Naturally processed peptides spanning the HPA-1a polymorphism are efficiently generated and displayed from platelet glycoprotein by HLA-DRB3*0101–positive antigen-presenting cells

Human platelet antigen (HPA)–1a and HPA-1b alloantigens are determined, respectively, by the polymorphism Leu or Pro at position 33 of the β₃ chain (glycoprotein IIIa [GPIIIa]) of the platelet integrin α_IIbβ₃.¹  Mismatch between fetal and maternal platelet antigens can lead to the production of maternal immunoglobulin G anti–HPA-1a alloantibodies that cross the placenta and cause thrombocytopenia in an HPA-1a–positive fetus.^2,3  Approximately 2% of white women are HPA-1b homozygous, and 10% of these develop anti–HPA-1a alloantibodies during pregnancy.^2,3  The estimated incidence of severe neonatal alloimmune thrombocytopenia (NAIT) is approximately 1 in 1100 neonates, and the clinical manifestations range from mild purpura to intracranial hemorrhage.^2,3  Maternal responsiveness to HPA-1a shows strong associations with major histocompatibility complex class II, with greater than 95% of responders carrying the DRB3*0101 allele encoding human histocompatibility leukocyte antigen (HLA)-DR52a,^4,5  and up to 94% may carry the DQB1*02 allele.⁶  Classical immunoglobulin G antibody responses are dependent on T-helper (Th) cells that recognize antigen as short, processed, linear peptides bound to major histocompatibility complex molecules on antigen-presenting cells (APCs). The identification of the immunodominant helper epitopes on the integrin sequence containing the HPA-1a polymorphic site would be an important step toward understanding and controlling the immune response to HPA-1a. Earlier studies have shown that short, synthetic HPA-1a peptides containing the Leu³³ polymorphism bind to recombinant DR52a molecules, whereas the Pro³³ version does not, with Trp²⁵ and Leu³³ predicted to be important anchor residues.⁷  The dominant synthetic HPA-1a peptides recognized by peripheral blood Th cells from women alloimmunized with anti–HPA-1a contain the Trp²⁵-Leu³³ core sequence, with Leu³³ preferentially at the C-terminus.⁸  Recent characterization of the fine specificity of Th cell clones derived from women with NAIT also mapped a putative core epitope spanning Trp²⁵-Leu³³.⁹  However, antigen processing within APCs is highly selective, with very few peptides typically generated from any given protein for efficient display to Th cells,^10–13  and there is no direct evidence that this core sequence is indeed presented from platelet glycoprotein. The aim of this study was therefore to determine whether DRB3*0101 homozygous APCs, pulsed with HPA-1a, naturally process and present peptides spanning the predicted Trp²⁵-Leu³³ core epitope.

DNA encoding HPA-1a amino acids 1-62 was cloned into the expression vector pGEX-6P-1, and Escherichia coli BL21(DE3) cells were transformed. This sequence corresponds to the complete plexin-semaphorin-integrin (PSI) domain of platelet GPIIIa, which spans the Leu³³ HPA-1a polymorphism and expresses the HPA-1a antigen.¹⁴  Because the PSI domain is predicted to be on the exterior of the GPIIIa protein¹⁴  and accessible to processing enzymes, we considered this fragment to be representative of the whole protein. The expression of recombinant HPA-1a fused with glutathione S-transferase was induced using isopropyl-β-d-thiogalactopyranoside. The recombinant fused protein was recovered from bacterial lysate and confirmed to bind anti–HPA-1a antibodies, which are sensitive to the conformation of the PSI domain, indicating the integrity of the HPA-1a antigen expression.¹⁵ 

HHKB,¹⁶  an Epstein-Barr virus–transformed B-cell line homozygous for HLA-DR B3*0101, was cultured to a concentration of 10⁷ cells/mL (total cells 0.7 × 10¹⁰), and then pulsed with 1 mg/mL HPA-1a recombinant antigen. Cells were maintained in RPMI 1640 at 4°C for 2 hours, and then 10 hours at 37°C in 5% CO₂, before washing and lysis in 20 mM Tris-HCl (pH 8)/1% Nonidet P-40/150 mM NaCl/5 mM EDTA (ethylenediaminetetraacetic acid), containing protease inhibitors (phenylmethylsulfonyl fluoride, aprotinin, leupeptin, and pepstatin A). Insoluble material was cleared by centrifugation, and the supernatant was filtered (0.45 μM). HLA-DR–peptide complexes were purified from the filtrate by immunoaffinity chromatography using monoclonal anti–HLA-DR antibody (L243) conjugated to protein A-Sepharose. Complexes were eluted from the column with 50 mM diethylamine (pH 11.5) and immediately neutralized in 2 M Tris-HCl (pH 7.4). The HLA-DR–bound peptides were acid eluted at 37°C for 3 hours in 10% acetic acid before ultrafiltration.

The peptide pool was separated by capillary reverse-phase (RP)–high performance liquid chromatography (HPLC) using a C-18, 3-cm × 2.1-mm Brownlee cartridge column. The eluted class II-associated peptide fractions were analyzed using an HCTultra PTM Discovery System (Bruker Daltonics), coupled to an UltiMate 3000 LC System (Dionex). Peptide peaks were detected and deconvoluted automatically using Data Analysis software (Bruker). Mass lists in the form of Mascot Generic Files were created automatically and used as the input to Mascot tandem mass spectrometry (MS/MS) Ions searches of the National Center for Biotechnology Information database.

Direct biochemical methods were used to identity any peptides processed and displayed from the HPA-1a antigen by APCs expressing the HLA-DRB3*0101 susceptibility allele. An HLA-DRB3*0101 homozygous B-cell line was pulsed with the recombinant HPA-1a PSI domain as antigen, the HLA-DR class II/peptide complexes were recovered, and the peptides were eluted for analysis. RP-HPLC of the peptide pool revealed, as expected, a heterogeneous mixture of sequences (Figure 1A), and 45 fractions were collected. Matrix-assisted laser desorption ionization–time of flight mass spectrometry demonstrated that the most abundant peptides had individual molecular masses of 1502-1831 Da, reflecting typical lengths of 13-15 amino acids. Further characterization using a quadrupole ion trap, equipped with an electrospray ion source, enabled sequence determination of individual peptides.

In fractions 21-23, HPA-1a–derived sequences scored highest in the Mascot search, among those peptides recovered. The profile of these abundant sequences identified a “nested set” of 4 naturally processed and presented HPA-1a peptides (Table 1). The sequences were up to 17 amino acids in length and spanned the Leu³³ polymorphism. Each peptide contained the core residues of the predicted dominant Th-cell epitope (Trp²⁵-Leu³³), and although heterogeneous at the N-terminal, all terminated at Arg³⁷. MS/MS analysis of the most abundant HPA-1a–derived peptide, a 16-mer with an m/z value of 934.15²⁺, is shown in Figure 1B through D.

There have been few other successful studies to identify the class II–associated peptides that are naturally processed from specific antigens, reflecting the technical difficulty in characterizing the complex peptide pools displayed by APCs. Other examples include the classic foreign antigens, hen egg lysozyme^10,17  and tetanus toxoid¹¹ ; the human autoantigens of the α-chain of type IV collagen¹² ; and GAD65¹³/insulin¹⁸ targeted, respectively, in Goodpasture's disease and type I diabetes. The current work is the first to report the sequences of naturally processed and presented peptides from a human alloantigen. The dominance of the HPA-1a–derived peptides, which were the most abundant sequences eluted from the APCs, is particularly striking, and will reflect both a high efficiency of processing and the predicted high affinity of the core epitope Trp²⁵-Leu³³ for the DR52a class II molecule expressing HLA-DRB3*0101. It is increasingly recognized that high affinity for class II alone does not ensure presentation of peptides, and that processing constraints play a key role in epitope selection.^12,13 

The HPA-1a peptides shown to be presented in this study form a “nested set” around the core epitope, a feature typical of the naturally processed peptides derived from other antigens.^12,13  This feature arises when class II molecules bind and protect a corresponding core sequence while this is still part of relatively large antigen fragments at an early stage of processing.¹⁹  Subsequent trimming of the fragments is variable, producing a characteristic set of peptides with different lengths spanning the core epitope.^20,21 

T-cell stimulation⁸  and peptide-binding²²  approaches with synthetic peptides have previously been used to map Th epitope(s) of HPA-1, and the polymeric residue Leu³³ of integrin β₃ was predicted to be a major anchor residue in the putative DRB3*0101 (DR52a) peptide-binding motif. In this study, we demonstrate for the first time that the predominant naturally processed and presented HPA-1a peptides span this predicted epitope. The efficient processing and presentation of the peptides are likely to be an important contributory factor in the immunogenicity of HPA-1a. Such peptides may also provide the basis for novel treatments to tolerize the corresponding Th response in HPA-1b1b women at risk of NAIT with an HPA-1a–positive fetus.

We are grateful to the Aberdeen University Proteomics Unit for advice on mass spectrometry analysis.

This work was supported by a PhD studentship from the Ministry of Health and Medical Education of Iran, and a grant from the Scottish National Blood Transfusion Service.

Contribution: G.A.S. performed experiments, discussed the data, and wrote the manuscript; and M.M., R.N.B., and S.J.U. designed the study, discussed the data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Stanislaw J. Urbaniak, Division of Applied Medicine, School of Medicine and Dentistry, Academic Transfusion Medicine Unit, Regional Transfusion Centre, Foresterhill Rd, Aberdeen, AB25 2ZW, United Kingdom; e-mail: s.j.urbaniak@abdn.ac.uk.