It is universally accepted that chloroplasts (cps) evolved via endosymbiosis of a cyanobacteria. After endosymbiosis, massive gene losses that were redundant between the protochloroplast derived from the cyanobacterial endosymbiont and primitive eukaryote cell had occurred, and transfer of essential genes from the ancestor to the nucleus decisively shaped the present plant nuclear and cp genome during evolution. Alternatively, nuclear genes with different organellar origins were replaced with cp-encoded genes, resulting in the present structures of the cp genome. Although such gene transfers and substitutions are important genetic events in evolution, little is known about this process.

It is considered that mitochondria also evolved via endosymbiosis of a α-proteobacteria in eukaryote. More genes (especially ribosomal-protein-encoding genes) are encoded in the mitochondrial genome of angiosperms than in those of vertebrates and fungi. Furthermore, the number of genes encoded in the mitochondrial genome varies among plant species. These clues suggest that gene transfer is still ongoing in angiosperms. Thus, the mitochondrial genome of angiosperms is a good tool for the study of gene transfer events from the mitochondria to the nucleus and provides a way of understanding the mechanism of gene transfer in eukaryote. Compared with angiosperm mitochondrial genomes, the genome structure and gene content of the cp genome are highly conserved in evolutionary distant lineages. Hence, little is known regarding the events involved in the transfer of genes from the cp to the nucleus. However, several cases of gene loss from the cp genome have been found, because of the increasing number of completely sequenced flowering plant cp genomes. This will allow us to analyze the process of gene transfer from the cp genome to the nucleus.

The present thesis is aimed at understanding the mechanisms underlying evolutionary gene substitution and transfer via the comparative analysis of two genes that encode the cp ribosomal proteins S16 (RPS16) and L32 (RPL32).

I. rps16 in the cp genome displays a variable status during evolution among Arabidopsis and its closely related species

rps16 is generally encoded by two exons (separated by one group II intron) that are located in the cp genomes of flowering plants. However, it has been reported that, in mono- and dicotyledonous plants, the cp-encoded RPS16 protein was replaced by the product of the nuclear-encoded rps16, which was transferred from the mitochondria to the nucleus before the early divergence of angiosperms. It is suggested that the present status of rps16 gene substitution in most angiosperm cp genomes is the intermediate stage.

In this study, I have identified the different functional statuses of rps16 in several cp genomes in the genus Arabidopsis and its close relatives. Eleven complete Brassicaceae cp genomes (Aethionema grandiflorum, Arabis hirsuta, Barbarea verna, Brassica rapa subsp pekinensis, Capsella bursa-pastoris, Crucihimalaya wallichii, Draba nemorosa, Lepidium virginicum, Lobularia maritima, Nasturtium officinale, and Olimarabidopsis pumila) and the partial chloroplast genome, including rps16, of Sinapis alba are available from current databases. Sequence comparison revealed that the cp-encoded rps16 genes of four Brassicaceae species have become pseudogenes, as these genes contain a deletion within their coding sequence in Arabis hirsuta, a nonsense mutation in Aethionema grandiflorum, and the complete loss of the second exon in D. nemorosa and L. maritima.

The fact that Arabis is phylogenetically close to Arabidopsis thaliana raised the possibility that the pseudogenization of cp-encoded rps16 observed in Arabis hirsuta might also occur in the Arabidopsis lineage. Further analysis of Arabidopsis thaliana and its close relatives (Arabidopsis arenosa, Arabidopsis lyrata, O. pumila, and Crucihimalaya lasiocarpa) has shown that pseudogenization occurred via the loss of the splicing capacity of the group II intron. The 5' splice site of the group II intron changed from GUGYG to GUACG in Arabidopsis thaliana. However, the splice site consensus sequences were conserved in other closely related species. RT-PCR was conducted to confirm the splicing activity of the group II intron among Arabidopsis thaliana and its close relatives. The results revealed that only the primary transcript was amplified in Arabidopsis thaliana, Arabis hirsuta, and O. pumila, suggesting the loss of the splicing of the intron in these plants. This raised the possibility of the widespread pseudogenization of rps16 in the angiosperm cp genomes via the loss of its splicing capacity, even when the rps16 encoded in the cp genome is transcriptionally active.

The estimated time of the divergence of Arabidopsis thaliana from all other Arabidopsis species is 3.0-5.8 million years ago (mya), and the time of the divergence of the Arabidopsis and Olimarabidopsis (Crucihimalaya) species is estimated at 10-14 mya. This suggests that the independent pseudogenization of the cp-encoded rps16 in Arabidopsis thaliana and O. pumila via dysfunctional splicing occurred within the last 5.8 and 14 myr, respectively. The onset of cp-encoded rps16 gene substitution was minimally estimated in previous works at 140-150 mya. Considering the time of divergence of Arabidopsis thaliana from O. pumila, the nuclear genome gained an rps16 copy~140 mya and the cp and nuclear copies have coexisted (perhaps redundantly) since then. However, in the last 5.8-14 myr, the cp-encoded rps16 copies have become recognizable pseudogenes in Arabidopsis thaliana and O. pumila. This suggests that the process of complete gene substitution of cp-encoded rps16 lasted for over 126 myr in Arabidopsis thaliana and O. pumila.

Why does the loss of rps16 from the cp genome seem to have accelerated in evolutionarily recent times in the Arabidopsis lineage? It was predicted that the level of inbreeding is positively associated with the level of functional transfer (and loss) of organellar genes. Interestingly, this study revealed that self-compatible plants tend to lose rps16 from their cp genomes, whereas self-incompatible plants tend to retain rps16 in their cp genomes. Self-compatibility may be one of the explanations for the acceleration of rps16 gene loss from the cp genome in Brassicaceae, although the underlying mechanism remains completely unknown.

II. Comparative genomic analysis of gene transfer of chloroplast rpl32 in Malpighiales demonstrates its parallel retention and progressive pseudogenization

rpl32, which was first characterized for its location between ndhF and trnL on the small single-copy region of the cp genome of Tobacco, is generally encoded by the cp genome in flowering plants. However, previous studies in Malpighiales revealed that this gene was functionally transferred to the nucleus in Bruguiera gymnorrhiza and in the genus Populus (P. alba and P. trichocarpa). In B. gymnorrhiza, cp rpl32 is encoded via alternative splicing at the seventh intron of the cp Cu-Zn superoxide dismutase gene (sod-1), encoded in the nucleus. On the other hand, the sod-1 sequence containing cp rpl32 is duplicated and subfunctionalized in the Populus genus. It has been strongly suggested that cp rpl32 has acquired the sequence that encodes the transit peptide from the sod-1 before the divergence of Malpighiales.

To confirm the status of cp- and nuclear-encoded cp rpl32, comparative genomic analysis of cp rpl32 was conducted in eight species (Passiflora citrina, Euphorbia sieboldiana, Calophyllum inophyllum, Acalypha hispida, Hypericum erectum, Viola mandshurica, Manihot esculenta, and Ochna serrulata) from six distinct families of Malpighiales.

Genomic PCR was conducted to determine whether rpl32 was lost from the cp genome of seven species (O. serrulata, C. inophyllum, H. erectum, E. sieboldiana, M. esculenta, V. mandshurica, and P. citrina). rpl32 was found in the cp genomes of C. inophyllum and M. esculenta and their expression was detected using RT-PCR, suggesting that the active rpl32 was retained in these two species. However, cp rpl32 was inactivated in the five remaining species. Three species (O. serrulata, H. erectum, and E. sieboldiana) lost rpl32 completely from their cp genomes. Pseudo-rpl32 was found in the cp genomes of V. mandshurica and P. citrina. These observations suggest that cp rpl32 may have been transferred to the nucleus in P. alba and in most Malpighiales species.

Previous works suggest that the integration of cp rpl32 into the seventh intron of sod-1 occurred in the nuclear genome of Malpighiales. Genomic PCR was conducted to determine whether cp rpl32 was integrated into the seventh intron position of sod-1 in eight species using the primer pairs designed from the conserved sod-1 and nuclear-encoded cp rpl32 sequences among Malpighiales species. Cp rpl32 was encoded in the seventh intron position of sod-1 in six species (C. inophyllum, A. hispida, E. sieboldiana, M. esculenta, V. mandshurica, and P. citrina). RT-PCR was conducted to confirm the expression of nuclear-encoded cp rpl32 in C. inophyllum, E. sieboldiana, M. esculenta, and V. mandshurica. Direct sequencing of each RT-PCR product revealed that all products were amplified from the transferred nuclear cp rpl32 integrated into the seventh intron position of sod-1. This result suggests that rpl32 gene transfer to the seventh intron of sod-1 occurred widely in Malpighiales, as predicted.

In M. esculenta, RT-PCR analysis revealed that the entire sequence of the seventh intron of the rpl32 mRNA was not spliced out and that five amino acids deduced from the RT-PCR product were deleted in the conserved RPL32 domain. A sequence identical to that of the RT-PCR product of nuclear-encoded cp rpl32 was detected in the whole-genome sequence of M. esculenta, which is available from the Joint Genome Initiative (http://www.jgi.doe.gov/CSP/), and no other cp rpl32 genes were detected in the nuclear genome. These results suggest that nuclear-encoded cp rpl32 is inactivated in M. esculenta. To the best of my knowledge, this is the first example of the inactivation of a gene transferred from the cp to the nucleus.

The dual expression of cp rpl32 encoded in the cp and nuclear genomes was observed in C. inophyllum. Hence, cp rpl32 gene transfer was identified as the intermediate stage in C. inophyllum. The order Malpighiales comprises around 700 genera and over 16,000 species in 30 families. Comparative analysis strongly suggests that the gene transfer of cp rpl32 to the seventh intron of sod-1 in the nucleus occurred in the common ancestor of Malpighiales. The estimated time of diversification of Malpighiales is around 114 mya. Therefore, the intermediate stage of cp rpl32 has been ongoing over the past 114 myr in C. inophyllum.

The study of two cp ribosomal genes, rps16 and rpl32, allowed the estimation of the period of gene transfer. This study provides novel evidences for the processes involved in gene transfer and substitution from the cp to the nucleus and the requirement of an extra-long period for the successful completion of these processes.

本論文は,葉緑体遺伝子の核ゲノムへの転移と置換について,2種の遺伝子(rps16, rpl32)に注目して,さまざまな植物種の核ゲノムおよび葉緑体ゲノムの構造と転写産物を比較することにより,進化の過程で起こった変化を示しメカニズムの一端を明らかにしたものである.

第1章では,緒言として細胞内共生説とオルガネラゲノムの進化に関するこれまでの知見を概説し,「葉緑体遺伝子の転移」および「葉緑体遺伝子の置換」が具体的な遺伝子を例にして定義されている.原始真核細胞にαプロテオバクテリアの祖先種が取り込まれ共生することでミトコンドリアを持つ真核細胞が成立し,その後さらにラン藻の祖先種が取り込まれて共生することで葉緑体となり植物細胞が成立した.進化の過程で,ラン藻が持っていた遺伝子の多くは葉緑体から脱落し,一部は原始真核細胞由来の遺伝子産物が葉緑体で機能するようになった.また,一部の葉緑体の遺伝子は核ゲノムに移行しその遺伝子産物が葉緑体に輸送されて機能するようになった.このように,葉緑体ゲノムから遺伝子が核ゲノムに移行し,葉緑体ゲノムからその遺伝子が欠失して「葉緑体遺伝子の転移」が完了する.また,核ゲノム上の原始真核細胞由来あるいはミトコンドリア由来のオルソログの産物が葉緑体で機能するようになり,対応する葉緑体ゲノム上の遺伝子が欠失することで,「葉緑体遺伝子の置換」が完了する.

第2章では,Arabidopsis属およびその近縁種におけるrps16遺伝子の置換について,比較ゲノム解析を中心とした研究結果が示されている.葉緑体のリボソームタンパク質S16をコードする遺伝子(cp rps16)は,多くの植物ではグループIIイントロンにより分割された2個のエクソンからなる遺伝子として葉緑体ゲノムに存在する.ところが,Medicago truncatulaとPopulus albaでは,葉緑体ゲノムからはrps16が欠失し,ミトコンドリアから核ゲノムへ転移したrps16がミトコンドリアと葉緑体の双方に輸送され機能していることが報告された.イネおよびシロイヌナズナではミトコンドリア由来の核遺伝子の産物であるRPS16が同様に葉緑体へ輸送されているが,葉緑体ゲノムにはrps16の配列が保持されていた.以上のことは,被子植物の種分化の早い段階でミトコンドリアから核に遺伝子転移したrps16の遺伝子産物が葉緑体でも機能するようになったが,葉緑体ゲノムに残ったrps16は,種間でその存否に違いがあることを示していた.そこで,本研究では,シロイヌナズナを含むArabidopsis属とその近縁の19種の植物における葉緑体ゲノム上のrps16の構造の解析をおこなった.その結果,Arabis hirsuta, Aethionema grandiflorum, Draba nemorosa, Lobularia maritimaの4種についてはコーディング配列内の塩基の欠失や第2エクソンの欠落,ストップコドンの挿入などの変異が認められ,偽遺伝子化していることが分かった.また,RT-PCR産物と葉緑体ゲノムの塩基配列を調べたところ,Arabidopsis thaliana, Olimarabidopsis. pumilaではイントロンが正常にスプライシングされいないことが判明した.これらの結果は,葉緑体遺伝子の置換に伴う葉緑体ゲノム上のrps16の偽遺伝子化が,Arabidopsis属内で種ごとに独立して進んでおり,偽遺伝子化のメカニズムはコーディング配列の変異だけではなく,グループIIイントロンのスプライシング異常によっても起こりうることを示している.

第3章では,キントラノオ目に属する8種の植物種について,葉緑体リボソームタンパク質L32をコードする遺伝子(rpl32)の遺伝子転移の比較ゲノム解析が行われた.多くの植物では,rpl32は葉緑体ゲノム上のndhFとtrnLの間に座乗している.しかし,キントラノオ目に属するBruguiera gymnorrhiza, Populus alba, Populus trichocarpaでは,葉緑体ゲノムから核ゲノム上のCu-Zn superoxide dismutase遺伝子(sod-1)の第7イントロンに転移していることが先行研究により明らかとなっていた.B. gymnorrhizaでは,この遺伝子はオルタナティブスプライシングにより機能的なSODとRPL32タンパク質が翻訳され,両者ともに葉緑体に輸送されている.またP. albaおよびP. trichocarpaでは,転移した遺伝子が重複した後,エクソンの欠失を経てSOD,RPL32をコードする遺伝子がそれぞれ分化したと考えられていた.本研究では,キントラノオ目の他の8種の植物(Passiflora citrina, Euphorbia sieboldiana, Calophyllum inophyllum, Acalypha hispida, Hypericum erectum, Viola mandshurica, Manihot esculenta, Ochna serrulata)について,核ゲノムおよび葉緑体ゲノム上のrpl32のゲノム配列を比較するとともに,それらの転写産物を解析した.O.serrulata, H.erectum, E. sieboldianaの3種は,P. albaやP. trichocarpaと同様に,葉緑体ゲノムから遺伝子が完全に脱落していた.一方,C. inophyllum, M. esculenta, V. mandshurica, P. ctrinaは,葉緑体ゲノムにrpl32配列を保持していた.ただし,V. mandshurica, P. ctrinaの2種については,ゲノム配列内に塩基置換や欠失挿入が認められ,偽遺伝子化していた.核ゲノム側のrpl32についても同様の比較をおこなった.その結果,調べられたすべての種において,sod-1の第7イントロンへの挿入とオルタナティブスプライシングによる転写産物が確認された.また,C. inophyllum, E. sieboldianaでは,遺伝子が重複しており,C. inophyllumの片方のコピーは偽遺伝子化していた.さらに,M. esculentaでは,オルタナティブスプライシングに異常が認められ,機能的なRPL32が翻訳されないことが示唆された.以上の結果から,以下のことが推察された.キントラノオ目の共通祖先において,rpl32は葉緑体ゲノムから核のsod-1の第7イントロンに移行し,オルタナティブスプライシングによって機能的なRPL32が産生されるようになった.その後,rpl32の挿入を持つsod-1の重複と偽遺伝子化あるいは機能分化がそれぞれの種で起こり,それに対応して葉緑体側に残されたrpl32配列の偽遺伝子化や脱落が起こった.M. esculentaにおいては,核側で機能的に発現していたrpl32がオルタナティブスプライシングの異常によって機能を失い,葉緑体ゲノム上に無傷のまま残存していたrpl32に依存することとなった.M. esculentaは,結果的には祖先型の構成に戻ったことになる.

以上要するに,本論文は葉緑体遺伝子の転移および置換が現在も種ごとに独立して進行中であり,スプライシング異常が偽遺伝子化のメカニズムになりうること,また,遺伝子転移および置換には想像されていた以上の時間を要することを明らかにした.これらの知見は,作物ゲノムの進化や環境適応機構を解明するための基盤となることから,学術的価値が高い.したがって,審査委員一同は本論文が博士(農学)の学位論文として価値があるものと認めた.