赤塚 美樹

<title>Abstract</title> Allogeneic stem cell transplantation (allo-SCT) is one of the most effective therapeutic options for blood cell cancers. While its major anti-leukemic benefits are obtained from allo-immune reactions against leukemic cells, or GVL, the same kind of allo-reactions could be also directed to normal host tissues, giving rise to a severe complication, know as graft versus host disease (GvHD). In HLA-matched transplantation, the development of both reactions absolutely depends on the presence of one or more mismatched minor histocompatibility antigens (mHAgs) and could be further modified by other genetic as well as environmental factors, including for example, cytokine polymorphisms and GvHD prophylaxis. Thus, in view of better preventing GvHD and specifically targeting allo-immunity to the tumor component, it is critical to understand what mHAgs are mismatched and responsible for the development of GVHD or GVL and what genetic factors can influence the overall reactions. To address these questions, we conducted whole genome association studies by genotyping more than 500,000 SNPs in donors and recipients of 1598 unrelated transplants from Japan Marrow Donation Program (JMDP). All transplants were matched for HLA-A, B, C, DRB1 and DQB1, while 1033 (63%) transplants were mismatched for HLA-DPB1. 656 (41.7%) and 245 (14.9%) of transplants had developed grade II–IV and III–IV of acute GvHD (aGvHD), respectively. Overall call rates exceeded 98% both in donors and in recipients. Unobserved HapMap PhaseII SNPs were rigorously imputed using genotyped SNPs. After excluding those SNPs showing &lt;95% call rate, deviation from Hardy-Weinberg equilibrium, or &lt;5% minor allele frequency, 1,276,699 SNPs were tested for association with development of acute and chronic GvHD, relapse, and overall survival, by calculating LogRank statistics for each SNP according to single genotypes in donors and recipients or based on mismatch in genotypes between donor and recipient. Statistical thresholds for genome-wide-P value of 0.05 were determined empirically by doing 1,000 permutations for each analysis. In the analysis of mismatched genotypes, SNPs around the HLA-DPB1 locus uniquely showed a strong association with the development of &gt;grade II aGvHD with the maximum P-value of 1.81 × 10−9 at rs6937034, and thus, successfully captured the association of DPB1 allele mismatch as directly defined by HLA typing (HR = 1.91, P= 2.88 × 10−13). To facilitate the identification of target mHAgs for aGvHD, we performed subgroup analysis, where association tests were confined to those transplants sharing particular HLA types based on the fact that recognition of mHAgs is restricted to particular HLA contexts (HLA restriction). Six loci was identified as candidate mHAg loci whose mismatch may confer increased risk for development of aGvHD. These included rs17473423 on chr12 associated with an A*2402/B*5201/Cw*1202/DRB1*1501/DQB1*0601 allele set shared in ~40% of unrelated transplants in Japanese (grade III–IV aGvHD with maximum P=3.99 × 10−13), rs9657655 on chr9 associated with another common allele in Japanese, A*3303/B*4403/Cw*1403 (grade III–IV aGvHD with maximum P=8.56 × 10−10), and other four loci associated with DQB1*0501, Cw*0102, B*5201, and Cw*1202. Two SNPs in patients were also found to be associated with aGvHD, rs5998746 on chr22 (P=3.41 × 10−8) and rs11873016 on chr18 (P=1.26 × 10−8), although no donor SNPs showed significant associations). Similarly, we identified four candidate SNPs associated with the development of severe cGvHD or relapse. Current study provided a unique opportunity in that combination of two different genotypes, not merely genotypes of single individuals, that is associated with particular disease phenotypes, is explored by whole genome association scans. Although further replication studies and biological confirmation are required, our results suggest that whole genome association studies of allo-SCT could provide a novel clue to understanding the genetic basis of allo-SCT.

<title>Abstract</title> Minor histocompatibility antigens (mHags) are the molecular targets of allo-immunity associated with major anti-tumor activities in hematopoietic stem cell transplantation (HSCT), but are also involved in the pathogenesis of graft-versus-host disease (GVHD). They are typically defined by the host’s SNPs that are not shared by the donor and immunologically recognized by cytotoxic T-cells isolated from the post-HSCT patients. However, despite their critical importance in transplantation medicine, fewer than 20 mHags have been identified during the past 20 years due to the lack of an efficient method for their isolation. Here we developed a novel method in which the large data set from the International HapMap Project can be directly used for genetic mapping of novel mHags. Concretely, immortalized B lymphoblastoid cell lines (LCLs) from a HapMap panel are grouped into mHag positive (mHag+) and negative (mHag−) subpanels according to their susceptibility to a cytotoxic T-cells (CTL) clone as determined by conventional chromium release cytotoxicity assays (CRAs), and the target mHag locus could be directly identified by association scan (indicated by χ2 statistic) using the highly qualified HapMap data set having over 3,000,000 SNP markers. The major concern about this approach arises from the risk of overfitting observed phenotypes to one or more incidental SNPs from this large number of the HapMap SNPs. To address this problem, we first estimated the maximum sizes of the test statistics under the null hypothesis (i.e., no associated SNPs within the HapMap set) empirically by simulating 10,000 case-control HapMap panels in different experimental conditions, and compared them with the expected size of test statistic values from the marker SNPs associated with the target SNP, assuming different linkage disequilibrium (LD), or values in between. Except for those mHags having very low minor allele frequencies (MAF) below ~0.05, the possibility of overfitting is progressively reduced as the number of LCLs increases, allowing for unique identification of the target locus in a broad range of values. To demonstrate the feasibility of this method, we tried to map the locus for HA-1H mHag, by actually immunophenotyping 58 LCLs from the JPT+CHB HapMap panel with CRAs using HLA-A*0206-restricted LCL (CTL-4B1). As expected, the genome-wide scan clearly indicated a unique association within the HMHA1 gene, showing a peak χ2 statistic of 52.8 (not reached in 100,000 permutations) at rs10421359. Next, we applied this method to mapping novel mHags recognized by HLA-B*4002-restricted CTL-3B6 and HLA-A*0206-restricted CTL-1B2, both of whose target mHags had not been identified. The peak in chromosome 19q13.3 for the CTL-3B6 set showed the theoretically maximum χ2 value of 50 (not reached in 100,000 permutations) at rs3027952, which was mapped within a small LD block of ~182kb containing a single gene, SLC1A5, as a candidate mHag gene. In fact, when expressed in HEK293T with HLA-B*4002 transgene, recipient-derived, but not donor-derived, SLC1A5 cDNA was able to stimulate interferon-γ secretion from CTL-3B6, indicating that SLC1A5 encodes the target mHag recognized by CTL-3B6. Conventional epitope mapping finally identified an undecameric peptide, AEATANGGLAL, which was further confirmed by epitope reconstitution assays. The target mHag locus for CTL-1B2 was identified at the peak (max χ2 = 44, not reached in 100,000 permutations) within a 598 kb block on chromosome 4q13.1, and coincides with the locus for a previously reported mHag, UGT2B17. Our epitope mapping by using UGT2B17 cDNA deletion mutants, prediction of candidate epitopes by HLA-binding algorithms and epitope reconstitution assays successfully identified a novel nonameric peptide, CVATMIFMI. Our results demonstrate how effectively the HapMap resources could be used for genetic mapping of clinically relevant human traits. This method may be also applied to disclosing other relevant human variations, if an accurate bioassay is applied to discriminate them. We anticipate our method based on the HapMap scan greatly accelerates isolation of novel mHags, which could be used for the development of selective allo-immune therapies to intractable blood cancers, circumventing potentially life-threatening GVHD, while harnessing its anti-tumor effects. Such knowledge on mHags should also promote our understanding of allo-immunity.

<title>Abstract</title> Differential graft versus host response after bone marrow transplantation (BMT) is key to leukemia patient survival. The development of specific graft versus leukemia response targeting tissue-specific minor histocompatibility antigens (mHAgs) is a novel therapeutic approach to treating hematopoietic malignancies. A pilot intensive immune cell therapy was applied to an acute myeloid leukemia (AML) patient who relapsed around day 43 after receiving unrelated allogeneic BMT with three allele disparities for HLA-A, C and DQB1. The patient received more than thirty infusions of dendritic cell-co-cultured donor-derived immune cells including cytotoxic T lymphocytes (CTL) generated against AML lysate-pulsed dendritic cells (DC), AML-DC fusion, and DC pulsed with HLA-A*2402-restricted ACC1Y mHAg epitope. The latter was expressed by recipient but not by donor hematopoietic cells. ACC1Y-peptide-major histocompatibility complex (MHC) multimer staining detected CTL in the patients peripheral blood at different time points during treatment. Four months after the first mHAg-CTL infusion, CTL-resistant AML cells were detected in relapsed pleural effusion and peripheral blood. Further analysis of the relapsed AML cells revealed complete loss of surface class I HLA-A*2402 and class II HLA molecules. Although the AML-specific immune cells were effective in killing the original AML in CTL assays in vitro, the late relapsed AML became resistant to the early AML-primed CTL. This result suggests that rapidly evolving AML cells may escape the intensive anti-leukemia allo-immune pressure including the mHAg-specific CTL through selective loss of A*2402 expression - the ACC1Y-restricted HLA locus. As the patient suffered ongoing multi-focal extramedullary relapse in muscles, sub-mucosal tissue, intestine, spinal cord, and brain, rare cancer cells resistant to mHAg-specific CTL could evolve quickly. Our study indicates that immune escape through diminished HLA expression, which resulted in lack of cancer cell killing by AML-specific immune cells, including the mHAg-restricted CTL, had developed during the therapy. Therefore, leukemia patients with high load of AML or multi-focal extramedullary relapse may not endure prolonged benefit from a single arm immune cell therapy.

<title>Abstract</title> Minor histocompatibility antigens (mHAgs) with expression limited to hematopoietic cells represent attractive targets for immunotherapy to induce selective graft-versus-leukemia (GVL) reactions. Here we report the identification of a novel mHAg which is recognized by an HLA-B*4403-restricted CTL clone. Microsatellite allele image analysis of two DNA pools generated from CTL-defined mHAg positive and mHAg negative groups was performed using microsatellite markers set at 100 kb intervals within the segment initially mapped by two-point genetic linkage analysis and detailed mapping of the chromosomal recombinant points. This approach defined a 0.53 Mbp region of chromosome 18q21–22 containing 12 candidate genes potentially encoding the mHAg, although the target gene could not be identified. Subsequently, cDNA expression cloning studies demonstrated that the CTL epitope of interest was encoded by a novel allelic splice variant of XM_209104, hereafter designated as XM_209104-av. Indeed, this gene was found to lie within the region predicted by microsatellite allele image analysis. The immunogenicity of the epitope was generated by differential protein expression due to alternative splicing, which was completely controlled by one intronic single nucleotide polymorphism (SNP) located in the consensus 5′ splice site adjacent to an exon. To our knowledge, this is the first example of a mHAg controlled by a SNP located in a region other than coding sequences. Because the CTL lysed also HLA-B*4402 positive, mHAg positive B-LCLs, this novel epitope peptide can bind to not only HLA-B*4403 but also HLA-B*4402 which is a relatively common HLA-B allele in Caucasian populations. Finally, the finding that the novel XM_209104-av showed low or no expression in normal tissues including resting hematopoietic cells, but significantly higher expression in primary acute leukemia cells, especially those of myeloid lineage, suggest that this novel epitope may be an attractive therapeutic target for immunotherapy not only as a minor H antigen but also as a leukemia-associated antigen.

<title>Abstract</title> We recently identified a human minor histocompatibility (H) antigen encoded by the UDP glycosyltransferase 2 family, polypeptide B17 (UGT2B17) gene. UGT2B17 is a metabolic enzyme that is highly expressed in liver and colon, and at low levels in hematopoietic cells. UGT2B17 conjugates exogenous and endogenous compounds including steroid hormones and facilitates their elimination. The immunogenicity of the UGT2B17 was due to differential expression of the protein in donor and recipient cells as a consequence of a homozygous deletion of the UGT2B17 gene in the recipient. Deletion of this gene in a subset of normal individuals suggested two mechanisms by which UGT2B17 could affect the outcome of allogeneic BMT. First, the expression of UGT2B17 in recipient but not donor cells could provide minor H antigens and increase GVHD. Second, the presence of UGT2B17 in the donor or recipient could affect the metabolism, function, or toxicity related to endogenous hormones or drugs administered with BMT. The purposes of this study were to determine the frequency of homozygous UGT2B17 deletion in healthy individuals and patients with hematological disease and whether a deficiency of UGT2B17 in donor and/or recipient affected the outcome of allogeneic BMT. A total of 435 recipients were selected from the patients who received unrelated BMT through the Japanese Marrow Donor Program between January 1993 and March 2000. Selection criteria were (1) matched at HLA-A, B, C and DRB1 genotypes, (2) unmanipulated marrow graft, (3) cyclosporine A or tacrolimus as GVHD prophylaxis, and (4) available DNA was stored. DNA from a total of 358 of the donors was also available. Homozygous deletion of the UGT2B17 gene determined by SSP-PCR was found in 320 (84.8%) of 377 donors and 358 (82.3%) of 435 patients (P=0.32, χ2 test). These frequencies of UGT2B17 deletion were substantially higher than the 11% observed in the Caucasian population in a previous study. A multivariate analysis (Cox proportional hazard model) showed no significant association between UGT2B17-mismatch in the GVHD direction and the incidence of acute GVHD (grade II-IV or III-IV), chronic GVHD, relapse, or survival. However, a deletion of UGT2B17 in the donor was a favorable factor for transplant-related mortality (TRM) (relative risk, 0.530; 95% CI, 0.334 to 0.841; P=0.007), survival (0.567; 0.387 to 0.831; P=0.036) and disease-free survival (DFS) (0.647; 0.443 to 0.945; P=0.024). We also analyzed these outcomes in the subset of patients with standard-risk malignant disease by the Kaplan-Meier method. TRM, survival, and DFS at eight years after transplantation were 23.7% for patients transplanted from a UGT2B17-deleted donor vs 52.9% for patients with a UGT2B17-positive donor (P=0.001); 68.3% vs 37.5% (P=0.0009); and 66.9% vs 37.5% (P=0.0009), respectively. These data suggest that the UGT2B17 enzyme in donor-derived blood cells may affect the metabolism of exogenously administered or endogenous molecules and adversely influence the outcome of unrelated BMT. Analysis in another patient population will be important to confirm our results.