The Application of Mass Spectrometry-Based Label-Free Quantitative Proteomic Strategies in Cancer Stem Cell Research.

The application of mass spectrometry(MS) has dramatically expanded the frontiers of the analysis of protein expression in biological samples. This thesis addresses the application of a MS-based label-free strategy to analyze complex biological samples. Both technical challenges and computational challenges are discussed. Chapter2 describes a comparative profiling study between two ovarian cell lines (MADH-2774 and TOV-112D) by coupling capillary isoelectricfocusing(cIEF) with reversed phase liquid chromatography followed by MS where 1749 and 1092 proteins are identified, respectively. The PI3K/AKT pathway is found to be predominant in MADH-2774 over TOV-112D by using Ingenuity Pathway Analysis(IPA). Chapter3 applies the technology developed in Chapter2 to study pancreatic cancer stem cells(CSCs). The limited sample quantity is overcome by a modification of the cell lysis procedure and the cIEF technique. The missing data problem where spectral count is equal to zero is also addressed by a modified data transformation algorithm, emphasizing the problem of optimizing the correction factor from an iterative process. A total of 763 and 1031 proteins are identified from the CSC group and the tumor group. Moreover, 169 proteins are found to be differentially expressed. The top network generated by IPA suggests a central role of NF-κβ. Chapter4 represents a more targeted study to investigate the alteration of the glycosylation patterns upon drug treatment in glioblastoma CSCs. Lectin microarray and affinity capture are employed to enrich the glycoproteins. In addition to the t-test, a generalized linear mixed-effect(GLMM) model is also explored to analyze the hierarchical data structure. 8 significant hits are captured by the t-test and 27 significant hits are captured by GLMM. Moreover, 6 of the 8 hits from the t-test are also present in GLMM, suggesting that GLMM is a more robust model to decrease the background noise and increase the differential testing accuracy. Chapter5 represents a more comprehensive study covering both discovery and verification phase. The label-free MS method is used in the discovery phase followed by the prioritization of protein candidates. Both immunoassay and multiple reaction monitoring are used to validate the changes upon treatment. A putative signaling network is constructed, suggesting a phenotype transformation towards non-tumorigenic cells.Ph.D.BioinformaticsUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/84562/1/dailan_1.pd