GVHD Prediction Based On HLA Three-dimensional Structure Modeling

Allogeneic haematopoietic stem cell transplantation (allo-HSCT) is a safe and effective method to cure a range of acute leukemia, and has a significant effectiveness for solid tumors, inherited metabolic diseases, and autoimmune diseases, et al. Allo-HSCT has widely applied prospect in clinical practice., while the curative effect of it is various for different individuals because of many factors. The most important factor is the severe acute graft-versus-host disease (GVHD) induced by major histocompatibility complex (MHC, HLA in human) mismatching of donor-recipient. The guidelines for unrelated adult donor hematopoietic cell transplants, based on retrospective analysis of large-sample clinical data, revised by the National Marrow Donor Program (NMDP) 2008 recommend that when possible, patients and unrelated donors should be matched at high resolution for HLA-A, -B, -C, and -DRB1 locus, and the eight alleles should be defined using DNA-based testing. When no perfect match donor was available, NMDP accept one allele mismatch at HLA-A, -B, -Cw, or -DRB1 locus as well. But there are no explicit answers to date about what specifical kind of HLA mismatch should be accepted, or what kind of HLA mismatch do not induce severe acute GVHD. It has always been the concerned issue in both clinical medicine and basic research fields how to assess the matching degree of any two HLA molecules, with HLA mismatch allo-HSCT becoming popular.When we couldn't predict the compatibility between the transplant recipient and the HLA mismatched donor based on the results of HLA gene typing, we had to resort to other evaluation methods, such as HLA functional matching, sequence matching, and structural matching, et al. The traditional mixed lymphocyte culture (MLC) assay and panel-reactive T cells (PRT) assay are individual assessment of the compatibility between HSCT recipients and the potential donors pre-transplant. And the cytology experiments results in vitro, on which those assessments are dependent, are variable. Therefore, neither assay is suitable for high-throughput screening of HSCT donors and for evaluating the compatibility of all HLA molecules. There's a rating system of HLA similarity based on differences of sequence, which is still not supported by retrospective clinical analysis very well. Recently there's another hypothesis to evaluate the compatibility of HLA molecules, which is awaited to predict the impact of HLA mismatch on HSCT outcomes based on HLA-Pep-TCR binding energy. Regretfully, lacking of clinical supports has aroused fierce opposition.According to the basic concept that the protein structure determines its function, a novel strategy to optimize allo-HSCT donor selecting based on differences of HLA molecules three-dimensional structure was proposed. We analyzed the functional compatibility of allogeneic HLA molecules through the differences of structure.Because of the extremely polymorphism, linkage disequilibrium, and national/regional diversity of MHC, the genetic polymorphism features of HLA in our own area and our own nation must be made clear firstly. HLA-A, -B, -Cw, -DRB1, -DQB1 and DRB3/4/5 loci were genotyped in 1014 Chinese Han population. Their gene frequencies, haplotype frequencies and linkage disequilibrium were also analyzed. The genetic distances between different populations were analyzed to evaluate their genetic relationships.Among all the detected HLA genes, A*02(0.33), B*15(0.14), Cw*03(0.25), DRB1*15(0.170), DQB1*06(0.218) are the popular gene groups distributing in Chinese Han population. A*02-B*46(0.071), A*02-Cw*01 (0.084), and B*46-Cw*01(0.095) are the predominant haplotypes in HLA I genes. Additionally, some of those haplotypes are statistically significant with strong linkage disequilibrium. In addition, DRB1*15 -DQB1*06 -DRB5 (0.137), DRB1*09 -DQB1*03 -DRB4 (0.129) are the main haplotypes in HLA II genes. Almost all HLA II gene haplotypes are in strong linkage disequilibrium, but the numbers of the varieties of HLA II gene haplotypes are less then HLA I genes. There were significant differences of HLA genetic distribution between different local Han populations, and similarly, the identical results were also found in the different ethnic populations of the world. With respect to genetic distance, distances within Han population of different areas were obviously smaller than that of different ethnic groups. All of these results reveal systemically the genetic polymorphism feature of Chinese Han population, which will be the basic reference data for the analysis of HLA structure differences and GVHD in following studies. Based on the HLA genes distribution results of Chinese Han population, we evaluated HLA molecule differences from the point of HLA structures, and discussed the relationship of HLA structure differences of donor-recipient and GVHD after allo-HSCT according to clinical data. Firstly, structures of all the 4556 molecules including HLA-A, -B, -Cw, -DRB1, -DPB1 and -DQB1 released from IMGT/HLA database up to Jan 2010 were modeled one by one with SWISS-MODEL server. And then the overall root mean square deviations (RMSD), which were used to evaluate the degree of structure differences of HLA molecules pairs in each locus, were calculated by Visual Molecular Dynamics (VMD) software aided with computer programming. The parameters of RMSD calculation were revised according to HLA molecular structure feature, and the amino acids residues located inα1/α2 domain (HLA I) orα1 domain (HLA II) interacting with neither peptide nor TCR were excluded in the calculation of RMSD-revised. A database management system integrating with HLA sequence alignment can also show the features of mutation amino acid residues was built to collect and sort out all the 1.6 million data. This system has been registered as HLA structure matching system (HLAStrucmark, v1.0), and obtained certificate for the registration of computer software in 2007.Then 16 HSCT data with HLA mismatches between related donor-recipient show that RMSD or RMSD-revised values is positive correlate to the degree of allograft rejection. That is, the more differences of HLA structure between donor-recipient are, the more severe rejection occurs after HSCT. In contrast, the more similar mismatched HLA molecules between donor-recipient are, the less rejection happens. However, retrospective case-control studies of large samples are better to draw more reliable conclusions when more samples are available.To validate that the level of HLA structure difference is correlated with the degree of GVHD with biology experiment data,we evaluated the compatibility of different HLA molecules with the cross reaction patterns of allogeneic reaction CTLs induced by different HLA-Pep complex. Firstly, a series of eukaryotic expression vectors inserted with full-length HLA-B cDNA gene were constructed, including HLA-B*1502, HLA-B*1518, HLA-B*3503 and HLA-B*4403 ,the popular gene type in Chinese Han population. The HLA I deficient cells Hmy2.CIR was transduced with HLA-B expression vectors using Amaxa Nucleofection system. Hmy2.CIR cells expressing HLA-B*1502, HLA-B*1518, HLA-B*3503 or HLA-B*4403 molecules pulsed with a nonameric consensus peptide (FLRGRAYGL) derived from Epstein-Barr virus, were used as APC cells to induce antigen specific CTLs respectively in vitro. And there were low binding affinity between the nonameric peptide and HLA-B molecules expressed on membrane of Hmy2.CIR. The FACS results indicated that percentage of CD8+ T lymphocytes subgroups up-regulated significantly after Hmy2.CIR inducing, and CD8+/CD4+ ratio converted to 10:7 compared with 4:10 before induction. Genes of T cell receptorβvariable region (TCRBV) families were scanned to observe the clonal changes of T cells before and after the induction and cultivation of T lymphocytes. The gene scan results showed that the 24 TCRBV families remained polyclonal before Hmy2.CIR inducing, while presented as oligoclonal distribution after inducing. These FACS and gene scan results suggested that HLA-B molecules pulsed with low binding affinity epitope peptide could induce clonal CTLs activation in vitro.The activated CTL cells induced in vitro exhibited various cross reactions while co-culturing with target cells expressing HLA-B*1502, HLA-B*1518, HLA-B*3503, HLA-B*4403 molecules pulsed with EBNA epitope peptide respectively. The cross-reaction patterns of different CTLs were distinctive in proliferation test and cytotoxicity test. The activated CTLs induced by Hmy2.CIR-B*4403-Pep exhibited unique HLA restriction, while CTLs induced by HLA-B*1502, B*1518 or B*3503 exhibited significant cross-reaction to each target cells. According to these data, HLA-B*4403 was un-compatibility with HLA-B*1502, B*1518, B*3503 in aspect of HLA restriction to TCR.The sequence alignment analysis show that there are many amino acid residue difference in the position of F epitope binding pocket between HLA-B*4403 and HLA-B*1502, B*1518, B*3503. The F pocket located in antigen binding groove contributes a lot to epitope anchoring and TCR recognition, and this point of view has been emphasized here. On the other hand, these results confirm the feasibility of the evaluation method which analyze HLA compatibility based on the CTLs cross-reaction patterns. The value of RMSD-revised between B*4403 and the other three HLA-B molecules were larger than the value of RMSD-revised within B*1502, B*1518, and B*3503. Data above suggested that HLA-B*4403 show somewhat unique feature comparing with the other three HLA-B molecules, and the results confirm the feasibility of the HLA compatibility evaluation method based on the HLA 3D structure analysis.In present study, we first explored and improved a novel strategy for evaluating HLA compatibility based on analysis of three-dimensional structure of HLA, which could optimize allo-HSCT donor selecting. This evaluating strategy has been affirmed by clinical data primarily. In addtion, we also created a pattern for HLA functional compatibility evaluation in vitro based on the cross-reaction scheme of CTLs, which could provid a scientific and effective platform for studying HLA compatibility in vitro. With the accumulation or collection of such research data, a complex relationship map of HLA similarity or compatibility could be built in the future, which would be helpful for the optimization allo-HSCT or organ donor selecting.