I. Introduction

Since human genome and several other organisms' genomes are completely sequenced, gene regulation is realized to be more immense and more complex than expected. The functional properties are known not only on protein coding sequences but also non protein coding or untranslated regions. Recently, ten of thousands of non protein coding RNAs (ncRNAs) have been identified. Also in our large-scale studies of human transcriptome, large numbers of putative ncRNAs have been observed, especially from intergenic regions. However, solid evidences for the in vivo existence of them are inadequate, and knowledge on the transcriptional mechanism of the ncRNAs remains elusive.

In this study, the intergenic transcription start site (TSS) clusters, identified by TSS Seq analysis of twelve human cell lines and tissues, were characterized. By multifaceted use of Illumina GA sequencer (Fig.1), made it possible to classify the TSSs depending on possible biological relevance.

II. Materials and Methods

In order to perform genome-wide integrated analysis of the intergenic TSSs, various techniques were combined with the massively paralleled sequencing technology, Illumina GA (Fig.1), namely;

a) TSS Seq; 139,446,730 36 bp-TSS tags, uniquely mapped to the human genome (hg18) without any mismatch from six cell lines (a colon cancer cell line, DLD-1; an embryonic kidney cell line, HEK293; a breast cancer cell line, MCF-7; a normal fibroblast cell, TIG-3; a Burkitt's lymphoma cell line, Ramos; a lung epithelial cell line, BEAS2B) and six normal tissues (brain, heart, kidney, fetal brain, fetal heart, fetal kidney) of humans. Additional, 36,761,810 36 bp-TSS tags from DLD-1, MCF7, HEK293, and TIG3 cells which were cultured in 1% O2.

b) ChIP Seq for RNA polymerase II; 10,687,815 tags for IP and 6,506,186 tags for background control from DLD-1 cells

c) Nucleosome Seq; 19,570,149 single-end tags from DLD-1 cells

d) Shotgun Seq; 3,382,901 shotgun tags

e) RNA Seq; 20,094,475 tags and 14,879,174 tags for nuclear and cytoplasmic fraction from DLD-1 cells

f) Small RNA Seq; 604,141 21-25 bp size-selected RNA tags from DLD-1 cells

RT-qPCR; expression profiling of the iTSCS and neighboring RefSeq genes were performed in 20 kinds of human normal tissues (Clontech Human Total RNA Master Panel II; Adrenal gland, Bone marrow, Brain(whole), Fetal Brain, Fetal liver, Heart, Kidney, Liver, Lung (whole), Placenta, Prostate, Salivary gland, Skeletal muscle, Testis, Thymus, Thyroid gland, Trachea, Uterus, Colon w/ mucosa, and spinal cord)

III. Results

Integrative transcriptome analysis of the intergenic TSCs (iTSCs) revealed that not all of the iTSCs have the same characters. Figure 2 represents the scheme of the classification and the used methods. The findings in this study (will be described in more details below) demonstrated that possible biological relevance iTSCs had distinctive features, thus could be discriminated from the other noise-level iTSCs.

i) iTSCs of >5ppm are only a minor population of the entire iTSCs

Base on 140 millions of TSS Seq tags, 371,849 iTSCs (TSS tag clustered into 500 bp bins) were characterized. However, the number of iTSCs with significant number of TSS tags, >5 parts per million (ppm), at least in one cell type was only 6,039.

ii) iTSCs have strong likelihood of representing TSSs of long ncRNAs

Overlapping analysis of the iTSCs position with the 5'ends of the cDNA collections; MGC and FLJ, and complete cDNA sequencing of shotgun sequence tags, revealed that 395, 1,617, and 464 iTSCs overlapped with the 5'-ends of the MGC, FLJ, and shotgun cDNAs, respectively.

iii) Surrounding sequences of iTSCs of >5ppm have several distinctive features

By analyzing characters of the surrounding sequences of the iTSCs of >5ppm, many characteristic properties were found differently from iTSCs of <5ppm as shown in Table 1.

The ChIP Seq analysis also showed diversity of Pol II bindings in the nearby regions of iTSCs of >5ppm and <5ppm. The results showed that 376 (40%) out of 931 iTSCs which were expressed at >5ppm in DLD-1 overlapped the pol II binding sites, while only 3,305 (7%) out of 46,416 iTSCs of <5ppm overlapped them. Interestingly, among these 3,305 cases, although iTSCs were of <5ppm in DLD-1 cells, those in other cell types were of >5ppm in 305 cases.

iv) Majority of iTSCs of >5ppm expressed as tissue-preferred manner

By using digital tag count of the TSS-Seq, 3,739 (62 %) of iTSCs were found expressed at >5ppm only in one cell type. RT-qPCR assays were performed in additional 20 kinds of human normal tissues revealed that clear signals of 44 out of 67 iTSCs of >5ppm were observed at least in one tissue.

v) A quarter of iTSCs of >5ppm are located in the proximal regions of the RefSeq genes, which their transcription is likely to be influenced from open chromatin of the neighboring transcription

23% (1,382 cases) of the iTSCs of >5ppm were found located within 10 kb downstream on the same strand of the RefSeq genes (RefSeq-linked iTSCs), while this frequency was found only 7% (24,486 cases) in iTSCs of <5ppm. The expression patterns of iTSCs of and their neighboring RefSeq genes were compared by calculating Pearson's correlation of the respective TSS tag counts. The expressions of the RefSeq-linked iTSCs were found generally resembled to the expressions of their nearest RefSeq genes (Fig3A). To further observe the expression correlation in wider cell types, the real time RT-PCR assays were done in 20 normal tissues. Again, general correlation of the expression patterns between the RefSeq-linked iTSCs of >5ppm and the nearby RefSeq genes was observed (Fig3B), on the other hand no such a correlation was found for the non-RefSeq-linked iTSCs of >5ppm.

By nucleosome Seq analysis, formation of the nucleosome-free regions were found just upstream of the TSSs the non-RefSeq-linked TSCs of >5ppm and periodic positioning of nucleosome immediately downstream of the TSS. On the contrary, no such an orderly nucleosome structure was observed in surrounding areas of the RefSeq-linked iTSCs >5ppm either the iTSCs of <5ppm.

vi) Biological relevance of the majority of the non-RefSeq-linked iTSCs is not serving as miRNA precursor

The overlapping analysis of iTSCs with previously reports small RNAs; miRNAs and snoRNAs, and small RNA Seq data was performed. However, for only 22 miRNAs, 11 snoRNAs, and 2 small RNA Seq, complete cDNA sequences directly showed that the mature forms of the small RNAs were actually located within the transcripted regions following the non-RefSeq-linked iTSCs.

vii) Large numbers of iTSC transcripts are exported into cytoplasm

Sub-cellular localization examined by RNA-Seq of nuclear and cytoplasmic fractions in DLD-1 cells, showed that 59% of non-RefSeq-linked iTSCs overlapped at least 3 RNA Seq tags from the cytoplasmic fraction. Of these, 99 iTSCs (17%) were particularly enriched in cytoplasm (p<0.01).

viii) Numbers of non-RefSeq-linked iTSCs showed differential expression in hypoxic response

By comparing digital tag count of the TSS-Seq of normoxic state and hypoxic state of DLD-1, MCF7, HEK293, and TIG3 cells found that in total, 508 and 365 iTSCs of >5ppm showed expression induction or repression by more than 5 fold in response to the hypoxic shock in at least one cell type.

For further study on the biological roles of iTSCs involve in hypoxic response, the differential expression pattern of iTSCs and RefSeq genes were compared by calculating Pearson's correlation of the respective TSS tag counts. Then particular Gene Ontology (GO) terms of RefSeq genes which showed associated expression with hypoxic induced/repressed iTSCs were examined. Interestingly, while various GO terms were associated, 53 iTSCs were found correlated to the term of "hemophilic cell adhesion". Therefore, one of the biological roles of iTSCs in response to hypoxia might relate with cell-cell interaction control.

IV. Discussion

In this study, I described the attempt of characterizing the iTSCs by multifaceted use of massively paralleled sequencer. Integrated analysis of various types of transcriptome data revealed the diversity of iTSC properties, particularly for non-RefSeq-linked iTSCs of >5ppm. Non-RefSeq-linked were found to be determinedly and independently regulated, which should be essential to realize biologically relevant transcriptions. This study also demonstrated that integrative transcriptome data was a useful starting point for further study on biological roles of the intergenic transcripts.

Fig.1. Schematic representation of the multifaceted use of massively paralleled sequencer

Fig.2. Schematic representation of the integrative transcriptome analysis of the iTSCs

Table 1. Characteristic features observed in the iTSCs of >5ppm and <5ppm

Fig.3. Correlation between the iTSCs and the nearest RefSeq genes

Fig.4.Example of hypoxic induction iTSCs

本論文は、遺伝子間領域に存在する転写開始点の特徴量について、次世代シークエンサー・イルミナGAを用いて、詳細な解析を行ったものである。イルミナGAから産出された約1億5000万の塩基配列データの解析結果、これらの遺伝子間転写開始点から転写される転写産物は明確なタンパク質コード領域を持たず、いわゆるnon-coding RNAに対応するものと考えられた。塩基配列データの計算機的解析により、これまで画一の集団と考えられていた遺伝子間転写開始点群は、それぞれ特徴量の異なるいくつかのグループから形成されることが示された。また、明確に生物学的機能の示唆される群は、それぞれの特徴量の組み合わせについての統計的な解析から区分可能であった。同様に次世代シークエンサーを用いたクロマチン免疫沈降解析、ヌクレオソーム構造解析を始めとする多角的な次世代シークエンサーを用いた解析の結果、これらの転写開始点群近傍には、その転写を合目的的に保証すると考えられる特徴的なゲノム構造が存在していることが明らかとなった。さらに、本論文では、低酸素状態で培養されたヒト癌細胞、4種類について、今回提案された多角的トランスクリプトーム解析を実践し、癌細胞の低酸素応答においては、遺伝子間に見出された転写開始点群は、細胞接着に関係すると考えられる遺伝子群とその発現パターンが密接に相関することを明らかにした。

本論文は、近年のゲノム/トランスクリプトーム解析から大量に同定されつつも、その生物学的意義の不明であった遺伝子間の転写産物に対し、初めて次世代シークエンサーを用いて行われた解析であり、タンパク質をコードしない転写産物の生物学的意義、また遺伝子間領域におけるゲノムからの転写制御様式の全体像の解明に向けて大きく貢献するものであると考えられたために、博士(生命科学)を授与するのに適当であると判断された。