The BTB domain is a well-conserved protein-protein interaction motif. There are 43 BTB-ZF transcription factors in the human proteome. Many of these transcription factors play crucial roles in cancer and developmental processes. The purpose of this project is to identify lists of interactors of the BTB domains of six BTB-ZF proteins with high confidence using a mass spectrometry based approach. The BTB domains BCL6, PLZF, Kaiso, LRF, FAZF, and Miz1 were studied. This study was able to identify 142 putative interactors. The list of putative interactor proteins participates in a wide array of biological functions. Selected putative interactors of the BCL6BTB were also validated using biochemical techniques. In conclusion, this project was able to provide an analysis of the protein-protein interactions mediated by the BTB domains of six BTB-ZF transcription factors. The information generated is valuable to guide future functional and structural studies of the BTB domains.

The BTB domain is a well-conserved protein-protein interaction motif. There are 43 BTB-ZF transcription factors in the human proteome. Many of these transcription factors play crucial roles in cancer and developmental processes. The purpose of this project is to identify lists of interactors of the BTB domains of six BTB-ZF proteins with high confidence using a mass spectrometry based approach. The BTB domains BCL6, PLZF, Kaiso, LRF, FAZF, and Miz1 were studied. This study was able to identify 142 putative interactors. The list of putative interactor proteins participates in a wide array of biological functions. Selected putative interactors of the BCL6BTB were also validated using biochemical techniques. In conclusion, this project was able to provide an analysis of the protein-protein interactions mediated by the BTB domains of six BTB-ZF transcription factors. The information generated is valuable to guide future functional and structural studies of the BTB domains.

Systematic identification of protein-protein interactions (PPIs) on a genome scale has become an important focus of biology, as the majority of cellular functions are mediated by these interactions. Several high throughput experimental techniques have emerged as effective tools for querying the protein-protein interactome and can be broadly categorized into those that detect direct, physical protein-protein interactions and those that yield information on the composition of protein complexes. Tandem affinity purification followed by mass spectrometry (TAP/MS) is an example of the latter that identifies proteins that co-purify with a given tagged query (bait) protein. Though TAP/MS enables these co-complexed associations to be identified on a proteome scale, the amount of data generated by the systematic querying of thousands of proteins can be extremely large. Data from multiple purifications are combined to form a very large network of proteins linked by edges whenever the corresponding pairs might form an association. Only a fraction of these pairwise associations correspond to physical interactions, however, and further computational analysis is necessary to filter out non-specific associations. This thesis examines how differing computational procedures for the analysis of TAP/MS data can affect the final PPI network, and outlines a procedure to accurately identify protein complexes from data consolidated from multiple proteome-scale TAP/MS experiments in the budding yeast \textit{Saccharomyces cerevisiae}. In collaboration with the Greenblatt and Emili laboratories at the University of Toronto, this methodology was extended to yeast membrane proteins to derive a comprehensive network of 13,343 PPIs and 720 protein complexes spanning both membrane and non-membrane proteins.

Down syndrome (DS), caused by an extra chromosome 21, affects 1 in 750 live births, and is characterized by cognitive impairment as well as several congenital defects. Currently, little is known about the molecular pathogenesis of DS and no direct genotype-phenotype relationship has yet been confirmed. The current screening test for DS subjects many women to undergo invasive procedures such as amniocentesis due to suboptimal sensitivity and specificity. Therefore, this study aimed to discover novel biomarkers to improve screening tests, and to discovery dysregulated molecular pathways in DS-affected fetus to better understand pathogenesis. To achieve this objective, proteomic analyses of amniotic fluid (AF) and amniotic fluid cells (amniocytes) were performed using mass spectrometry (MS), which allows discovery of a large number of proteins in complex biological samples. Since AF contains the most information of the developing fetus, we first generated the most comprehensive list of proteins present in AF by using high resolution MS. We then performed quantitative analyses of proteins from AF as well as amniocytes to reveal novel biomarkers and clues to altered molecular mechanisms of DS. Comparison between the proteome of AF from unaffected and DS-affected pregnancies allowed selection of 60 candidate biomarkers based on spectral counting. Two candidates, APP and TNC-C, were verified by immunoassays to show two-fold increase in AF from DS-pregnancies. Additionally, CPA4, MUC13, CEL, DPP4 and MMP2 were verified to be differentially expressed in trisomy 21-AF via selected reaction monitoring assays using triple-quadruple mass spectrometer. Amniocytes from DS-affected and unaffected fetuses were also quantitatively analyzed by using Stable Isotope Labelling of Amino acids in Cell culture technique. Over 4900 proteins were identified from amniocyte lysate and supernatant by LTQ-Orbitrap mass spectrometer, and 85% of these proteins were quantified based on MS/MS spectra ratios of peptides containing isotope-labelled amino acids. Proteins that consistently showed aberrant expression from affected amniocytes have been selected for further verification and molecular network analyses since they may play a role in DS pathogenesis.

Analysis of complex protein samples is demanding due to the wide dynamic range of expression levels and the limited detection range of technology. Proteomics relies heavily on the development of new fractionation strategies to help reduce complexity, and overcome the technological and biological challenges associated with proteome analysis. Here, the development of a prototype instrument named ??Microflow MF10?? was explored to enrich for particular classes of proteins. The MF10 was found to have a number of advantages over commercially available fractionation systems. Due to the reduced separation electrode distance, fractionation was rapid, occuring within ~0.125 kVH over 2-6 fractions under native conditions but longer under denaturing conditions. As low as 2 ng peptide could be fractionated with recovery for downstream analysis achievable. The ability to alter protein charge by changing the pH (acidic (pI 3.6) to basic environments (pI 10.4)) allows selection of proteins based on charge/mobility, size, shape, buffer ionic strength, pH and field strength. Proteins <10 kDa are also not routinely analysed because current technology is unable to cater for this region of the proteome. Peptide enrichment using the MF10 was achieved using a 7-protein/peptide standard mix (1-25 kDa), to the 1-5 kDa fraction with simultaneous fractionation of the higher mass protein standards. Plasma was also used to enrich for the peptidome (< 5 kDa) in the presence of the proteome. Enrichment of 73 proteins inclusive of 22 proteins in the 1-25 kDa fraction was achieved compared to a total equivalent of 42 proteins from unfractionated plasma. Rare samples (≤ 106 cells) from stem cell populations or derived clinically are challenging due to the absolute limits in protein copy number and abundance. CD34+ haematopoietic stem cells and CD4+/CD8+ T-cells were used to develop fractionation methods and elucidate the cell differentiation process. MF10 fractionation and analysis by SDS-PAGE and LC-MS/MS revealed 24 differentially expressed proteins between the 3 cell populations, which may be involved in cell differentiation. To quantify these expression differences, iTRAQ with 2-D LC-MS/MS was applied. This study has highlighted the challenges associated with samples of limited quantity. It has been successful in understanding the effects of various conditions on the electrophoretic mobility of proteins, which in proteomics, has remained largely unexplored.

Insulin stimulates glucose uptake in muscle and adipose tissue by translocating glucose transporter 4 (GLUT4) from an intracellular storage sige to the cellsurface. Studies have shown that insulin-activation of PI3K is required for stimulatin of GLUT4 translocation, however, platelet-derived growth factor (PDGF) also activates PI3K in adipocytes without significant effects on GLUT4 localisation. The aim of this study is to examine the hypothesis that insulin stimulates anunique signalling pathway that regulates GLUT4 translocation in adipocytes, and to identify novel signalling candidates in this pathway. A proteomic approach was employed to compare the effect of insulin versus PDGF on protein phosphorylation in 3T3-L1 adipocytes, using high-resolution two-dimensional gel electrophoresis (2-DE) in combination with 32P-metabolic labelling and autoradiography. The effect of wortmannin, an inhibitor of PI3K, on insulin-induced phosphorylation was also examined as a further screen for selection of candidate phosphoproteins for insulin-stimulation of GLUT4 translocation. Three subcellular Insulin stimulates glucose uptake in muscle and adipose tissue by translocating glucose transporter 4 (GLUT4) from an intracellular storage site to the cell surface. Studies have shown that insulin-activation of PI3K is required for stimulation of GLUT4 translocation, however, platelet-derived growth factor (PDGF) also activates PI3K in adipocytes without significant effects on GLUT4 localisation. The aim of this study is to examine the hypothesis that insulin stimulates an unique signalling pathway that regulates GLUT4 translocation in adipocytes, and to identify novel signalling candidates in this pathway. fractions were extensively examined, the plasma membrane (PM) fraction, the cytosol fraction and a high speed pellet (HSP) fraction that contains the insulin-responsive pool of GLUT4. These studies showed that phosphorylation of 18 proteins were consistently increased by insulin treatment, with 8 proteins in the cytosol, 10 proteins in the HSP and none in the PM. The phosphorylation of 6 cytosolic and 2 HSP insulin-stimulated phosphoproteins were also significantly increased by PDGF treatment. Wortmannin inhibited the insulin-stimulated phosphorylation of all but 2 proteins. Ten insulin-stimulated phosphoproteins were wortmannin-sensitive and PDGF-insensitive, and were selected as potential candidates for involvement in insulin-stimulated GLUT4 translocation. The molecular identity of insulin-stimulated phosphoproteins were initially examined by immunoblotting. Two cytosolic 2D gel spots (C65 and C79) were identified as mitogen-activated protein kinase (MAPK) isoforms by immunoblotting. Importantly, treatment-induced changes in the phosphorylation of these spots matched those reported for MAPK in adipocytes, confirming that our phosphoprotein mapping accurately reflects cellular phosphorylation. As the observed molecular mass and isoelectric point (pI) of most insulin-regulated phosphoproteins did not match known insulin-signalling proteins, we attempted to identify these by protein sequencing. In agreement with the low abundance of signalling proteins, most of the insulin-regulated phosphoprotein spots were not Coomassie-stained, even on preparative 2D gels. Thus, corresponding spots were pooled from numerous preparative 2D gels, before tryptic digestion and anlaysis by liquid chromatography-coupled tandem mass spectrometry (LC-MS). One of the more abundant candidate phosphoproteins was identified as ATP-citrate lyase (ACL), an enzyme involved in fatty acid synthesis previously shown to be an insulin-regulated phosphoprotein. Insulin treatment also results in the dephosphorylation of several phosphoproteins, one of which was identified as eukaryotic translation elongation factor 2 (eEF2). ACL and eEF2 phosphorylation were found to be regulated in an insulin-specific and wortmannin-sensitive manner in adipocytes. Recent studies suggest that protein kinase B (PKB) is activated downstream of PI3K, and participates in insulin-stimulated glucose uptake. However, the molecular mass and pI of phosphoproteins mapped in this study did not match any of the known PKB isoforms. To investigate the regulation of PKB isoform in adipocytes, studies were performed with PKB antibodies. PKBb expression was found to be induced upon differentiation of 3T3-L1 adipocytes, where insulin, but not PDGF stimulated its phosphorylation. In contrast, PKBa expression was high in fibroblasts, where PDGF was more efficacious than insulin in inducing its phosphorylation. These results suggest that PKBb, rather than PKBa, is the most pertinent isoform for insulin action in adipocytes. PDGF exerted no detectable effect on the phosphorylation of PKBb in 3T3-L1 adipocytes, as determined by phospho-specific antibodies, 32P-labelling and shifts in electrophoretic mobility and pI. Furthermore, insulin stimulated the translocation of PKBb to membrane fractions of adipocytes, whereas PDGF was without effect. Together, these results strongly suggest that PDGF does not activate PKBb in 3T3-L1 adipocytes. Using 2-DE, we have obtained data which support the hypothesis that insulin stimulates an unique signalling pathway in adipocytes. Three potential models are proposed to explain the insulin specificity: targeting of PI3K to pre-assembled signalling complexes in the HSP, activation of specific PI3K isoforms, and activation of two PI3K-dependent signalling pathways. These models remain to be validated in future studies. Candidate phosphoproteins for insulin regulation of glucose metabolism have been mapped using 2-DE. While only abundant metabolic enzymes have been identified so far, methods for phosphoprotein purification and identification have been established. The candidate phosphoproteins mapped in the HSP fraction are likely to be low abundance signalling proteins, and further preparative isolation of these phosphoproteins is required in order to obtain enough protein for identification of these candidates. In conclusion, the present study has demonstrated the successful application of proteomic techniques in cell signalling research.

Thesis (Ph. D.)--University of Hong Kong, 2007. / Also available online.

Thesis (M. Phil.)--University of Hong Kong, 2008. / Also available in print.

Thesis (Ph. D.)--University of Hong Kong, 2008. / Also available in print.

Thesis (Ph. D.)--Ohio State University, 2008.