MHC class I downregulation represents a significant challenge for successful T cell–based immunotherapy. T cell epitopes associated with impaired peptide processing (TEIPP) constitute a novel category of immunogenic Ags that are selectively presented on transporter associated with Ag processing–deficient cells. The TEIPP neoepitopes are CD8 T cell targets, derived from nonmutated self-proteins that might be exploited to prevent immune escape. In this study, the crystal structure of H-2Db in complex with the first identified TEIPP Ag (MCLRMTAVM) derived from the Trh4 protein has been determined to 2.25 Å resolution. In contrast to prototypic H-2Db peptides, Trh4 takes a noncanonical peptide-binding pattern with extensive sulfur–π interactions that contribute to the overall complex stability. Importantly, the noncanonical methionine at peptide position 5 acts as a main anchor, altering only the conformation of the H-2Db residues Y156 and H155 and thereby forming a unique MHC/peptide conformer that is essential for recognition by TEIPP-specific T cells. Substitution of peptide residues p2C and p5M to the conservative α-aminobutyric acid and norleucine, respectively, significantly reduced complex stability, without altering peptide conformation or T cell recognition. In contrast, substitution of p5M to a conventional asparagine abolished recognition by the H-2Db/Trh4-specific T cell clone LnB5. We anticipate that the H-2Db/Trh4 complex represents the first example, to our knowledge, of a broader repertoire of alternative MHC class I binders.
Major histocompatibility complex class I (MHC-I) molecules selectively bind peptides derived from endogenously expressed proteins and present them at the cell surface to CD8+ CTLs and NK cells, enabling immune recognition and elimination of infected or cancerous cells. Ag processing and presentation include degradation of intracellular proteins by the multicatalytic proteasome into a repertoire of peptides that are further trimmed in the cytosol by several different proteases (1). A selection of peptides is thereafter actively transported into the endoplasmic reticulum (ER) by the transporter associated with Ag processing (TAP) complex, where they can be further trimmed by the ER aminopeptidase associated with Ag processing. Finally, peptides are loaded onto MHC-I by an ensemble of proteins including tapasin, calnexin, calreticulin, and ERp57 (2), and MHC-I/peptide complexes (pMHC) are thereafter exported to the cell surface via the Golgi network (3).
The efficacy of CTL responses toward tumors depends on successful Ag processing and MHC-I presentation of tumor-associated Ags by malignant cells. At least five classes of tumor-associated Ags have been identified, including mutated Ags, Ags encoded by cancer-germline genes, peptides derived from oncogenic viruses, differentiation Ags, and Ags aberrantly over expressed in tumors (4). However, dysregulation of MHC-I Ag processing, often as a result of defects in the function of TAP, tapasin, and/or proteasome, appears to be the major mechanism for downregulation of MHC-I on the surface of tumors, which can reduce CTL responses and ultimately result in tumor progression (5, 6).
A unique category of CTL that can prevent tumor escape by targeting an alternative repertoire of pMHC at the surface of TAP-deficient cells has been previously identified (7–11). Although derived from nonmutated endogenous proteins, the T cell epitope associated with impaired peptide processing (TEIPP) neoantigens are not presented on normal cells (11) and act as immunogenic epitopes. The induction of TEIPP-specific CTL responses resulted in selective eradication of TAP-deficient tumors in vivo (7, 11), suggesting the possibility to combine a TEIPP-specific CTL repertoire with a conventional CD8+ CTL immunotherapy strategy to prevent tumor immune escape. It has been recently demonstrated that impairment of TAP function, as commonly found in cancers and virus-infected cells, lowers the otherwise naturally occurring resistance for peptides from alternative processing routes, allowing for MHC-I presentation of other peptide sources (12, 13).
The first described TEIPP Ag (MCLRMTAVM), derived from the C-terminal portion of a minor splice variant of the commonly expressed cellular protein Trh4, was identified using synthetic peptide library screens and bioinformatics (11, 14). Hence its processing does not require proteasome cleavage, nor transport by TAP. Indeed, the endogenous protein Trh4 is ubiquitously present in a broad range of cells, and the Trh4-derived TEIPP peptide (hereafter named Trh4) used in the current study is directly liberated by the signal peptide peptidase in the ER membrane for MHC-I loading in the ER (15). Although this process also takes place in cells with intact TAP function, Trh4 is only presented by MHC-I on cells that harbor impaired TAP function (11).
Most of the hitherto identified canonical epitopes presented on H-2Db are 9–11 aa long with an asparagine at position 5 and a cysteine, a methionine or a leucine at the C-terminal position, acting as main anchor positions. The main anchor asparagine p5N usually forms forklike hydrogen bonds with the H-chain residue Q97 (16, 17). In contrast, the sequence of the Trh4 peptide epitope does not conform to the conventional H-2Db binding motif, with several sulfur-containing residues: methionines at positions 1, 5, and 9 and a cysteine at position 2. The crystal structure of H-2Db in complex with Trh4 reveals that this noncanonical epitope makes use of extensive sulfur–π interactions, stabilizing with higher efficiency H-2Db compared with prototypic peptides such as the immunodominant lymphocytic choriomeningitis virus (LCMV)–derived peptide gp33 (18–20). Similarly to p5N in canonical H-2Db peptides, the methionine residue p5M in Trh4 acts as a main anchor. However, the different conformation of p5M in Thr4 alters the conformation of the H-2Db residues Y156 and H155 and creates a unique MHC/peptide conformer that is essential for recognition by TEIPP-specific T cells. Substitution of p5M to a conventional asparagine abolished recognition by the H-2Db/Trh4-specific T cell LnB5, indicating the importance of the interplay between p5M and H155 for adequate TEIPP T cell recognition.
Materials and Methods
Preparation and crystallization of the H-2Db/Thr4 MHC-I complex
The peptide Trh4 (MCLRMTAVM) as well as the norleucine (NLE; MCLR-NLE-TAVM; Trh4-p5NLE) and the α-aminobutyric acid (ABU; M-ABU-LRMTAVM; Trh4-p2ABU) altered peptide variants of Trh4 were purchased from Genscript (Piscataway, NJ). Refolding and purification of the H-2Db/Trh4, H-2Db/Trh4-p5NLE, and H-2Db/Τrh4-p2ABU complexes were performed according to previously published protocols (20–23). All crystals were obtained by using the hanging drop vapor diffusion method. The best crystal for H-2Db/Trh4 appeared in 1.9 mol ammonium sulfate and 0.1 mol Tris-HCl (pH 7.5) at room temperature, for H-2Db/Trh4-p5NLE in 1.6 mol ammonium sulfate, 0.1 mol Tris-HCl (pH 8), and 0.5 mol NaCl in 4°C, and for H-2Db/Trh4-p2ABU in 1.5 mol ammonium sulfate, 0.1 mol Tris-HCl, and 0.5 mol NaCl (pH 7.5) in 4°C. For H-2Db/Trh4, 4 μl 1.8 mg/ml protein solution in 20 mmol Tris-HCl (pH 7), mixed with 2 μl crystallization reservoir solution, was equilibrated against 1 ml well solution at 20°C. For H-2Db/Trh4-ABU and H-2Db/Trh4-NLE, 2 μl 4.35 and 2.55 mg/ml protein solution, respectively, both in 20 mmol Tris-HCl and 150 mmol NaCl (pH 8), were mixed with 1.7 μl reservoir solution.
Data collection and processing
Data collection for H-2Db/Trh4, H-2Db/Trh4-p2ABU, and H-2Db/Trh4-p5NLE was performed under cryogenic conditions (temperature 100 K) at beam lines ID14-2 (European Synchrotron Radiation Facility [ESRF], Grenoble, France), MX 14.1 (Bessy, Helmholtz-Zentrum Berlin, Germany), and ID30A-1 through MXPressE automatic data collection (ESRF) to 2.25, 1.98, and 2.0 Å resolution, respectively. Crystals were soaked in a cryoprotectant solution containing 20% glycerol before freezing. A total of 480 images were collected with 0.5° oscillation per frame for H-2Db/Trh4. A total of 1050 images were collected for H-2Db/Trh4-p5NLE, and 2000 H-2Db/Trh4-p2ABU images were collected with 0.15° and 0.1° oscillation per frame, respectively. Data were processed with MOSFLM (24) and AIMLESS (25) from the CCP4 suite. Although the H-2Db/Trh4 MHC complex crystallized in the monoclinic space group P21, both H-2Db/Trh4-p2ABU and H-2Db/Trh4-p5NLE crystallized in the space group I2 with similar unit cell parameters. Data collection statistics for all MHC complexes are provided in Supplemental Table I.
Crystal structure determination and refinement
The crystal structures of H-2Db/Trh4, H-2Db/Trh4-p5NLE, and H-2Db/Trh4-ABU were determined by molecular replacement using Phaser (26) and the H-2Db/gp33 complex with the peptide omitted (Protein Data Bank code 1S7U) (18, 19) as a search model. The program Phaser determined four molecules in the asymmetric unit for H-2Db/Trh4 and only two molecules for H-2Db/Trh4-ABU and H-2Db/Trh4-p5NLE. Five percent of the total amount of reflections was set aside for monitoring refinement by Rfree. Refinement of the crystal structures was performed using REFMAC5 (27). A clearly interpretable electron density was observed in the peptide binding clefts, and the peptides could be unambiguously modeled in all MHC complexes within the asymmetric units. Water molecules were added using COOT (28) and their position inspected manually. The stereochemistry of the final models was verified with COOT. The final refinement parameters are presented in Supplemental Table I. Figures were created using the program PyMOL (29).
Peptide/MHC cell-surface stabilization assays
RMA-S cells were incubated with 10−6 mol each peptide at 26°C for 12 h in 5% CO2. Cells were thereafter incubated at 37°C for 1 h, then washed twice with PBS, and resuspended in prewarmed (37°C) peptide and serum free AIM-V medium containing 5 μg/ml brefeldin A (Sigma-Aldrich). Cells, collected at time points 1, 2, 3, 4, and 6 h, were stained with anti-H-2Db (KH95), washed with PBS, and fixed using 1% paraformaldehyde. Cell-surface expression of H-2Db was determined using a BD FACSCalibur (BD Biosciences). The mean fluorescence intensity (MFI) of H-2Db expression for the indicated peptide concentrations was subtracted from the observed MFI on cells without peptide and then was divided by the observed MFI on cells without peptide as an estimate of peptide expression ([MFIpeptide − MFIcells]/MFIcells). The maximum peptide expression value for each peptide was defined as 100% MHC/peptide expression level at the cell surface. Data were analyzed using Cell Quest Pro (BD Biosciences). The HIV-derived H-2Dd–restricted epitope P18-I10 (RGPGRAFVTI) (22, 30) and the H-2Db–restricted LCMV-derived immunodominant peptide gp33 (KAVYNFATM) (18) were used as a negative and positive controls, respectively.
Thermostability measurements using circular dichroism
Circular dichroism measurements were performed in 20 mmol K2HPO4/KH2PO4 (pH 7.5) using protein concentrations between 0.15 and 0.25 mg/ml (31). Spectra were recorded with a JASCO J-810 spectropolarimeter (JASCO Analytical Instruments, Great Dunmow, U.K.) equipped with a thermoelectric temperature controller in a 2-mm cell. pMHC denaturation was measured between 30 and 70°C at 218 nm with a gradient of 48°C/h at 0.1°C increments and an averaging time of 8 s. The melting curves were scaled from 0–100% and the melting temperature (Tm) values extracted as the temperature at 50% denaturation. Curves and Tm values are an average of at least three measurements from at least two independent refolding assays/complex. Spectra were analyzed and figures created using GraphPad Prism 6 (GraphPad Software). The curves were made using nonlinear sigmoidal fit to the scaled denaturation data.
T cell recognition assays
The CD8+ T cell clone LnB5, specific for H-2Db/Trh4, was generated and cultured as described previously (15). The LnB5 T cell clone (3 × 103) was mixed overnight in 100 μl culture medium in the presence of titrated concentrations of Trh4–peptide variants and freshly isolated splenocytes (50 × 103). The following day, levels of IFN-γ in the supernatants were analyzed by sandwich ELISA, as described before (15). The presented data represent mean values obtained from triplicate test wells, and error bars represent SD of these values.
The overall structure of H-2Db/Trh4 reveals a noncanonical binding pattern
The crystal structure of H-2Db in complex with the TEIPP epitope Trh4 (MCLRMTAVM) was determined to 2.25 Å resolution. Details of data collection and additional indicators of the quality of the crystal structure are provided in Supplemental Table I. The overall three-dimensional architecture of the H-2Db/Trh4 complex is typical of classical MHC-I molecules (Fig. 1A), and comparison with several other H-2Db/peptide complexes revealed that the spatial positioning of the different subdomains that comprise the MHC complex is similar, as illustrated by an overall root mean square deviation of the peptide binding cleft (residues 1–175) of H-2Db/Trh4 with the prototypic H-2Db/gp33 (18, 19, 21) of 0.45 Å. The electron density map of the peptide-binding cleft of H-2Db/Trh4 was of good quality allowing unambiguous positioning of the bound peptide (Fig. 1A), except for the exposed side chain of the arginine residue at position 4 (p4R), which adopted several possible conformations.
At first sight, the peptide Trh4 binds to H-2Db in a similar manner to classical peptides, forming hydrogen bonds with several H-2Db residues along the cleft as well as conserved hydrogen bond networks between the N and C termini and H-2Db residues in pockets A and F, respectively (Supplemental Fig. 1). However, in contrast to all previously reported H-2Db epitopes, Trh4 does not contain an asparagine at position 5 and comprises an unusually large number of sulfur-containing residues at positions 1, 2, 5, and 9. The Trh4 peptide makes primary use of positions p2C, p5M, and p9M for binding, interacting with specific aromatic amino acids in H-2Db (Fig.1B). The use of a cysteine and methionine as secondary and main anchor residues at positions 2 and 5, respectively, has to our knowledge not been previously described. Finally, residues p4R and p6T protrude toward the solvent, readily available for interactions with TCRs (Fig. 1A).
Extensive sulfur–π interactions are formed between Trh4 and H-2Db; implications for TCR recognition
The N terminus of Trh4 forms hydrogen bonds with H-chain residues Y7, Y159, and Y171. The side chain of residue p1M forms van der Waals and sulfur–π interactions with the aromatic ring of H-2Db residue W167 (Fig. 1C). The side chain of peptide residue p2C protrudes into the B-pocket of H-2Db, composed by residues Y7, E9, S24, Y45, and E63. It has been previously demonstrated that the negatively charged residue E63 plays an important role in defining the characteristics of the B-pocket in H-2Db, selecting for relatively smaller residues such as alanine and serine and preventing binding of epitopes with negatively charged residues at position 2 (32). Although the main chain of p2C forms hydrogen bonds with E63 and K66 as well as van der Waals interactions with Y159, the sulfhydryl group of p2C forms sulfur–π interactions with the phenol ring of Y45. Furthermore, the sulfur atom also forms a hydrogen bond with the side chain of E63, contributing further to the binding of p2C within the B-pocket of H-2Db (Fig. 1D).
The C-pocket in H-2Db, formed by residues Q9, Q97, and S99, is usually occupied by an asparagine that acts as main anchor residue, forming forklike hydrogen bonds with Q97 (18). In contrast, the methionine p5M in Trh4 takes a different conformation compared with canonical peptides with an asparagine at position 5. The main chain nitrogen atom of p5M forms a hydrogen bond with Q70 (Figs. 1E, 2). Importantly, it also forms hydrophobic and sulfur–π interactions with the side chains of H-2Db residues W73, F116, and Y156. As a consequence, residue Y156 bends up toward the solvent by 1.8 Å, resulting in a significant shift of the side chain of residue H155, previously shown as important for TCR recognition (33) (Fig. 2). In fact, besides the modifications at positions 155 and 156, no differences were found between the H-chains of H-2Db/Trh4 and H-2Db/gp33, nor with any of the hitherto structurally determined H-2Db/peptide complexes, which suggests that LnB5 specificity focuses on peptide residues and H155. Thus, structural analysis of H-2Db/Trh4 suggests that presence of a methionine residue at p5 affects the position of key H-2Db TCR-interacting residues, forming a unique MHC/peptide conformer that could be essential for recognition by TEIPP-specific T cells (Fig. 2). Finally, besides the network of hydrogen bonds formed between the carboxyl moiety of p9M and residues N80, Y84, and K146, the side chain of this peptide residue also acts as a main anchor residue forming hydrophobic and sulfur–π interactions with H-2Db residues W73 and Y123 (Fig. 1F).
In conclusion, the unconventional motif revealed by the crystal structure of the H-2Db/Thr4 complex suggests that several sulfur–π interactions formed between p1M, p2C, p5M, and p9M and H-2Db H-chain residues may extensively contribute to the binding of this TEIPP-associated epitope. Furthermore, the presence of a methionine at position 5 alters significantly the conformation of residues Y156 and H155, the latter known to be important for T cell recognition (33), forming altogether a novel MHC/peptide conformer that could be essential for recognition by TEIPP-specific T cells.
Sulfur–π interactions formed with Trh4 residues p2C and p5M are important for complex stability
In order to assess the importance of sulfur–π interactions for Thr4 binding to H-2Db, several altered peptide ligands (APLs) were created. Because stabilization of MHC-I by modified peptides correlates with immunogenicity (34–36), the capacity of all peptides to stabilize cell-surface expression of H-2Db was assessed (Fig. 3). Our results demonstrate that although lacking the fundamental asparagine residue at position 5, Trh4 stabilizes H-2Db better than the LCMV-derived immunodominant epitope gp33 (Fig. 3A). Interestingly, substitution of p5M to an asparagine reduced the stabilization capacity of the altered version Trh4-p5N compared with Trh4, to similar levels as gp33. Residues p1M and p5M were substituted to the methionine analog NLE in Trh4-p1NLE and Trh4-p5NLE, respectively, and residue p2C was exchanged for the cysteine analog ABU in Thr4-p2ABU. Although these substitutions conserve size and overall form, they abolish any possible formation of sulfur–π interactions. Although the cell-surface stability of H-2Db/p1NLE was not altered compared with the wild-type complex, both the p2ABU and p5NLE variants significantly reduced H-2Db cell surface stability (Fig. 3B), demonstrating the importance of sulfur–π interactions at peptide positions 2 and 5. Thermostability measurements using circular dichroism of soluble H-2Db in complex with Trh4 or the APLs Thr4-p2ABU and Trh4-p5NLE demonstrated that both modifications decreased complex stability, reducing Tm by 3.1 and 1.3°C, respectively (Fig. 3C). This shift in stabilization capacity is well in line with the measured differences in complex stability in the cellular assays.
The conformations of Thr4-p2ABU and Trh4-p5NLE are similar to Trh4, conserving equivalent TCR recognition
In order to exclude the possibility that the introduced substitutions at positions 2 and 5 altered the conformation of the modified peptides, the crystal structures of H-2Db in complex with Thr4-p2ABU and Trh4-p5NLE were both determined to 2.0 Å resolution (Supplemental Table I). The electron density of both structures was of good quality, allowing for unambiguous positioning of all residues (Supplemental Fig. 2). Comparison of both structures with H-2Db/Trh4 revealed striking overall similarity, with root mean square deviation values for the Cα atoms corresponding to H-chain residues 1–175 of 0.33 and 0.26 Å for H-2Db/Thr4-p2ABU and H-2Db/Trh4-p5NLE, respectively. The main chain conformations of Trh4-p5NLE and Thr4-p2ABU are similar to Trh4. The p5NLE and p2ABU residues adopt exactly the same conformation as p2C and p5M, respectively, in Trh4 (Fig. 4). In particular, the conformations of H-2Db residues Y156 and H155 were conserved in both H-2Db/Thr4-p2ABU and H-2Db/Trh4-p5NLE compared with H-2Db/Trh4, resulting in similar MHC/peptide conformers. Thus, structural comparison demonstrates that the measured reduction in stabilization capacity for both altered peptides (Fig. 3) is exclusively due to the removal of sulfur–π interactions.
The APLs were tested for recognition by the H-2Db/Trh4-specific T cell clone LnB5 (15) (Fig. 5). Titrated amounts of Trh4, p1NLE, p5NLE, and p9NLE were loaded on splenocytes and tested for their capacity to activate LnB5 T cells, by quantifying IFN-γ release in culture supernatants (Fig. 5A). All of the NLE peptide variants induced strong activation of LnB5 T cells with efficiencies comparable to Trh4. Furthermore, an APL in which all methionine residues (p1, p5, and p9) were simultaneously mutated to NLE-stimulated LnB5 T cells to comparable extents compared with Trh4. Similarly, substitution of the cysteine at p2 with ABU did not influence T cell recognition (Fig. 5B), indicating that the NLE and ABU modifications did not affect recognition by the LnB5 TCR. Thus, although the multiple sulfur–π interactions formed between Trh4 and H-2Db are important for complex stability, T cell recognition is not affected, most probably due to similar MHC/peptide conformers.
Substitution of Trh4 residue p5M to asparagine abolishes recognition by the H-2Db/Trh4-specific T cell clone LnB5
To assess the relative importance of each Trh4 residue for recognition by the H-2Db/Trh4-specific T cell clone LnB5, each amino acid of the peptide was individually substituted to an alanine (Fig. 5C). Target cells were supplemented with 10-fold dilutions of each APL, and the capacity of LnB5 T cells to produce IFN-γ was analyzed. Although substitution of p1M, p2C, and p7G did not alter release of IFN-γ, mutations of p4R, p5M, p6T, and p8V abolished recognition by LnB5 T cells, even at concentrations as high as 1 μg/ml. The importance of residues p4R and p5M for recognition by LnB5 was further substantiated by substitution to leucine and asparagine, respectively, also abolishing TCR recognition (Fig. 5D).
The crystal structure of H-2Db/Trh4 provides a molecular platform to understand these functional results. Residues p4R, p6T, and p8V all protrude toward the solvent-enabling interaction with the LnB5 TCR. Possibly the main chain of residue p7A may also interact with the TCR through interactions with its main chain atoms (Fig. 1A). Interestingly, mutation of p1M to alanine did not affect recognition by LnB5 (Fig. 5C), supporting the notion that the CDR loops of LnB5 focus on the central and C-terminal residues of the peptide. Although modification of p2C to alanine did not impair at all TCR recognition, modification of p5M to alanine abolished recognition. Substitution of the cysteine residue to alanine does most probably not alter the conformation of this altered peptide version and, although it possibly reduces the stability of the complex, the modified peptide is still recognized by LnB5. Indeed, according to the immune epitope database (www.iedb.org), alanine residues at peptide position 2 are overrepresented in H-2Db–restricted epitopes and can fit snuggly within the B-pocket of H-2Db.
Most peptides that bind H-2Db contain an asparagine as main anchor residue at position 5. We demonstrate in this study that the methionine at peptide position 5 in Trh4 alters significantly the conformation of H chain residues Y156 and H155, resulting in a unique MHC/peptide conformer that could be essential for recognition by TEIPP-specific T cells. Substitution of p5M to an asparagine, which reduces complex stability (Fig. 4), most probably results in a conventional positioning of the side chain of p5N, as described in all previously determined crystal structures of H-2Db/peptide complexes. This modification most probably re-establishes a conventional MHC/peptide conformer, altering back the positioning of the H-chain residues Y156 and H155 to prototypic conformations. Indeed, substitution of p5M to asparagine abolished recognition by the H-2Db/Trh4-specific T cell clone LnB5 (Fig. 5D). In conclusion, a methionine residue at peptide position 5 in Trh4 results in the formation of a unique H-2Db/Trh4 conformer that is specifically recognized by the TEIPP-specific T cell clone LnB5.
Inhibition of conventional TAP-mediated peptide transport is one of the bottlenecks of T cell–based therapeutic approaches against cancer and viral infections. The majority of conventional peptides are processed by the proteasome-TAP–dependent pathway and represent a preselected repertoire prior to the loading of high affinity peptides to MHC molecules in the ER. However, absence of TAP blocks their influx into the ER and peptides derived from alternative sources take over the peptide repertoire. As these peptides are generally derived from proteasome-independent mechanisms, it is possible to identify epitopes that can have escaped central and peripheral tolerance (12, 13). The identification of a novel category of TEIPPs that are selectively presented on TAP-deficient cells represents a significant advance in our search for novel cancer-specific targets. The Trh4 epitope, presented only on TAP-deficient tumor cells, may clearly belong to this interesting TEIPP target repertoire and might serve as ideal CD8 T cell Ags to exploit for immunotherapy of tumor immune escape variants.
Most of the hitherto identified immunogenic H-2Db and HLA-A2–restricted TEIPP-epitopes are derived from either protein signal sequences that are processed by signal peptidases in the ER or from transmembrane proteins residing in the ER (10, 11, 13, 37, 38). The Trh4 protein is such an ER membrane spanning protein of which the C terminus is loaded unto H-2Db in a peptide-transporter–independent way. Recently, the processing mechanism of this peptide was unraveled, establishing that liberation of the C-terminal segment of the Trh4 protein that protrudes into the lumen of the ER is mediated by the signal peptide peptidase SPP (15). Interestingly, N-terminally encoded signal sequences of secreted proteins are also released by signal peptide peptidase, implying a dominant role for this enzyme in TAP-independent presentation of TEIPP peptides (39). This does not exclude the possibility for other TEIPP peptides to be processed via the proteasome and metalloproteinases, although these exact pathways have not been elucidated at the molecular level yet (40). Importantly, TEIPP peptides are immunogenic because they fail to be presented by cells with a proficient processing machinery. Surface display by MHC-I molecules is only observed after defects in this pathway, especially after blockade of the peptide transporter TAP.
In this study, we provide a structural description for the first identified TEIPP epitope derived from the Trh4 protein, restricted to H-2Db. The Trh4 peptide has an unusual amino acid composition in which 44% of the epitope is composed of sulfur-rich residues (MCLRMTAVM). The surface stability of the H-2Db/Trh4 complex is higher compared with H-2Db/gp33, well established as a very stable MHC complex. Besides a similar amount of hydrogen bond and hydrophobic interactions formed in the H-2Db/Trh4 complex compared with all known three-dimensional structures of H-2Db in complex with canonical epitopes, a large amount of sulfur–π interactions are also formed between the sulfur-containing peptide residues p1M, p2C, p5M, and p9M and H-chain residues. Sulfur aromatic interactions are fairly common in protein structures and have been identified as weakly polar interactions stronger than van der Waals interactions (19, 20). In this study, substitution of p2C and p5M to ABU and NLE, respectively, reduced complex stability, demonstrating their importance for the stabilization capacity of Trh4. Importantly, structural comparison of Trh4 with the two APLs Trh4-p2ABU and Trh4-p5NLE demonstrated that the introduced modifications did not affect their conformations.
After asparagine, methionine is the second most preferred residue as a main anchoring residue at position 5 for H-2Db epitopes. To our knowledge, our study provides the first structural description of such a p5M-containing peptide. Although the shallow and at first sight relatively polar environment within the C-pocket of H-2Db provides a preference for p5N, it is still fully possible, as demonstrated within the current study, to stabilize efficiently H-2Db complexes using a methionine at position 5 through the formation of sulfur–π and hydrophobic interactions. Instead of the canonical forklike hydrogen bond interactions formed between p5N and H-chain residue Q97, the side chain of p5M interacts with the phenol rings of residues W73 and Y156, resulting in significant conformational modifications of Y156 and of H155, the latter a well-established major contact for interactions with TCRs (33). Importantly, the observed conformational modifications, although isolated only to two H-chain residues, form a novel conformer that is different from all the previously described H-2Db/peptide complexes. We hypothesize in this study that this conformer, unique for H-2Db/Trh4, is essential for the high specificity of the TCR LnB5.
Future determination of the ternary LnB5/H-2Db/Trh4 complex will provide detailed molecular information underlying the structural and functional results presented within the current study. Furthermore, we have previously demonstrated that substitution of peptide position 3 in the melanoma-associated H-2Db-restricted gp100-derived epitope EGSRNQDWL to a proline increased significantly the stability of the complex and provoked efficient in vivo responses toward B16 tumors (36). The molecular basis underlying these effects were clearly due to the formation of CH–π interactions between the side chain of the proline p3P and the side chain of the H-2Db tyrosine residue Y159 (41). We will apply this strategy on the Trh4 peptide in hope of inducing stronger TCR responses toward TAP-deficient tumor cells.
We thank the ESRF (Grenoble, France) for access to the synchrotron beam lines ID14-2 and ID30A-1 and Bessy (Helmholtz-Zentrum Berlin, Berlin, Germany) for the MX 14.1 as well as the crystallization facility in the Protein Science Facility, Karolinska Institutet. We also thank Dr. Jan Wouter Drijfhout for all peptide production.
This work was supported by the Swedish Research Council (to A.A.), the Swedish Cancer Society (to A.A.), the Dutch Cancer Society (to T.v.H., UL 2010-4785), and the Portuguese Foundation for Science and Technology (to C.C.O., SFRH/BD/33539/2008). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The crystal structures presented in this article have been submitted to the Protein Data Bank (http://www.rcsb.org/pdb/home/home.do) under accession numbers 5E8N, 5E8O, and 5E8P.
The online version of this article contains supplemental material.
Abbreviations used in this article:
altered peptide ligand
European Synchrotron Radiation Facility
lymphocytic choriomeningitis virus
mean fluorescence intensity
MHC class I
MHC class I/peptide complex
transporter associated with Ag processing
T cell epitope associated with impaired peptide processing
The authors have no financial conflicts of interest.