Major Histocompatibility Complex (MHC) is a genomic region found in jawed vertebrates and harbors various immunologically important genes. All jawed vertebrates from sharks to mammals have the evolutionarily conserved basic gene organization of the MHC. Teleosts are exceptional in that their orthologs of the mammalian MHC-encoded genes are dispersed on several chromosomes. Even in teleost, however, the MHC class IA genes and several other genes (transporter associated with antigen presentation 2 (TAP2), proteasome subunit beta type (PSMB8, PSMB9, PSMB10), TAP binding protein (TAPBP)) encoding for the molecules directly involved in class I antigen presentation are linked, defining the teleost MHC class I region. The uniquely derived genomic organization of the teleost MHC among the jawed vertebrates makes it a curious target for the structural analysis to understand the evolution of the jawed vertebrate MHC.

Oryzias species commonly known as medaka (ricefish), is a small egg-laying organism distributed throughout the east Asia and more than 20 Oryzias and closely related Xenopoechilus species have been described thus far which are classified into three species groups, the latipes, javanicus and celebensis, based on the karyotypes, allozymes, and mitochondrial, and nuclear DNA sequences. Previous comparative analysis of the MHC class I region at the nucleotide sequence level between two inbred strains of Oryzias latipes (latipes group), Hd-rR and HNI, revealed that an approximately 100 kb block harboring the PSMB10, PSMB8 and two class IA (UAA, UBA) genes were so diverged between these two strains that almost no sequence similarity was noticed in the intronic and intergenic regions. To understand the evolution of the MHC class I region in genus Oryzias at the nucleotide sequence level, I determined the complete nucleotide sequences of the BAC clones covering the MHC class I region of O. dancena (javanicus group) and O. luzonensis (latipes group).

Sequence information of the MHC class I region were determined using the BAC clones isolated from the three genomic libraries designated as IMBX and IMBY for O. dancena and LMB for O. luzonensis. Screening of these libraries to identify the BAC clones containing the MHC class I genes were carried out using the primer sets specific to the PSMB9, PSMB8, class IA (UAA) and TAPBP genes. The sequencing of the identified BAC clones were carried out using the shotgun approach. In brief, the purified BAC DNA was partially digested with Sau3AI and fragments ranging from 3-5 kb and 6-8 kb were gel purified, cloned into the pGEM3 zf(+) vector and transformed into the DH5α cells. About 1600-1800 sequence reads were collected for each BAC clone, and were assembled into contigs and scaffolds using Phred/Phrap/Consed software. The remaining gaps between the contigs were filled by primer-walking using the BAC DNA as the template. Identification of the genes from the assembled region was carried out by homology search (BLASTn, TBLASTX) and further supported by gene prediction softwares (Genscan, Fgenesh). RT-PCR analysis was performed to confirm the predicted intron-exon boundaries. In addition, RT-PCR analysis was also performed to detect polymorphism of the class IA genes.

Twelve and eight clones were identified from the O.dancena IMBY and IMBX libraries, respectively. Based on preliminary mapping of these BAC clones, IMBY 58G24, IMBY 68M2 and IMBX 79J15 were chosen for sequence analysis. Insert size of IMBY 58G24 and IMBY 68M2 were 157,124 bp and 201,149 bp, respectively, and these sequences showed 100% identity in the 17,510 bp of overlapping region. The combined 340,763 bp sequence covered the region from exon 8 of the RXRB gene to further downstream of the TCF19 gene. The IMBX 79J15 BAC clone had 138,052 bp insert and covered the region from exon 9 of the RING3 gene to the intergenic region between the MHC class I (O. latipes UBA) and TAPBP genes. Approximately 4 kb sequence up to exon 6 of the TAPBP gene was determined by primer walking using O. dancena genomic DNA as template. The 141,664 bp sequence was composed from these sequences.

The sequence analysis of the MHC class I region of O. luzonensis was carried out using the LMB 52F15 and LMB 66F20 BAC clones. These two clones showed 100% nucleotide sequence identity at the approximately 15 kb of overlapping region. 193,474 bp continuous sequence was composed of the entire sequence of LMB 66F20 and the extra region of LMB 52F15, and covered the region from exon 2 of the UDA gene to exon 3 of the TAPBP gene.

The previous study in O. latipes reported the presence of the highly diverged dichotomous haplotypic lineages of the PSMB8 and PSMB10 genes termed as d- and N-, originally reported in Hd-rR and HNI, respectively. Based on the typing of the PSMB8 gene, the IMBY sequence was judged to represent the N- haplotypic lineage of O. dancena, whereas the IMBX sequence represented the d- lineage of O. dancena. Similarly, the LMB sequence represented the d- lineage of O. luzonensis. The order and orientation of the predicted genes in the assembled sequences of d- and N- haplotypes of O. dancena, and dhaplotype of O. luzonensis were identical to those of d- and N- haplotypes of O. latipes, except for the number of the MHC class IA genes which showed species-specific variation. Whereas one UAA and one UBA genes were present in the HNI and Hd-rR strains (O. latipes), IMBX sequence contained four copies of the UAA gene Orda-UAA1*0201, Orda-UAA2*0201N, Orda-UAA3*0201, and Orda-UAA4*0201. The IMBY sequence also contained four copies of the UAA gene, Orda-UAA1*0101S, Orda-UAA2*0101N, Orda-UAA3*0101N and Orda-UAA4*0101. No UBA like class I gene was detected in O. dancena. The LMB sequence contained three UAA genes i.e. Orlu-UAA1*0101, Orlu-UAA2*0101, and Orlu-UAA3*0101, and a single UBA gene, Orlu-UBA*0101N.

Dotplot comparison among these haplotypes detected the diagonal line on both sides of the central block showing no clear diagonal line, indicating that the most of the MHC class I region of Oryzias shows usual evolutionary pattern. The central variable block extended from PSMB10 to the beginning of the TAPBP and included the genes directly involved in the MHC class I antigen presentation, PSMB10, PSMB8 and class IA genes. The class IA genes in the variable block showed species-specific variation not only in copy number but also in its α3 domain sequence. On the other hand, phylogenetic analysis of the deduced amino acid sequence of the mature PSMB10, and PSMB8 showed d- and Nlineage specific clustering, indicating that these genes shows trans-species dimorphism between O. latipes and O. dancena.

In the current study, I explored the evolution of Oryzias's MHC class I region using the sequence information from the d- and N- lineages of O. latipes, d- lineage of O. luzonensis and d- and N- lineages of O. dancena. This study indicated that the Oryzias MHC class I region can be divided into conserved block and variable block. The variable block can be further subdivided into two sub-blocks the class I, and the PSMB10-PSMB8 The class I sub-block showed the most striking inter-species variation. Thus, copy number and type of the class IA genes varied from species to species; 1 UAA and 1 UBA for O. latipes, 3 UAA and 1 UBA for O. luzonensis, and 4 UAA and no UBA for O. dancena. The simplest explanation for this situation may be that the last common ancestor of these three species had only one class IA gene that duplicated or multiplied independently in each lineage. However, analysis of other Oryzias species is required to clarify the number and type of original class IA genes. The "class I" sub-block started from little before the 3` end of the first class I gene proximal to the PSMB8 gene side and extended to the intergenic region of the last class I gene and the TAPBP gene. The "PSMB10-PSMB8" sub-block spanned from the 6th intron of the PSMB10 gene to the 3` end of the PSMB8 gene, and was previously found to show dichotomous haplotypic lineage in O. latipes. Also, these dichotomous haplotypic lineages showed trans-species evolution in the latipes and celebensis species groups. Here I showed that these dichotomous haplotypic lineages are also found in O. dancena, belonging to the third Oryzias species group, the javanicus group. Therefore it is likely that these dichotomous haplotypic lineages was present in the common ancestor of all Oryzias species, and were transmitted from an ancestral species to a descendent species, providing a typical example of balancing selection.

Thus, the MHC class I regions of Oryzias species provide a unique opportunity to perform phylogenetic analysis of the three tightly linked sub-blocks evolving independently. Especially, the co-existence of the trans-species dimorphic sub-block and the species-specific variable sub-block is of interest. These gene are involved in the MHC class I antigen presentation and suggested that the two different kinds of selective pressure are working on these two sub-blocks. Although, the merit of the close linkage between these two sub-blocks could be to keep co-segregation of the functionary compatible PSMB and class IA alleles, independent evolution of the PSMB and class IA genes makes this explanation unlikely. Another explanation is that the close linkage helps rapid fixation of advantageous class IA new alleles. Demonstration of the actual difference in cleaving specificity of PSMB8 of two lineages as well as elucidation of peptide binding spectrum of class IA alleles of wild population are required to clarify the evolutionary merit of this linkage.

主要組織適合性抗原複合体(Major Hitocompatibility Complex, MHC)は有顎脊椎動物に固有のゲノム領域であり、Tリンパ球に抗原を提示するMHCクラスI, II 分子の遺伝子を始めとして、免疫反応に重要な役割を果たす多くの遺伝子が集積している。このようなゲノム領域がいかにして形成されたかは興味深い問題であり、これまでの比較免疫学的な研究から、MHCの基本的な構成は多くの有顎脊椎動物において保たれているものの、それぞれの分類群毎に派生的な特徴を示すことが明らかになっている。特にこれまでにMHC構造の種内多型、近縁種間での進化についての解析の主な対象とされてきた哺乳類は、クラスI遺伝子とその抗原提示に関わるプロテアソームサブユニット遺伝子であるPSMB8との間の緊密な連鎖が失われた特殊なMHC構造を示し、哺乳類で得られた知見がどこまで他の脊椎動物に普遍化出来るかは疑問視されるに至っている。本論文は、クラスI、 PSMB8両遺伝子の緊密な連鎖が保たれている硬骨魚類のメダカ近縁種のMHCクラスI領域の構造解析を行い、先行研究のあるメダカのものと比較することにより、MHCの基本的な進化過程を明らかにしたものである。学位論文の主要部分は、Introduction, Materials and methods, Results, Discussionよりなり、得られた成果とその意義が述べられている。

東アジアに約20種存在するメダカ属各種は、核型、ミトコンドリアのDNA配列等により、3つの種群、メダカ種群、ジャワメダカ種群、セレベスメダカ種群に分けられている。本研究ではBACライブリーが利用可能であった、メダカ種群に属するルソンメダカと、ジャワメダカ種群に属するインドメダカが解析された。ルソンメダカは2つの一部重複するBACクローンから、193,474bpの連続配列が得られ、メダカにおいてd型、N型と呼ばれる著しく配列の異なる二型性を示すPSMB8遺伝子は明らかにd型と分類された。インドメダカは2つのBACライブラリーを使い、一方のライブラリーからはd型のPSMB8遺伝子を含む一つのBAC配列、141,664bpが決定された。もう一方のライブラリーからは2つのBACクローンから340,769bpの連続配列が決定されたが、PSMB8遺伝子はN型であった。これまでインドメダカの属するジャワメダカ種群からはN型のPSMB8遺伝子は確認されておらず、この結果により初めてPSMB8遺伝子の二型性は数千万年前とされるメダカ種群、ジャワメダカ種群の分岐以前から存在し、trans-speciesに伝えられてきたことが明らかになった。本研究により決定された3つのMHCクラスI領域の配列と、これまでメダカで明らかにされていた2つの配列を互いに比較した結果、PSMB10, PSMB8及びいくつかのclass IA遺伝子を含む約100 kbの領域は配列間で大きな違いを示しMHC領域特有の進化をしているが、それ以外の領域は比較的高い保存性を保ち、通常の非MHC領域と同様の進化をしていることが示唆された。また、MHC特有の進化を示す領域のうち、PSMB10, PSMB8遺伝子は種を超えた二型性を示すのに対して、class IA遺伝子はそれぞれの種に固有なコピー数を示した。メダカにはUAAとUBAと呼ばれる2種類のclass IA遺伝子が存在するが、ルソンメダカにはUAAが3コピー、UBAが1コピー存在し、またインドメダカにはUBAは無く、UAAが4コピー存在した。class IA遺伝子の系統樹解析では、同種の遺伝子がクラスターを形成し、それぞれの種内でhomogenizationが行われていることが示唆された。以上の結果はメダカ属のMHCクラスI領域は3つの亜領域に分かれ、それぞれが異なる淘汰圧のもとに互いに独立に進化していることを示している。

これらの研究成果は、MHC の基本構造を保持する動物において初めて詳細にその進化過程を明らかにしたものとして、脊椎動物全体を通してのMHCの進化の理解に重要な寄与をするものである。

なお、本論文は、野中真弓・野中 勝との共同研究であるが、論文提出者が主体となって分析および検証を行ったもので、論文提出者の寄与が十分であると判断する。

したがって、博士(理学)の学位を授与できると認める。