290
5
Heemels, M.-T., T. N. M. Schumacher, K. Wonigeit, H. L. Ploegh.
1993
. Peptide translocation by variants of the transporter associated with antigen processing.
Science
262
:
2059
6
Neefjes, J. J., F. Momburg, G. Hämmerling.
1993
. Selective and ATP-dependent translocation of peptides by the MHC-encoded transporter.
Science
261
:
769
7
Groettrup, M., A. Soza, U. Kuckelkorn, P.-M. Kloetzel.
1996
. Peptide antigen production by proteasome: complexity provides efficiency.
Immunol. Today
17
:
429
8
Suh, W.-K., M. F. Cohen Doyle, K. Fruh, K. Wang, P. A. Peterson, D. B. Williams.
1994
. Interaction of MHC class I molecules with the transporter associated with antigen processing.
Science
264
:
1322
9
Ortmann, B., M. J. Androlewicz, P. Cresswell.
1994
. MHC class I/β2-microglobulin complexes associate with TAP transporters before peptide binding.
Nature
368
:
864
10
Jackson, M. R., M. F. Cohen Doyle,P. A. Peterson, and D. B. Williams. 1994. Regulation of MHC class I transport by the molecular chaperone, calnexin (p88, IP90). Science 263:384.
11
Nössner, E., P. Parham.
1995
. Species-specific differences in chaperone interaction of human and mouse major histocompatibility complex class I molecules.
J. Exp. Med.
181
:
327
12
Sadasivan, B., P. J. Lehner, B. Ortmann, T. Spies, P. Cresswell.
1996
. Roles for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class I molecules with TAP.
Immunity
5
:
103
13
Fischer Lindahl, K..
1993
. Peptide antigen presentation by non-classical class I molecules.
Semin. Immunol.
5
:
117
14
Stroynowski, I..
1990
. Molecules related to class-I major histocompatibility complex antigens.
Annu. Rev. Immunol.
8
:
501
15
Bradbury, A., K. T. Belt, T. M. Neri, C. Milstein, F. Calabi.
1988
. Mouse CD1 is distinct from and co-exists with TL in the same thymus.
EMBO J.
7
:
3081
16
Rötzschke, O., K. Falk, S. Stevanovic, B. Grahovac, M. J. Soloski, G. Jung, H.-G. Rammensee.
1993
. Qa-2 molecules are peptide receptors of higher stringency than ordinary class I molecules.
Nature
361
:
642
17
Imani, F., M. J. Soloski.
1991
. Heat shock proteins can regulate expression of the Tla region-encoded class Ib molecule Qa-1.
Proc. Natl. Acad. Sci. USA
88
:
10475
18
Vidovic, D., M. Roglic, K. McKune, S. Guerder, C. MacKay, Z. Dembic.
1989
. Qa-1 restricted recognition of foreign antigen by a γδ T-cell hybridoma.
Nature
340
:
646
19
Pamer, E. G., C. R. Wang, L. Flaherty, K. Fischer Lindahl, M. J. Bevan.
1992
. H-2 M3 present a Listeriamonocytogenes peptide to CD8+ lymphocytes.
Cell
70
:
215
20
Kurlander, R. J., S. M. Shawar, M. L. Brown, R. R. Rich.
1992
. Specialized role of a murine class I-b MHC molecule in procaryotic host defenses.
Science
257
:
678
21
Kaliyaperumal, A., R. Falchetto, A. Cox, R. Dick, II, J. Shabanowitz, Y. Chien, L. Matis, D. F. Hunt, J. A. Bluestone.
1995
. Functional expression and recognition of nonclassical MHC class I T10b is not peptide-dependent.
J. Immunol.
155
:
2379
22
Weintraub, B. C., M. R. Jackson, S. M. Hedrick.
1994
. γδ T cells can recognize nonclassical MHC in the absence of conventional antigenic peptides.
J. Immunol.
153
:
3051
23
Bahram, S., M. Bresnahan, D. E. Geraghty, T. Spies.
1994
. A second lineage of mammalian major histocompatibility complex class I genes.
Proc. Natl. Acad. Sci. USA
91
:
6259
24
Beckman, E. M., S. A. Porcelli, C. T. Morita, S. M. Behar, S. T. Furlong, M. B. Brenner.
1994
. Recognition of a lipid antigen by CD1-restricted αβ+ T cells.
Nature
372
:
691
25
Wei, X. H., H. T. Orr.
1990
. Differential expression of HLA-E, HLA-F, and HLA-G transcripts in human tissue.
Hum. Immunol.
29
:
131
26
Ulbrecht, M., T. Honka, S. Person, J. P. Johnson, E. H. Weiss.
1992
. The HLA-E gene encodes two differentially regulated transcripts and a cell surface protein.
J. Immunol.
149
:
2945
27
Ulbrecht, M., J. Kellermann, J. P. Johnson, E. H. Weiss.
1992
. Impaired intracellular transport and cell surface expression of nonpolymorphic HLA-E: evidence for inefficient peptide binding.
J. Exp. Med.
176
:
1083
28
Perarnau, B. M., A. C. Gillet, R. Hakem, M. Barad, F. A. Lemonnier.
1988
. Human β2-microglobulin specifically enhances cell-surface expression of HLA class I molecules in transfected murine cells.
J. Immunol.
141
:
1383
29
Knarr, G., M.-J. Gething, S. Modrow, J. Buchner.
1995
. BiP binding sequences in antibodies.
J. Biol. Chem.
270
:
27589
30
Kioussis, D., F. Wilson, C. Daniels, C. Leveton, J. Taverne, J. H. Playfair.
1987
. Expression and rescuing of a cloned human tumour necrosis factor gene using an EBV-based shuttle cosmid vector.
EMBO J.
6
:
355
31
Bunch, T. A., Y. Grinblat, L. S. B. Goldstein.
1988
. Characterization and use of the Drosophila metallothionin promoter in cultured Drosophila melanogaster.
Nucleic Acids Res.
16
:
1043
32
Steller, H., V. Pirrotta.
1985
. A transposable P vector that confers selectable G418 resistance to Drosophila larvae.
EMBO J.
4
:
167
33
Matsumura, M., Y. Saito, M. R. Jackson, E. S. Song, P. A. Peterson.
1992
. In vitro peptide binding to soluble empty class I major histocompatibility complex molecules isolated from transfected Drosophila melanogaster cells.
J. Biol. Chem.
267
:
23589
34
Falo, L. D., Jr, L. J. Colarusso, B. Benacerraf, K. L. Rock.
1992
. Serum proteases alter the antigenicity of peptides presented by class I major histocompatibility complex molecules.
Proc. Natl. Acad. Sci. USA
89
:
8347
35
Sherman, L. A., T. A. Burke, J. A. Biggs.
1992
. Extracellular processing of peptide antigens that bind class I major histocompatibility molecules.
J. Exp. Med.
175
:
1221
36
Suh, W.-K., E. K. Mitchell, Y. Yang, P. A. Peterson, G. Waneck, D. B. Williams.
1996
. MHC class I molecules form ternary complexes with calnexin and TAP and undergo peptide-regulated interaction with TAP via their extracellular domains.
J. Exp. Med.
184
:
337
37
Shawar, S. M., R. G. Cook, J. R. Rodgers, R. R. Rich.
1990
. Specialized functions of MHC class I molecules I. An N-formyl peptide receptor is required for construction of the class I antigen Mta.
J. Exp. Med.
171
:
897
38
Vyas, J. M., R. R. Rich, D. D. Howell, S. M. Shawar, J. R. Rodgers.
1994
. Availability of endogenous peptides limits expression of an M3a-Ld major histocompatibility complex class I chimera.
J. Exp. Med.
179
:
155
39
Wang, C.-R., A. R. Castaño, P. A. Peterson, C. Slaughter, K. Fischer Lindahl, J. Deisenhofer.
1995
. Nonclassical binding of formylated peptide in crystal structure of the MHC class Ib molecule H2-M3.
Cell
82
:
655
40
Ojcius, D. M., C. Delarbre, P. Kourilsky, G. Gachelin.
1994
. Major histocompatibility complex class I molecules and resistance against intracellular pathogens.
Crit. Rev. Immunol.
14
:
193
41
Bouvier, M., D. C. Wiley.
1994
. Importance of peptide amino and carboxyl termini to the stability of MHC class I molecules.
Science
265
:
398
42
Conolly, D. J., L. A. Cotterill, P. J. Dyson, R. A. Hederer, C. J. Thorpe, P. J. Travers, J. H. McVey, P. J. Robinson.
1994
. A cDNA clone encoding the mouse Qa-1a histocompatibility antigen and proposed structure of the putative peptide binding site.
J. Immunol.
151
:
6089
43
Collins, E. J., D. N. Garboczi, D. C. Wiley.
1994
. Three-dimensional structure of a peptide extending from one end of a class I MHC binding site.
Nature
371
:
626
44
Braud, V., E. Y. Jones, A. McMichael.
1997
. The human major histocompatibility complex class Ib molecule HLA-E binds signal sequence derived peptides with primary anchor residues at positions 2 and 9.
Eur. J. Immunol.
27
:
1164
45
Ruppert, J., J. Sidney, E. Celis, R. T. Kubo, H. M. Grey, A. Sette.
1993
. Prominent role of secondary anchor residues in peptide binding to HLA-A2.1 molecules.
Cell
74
:
929
46
Corr, M., L. F. Boyd, S. R. Frankel, S. Kozlowski, E. A. Padlan, D. H. Margulies.
1992
. Endogenous peptides of a soluble major histocompatibility complex class I molecule, H-2Lds: sequence motif, quantitative binding, and molecular modeling of the complex.
J. Exp. Med.
176
:
1681
47
Robinson, R. A., D. R. Lee.
1996
. Studies of tum peptide analogs define an alternative anchor that can be utilized by Ld ligands lacking the consensus P2 anchor.
J. Immunol.
156
:
4266
48
Neisig, A., R. Wubbolts, X. Zang, C. Melief, J. Neefjes.
1996
. Allele-specific differences in the interaction of MHC class I molecules with transporters associated with antigen processing.
J. Immunol.
156
:
3196
49
Aldrich, C. J., A. DeCloux, A. S. Woods, R. J. Cotter, M. J. Soloski, J. Forman.
1994
. Identification of a Tap-dependent leader peptide recognized by alloreactive T cells specific for a class Ib antigen.
Cell
79
:
649
50
Momburg, F., J. Roelse, J. C. Howard, G. W. Butcher, G. J. Hämmerling, J. J. Neefjes.
1994
. Selectivity of MHC-encoded peptide transporters from human, mouse and rat.
Nature
367
:
648
51
Rötzschke, O., K. Falk, S. Stevanovic, V. Gnau, G. Jung, H. G. Rammensee.
1994
. Dominant aromatic/aliphatic C-terminal anchor in HLA-B*2702 and B*2705 peptide motifs.
Immunogenetics
39
:
74
52
Hughes, E. A., C. Hammond, P. Cresswell.
1997
. Misfolded major histocompatibility complex class I heavy chains are translocated into the cytoplasm and degraded by the proteasome.
Proc. Natl. Acad. Sci. USA
94
:
1896
53
Jiang, H., R. Ware, A. Stall, L. Flaherty, L. Chess, B. Pernis.
1995
. Murine CD8+ T cells that specifically delete autologous CD4+ T cells expressing Vβ8 TCR: a role of the Qa-1 molecule.
Immunity
2
:
185
54
Houlihan, J. M., P. A. Biro, H. M. Harper, H. J. Jenkinson, C. H. Holmes.
1995
. The human amnion is a site of MHC class Ib expression: evidence for the expression of HLA-E and HLA-G.
J. Immunol.
154
:
5665