Investigation of leukocyte transcriptomes using serial analysis of gene expression

Thesis

Two tag-based gene expression technologies (serial analysis of gene expression, SAGE, and massively parallel signature sequencing, MPSS) were used to profile a variety of lymphocyte transcriptomes. By combining these libraries with publicly available genome and transcriptome data, both immunological and general aspects of gene expression could be considered. Unexpectedly, analysis of the expression of currently known cell surface components and the proteins corresponding to "immune specific" tags in a cytotoxic T-cell (CTL) library suggested that the current knowledge of the immune specific composition of the cell surface components of a resting CTL is largely complete. An analysis of the "immune specific" tags in a natural killer cell library revealed that a small number of tags could not be matched to any previously sequenced transcripts, suggesting the presence of functionally important but previously uncharacterised transcripts or exons in these cells. To examine the entire transcriptome large libraries are required, implying that MPSS would be the most appropriate technology. A comparison of libraries produced by the two tag-based technologies to characterise CD4⁺ T-cells revealed a relatively poor correlation, suggesting bias in the two techniques. Investigations into this bias led to the conclusion that despite its great depth, the random sampling events involved in the production of a library limit the breadth of MPSS sampling, making it inappropriate for characterising entire transcriptomes. Finally, with the availability of large LongSAGE libraries it is now possible to examine weakly expressed transcript classes. A panel of LongSAGE libraries was used to conduct the first large scale quantitative study of the expression of cis-natural antisense transcripts (cis-NATs). cis-NATs were found to be expressed at approximately one tenth of the level of sense transcripts and across the panel of libraries cis-NATs were found for approximately two thirds of all observed sense transcripts. This suggests antisense transcription is more widespread than previously thought.

Authors: Hene, L

• Publication date: 2005
• DOI: 10.5287/ora-qav0vzyq6
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English