The Minor Histocompatibility Antigen HA-1: A Diallelic Gene with a Single Amino Acid Polymorphism

See allHide authors and affiliations

Science  13 Feb 1998:
Vol. 279, Issue 5353, pp. 1054-1057
DOI: 10.1126/science.279.5353.1054


The minor histocompatibility antigen (mHag) HA-1 is the only known mHag for which mismatching is correlated with the development of severe graft versus host disease (GvHD) after human leukocyte antigen–identical bone marrow transplantation. HA-1 was found to be a nonapeptide derived from an allele of the KIAA0223gene. The HA-1–negative allelic counterpart encoded byKIAA0223 had one amino acid difference from HA-1. Family analysis with HA-1 allele-specific polymerase chain reaction showed an exact correlation between this allelic polymorphism and the HA-1 phenotype. HA-1 allele typing of donor and recipient should improve donor selection and allow the determination of bone marrow transplantation recipients with high risk for HA-1–induced GvHD development.

Bone marrow transplantation (BMT) is the current treatment for hematologic malignancies, severe aplastic anemia, and immune deficiency disease. A frequent and life-threatening complication after allogenic human leukocyte antigen (HLA)–identical BMT is GvHD (1). Disparities in genes other than the major histocompatibility complex (MHC), referred to as minor histocompatibility antigens (mHags), are clearly involved in the development of GvHD. The mHags are recognized by T cells and were shown to be peptides derived from intracellular proteins presented by MHC molecules (2-4). A retrospective analysis revealed a significant association between mismatching for the mHag HA-1 and the development of severe GvHD after HLA-identical BMT (5). HA-1 is recognized by HLA-A*0201–restricted cytotoxic T cells (CTLs), is present in 69% of the HLA-A*0201–positive population, and is only expressed by cells of hematopoietic origin (6, 7). Family analysis demonstrated a Mendelian mode of inheritance for HA-1 and segregation independent from the MHC complex (8). To identify the mHag HA-1, HLA-A*0201 molecules were purified from the HA-1–expressing Epstein-Barr virus–transformed B lymphoblastoid (EBV-BL) cell line Rp (9). The HLA-A*0201–bound peptides were isolated and fractionated by multiple rounds of reversed phase high-performance liquid chromatography (HPLC) with different organic modifiers and different gradients (10).

The fractions were analyzed for their capacity to reconstitute HA-1–specific lysis of T2 cells, by an HA-1–specific CTL clone, in a51Cr-release assay (Fig. 1, A to C) (11). The two active fractions (33 and 34) of the third HPLC fractionation were analyzed by microcapillary HPLC–electrospray ionization tandem mass spectrometry (12). Because more than 100 different peptides were present in these fractions, about 40% of fraction 33 was chromatographed with an on-line microcapillary column effluent splitter (2,13) and simultaneously analyzed by tandem mass spectrometry and 51Cr-release assay (Fig. 1D). Five peptide species [at mass-to-charge (m/z) ratios of 550, 520, 513, 585, and 502] were present in fractions that contained HA-1–reconstituting activity and absent in fractions without activity. To determine which of the five candidates was the HA-1 peptide, we fractionated a second HA-1 purification from the EBV-BL cell line Blk (9) under similar conditions, except that the third dimension separation was done on a microcapillary HPLC column. Mass spectrometric analysis of the peptides in the single peak of reconstituting activity revealed that the only candidate species that was present was that of m/z 513 (14).

Figure 1

Reconstitution of HA-1 with HPLC-fractionated peptides eluted from HLA-A*0201 molecules in a 51Cr-release assay with mHag HA-1–specific T cell clone 3HA15. (A) Peptides were eluted from 9 × 1010 HA-1– and HLA-A*0201–positive Rp cells and separated by reversed phase HPLC with HFBA as an organic modifier and a gradient of 1% acetonitrile/min. (B) Fraction 24 of the first HPLC dimension that contained HA-1 activity was further fractionated by reversed phase HPLC with TFA as an organic modifier and a gradient of 0.7% acetonitrile/min. (C) HA-1–containing fraction 27 of the second gradient was further chromatographed with a third shallower gradient consisting of 0.1% acetonitrile/min. Background lysis of T2 by the CTL in the absence of any peptides was in (A) 3%, and in (B) and (C) 0%. Positive control lysis was in (A) 99%, in (B) 74%, and in (C) 66%. (D) Determination of candidate HA-1 peptides. HPLC fraction 33 from the separation in (C) was chromatographed with an on-line microcapillary column effluent splitter and analyzed by electrospray ionization mass spectrometry and a 51Cr-release assay. HA-1–reconstituting activity as percent specific release was compared with the abundance of peptide candidates measured as ion current.

The m/z 513 peptide was analyzed by collision-activated dissociation (CAD) analysis and determined to have the sequence VXHDDXXEA (X stands for either Ile or Leu, which cannot be distinguished by mass spectrometry under these conditions) (Fig. 2A) (15). Of all possible isomers containing either Ile or Leu in place of X in this sequence, only synthetic peptide VLHDDLLEA coeluted with the naturally processed peptide m/z 513 on the microcapillary HPLC column (14). In a 51Cr-release assay, synthetic peptide VLHDDLLEA was recognized by three different HA-1–specific CTL clones, derived from two unrelated individuals, with a half maximal activity at 150 to 200 pM (Fig. 2B). Thus, the mHag HA-1 is the nonapeptide VLHDDLLEA.

Figure 2

Sequencing of mHag HA-1 peptide by tandem mass spectrometry. (A) CAD mass spectrum of peptide candidate with m/z of 513. (B)Reconstitution assay with different concentrations of synthetic mHagHA-1 peptide with three HA-1–specific T cell clones, 3HA15, clone 15, and 5W38. Background lysis of T2 by CTLs in the absence of any peptide was for 3HA15, 4%; for clone 15, 10%; and for 5W38, 2%. Positive control lysis was for 3HA15, 46%; for clone 15, 47%; and for 5W38, 48%.

Database searches revealed that the HA-1 peptide VLHDDLLEA was identical at eight of nine residues with the peptide VLRDDLLEA, which is encoded by the partial cDNA sequence designatedKIAA0223 from the acute myelogenous leukemia KG-1 (16). Because HA-1 has a population frequency of 69%, we reasoned that the VLRDDLLEA peptide sequence might represent the HA-1–negative allelic counterpart present in the remaining 31% of the population. To test this hypothesis, we analyzed the cDNA sequence spanning this region of KIAA0223 in EBV-BL cell lines derived from a presumed HA-1 homozygous positive (HA-1+/+) individual, from an HA-1 homozygous negative (HA-1–/–) individual, and from the KG-1 cell line (17). Six of six cDNA sequences from the HA-1–/– individual showed 100% homology with the reported KIAA0223 sequence. In contrast, six of six sequences amplified from the HA-1+/+ individual displayed a two-nucleotide difference from the KIAA0223 sequence (CTGCA instead of TTGCG), leading to the amino acid sequence VLHDDLLEA. This sequence was identical to that of the peptide identified by mass spectrometry as the HA-1 mHag. The KG-1 cell line expressed both sequences. These results are consistent with the hypothesis that the mHag HA-1 sequence VLHDDLLEA is encoded by an allele (designated HA-1H) of the cDNA sequence KIAA0223. We designated the cDNA sequence of KIAA0223 present in the data bank as HA-1R.

Further support for this hypothesis was obtained by screening a family consisting of the parents and four children, which had been previously typed for HA-1 with HA-1–specific CTLs, for their KIAA0223sequence polymorphism (18). All HA-1–negative members of family 1 displayed the HA-1R sequence, whereas all HA-1–positive members expressed both HA-1H and HA-1R sequences. We subsequently designed HA-1 allele–specific polymerase chain reaction (PCR) primers to evaluate a second family previously typed for HA-1 (19) (Fig.3A). The screening of this family also revealed an exact correlation of the HA-1 phenotype and theKIAA0223 gene polymorphism. Together these results indicate that the KIAA0223 gene forms a diallelic system of which the HA-1H allele leads to recognition by the mHag HA-1–specific T cell clones.

Figure 3

TheKIAA0223 polymorphism exactly correlated with mHag HA-1 phenotype. (A) HA-1 allele–specific PCR reaction in a HA-1–typed family correlated exactly with the HA-1 phenotype. The sizes of the resulting PCR products were consistent with the expected sizes deduced from the cDNA sequence. (B) Transfection of the HA-1H allele of KIAA0223 leads to recognition by mHag HA-1–specific T cells. The HA-1H and the HA-1R coding sequence of KIAA0223were transfected together with HLA-A*0201 into HeLa cells. After 3 days the HA-1–specific CTL clones 5W38 and 3HA15 were added, and after the 24 hours TNF-α release was measured in the supernatant. The clone Q66.9 is specific for the influenza matrix protein residues 58 to 66. No TNF-α production was observed after transfection of the pcDNA3.1(+) vector alone.

To prove that the HA-1H allele of KIAA0223encodes the mHag HA-1, we cloned the appropriate regions of both the HA-1H and the HA-1R alleles in eukaryotic expression vectors and transiently transfected them in HA-1–negative HeLa cells in combination with HLA-A*0201 (20). The HeLa cells transfected with the HA-1H sequence, but not the HA-1R sequence, were recognized by two HA-1–specific CTL clones, as revealed by a tumor necrosis factor-α (TNF-α) release assay (Fig. 3B). The latter absence of recognition was expected, because exogenous VLRDDLLEA peptide was not recognized by HA-1–specific CTL clones 5W38 and clone 15 and was only recognized by 3HA15 at peptide concentrations 10,000 times that necessary for the VLHDDLLEA peptide (21). Thus, the mHag HA-1 is encoded by the HA-1H allele of the KIAA0223 gene.

Although polymorphism is present at the genetic level, the functional polymorphism is determined by HLA-A*0201–restricted T cell reactivity. Therefore, to determine whether the HA-1R peptide VLRDDLLEA is presented by HLA-A*0201, we eluted peptides from HLA-A*0201 molecules from a HA-1Rhomozygous EBV-BL cell line, fractionated them by reversed phase HPLC, and analyzed them by mass spectrometry. The synthetic VLRDDLLEA peptide was used as a marker. Peptide VLRDDLLEA could not be detected in the HLA-A*0201–eluted peptides (22), indicating that this peptide is not presented or is presented in very low amounts by HLA-A*0201 on the cell surface. This is most likely the result of a binding affinity of peptide VLRDDLLEA [50% inhibitory dose (IC50) of 365 nM] that was 1/12 that of peptide VLHDDLLEA (IC50 of 30 nM) for HLA-A*0201 (Fig.4) (23). In addition, one amino acid polymorphism can result in different proteasomal processing and differences in peptide presentation (24). The absence of the HA-1 R peptide in HLA-A*0201 suggests that this allele can be considered as a null allele with regard to HLA-A*0201–restricted T cell reactivity. This further suggests that only BMT from an HA-1R/R donor to an HA-1H/H or HA-1H/R recipient and not the reverse would be significantly associated with GvHD. This was indeed observed in a retrospective study in which HLA-A*0201–positive BMT pairs were typed for HA-1 (5). However, it remains possible that other peptides that contain the HA-1R allelic difference bind to other HLA alleles and are recognized by T cells.

Figure 4

Binding of HA-1H and HA-1R peptides to HLA-A*0201. HA-1H and HA-1R peptides were assayed for their ability to inhibit the binding of fluorescent peptide FLPSDCFPSV to recombinant HLA-A*0201 and β2-microglobulin in a cell-free peptide binding assay. One representative experiment is shown. The IC50 was determined on the basis of the results of four experiments and was 30 nM for VLHDDLLEA and 365 nM for VLRDDLLEA.

Only a few non–sex linked mHag-encoding genes have been identified so far. In mice, two maternally inherited mHags are encoded by two mitochondrial genes with two and four alleles (25). Recently, the H13 locus was defined as a H-2 Db–binding nonapeptide differing in one amino acid from its allele. Reciprocal T cell responses could be elicited (26). The human mHag HA-2 has only been sequenced on the peptide level (2). The identification of the gene encoding the mHag HA-1 is the first example of a human non–sex linked mHag that is derived from a polymorphic gene. The HA-1–encoding KIAA0223 gene has at least two alleles differing in two nucleotides that lead to a single amino acid difference.

Although the number of different human mHags is probably high, it is envisaged that only few immunodominant mHags can account for the risk for GvHD (27). HA-1 demonstrates an immunodominant behavior. First, CTL clones reactive to HA-1 were obtained from peripheral blood lymphocytes of three individuals, each transplanted across a multiple and probably distinct mHag barrier (6). Second, in the study mentioned earlier investigating the influence of mHags HA-1 to HA-5 mismatching on the development of GvHD, a mismatch of only HA-1 was significantly associated with GvHD in adult patients (5). The immunodominance of HA-1 was recently confirmed by others. Roosneket al. demonstrated HA-1–specific T cell clones in two out of two HA-1 mismatched HLA-identical BMT recipients (28). The direct applicability of the identification of the HA-1 alleles is the typing before BMT of HLA-matched donor-recipient combinations. This will improve bone marrow donor selection as well as prediction of development of HA-1–induced GvHD.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: ihbsecr{at}


View Abstract

Navigate This Article