Identification of Biological Relevant Minor Histocompatibility Antigens within the B-lymphocyte–Derived HLA-Ligandome Using a Reverse Immunology Approach

Allogeneic HLA-matched hematopoietic stem cell transplantation (allo-SCT) and subsequent donor lymphocyte infusion (DLI) to eradicate persistent or relapsed malignant cells are considered an effective curative treatment for patients with high-risk hematologic malignancies (1). The curative potential of this therapy has been attributed to the recognition of malignant cells by donor T cells. Detailed analyses of T-cell immune responses in patients responding to DLI have demonstrated that donor T cells can recognize minor histocompatibility antigens (MiHA) presented by HLA molecules on malignant cells. MiHA are peptides derived from polymorphic proteins that differ between donor and recipient due to single-nucleotide polymorphism (SNP; refs. 2–5). Previously, it has been demonstrated that T cells directed against MiHA that are ubiquitously expressed can mediate life-threatening GVH (6), whereas T cells directed against MiHA with hematopoietic restriction may mediate graft-versus-leukemia (GVL) response in absence of GVHD (7).

Although significant strides in MiHA discovery have been made (8, 9), a major limitation for clinical implementation is the small number of identified MiHA derived from genes that are exclusively expressed in hematopoietic cells. To allow the selective analysis of hematopoietic-restricted MiHA, we and others have used reverse immunologic approaches in which predicted polymorphic peptides are the starting point and peptide candidates are subsequently screened for their capacity to induce a specific T-cell response. This approach has the potential to screen for T cells recognizing MiHA that are exclusively expressed by hematopoietic cells. However, it has been reported that when such peptide predictions are solely based on computer algorithms that predict peptide–HLA binding affinity and proteolytic cleavage, the detected T-cell responses are often directed against epitopes that are not naturally processed and therefore fail to lyse malignant target cells (10–12). To circumvent this peptide selection problem, we previously introduced mass spectrometry (MS)-based HLA-ligandomes as a reliable source for naturally processed and presented peptides and developed an algorithm that can be exploited to identify T cells specific for potential clinically relevant MiHA (13, 14).

In this study, we selected the 25 most promising MiHA candidates from a large set of B lymphocyte–derived HLA class I eluted peptides (15). To validate these MiHA candidates, it is essential to demonstrate their immunogenic potential by isolation of high avidity–specific T cells. We therefore screened the T-cell receptor (TCR) repertoire of 16 unrelated donors for the presence of T cells specific for these MiHA candidates. MiHA-specific MHC multimer–positive T cells were isolated from peripheral blood by magnetic-activated cell sorting (MACS). Subsequently, functional testing of MHC multimer sorted and expanded T-cell clones demonstrated the immunogenic potential of LB-CLYBL-1Y, LB-TEP1-1S, and 2 previously described MiHA. For one of the newly defined MiHA, we were able to confirm the biologic relevance by demonstrating MHC multimer–positive T cells in a patient suffering from acute myeloid leukemia (AML) after treatment with allo-SCT and subsequent DLI. Our data illustrate that with this reverse immunology approach, biologically relevant MiHA can be identified as well as MiHA that are not frequently induced in vivo but can potentially be used for immunotherapeutic strategies.

Peripheral blood was obtained from different individuals after informed consent (Sanquin Reagents, Amsterdam; and Leiden University Medical Center, Leiden, The Netherlands). All experiments were approved by the local medical ethics committees. Blood samples were HLA-typed by high-resolution genomic DNA typing. Peripheral blood mononuclear cells (PBMC) were isolated by ficoll gradient separation and cryopreserved for further use. T cells were cultured in T-cell medium consisting of Iscove Modified Dulbecco Media (IMDM; Lonza) supplemented with 5% human serum (HS), 5% FBS, 100 IU/mL IL2 (Proleukin), 2 mmol/L l-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). Epstein-Barr virus–transformed lymphoblastoid B cell lines (EBV-LCL) and phytohemagglutinine (PHA; Murex Diagnostics) blasts were generated using standard procedures. After generation, stock ampules are frozen for which the genetic identity is confirmed by in-house WGA profiling (Illumina Human1M-duo array) as previously described (9). The T2 cell line was cultured in IMDM with penicillin/streptomycin and 5% FCS. Cell lines are no longer cultured then 3 months.

The peptides used for this study are derived from a newly established large peptide library that has been recently described by Hassan and colleagues (15). In short, peptide elution, reverse-phase high-performance liquid chromatography (RP-HPLC), and tandem mass spectrometry (MS/MS) were carried out as previously described (16, 17). Briefly, 60 × 10⁹ HLA-A*0201 and HLA-B*0702–positive EBV-LCL were washed with ice-cold PBS and pellets were stored at −80°C until use. Peptide HLA class I complexes were purified from cell lysate by affinity chromatography. Subsequently, peptides were eluted from HLA molecules and separated from the HLA heavy chain fragments and β2-microglobulin by size filtration. The peptide mixture was separated by various first-dimension separation techniques, after which the peptides were measured by on-line nanoHPLC-MS/MS.

MiHA-candidates were selected on the basis of our recently developed MiHA selection algorithm from the recently established peptide elution library (13). Briefly, the tandem mass spectra of eluted peptides were submitted to the HSPVdb (18), a database optimized for finding polymorphic peptides. To select for MiHA candidates within this set, we evaluated the polymorphic peptides using strict threshold scores for mass spectrometry defined sequence reliability (BMI ≥ 30, ppm ≤ 2.0), peptide length (8–11 amino acids), predicted peptide–HLA affinity (<500 nmol/L), allele frequencies of the SNP encoding polymorphism (0.05%–0.7%), and specific gene exclusion criteria (no extreme polymorphic genes). After confirming their integrity by comparing the tandem mass spectra of the synthetic peptides with that of the eluted counterparts, the top 25 MiHA candidates, with the highest threshold scores, were selected for further analysis.

All peptides were synthesized in-house using standard Fmoc chemistry. Recombinant HLA-A*0201 and HLA-B*0702 heavy chain and β₂m were in-house produced in Escherichia coli. MHC class I refolding was performed as previously described with minor modifications (19). MHC class I complexes were purified by gel filtration HPLC in PBS and stored at 4°C. The peptide HLA-A*0201 or HLA-B*0702 binding affinity was assessed by subjecting prefolded UV-sensitive peptide–MHC complexes (100 μg/mL) to 366 nm UV light (Camag) for 1 hour in the presence of the peptide of interest (200 μmol/L; 20). As controls, the CMV PP65 NLVPMVATV, CMV PP65 TPRVTGGAM, modified MART1 ELAGIGILTV, and wild-type MART1 AAGIGILTV peptide were used. After exchange, samples were spun at 16,000 × g for 5 minutes, and supernatants were used to assess HLA monomer rescue using a bead assay as previously described (21).

Before isolation, PBMC samples were stained with phycoerythrin (PE)-coupled MHC multimers for 1 hour at 4°C. Subsequently, cells were washed and incubated with anti-PE beads (Miltenyi Biotec). PE-positive T cells were isolated by MACS, according to the manufacturer's protocol. Positive fraction was cultured with irradiated autologous feeder cells in T-cell medium supplemented with 5 ng/mL IL15 (Biosource) and anti-CD3/CD28 Dynabeads (Invitrogen). After 2 weeks, T-cell cultures were analyzed by MHC multimer combinatorial coding flow cytometry. Data acquisition was performed on an LSR-II flow cytometer (BD Biosciences), and MHC multimer–positive T cell populations were single-cell sorted on a FACSAria (BD) into round-bottomed 96-well plates containing 100 μL T-cell medium supplemented with 800 ng/mL PHA and 1 × 10⁵ irradiated feeder cells. TCR-β variable chain (TCR-Vβ) usage was investigated by flow cytometry using specific monoclonal antibodies as included in the TCR-Vβ repertoire kit (Beckman Coulter).

For analysis of IFNγ and granulocyte macrophage colony-stimulating factor (GM-CSF) production, 5,000 T cells were stimulated with 20,000 EBV-LCL or 10,000 fibroblasts loaded with different concentrations of peptides in round-bottomed 96-well plates. Before stimulation, fibroblasts were either pretreated with IFNγ (100 IU/mL) or not for 5 days. For recognition of primary malignant cells, 1,000 T cells were stimulated with 5,000 malignant cells in a 384-well plate. After 18 hours, supernatants were harvested, and the concentration of IFNγ and GM-CSF were measured by an ELISA (Sanquin Reagents). An arbitrary detection limit was set to 100 pg/mL for both cytokines.

A panel of 100 HLA-typed EBV-LCL was selected for SNP screening as previously described (9). Briefly, gDNA was isolated of 5 × 10⁶ EBV-LCL cells using Gentra Puregene Cell Kit (Qiagen), and PCR-free whole-genome amplification (WGA) was performed. The DNA samples were hybridized to Illumina Human1M duo arrays. For gene expression analysis, total RNA was isolated using small and micro-scale RNAqueous isolation kits (Ambion) and amplified using the TotalPrep RNA Amplification Kit (Ambion). After preparation using the whole-genome gene expression direct hybridization assay (Illumina), cRNA samples were dispensed onto Human HT-12 v3 Expression BeadChips (Illumina) according to the manufacturer's protocol. Sample collection was performed as previously described (22–24). Microarray gene expression data were analyzed using R 2.15.

We have recently reported a MiHA selection algorithm to be able to select MiHA candidates from a library of HLA class I eluted peptides (13). This MiHA selection algorithm comprises several evaluation steps that are summarized in Material and Methods. This algorithm was used to screen our newly established library of eluted HLA class I peptides derived from multiple HLA-A*0201 and B*0702–positive EBV-LCLs, to select for potential MiHA candidates (15). To validate this newly established library of approximately 16,000 eluted HLA class I peptides comprising mainly monomorphic peptides, we first screened for the presence of known MiHA. Peptide sequences or their relevant length variants were identified for 10 of 13 MiHA that were expressed by the EBV-LCLs as revealed by SNP genotyping (Supplementary Table SI; refs. 3, 8, 9), illustrating the high quality of this peptide elution library.

In the next step, we selected a set of 25 MiHA candidates using the MiHA selection algorithm, including 22 novel MiHA candidates as well as the previously reported LB-NISCH-1A, LB-ERAP1-1R and LB-GEMIN4-1V MiHA (Supplementary Table SII). We first analyzed the capacity of the 25 MiHA candidates to stabilize a peptide–HLA complex in an HLA binding assay that is based on UV-induced conditional ligand cleavage as described previously (10, 21, 25). After UV exchange, the HLA-binding affinity of the tested peptides was normalized to the high-affinity control peptides CMV-PP65-NLV and CMV-PP65-TPR for HLA-A*0201 and B*0702, respectively (Supplementary Fig. S1A and S1B). MiHA candidates were selected when their HLA-binding affinity exceeded that of the MART1-WT-AAG peptide, which binds with low affinity to HLA-A*0201. HLA-binding affinity as measured with HLA rescue scores exceeded that of the MART1-WT-AAG control for 8 HLA-A*0201 and 13 B*0702 binding MiHA candidates. Peptide sequences and allele frequencies of the MiHA are shown in Table 1.

To validate the 21 peptides as MiHA with immunogenic potential, we generated MHC multimers and analyzed the T-cell repertoire of 16 healthy HLA-A*0201 and B*0702–positive donors for MHC multimer reactive T-cells. MiHA-specific T cell lines were generated by incubating 100 × 10⁶ PBMC with a specific set of MHC multimers, followed by enrichment of MHC multimer–positive cells on a magnetic column. To allow the isolation of high-avidity T-cell populations, the set of MHC multimers was specifically adjusted for each PBMC sample to cover only those MiHA for which the encoding SNP was screened homozygous negative in the respective donor, as in this individual's T-cell repertoire, high-avidity MiHA-specific T cells will not have been deleted because of negative selection. The set of PBMC donors was specifically adjusted to cover as many applicable donors per MiHA candidate. Unfortunately, no homozygous negative donors were found for the SCRIB-1L MiHA candidate. The MHC multimer–enriched T cells were expanded for 14 days in presence of αCD3/28 beads, IL2 and IL15. To increase the frequency of MHC multimer–positive T cells, a second pull-down was performed at day 14 using the identical initial set of MiHA candidate–specific MHC multimers. After both rounds of MHC multimer enrichment, we analyzed the expanding T cell lines for the presence of MHC multimer–positive T cells by FACS. A representative FACS analysis after 2 rounds of MHC multimer enrichment is demonstrated in Fig. 1A in which 4 MHC multimer–positive T-cell populations specific for HLA-B*0702 binding MiHA candidates are detected in donor OMH. After the first MHC multimer pull down, MHC multimer–positive T-cell populations were detected specific for 11 of the 20 tested MiHA candidates in one or more T-cell cultures. MHC multimer–positive T-cell frequencies varied between 0.01% and 5.0% of total CD8⁺ T cells (Supplementary Table SIII). These low frequencies are most likely due to the very low frequency of MHC multimer–positive T cells in the naïve repertoire. After the second MHC multimer pull down, MHC multimer–positive T-cell populations were detected specific for an increased number of 16 of the 20 tested MiHA candidates with frequencies up to 85% of total CD8⁺ T cells (Supplementary Table SIII). As demonstrated in Fig. 1B, T cells reactive with the CLYBL-1Y and LB-GEMIN4-1V MHC multimer were frequently detected in 8 of the 16 and 6 of the 9 enriched T cell lines, respectively. In contrast, T cells specific for the other MiHA appeared more restricted to a few donors.

Next, a total number of 806 MHC multimer–positive T-cell clones representing all detected T-cell populations were generated by FACS. To demonstrate recognition of potential MiHA candidates, all 806 T-cell clones were stimulated with specific antigen, and cytokine production was determined as a measure for antigen-specific reactivity. Because the T-cell clones were most likely derived from the naïve T-cell repertoire and have presumably not all acquired the capacity to produce IFNγ upon antigen encounter, we first screened all T-cell clones for their potential to secrete IFNγ after αCD3/28 stimulation. As demonstrated in Fig. 2A, the generated T-cell clones demonstrated a broad range of IFNγ secretion. To investigate whether GM-CSF was of additional value to improve the screening efficiency, the GM-CSF production of part of the T-cell clones with variable IFNγ secretion potential was measured after αCD3/28 stimulation. The results demonstrate that a substantial number of T-cell clones with poor intrinsic IFNγ production were able to produce pronounced GM-CSF levels (Fig. 2B). Therefore, by using both IFNγ and GM-CSF as a readout, we could increase the number of MHC multimer–positive T-cell clones that could be screened. By setting an arbitrary detection limit to 100 pg/mL for both cytokines, we increased the number of T-cell clones that could subsequently be screened for MiHA-specific reactivity from approximately 85% (IFNγ only) to 98% (IFNγ and GM-CSF). By including GM-CSF release as selection criteria, 108 more T-cell clones could be screened for MiHA-specific reactivity.

To demonstrate the capacity of the isolated MHC multimer–positive T-cell clones to recognize specific MiHA peptides, we stimulated all T-cell clones with HLA-A*0201 and B*0702–positive T2 target cells loaded with titrated concentrations of specific peptide. MiHA-specific T-cell reactivity was observed for 8 of 16 tested MiHA candidates, and either IFNγ or GM-CSF was detected after stimulation with CLYBL-1Y, ERAP1-1R, GEMIN4-1V, NISCH-1A, HMMR-1V, C18orf21-1A, APOBEC3H-1K, and TEP1-1S MiHA peptides. For each donor, the MiHA-specific T-cell clone that demonstrated the highest peptide avidity is shown (Fig. 3A–E). MiHA-specific T-cell clones demonstrated variable peptide avidity and half maximum cytokine production (IC₅₀) varied between an IC₅₀ of approximately 100 pmol/L for the high-avidity T-cell clone K156 specific for GEMIN4 (Fig. 3C) and an IC₅₀ of approximately 62.5 nmol/L for the low-avidity T-cell clone K337 specific for C18orf21 (Fig. 3E). For the CLYBL-1Y MiHA candidate, MHC multimer–positive T-cell clones were successfully isolated from 4 different donors. High-avidity CLYBL-1Y–specific T-cell clones were isolated from donor ABM, whereas only low-avidity or nonreactive T cells were isolated from donor UDN, EPP, and FHT, respectively (Fig. 3A). For the previously described MiHA ERAP1-1R, both high- and low-avidity T-cell clones were isolated from the 2 donors that demonstrated MHC multimer–positive T-cell populations after pull down. For the previously described MiHA GEMIN4-1V, high-avidity T-cell clones were only isolated from 4 of 6 donors that demonstrated MHC multimer–positive T-cell populations after pull down (Fig. 3B and C; low-avidity T-cell clones not shown). The NISCH-1A, HMMR-1V, C18orf21-1A, and APOBEC3H-1K–specific T-cell clones were each successfully isolated from one donor and exhibited high (HMMR-1V) to low peptide avidity (NISCH-1A, C18orf21-1A, and APOBEC3H-1K; Fig. 3D). For the TEP1-1S MiHA candidate, high-avidity T-cell clones were isolated from donor EPP. T-cell clone K091 demonstrated high peptide–specific GM-CSF production and low IFNγ production (Fig. 3E). No peptide-specific T-cell clones specific for the other MiHA candidates were observed.

All T-cell clones that demonstrated peptide-specific cytokine production were stimulated with a panel of 10 SNP-genotyped HLA-A*0201 and B*0702–positive EBV-LCL target cells to screen for reactivity against endogenously processed and presented peptide. As demonstrated in Fig. 3F, G, H, and J, all high-avidity T-cell clones specific for CLYBL-1Y, TEP1-1S, ERAP1-1R, and GEMIN4-1V demonstrated recognition of all target cells that endogenously process and present their respective MiHA, whereas targets that were negative for the MiHA were not recognized. Surprisingly, the high-avidity T-cell clone K507 specific for HMMR-1V did not show a recognition pattern that correlated with MiHA expression. This may be caused by the absence of endogenously presented MiHA peptide by some of the SNP-positive EBV-LCL or by the recognition of allo-HLA molecules expressed by EBV-LCL. No or only marginal target cell recognition was observed for the NISCH-1A, C18orf21-1A, and APOBEC3H-1K–specific T-cell clones (Fig. 3I). These results indicate that the MiHA CLYBL-1Y and TEP1-1S represent potentially immunologic relevant MiHA candidates.

To validate the biologic relevance of the MiHA candidates, we analyzed the peripheral blood of patients suffering from various hematopoietic malignancies that received an allo-SCT and DLI and demonstrated a clinical response revealed by declining patient chimerism, for the detection of MiHA-specific T cells. Patients were only screened with MiHA-specific MHC multimers when they were positive for the MiHA and received a DLI from a donor who was homozygous-negative for the SNP encoding MiHA (Supplementary Table SIII). Tested patient samples were obtained during or after the peak response, 5 to 8 weeks after treatment with DLI. For the newly identified CLYBL-1Y MiHA, MHC multimer–positive T cells were detected in 1 of the 3 screened patients (Fig. 4A) with a frequency of 0.34% of total CD8⁺ T cells at day 41 after DLI (AML patient MBF, Fig. 4B). For the previously described ERAP1-1R and GEMIN4-1V MiHA, MHC multimer–positive T cells were detected in 1 of the 2 and 1 of the 4 screened patients [both multiple myeloma (MM) patient CUB at day 86], respectively (Fig. 4A and B). Detected frequencies of circulating MHC multimer–positive T cells ranged between 0.04% and 0.34% of total CD8⁺ T cells. For the other 17 MiHA candidates with validated HLA-binding affinity and SNP occurrence, including the newly identified MiHA TEP1-1S, no MHC multimer–positive T cells were detected in the peripheral blood of 1 to 5 screened patients (Fig. 4A and B and Supplementary Table SIII).

To investigate whether the MHC multimer–enriched T cell clones exerted comparable peptide-specific avidity as the in vivo generated patient-derived MiHA-specific T cells, we generated CLYBL-1Y–specific T-cell clones by single-cell sorting of CLYBL-1Y MHC multimer–positive T cells from patient MBF. After expansion, the T-cell clones were stained with the MHC multimer and TCR-Vβ monoclonal antibodies. The T-cell clone K264, which was generated by MHC multimer enrichment from donor ABM, demonstrated similar MHC multimer staining intensity as the patient-derived K2-339 clone (Fig. 5A), but a difference in TCR Vβ usage; clones derived from donor ABM expressed TCR Vβ22 and patient MBF–derived clones were TCR-Vβ1–positive. Clones K264 and K2-339 were stimulated with T2 cells loaded with titrated concentrations of either the specific or the allelic variant peptide and IFNγ production was measured (Fig. 5B). Both T-cell clones demonstrated high CLYBL-1Y–specific peptide reactivity, with IC₅₀ varying between 1 and 4 pmol/L, whereas the allelic CLYBL-1D variant was not or hardly recognized by both T-cell clones, demonstrating that T-cell clones derived from an unprimed setting can be equally potent as T cells derived from an in vivo primed setting.

To investigate the expression pattern of the CLYBL and TEP1 genes, we performed a microarray gene expression array using a panel of primary and cultured malignant (and nonmalignant) hematopoietic and non-hematopoietic cells (Supplementary Fig. S2A and S2B). The data showed that the CLYBL gene is broadly expressed in hematopoietic and non-hematopoietic cells. Expression of the TEP1 gene was not significantly measured in the majority of the samples. To investigate whether the CLYBL-1Y- and TEP1-1S–specific T-cell clones were able to recognize hematopoietic malignant cells, they were stimulated with primary chronic myelogenous leukemia (CML), AML, and acute lymphoblastic leukemia (ALL) cells derived from different MiHA-positive and -negative patients who were positive for the restricting HLA molecule. As a control, T-cell clones were also tested for recognition of EBV-LCL generated from the same individuals. Both the high-avidity T-cell clones CLYBL K264 and TEP1 K091 demonstrated MiHA-specific recognition of primary hematopoietic malignant cells (Fig. 5C and E). No reactivity was observed against MiHA-negative target cells. These data indicate that the tested MiHA can be presented in the context of HLA at the surface of leukemic cells and may therefore serve as direct targets of CD8⁺ T cells involved in a GVL response. In addition, the potential of the MiHA to serve as target in GVHD was estimated and the MiHA-specific T-cell clones were tested for recognition of non-hematopoietic fibroblasts. To mimic the proinflammatory cytokine milieu, early after transplant or during potent GVHD responses, fibroblasts were pretreated with IFNγ. Although both the CLYBL K264 and TEP1 K091 T-cell clone poorly recognized nontreated fibroblasts, they clearly recognized those that were IFNγ pretreated (Fig. 5D and F). The in vivo generated CLYBL K2-339 demonstrated a similar recognition pattern of non-hematopoietic cells (data not shown). As patient MBF demonstrated both a clinical response and GVHD after DLI, we speculate that the CLYBL-1Y specific T cells may have recognized both malignant and normal non-hematopoietic cells. These data indicate that both CLYBL-1Y and TEP1-1S may be considered as MiHA with potential therapeutic value under noninflammatory conditions, but they may participate in toxic GVHD responses in a proinflammatory environment. A feature that renders these MiHA not preferable for immunotherapy but set them as potential candidates for new GVHD biomarkers.

In this study, we demonstrate the immunogenicity of 2 predefined MiHA candidates that were predicted using a reverse immunology approach. The biologic relevance of at least one of them, the CLYBL-1Y MiHA, was demonstrated by the detection of a substantial MHC multimer–positive T-cell population specific for this MiHA in a patient suffering from AML who experienced an antileukemic response after treatment with allo-SCT and subsequent DLI.. These in vivo primed T cells demonstrated to be high-avidity T cells specific for CLYBL-1Y. For the TEP1-1S MiHA, only 2 patients could be screened for the presence of TEP1-1S–specific MHC multimer–positive cells. Therefore, absence of an in vivo induced immune response against TEP1-1S could either be due to the low number of patients that could be screened or due to subdominance of TEP1-1S in immune responses. Our reverse immunology approach has the advantage to allow identification of subdominant MiHA, as the T-cell repertoire of patients that are screened in forward immunology approaches is skewed toward highly immunodominant MiHA-specific responses. Subdominant MiHA, however, may be of therapeutic interest as they can be exploited in potential peptide vaccination or adoptive T-cell therapies when they demonstrate promising gene expression patterns.

A major limitation for the identification of large numbers of MiHA using a reverse immunology approach may be the low frequency of high-avidity MiHA-specific T cells within an individual's T-cell repertoire. By performing 2 rounds of MHC multimer enrichments each followed by a 10-day expansion step, we increased the number of MHC multimer–positive T cells that was isolated, as we observed that after the second enrichment round previously undetected MHC multimer–positive T-cell populations were found. In addition, by measuring GM-CSF in addition to IFNγ as readout for T-cell reactivity, we were able to increase the number of MHC multimer–positive T-cell clones that could be screened for functional activity. However, the failure to isolate high-avidity T cells for the previously identified LB-NISCH-1A MiHA indicates that isolation of high-avidity T cells may still be a matter of chance (13). Alternatively, T cells may be tolerant for some MiHA due to molecular mimicry with non-polymorphic epitopes or due to their failure to discriminate between both allelic variants of an SNP encoding MiHA (26), explaining why not all MiHA candidates will be of clinical relevance.

For 8 of the 20 MiHA candidates, MHC multimer–positive T-cell clones were isolated that demonstrated MiHA-specific peptide reactivity. The different MHC multimer–positive T-cell clones however demonstrated functional heterogeneity. We have recently demonstrated that CMV-specific MHC multimer–positive T cells isolated from CMV-negative individuals by MHC multimer enrichments also demonstrated a large variation in functional avidity (27). This heterogeneity in functional activity of the MHC multimer–positive T-cell populations may be specific for T cells isolated from the naïve repertoire as memory T cells are skewed toward a high-avidity range as a result of antigen encounter in vivo. Nonresponsiveness of MHC multimer–positive T-cell clones has also been reported by others for T cells derived from the naïve repertoire (28, 29). The discrepancy between MHC multimer reactivity and T-cell functionality can most likely be explained by the staining with multimerized MHC–peptide complexes. Because MHC monomers do not stably bind to TCRs, multimerization is necessary to identify antigen-specific T cells. However, multimerization of MHC–peptide complexes alters the TCR–MHC–peptide dissociation on- and off-rate kinetics and will result in increased binding avidity of the multimerized MHC–peptide complex to surface TCR. By measuring the strength of TCR binding to monomeric peptide–MHC complexes using the Streptamer K_off rate assay, we recently demonstrated that the dissociation kinetics correlated with the observed functional avidity of different MHC multimer–positive T-cell clones and that T cells demonstrating lower dissociation rates confer significantly better antigen-specific reactivity than those with fast dissociation rates (27, 30). The integration of such or alternative methodologies which allow the controlled disassembly of MHC multimers directly after T-cell isolation and subsequent FACS of T cells demonstrating a low TCR–MHC–peptide dissociation rate may further increase the efficiency of future approaches to identify high-avidity T-cell clones with a predefined specificity. In conclusion, our data illustrate that with the reverse immunology approach presented in this study, biologically relevant MiHA can be identified as well as MiHA that are not frequently induced in vivo but can potentially be used for immunotherapeutic strategies.

No potential conflicts of interest were disclosed.

Conception and design: P. Hombrink, C. Hassan, M.G.D. Kester, J.H. Frederik Falkenburg, P.A. van Veelen, M.H.M. Heemskerk

Development of methodology: P. Hombrink, M.G.D. Kester, P.A. van Veelen, M.H.M. Heemskerk

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Hombrink, C. Hassan, M.G.D. Kester, L. Jahn, M.J. Pont, A.H. de Ru, M. Griffioen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Hombrink, C. Hassan, M.G.D. Kester, P.A. van Veelen, M.H.M. Heemskerk

Writing, review, and/or revision of the manuscript: P. Hombrink, C. Hassan, M.G.D. Kester, J.H. Frederik Falkenburg, P.A. van Veelen, M.H.M. Heemskerk

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Hombrink, M.G.D. Kester, C.A.M. van Bergen

Study supervision: P. Hombrink, J.H. Frederik Falkenburg, P.A. van Veelen, M.H.M. Heemskerk

This project was supported by the Dutch Cancer Society grant no. 07-3825 and the Landsteiner Foundation for Blood Transfusion Research grant no. LSBR0713.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.