Rainbow trout (Oncorhynchus mykiss) is a cold-water aquaculture species which is of a considerable economic importance. It is originally a native fish in the North America. It became one of the most popular aquaculture species following the introduction to all continents since 1874. The global annual production of rainbow trout was almost tripled from 259,161 tons in 1989 to 732,432 tons in 2009 (FAO, 2009).

The first introduction of rainbow trout to Japan was in 1877, and its mass production started in 1950s (Matsuda, 1992), when a strain named Donaldson established by Dr. Donaldson in the University of Washington (Donaldson et al., 1957) was introduced to Japan. The Donaldson strain showed several advantages such as increased growth rate, disease resistance and enhanced egg production. Since 1966, Miyazaki Prefectural Fisheries Experimental Station developed a thermally selected strain of rainbow trout based on the Donaldson strain by traditional selective breeding. This strain acquired a degree of upper temperature tolerance as revealed by the fact that the strain grows normally and also feeds actively at 24°C, in contrast to the optimum temperature of water for the normal strain below 20°C. Additionally, the thermally selected strain survived from occasional exposure to heated water at 30~35°C for 1 to 5 min, which is much higher than the upper limit of normal survival water temperature around 24°C (Ineno et al., 1993). A related research indicated that mitochondrial cytochrome c oxidase subunit II (COXII) gene was associated with upper temperature tolerance in eggs and embryos (Ikeguchi et al., 2005). However, the molecular mechanism involved in this thermal tolerance has been still unknown.

The objective of the present study was to investigate genes related to upper temperature tolerance in rainbow trout at juvenile and adult stages. First, the expression levels of COXII gene in five different tissues from juveniles were compared between normal and thermally selected strains. Next, thermally selected diploid rainbow trout genome and gynogenetic double haploid genome of the same strain were sequenced by next-generation sequencing (NGS) technologies. Genome-wide microsatellite candidates were exploited with the sequences of diploid genome of thermally selected strain by a GS-FLX sequencer (Roche). Then, genome de novo assembly of thermally selected gynogenetic rainbow trout was conducted using Genomic Analyzer II (GAII) and HiSeq 2000 sequencing systems (Illumina). Moreover, the comparative transcriptome analysis before and after exposure to high temperatures were performed using various tissues from both normal and thermally selected strains to examine the genome-wide gene expression patterns at upper temperatures.

1. COXII mRNA expression analysis

The different expression levels of COXII mRNA between thermally selected and normal strains were analyzed by real-time PCR using brain, heart, liver, muscle, and skin tissues from one year-old juveniles (length, 9.9-13.9 cm; weight,13.2-31.0 g). After triplicated real-time PCR measurements, the expression levels were calculated by the ΔΔCT method with elongation factor-1 alpha gene as an internal standard. When the the data between the two strains were analyzed by student's t-test, no significant differences were found in all tissues examined.

2. Generation of gynogenetic double haplotype fish

In order to generate high quality genomic sequences of rainbow trout, we attempted to use a double haploid fish as a genomic DNA source. For this purpose, a gynogenetic rainbow trout offspring (length, 2.1 cm; weight, 1.3 g) from the thermally selected strain was produced by the fertilization with UV-irradiated masou salmon (Oncorhynchus masou) sperms followed by the blockage of the first egg cleavage. Prior to Illumina GAII and HiSeq 2000 sequencing, microsatellite genotypes of the gynogenetic sample were verified. Genomic DNAs of the gynogenetic juvenile and three normal diploid fish were examined using 16 polymorphic microsatellites which were selected from a published linkage map (Rexroad et al., 2008). These microsatellite markers were amplified by PCR, and their genotypes were examined by fragment analysis using an ABI 3100 sequencer and Peak scanner software. As a result, all 16 microsatellites were homozygous for the gynogenetic fish, while most of them in the three diploid fish were heterozygous, indicating that the gynogenetic fish had the double haploid genome.

3. Next-generation sequencing

The rainbow trout genome was sequenced by 3 different next-generation sequencing platforms, GS-FLX, GAII and HiSeq 2000 sequencers. For the GS-FLX sequencing, genomic DNA was extracted from pelvic fin tissues of a three-year-old female of the thermally selected diploid strain. It was used as a genomic reference for genome-wide genetic marker discovery. Five rounds of the 454 GS-FLX sequencing generated 4,634,401single-end reads with total length of 1,531,336,345 bp (average read length, 330bp). Quality filtering (quality score, 20<) and size selection (discarding the sequences of < 100 bp) for the raw reads were performed using FASTX-toolkit software. After the removal of duplicated sequences by CD-HIT, 1.4 Gb of high quality sequences were obtained using a GS-FLX sequencer.

The genomic DNA of the gynogenetic fish (length, 2.1 cm; weight, 1.3 g) was prepared for sequencing with Illumina sequencers. An Illumina GAII sequencer produced 43,722,295,364 bp paired-end data (insert size, 400 bp and 700 bp; read length, 101 bp). Another 201,354,918,016 bp mate-pair reads were generated by an Illumina Hiseq 2000 sequencer (insert size, 2kb and 5kb; read length, 76 bp). Quality filtering and low quality base masking (quality score, 20; the base with a quality score lower than 20 was masked by N) for the data were performed by FASTX-toolkit software.

In total, 246.4 Gb sequence data were obtained by the three sequencing platforms, which covered 102.7x of the estimated rainbow trout genome size, 2.4 Gb, which will be described later in more details.

4. Massive genetic marker discovery

To discover the microsatellite genetic markers on the rainbow trout genome, the sequenced reads using a GS-FLX sequencer were loaded into MIcroSAtellite identification tool (MISA) to search for the potential microsatellite candidates. The microsatellite candidates met the requirement of repeat unit (2~8 bp) over 3 times and had the minimum length of 20 bp. Total 215,024 microsatellites were detected, which accounted for 0.65% of the 1.4 Gb rainbow trout genomic sequences. Dinucleotide microsatellite candidates that accounted for 70.7% of all discovered candidates were dominated among all kinds of repeats, which were followed by tetra- (12%) and pentanucleotide (5.7%) microsatellite candidates. The primers of all candidates that had sufficient flanking regions were designed by the integration of MISA and Primer3.

5. Genome de novo assembly

The accurate estimation of the rainbow trout genome size is preferable for the following de novo assembly step. To evaluate the genome size, k-mer distribution histogram was useful. As k-mers represent k nucleotides generated from sequencing reads iteratively, the occurrence of k-mers reflects the sequencing depth. Thus, the actual sequencing coverage is observed in the k-mer distribution histogram. Generations of k-mers (k = 20, 30) were carried out and, then, their frequencies were calculated by Jellyfish software. The peak frequencies of 20-mer and 30-mer for the sequencing reads of the gynogenetic fish were estimated to be 6.7 and 5.4, respectively. The rate of sequencing errors was also estimated from the histogram. The genome size was calculated with the total number of k-mers, the peak frequencies of k-mer, and the error rate. As a result, the estimated genome size of rainbow trout was about 2.4 Gb.

All of pair-end (400 bp and 700 bp insert size) and part of mate-pair (2kb insert size) sequencing data from the gynogenetic fish were used to perform genome de novo assembling using SOAPdenovo software. In order to optimize settings of de novo assembly, k-mers of 51~71 were selected to generate contigs by de Bruijn graph algorithm. Subsequently, the distance information of all pair-end and mate-pair reads (400 bp, 700 bp and 2kb) were loaded into Soapdenovo to perform scaffolding. These processes were performed step by step from short (400 bp) to long (2kb) sizes.

When the k-mer of 59 was employed, the longest total contig size of 1,861,100,116 bp was obtained, which accounted for 77.5% of the genome size of rainbow trout, whereas the 51-mer showed the highest coverage of scaffold (2,239,632,961 bp, accounting for 93%). The N50 lengths of contigs and scaffolds of this assembly were 577 bp and 15,046 bp, respectively. There were still about 0.5 Gb gaps in the assembled sequences.

6. De novo transcriptome assembly of normal and thermally selected strains

Total RNAs were extracted from brain, heart, liver, muscle and gill of both normal and thermally selected strains before and after exposure to high temperature at 26°C for 30min. cDNA libraries were constructed using these total RNAs and subsequently sequenced with a HiSeq 2000 sequencer using the pair-end method with 150 bp insert size. The total sequence lengths of the samples ranged from 8,081,204,488 bp (gill tissues of thermally selected strain after exposure to high temperatures) to 11,699,637,448 bp (muscle tissues of the normal strain before exposure to high temperatures). To obtain high quality sequences, quality selections to filter out low quality reads were performed. For each sample, 50M reads data were randomly selected to perform further analysis. These selected 50 M paired sequences were applied to Velvet and Oases software packages to generate the contigs of transcripts. Among five different tissues, brain tissues produced the most varieties of contigs of transcripts : 55k, 52k, 52k, and 49k contigs of 1000-bp or more in size were obtained in normal-before, normal-after, thermally-selected-before, and thermally-selected-after fish, respectively. On the contrary, the numbers of contigs from muscle tissues were 13k, 11k, 12 and 11k, respectively.

The homology search of these de novo assembled transcripts was conducted using BLASTX (E-value = 1E-10) against the NCBI protein database. Then, they were annotated by Blast2Go to assign gene ontology (GO) terms and were mapped on the known KEGG pathway. There were also found many novel genes that showed low similarity with known genes in the NCBI database. Consequently, detailed expression profiles of annotated and unannotated genes for each tissue were constructed by mapping quality-filtered raw reads back on these de novo assembled transcripts using Bowtie software. Liver tissues showed 673 known genes that exhibited at least two-fold changes of expression after exposure to high temperatures, including many heat shock protein family genes and BAG family protein regulator genes. These results are very useful to find genes responsible for upper temperature tolerance of rainbow trout.

7. Conclusions

In this study, both genomic and transcripome analyses of the thermally selected rainbow trout were investigated. For the survey of COXII mRNA expression, no significant differences were observed in brain, heart, liver, muscle and skin tissues between the normal and thermally selected juveniles in contrast to the significant expressions observed in embryos at 2 to 4 cell stages of the thermally selected strain (Ikeguchi et al., 2005). The discovered microsatellite candidates will be helpful to assist future marker-assist selections, because there has been still no useful genome reference of rainbow trout. Thus, the newly assembled draft genome sequence of rainbow trout in this study can be utilized as a reference to assist further genetic and genomic studies of rainbow trout. By the comparative gene expression analysis, lists of candidate genes for the thermal tolerance such as heat shock protein 90 alpha gene and DNAj homolog subfamily b member gene were presented. Future studies based on these results will provide more detailed information for the genes and mutations that are responsible for upper temperature tolerance in fish.

近年、表面化してきた地球温暖化によって各種動植物の生育・飼育環境が変化し、その生存/生産に多大な影響が生じることが懸念されている。ニジマスは水産業的に重要な冷水性魚であり、世界各地で養殖されているが、ニジマスも地球温暖化による影響を免れ得ないことが懸念されている。現在地球温暖化に対する対策として二酸化炭素削減のための各種取り組みがなされているが、生物学的取り組みとして温暖化に対して耐性のある個体を選抜育種等の手段で準備する試みが水産生物を中心に我が国で始まっている。その一環として、宮崎県で選抜育種されてきた高温耐性ニジマスが注目されている。本申請者は同高温耐性ニジマスにおいて、最終的に高温耐性を司る遺伝的要因を同定すべく、ニジマスのゲノム解析、トランスクリプトーム解析を推進し、以下の内容から構成された博士論文を提出した。

1 COXII遺伝子の発現解析

先行研究において初期発生の時点ではCOXII遺伝子のレベルが高温耐性ニジマスと標準ニジマスで大きく異なっていることが示されていた。そこで幼魚でCOXII遺伝子の発現レベルが両者で異なっているか検討したが、有意な差は認められなかった。このことから、少なくとも幼魚ではCOXII遺伝子の発現と高温耐性形質との間に強い関連はないものと考えられた。

2 雌性発生個体の作出と接合様相の確認

全ゲノムショットガン法によるゲノムアセンブルに資するため、アセンブルの障害となる「多型」のない第一卵割阻止型雌性発生ニジマスを宮崎県水産試験場に依頼して作出した。その雌性発生個体を調べるため、マイクロサテライトメーカーをPCR後にラベルする手法を導入した。16種類のマイクロサテライトマーカーを用いて多型解析を行なった結果、作出した個体が目的とするホモ接合(ダブルハプロイド)個体であることが確認された。

3 次世代シーケンサーによるシーケンシングと新規アセンブル

通常のヘテロ接合個体より抽出したゲノムDNAを出発材料に、次世代シーケンサー454 GS-FLXを用いてシーケンシングを行ない、463万リード、平均リード長330bp、のべ約1.5Gbの塩基配列を得た。これらのデータから約20万種類のマイクロサテライト多型マーカー候補が得られた。

次に作出したホモ接合個体からゲノムDNAを抽出し、Illumina GAを用いて、平均400bpと700bpのゲノムDNA断片のペアーエンドシーケンシングを行った。これらのデータを20-30塩基長単位で1塩基ずつスライドさせて構成した塩基配列の集合を構築し、同一の配列の出現頻度を解析することで、シーケンスのエラー率とゲノムサイズの推定を行なうことができた。エラー率は約0.5%、ゲノムサイズは約2400Mbと推定され、ゲノムサイズについては既報の推定値とほぼ一致した。さらに、平均2kbと5kbのメートペアライブラリーを調製し、同様にシーケンスに供した。これらによりのべ100倍ゲノム相当の塩基配列データが得られた。得られた塩基配列のうち約60倍ゲノム分のデータを用いて、アセンブルソフトSoapdenovoによりde novoアセンブリを行った。アセンブルはサイズ推定のステップで行なったのと同様、塩基配列データを分割してからde Bruijnのグラフを作成するプロセスで行なうが、分割単位であるk-mer値を様々検討した結果、k=59 bp前後で最適な結果が得られた。この条件によりアセンブルで得られたScaffoldサイズの合計は2.2Gb、N50値は15,046b、推定ゲノム被覆度は91.9%であった。

これらに加えて、一塩基多型(SNP)を主とした多型データを得るため、高温耐性系(ヘテロ接合個体)、標準系のシングルリードシーケンシングを行い、データを蓄積した。

4 cDNA配列の新規アセンブルと頻度解析

高温耐性系および標準系それぞれ一個体につき、熱ストレス付加前後の各組織から、それぞれ50Mペア以上のリードをえた。これらからクオリティーの低位のデータを除去し、各系統・組織・条件につき25Mのシーケンスデータを用いてリードのアセンブル、頻度解析を行なった。特に試験的な解析でヒートショックタンパク質(heat shock protein, HSP)遺伝子で特徴的な結果が見いだされた鰓について、より詳細な解析を進めた。熱ストレス付加前・鰓に関して遺伝子の発現頻度を調べた結果、高温耐性系で標準系より発現が2倍以上多い遺伝子が324種類同定されたが、特に複数のHSP遺伝子が顕著な差を示した。本結果から、HSPの恒常的な発現上昇がニジマスに高温耐性を付与している可能性が示された。

以上の成果は学術上、応用上資するところが大きく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あるものと認めた。