Using whole-genome wide gene expression profiling for the establishment of RNA fingerprints

Abstract

<p>This thesis introduces the term RNA fingerprint and applies this concept to different scientific questions in immunology and medical diagnostics. A RNA fingerprint comprises observed transcriptional changes in response to a molecular signal. Molecular signals include activated oncogenic pathways by introducing the oncogene as a transgene, receptor ligand interactions, treatment of cells with inhibitory factors or responses of cells to different diseases. In this thesis four different concepts of RNA fingerprints are introduced. The first concept deals with <em>in vitro</em> generated gene signatures for different T cell inhibitory molecules, including TGFβ and PD-1 which are then termed RNA fingerprints of these molecules. By applying supervised and unsupervised classification methods based on the RNA fingerprints of both, TGFβ and PD-1 it is then shown that T cells derived from patients with Hodgkin’s lymphoma are under the influence of both, TGFβ and PD-1. The concept is then extended to a disease-specific RNA fingerprint in a diagnostic setting. Here a lung cancer specific RNA fingerprint is developed to predict the occurrence of lung cancer prior to clinical manifestation. A further concept deals with the use of pre-defined RNA fingerprints. These can be extracted from biological databases that include information about genes belonging to special pathways or groups of genes with similar functions. I have developed a new and very simple gene-class testing method, GOAna, which is based on RNA fingerprints provided by the Gene Ontology (GO) Consortium. Using GOAna, it is possible to perform an unbiased analysis based on all branches of GO. The last concept introduces the idea of considering the microarray experiment itself as a RNA fingerprint. I hypothesized that all transcriptional changes which are revealed by a microarray experiment can serve as a RNA fingerprint and can decipher underlying signaling mechanisms. The presented gene-class testing algorithm is extended by a network-construction algorithm to determine key player genes which link the identified significant gene spaces. Using this approach a key player within the PGE2 signaling pathway in CD4+ T cells is identified and experimentally validated. Additionally, a software package, IlluminaGUI, which allows the researcher to establish and apply RNA fingerprints to gene expression data derived from Illumina’s Sentrix BeadChip technology is introduced. IlluminaGUI is implemented as a graphical user interface and is intended to enable the interested life scientist who is not familiar with a command line based environment like the R language to analyze microarray experiments. Finally, critical issues concerning the used technology are raised. All described approaches for the creation of RNA fingerprints are heavily dependent on the reliability of the microarray format used for the study. Here the continuity of RNA fingerprints is discussed when a new version of a microarray with updated probe content becomes available. </p>