An Antigen Produced by Splicing of Noncontiguous Peptides in the Reverse Order

See allHide authors and affiliations

Science  08 Sep 2006:
Vol. 313, Issue 5792, pp. 1444-1447
DOI: 10.1126/science.1130660


CD8-positive T lymphocytes recognize peptides that are usually derived from the degradation of cellular proteins and are presented by class I molecules of the major histocompatibility complex. Here we describe a human minor histocompatibility antigen created by a polymorphism in the SP110 nuclear phosphoprotein gene. The antigenic peptide comprises two noncontiguous SP110 peptide segments spliced together in reverse order to that in which they occur in the predicted SP110 protein. The antigenic peptide could be produced in vitro by incubation of precursor peptides with highly purified 20S proteasomes. Cutting and splicing probably occur within the proteasome by transpeptidation.

Most nucleated cells display on their surface a broad repertoire of peptides derived from proteasome-mediated degradation of intracellular proteins and bound to major histocompatibility complex (MHC) class I molecules, which are known in humans as human leukocyte antigen (HLA) class I molecules. Surveillance of this repertoire by CD8+ T lymphocytes allows the adaptive cellular immune system to detect and to eliminate cells containing foreign or abnormal proteins (1, 2). The search for antigens recognized by CD8+ T cells has focused on contiguous fragments of proteins expressed in malignant or infected cells. Recently, two antigenic peptides recognized by antitumor CD8+ T cells were each found to be composed of the fusion of peptide fragments of the respective parental proteins after excision of an intervening segment (3, 4). In one case, the excision and splicing reactions were shown to occur in the proteasome (4).

A CD8+ cytolytic T-lymphocyte (CTL) clone, termed DRN-7, was isolated from a recipient of MHC-matched allogeneic hematopoietic cell transplantation (HCT) (5). In this setting, donor T cells recognizing minor histocompatibility (H) antigens, which are peptides presented on recipient cells and encoded by polymorphic non-MHC genes, can cause graft-versus-host (GVH) disease and graft-versus-leukemia (GVL) reactions (6). CTL DRN-7 was found to recognize an HLA-A*0301–restricted minor H antigen expressed by hematopoietic cells and to inhibit engraftment of HLA-A*0301+ human acute myelogenous leukemia cells in nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice; these findings suggest that the minor H antigen may be a GVL target (7). To identify the gene that encodes this antigen, we screened a cDNA library constructed from the Epstein-Barr virus (EBV)–transformed B cells of the recipient. Plasmid DNA from this library was transfected into COS-7 cells, together with DNA encoding HLA-A*0301, and the transfectants were tested for their ability to stimulate CTL DRN-7. We identified a positive pool and screened DNA from individual colonies isolated from this pool. A cDNA, termed AH9C10, stimulated HLA-A*0301–dependent tumor necrosis factor (TNF) release from CTL DRN-7 (Fig. 1A). This cDNA corresponded to nucleotides 730 to 1376 of the transcript of gene SP110, which encode residues 213 to 425 of the SP110 nuclear body protein (8, 9) (Fig. 1A). Testing truncated constructs of AH9C10 for their ability to stimulate CTL DRN-7 further localized the antigen-encoding region to a 60-nucleotide (nt) interval encoding amino acids 286 to 305 of SP110 (Fig. 1A).

Fig. 1.

The A996 allele of SP110 encodes the HLA-A*0301–restricted minor H antigen recognized by CTL clone DRN-7. (A) COS-7 cells were transfected with an HLA-A*0301 cDNA and a partial SP110 cDNA, termed AH9C10, encoding residues 213 to 425 of the predicted SP110 protein, or with minigene constructs derived from AH9C10 and encoding the indicated intervals. Twenty-four hours after transfection, CTL DRN-7 was added, and TNF production was measured 24 hours later. (B) Nucleotide sequence of a fragment of the SP110 alleles carried by the transplant recipient and transplant donor. The A/G polymorphism at nucleotide 996 (bold), and the corresponding amino acid polymorphism at position 299 are shown. This polymorphism is reported in databases as rs1365776. The Bst OI restriction site used for genotyping is indicated. (C) Correlation between the presence of the A996 SP110 allele and recognition by CTL DRN-7. (Top) Lysis of 51Cr-labeled EBV–B cells from 11 representative HLA-A*0301 individuals, including the transplant recipient (DRN) and donor (CAN), by CTL DRN-7 at an effector-to-target ratio (E/T) of 10. (Bottom) Genotype of each individual at the A996G polymorphism in SP110.

Most human minor H antigens that have been defined result from nonsynonymous polymorphisms in the coding region of normal genes. Sequencing of the SP110 alleles in EBV-transformed B cell lines derived from the transplant recipient and donor identified an A/G polymorphism at nucleotide 996 in the SP110 coding sequence (Fig. 1B). This polymorphism was contained within the 60-nt interval encoding the antigen recognized by CTL DRN-7 and created an amino acid substitution of Gly for Arg at position 299 (R299G) of the SP110 protein (10). The genotype at this polymorphism was determined in EBV–B cell lines from 64 other HLA-A*0301 individuals. A perfect correlation was observed between the presence of at least one A996 allele and susceptibility to lysis by CTL DRN-7 (Fig. 1C). In the 66 individuals tested, the frequencies of the A996 and the G996 alleles were 0.54 and 0.46, respectively.

We attempted to identify the antigenic peptide encoded by the critical 60-nt interval in the A996 allele of SP110 by synthesizing a series of overlapping peptides that collectively spanned the predicted 20–amino acid fragment and by testing each one for recognition by CTL DRN-7 after loading onto donor EBV–B cells. None of these peptides sensitized EBV–B cells to lysis (table S1). Considering that a posttranslational modification might be required for antigenicity, we tried to enable this modification by introducing the 20–amino acid peptide STPKRRHKKKSLPRGTASSR inside donor EBV–B cells by electroporation (4). Cells electroporated with this peptide were strongly recognized by CTL DRN-7 (Fig. 2A), whereas cells electroporated with peptide STPKRRHKKKSLPGGTASSR, encoded by the G996 allele, were not. The putative modification involved intracellular processing, because it was not observed when the electric shock was omitted. Moreover, in experiments where intracellular loading of the precursor peptide was obtained by prolonged incubation with very high doses of the peptide, we observed that recognition of loaded cells was dependent on the expression of transporter associated with antigen processing (TAP), which indicates a cytosolic processing step (fig. S1).

Fig. 2.

Recognition by CTL DRN-7 of a peptide produced by reordering and splicing of two noncontiguous SP110 fragments. (A) The indicated peptides (200 μg/ml) were electroporated into donor CAN EBV–B cells (solid bars) or loaded exogenously (open bars). CTL DRN-7 was added and the production of IFN-γ was measured after overnight incubation. (B) 51Cr–labeled CAN EBV–B cells were loaded for 30 min with the indicated concentrations of synthetic peptides comprising noncontiguous fragments of SP110 joined together in either the same or the reverse order as that with which they appear in the SP110 protein. CTL DRN-7 was added at an E/T of 5, and lysis was measured after 4 hours. (C) Peptides were eluted from HLA class I molecules purified from EBG EBV–B cells, separated into fractions by HPLC successively on two different columns, and the fractions were tested for recognition by CTL DRN-7 (top). 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 peptides SLPRGTSTPK, SLPRGTSTPKR, and SLPRGTSTPKRR (60 pmol each), were injected under the same HPLC conditions, and the fractions were tested for CTL recognition (bottom).

The transfection of truncated cDNA constructs had indicated that both ends of the 20–amino acid fragment were required for antigenicity (Fig. 1A). This suggested that splicing of two peptide fragments contained within STPKRRHKKKSLPRGTASSR might produce the antigen. We electroporated pairs of peptides making up nonoverlapping segments of STPKRRHKKKSLPRGTASSR into donor EBV–B cells and tested the electroporated cells for recognition by CTL DRN-7. Two decameric peptides, STPKRRHKKK (residues 286 to 295) and SLPRGTASSR (296 to 305), stimulated CTL DRN-7 when electroporated simultaneously, but not when electroporated singly (Fig. 2A). Thus, the antigenic peptide was made from the splicing of two distinct fragments. To identify the minimal fragments required, we electroporated a series of peptides truncated at either end (Fig. 2A). Removal of the N-terminal serine of either peptide abrogated recognition. Sequential removal of C-terminal residues revealed that the combination of STPK and SLPRGTAS retained the ability to stimulate CTL DRN-7. We hypothesized that the antigenic peptide was a spliced product resulting from linkage of these two fragments or parts of them. We synthesized a series of peptides containing such fragments, including peptide STPKSLPRGT, and loaded them directly onto target cells at concentrations up to 10 μg/ml. CTL DRN-7 did not recognize any of these peptides (Fig. 2B and table S2).

Peptide STPKSLPRGT did not contain a good HLA-A3–binding motif, which has leucine in position two and lysine at the C terminus (11, 12). We noted that if its two constituent fragments were linked in the reverse order, the resulting peptide, SLPRGTSTPK, would contain a perfect HLA-A3–binding motif. We synthesized and loaded a peptide with this reordered sequence onto target cells, and we observed efficient recognition by CTL DRN-7, with half-maximal lysis at a peptide concentration of ∼40 pM (Fig. 2B). Peptide SLPGGTSTPK, which would be derived from the G996 allele of SP110, was only weakly recognized at high concentrations. We tested additional reordered and spliced peptides of various lengths and observed that peptide SLPRGTSTPK was the optimal peptide, even though the longer peptides SLPRGTSTPKRR and SLPRGTSTPKR were also recognized by CTL DRN-7 at slightly higher concentrations (Fig. 2B and fig. S2).

To determine whether this spliced and reordered SP110 peptide was identical to the peptide naturally presented at the cell surface, we isolated MHC class I molecules from HLA-A*0301+ EBV–B cells that were homozygous for the A996 SP110 allele. We eluted peptides with acid and fractionated them by high-performance liquid chromatography (HPLC) on a C18 column. We identified a fraction that was able to stimulate interferon-γ (IFN-γ) release by CTL DRN-7 when loaded onto target cells. When synthetic peptide SLPRGTSTPK was chromatographed under the same HPLC conditions, the same fraction stimulated CTL DRN-7. However, we observed that peptides SLPRGTSTPK, SLPRGTSTPKR, and SLPRGTSTPKRR had similar retention times on the C18 column. We then reinjected the positive fraction obtained with the eluates into a porous graphitic carbon column, which better separates such highly hydrophilic peptides. Once again, a fraction was identified that stimulated CTL DRN-7 (Fig. 2C). When synthetic peptide SLPRGTSTPK was chromatographed on this column, the same fraction stimulated CTL DRN-7 (Fig. 2C). This was not the case with synthetic peptides SLPRGTSTPKR or SLPRGTSTPKRR. Thus, synthetic peptide SLPRGTSTPK corresponded to the peptide naturally presented to CTL DRN-7.

Previous work has identified the role of the proteasome in the production of a spliced antigenic peptide (4). Pretreatment of donor EBV–B cells with the irreversible proteasome inhibitor lactacystin before electroporation of STPKRRHKKKSLPRGTASSR inhibited their ability to stimulate CTL DRN-7 (fig. S3). To directly evaluate the ability of the proteasome to produce the SP110 peptide, we incubated purified 20S proteasomes with peptide STPKRRHKKKSLPRGTASSR and examined recognition by CTL DRN-7 of target cells loaded with the digests. Digests obtained after 1 or 2 hours of incubation were strongly recognized by CTL DRN-7, which indicated that the antigenic peptide had been produced in vitro (Fig. 3A). This was not the case when the incubation was performed in the presence of lactacystin, or with peptide STPKRRHKKKSLPGGTASSR, which is encoded by the G996 SP110 allele (Fig. 3A).

Fig. 3.

Production of the antigenic peptide SLPRGTSTPK by the proteasome. (A) Recognition by CTL DRN-7 of digests obtained by incubating the indicated peptides with purified 20S proteasomes. The digests collected at the indicated time points were loaded onto CAN EBV–B cells and tested for recognition by CTL DRN-7. (B) Peptide fragments detected by MS after incubation of precursor peptide STPKRRHKKKSLPRGTASSR with 20S proteasomes for 60 min. Cleavage sites are indicated by vertical lines and those that are relevant for the production of the antigenic peptide, by arrows. The experimental conditions did not allow fragment quantification or the detection of small fragments (below four to five residues). A similar fragmentation pattern was observed at various digestion times. (C) MS/MS fragmentation spectrum of the doubly charged ion with m/z 5222+ observed in the digest obtained after a 180-min incubation of 20S proteasomes with peptides STPK and SLPRGTASSR (top), and fragmentation spectrum of the doubly charged ion (m/z 5222+) of the synthetic decamer SLPRGTSTPK (bottom). The fragments that were detected are indicated above the peptide sequence for N-terminal b or c ions and below for C-terminal y ions. Ion with m/z 5042+ is a doubly dehydrated derivative of ion 5222+.

We used HPLC coupled to mass spectrometry (MS) to identify the fragments present in the digests (Fig. 3B). This analysis revealed at least eight cleavages in the peptide, including the three cleavages required to liberate fragments STPK and SLPRGT and to allow their splicing to produce the antigenic peptide SLPRGTSTPK. This peptide was not detected by HPLC-MS in those digests. However, by using HPLC combined with tandem mass spectrometry (MS/MS), we identified the antigenic peptide SLPRGTSTPK in a digest obtained by incubation of proteasomes with peptides STPK and SLPRGTASSR. This digest was also strongly recognized by CTL DRN-7 (Fig. 3A). The antigenic peptide was detected as a doubly charged ion with m/z 522, whose retention time and fragmentation pattern were identical to those of the corresponding synthetic peptide (Fig. 3C). Thus, the proteasome can perform the splicing reaction required to produce antigenic peptide SLPRGTSTPK.

Our prior study of a spliced peptide indicated that splicing occurred inside the proteasome by transpeptidation involving an acyl-enzyme intermediate (4). This intermediate, which is transiently formed on the hydroxyl group of the side chain of the catalytic threonine, is rapidly hydrolyzed during proteolysis. In the splicing reaction, the N-terminal group of the other peptide fragment competes with water molecules to perform a nucleophilic attack of the ester bond of the intermediate, which results in a transpeptidation reaction producing the spliced peptide. In theory, this mechanism could also allow for a reordering of the peptide fragments before their ligation (Fig. 4A). The acyl-enzyme intermediate would involve fragment SLPRGT; fragment STPK would be first liberated by hydrolysis and would attack the ester bond of the intermediate with its amino group. If this model accounts for the production of the SP110 peptide, the energy required to create the new peptide bond of the spliced peptide should be recovered, through the ester bond of the intermediate, from the energy liberated by the cleavage between Thr301 and Ala302. A proteasomal digest of peptides STPK and SLPRGT, which correspond to the final fragments of the spliced peptide and would not need any additional cleavage, failed to produce the antigenic peptide, which confirmed the need to recover the energy of a cleaved bond (Fig. 4B). In contrast, a digest involving STPK and SLPRGTASSR did produce the antigen, whereas a digest of STPKRR and SLPRGT did not (Fig. 4B). Thus, cleavage of the bond between Thr301 and Ala302 is necessary to provide the energy required to create the new bond of the spliced peptide. Acetylation of the N-terminal group of STPK prevented its ability to produce the antigenic peptide when digested with SLPRGTASSR (Fig. 4B). Similarly, acetylation of the N terminus of the single peptide STPKRRHKKKSLPRGTASSR abolished its ability to produce the antigenic peptide (Fig. 4C). Thus, consistent with our model, a free N-terminal group is required on fragment STPK to perform the nucleophilic attack of the acylenzyme intermediate.

Fig. 4.

Mechanism of peptide splicing. (A) Model of the splicing reaction inside the proteasome. The balls represent the catalytically active β-subunits of the proteasome with the hydroxyl group of the side chain of the N-terminal threonine. (B) Various synthetic peptides were combined in a pairwise manner and incubated with 20S proteasomes. Digests were tested for recognition by CTL DRN-7. Ac-STPK, N-α-acetylated peptide STPK. (C) The indicated precursor peptide or its N-α-acetylated derivative were incubated with 20S proteasomes. Digests were tested for recognition by CTL DRN-7. Results are expressed as a function of the percentage of degradation of the precursor peptide, as measured by HPLC-MS. Digestion times were 0, 120, 240, and 360 min.

Our results indicate that the reordered spliced peptide SLPRGTSTPK derived by proteasomal processing of the Arg299 SP110 protein is the naturally processed antigen recognized by CTL DRN-7 (supporting online text) and that the splicing reaction occurs in the proteasome by transpeptidation involving an acyl-enzyme intermediate. In contrast to the two previous examples of spliced peptides (3, 4), the peptide recognized by CTL DRN-7 is produced by ligation of two noncontiguous peptide fragments in the reverse order. The observation that this antigen is expressed in some normal cells (5) indicates that peptide splicing is not restricted to tumor cells. The proteasome, by virtue of its proteolytic capacity, participates in the generation of active transcription factor domains from inactive precursors (1317), controls the levels of numerous regulatory proteins, and serves as the major source of peptides recognized by CD8+ T cells. The ability of the proteasome to splice together peptide fragments from a protein in either the initial or reverse order has profound implications for the diversity of peptides that can be presented on the cell surface for recognition by CD8+ T cells and could also have other un-anticipated consequences.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3

Tables S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article