|Home | About | Journals | Submit | Contact Us | Français|
Dysregulated protein phosphorylation is a hallmark of malignant transformation. Transformation can generate major histocompatibility complex (MHC)-bound phosphopeptides that are differentially displayed on tumor cells for specific recognition by T cells. To understand how phosphorylation alters the antigenic identity of self peptides and how MHC class II molecules present phosphopeptides for CD4+ T cell recognition, we determined the crystal structure of a phosphopeptide derived from melanoma antigen recognized by T cells-1 (pMART-1), selectively expressed by human melanomas, in complex with HLA-DR1. The structure revealed that the phosphate moiety attached to the serine residue at position P5 of pMART-1 is available for direct interactions with T cell receptor (TCR), and that the peptide N-terminus adopts an unusual conformation orienting it toward TCR. This structure, combined with measurements of peptide affinity for HLA-DR1 and of peptide–MHC recognition by pMART-1-specific T cells, suggests that TCR recognition is focused on the N-terminal portion of pMART-1. This recognition mode appears to be distinct from that of foreign antigen complexes but is remarkably reminiscent of the way autoreactive TCRs engage self or altered self peptides, consistent with the tolerogenic nature of tumor–host immune interactions.
A variety of post-translational modifications of naturally processed peptides displayed by major histocompatibility complex (MHC) class I or II molecules have now been described, including glycosylation, deamidation, cysteinylation and phosphorylation.1 Furthermore, peptides bearing these post-translational modifications can be discriminated from their unmodified homologs by T cells. Among the post-translationally modified peptides identified to date, phosphopeptides2-4 are of particular interest since dysregulated phosphorylation is one of the hallmarks of malignant transformation and contributes directly to oncogenic signaling cascades involved in cell growth, differentiation and survival.5,6 Indeed, phosphopeptides directly isolated from the human MHC class I molecule HLA-A2 include those derived from proteins involved in oncogenic signaling and cell cycle regulation.2,3 Because many of these phosphopeptides are differentially displayed on cancer cells,3 they provide a new cohort of targets for cancer immunotherapy.1,7
Until very recently, it was unknown whether MHC class II molecules could present phosphopeptides for specific recognition by CD4+ T cells, whose activation and recruitment are critical for the development of effective and long-lasting anti-tumor immunity.8,9 In one study, mass spectrometric sequencing was used to demonstrate the existence of HLA-DR-associated phosphopeptides on two pairs of autologous human melanoma and Epstein-Barr virus (EBV)-transformed B lymphoblastoid lines.4 The 150 unique phosphopeptides identified derived from 53 different source proteins representing all cellular compartments. As characteristic of nonphosphorylated MHC class II-restricted epitopes, most of the phosphopeptides occurred in nested sets, and their average length was 16 amino acids (range 8–28). Significantly, the majority of source proteins support vital cellular functions, such as metabolism, cell cycle regulation and signal transduction.4 Similar results were reported in a separate study of human MHC class II-restricted phosphopeptides derived from one melanoma and one B lymphoblastoid cell line,10 suggesting the generality of phosphopeptide presentation by MHC class II molecules.
The ability of human CD4+ T cells to specifically recognize MHC class II-restricted phosphopeptides was first demonstrated using as an example an HLA-DR1-restricted phospho-MART-1 (melanoma antigen recognized by T cells-1; also known as Melan-A) peptide (pMART-1100–111; APPAYEKLpSAEQ, where pS is phosphoserine) that was isolated from a cultured melanoma line.4 MART-1 is of special interest because its selective expression by cells of the melanocytic lineage has made it a prime target for immunotherapeutic approaches to melanoma, including vaccines and adoptive T cell transfer.11,12 CD4+ T cells recognizing pMART-1 presented by HLA-DR1 were highly specific for the phosphate moiety of the peptide. Importantly, these T cells recognized intact melanoma cells expressing MART-1 and HLA-DR1, indicating the presence of sufficient quantities of pMART-1 peptide–MHC complexes at the cell surface to trigger T cell signaling.
Establishing the molecular basis for phosphopeptide presentation and recognition will enable the rational design of new cancer immunotherapies targeting this category of tumor-derived epitopes. Recent crystal structures of several phosphopeptide–HLA-A2 complexes showed that the phosphate moiety formed an integral part of these structures, stabilizing interactions with the MHC class I molecule.13,14 Here we report the first structure of a phosphopeptide–MHC class II complex, involving pMART-1100–114 bound to HLA-DR1. This structure, in conjunction with measurements of peptide–MHC affinity and T cell recognition of truncated and substituted pMART-1 peptides, reveals the basis for presentation of tumor-associated MHC class II-restricted phosphopeptides to CD4+ T cells.
To understand how phosphorylation influences antigenic identity, we determined the crystal structure of pMART-1100–114 (APPAYEKLpSAEQSPP) in complex with HLA-DR1 to 2.1 Å resolution (Table 1; Fig. 1a). This peptide belongs to a nested set of four phosphopeptides derived from the C-terminus of MART-1 that range in length from 12 to 17 residues, all starting at Ala100. The nested peptides were eluted from HLA-DR-peptide complexes isolated from a human melanoma line expressing a single DR molecule (HLA-DRβ1*0101), thereby assuring unambiguous HLA allele restriction, as well as from a second melanoma line expressing HLADRβ1*0101, β1*0404 and β4*0103.4 Phospho-MART-1 peptides were not found on autologous EBV-transformed B cells, consistent with the selective expression of MART-1 in melanocytic lineage cells. The choice of pMART-1100–114 for structural studies was based on its higher affinity for HLA-DR1 (IC50 = 102 nM) compared to three other naturally-processed pMART-1 peptides: pMART-1100–111 (459 nM), pMART-1100–115 (142 nM), and pMART-1100–116 (327 nM). This facilitated assembly of stable pMART-1–HLA-DR1 complexes for crystallization. The affinities of pMART-1 peptides for HLA-DR1 are considered to be in the intermediate range, as defined by the IC50 values of non-phosphorylated peptides for HLA-DR molecules.15
Continuous and unambiguous electron density was observed for the entire 15-residue pMART-1100–114 peptide, including the N- and C-termini and the phosphate moiety attached to Ser108 (Fig. 1b). The phosphopeptide adopts an extended conformation in the HLA-DR1 binding groove resembling that of non-phosphorylated peptides bound to MHC class II molecules (Fig. 1c). The primary anchor residues are Tyr104 (P1) and Ala109 (P6), in agreement with the requirement for an aromatic or large hydrophobic residue at P1 and a small uncharged residue at P6 for efficient binding to HLA-DR1.15,16 The secondary anchors are Leu107 (P4), Glu110 (P7), and Ser112 (P9). This binding register is likely to be maintained in nonphosphorylated counterparts of pMART-1100–114, since the side chain of the phosphorylated Ser108 residue at P5 projects into solvent, away from the peptide-binding groove of HLA-DR1 (Fig. 1a,c).
The conformation of pMART-1100–114 was directly compared with those of peptides from other peptide–MHC class II complexes16-20 by superposition of the α1/β1 domains of the class II molecules (Fig. 2). An unusual feature of the phospho-MART-1 peptide is that N-terminal residues P-3 and P-2, which lie outside the binding groove, are well ordered in the structure (Fig. 1b), despite the absence of crystal contacts to the N-terminus. In other peptide–MHC class II complexes, by contrast, these residues (when present) are generally not clearly defined in the electron density, indicating flexibility. Another notable feature of pMART-1 is that its main chain makes a nearly 90° turn at P-2 Pro102, causing P-4 Ala100 and P-3 Pro101 to point upward from the α1/β1 platform of HLA-DR1 (Fig. 2). This orientation suggests the possibility of TCR contacts to the ordered N-terminus of the peptide (Fig. 1b), as discussed later.
No direct contacts were observed between the phosphate moiety of P5 pSer108 and the phospho-MART-1 peptide or HLA-DR1. However, P5 pSer108 is stabilized by ordered water molecules that form bridging hydrogen bonding interactions (Fig. 1d). In particular, the P5 pSer108 O2P atom forms a water-mediated hydrogen bond with the main chain oxygen atom of P3 Lys106. Similarly, the P5 pSer108 O1P atom interacts with the Nδ2 atom of HLA-DR1 Asn62α via two bridging water molecules. By contrast, the reported phosphopeptide–HLA-A2 complexes include multiple direct contacts between the phosphate group and the α1 and α2 helices of the MHC class I molecule, in addition to water-mediated interactions.13,14
Phosphorylated MART-1 peptides bind HLA-DR1 with 2–4-fold lower affinity than their non-phosphorylated counterparts (Fig. 3). This modest reduction may be attributed to electrostatic repulsion between the negatively charged phosphate group of P5 pSer108 and the nearby carboxylate group of P7 Glu110 (Fig. 1d). In phosphopeptide–HLA-A2 complexes, by comparison, phosphorylation was sometimes (but not always) found to increase affinity for MHC class I,13,14 suggesting that certain MHC class II-restricted phosphopeptides might also display phosphorylation-dependent binding to MHC.
The phosphate group of P5 pSer is fully exposed to solvent and therefore potentially available for direct interactions with TCR. Indeed, the phosphate is a critical determinant for TCR recognition, as demonstrated using a CD4+ T cell clone (D7-F6) specific for pMART-1 (Fig. 3). These cells secreted the cytokine granulocyte-macrophage colony-stimulating factor (GM-CSF) in response to phosphorylated MART-1 peptides pulsed onto DR1-expressing antigen-presenting cells (APCs), but not to their non-phosphorylated homologs.
Having established that phosphorylation is essential for TCR recognition, we next examined recognition of a series of synthetic overlapping phospho-MART-1 peptides, in order to identify other determinants of T cell reactivity (Fig. 4a,b). Similar to pMART-1100–116, a series of naturally occurring C-terminally truncated phosphopeptides (pMART-1100–115, pMART-1100–114, pMART-1100–111) and the synthetic peptide pMART-199–110 were also capable of stimulating T cell reactivity, indicating that residues Gln111 (P8) through Tyr116 (P13) are not involved in critical interactions with TCR. In the pMART-1100–114–HLA-DR1 structure, P8 Gln111 and P11 Pro114 (like P5 pSer108) are ~70% accessible to solvent and therefore potentially available for contacts to TCR, whereas P9 Ser112 and P10 Pro113 are ~80% buried (P9 Ser112 is a secondary anchor residue) and unlikely to contact TCR directly. Indeed, removal of P9 Ser112 and P10 Pro113 (compare pMART-1100–114 with pMART-1100–111) resulted in a 4-fold loss of affinity for HLA-DR1 but a paradoxical increase in T cell activation (Fig. 4b). Residues P12 Pro115 and P13 Tyr116, although absent from the pMART-1100–114–HLA-DR1 structure, are expected to extend beyond the peptide-binding groove.
In sharp contrast to C-terminal residues P8 to P13, N-terminal residues P-4 to P7 contain important determinants for T cell recognition, besides the critical P5 pSer108 residue. Thus, removal of P7 Glu110 (compare pMART-199–110 with pMART-198–109) abrogated T cell stimulation, even though this truncation reduced affinity for HLA-DR1 by less than 2-fold (Fig. 4a). In the pMART-1100–114–HLA-DR1 structure (Fig. 1c), the side chain of P7 Glu110 projects from the surface of the MHC molecule and could interact with TCR. Removal of N-terminal residues P-4 Ala100, P-3 Pro101, P-2 Pro102, and P-1 Ala103 (hereafter referred to as the APPA motif) effectively abolished TCR recognition, as evident from comparing T cell stimulation by pMART-1100–114 versus pMART-1103–114, and pMART-1100–115 versus pMART-1104–115 (Fig. 4a). Since truncation of residues P-4 to P-1 did not reduce binding to HLA-DR1 (indeed, a significant increase in MHC binding affinity was noted with removal of P-4 Ala100), these four N-terminal residues must include structural determinants required for TCR recognition.
To assess the contribution of individual residues of the APPA motif to TCR recognition, P-4 Ala100 and P-1 Ala103 of pMART-1100–114 were each replaced by isoleucine, while P-3 Pro101 and P-2 Pro102 were replaced by alanine. Except for the replacement at P-4, these substitutions significantly reduced T cell stimulatory capacity compared to wild type pMART-1100–114, suggesting interactions with TCR (Supplementary Fig. 1). Most pronounced was the effect of substituting P-3 Pro101, which contributes directly to the unusual conformation of the APPA motif, whereby the peptide main chain makes a 90° turn at P-2 Pro102 directing N-terminal residues P-4 Ala100 and P-3 Pro101 towards the TCR (Fig. 2). In addition, although replacing P-4 Ala100 by the chemically similar isoleucine residue did not significantly affect T cell recognition, removal of P-4 Ala100 caused a total loss of T cell reactivity (Fig. 4a).
The marked focus on the N-terminal portion of pMART-1100–114 by TCR D7-F6 closely resembles peptide recognition by autoreactive TCRs specific for self antigens, including tumor antigens, presented by MHC class II molecules.21 In particular, X-ray crystallographic studies of autoimmune TCRs specific for myelin basic protein associated with multiple sclerosis,22,23 and of a tumor-specific TCR that recognizes a somatically mutated human melanoma antigen,24 have revealed substantial alterations in the topology of TCR binding to peptide–MHC compared to anti-foreign TCRs.21 In the self-reactive TCR–peptide–MHC complexes, the TCR is skewed toward the peptide N-terminus and the MHC class II β chain helix relative to its central position in anti-foreign TCR–peptide–MHC complexes, resulting in suboptimal TCR binding that may have enabled escape from negative thymic selection. The concentration on the N-terminal half of pMART-1 is therefore highly reminiscent of the way autoreactive TCRs engage self or altered self peptides.21
MART-1 protein plays an essential role in regulating mammalian pigmentation25 but was not known to be phosphorylated prior to the discovery of phospho-MART-1 peptides complexed with HLA-DR1.4 Tissue expression patterns of phospho-MART-1 protein versus its non-phosphorylated counterpart have yet to be determined. If MART-1 is expressed in the thymus and phosphorylation occurs after thymic development, then this post-translational modification will create a neo-epitope, most likely by increasing TCR affinity for MART-1–HLA-DR1 ligands beyond the threshold required for efficient T cell activation in the periphery. Alternatively, if pMART-1 is expressed in the thymus during T cell development, then surviving reactive TCRs are predicted to have suboptimal peptide–MHC binding characteristics. Future efforts to determine the structure of TCR D7-F6 bound to pMART-1100–114–HLA-DR1 and to measure the biophysical properties of this complex should provide relevant insights.
This work was supported by National Institutes of Health grants AI073654 (to S.L.T. and R.A.M.), AI036900 (to R.A.M.), CA134060 (to V.H.E.) and AI33993 (to D.F.H.), and by a grant from the Melanoma Research Alliance Foundation (to V.H.E., D.F.H. and S.L.T.). We thank H. Robinson (Brookhaven National Synchrotron Light Source) for X-ray data collection. Support for beamline X29 comes from the Offices of Biological and Environmental Research and of Basic Energy Sciences of the U.S. Department of Energy, and from the National Center for Research Resources of the National Institutes of Health.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Atomic coordinates and structure factors for the pMART-1100–114–HLA-DR1 complex have been deposited in the Protein Data Bank under accession code 3L6F.