SEARCH

SEARCH BY CITATION

Keywords:

  • Antigen presentation/processing;
  • CD8 T cells;
  • Immunotherapy;
  • Mass spectrometry;
  • MHC

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Human leukocyte antigens (HLA) have long been grouped into supertypes to facilitate peptide-based immunotherapy. Analysis of several hundreds of peptides presented by all nine antigens of the HLA-B44 supertype (HLA-B*18, B*37, B*40, B*41, B*44, B*45, B*47, B*49 and B*50) revealed unique peptide motifs for each of them. Taking all supertype members into consideration only 25 out of 670 natural ligands were found on more than one HLA molecule. Further direct comparisons by two mass spectrometric methods – isotope labeling as well as a label-free approach – consistently demonstrated only minute overlaps of below 3% between the ligandomes of different HLA antigens. In addition, T cell reactions of healthy donors against immunodominant HLA-B*44 and HLA-B*40 epitopes from EBV lacked promiscuous T-cell recognition within the HLA-B44 supertype. Taken together, these results challenge the common paradigm of broadly presented epitopes within this supertype.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Peptides presented on human leukocyte antigen (HLA) class I molecules are the final result of antigen processing and are usually generated under the participation of the cytosolic proteasomes and aminopeptidases. About 1% of peptides produced in the cytosol are translocated to the ER via TAP 1. Upon undergoing further N-terminal trimming, the peptides are finally loaded onto HLA class I molecules. Thus, peptides presented on the cell surface have succeeded in following various rules along the pathway of antigen processing. They are determined by the cleavage specificities of the proteasome and by cytosolic aminopeptidase activity. The transport via TAP shows strong preferences for hydrophobic or charged residues in both the C-terminal and the second position 2. Selectivity also applies to trimming, as peptide bonds between any amino acid and Pro cannot be cleaved by peptidases of the ER, and HLA class I molecules themselves serve as templates for their to-be-ligands 3. Finally, peptides build stable complexes with HLA molecules only if they fit into the binding groove, which depends on the nature of the so-called pockets.

The allele-specific amino acid composition of these pockets determines both their polarity and stereochemistry and, consequently, also the residues of the peptide that are allowed to protrude into these pockets. The peptide motif describes such primary anchors of the peptide, which have the strongest effect on ligand binding as well as less constricted but still nonetheless important auxiliary anchors. The extensive polymorphism of HLA genes provides the basis for similar variability among peptide motifs. With an increasing number of HLA alleles characterized in more detail these were classified into several groups. So-called supertypes can be defined according to either structure or function 4. The latter approach groups HLA alleles with regard to their peptide motifs and is hence more closely related to their function in antigen presentation to T cells. It has been proposed that a peptide which binds to one member of the supertype should, with a good probability, also be capable of binding to the others 5.

Immunotherapeutic approaches to cancer or viral infections are largely based on immunogenic peptides presented on HLA molecules and eliciting T-cell responses that destroy transformed or virally infected cells. A major drawback of this safe and technically simple approach is the HLA restriction of the respective epitope, which has to be considered. Consequently, individualized treatment of cancer patients has been proposed 6.

The concept of HLA supertypes with promiscuous ligands promises to overcome such difficulties. One peptide could be used for the vaccination of a large number of patients all expressing HLA antigens that are different but belong to the same supertype. Hence, many studies have investigated such cross-binding peptides by using binding assays, by T-cell stimulations, or by the prediction of promiscuous peptides 5, 7, 8.

The first supertypes, namely A2, B7 and A3, were defined in 1995 and 1996. Based on peptide binding motifs, the HLA-B44 supertype comprising HLA-B37, B41, B44, B45, B47, B49, B50, B60 and B61 was first proposed in 1996 9. This definition took into account pool-sequencing data in the case of B37, B40/B60, and subtypes of B*44. In the meantime, HLA-peptide-binding assays were applied to characterize binding motifs of six members of the supertype in detail and identify cross-reacting peptides 5. Natural ligands have been sequenced by mass spectrometry 10, 11. HLA-B*4101 was newly characterized by pool sequencing 12.

Present knowledge of peptide motifs of HLA-B44 supertype members covers only certain members of the HLA-B44 supertype and is based more on binding studies than natural ligands. Despite these deficiencies, peptides with degenerate B44 supertype binding capacities have long been thought to be capable of facilitating peptide-based immunotherapeutic and diagnostic strategies. The HLA-B44 supertype contains some of the most frequent HLA alleles 13. Calculations using the algorithm presented by Schipper et al. 14 result in a high population coverage throughout all ethnic populations and thus demonstrate the relevance for clinical applications (Supporting Information Fig. 1). To scrutinize the HLA-B44 supertype definition and its putative implications we analyzed 670 peptides presented on different HLA allotypes from all nine supertype antigens to enable the definition of peptide motifs based on natural ligands. In addition, a direct molecular comparison of peptide repertoires of different HLA allotypes was achieved using differential isotope labeling as well as label-free mass spectrometric approaches that revealed only marginal overlaps. Finally, we tested the recognition of immunodominant, HLA-B*4402- or HLA-B*4001-restricted EBV epitopes demonstrating a lack of promiscuous T-cell responses within the HLA-B44 supertype.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Uniqueness of peptide motifs of HLA-B44 supertype antigens

To determine the peptide motifs of all nine members of the supertype, large numbers of ligands were analyzed. HLA-presented peptides were isolated from different cell lines and primary tissues – mainly B-lymphoblastoid cell lines (B-LCL) and renal cell carcinoma (RCC) specimen – by detergent lysis and subsequent immunoaffinity purification using HLA-specific antibodies. Peptides were then sequenced by liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS). Finally, peptide motifs of individual allotypes were determined or refined from a total of 670 ligands (Fig. 1). On average, 74 ligands were sequenced per HLA allotype with peptide numbers varying between 18 and 184 (median number of peptides per allotype: 60). All peptide sequences as well as the HLA allotypes and sources from which they were isolated are given in Supporting Information Table 1.

thumbnail image

Figure 1. Peptide motifs and length profiles of HLA-B44 supertype allotypes (A–J). Allele-specific peptide motifs show amino acid frequencies over nine positions of nonamers. Character sizes represent amino acid frequencies in anchor positions. In non-anchor positions amino acids are only indicated if frequencies exceed 30%. The bar diagrams show peptide length distributions with the number of peptides on the y-axis and the bars representing octamers, nonamers, decamers and combined undecamers and dodecamers.

Download figure to PowerPoint

In general, members of the HLA-B44 supertype favor acidic amino acids, mostly Glu, in the first anchor position, P2. The second anchor position at the C-terminus, PΩ, is occupied by a variety of hydrophobic residues. However, a closer look at individual allotypes reveals a multiplicity of unique features of their peptide motifs. Although most members of the supertype almost exclusively present ligands with Glu in P2, HLA-B*37 and B*4701 favor Asp with 60 and 95%, respectively, over Glu. The uniqueness is even more pronounced regarding the C-terminal anchor. HLA-B*4101, B*4501 and B*5001 prefer small amino acids such as Ala and Pro, while HLA-B*1801 and B*4402 accept aromatic residues predominantly in this position. The strong preferences for Ile, Leu or Phe in PΩ contribute to the individuality of HLA-B*4901, B*4001, B*4701 and B*37, respectively. For some allotypes, P1 represents an auxiliary anchor, which is acidic in the case of HLA-B*1801 and basic for HLA-B*4101 and B*4701. HLA-B*4701 is the only molecule with a basic secondary anchor in PΩ−1, the peptide position directly preceding the C-terminus. Furthermore, Leu is preferred in P3 of HLA-B*37 and B*4701 as well as Arg and Leu in P5 among ligands of HLA-B*37 and B*4701, respectively. For HLA-B*4001, the preference for Val, Leu and Ile suggests a hydrophobic auxiliary anchor in P7. In addition to the individuality based on amino acid composition, peptide length profiles also differ remarkably among the allotypes of the HLA-B44 supertype. Octamers are preferred at the expense of nonamers by HLA-B*1801 and B*37, while there is a shift toward longer peptides in the case of HLA-B*4001 and B*4402.

Molecular basis of ligand–HLA interactions

The different binding specificities of the HLA-B44 supertype members can to some extent be explained by evaluation of the pocket structure of the different alleles (Supporting Information Table 2). Considering P1, the high incidence of Glu and Asp among HLA-B*1801 ligands can be explained by the positive net charge of pocket A caused by the basic His171. All other members of the B44 supertype display negatively charged P1 pockets, but differ in their preference for acidic side chains.

The exclusive occupation of P2 by Glu and occasionally Asp is explained by a salt bridge between P2 and Lys45 of the HLA molecule, which determines the negative polarity of P2 15. Only in HLA-B*1801 and B*37 is Lys replaced by Thr. It has been proposed that pocket geometry and composition may render His9 exceptionally important as it can interact with the carboxyl group of a P2 acidic side chain 16, 17.

The size of hydrophobic side chains in PΩ of ligands is largely influenced by the size of amino acids 81, 95 and 116 of the HLA-α chain 18. Trp95 or Tyr116 appear to exclude aromatic amino acids from this pocket due to their size, which applies to HLA-B*4001, HLA-B*4101, HLA-B*4501, B*4901 and B*5001. Phe116 of HLA-B*37 tolerates Phe in PΩ as it is not as spacious as Tyr. For HLA-B*1801, the small Ser116 allows aromatic anchors and, due to the interaction of hydroxyl groups, favors Tyr more than any other member of the supertype. For Asp116, all aromatic anchor residues are allowed and sometimes strongly preferred, as in HLA-B*4402; but as indicated by the preference of HLA-B*4701 for Leu, they are not compulsory. The combination Leu81-Trp95-Tyr/Leu116 narrows pocket F by their bulky hydrophobic side chains resulting in a preference for small side chains in ligands of HLA-B*4101, B*4501 and B*5001. Furthermore, polymorphisms within pocket F might not only influence the peptide repertoire directly by physico-chemical interactions but also by changing the dependence of peptide loading on tapasin 18, 19.

No other HLA-A, -B or -C antigens possess Lys45 or His9 (http://www.anthonynolan.com/HIG/data.html) 20. Hence, it can be expected from their pocket structure that they do not present peptides with acidic residues in P2. This is further supported by peptide motifs of 12 HLA-A, 21 non-B44 HLA-B and 5 HLA-C antigens as well as by hundreds of natural ligands of even more HLA antigens which do not contain glutamic or aspartic acid in the first anchor position (http://www.syfpeithi.de). The composition of the HLA-B44 supertype is thus considered exhaustive.

Marginal overlap of peptide repertoires among different allotypes

Peptide motifs already suggested a high degree of individuality of different allotypes. This is also supported by HLA ligand repertoire analyses. The collection of sequences already used for peptide motif definitions was further analyzed with respect to peptides found on more than one HLA allotype. Surprisingly, of 670 ligands presented on different HLA molecules of the B44 supertype, only 25 were found on more than one allotype (Fig. 2). Merely 2 of these 25 were detected on more than two different HLA molecules. By contrast, when comparing RCC tissues from different patients expressing HLA-B*1801, an overlap of more than 50% of peptide sequences was determined: RCC098 shared 24/47 peptides (51%) with RCC189 while RCC115 shared 12/15 HLA-B*1801 ligands (80%) with RCC189.

thumbnail image

Figure 2. Only 25 promiscuous peptides identified among 670 natural ligands. Schematic representation of the overlap of HLA ligandomes of different allotypes. Ligands detected on two different HLA alleles are depicted within the intersecting square of the corresponding rows and columns. The total number of ligands characterized for each allotype is indicated in parentheses.

Download figure to PowerPoint

Limitations of this approach are the numerous sources used for peptide isolation, which might be responsible for the missing overlap. By contrast, our experience tells us that peptide motifs are independent of the tissue type or cell line used, which is confirmed by comparison of HLA-B*4001 ligands from either RCC or the B-LCL EMJ. Further support comes from the comparison of HLA-B*1801 peptides presented on solid tissues analyzed in this study and ligands derived from soluble HLA-B*1801 from a transfected B-LCL characterized by Hickman et al. 10. Nevertheless, to exclude considerable influences of the type of sources used, two other mass spectrometric approaches were employed to investigate more directly the overlap of peptide repertoires among different allotypes of the HLA-B44 supertype expressed by comparable sources, namely B-LCL.

The label-free data-dependent ion selection (DDIS) approach was applied to determine the overlap between three HLA-B44 supertype allotypes expressed on three B-LCL: EMJ (HLA-B*4001), Awells (HLA-B*4402) and BM15 (HLA-B*4901) (Fig. 3). These cell lines are homozygous with regard to the HLA-B locus. Peptides were isolated using the HLA-B,-C-specific antibody B1.23.2. From our experience B1.23.2 is in fact HLA-B,-C-specific as we never observed peptides in these extractions that could be assigned to HLA-A molecules. Furthermore, because of the very low HLA-C surface expression 21, 22, the bulk of isolated peptides derives from HLA-B molecules. For each sample, several LC-MS/MS experiments with DDIS were carried out consecutively and in random order. Peptide precursors selected for fragmentation were compared between two and three experiments of different samples, respectively (Supporting Information Fig. 2). In total, 6159 fragmentation spectra resulting from seven different LC-MS/MS experiments (with 745–946 spectra each) were used for pairwise as well as threefold comparisons. The overlap of identical spectra from two different cell lines was always below 3%. In particular, the overlap among HLA-B*4901 and B*4402 was exiguous (0.6%). Furthermore, merely 0.6% of spectra were detected in all three samples. Two repetitive DDIS experiments usually yield 50–60% of identical spectra (technical reproducibility, Supporting Information Fig. 2).

thumbnail image

Figure 3. Label-free DDIS approach reveals only marginal overlap of the HLA ligandomes of different HLA-B44 supertype allotypes. Peptides from three different HLA-B44 supertype allotypes were analyzed by LC-MS/MS experiments with DDIS. Precursors selected for fragmentation were compared between two or three different experiments on the basis of retention time, m/z value and fragmentation pattern. Numbers of compared fragmentation spectra are given in parentheses. Percentages of overlap are given as mean±SEM (n=8 for pairwise comparisons, n=72 for threefold comparison).

Download figure to PowerPoint

The differential N-terminal isotope coding (dNIC) approach uses stable isotope labeling for pairwise comparison of peptides from two samples. Ligands extracted from HLA-B*4901 and B*4101 were compared (Fig. 4). HLA-presented peptides were isolated from the B-LCL BM15 and LCL026 using the antibody B1.23.2. Peptides derived from BM15 were modified with the light form of the nicotinylation reagent, those from LCL026 with the deuterated, heavy form. Equal amounts of peptides were mixed for an liquid chromatography-coupled mass spectrometry (LC-MS) experiment, which was then analyzed with respect to peptide pairs, which indicate the presence of identical peptides in both samples. All in all 513 single signals (peptides found in only one sample) but only two signal pairs (peptides found in both samples) were detected (0.3%).

thumbnail image

Figure 4. The dNIC approach demonstrates the tenuous overlap of HLA ligands extracted from HLA-B*4901 and B*4101. HLA-presented peptides were analyzed in an LC-MS experiment after differential modification with light and heavy isotope label (H4-/D4-NIC). Five hundred and thirteen peptide pairs as well as two unique peptides were identified. Seventy-eight peptides were sequenced by separate LC-MS/MS experiments, only one sequence was identified in both samples.

Download figure to PowerPoint

In order to improve the assessment of the extent of overlap detected in the dNIC approach and to integrate it into a comprehensive view, several control experiments were carried out. The LCL721.45 cell line was used for two independent extractions at different time points. Differential modification of these two samples yielded more than 95% of signal pairs (biological reproducibility, data not shown). Ratios of these signal pairs showed that 95% of all peptides varied no more than 1.4-fold between two samples (Supporting Information Fig. 3). Differential modification of peptides derived from the same allotype but from two different B-LCL results in 20–30% of signal pairs (data not shown).

Within the dNIC approach, sequence analysis of individual peptides was also performed. As LCL026 is not homozygous with regard to the HLA-B locus, some single signals might derive from HLA-B*5101-bound peptides. Therefore, in separate LC-MS/MS experiments, sequences of 56 HLA-B*4901-presented and 22 HLA-B*4101-presented ligands were determined. Only one of these peptides showed a signal pair in the LC-MS experiment and was thus present in both samples.

Evaluation of source proteins

BLAST (Basic Local Alignment Search Tool) searches revealed that the sequenced HLA ligands were derived from more than 500 different proteins. Seventy-five proteins give rise to more than one peptide. Table 1 presents several examples.

Table 1. Multiple HLA ligands from the same protein are presented on different HLA alleles of the HLA-B44 supertype
 Ligands presented on HLA-
Gene symbola)B*1801B*37B*4001B*4101B*4402B*4501B*4701B*4901B*5001
  • a)

    a) MYL6, myosin, light polypeptide 6, alkali, smooth muscle and non-muscle; VIM, vimentin; SPTBN1, spectrin, beta, non-erythrocytic 1; TLN1, talin1.

MYL6DEMNVKVL AEFKEAFQL AEFKEAFQLF  AEIRHVLVTL 
   AEIRHVLVTL    YEELVRMV 
   YEELVRMVL      
VIMLERKVESLRETNLDSLPREKLQEEML EEIAFLKKL(M)EENFAVEA REYQDLLNV 
SPTBN1DEKSIITY   EEASLLHQF KDLTSVMRLYEERVQAVV 
 DEMKVLVL        
 EEASLLHQF        
TLN1DEYSLVREL  SEAKPAAVAASEIEAKVRY   QEISHLIEP

For example, three different peptides derived from MYL6 (myosin, light chain 6, alkali, smooth muscle and non-muscle) are HLA-B*4001-restricted. One of them is also found on HLA-B*4901. HLA-B*4901 and B*4402 display further length variants while a very different peptide is found on HLA-B*1801. In the case of VIM (vimentin), six very different peptides and an elongated variant are presented by six different HLA allotypes. Five ligands derived from SPTBN1 (spectrin, beta, non-erythrocytic 1) and four TLN1 (talin 1)-derived peptides are each displayed by four supertype members. Thus, different allotypes present multiple peptides from the same protein.

CD8 T-cell responses against EBV-derived peptides restricted by HLA-B*44 or B*40

EBV-specific CD8 T-cell responses can frequently be detected in IFN-γ ELISPOT assays without the need to determine the infection status of the donors since a high proportion of individuals have a history of infection. We have tested reactions against three HLA-B*44-restricted peptides (AEGGVGWRHW, EBNA6162–171; EENLLDFVRF, EBNA6281–290; VEITPYKPTW, EBNA4657–666) and one HLA-B*40-restricted peptide (IEDPPFNSL, LMP2200–208) from EBV antigens in 130 blood donors 23. The donors were required to express exactly one allele of the HLA-B44 supertype but were otherwise selected randomly. Since ex vivo ELISPOT assays with the peptide VEITPYKPTW yielded positive responses of HLA-B*44 donors only, we included a presensitization step for the three other peptides to expand specific memory T cells and enhance weak responses. Nevertheless, we detected CD8 T-cell responses against two of these epitopes only in donors positive for the respective restricting allele (Fig. 5, Table 2), and in none of the donors carrying other alleles of the supertype. For EENLLDFVRF, two responses on other alleles of the supertype were observed, but the differences in the frequency of responses between donors positive and negative for HLA-B*44 are significant (Fisher's exact test, p=7.3×10−3, n=64).

thumbnail image

Figure 5. T-cell recognition of EBV epitopes is tightly restricted to the appropriate HLA allotype. CD8 T-cell responses to three HLA-B*44-restricted (A, C and D) and one HLA-B*40-restricted epitope (B) from EBV were assayed by IFN-γ ELISPOT after recall stimulation (A–C) or ex vivo (D) from donors carrying different antigens of the HLA-B44 supertype. Carriers of non-restricting alleles have been grouped for visualization purposes as well as for statistical analysis. The logarithmic y-axis shows the spot count index that is the ratio of the mean spot number per well to the mean spot number of the HIV control representing the background. Furthermore, mean spot numbers above 10 (•) and below 10 (○) were distinguished. Reactions were regarded as positive if the mean spot number per well was at least 10 and the spot count index was at least more than 3 (for ex vivo analysis) or 5 (for recall stimulations). In all four cases, the frequency of positive responses is significantly higher in donors carrying the restricting allele than in all others (Fisher's exact test, see Table 2 for further details).

Download figure to PowerPoint

Table 2. T-cell recognition of EBV epitopes is tightly restricted to the appropriate HLA allotype
 Responding/tested donors positive for HLA- 
Peptide sequenceB*18B*37B*40B*41B*44B*45B*47B*49B*50p-Value (n) (Fisher's exact test)
AEGGVGWRHW0/120/70/120/317/280/40/30/60/33.4×10−10 (78)
EENLLDFVRF0/121/41/129/280/40/47.3×10−3 (64)
IEDPPFNSL0/60/1013/150/210/20/43.3×10−11 (58)
VEITPYKPTW0/40/60/20/29/160/10/50/30/14.2×10−5 (40)

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Along with the increasing number of HLA alleles characterized in more detail, it became clear that different alleles can be classified into several supertypes. In this study we verified the assignment of nine HLA antigens to the B44 supertype based on their peptide motif deduced from 670 natural ligands. The refined motifs of HLA-B*37, B*4001, B*4101, B*4402 and B*4501 are in good agreement with earlier results from Edman pool sequencing and known epitopes. Ligand specificities of HLA-B*4701, B*4901 and B*5001 have not been described before and thus fill a major gap in our knowledge of the HLA-B44 supertype. HLA-B*1801 has already been characterized from a cell line by Hickman et al. 10. Interestingly, despite different methodologies and usage of either a cell line or solid tissues, both approaches yielded similar peptide motifs from natural ligands.

In contrast, we observed several differences to HLA-peptide-binding studies 5 where Gly was found among the preferred amino acids at the C-terminus of the peptides: none of our 121 HLA-B*1801-presented ligands showed a Gly in PΩ. Such discrepancies between analysis of natural ligands and binding studies have already been reported 24. Mere binding to a certain HLA molecule is not sufficient for natural presentation since critical steps of antigen processing other than HLA binding (such as proteolytic activities and specificities as well as transport preferences) do play a major role.

HLA-B*18, B*37, B*40, B*41, B*44, B*45, B*47, B*49 and B*50 indeed can be grouped into one supertype sharing acidic anchors in P2 and hydrophobic anchor residues in PΩ. Nevertheless, a variety of allotype-specific features associated with primary and auxiliary anchor positions and with peptide length distributions contributes to the uniqueness of each individual supertype members. This individuality is also implied by the small proportion of sequenced ligands found on more than one HLA allotype (25 out of 670).

To rule out that the missing overlap of peptide repertoires might be due to low peptide numbers for some allotypes, the label-free DDIS approach was used to investigate the overlap of different B44 supertype members. Comparison of thousands of MS/MS spectra showed an overlap of below 3%, while repetitive LC-MS/MS experiments of the same sample usually yield 50–60% of identical spectra. Our DDIS experiment revealed that spectra common to HLA-B*4901 and B*4402 were rare (0.6%), while spectra of HLA-B*4001 ligands were found slightly more often in the other two samples (2.3–2.4%). A similar tendency might already be estimated from the promiscuous peptides (Fig. 2). This observation probably reflects the degree of uniqueness of peptide motifs of different HLA-B44 supertype members. While Ile (HLA-B*4901) and aromatic residues (HLA-B*4402) in PΩ are rather special to these molecules, Leu, which dominates pocket F in HLA-B*4001, is also found among ligands of the other two HLA allotypes.

Furthermore, using the dNIC approach HLA-B*4901 and B*4101 ligands were differentially modified by isotope labeling allowing a direct comparison of their peptide repertoires. Again, among 513 signals only 2 were found to overlap between the two B-LCLs (0.4%). For comparison, differential modification of two samples extracted from the same cell line yielded more than 95% of signal pairs. The same experiment with peptides derived from the same allotype but from two different B-LCL results in 20–30% of signal pairs. This shows that although the HLA peptide repertoire of an allotype contains cell line-specific ligands there is a pronounced cell line-independent proportion which is two orders of magnitude above the overlap among different allotypes. Thus, several different approaches consistently reveal only a marginal overlap of HLA ligandomes.

Ideally, at least one peptide of each protein or even isoform should be presented on the cell surface to make any intracellular changes detectable for T cells. Similar HLA subtypes of the same supertype do not necessarily use the same ligand from a protein because the overlap of peptide repertoires is so minute (Table 1). Instead, multiple peptides from the same protein are presented by different allotypes. In addition there are few examples of peptides presented in the form of different length variants like AEFKEAFQL(F), which is presented by HLA-B*4001 (with C-terminal Leu) and B*4402 (with C-terminal F). This might reflect an adaptation to differing preferences for the C-terminal anchor, which fits well into the general picture.

Nevertheless, the overlap might be more pronounced than is apparent at first sight due to peptides of low abundance going undetected in all approaches. We cannot rule out the possibility that the overlap could slightly increase if all peptides, including those currently undetectable, are characterized. However, the overlap detected within our approaches is so small that it is quite obvious that most peptides are not presented on various allotypes, and that ligand profiles of different HLA allotypes are fairly unique.

The consequence of only tenuous overlaps of peptides presented on different alleles of the supertype is also illustrated by analysis of CD8 T-cell responses. Our results demonstrate that one HLA-B*40-restricted and three HLA-B*44-restricted peptides derived from EBV-encoded proteins are dominant epitopes only in donors with the cognate allele of the HLA-B44 supertype. No specific T-cell responses against three of these peptides could be detected in carriers of other alleles of this supertype. Only minor promiscuous recognition of the fourth peptide was observed with a significant difference between donors positive and negative for HLA-B*44. For HLA-B*4402 and B*4403 it has shown that even a single polymorphism modifies not only the peptide repertoire but also T-cell recognition 15.

A supermotif average relative binding (SARB) value has been proposed as a measure for the probability of degenerate binding of peptides to six alleles of the HLA-B44 supertype 5. Thirteen of 14 peptides with SARB values higher than 10 have been found to be degenerate binders in that study. The peptides used in our study possess SARB values from a wide range (AEGGVGWRHW: 39, EENLLDFVRF: 12, IEDPPFNSL: 0.13, VEITPYKPTW: 0.16). Even a peptide with an SARB value far higher than 10, AEGGVGWRHW, shows no promiscuous recognition, again implying that binding studies are not necessarily consistent with observed natural ligands and T-cell reactions.

The marginal overlap of peptides between different allotypes of the HLA-B44 supertype, together with the unique features of their overall peptide motifs as well as the lack of promiscuous recognition of epitopes by CD8 T cells, argue against the occurrence of broadly promiscuous ligands in vivo. According to our results, the existence of a peptide that is presented by all members of the supertype and is specific for the potential therapeutic application as well as capable of eliciting T-cell responses with multiple HLA restrictions is highly unlikely.

Thus the common paradigm of HLA supertypes with promiscuous ligands 25 promising to overcome difficulties of peptide-based immunotherapy has to be questioned. Despite rough similarities of binding specificities among HLA supertype members, immunotherapeutic approaches do require tailoring to the HLA allotypes of individual patients.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Cell lines and antibodies

The HLA-defined B-LCL BM15 (International Histocompatibility Workshop, IHW-No. 9040, HLA-B*4901), Awells (IHW-No. 9090, HLA-B*4402), WIL-JON (HLA-B*4101), EMJ (IHW-No. 9097, HLA-B*4001), LCL026 (HLA-B*4101), LCL721.45 (HLA-A*02, -B*51) and the human melanoma line M18 (European Searchable Tumour Line Database, ESTDAB-133, HLA-B*4901) were maintained in RPMI 1640 medium containing 10% fetal bovine serum and supplemented with 2 mM L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin. Cell lines were established from patients RCC026 (HLA-B*4101) and RCC068 (HLA-B*4501).

The antibodies W6/32 (anti-HLA class I) 26 and B1.23.2 (anti-HLA-B, -C) 27 were purified from hybridoma culture supernatants using protein A sepharose beads (GE Healthcare, Uppsala, Sweden).

Patients and tumor specimens

Surgically removed RCC and colorectal carcinoma specimens were provided by the Department of Urology or Department of Surgery, University of Tübingen. The local ethical committee approved this study, and informed consent was obtained from the patients. Pathological staging and grading were performed by the Department of Pathology and low-resolution HLA typing was done by the Institute of Clinical and Experimental Transfusion Medicine, University of Tübingen.

Elution and analysis of HLA class I-bound peptides

Frozen cell pellets (2×1010−5×1010 cells) and shock-frozen tumor samples were essentially processed as described 28. For pairwise sample comparison peptides were differentially modified as described by light or heavy dNIC-NHS (1-([1H4/2D4] nicotinoyloxy) succinimide, D represents deuterium) 29. Peptides were analyzed online by a reversed phase Ultimate HPLC system (Dionex, Amsterdam, The Netherlands) or nanoLC 2D system (Eksigent Technologies, Dublin, CA, USA) coupled to a hybrid quadrupole orthogonal acceleration time-of-flight mass spectrometer (Q-TOF I; Micromass, Manchester, UK) equipped with a micro-electrospray ionization source, as described 29.

Modification and mixing of peptides

For experiments aiming for the comparison of HLA ligandomes of two different samples peptides were guanidinated and differentially modified by light or heavy dNIC-NHS (1-([1H4/2D4] nicotinoyloxy)succinimide, D represents deuterium) as described elsewhere 29, 30. To enable mixing of peptides from the different samples for LC-MS experiments in a total peptide ratio of 1:1, absorption of isotope-labeled peptide pools was determined at 260 nm.

MS

Peptides were analyzed online by a reversed phase Ultimate HPLC system (Dionex, Amsterdam, The Netherlands) or nanoLC 2D system (Eksigent Technologies, Dublin, CA, USA) coupled to a Q-TOF I (Micromass, Manchester, UK) equipped with a micro-electrospray ionization source, as described previously 29. Mixtures of modified peptides from two different samples were measured in an LC-MS experiment without fragmentation. For sequence analysis nicotinylated as well as unmodified peptides were analyzed in individual LC-MS/MS experiments with DDIS. For comparison of peptides from three samples without modification all LC-MS/MS experiments were carried out consecutively with random order of samples.

Analysis of MS data

Fragment spectra were analyzed manually as described 5 and database searches (National Center for Biotechnology Information, Expressed Sequence Tag) were carried out using Multiple Alignment System for Protein Sequences Based on Three-way Dynamic Programming (MASCOT, http://www.matrixscience.com).

Consecutive LC-MS/MS experiments with DDIS of unmodified samples were compared with regard to precursors selected for fragmentation. To calculate the overlap of fragmentation spectra for two LC-MS/MS experiments, A and B, the number of spectra from experiment A with a matching spectrum in experiment B was divided by the total number of spectra from experiment A and vice versa.

Motif determination and visualization

Peptides were assigned to HLA allotypes taking into consideration the HLA typing of the sample, the use of specific antibodies and the anchor residues known from pool sequencing and published in SYFPEITHI database 31. Furthermore binding studies 5 as well as epitope predictions were considered if available (http://www.syfpeithi.de, http://www-bimas.cit.nih.gov/molbio/hla_bind/, http://www.cbs.dtu.dk/services/NetMHC, http://www-bs.informatik.uni-tuebingen.de/SVMHC/, http://tools.immuneepitope.org/analyze/html/mhc_binding.html).

Absolute and relative amino acid frequencies were determined for all peptide positions. Figures of HLA motifs depict anchor positions and preferred residues with character sizes proportional to frequencies. Preferred amino acids are only indicated if frequencies exceed 30%.

Proteomic data analysis

Protein-protein and protein-translated database BLAST searches (http://www.ncbi.nlm.nih.gov/blast) were carried out to determine the source proteins for all peptide sequences. Each source protein was assigned the appropriate Entrez GeneID. If a peptide sequence was contained in more than one GeneID only the lowest number was considered to provide a systematic procedure and avoid any selection bias.

PBMC isolation and T-cell culture

Buffy coats were obtained from healthy blood bank donors of known HLA class I. The local ethical committee approved this study. PBMC were isolated from fresh buffy coats using standard gradient separation and cryopreserved in fetal bovine serum with 10% DMSO.

T-cell medium consisted of RPMI 1640 containing HEPES and L-glutamine with 10% heat-inactivated human serum, 50 U/mL penicillin, 50 μg/mL streptomycin and 20 μg/mL gentamicin.

Peptide synthesis

Peptides were synthesized by standard Fmoc chemistry using the 433A Peptide Synthesizer (Applied Biosystems, Weiterstadt, Germany) or the Economy Peptide Synthesizer EPS 221 (ABIMED, Langen, Germany) and analyzed by HPLC (Varian Star; Zinsser Analytics, Munich, Germany) and MALDI-TOF MS (future, GSG, Bruchsal, Germany).

Peptide presensitization of PBMC

Thawed PBMC were taken into culture on day 1 with 2.5 ng/mL IL-4 or 5 ng/mL IL-4 and 5 ng/mL IL-7. On day 2, the same amount of cytokines together with 1 μg/mL of peptide was added. Two nanograms per microliter IL-2 were added on days 4 and 6 or on days 4, 6 and 8. Cells were used for ELISPOT on day 13.

IFN-γ ELISPOT assay

The 96-well nitrocellulose plates (Millipore, Bedford, MA, USA) were coated with mouse anti-human IFN-γ. The 5×105 PBMC/well were stimulated with 1 μg/mL peptide for 26 h. Additional Ab incubations and development of the ELISPOT were performed according to the manufacturer's instructions (Mabtech, Nacka, Sweden). Plates were analyzed using the Immunospot Image Analyzer and ImmunoSpot Software Version 3.2e (both Cellular Technology, Cleveland, OH, USA). If spots were too numerous to count, the spot count index was defined as 2000. Reactions were regarded as positive if the mean spot number per well was at least 10 and more than three or five times the mean spot number of the HIV control. ELISPOT data were analyzed using Fisher's exact test.

Statistical analysis

ELISPOT data were analyzed using a two-sided Fisher's exact test with α=0.05. This test is used for the analysis of categorical data (positive versus negative reactions) with small sample sizes. p-Values were calculated by comparing donors carrying the restricting allele with donors carrying all other alleles.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank H. Planatscher and L. T. Gerstmayr for bioinformatics support; C. Gouttefangeas for ELISPOT presensitization protocols; E. Nößner for LCL026; C. A. Müller for high resolution typing; F. Altenberend, J. Makowski, V. Meyer and J. Stickel for additional experimental support; T. Stehle for critical reading of the manuscript. This work was supported by the European Union (LSHB-CT-2003-503321 to St. St.; LSHB-CT-2004-503319 to H.-G. R.) and the Deutsche Forschungsgemeinschaft (SFB 685 to St. St., A. S. and H.-G. R., GK 794 to H.-G. R.).

Conflict of interest: The authors declare no financial or commercial conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Reits, E., Griekspoor, A., Neijssen, J., Groothuis, T., Jalink, K., van Veelen, P., Janssen, H. et al., Peptide diffusion, protection, and degradation in nuclear and cytoplasmic compartments before antigen presentation by MHC class I. Immunity 2003. 18: 97108.
  • 2
    van Endert, P. M., Riganelli, D., Greco, G., Fleischhauer, K., Sidney, J., Sette, A. and Bach, J. F., The peptide-binding motif for the human transporter associated with antigen processing. J. Exp. Med. 1995. 182: 18831895.
  • 3
    Serwold, T., Gonzalez, F., Kim, J., Jacob, R. and Shastri, N., ERAAP customizes peptides for MHC class I molecules in the endoplasmic reticulum. Nature 2002. 419: 480483.
  • 4
    Doytchinova, I. A., Guan, P. and Flower, D. R., Identifying human MHC supertypes using bioinformatic methods. J. Immunol. 2004. 172: 43144323.
  • 5
    Sidney, J., Southwood, S., Pasquetto, V. and Sette, A., Simultaneous prediction of binding capacity for multiple molecules of the HLA B44 supertype. J. Immunol. 2003. 171: 59645974.
  • 6
    Weinschenk, T., Gouttefangeas, C., Schirle, M., Obermayr, F., Walter, S., Schoor, O., Kurek, R. et al., Integrated functional genomics approach for the design of patient-individual antitumor vaccines. Cancer Res. 2002. 62: 58185827.
  • 7
    Burrows, S. R., Elkington, R. A., Miles, J. J., Green, K. J., Walker, S., Haryana, S. M., Moss, D. J. et al., Promiscuous CTL recognition of viral epitopes on multiple human leukocyte antigens: biological validation of the proposed HLA A24 supertype. J. Immunol. 2003. 171: 1407-1412.
  • 8
    Takedatsu, H., Shichijo, S., Katagiri, K., Sawamizu, H., Sata, M. and Itoh, K., Identification of peptide vaccine candidates sharing among HLA-A3+, -A11+, -A31+, and -A33+cancer patients. Clin. Cancer Res. 2004. 10: 11121120.
  • 9
    Sidney, J., Grey, H. M., Kubo, R. T. and Sette, A., Practical, biochemical and evolutionary implications of the discovery of HLA class I supermotifs. Immunol. Today 1996. 17: 261266.
  • 10
    Hickman, H. D., Luis, A. D., Buchli, R., Few, S. R., Sathiamurthy, M., VanGundy, R. S., Giberson, C. F. and Hildebrand, W. H., Toward a definition of self: proteomic evaluation of the class I peptide repertoire. J. Immunol. 2004. 172: 29442952.
  • 11
    Krüger, T., Schoor, O., Lemmel, C., Kraemer, B., Reichle, C., Dengjel, J., Weinschenk, T. et al., Lessons to be learned from primary renal cell carcinomas: novel tumor antigens and HLA ligands for immunotherapy. Cancer Immunol. Immunother. 2005. 54: 826836.
  • 12
    Bade-Doeding, C., DeLuca, D. S., Seltsam, A., Blasczyk, R. and Eiz-Vesper, B., Amino acid 95 causes strong alteration of peptide position Pomega in HLA-B*41 variants. Immunogenetics 2007. 59: 253259.
  • 13
    Cao, K., Hollenbach, J., Shi, X., Shi, W., Chopek, M. and Fernandez-Vina, M. A., Analysis of the frequencies of HLA-A, B, and C alleles and haplotypes in the five major ethnic groups of the United States reveals high levels of diversity in these loci and contrasting distribution patterns in these populations. Hum. Immunol. 2001. 62: 10091030.
  • 14
    Schipper, R. F., van Els, C. A., D'Amaro, J. and Oudshoorn, M., Minimal phenotype panels. A method for achieving maximum population coverage with a minimum of HLA antigens. Hum. Immunol. 1996. 51: 9598.
  • 15
    Macdonald, W. A., Purcell, A. W., Mifsud, N. A., Ely, L. K., Williams, D. S., Chang, L., Gorman, J. J. et al., A naturally selected dimorphism within the HLA-B44 supertype alters class I structure, peptide repertoire, and T cell recognition. J. Exp. Med. 2003. 198: 679691.
  • 16
    Thorpe, C. J. and Travers, P. J., Prediction of an HLA-B44 binding motif by the alignment of known epitopes and molecular modeling of the antigen binding cleft. Immunogenetics 1994. 40: 303305.
  • 17
    Zhang, C., Anderson, A. and DeLisi, C., Structural principles that govern the peptide-binding motifs of class I MHC molecules. J. Mol. Biol. 1998. 281: 929947.
  • 18
    Zernich, D., Purcell, A. W., Macdonald, W. A., Kjer-Nielsen, L., Ely, L. K., Laham, N., Crockford, T. et al., Natural HLA class I polymorphism controls the pathway of antigen presentation and susceptibility to viral evasion. J. Exp. Med. 2004. 200: 13-24.
  • 19
    Williams, A. P., Peh, C. A., Purcell, A. W., McCluskey, J. and Elliott, T., Optimization of the MHC class I peptide cargo is dependent on tapasin. Immunity 2002. 16: 509-520.
  • 20
    Robinson, J., Waller, M. J., Parham, P., de Groot, N., Bontrop, R., Kennedy, L. J., Stoehr, P. and Marsh, S. G., IMGT/HLA and IMGT/MHC: sequence databases for the study of the major histocompatibility complex. Nucleic Acids Res. 2003. 31: 311314.
  • 21
    Snary, D., Barnstable, C. J., Bodmer, W. F. and Crumpton, M. J., Molecular structure of human histocompatibility antigens: the HLA-C series. Eur. J. Immunol. 1977. 7: 580585.
  • 22
    McCutcheon, J. A., Gumperz, J., Smith, K. D., Lutz, C. T. and Parham, P., Low HLA-C expression at cell surfaces correlates with increased turnover of heavy chain mRNA. J. Exp. Med. 1995. 181: 20852095.
  • 23
    Bihl, F. K., Loggi, E., Chisholm III, J. V., Hewitt, H. S., Henry, L. M., Linde, C., Suscovich, T. J. et al., Simultaneous assessment of cytotoxic T lymphocyte responses against multiple viral infections by combined usage of optimal epitope matrices, anti- CD3lmAb T-cell expansion and “RecycleSpot”. J. Transl. Med. 2005. 3: 20.
  • 24
    Davenport, M. P., Smith, K. J., Barouch, D., Reid, S. W., Bodnar, W. M., Willis, A. C., Hunt, D. F. and Hill, A. V., HLA class I binding motifs derived from random peptide libraries differ at the COOH terminus from those of eluted peptides. J. Exp. Med. 1997. 185: 367371.
  • 25
    Sidney, J., del Guercio, M. F., Southwood, S., Engelhard, V. H., Appella, E., Rammensee, H. G., Falk, K. et al., Several HLA alleles share overlapping peptide specificities. J. Immunol. 1995. 154: 247259.
  • 26
    Barnstable, C. J., Bodmer, W. F., Brown, G., Galfre, G., Milstein, C., Williams, A. F. and Ziegler, A., Production of monoclonal antibodies to group A erythrocytes, HLA and other human cell surface antigens – new tools for genetic analysis. Cell 1978. 14: 920.
  • 27
    Rebai, N. and Malissen, B., Structural and genetic analyses of HLA class I molecules using monoclonal xenoantibodies. Tissue Antigens 1983. 22: 107117.
  • 28
    Weinzierl, A. O., Maurer, D., Altenberend, F., Schneiderhan-Marra, N., Klingel, K., Schoor, O., Wernet, D. et al., A cryptic vascular endothelial growth factor T-cell epitope: identification and characterization by mass spectrometry and T-cell assays. Cancer Res. 2008. 68: 24472454.
  • 29
    Lemmel, C., Weik, S., Eberle, U., Dengjel, J., Kratt, T., Becker, H. D., Rammensee, H. G. and Stevanovic, S., Differential quantitative analysis of MHC ligands by mass spectrometry using stable isotope labeling. Nat. Biotechnol. 2004. 22: 450454.
  • 30
    Weinzierl, A. O., Lemmel, C., Schoor, O., Müller, M., Krüger, T., Wernet, D., Hennenlotter, J. et al., Distorted relation between mRNA copy number and corresponding major histocompatibility complex ligand density on the cell surface. Mol. Cell Proteomics 2007. 6: 102113.
  • 31
    Rammensee, H., Bachmann, J., Emmerich, N. P., Bachor, O. A. and Stevanovic, S., SYFPEITHI: database for MHC ligands and peptide motifs. Immunogenetics 1999. 50: 213219.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors.

FilenameFormatSizeDescription
eji_200838632_sm_supplInfo.pdf360KsupplInfo

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.