{"created":"2023-06-20T13:20:56.892527+00:00","id":1015,"links":{},"metadata":{"_buckets":{"deposit":"6fef5ac6-4567-4adf-b65e-8e5b270b5ae1"},"_deposit":{"created_by":1,"id":"1015","owners":[1],"pid":{"revision_id":0,"type":"depid","value":"1015"},"status":"published"},"_oai":{"id":"oai:ir.soken.ac.jp:00001015","sets":["2:430:20"]},"author_link":["0","0","0"],"item_1_creator_2":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"小笠原, 理"}],"nameIdentifiers":[{}]}]},"item_1_creator_3":{"attribute_name":"フリガナ","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"オガサワラ, オサム"}],"nameIdentifiers":[{}]}]},"item_1_date_granted_11":{"attribute_name":"学位授与年月日","attribute_value_mlt":[{"subitem_dategranted":"2005-09-30"}]},"item_1_degree_grantor_5":{"attribute_name":"学位授与機関","attribute_value_mlt":[{"subitem_degreegrantor":[{"subitem_degreegrantor_name":"総合研究大学院大学"}]}]},"item_1_degree_name_6":{"attribute_name":"学位名","attribute_value_mlt":[{"subitem_degreename":"博士（理学）"}]},"item_1_description_12":{"attribute_name":"要旨","attribute_value_mlt":[{"subitem_description":"The advent of whole-genome sequencing and large-scale profiling of gene expression has revealed several unexpected phenomena in the genome that had never been discovered from the studies of a small number of genes. With the examination of the factors responsible for the formation of such new phenomena, it is conceivable that new relationship between underlying biological processes will be elucidated. In this thesis, I examine such phenomena in the transcriptome of a large variety of organisms, using data from public databases and those obtained in our laboratory and I report unexpected relationships in the transcriptome evolution. <br /><br />Zipf's law of transcriptome is one of the phenomena that have been revealed by genome-wide studies of gene expression. This law states that there is a relationship between the transcript frequency (f) and abundance rank (r) represented as f=k/r<SUP>b</SUP>, where k is a constant and b is a constant parameter that represents the absolute value of the slope in a log-log plot of transcriptome frequencies. I reported in my published paper that this law was applicable to all human normal tissues that I observed. Further, in muscle and liver, which are primarily composed of a homogeneous population of differentiated cells, the slope parameter b was nearly equal to 1. In cell lines, epithelial tissue and compiled transcriptome data, only high-rankers deviate from the law. In addition to my work, several other authors also reported this law in other species. It has been known that this law is applicable to a large variety of species, such as vertebrate s (Homo sapiens, Mus musculus and Rattus norvegicus), invertebrates (Drosophila melanogaster and Caenorhabditis elegans), other eukaryotes (Saccharomyces cerevisiae and Arabidopsis thaliana) and even to bacteria (Escherichia coli). It is remarkable that the observed slope parameter b is almost unique (b〓1) irrespective of the species investigated. <br />To explain the factors responsible for the formation of the law, I proposed an evolutionary model of Zipf's law of transcriptome. In this model, Zipf's law could be replicated based on three assumptions. The first assumption states that the baseline expression level of each gene in a cell is coded in the genome sequence such as in the cis-elements or enhancer regions; therefore, the expression levels of genes are affected by mutations, and these are inherited to the offspring. This assumption is supported by the fact that in a large variety of organisms, the expression levels of genes have abundant natural variation and familial aggregation. In addition, it is known that the location of the cis-element and/or the trans-factor of the genes can be determined with the quantitative trait locus (QTL) analysis in which the expression level of each gene is treated as a quantitative trait. The second assumption states that the expression level changes in stochastic proportion to its intensity. This assumption is supported by the observation that expression differences accumulate at a constant ratio in primates and rodents. The third assumption states that the number of expressed genes in a cell type is nearly constant throughout the evolutionary process and that any functional gene is prohibited from losing its gene expression ability. By the Monte-Carlo simulation of the model, I showed that a stable distribution of f=0.1/r was obtained from these three assumptions, regardless of the initial distribution. To demonstrate that the uniqueness of slope parameter b among variety of species can be replicated from the evolutionary model, I conducted a Monte-Carlo simulation to determine the condition for converging the distribution with the slope parameter b〓1. In my model, the slope parameter b depends on the number of mRNA molecules in a cell (M), the number of genes expressed in the cell (G), and the permissible lower limit of expression level (L). When the value of parameter M is fixed to 300,000, as in the case of a typical human cell, and L is set to a sufficiently small value, i.e., 1-2 copies/cell, the distribution converged to b〓l over a wide range of values of parameter G, i.e., from 10,000 to 50,000 genes in a cell. This is the reason for the universality of b, which is predicted by the evolutionary model of Zipf's law. <br />At approximately the same time, several authors (Kuzunetosov 2003, Frusawa and Kaneko 2003, Ueda 2004) proposed other models of Zipf's law of transcriptome independently (see reference in the thesis 20, 21, 22). All of their models attributed the Zipf's law of transcriptome to the gene expression dynamics in each cell (the dynamics model, hereafter). They argued that the change in gene expression level in each cell follows the formulation of geometrical Brownian movement. However, in addition to the lack of reliable biological evidence for the dynamics model, I pointed out that the dynamics model cannot replicate the observed Zipf's law of transciptome even in mathematical sense, contrary to the authors' assertion. From the dynamics model, it follows that the rank order of expression level in each cell independently diverged at random, even if the same type of cell was considered. It is noteworthy that the observations of expression level distribution were obtained from a mixture of millions of cells. The central limit theorem of probability theory states that the distribution obtained from such a mixture of a large number of completely diverged samples should be a normal distribution. Therefore, if the dynamics model is valid, the observed distribution of the expression level of genes should follow a normal distribution, not a Zipf's law distribution. Such divergence does not occur in my evolutionary model; hence, Zipf's law is replicated. <br />Obviously, the determination of the correct cause of Zipf's law of transcriptome critically influences the direction of further investigation. Based on the dynamics model of Zipf's law, Ochiai et al. (2004) derived a formula that describes the elementary process of gene expression dynamics in a cell. Determining such a formula is crucial for estimating gene regulatory networks from time course data of gene expression profiles. However, if my assertion is valid, the proposed formula will lose its ground. If my evolutionary model is accepted, Zipf's law of transcriptome would be related to the neutral model of transcriptome evolution, proposed by Khaitovich et al. (2004) (53). They discovered a clocklike accumulatiqn of gene expression divergence within primates and rodents (53). These results were in agreement with the observation of Rifkin et al. (2003), who reported that differences in gene expression were consistent with phylogenetic relationships among Drosophila species (l2). Zipf's law of transcriptome can be viewed as a new support for the clocklike accumulation of expression diversity, because the assumption of my evolutionary model is nearly equivalent to the neutral model of transcriptome evolution. <br /><br />In the evolutionary model of Zipf's law, I focused only on the evolutionary change in expression levels of genes in a cell. Next, I tried to extend my study to the expression patterns in various tissues (anatomical expression pattern, hereafter). To investigate the evolution of anatomical gene expression patterns, I focused on housekeeping genes as a special set of genes that were definitely expressed and function in all cell types. <br />Identification of housekeeping genes from large-scale expression profiles was first exemplified by Velculescu et al. (1999) who used the SAGE method (24). This was followed by Warrington (2000) and Hsiao (2001) who used oligonucleotide microarrays (25, 26). These studies opened up new opportunities to explore the relationship between expression patterns and other features of genes, such as gene length, sequence divergence, location in the chromosomes, and so on. Several sets of housekeeping genes were published along with such studies, but it has rarely been well discussed whether or not the analyzed set of genes is a non-biased representative of housekeeping genes. In fact, by comparing the two published screenings for housekeeping genes, one based on the GeneChip method and the other based on the SAGE method, I found that there was a low concordance between the results of the two screening methods. I also found that, in both processes, there was poor sensitivity in the identification of housekeeping genes. Therefore, I examined the causes of this inconsistency, and by tuning the parameters for housekeeping gene selection, I compiled a more reliable set of housekeeping genes. In this study, I found a good correlation between the observed breadth of gene expression (the number of organs in which gene expression was detected) and the expression level of genes, even in a set of known housekeeping genes. Based on this, I concluded that the expression level of a gene seriously affects the apparent breadth of its expression. This was particularly manifested in the result where I succeeded in doubling the number of housekeeping gene candidates (from 2,792 to 5,537) without losing specificity. The newly identified housekeeping genes (new HK) and the previously identified housekeeping genes (old HK) shared features in terms of constancy of expression abundance among tissues (expression evenness), cellular localization of products, and the fraction of genes that have CpG islands at their transcription start sites. Estimated contaminants, which comprise approximately 12%-20% of either new or old HK, were genes that were unique to widely distributed cells rather than those that were common to a wide variety of cells.<br /><br />Main points of this thesis are summarized as follows. <br />Part I<br />1. I reported that mRNA frequencies in human normal tissues obeyed the Zipf's law.  Especially, in the organs that are primarily composed of a homogeneous population of differentiated cells, the slope parameter was nearly equal to 1.  <br />2. I proposed a new theoretical model for explanation of the factors responsible for the formation of the Zipf's law. It is the evolutionary model of the Zipf’s law of transcriptome. Further, I gave several experimental supports for the each assumption of the model.  <br />3. I concluded that the gene expression dynamics models for the Zipf's law of transcriptome are not valid because they can not replicate the Zipf's law even in mathematical sense. <br />Part II <br />In order to extend my study from the gene expression strength in a tissue type to the expression patterns in various tissues (anatomical expression pattern), I focused on housekeeping genes as a special set of genes that were definitely expressed and function in all cell types. <br />1. By comparing the two representative large scale screenings for housekeeping genes in the human genome, I found that there was a low concordance between the results of the two screening methods and there was poor sensitivity in the identification of housekeeping genes. <br />2. I demonstrated that the cause of the low concordance was that the expression level of a gene seriously affects the apparent expression breadth (the number of tissues in which the gene was expressed), because there was a good correlation between the observed breadth of gene expression and the expression level of genes, even in a set of known housekeeping genes. <br />I compiled a new and more reliable set of housekeeping genes, and I succeeded in doubling the number of housekeeping gene candidates (from 2,792 to 5,537) without losing specificity. <br />","subitem_description_type":"Other"}]},"item_1_description_18":{"attribute_name":"フォーマット","attribute_value_mlt":[{"subitem_description":"application/pdf","subitem_description_type":"Other"}]},"item_1_description_7":{"attribute_name":"学位記番号","attribute_value_mlt":[{"subitem_description":"総研大乙第147号","subitem_description_type":"Other"}]},"item_1_select_14":{"attribute_name":"所蔵","attribute_value_mlt":[{"subitem_select_item":"有"}]},"item_1_select_8":{"attribute_name":"研究科","attribute_value_mlt":[{"subitem_select_item":"生命科学研究科"}]},"item_1_select_9":{"attribute_name":"専攻","attribute_value_mlt":[{"subitem_select_item":"18 遺伝学専攻"}]},"item_1_text_10":{"attribute_name":"学位授与年度","attribute_value_mlt":[{"subitem_text_value":"2005"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"OGASAWARA, Osamu","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2016-02-17"}],"displaytype":"simple","filename":"乙147_要旨.pdf","filesize":[{"value":"494.9 kB"}],"format":"application/pdf","licensetype":"license_11","mimetype":"application/pdf","url":{"label":"要旨・審査要旨","url":"https://ir.soken.ac.jp/record/1015/files/乙147_要旨.pdf"},"version_id":"d047a965-6ab5-416a-a0ac-c286948c99bb"},{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2016-02-17"}],"displaytype":"simple","filename":"乙147_本文.pdf","filesize":[{"value":"3.9 MB"}],"format":"application/pdf","licensetype":"license_11","mimetype":"application/pdf","url":{"label":"本文","url":"https://ir.soken.ac.jp/record/1015/files/乙147_本文.pdf"},"version_id":"e6cabef1-fa86-481c-8eb7-d087016ad417"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"eng"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"thesis","resourceuri":"http://purl.org/coar/resource_type/c_46ec"}]},"item_title":"Statistical Analysis of Anatomical Expression Pattern and Expression Strength.","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"Statistical Analysis of Anatomical Expression Pattern and Expression Strength."},{"subitem_title":"Statistical Analysis of Anatomical Expression Pattern and Expression Strength.","subitem_title_language":"en"}]},"item_type_id":"1","owner":"1","path":["20"],"pubdate":{"attribute_name":"公開日","attribute_value":"2010-02-22"},"publish_date":"2010-02-22","publish_status":"0","recid":"1015","relation_version_is_last":true,"title":["Statistical Analysis of Anatomical Expression Pattern and Expression Strength."],"weko_creator_id":"1","weko_shared_id":-1},"updated":"2023-06-20T16:09:26.309146+00:00"}