Final Antigenic Melan-A Peptides Produced Directly by the Proteasomes Are Preferentially Selected for Presentation by HLA-A*0201 in Melanoma Cells

Digestion of synthetic peptides, analyses by on-line HPLC-mass spectrometry (HPLC-MS/MS), and quantification

Wild-type and A27L substituted precursor peptides (3.5 μg), Melan-A_15–40 and Melan-A_15–40^A27L, respectively, were digested by 4 μg of proteasomes isolated from human erythrocytes in 13 μl of 25 mM Tris, pH 7.6. Proteasome affinity purification and quantification were performed as previously described (8, 13, 14). The purity of the proteasome preparations was assessed by SDS-PAGE and staining. The reaction was stopped after 1 h with 1% trifluoroacetic acid (TFA). After lyophilization, the digests were resuspended in 12 μl of H₂O/CH₃CN (50/50, v/v) and diluted to 48 μl in CH₃CN/H₂O/HCOOH (5/95/0.05, v/v/v). Twelve microliters of a 20 μM solution of an internal standard synthetic peptide in CH₃CN/H₂O/HCOOH (5/95/0.05, v/v/v) was then added to the digest. Fifteen microliters was analyzed by online HPLC-UV-electro-spray ionization-MS/MS. Peptides were separated on a Pepmap LC Packings (Dionex, Amsterdam, The Netherlands) C₁₈ column (1 mm × 15 cm) at a flow rate of 40 μl/min. The separation was achieved with a gradient elution of 10–50% for 40 min (A is 5/95/0.05 CH₃CN/H₂O/HCOOH, B is 70/30/0.05 CH₃CN/H₂O/HCOOH; v/v/v). UV absorption was measured online at 215 nm with electro-spray ionization-MS/MS detection performed on an LCQ ion trap mass spectrometer (ThermoFinningan, San Jose, CA). Data acquisition was performed in a data-dependent mode consisting of a full-scan MS over the mass range m/z 220-2000, followed by a zoom-scan MS and a full-scan MS/MS in a dynamic exclusion mode of the most intense ion detected in the full-scan MS. MS/MS data were acquired using a 3 m/z ion isolation window, a relative collision energy of 35%, and a dynamic exclusion duration of 0.5 min. After identification by mass spectrometry, quantification of the peptides was performed by measuring the surface area (millivolts per second) of the relevant peaks on the UV chromatograms and by measuring the ion current (arbitrary units) corresponding to a specific m/z value in the full-scan MS. For nonoverlapping UV peaks, the variations in relative abundances of different peptide species determined by UV and MS analyses were similar. In case of peptide coelution during the HPLC separation, quantification was based on MS data only.

Construction of plasmids and recombinant vaccinia

The ubiquitin/protein/reference (UPR)-based pEGFP vector used in this study is identical with the vector described previously (12), except that the GFP coding sequence was replaced by the enhanced GFP (EGFP). All constructs were obtained by annealing complementary synthetic oligonucleotides containing at their 5′ and 3′ ends SacII and BamHI restriction sites, respectively. The oligonucleotides contained two codons for Gly at the 5′ end (except for construct shown in Fig. 2⇓B; see below) and a stop codon at the 3′ end. Insertion of these fragments into the SacII-BamHI sites of pEGFP/ubiquitin (Ub) led to the generation of plasmids directing the expression of the reference protein EGFP-Ub and the Melan-A fragments spanning aa 26–35 (M26–35^A27L), aa 26–40 (M26–40^A27L), aa 20–35 (M20–35^A27L), and aa 22–35 (M22–35^A27L; Fig. 2⇓A). In the construct Ub-M26–35^A27L (Fig. 2⇓B), the two C-terminal Gly of ubiquitin were removed, inhibiting the cleavage of EGFP-Ub by ubiquitin peptidase (15). In all constructs, the amino acid Ala²⁷ of Melan-A was substituted by a leucine (L). The sequence of all inserts was confirmed by DNA sequencing.

The generation of recombinant vaccinia virus was described previously (12). Viruses contain the UPR-based expression cassette consisting of GFP-Ub. Cells were infected at a multiplicity of infection of 10 for 4 h before the addition of CTL.

Cell lines

The human melanoma cell line NA8-MEL was maintained in RPMI 1640 medium (Invitrogen Life Technologies, Carlsbad, CA) supplemented with 10% FCS, 10 mM HEPES (Invitrogen Life Technologies), and 1% penicillin-streptomycin (Invitrogen Life Technologies). The CTL used in the IFN-γ release assay is a murine CTL line obtained by weekly restimulations of cells after immunization of HLA-A*0201/K^b transgenic mice (16) with vaccinia virus coding for Melan-A_26–35^A27L. The CTL line was maintained in DMEM (Invitrogen Life Technologies) supplemented with 10% FCS, 10 mM HEPES (Life Technologies), 1% penicillin-streptomycin (Life Technologies), 2% 2-ME, and 30 U/ml IL-2 (produced at our institute).

The human CTL clone A42 was provided by Dr. L. Rivoltini (Unit of Immunotherapy of Human Tumors, Istituto Nazionale Tumori, Milan, Italy) and has been described previously (17). The line EAA and the clone A42 were maintained by culture in DMEM supplemented with 8% human A⁺ pooled sera and 10³ U/ml human rIL-2.

CTL assay, IFN-γ release assay, and ELISA

NA8-MEL cells were seeded at a density of 2 × 10⁴ cells/well in a 96-well plate the day before transfection. Plasmids (0.1 μg/well) were transfected in duplicate into cells with FuGene according to the manufacturer’s protocol (FuGene transfection reagent; Amersham Biosciences, Arlington Heights, IL). Half the cells were pulsed with exogenous Melan-A_26–35^A27L peptide (1 μM final concentration) for 30 min at 37°C before the addition of CTL at an E:T cell ratio of 3:1 in DMEM containing 10% FCS, penicillin-streptomycin/HEPES, 5% IL-2, and 2% 2-ME. Twenty-four hours later, supernatants were harvested, and the concentration of IFN-γ in the supernatant was measured by ELISA. Microtiter plates were coated with 2 μg/ml anti-mouse IFN-γ mAbs (R46A2, produced at the institute) and blocked with 1% BSA in PBS. Serial 2-fold dilutions of the cell supernatant were added to the wells, and the concentration of IFN-γ was revealed using 1 μg/ml anti-murine IFN-γ-biotinylated rat Ig (ANI-803, produced at the institute) and HRP-conjugated streptavidin (Invitrogen Life Technologies, Gaithersburg, MD). The OD was measured at 490 nm. Duplicate values were generally within 15% of the mean.

Western blot analysis

An aliquot of each infected target cell (2 × 10⁶ cells) used in the CTL assay presented in Fig. 5⇓C was lysed in 100 μl of lysis buffer containing 20 mM Tris (pH 7.6), 150 mM NaCl, 5 mM EDTA, and 1% Nonidet P-40. The lysate was cleared by centrifugation at 11,000 × g for 10 min at 4°C. Ten microliters of the cleared lysate was separated by SDS-12% PAGE and subjected to immunoblotting. Due to the fact that the reference protein GFPha-Ub contained an influenza hemagglutinin (ha) epitope tag between the sequence of GFP and Ub, its expression was detected with the anti-ha mAb (1/2,000; Roche, Indianapolis, IN) and HRP-conjugated anti-mouse Ab (1/2,500; Amersham Biosciences) and revealed by ECL (Amersham Biosciences). The x-ray film was scanned and the bands corresponding to GFPha-Ub were quantitated densitometrically using AIDA Image Analyzer software (Raytest, Straubenhardt, Germany).

Flow cytometry

Flow cytometry was performed on FACScan devices using CellQuest software (BD Biosciences, San Jose, CA). Transfected cells were washed twice, resuspended in PBS/3% FCS, and directly analyzed. Identification of live cells was based on the forward and side scatter of mock-transfected cells. Sorting of transfected cells was performed using a FACSAria cell sorter (BD Biosciences). Cells (10⁷) were washed twice, resuspended in PBS/3% FCS, and directly sorted according to their EGFP expression level into EGFP^high and EGFP^low subpopulations. Sorted cells were allowed to recover for 60 min in complete medium before the addition of specific CTL.

Peptide digestion by incomplete cytosol, HPLC separation, and MS

Purified synthetic peptides were quantified by amino acid analysis. Five nanomoles of peptide Melan-A_26–35, Melan-A_26–35^A27L, Melan-A_22–35, Melan-A_22–35^A27L, and RU1_29–42 were digested in 40 μl of incomplete cytosol (cytosol without proteasome) corresponding to 3 × 10⁶ cell equivalents. Preparation of proteasome-depleted cytosol was described previously (15) and contained <5% of the remaining proteasomal activities, based on the proteolysis of fluorogenic substrates (data not shown). For each peptide, an aliquot was removed at time zero and served as the reference for quantitative analysis. Digestions were allowed to proceed for 2.5, 5, 10, and 20 min at 37°C. At each time point, the reaction was stopped with 1% TFA. The digests were diluted with acetic acid/water (50/50, v/v) to a final volume of 120 μl, and 100 μl was injected on the HPLC. HPLC separations were performed on an Agilent system (1100 series) using a C₁₈ small pore Vydac (250 × 4 mm; 5 μm; 201SP54; Hesperia, CA) column. The column was eluted at a flow rate of 1 ml/min by a linear gradient of acetic acid/TFA/acetonitrile/H₂O (0.1/0.005/80/20, v/v/v/v) on acetic acid/TFA/H₂O (0.1/0.005/100, v/v/v/), rising within 60 min from 0 to 100%. The OD of the eluate was monitored at 214 nm. Fractions were collected using an Agilent (G1364A) fraction collector on 96-well plates and subjected to MALDI-TOF mass spectrometric analysis as previously described (15).

Degradation of Melan-A_26–35 and Melan-A_26–35^A27L precursors by proteasomes produces minimal and N-terminally extended antigenic peptides

We first analyzed and compared degradation of the wild-type and mutated Melan-A_26/27–35 precursors by purified proteasomes in vitro. The 26-aa-long precursors, encompassing aa 15–40, were digested with standard proteasomes and subjected to on-line HPLC-UV-MS/MS analyses. As shown in Fig. 1⇓, the major single proteasomal cleavage site was detected at the exact C terminus of the wild-type and mutated antigenic peptides, producing the fragments Melan-A_15–35 and Melan-A_27–35^A27L (and its complementary fragment ILGVL). Five fragments were generated with that C terminus: the final antigenic peptides Melan-A_26–35 and Melan-A_27–35, two N-terminally extended fragments corresponding to the peptide Melan-A_22–35 and Melan-A_20–35, and the 8-mer Melan-A_28–35, a fragment too short to bind to HLA-A*0201 (Fig. 1⇓, A and B). The peptides Melan-A_28–35, Melan-A_27–35, Melan-A_26–35, and Melan-A_22–35 were produced in similar amounts, whereas less of the fragment Melan-A_20–35 was generated. The same cleavages were also observed with the mutated precursor Melan-A_15–40^A27L, although the relative ratio of Melan-A_26–35^A27L vs Melan-A_27–35^A27L was strongly biased in favor of the 10-mer (Fig. 1⇓, C and D). The N-terminally extended precursors were also produced, as was the internal 8-mer Melan-A_28–35. We concluded from this experiment that both wild-type and mutated minimal antigenic Melan-A peptides were produced by the proteasomes along with two N-terminally extended intermediates.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Degradation of Melan-A by proteasomes. A, HPLC profile of the fragments produced by the degradation of the synthetic peptide Melan-A_15–40 by standard proteasomes. The identities of relevant peaks are labeled I–VI and correspond to the fragments indicated in B. The star describes the elution peak of an internal marker, and the peak labeled ILGVL corresponds to the C-terminal fragments 36–40. B, Schematic representation of the fragments detected in A and displaying the correct C termini. The dotted lines indicate the cleavage sites, and the horizontal bars show the length of the fragment. The thickness of the bars is indicative of the relative amount of each fragment compared with the most abundant one (fragments 15–35), which was arbitrarily set at 1. The Roman numerals correspond to the fragments produced by the proteasomes. The position of the amino acids within the Melan-A protein is indicated above the sequence. Only fragments displaying the correct C terminus of the antigenic peptides are shown. C, Same as A, except that the synthetic peptide Melan-A_15–40^A27L was used in the digestion. D, Same as B for Melan-A_15–40^A27L.

Previous studies reported that peptide fragments produced by proteasomes in vitro could also be isolated from whole-cell extracts, suggesting that the production of peptides in vitro is representative of the intracellular processing (18, 19). Because the final Melan-A peptides and intermediates were produced by the proteasomes in vitro, we determined which one, if any, would be preferentially selected for loading onto HLA-A*0201 molecules in melanoma cells.

Recognition of cells expressing Melan-A_26–35^A27L and extended precursors by specific CTL

Based on the peptide fragments described in Fig. 1⇑D, we constructed a series of plasmids to express the corresponding fragments in mammalian cells (Fig. 2⇓). We selected four sequences, two corresponding to the fragments carrying additional four and six N-terminal amino acids, respectively, and one corresponding to the final antigenic decamer. Because the proteasomes have been shown to limit the supply of antigenic peptides, we also compared the presentation efficiency of a peptide that only required proteasomal processing with those that only required N-terminal trimming. Therefore, we included a plasmid coding for a fragment that displayed the final N terminus, but carried an additional five amino acids at the C terminus of the antigenic peptide. All constructs were expressed from a UPR-based plasmid, which produces equimolar amounts of the reference protein EGFP-Ub and the fragment of interest (14, 20). This strategy circumvents the need for N-terminal Met and allows measurement of the transfection efficiency and expression levels of the EGFP-Ub moieties by flow cytometry. We focused on the processing of Melan-A peptides carrying the A27L mutation because we had previously shown that the recognition of transiently transfected cells expressing the wild-type antigenic peptide Melan-A_26–35 did not efficiently stimulate the production of cytokines by specific CTL (7, 12).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

DNA constructs used in the study. A, Plasmids coding for the two N-terminally extended intermediates M20–35^A27L and M22–35^A27L, the antigenic peptide M26–35^A27L and the C-terminally extended fragment M26–40^A27L were generated. All constructs were expressed from an UPR-based plasmid, which contained the ha-tagged EGFPha-Ub as reference protein. This construct produces equimolar amounts of two distinct polypeptides, EGFPha-Ub and the Melan-A peptides. The arrow indicates the cleavage site of the Ub-specific protease. B, The antigenic peptide M26–35^A27L expressed from a noncleavable UPR-based plasmid.

Human NA8-MEL melanoma cells (HLA-A*0201 positive, Melan-A negative) were transiently transfected with those constructs. Transfection efficiency and expression level were assessed by flow cytometry (Fig. 3⇓A). Approximately 20% of the cells expressed EGFP-Ub, and the level of expression was similar for each construct, as judged by the mean fluorescence of the fluorescent cell pool. Nevertheless, the expression level in individual cells varied significantly within a transfected population, as evidenced by the width of the fluorescence histogram.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Expression and presentation of Melan-A peptides to specific CTL. A, NA8-MEL cells were transiently transfected with the constructs described in Fig. 2⇑. The transfection efficiency (percentage) and the average expression level (MF) were calculated based on the fluorescence emitted by the reference protein EGFPha-Ub. The constructs are indicated above each plot. B, Transfected cells were incubated in duplicate with Melan-A-specific CTL at an E:T cell ratio of 3:1, and the production of IFN-γ by the CTL was measured by ELISA. As controls, cells transfected with EGFPha-Ub (−) and mock-transfected cells loaded with saturating amounts of synthetic peptide Melan-A_26–35^A27L were used. A representative result of three independent experiments is shown. C, Cells transiently transfected with M26–35^A27L and Melan-A^A27L were sorted by FACS into an EGFP^bright and EGFP^dim subpopulation, based on fluorescence intensity emitted by the EGFPha-Ub moiety. D, The bright (EGFP high) and dim (EGFP low) subpopulations expressing either M26–35^A27L or Melan-A^A27L were separately incubated with specific CTL, at an E:T cell ratio of 3:1, and the production of IFN-γ was measured by ELISA. Mock-transfected cells (−) were used as a negative control. A representative result of two independent experiments is shown.

Cells from the same transfection as that used for flow cytometry were assayed for their recognition by specific anti-Melan-A CTL (Fig. 3⇑B). CTL were incubated for 24 h with the transfected cells at a ratio of 3:1, and the secretion of IFN-γ by the CTL was measured by ELISA. Cells expressing the antigenic decamer (M_26–35^A27L) were efficiently recognized, as indicated by the level of IFN-γ release. This level was similar to that obtained by incubating CTL with cells loaded with saturating amounts of exogenous Melan-A_26–35^A27L peptide. The recognition of cells expressing the C-terminally extended precursor M_26–40^A27L, which requires a single proteasomal cleavage for the production of the antigenic peptide, was lower, confirming that the proteasomal cleavage is limiting the availability of the antigenic peptide. Surprisingly, cells expressing the two N-terminally extended precursors, M_20–35^A27L and M_22–35^A27L, were recognized less efficiently than cells expressing M_26–35^A27L. These results suggest that conversion of the N-terminally extended intermediates into the final antigenic Melan-A peptide is inefficient in these cells. Alternatively, rapid proteolytic cleavage within the antigenic peptide could limit presentation of the N-terminally extended products.

Sorting of the subpopulations of transfected cells expressing M_26–35^A27L or the full-length protein Melan-A^A27L and exhibiting bright or dim fluorescence intensities by FACS (Fig. 3⇑C) showed that both populations could be recognized by anti-Melan-A CTL, but that the amount of IFN-γ secreted by CTL was 3 times higher when incubated with bright cells than with dim cells (Fig. 3⇑D). In both subpopulations, cells expressing M_26–35^A27L were recognized more efficiently than those expressing the full-length protein.

To control that the efficient presentation of cells expressing M26–35^A27L was caused by release of the EGFP-Ub moiety from the final antigenic peptide and not by rapid degradation of the uncleaved fusion protein, we introduced a mutation in the plasmid M26–35^A27L, which resulted in the production of an uncleavable fusion protein, EGFP-Ub-M26–35^A27L (Fig. 2⇑B). This mutation, which removes the two C-terminal Gly residues of ubiquitin, prevents cleavage of the EGFP-Ub moiety by ubiquitin proteases. The effects of this mutation have been described previously (15). NA8-MEL cells were transfected with this plasmid and exposed to anti-Melan-A CTL (Fig. 3⇑B). Recognition of the transfected cells by CTL was poor, confirming that the efficient recognition of cells expressing M26–35^A27L was due to the efficient release of the antigenic peptide in its final form. This also confirmed that the differences observed between the fully processed peptide and the N-terminally extended ones were due to the peptide-flanking sequences liberated in the cytosol.

Trimming of Melan-A_26–35 and Melan-A_26–35^A27L precursors by cytosolic proteases

A possible cause of the apparent inefficient presentation of the antigenic Melan-A peptide in cells expressing the N-terminally extended precursors could be their different sensitivities to cytosolic peptidases. To address this question, we incubated synthetic peptides corresponding to the 14-mers Melan-A_22–35 and Melan-A_22–35^A27L as well as the 10-mers Melan-A_26–35 and Melan-A_26–35^A27L with NA8-MEL cytosol depleted of proteasomes. As control, we incubated an N-terminally extended 14-mer precursor of the HLA-B51-restricted peptide RU1_34–42. This precursor was shown previously to be trimmed to the antigenic nonamer by the sequential activity of two different cytosolic peptidases, tripeptidyl peptidase II and puromycin-sensitive aminopeptidase (13, 15). The two 14-mer Melan-A wild-type and mutated peptides were degraded at a significantly lower rate than the RU1 peptide and the 10-mer Melan-A peptides (Fig. 4⇓, A and B, respectively). The half-life of the 14-mer Melan-A peptides was ∼10 min, whereas the half-life of the 14-mer RU1 peptide was ∼5 min, similar to the half-life of the antigenic 10-mers. At each time point, the peptide digest was separated by HPLC, and the content of each detected peak was identified by mass spectrometry. As shown in Fig. 4⇓B, the species corresponding to the final antigenic peptide sequence 26–35 was produced from the precursor Melan-A_22–35 and was detectable after 20 min degradation. For Melan-A_22–35^A27L, the final antigenic peptide could be identified after 5 min. In sharp contrast, the fragment corresponding to the final antigenic peptide RU1_34–42 was already clearly quantifiable after 2.5 min. The exact composition of the fractions containing the final antigenic peptides for each of the precursors is detailed in Fig. 4⇓C. For Melan-A_22–35 (upper row), deletion fragments were already detected at time zero. However, several other fragments appeared over time, including the final antigenic peptide at 20 min. For Melan-A_22–35^A27L (middle row), the final antigenic peptide was already detectable at 5 min, among other fragments. Finally, the final antigenic peptide produced from the RU1 precursor (lower row) was detected at 0, 2.5, and 5 min. Despite its detection by mass spectrometry at time zero, the final antigenic peptide is nevertheless produced in the course of the digestion, and its actual amount can be derived from the HPLC profile shown in Fig. 4⇓B, because no other fragment coeluted. The RU1_29–42 peptide was completely digested at 20 min. The antigenic 9-mer Melan-A_27–35 (m/z 815) and Melan-A_27–35^A27L (m/z 857) could not be detected by MALDI-TOF MS, because this region contains several intense peaks corresponding to matrix ions. Because MALDI-TOF mass spectrometry does not allow the identification of fragments below a mass of 850 Da, we cannot completely rule out that a rapid internal cleavage could take place in the cytosol. Nevertheless, the degradation kinetics of the Melan-A precursors correlated well with the efficiency by which the final antigenic peptides were generated.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Degradation of Melan-A peptides by cytosolic peptidases. A, Synthetic peptides corresponding to Melan-A_22–35 and Melan-A_26–35 (left panel) as well as Melan-A_22–35^A27L and Melan-A_26–35^A27L (right panel) were incubated with proteasome-depleted cytosol for 0, 2.5, 5, 10, and 20 min at 37°C. The 14-mer peptide RU1_29–42 was used as a reference. At each time point, the reaction was stopped, and the peptide was separated by HPLC. The percentage of precursor peptide remaining at each time point is calculated based on the amount of peptide present at time zero. Average values (n = 2) with mean error are shown. The dotted line indicates the 50% level. B, Samples shown in A were separated by HPLC, and the HPLC profiles at 0, 2.5, 5, and 20 min are shown for Melan-A_22–35 (left column), Melan-A_26–35^A27L (middle column), and RU1_29–42 (right column). The main peak of each digestion contained the precursor peptide and served to determine the degradation curve shown in A. The species detected in the relevant fractions are indicated as P for the precursor and as P-1, P-2, and P-3 for fragments lacking one, two, and three N-terminal amino acids, respectively. The gray boxes correspond to the final antigenic 10-mer peptide of Melan-A and 9-mer of RU1. For digestion of the RU1 peptide, a detailed view of the relevant region is shown in the inset. Under these separating conditions, the RU1 peptide elutes in two peaks. We do not know the reason for this. Because only one species could be detected by MS, the entire double peak was integrated for the quantitation shown in A. A representative experiment of two is shown. C, The composition of each HPLC fraction containing the final antigenic peptides was assessed by MALDI-TOF MS. The identity of each fragment is labeled as in B. Double peaks correspond to the peptide and its Na⁺ adduct (+22 Da). The stars identify contaminating peaks.

Melan-A_26–35 is inefficiently converted into Melan-A_27–35 by intracellular proteases

To confirm that the expression of an N-terminally extended Melan-A precursor leads to poor recognition of target cells by specific CTL, we took advantage of the fine specificity of a previously described CTL clone, clone A42 (17). This clone was shown to recognize Melan-A_27–35 more efficiently than Melan-A_26–35, which can be considered, under these circumstances, as an N-terminally extended intermediate. We first established that this tumor-reactive clone indeed recognizes the nonamer Melan-A_27–35 more efficiently than the decamer Melan-A_26–35 (Fig. 5⇓A). We then tested whether melanoma cells expressing the nonamer Melan-A_27–35 or the decamer Melan-A_26–35 would be recognized with similar efficiency. To this end, we infected NA8-MEL cells with recombinant vaccinia virus coding for the wild-type versions of the decamer Melan-A_26–35, the nonamer Melan-A_27–35, and the full-length Melan-A protein and incubated them with CTL clone A42 (Fig. 5⇓B). As expected, the cells expressing the antigenic nonamer were efficiently recognized by the CTL clone A42, but not cells expressing the decamer Melan-A_26–35. NA8-MEL cells infected with recombinant vaccinia virus coding for the 10-mer Melan-A_26–35^A27L were not recognized either (Fig. 5⇓C). However, cells expressing the full-length proteins Melan-A and Melan-A^A27L were recognized by the CTL clone A42. To ensure that the cells used for the experiment depicted in Fig. 5⇓C were expressing similar amounts of Melan-A proteins or peptides, we took advantage of our UPR-based constructs, in which Melan-A is expressed as C-terminal fusions of the ha-tagged reporter GFP-Ub (12, 20). We therefore performed a Western blot analysis on an aliquot of cells used in Fig. 5⇓C. As shown in Fig. 5⇓D, infected cells were expressing similar amounts of the reporter proteins. Because the reporter and the Melan-A products are expressed from the same mRNA, the amount of detected reporter is equimolar to the amount of Melan-A in the cells. This observation led us to conclude that although the decamers Melan-A_26–35 and Melan-A_26–35^A27L are expressed, they are not efficiently converted into the antigenic nonamers in infected cells, supporting our previous findings that cells expressing the N-terminally elongated precursors of Melan-A were inefficiently recognized by CTL. However, we cannot completely rule out that the decamer Melan-A_26–35 escapes N-terminal trimming by rapid binding to HLA-A*0201 molecules. Cells expressing the full-length protein were also recognized by the CTL clone A42, suggesting that along with other fragments, these cells produce the antigenic nonamer from the full-length protein. In parallel, cells infected with the same constructs were incubated with the CTL line EAA that indiscriminately recognizes the nonamer and decamer peptides. Contrary to what was observed with clone A42, cells expressing the different constructs were recognized with similar efficiency, indicating that infected NA-8 MEL can present Melan-A_26–35 as well as Melan-A_27–35 (Fig. 5⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Recognition of melanoma cells expressing Melan-A_27–35, wild-type and A27L-mutated Melan-A_26–35, and full-length Melan-A. A, The fine specificity of the anti-Melan-A CTL clone A42 was tested by exogenous loading of HLA-A*0201⁺ Melan-A⁻ T2 cells with peptides Melan-A_27–35 and Melan-A_26–35, respectively. B, Recognition of NA8-MEL cells infected with recombinant vaccinia expressing Melan-A_27–35, Melan-A_26–35, and full-length Melan-A. The left panel represents the target cell lysis by CTL clone A42. The right panel corresponds to the target cell lysis by CTL line EAA, which recognizes Melan-A_27–35/HLA-A*0201 and Melan-A_26–35/HLA-A*0201 with similar efficiency. C, Same as in B, except that NA8-MEL cells infected with recombinant vaccinia virus coding for the mutated Melan-A^A27L and Melan-A_26–35^A27L were included. D, Expression level of the reference protein GFPha-Ub in NA8-MEL cells infected with recombinant vaccinia virus. An aliquot of the cells used in C was lysed and subjected to Western blot analysis using the anti-ha mAb that detects the UPR-based reference protein GFPha-Ub (see text for details). The relative intensity of the band corresponding to GFPha-Ub is indicated below each lane. The intensity of the band detected in cells expressing Melan-A_27–35 was arbitrarily set at 1. For each panel, a representative result of two independent experiments is shown.

These data demonstrate 1) that Melan-A_26–35, which can be considered an N-terminally extended precursor of Melan-A_27–35, is not converted to the antigenic 9-mer in melanoma cells; and 2) that both 10- and 9-mer Melan-A peptides can be presented by HLA-A*0201 at the surface of melanoma cells.