(410e) Searching for Self-Renewal and Pluripotency Defining Genes in Transcriptome Space of Embryonic Stem Cells

Embryonic stem cells are characterized by their ability to self-renew indefinitely in culture and differentiate into all the three germ layers (pluripotency). The cellular state that endows these cells with these characteristics is still not well understood. Recent revelation of cellular reprogramming during the formation of induced pluripotent stem cells adds another dimension to this problem but also provides new insights into the molecular circuitry responsible for these characteristics. Microarray data from different embryonic and induced pluripotent stem cell studies was compiled from public repositories. Linear and Robust Multi-array Average normalizations were performed on the compiled datasets. Non-negative matrix factorization, a tool for factorizing high-dimensional data, was employed to identify predominant biological trends in the data. Consequently, samples were segregated into classes that were indicative of different degrees of potency. Subsequently, differential expression analysis was conducted between the predicted classes of interest using Significance Analysis of Microarrays (SAM). Functional enrichment amongst the differentially expressed genes was determined by using Metacore and Ingenuity Pathway Analysis. Genes upregulated in the pluripotent stem cells of both species (human and mouse) showed enrichment in cellular functions including DNA damage repair, G1 to S phase progression and development associated TGFβ/Activin signaling.

ARACNE, developed by Basso et al. (Nature Genetics.2005), uses gene expression profiles to identify cellular regulatory networks. This algorithm was used to predict genes whose expression is closely correlated to that of stem cell-associated genes such as Oct4, Nanog, and Sox2 etc. Genes exhibiting very similar expression profiles can be manipulated at the protein level in very different ways.

Analysis of transcription-level data is ultimately limited due to the lack of information about post-transcriptional or post-translational regulation. However, integration of transcriptome signatures with other sources of genomic and proteomic evidence (e.g. protein-protein interactions or phenotype data) can address this limitation. To search for network-level mechanisms explaining the observed changes in expression, genes exhibiting stem-cell associated changes in expression were overlayed on a mouse functional genetic network consisting of over 20000 linkages among protein-coding genes based on a diverse collection of genomic and proteomic data (Guan et al. PLoS Comput Biol. 2008). Integration of transcriptome data with this comprehensive functional network was used to corroborate the results from ARACNE.

These novel associations predicted by ARACNE and the functional network can be used to develop new hypotheses to test experimentally. This demonstrates that the vast amount of diverse public transcriptome data can be utilized to gain novel insights into stem cell-associated complex biological processes through the use of statistical and functional genomics tools.