Differential Impact of HLA-B*44 Allelic Mismaches at Position 156 on Peptide Binding Specificities and T-Cell Diversity

Single amino acid mismatches at position 156 on the alpha 2 helix of B*44 variants have been described to cause immunological episodes. The magnitude of permissivity between B44/156 variants differs from peptide presentation independent of the peptide loading complex (B*44:28) to influencing clinical episodes (B*44:02 vs. B*44:03). We here investigated if the single exchange of an Asp>Glu as occurring in B*44:35 at residue 156 would force an immune response in vitro.


Introduction
Allogeneic hematopoietic stem cell transplantation (HSCT) is used as a therapeutic strategy for treating many haematological malignancies and certain other immune related disorders. Selection of donors for HSCT involves rigorous assessment and matching of human leukocyte antigen (HLA) types between the donors: recipient pair. The best outcomes following transplantation have been achieved when an HLA identical sibling donor is used for transplantation. However, there is to date still only a 30% chance of finding such a donor and in most instances HLA mismatched stem cells from related or unrelated donors are used for transplantation [1]. Such kinds of transplants are therefore often associated with severe graft-versus-host-disease (GVHD) and transplant rejection [2]. Many clinical studies have demonstrated that the effect of HLA mismatches depends on the nature of exchanged amino acids (AA) and their location within the heavy chain of the class I molecule [3]. As of today, there is no strategy available for the selection of the less permissive class I mismatch when no identical donor but multiple mismatched donors are available due to the enormous degree of HLA class I polymorphisms (n) and the quantity of possible mismatches [magnitude = ½•n• (n-1)] that can occur.
Foreign tissue is recognized through several mechanisms, involving direct and indirect recognition of HLAs and/ or minor histocompatibility antigens (mHags) [4][5][6]. Retrospective analysis of donor: recipient pairs have demonstrated that the strongest effects following transplantation are HLA-dependent [4][5][6][7]. Alloreactivity following HSCT from HLA-identical donor:recipient pairs can be attributed to the immunological effect drived from mHAGs, whereas that following HSCT from HLA mismatched unrelated donors is caused by the cumulative effects of both HLA mismatching and mHags (Table 1). In such a scenario, the magnitude of HLA mismatch effect alone remains unclear and is extremely difficult to separate out from the enormous number of potential mHags present in the human genome.
There is growing evidence highlighting the impact of single AA polymorphisms occurring at certain key residues within the class I molecules in influencing post-transplantation outcomes. One such polymorphism occurs at position 156, located at the center of α2 helix within the HLA-A*24, B*35 and B*44 groups there are selected alleles that differ exclusively by an AA exchange at position 156. Interestingly these "micropolymorphisms" are able to alter the magnitude of permissivity between the B*44 subtypes by selecting and presenting an alternate peptide repertoire independently of the peptide loading complex (B*44:28 156Arg ) [9]. This lack of permissivity has also been highlighted in clinical episodes (e.g B*44:02 156Asp v/s B*44:03 156Leu ) [10]. Currently, there is no reliable method that can be used for selective evaluation of allogenicity due to single mismatches within the HLA class I molecule and to distinguish their effects from the influence of mHAGs. We therefore developed an in vitro system that we termed "the artificial allogeneic system" (Figure 1). The basic principle of our in vitro model is that autologous donor cells from HLA-B*4402 positive donors are transduced with lentiviral constructs encoding for an additional membrane-bound allogeneic HLA class I molecule ( Figure 2), in this case HLA-B*44:03 or B*44:35. The recombinant allo-HLA molecule displayed in the model therefore the sole alloreactive feature. We wanted to determine if the observed allogenicity if any is reflected on the peptide binding features exhibited by these allotypes. The outcome following transplantation is to a large extent influenced by the T cell recognition of class I alleles as self/ antigenic [11] and T-cell recognition of these different alleles is in turn influenced by the ability of these variants to bind peptides with distinct AA sequences [12]. Therefore, we can define histocompatibility based on the repertoire of peptides that are selected and presented by these different allotypes. Our aim was to integrate the data obtained from mass spectrometric sequencing of endogenous peptide repertoire presented by these polymorphic variants along with observed T cell responses in order to determine permissible/ non-permissible mismatches for transplantation within the B*44 allelic group. This strategy is likely to be a highly valuable instrument for the validation of computational allogenicity prediction tools.

Designing lentiviral vectors
cDNA from a HLA-B*4402 positive donor (exons 1-7) was amplified by PCR using the primers HLA-B1-TAS (5` GAGATGCGGGTCACGGCG 3`) and HLA-B-TAAS-E7 (5` TCAAGCTGTGAGAGACACATCAG 3`). The PCR product was ligated into the eukaryotic expression vector pcDNA3.1V5/His using the pcDNA3.1V5/His TA-cloning Kit (Invitrogen, Karlsruhe, Germany). The B*4402 insert was ligated into the pRRL.PPT.SFFV. mcs.pre vector and the insertion verified by sequencing (forward and reverse direction) using an ABI PRISM 377 sequencer (Applied Biosystems, Foster City, CA, USA and its effects must be taken into account. The HLAdifferential and mHAG-identical situation does not occur, and therefore in a natural setting, it is not possible to determine the HLA mismatch effects exclusively. The lower line emphasizes the characteristics of the artificial allogeneic system, where the magnitude of HLA mismatch effects can be determined independent of mHAG effects.  Production of lentiviral particles in HEK 293T cells 6 x10 6 HEK293T cells/10 cm dish were plated out along with 5 ml DMEM/ 10% FCS media (+G418 100 μg/ ml). Lipofectamine (Invitrogen) was added dropwise into a 15 ml falcon tube containing OPTI-MEM (Gibco, Life Technologies GmbH, Darmstadt, Germany) and incubated for 5 min. 10 μg of transfer vector (pRRL HLA-B*44:02 or HLA-B*44:03 or B*44:35) packaging plasmid: 5 μg psPAX2 and envelope plasmid: 5 μg pMD2G were added to another 15 ml falcon tube containing OPTI-MEM. The lipofectamine-OPTI-MEM mixture was added dropwise into falcon tube containing the plasmids, mixed and incubated for 20 min before adding to the HEK293T seeded plates. The media was changed 8 h post transfection. After 24 h, the supernatant containing viral particles was collected and filtered through a 0.45 μm filter (Millipore GmbH, Schwalbach, Germany) and concentrated by centrifuging overnight at 16°C at 10,000 rpm. Lentiviral pellet obtained was dissolved in RPMI/ 2% AB-sera (Invitrogen; c.c. pro, Oberdorla, Germany).
Lentiviral particles for production of soluble HLA class I molecules were produced by transfecting 5x10 6 HEK293T cells with 10 μg of transfer vector (pRRL sHLA-B*44:03 or sHLA-B*44:35) packaging plasmid: 5 μg psPAX2 and envelope plasmid: 5 μg pMD2G. Following media exchange, supernatant containing lentiviral vector was collected, filtered and concentrated by centrifuging overnight at 16°C at 10.000 rpm. The lentiviral pellet was resuspended in RPMI 1640 medium supplemented with 10% FCS and L-glutamine (Invitrogen). The diluted cell suspension was added dropwise over 15 ml of leucosep (Greiner Bio-One, GmbH) in a 50 ml falcon tube and centrifuged at 2000 rpm for 15 min at room temperature without breaking. Following centrifugation, the upper plasma layer was aspirated carefully leaving behind the interphase containing mononuclear cell layer (lymphocytes, monocytes and thrombocytes) undisturbed. The mononuclear cell layer was resuspended in 15 ml of PBS and centrifuged at 1400 rpm for 15 min. The cell pellet was dissolved in 25 ml of PBS and centrifuged at 800 rpm for 15 min. The supernatant was discarded and the cell pellet was resuspended in RPMI/ 2% AB-sera and the cell number was determined. On day 0, freshly isolated PBMCs supplemented with RPMI/ 2% AB-sera were plated on falcon cell culture dishes (BD Biosciences), precoated with Sandoglobulin (30 µg/ml) (Sandoz, Holzkirchen, Germany) and incubated for 1 -1.5 h in a humidified 37°C CO 2 incubator. Monocytes were adherent at this stage, remaining nonadherent cells in the supernatant have been removed. The monocytes were cultured in RPMI/ 2% AB-sera (Invitrogen; c.c.pro) over night. On day 1, monocyte containing media was removed, filtered and stored at -20°C. Monocytes were cultured in fresh RPMI/ 2% AB-sera supplemented with IL-4 (500 U/ ml) (PeproTech GmbH, Hamburg, Germany) and GM-CSF (800 U/ ml) (PeproTech GmbH, Hamburg, Germany). On day 4 if the cells were still adherent, media was removed completely and fresh RPMI/ 2% AB-sera supplemented with IL-4 (500 U/ ml) and GM-CSF (800 Uml) were added to the cells. If the cells were in suspension, additional media along with the cytokines for the entire volume was added. On day 5, the supernatant containing mature dendritic cells (DCs) was collected and centrifuged at 800 rpm. The cells were counted and plated at a density of 1 x 10 6 cells per well in a 6 well cell culture plate supplemented with media containing the cytokines. On day 6, the monocyte conditioned media (from day 1) was added to the cells. Mature DCs were transduced with lentiviral vectors on day 7. The lentiviral transduction of the DCs was performed adding the dissolved lentiviral-pellet in the presence of 8 μg/ mL protamine sulfate (Sigma-Aldrich) to the DC plates followed by incubation for 8 h. Transduced DCs were removed from the plates and stimulation assay performed by co-cultivating recombinant autologous DCs and autologous CD3+ cells.

Isolation of CD3+ cells
PBMCs were isolated from blood of healthy donors by density gradient centrifugation as explained above. CD3+ T cells were isolated from PBMCs using Pan T Cell Isolation Kit II (Miltenyi Biotec, Bergisch Gladbach, Germany) and following the protocol from the supplier. In brief, PBMCs were washed with MACS buffer (Miltenyi Biotec, Bergisch Gladbach, Germany) and centrifuged at 1,200 rpm. Cell numbers were determined and the cells were incubated with biotin-antibody-cocktail (10 µl/10 7 cells) for 10 min at 4°C. 30 µl of MACS buffer and 20 µl of anti-biotin-microbeads were added per 10 7 cells and incub ated for 15 min at 4°C. The cells were washed, centrifuged at 1,200 rpm and the pellet was resuspended in MACS buffer. The labeled cells were passed through a magnetic separation column and flow through containing the CD3+ cells were resuspended in RPMI (Invitrogen) supplemented with 10% human AB sera (c.c.pro). These CD3+ T cells were matured to CD8+ T cells and used further as effector cells in ACT assays and for analysis of T-cell receptor Vβ repertoire.

ACT assay -Assay for cytotoxicity
For the ACT assay, target cells were stained with CFSE (Carboxy-Fluorescein Diacetat Succinimidyl Ester) (Invitrogen) to a final concentration 2 µM. The reaction was stopped 3 min post-labeling and the cells were washed with PBS. Effector cells (stimulated CD8+ T-cells) were counted and the cytotoxicity assay was set up with a target: effector ratio 1:10. The cells were incubated for 24 h at 37°C, washed and incubated with 7-aminoactinomycin (7-AAD) (Life Technologies, GmbH. Germany). Percentage of cytotoxic lysis was measured by flow cytometry using FacsCanto and analysed using FlowJo.

Analysis of T-cell receptor Vβ repertoire
T-cell receptor (TCR) Vβ gene expression was assessed using TCRExpress™ Clonality Detecting Kit according to manufacturer's protocol (BioMed Immunotech, Alachua, FL). In brief, naïve CD3+ T cell population were co-cultured with autologous dendritic cells expressing recombinant HLA-B*44:02 or B*44:03 or B*44:35 allele for 10 days. On day 10, RNA was extracted from 5*10 6 CTL clones, following which cDNA was synthesized and the TCR genes were amplified in a single step of 35 PCR cycles. Specific TCR genes were amplified using TCR Vβ gene-specific oligonucleotides by nested PCR for an additional 25 cycles. PCR products were resolved on a 4% high resolution agarose gel. The banding patterns were assessed and the T cell clonality was determined.

Characterization of HLA-B*44-derived peptides
Peptides obtained from sHLA-B*44 complexes from 2 mg of total protein were differentiated into low and high binding peptides. The trimeric elution fractions were filtered through a 10 kDa MWCO membrane (Millipore, Schwalbach, Germany) and the peptides detected in the flow-through were considered to be low binding. The retentate containing HLA-B*44 trimeric complexes were then treated with 0.1% trifluoroacetic acid (TFA) to elute high binding peptides. Peptides were separated by filtering through a 10 kD MWCO YM membrane (Millipore). Flow through fractions containing the low or high affinity peptides were subjected to Mass Spectrometry using an Eksigent nano-LC Ultra 2D HPLC coupled to an Orbitrap ion trap (Thermo Fischer, Waltham, Massachusetts, USA) providing a very high mass accuracy (< 5 ppm). Mascot software [13] incorporating the IPI human and the respective decoy databases was used to carry out database queries.

Molecular modelling
Modelling of the B*44:35 156Glu allele was carried out with the program DeepView [14] using the available B*44:02 156Asp structure (PDB 1M6O) [15] as a template and mutating 156Asp to 156Glu using the internal rotamer library to find the best side chain orientations with minimum steric clashes. Each model was then subjected to energy minimization as implemented in DeepView. The graphics program PyMOL (http:// www.pymol.org) was used to generate all structure figures.  (Figures 4a-4c). Stimulated CD8+ T cells also gained the expression of activation marker CD69 by day 8 (Figure 4e).  sHLA-B*44 molecules were generated by stable lentiviral transduction of B-LCLs to express these molecules, followed by large scale production of these cells in bioreactors. Supernatant containing sHLA molecules were purified by affinity chromatography using mAb w6/32. Trimeric complexes were eluted using 0.1 M glycine / HCl buffer (pH 2.7). Elution fractions were filtered through a 10 kDa MWCO membrane and the peptides detected in the flow-through were considered to be low binding ones. The retentate containing trimeric complexes were then treated with 0.1 % trifluoroacetic acid (TFA) to elute high binding peptides. Peptides were purified by RP-HPLC and subjected to mass spectrometry using an Eksigent nano-LC Ultra 2D HPLC coupled to an Orbitrap ion trap.

Peptide features
Outcome following transplantation is majorly influenced by T cell recognition of class I alleles bound to peptides with distinct amino acid sequences as self/ antigenic. We used peptide sequencing as a means to determine if polymorphic differences occurring at position 156 in B*44:03 or B*44:35 alleles influenced the peptide binding specificities of these molecules by comparing the peptides eluted from them with B*44:02 derived peptides published earlier [9]. A total of 2 mg of Peptides derived from sHLA-B*44/ 156 variants were classified into low and high binding peptides based on the criteria metioned in the materials and methods section. Peptides were purified by RP-HPLC and sequenced using Eksigent nano-LC Ultra 2D HPLC coupled to an Orbitrap ion trap providing a very high mass accuracy.
The profile of peptides recovered from B*44:35 molecules showed 24 peptides that were low binding and 56 high binding peptides. Interestingly, we were not able to recover any low binding peptides from B*44:03. However, upon TFA treatment, we were able to identify 38 peptides of high affinity.

Peptides eluted from sHLA-B*44:35 demonstrate a strong shift in the binding preference at the peptide C terminus
Peptides eluted from sHLA-B*44:02 or B*44:03 demonstrated exclusively Glu at position 2, while B*44:35-derived peptides exhibited a small fraction of other residues such as Lys/ Leu in addition to Glu based on the frequency of residues at this position (Figures 7a and 7c). Major anchor residues at the C terminus among the peptides derived from B*44:02 was found to be Trp (~ 45%) or Tyr (~ 25%) while B*44:03 demonstrated higher preference for Tyr (38%), followed by Phe (36%) at this position and only a small percentage of Trp (9%) (Figures 7b and 7d). We observed a strong shift in binding prefence at the C terminus among the peptides eluted from B*44:35. Leu (26%), Lys (26%) and Trp (22%) were found to occur most frequently at the C terminus among the low binding peptides recovered from this allotype (Figure 7b). Among the high binding-B*44:35-derived peptides, Lys (51%) was found to be the major anchor residue at the C terminus, followed by Tyr (24%) and Phe (20%) (Figure 7d). Also, Lys at C Low binding peptides -C terminus

sHLA-B*44:35 alleleic variant presents peptides of extraordinary length
Comparison of peptides lengths demonstrated that the low binding peptides eluted from sHLA-B*44:02 were all of canonical length, with nona (23%) and decamer (64%) peptides occuring more frequently. In addition, this allotype was also found to present a small fraction of octamer peptides (6%). In comparision, octamer peptides were absent among the low binding peptides eluted from B*44:35. We were able to identify a significant percentage of non-canonical peptides (24%) amongst this peptide subset, in addition to peptides of canonical lengths (Figure 8a). Majority of the high binding peptides presented by B*44:02 and B*44:03 were of canonical length (~ 90%). In contrast, we found that B*44:35 variants presented a significant proportion (~ 40%) of high binding peptides of extraordinary length, with the longest peptide containing up to 28 AAs (Figure 8b).

A small percentage of the overall peptide repertoire is shared between the three different B*44/156 allotypes
As the B*44/156 variants were all expressed in the same cell line (B-LCL 721.221), their peptide sources were identical, although these subtypes were found to share only a small subset of their overall peptide repertoire (3%). B*44:02 was found to share ~ 8% of its repertoire with peptides recovered from B*44:03 and 6% with B*44:35. We found a slightly higher percentage of overlapping peptide repertoire shared between the B*44:03 and B*44:35 subtypes (11%) (Figure 9).

Molecular modelling of the B*44:35 156Glu micropolymorphism
Our previously described bioinformatic tool "Pocketcheck" [8] is a database of contact data for deposited HLA-class I crystal structures and highlights those heavy chain residues that frequently contact particular peptide residue positions [8]. From these data it is clear that position 156 can contact positions 3, 4, 5, 6 and/or 7 of a bound peptide but is unable to directly contact the C-terminal (PΩ) position of a peptide. Since peptide sequencing from the B*44:35 156Glu allele caused an unexpected change in the PΩ anchor compared to B*44:02 156Asp we generated a molecular model of the B*44:35 156Glu allele to understand the altered molecular features now selecting Lys at this position.
The substitution of Asp (B*44:02) with Glu (B*44:35) at position 156 in our modelling generated a steric clash between 156Glu and 114Asp, suggesting that 114Asp is forced to adopt an alternate conformation ( Figure 10). This subtle change appears to ripple across the floor of the peptide binding groove and we predict that the effect manifests itself by disrupting the contact between 114Asp and 116Asp that is partially mediated through an active water molecule. The resulting geometry of 116Asp now suggests a more favourable orientation to bind peptides containing a C-terminal Lys residue.

Discussion
Donor selection process during HSCT involves rigorous assessment and matching of HLA types between the donor:recipient pair [3]. The best clinical results following transplantation have been achieved when HLA matched sibling donor is used for transplantation. However, there is only a 30% chance of finding such a donor and in most instances, stem cells from the most compatible HLAmismatched related/ unrelated donor are used for transplantation [1]. Polymorphisms occurring in HLA class I molecules determine peptide selection and thus the conformation of the whole pHLA complex. The effect of individual polymorphisms remains poorly understood due to discrepancies arising post-transplantation [16,17]. Therefore, it is of crucial importance to determine the effect of every individual amino acid (AA) polymorphism occurring within the class I molecules to evaluate the best permissible mismatches in order to reduce/avoid post-transplant complications.
To date, there is an increase in the number of stem cell donors and availability of high resolution four-digit HLA typing, supporting the chance to find a well matched donor: recipient pair. However, there is currently no strategy to determine the allogenicity of selected single AA exchanges to support clinical decisions during a transplantation scenario. If several donors, each with different HLA class I mismatches are available, the relative alloreactivity between different mismatches cannot be predicted due to lack of adequate in vitro and in vivo data for ranking mismatch effect. Therefore, we aim to resolve this issue by integrating the observed allogeneic T cell responses in vitro along with the peptide sequencing data to determine if a particular mismatch might be acceptable for transplantation. Our artificial allogeneic system provides a tool to distinguish the effects of mHAGs from those of HLA mismatches. In this system, observed allogeneic T cell responses can be attributed exclusively due to the specific mismatched AA between the donor and recipient HLA class I molecule. In order to validate our artificial allogeneic system, we chose the HLA-B*44 group as it is most commonly occurring group in the Caucasian population and contains allelic subtypes with single AA mismatches occurring at  position 156 within the α2 helix of HLA class I heavy chain. AA at position 156 is regarded to be critical in the donor selection process because mismatches at this position are found to cause alloreactive T cell responses [10,18]. Within the HLA-B*35 group, it was observed that the HLA-B*35:01 156Leu versus B*35:08 156Arg mismatch elicited strong cytotoxic T cell responses, most likely due to conformational changes occurring within the peptide binding region (PBR) [19]. This conformational change alters the peptide-binding affinity, resulting in presentation of peptides with allele-specific epitopes differing both in their lengths and immunogenicity [20]. Within the B*44 group it was shown that mismatching HLA-B*44:02 156Asp with B*44:03 156Leu was found to result in transplant rejection and acute GvHD, indicating that they represent significant barriers to clinical transplantation [10]. Another naturally occurring polymorphism within this allelic group is B*44:35 156Glu that shares the same nature of polar, AA group at position 156. We wanted to determine if this AA exchange would be permissible in a transplantation setting using our artificial allogeneic system system. As outcome following HSCT is to a large extent influenced by the T cell recognition of peptides bound to HLA class I molecules as self/ allogeneic [11], histocompatibility can further be defined based on the features and repertoire of peptides presented by specific allotypes. We wanted to investigate if the observed allogenicity between the B*44/ 156 variants reflected on the peptides presented by them. Although the majority of HLA alleles present peptides terminating in hydrophobic AAs the proteasome has at least three different catalytic activities: trypsin-like (i.e. cleavage after basic AAs), chemotrypsin-like (i.e. cleavage after large, hydrophobic AAs) and peptidyl-glutamyl-peptide hydrolyzing activity (i.e. cleavage after acidic AAs). Here we used soluble HLA technology and investigated the endogenous peptides presented by B*44:02, B*44:03 and B*44:35 allotypes to determine if there are any significant alterations in the binding motifs and lengths in peptides presented by them. We found only 3% of the overall peptide repertoire to be shared between these allotypes. Systematic characterization of the ligands derived from the B*44 allotypes demonstrated that the p2 position in B*44:02 and B*44:03 was conserved and exclusively Glu occured at this position while in B*44:35, other residues such as Lys/ Leu also occured in addition to Glu. Our peptide elution data derived from B*44:35 highlighted a strong shift in the binding preference for Lys at the C terminus, while this residue was completely absent in peptides eluted from B*44:02 and B*44:03.
The observation of such a change in the peptide anchor in B*44:35 as a result of a highly conserved substitution at position 156 was unexpected. Our molecular modelling of the B*44:35 structure has however allowed us to postulate how this may be achieved and subtle differences in the conformations and energetics of the "floor" of the peptide binding grove clearly influence the ability of individual alleles to select specific peptide repertoires. The selection of Lys as a P anchor is relatively uncommon, compared to many other alleles, however the binding of peptides with Lys at the C terminus (27%) versus its complete absence has been demostrated for the ankylosing spondylytis (Bechterew's disease) which is associated with B*27:05 116Asp but not with its protective counterpart B*27:09 116His . Additionally it has been demonstrated in B*27:05 and B*27:09 alleles that the binding of peptides containing a Lys PΩ anchor appears to be weaker and thermodynamically less stable [21]. Such allosteric mechanisms might influence not only the association and dissociation of the trimeric complexes, but also the mode of recognition by the T-cell receptors [22]. Highlighted as sticks is the pΩ peptide anchor (yellow/ dark pink), the residues comprising the pΩ binding pocket (orange), residues 156Glu, 114Asp and 116Asp from the HLA-B*44:35156Glu model are coloured cyan. b) A close up view of the molecular details depicted in a), the steric clash by introducing 156Glu disrupts the 97Arg-114Asp-water-116Asp hydrogen bonding network (red dashes) by forcing a conformational change in 114Asp. 116Asp is then more likely to be in a receptive conformation to accept Lys as a pΩ anchor residue. Analysis of the peptide binding affinities showed that there were no peptides of low affinity that could be recovered from the B*44:03 molecules, indicating a role for Leu at 156 in increasing the stability of the peptide-HLA complex. Although the binding affinity of the peptide is not indicative of the relative immunogenicity of the peptide, higher affinity peptides generally extend the half-life of the peptide-HLA complex at the cell surface, thereby increasing the time available for T-cell receptor recognition. In addition to variations in the binding motifs and binding affinities, we also found marked differences in the length of peptides presented by these allotypes. The majority of pepides presented by B*44:02 and B*44:03 were 9-11 AAs long, with a small fraction of octameric peptides. B*44:35 variants on the other hand were found to present a significantly high number of peptides of extraordinary length, accounting for ~ 47% of peptides, with the longest peptide containing up to 28 AAs. In addition, HLA-B*44:35 is found to disfavour the binding and presentation of octamer peptides.
These data highlight the need for direct comparison of HLA alleles at many levels in order to determine how permissive individual micropolymorphisms can be. Indeed it is clear that neither direct sequence comparison nor molecular modelling would have predicted the change in PΩ anchor motif observed by direct sequencing of the bound peptide repertoire. Only then we can see, using the detailed structures of HLA molecules, how such subtle conformational changes and why the manipulation of a few hydrogen bonds within a protein could ultimately be responsible for successful transplantation and patient survival.