An Antigenic Peptide Produced by Peptide Splicing in the Proteasome

See allHide authors and affiliations

Science  23 Apr 2004:
Vol. 304, Issue 5670, pp. 587-590
DOI: 10.1126/science.1095522


CD8 T lymphocytes recognize peptides of 8 to 10 amino acids presented by class I molecules of the major histocompatibility complex. Here, CD8 T lymphocytes were found to recognize a nonameric peptide on melanoma cells that comprises two noncontiguous segments of melanocytic glycoprotein gp100PMEL17. The production of this peptide involves the excision of four amino acids and splicing of the fragments. This process was reproduced in vitro by incubating a precursor peptide of 13 amino acids with highly purified proteasomes. Splicing appears to occur by transpeptidation involving an acyl-enzyme intermediate. Our results reveal an unanticipated aspect of the proteasome function of producing antigenic peptides.

The antigenic peptides recognized by CD8 T cells result from the degradation of intracellular proteins and correspond to small fragments of these proteins. One interesting exception was recently reported by Hanada et al., who described an antigenic peptide composed of two noncontiguous segments of the parental protein (1). The production of this peptide implied that an intervening segment of 40 amino acids had been excised and that the products had been spliced, but the mechanism remained to be characterized. We describe here a second example of antigenic peptide produced by peptide splicing and show that the excision and splicing process is exerted by the proteasome, a multicatalytic peptidase complex that accounts for the bulk of protein degradation in the cytosol (2).

By stimulating blood lymphocytes of a melanoma patient with autologous tumor cells, we isolated a clone of CD8 cytolytic T lymphocytes (CTL) that recognized an antigen presented by human lymphocyte antigen (HLA)–A32 and encoded by SILV, the gene encoding glycoprotein gp100PMEL17 (Fig. 1) (3, 4). This antigen is present on several melanoma cell lines. To localize the peptide recognized by this clone, named CTL 14, we transfected COS-7 cells with plasmids encoding HLA-A32 and a series of truncated gp100 sequences, and we tested the transfected cells for recognition by the CTL (Fig. 1B) (5). The results pointed to a 16–amino acid segment located between positions 37 and 52. When HLA-A32+ target cells were incubated with a series of synthetic peptides contained in this segment, none was recognized by the CTL. We considered that the peptide antigenicity might depend on a posttranslational modification, and we tried to enable this putative modification by introducing synthetic peptides inside target cells by electroporation before testing for CTL recognition. We observed that the 13–amino acid peptide RTKAWNRQLYPEW (gp100 positions 40 to 52) (6), which was not recognized by CTL 14 after exogenous loading on autologous Epstein-Barr virus (EBV)–transformed B cells, was recognized after electroporation into these cells (Fig. 2A). Shorter peptides lacking the N- or the C-terminal residue remained negative after electroporation.

Fig. 1.

Recognition by CTL 14 of an antigen derived from gp100PMEL17. (A) Lytic activity of clone CTL 14 on autologous melanoma line LG2-MEL-5-35, autologous EBV-transformed B cells LG2-EBV, allogeneic HLA-A32+ gp100+ melanoma lines LB17-MEL and CP64-MEL, and autologous melanoma line LG2-MEL-220, which has lost expression of gp100. (B) Recognition by CTL 14 of COS-7 cells transfected with the HLA-A*3201 cDNA and either the full-length gp100 cDNA or subgenic fragments of gp100. Minigenes are named according to nucleotide positions in the open reading frame. Amino acid positions are indicated below the peptide sequence.

Fig. 2.

Identification of the peptide recognized by CTL 14. (A) LG2-EBV cells were electroporated with the indicated peptides and tested for recognition by CTL 14 (black bars). As control, peptide RTKAWNRQLYPEW was pulsed onto LG2-EBV cells in the same conditions except that electroporation was omitted (white bar). The alanine substitutions are indicated with a bold A and a subscript indicating the position of the substituted residue. (B) Recognition of peptide RTKQLYPEW by CTL 14. 51Cr-labeled LG2-EBV cells were incubated with the indicated peptides. CTL 14 was added at an effector-to-target ratio of 10. (C) Elution of the natural peptide recognized by CTL 14. Peptides eluted from HLA class I molecules purified from melanoma cells LG2-MEL-5-35 were separated by HPLC, and the fractions were tested for recognition by CTL 14 (top panel). To rule out contamination of the HPLC system, buffer was run on the column before the eluted samples, and the fractions were tested similarly. Synthetic peptide RTKQLYPEW (30 pmol), which was injected under the same HPLC conditions, was monitored by ultraviolet (UV) absorption, and the fractions were tested for CTL recognition (middle panels). A digest obtained after a 120-min incubation of 20S proteasomes with peptide RTKAWNRQLYPEW was separated and tested similarly (bottom panel). mUA, milli-units of absorbance; 13-mer peptide, 13–amino acid peptide.

To identify the residues that were important for antigenicity, we synthesized a series of peptides based on the sequence of the positive 13–amino acid peptide with an alanine substitution at each position and tested them for CTL recognition after electroporation (Fig. 2A). Whereas residues located near the N and the C terminus appeared essential for CTL recognition, the substitution of residues located in the middle region (positions 44 to 48) of the peptide affected antigenicity only marginally. Because of the recent findings of Hanada et al. (1), we considered that the modification required for antigenicity of the 13–amino acid peptide might involve the removal of those internal amino acids that appeared dispensable. We tested a series of peptides with internal deletions centered around positions 44 and 45 (Fig. 2B). We found that peptide RTKQLYPEW, the nonamer that would be produced from the 13–amino acid peptide RTKAWNRQLYPEW by removal of residues AWNR (positions 43 to 46), was efficiently recognized by CTL 14 after exogenous loading. Half-maximal lysis was achieved at a concentration of 1 nM.

To confirm that nonamer RTKQLYPEW was identical to the peptide presented by melanoma cells, peptides bound to HLA class I molecules purified from autologous tumor cells were eluted and separated by high-performance liquid chromatography (HPLC). The fractions of eluted peptides that stimulated the CTL were the same as those observed after running the synthetic nonamer under the same HPLC conditions (Fig. 2C).

To gain insight into the peptide-splicing mechanism, we considered the possible involvement of the proteasome, which is known to produce most antigenic peptides presented by HLA class I molecules (2). Two proteasome inhibitors, lactacystin and epoxomicin, prevented the recognition of target cells electroporated with the 13–amino acid peptide, indicating that proteasome activity was required for processing of this precursor peptide (fig. S1). The 13–amino acid peptide was then incubated with highly purified 20S proteasomes. The digests strongly stimulated production of interferon-γ (IFNγ) by CTL 14 (Fig. 3A). This was not the case when lactacystin was included in the digestion mixture. Thus, the proteasome produced the antigen in vitro, presumably by cleaving the precursor peptide and splicing two noncontiguous fragments. To confirm that the antigen produced by the proteasome was nonamer RTKQLYPEW, we separated the digest under the same HPLC conditions as before and tested the fractions with the CTL (Fig. 2C, bottom). The peptide recognized by the CTL in the digest appeared in the same fraction as the synthetic nonamer RTKQLYPEW. We then identified the peptide fragments present in the digests by HPLC coupled to mass spectrometry (MS). The observed fragments indicated cleavage after residues 42, 43, and 46 (Fig. 3B). The cleavages after 42 and 46 correspond to the boundaries of AWNR, the presumed excised fragment. Nonamer RTKQLYPEW has an expected mass-to-charge ratio (m/z) of 610.8. Only a faint signal was detected at m/z 610.8, but when this signal was further analyzed by tandem MS (MS/MS), it was identified as RTKQLYPEW on the basis of its fragmentation pattern, which was identical to that of the corresponding synthetic peptide (Fig. 3C). Thus, the proteasome appears to produce peptide RTKQLYPEW by excision and splicing, albeit at a very low efficiency. A comparison of the level of CTL activation obtained with the digests or with a titration curve of the synthetic nonamer provided a rough estimate of one spliced peptide produced from about 104 molecules of precursor peptide. The extreme sensitivity of CTL, which can recognize target cells presenting fewer than 10 molecules of antigenic peptide, probably explains why such a low-efficiency process leads to lysis of melanoma cells by CTL 14 (7, 8).

Fig. 3.

Role of 20S proteasomes in peptide splicing. (A) Production of the spliced peptide by incubation of 20S proteasomes in vitro with the 13–amino acid precursor RTKAWNRQLYPEW. Digests were pulsed onto LG2-MEL-220 target cells and tested for recognition by CTL 14. (B) Main peptide fragments detected by mass spectrometry after a 120-min digestion of the 13–amino acid precursor by 20S proteasomes. The lines indicate the various fragments observed, and their thickness is related to the amount of peptide observed. The experimental conditions did not allow the detection of fragments RTK and RTKA. (C) MS/MS fragmentation spectrum of the doubly charged ion with m/z 610.82+ observed in the digest obtained after a 60-min incubation of 20S proteasomes with 13–amino acid peptide RTKAWNRQLYPEW (top), and fragmentation spectrum of synthetic nonamer RTKQLYPEW (bottom). The fragments that were detected are indicated above the peptide sequence for N-terminal ions and below for C-terminal ions. (D) MS/MS fragmentation spectrum of the doubly charged ion with m/z 646.32+ observed in the same digest as in (C) (top) and fragmentation spectrum of synthetic decamer RTKAQLYPEW (bottom).

Presumably, the proteasome produces the antigenic peptide by cleaving the precursor peptide and catalyzing the formation of a peptide bond between two distant fragments. Because the production of the antigenic peptide occurs in vitro in the absence of exogenous adenosine 5′-triphosphate, the energy required to form the new peptide bond must be recovered from one of the bonds cleaved by the proteasome. This could occur by transpeptidation as follows: Cleavage by the proteasome is known to occur by nucleophilic attack of the peptide bond by the catalytic threonine, resulting in the formation of an acyl-enzyme intermediate in which a peptide fragment is attached by an ester bond to the catalytic threonine (9). Thus, the observed cleavage of precursor RTKAWNRQLYPEW after residue 42 produces an intermediate that comprises fragment RTK attached to the catalytic threonine (Fig. 4A). Normally, this intermediate is rapidly hydrolyzed. However, inside the confined catalytic chamber of the proteasome, the intermediate is surrounded by peptide fragments, such as QLYPEW resulting from cleavage after residue 46. The N terminus of this fragment could compete with water molecules and occasionally make a nucleophilic attack of the ester bond of the intermediate, thereby forming a new peptide bond and producing the antigenic peptide RTKQLYPEW (Fig. 4A).

Fig. 4.

Mechanism of peptide splicing. (A) Model of the peptide splicing reaction. (B) Various synthetic peptides were combined in a pairwise manner and incubated with 20S proteasomes. Digests were tested for recognition by CTL 14. Mass spectrometry confirmed the presence of RTKQLYPEW in the digests recognized by the CTL and its absence in the others. Ac-QLYPEW, N-α-acetylated peptide QLYPEW.

To test this model, we first incubated proteasomes with two distinct peptides each containing a different portion of the precursor peptide, RTKAWNR and AWNRQLYPEW. The digests were recognized by CTL 14 almost as efficiently as were the digests obtained with the 13–amino acid precursor (Fig. 4B). We then incubated proteasomes with peptides RTK and QLYPEW and observed that the antigenic peptide was not produced (Fig. 4B). This confirmed that splicing required recycling of the energy of a peptide bond. We also observed that the antigenic peptide was produced after incubation with RTKAWNR and QLYPEW but not with RTK and AWNRQLYPEW (Fig. 4B). This indicated that the energy of the new peptide bond was recycled from the bond between K42 and A43 and not from that between R46 and Q47. This supported the model of formation of an acyl-enzyme intermediate between RTK and the proteasome, which is attacked by fragment QLYPEW. That the N terminus of the latter fragment was indeed responsible for this nucleophilic attack was supported by the fact that its N-α-acetylation completely prevented the production of the antigenic peptide (Fig. 4B).

This model predicts that splicing does not depend on a particular sequence motif, but could occur with any fragment produced by the proteasome. We therefore examined whether the digests contained other spliced peptides. The other cleavage site of the 13–amino acid peptide occurred after residue 43, producing fragment RTKA (Fig. 3B). We searched the digests for the presence of peptide RTKAQLYPEW, which would result from splicing of RTKA with QLYPEW. A signal detected at the expected m/z for this peptide was identified by MS/MS as RTKAQLYPEW (Fig. 3D).

Protein splicing has been described in unicellular organisms as the autocatalytic excision of segments named inteins, which are up to several hundred amino acids in length (10). The peptide splicing described here differs in several respects. It affects much smaller segments and is not autocatalytic. Rather, it is catalyzed by the proteasome and therefore takes place during protein degradation.

Peptide identification efforts have provided many examples of antigenic peptides that do not simply correspond to fragments of conventional proteins, but rather result from aberrant transcription, incomplete splicing, translation of alternative or cryptic open reading frames, or post-translational modifications (1116). Peptide splicing is another mechanism that increases the diversity of antigenic peptides presented to T cells. It represents a new aspect of the proteasome function in antigen processing.

Supporting Online Material

Materials and Methods

Fig. S1


References and Notes

View Abstract

Navigate This Article