Browsing by Subject "Proteomics"

Currently there is a dearth of analytical techniques for studying protein-ligand interactions on the proteomic scale. Existing techniques, which rely on various calorimetry or spectroscopy methods, are limited in their application to the proteomic scale due to their need for large amounts of pure protein. Recently, several mass spectrometry-based methods have been developed to study protein-ligand interactions. These mass spectrometry-based methods overcome some of the limitations of existing techniques by enabling the analysis of unpurified protein samples. However, the existing mass spectrometry-based methodologies for the analysis of protein-ligand binding interactions are not directly compatible with current mass spectrometry-based proteomics platforms.

Described here is the development and application of a new technique designed to detect and quantify protein-ligand binding interactions with mass spectrometry-based proteomic platforms. This technique, termed SPROX (Stability of Proteins from Rates of Oxidation), uses an irreversible covalent oxidation labeling reaction to monitor the global unfolding reactions of proteins to measure protein thermodynamic stability. Two variations of the SPROX technique are established here, including one variation that utilizes chemical denaturant to induce protein unfolding and a second variation that utilizes temperature to denature proteins. The SPROX methodology is tested on five proteins including ubiquitin, ribonuclease A, bovine carbonic anhydrase II, cyclophilin A, and calmodulin. Results obtained on these model systems are used to determine the method's ability to measure the thermodynamic parameters associated with each protein's folding/unfolding reaction. Results obtained on calmodulin and cyclophilin A are used to determine the method's ability to quantify the dissociation constants of protein-ligand complexes.

The primary motivation for the development of the SPROX protocols in this work was to create a protein-ligand binding assay that could be interfaced with conventional mass spectrometry-based platforms. Two specific SPROX protocols, including a label-free approach and an oxygen-16/18 labeling approach, are developed and demonstrated using the thermal SPROX technique to analyze ligand binding in a model four-protein component mixture consisting of ubiquitin, ribonuclease A, bovine carbonic anhydrase, and cyclophilin A. The thermal SPROX technique's ability to detect cyclosporin A binding to cyclophilin A in the context of the model mixture is shown using both labeling approaches.

An application using the SPROX technique combined with a multi-dimensional protein identification technology (MudPIT)-based proteomics platform is also described. In this application, which utilized an isobaric mass tagging strategy, 325 proteins in a yeast cell lysate are simultaneously assayed for CsA-binding. This study was also used to investigate the protein targets of an already well-studied immunosuppressive drug, cyclosporin A. Two of the ten protein targets identified in this work are known to interact with CsA, one through a direct binding event and one through an indirect binding event. The eight newly discovered protein targets of CsA suggest a molecular basis for post-transplant diabetes mellitus, which is a side effect of CsA in humans.

Recently, several mass spectrometry-based proteomics techniques have been developed for the large-scale analysis of thermodynamic measurements of protein stability. This has created the possibility of characterizing disease states via differential thermodynamic stability profiles. Described here is the application of the Stability of Proteins from Rates of Oxidation (SPROX) technique to characterize mouse models of disease. The mouse models studied here are of normal aging and two genetically induced Parkinson’s Disease (PD) models.

Thermodynamic stability profiles were generated for 809 proteins in brain cell lysates from C57BL/6 mice at age 6- (n=7) and 18-months (n=9). The biological variability of the protein stability measurements was low, and within the experimental error of the SPROX technique. Remarkably, the large majority of the 83 brain protein hits were destabilized in the old mice, and the hits were enriched in proteins that have slow turnover rates (p<0.07). Furthermore, 70% of the hits have been previously linked to aging or age-related disease.

One of the PD mouse models involved characterizing the protein interactions induced by mutated leuceine-rich repeat kinase 2 (LRRK2) at a pre-symptomatic time point (3 months old). The models used were a control, overexpressed wildtype LRRK2, and overexpressed R1441G mutated LRRK2 (n=2 for all models). Comparative analyses on thermodynamic stability profiles of ~470 proteins revealed relatively few differences. In fact, the observed hit rate in each comparative analysis was close to that associated with the biological variability of the mice. However, four protein hits, dihydropyrimidinase-related protein 2, eukaryotic translation initiation factor 4A2, Rap1 GTP-GDP dissociation stimulator 1 and myelin basic protein, were identified with consistent thermodynamic stability in multiple mice within a biological state and as hits in multiple comparisons suggesting they are the most likely to be true positives.

The second PD mouse model studied was one in which the human α-synuclein protein, containing the known PD mutation A53T, was overexpressed. To characterize the disease progression of PD induced by this mutation, mice were sacrificed at 1 month (n=4), 6 months (n=4) and when they became symptomatic at 10-16 months (n=3). Thermodynamic stability profiles were generated for >850 proteins at each time point. The relative stabilities of these proteins were assayed in a series of comparative analyses involving mice at the different time points and the normally aged mice from above. In total 244 peptides were found to be differentially stabilized during PD progression. A subset of 52 peptide hits was identified to be of particular interest. Of these 52 peptides 22 were identified with early disease progression, 5 peptides showed late disease progression, 5 peptides reported a gradual difference in stability over disease progression and 20 peptides indicated no disease progression trend. More than 90% of the 32 peptides indicating a trend in disease progression showed progression related destabilization.

The results of this thesis help validate the use of thermodynamic stability measurements to capture disease-related proteomic differences in mice. Furthermore, these results establish a new biophysical link between the hit proteins identified and their role in aging, LRRK2 protein interactions, and PD progression.

Background

Methods

Results

Conclusions

Radial glial cells (RGCs) are essential for the generation and organization of neurons in the cerebral cortex. RGCs have an elongated bipolar morphology with basal and apical endfeet that reside in distinct niches. Yet, how this subcellular compartmentalization of RGCs controls cortical development is largely unknown. Here, we employ in vivo proximity labeling, in the mouse, using unfused BirA to generate the first subcellular proteome of RGCs and uncover new principles governing local control of cortical development. We discover a cohort of proteins that are significantly enriched in RGC basal endfeet, with MYH9 and MYH10 among the most abundant. Myh9 and Myh10 transcripts also localize to endfeet with distinct temporal dynamics. Although they each encode isoforms of non-muscle myosin II heavy chain, Myh9 and Myh10 have drastically different requirements for RGC integrity. Myh9 loss from RGCs decreases branching complexity and causes endfoot protrusion through the basement membrane. In contrast, Myh10 controls endfoot adhesion, as mutants have unattached apical and basal endfeet. Finally, we show that Myh9- and Myh10-mediated regulation of RGC complexity and endfoot position non-cell autonomously controls interneuron number and organization in the marginal zone. The first part of this study demonstrates the utility of in vivo proximity labeling for dissecting local control of complex systems, and reveals new mechanisms for dictating RGC integrity and cortical architecture. In the second portion of this work, we have developed a method for purification of endfeet from the embryonic mouse brain and employed it to discover the first global transcriptome of RGC endfeet. Analysis at E15.5 revealed that the network of localized mRNAs is much more extensive than previously appreciated. There are over 3,000 transcripts localized to RGC endfeet and 870 of them are highly enriched in the endfeet compared to the cell body. These data uncovered hundreds of new genes in endfeet and also reinforced our previous findings that cytoskeletal regulators and ECM components are especially important in endfeet. Exploration of the newly discovered localized transcripts will provide valuable insights into additional RGC functions and allow us to assess potential signaling interactions between endfeet and surrounding cells. We also propose a method for subcellular gene knockdown in which we can modulate mRNA levels of a gene of interest in the cell body and endfeet independently in vivo. Through these studies we have discovered vital roles for subcellular gene regulation in RGCs and developed tools to facilitate future studies.

The advancement of mass spectrometry-based protein stability profiling measurements within the past twenty years has led to the development of a suite of approaches that enables the evaluation of protein folding stability on a broad range of biological mixtures with varying complexity. These approaches include chemical and thermal denaturation techniques (SPROX and TPP, respectively) as well as proteolysis strategies such as limited proteolysis (LiP) and pulse proteolysis (PP) which have all been extensively used and evaluated for small molecule protein target discovery applications. However, the capabilities of these methods have yet to be fully evaluated in the characterization of disease phenotypes and other biological events such as post-translational modifications and RNA-protein interactions. A major focus of the work included in this dissertation has been the comparative analysis of the above techniques for the characterization of biological phenotypes. The application and comparative analysis of the above techniques to the characterization of RNA-protein interactions is also described.

In this work, I present four studies across the range of 'omics data types - a Genome- Wide Association Study for gene-by-sex interaction of obesity traits, computational models for transcription start site classification, an assessment of reference-based mapping methods for RNA-Seq data from non-model organisms, and a statistical model for open-platform proteomics data alignment.

Obesity is an increasingly prevalent and severe health concern with a substantial heritable component, and marked sex differences. We sought to determine if the effect of genetic variants also differed by sex by performing a genome-wide association study modeling the effect of genotype-by-sex interaction on obesity phenotypes. Genotype data from individuals in the Framingham Heart Study Offspring cohort were analyzed across five exams. Although no variants showed genome-wide significant gene-by-sex interaction in any individual exam, four polymorphisms displayed a consistent BMI association (P-values .00186 to .00010) across all five exams. These variants were clustered downstream of LYPLAL1, which encodes a lipase/esterase expressed in adipose tissue, a locus previously identified as having sex-specific effects on central obesity. Primary effects in males were in the opposite direction as females and were replicated in Framingham Generation 3. Our data support a sex-influenced association between genetic variation at the LYPLAL1 locus and obesity-related traits.

The application of deep sequencing to map 5' capped transcripts has confirmed the existence of at least two distinct promoter classes in metazoans: focused promot- ers with transcription start sites (TSSs) that occur in a narrowly defined genomic span and dispersed promoters with TSSs that are spread over a larger window. Pre- vious studies have explored the presence of genomic features, such as CpG islands and sequence motifs, in these promoter classes, and our collaborators recently inves- tigated the relationship with chromatin features. It was found that promoter classes are significantly differentiated by nucleosome organization and chromatin structure. Here, we present computational models supporting the stronger contribution of chro- matin features to the definition of dispersed promoters compared to focused start sites. Specifically, dispersed promoters display enrichment for well-positioned nucleosomes downstream of the TSS and a more clearly defined nucleosome free region upstream, while focused promoters have a less organized nucleosome structure, yet higher presence of RNA polymerase II. These differences extend to histone vari- ants (H2A.Z) and marks (H3K4 methylation), as well as insulator binding (such as CTCF), independent of the expression levels of affected genes.

The application of next-generation sequencing technology to gene expression quantification analysis, namely, RNA-Sequencing, has transformed the way in which gene expression studies are conducted and analyzed. These advances are of partic- ular interest to researchers studying non-model organisms, as the need for knowl- edge of sequence information is overcome. De novo assembly methods have gained widespread acceptance in the RNA-Seq community for non-model organisms with no true reference genome or transcriptome. While such methods have tremendous utility, computational complexity is still a significant challenge for organisms with large and complex genomes. Here we present a comparison of four reference-based mapping methods for non-human primate data. We explore mapping efficacy, correlation between computed expression values, and utility for differential expression analyses. We show that reference-based mapping methods indeed have utility in RNA-Seq analysis of mammalian data with no true reference, and that the details of mapping methods should be carefully considered when doing so. We find that shorter seed sequences, allowance of mismatches, and allowance of gapped alignments, in addition to splice junction gaps result in more sensitive alignments of non-human primate RNA-Seq data.

Open-platform proteomics experiments seek to quantify and identify the proteins present in biological samples. Much like differential gene expression analyses, it is often of interest to determine how protein abundance differs in various physiological conditions. Label free LC-MS/MS enables the rapid measurement of thousands of proteins, providing a wealth of peptide intensity information for differential analysis. However, the processing of raw proteomics data poses significant challenges that must be overcome prior to analysis. We specifically address the matching of peptide measurements across samples - an essential pre-processing step in every proteomics experiment. Presented here is a novel method for open-platform proteomics data alignment with the ability to incorporate previously unused aspects of the data, particularly ion mobility drift times and product ion data. Our results suggest that the inclusion of additional data results in higher numbers of more confident matches, without increasing the number of mismatches. We also show that the incorporation of product ion data can improve results dramatically. Based on these results, we argue that the incorporation of ion mobility drift times and product ion information are worthy pursuits. In addition, alignment methods should be flexible enough to utilize all available data, particularly with recent advancements in experimental separation methods. The addition of drift times and/or high energy to alignment methods and accurate mass and time (AMT) tag databases can greatly improve experimenters ability to identify measured peptides, reducing analysis costs and potentially the need to run additional experiments.

Synapses – the delicate connections between our neurons – adjust and refine their strength to shape our brains, our thoughts, and our memories. Proteomic and genetic techniques have revealed that this process, known as synaptic plasticity, is tightly controlled by signaling cascades that ultimately expand or contract actin networks within postsynaptic sites. In this dissertation, I advance the field of synaptic plasticity by focusing on presynaptic terminals, which are equal partners with their postsynaptic counterparts. To date, the study of presynaptic plasticity has been difficult due to the limited number of presynaptic signaling molecules currently identified (particularly those regulating the cytoskeleton), as well as the lack of tools to manipulate these molecules specifically within presynaptic terminals. I therefore developed new experimental approaches to tackle both of these hurdles. After mapping presynaptic cytoskeletal signaling pathways in the mouse brain, I discovered a new mechanism of presynaptic plasticity that is driven by action potential-coupled actin remodeling.

Presynaptic terminals cannot be biochemically purified away from postsynaptic sites. This has restricted previous presynaptic proteomic studies to isolated synaptic vesicles or other fractions, which have only identified a few actin signaling molecules. I thus turned to a new proteomic method called in vivo BioID. This approach is based on proximity-based biotinylation, which labels proteins in a compartment of interest as defined by a bait protein. My choice of presynaptic bait worked beautifully, leading to the mass spectrometry-based identification of 54 cytoskeletal regulators, most of which were previously not known to be presynaptic. The networks of presynaptic actin signaling molecules turn out to be just as richly diverse as those of the postsynapse. Many proteins also converge on a Rac1-Arp2/3 signaling pathway that leads to the de novo nucleation of branched actin filaments. This reveals that the presynaptic cytoskeleton consists of a dynamic, branched actin network.

This finding was unexpected because Rac1 and Arp2/3 have long-established roles in the development and plasticity of the postsynapse. This also makes it difficult to isolate the presynaptic functions of these proteins. I thus created optogenetic tools and electrophysiological strategies to acutely and bidirectionally manipulate their activity specifically within presynaptic terminals. I showed that presynaptic Rac1 and Arp2/3 negatively regulate the recycling of synaptic vesicles, thereby driving a form of plasticity known as short-term depression. I also showed that this mechanism is conserved between excitatory and inhibitory synapses, demonstrating it is a fundamental aspect of presynaptic function. Finally, I conducted a series of experiments using two-photon fluorescence lifetime imaging (2pFLIM) with a FRET-based biosensor of Rac1 activity. I discovered that calcium entry during action potential firing activates Rac1 within presynaptic terminals. This establishes a new mechanism of short-term depression that is driven by an action potential-coupled signal to the presynaptic cytoskeleton.

This dissertation thus combines proteomics, optogenetics, electrophysiology, and 2pFLIM-FRET to gain new insights into presynaptic plasticity. These findings have three significant implications. First, they challenge the prevailing view that the Rac1-Arp2/3 pathway functions largely at excitatory postsynaptic sites. This compels re-evaluation of how mutations in Rac1 and Arp2/3 cause neurological diseases such as intellectual disability and schizophrenia. Second, the genetic and optogenetic tools I developed are the first way to specifically modulate short-term depression, finally allowing for the exact functions of this form of plasticity to be determined in vivo. This has particular relevance for working memory, which has been theorized to be controlled by short-term presynaptic plasticity. Finally, this study provides a proteomic framework and blueprint of experimental strategies to conduct a systematic genetic analysis of the presynaptic cytoskeleton, which may finally unify the controversial theories about presynaptic actin function. In sum, the experimental strategies and resources that I developed highlight the multifaceted, sophisticated signaling that occurs in presynaptic terminals. This may yet shed light on how we remember our experiences, and why we are who we are.

In the past decade, several mass spectrometry-based proteomic techniques have been developed for the large-scale analysis of protein folding stabilities. The main focus of this dissertation is to develop and apply these large-scale protein folding stability approaches in drug target identification and disease biomarker discovery. One goal of this work is to develop a novel chemo-selection strategy to improve the bottom-up proteomics readout in proteome-wide limited proteolysis experiments. Another goal of this work is to apply these methods to the target identification of two drugs with known mode of action, and to the biomarker discovery of Parkinson’s disease.

Described in the first part of the dissertation is the development of a chemo-selective enrichment strategy to isolate the semi-tryptic peptides generated in mass spectrometry-based applications of limited proteolysis methods. The method is termed Semi-Tryptic Peptide Enrichment Strategy for Proteolysis Procedures (STEPP). The STEPP-PP workflow was evaluated in two proof-of-principle drug target identification experiments involving two well-studied drugs, cyclosporin A and geldanamycin. The STEPP-LiP workflow was evaluated in one proof-of-principle experiment on identification of protein conformational changes between a breast cancer cell line, MCF-7, and a normal cell line, MCF-10A. The STEPP protocol increased the number of semitryptic peptides detected in the LiP and PP experiments by 5- to 10-fold. The STEPP protocol not only increases the proteomic coverage, but also increases the amount of structural information that can be gleaned from limited proteolysis experiments. Moreover, the protocol also enables the quantitative determination of ligand binding affinities. When coupled to a one-pot data acquisition strategy, the one-pot STEPP-PP technique was found to have a very low false positive rate (i.e., 0.09%) in a proof-of-principle drug target identification experiments involving cyclosporin A and a yeast lysate.

The second part of this dissertation describes the application of protein folding stability approaches to the identification of protein targets of subglutinol A (a natural immunosuppressant) and manassantin A (a natural product with anti-cancer activity).

In the suglutinol A study, a combination of SPROX, TPP, CPP and STEPP-PP strategies was used to identified two consistent protein hits, deoxycytidine kinase (DCK) and exportin-2 (XPO2), from more than 2000 assayed proteins in a 2B4T cell lysate. The binding of DCK with subglutinol A was validated using a targeted gel-based pulse proteolysis experiment. A set of chemical biology experiments were performed to uncover the relation of this interaction with subglutinol A’s mode of action. It was shown that neither of the kinase activity, expression level or phosphorylation modification level of DCK was alternated by subglutinol A. However, the nuclear transportation of DCK was blocked by subgltutinol A. This reduction of DCK level in the cell nucleus possibly leads to the observed reduction of nuclear dCMP pool and the halted proliferation of sublgutinol A treated T cells.

In the manassantin A study, a combination of STEPP-LiP, STEPP-PP, one-pot STEPP-PP, one-pot SPROX and one-pot TPP strategies were performed to identify the protein target of the drug in a hypoxia-treated HEK293T cell lysate. These experiments assayed over 4000 proteins and found 4 protein hits for further validation of their interaction with manassantin A.

The third part of describes the utilization of the SPROX method to characterize the progression of PD in a mouse model of the disease in which the human α-synuclein protein with an A53T mutation was overexpressed. The thermodynamic stabilities of proteins in brain tissue cell lysates from Huα-Syn(A53T) transgenic mice were profiled at three time points including at 1 Month (n=9), at 6 Months (n=7), and at the time (between 9 and 16 Months) a mouse became Symptomatic (n=8). The thermodynamic stability profiles generated here on over 300 proteins were compared to the thermodynamic stability profiles generated on the same proteins from similarly aged wild-type mice using a two-way ANOVA analysis. A group of 22 proteins were identified with age-related protein stability changes, and a group of 11 proteins were found to be differential stabilized in the Huα-Syn(A53T) transgenic mouse model. The proteins differentially stabilized in the disease mouse model could potentially be used as Parkinson’s disease biomarkers upon further validation.

The identification and quantification of protein-protein interactions in large scale is critical to understanding biological processes at a systems level. Current approaches for the analysis of protein -protein interactions are generally not quantitative and largely limited to certain types of interactions such as binary and strong binding interactions. They also have high false-positive and false-negative rates. Described here is the development of and application of mass spectrometry-based proteomics metehods to detect and quantify the strength of protein-protein and protein-ligand interactions in the context of their interaction networks. Characterization of protein-protein and protein-ligand interactions can directly benefit diseased state analyses and drug discovery efforts.

The methodologies and protocols developed and applied in this work are all related to the Stability of Unpurified Proteins from Rates of amide H/D Exchange (SUPREX) and Stability of Protein from Rates of Oxidation (SPROX) techniques, which have been previously established for the thermodynamic analysis of protein folding reactions and protein-ligand binding interactions. The work in this thesis is comprised of four parts. Part I involves the development of a Histidine Slow H/D exchange protocol to facility SURPEX-like measurements on the proteomic scale. The Histidine Slow H/D exchange protocol is developed in the context of selected model protein systems and used to investigate the thermodynamic properties of proteins in a yeast cell lysate.

In Part II an isobaric mass tagging strategy is used in combination with SPROX (i.e., a so-called iTRAQ-SPROX protocol) is used to characterize the altered protein interactions networks associated with lung cancer. This work involved differential thermodynamic analyses on the proteins in two different cell lines, including ADLC-5M2 and ADLC-5M2-C2.

Parts III and IV of this thesis describe the development and application of a SPROX protocol for proteome-wide thermodynamic analyses that involves the use of Stable Isotope Labeling by Amino acid in cell Culture (SILAC) quantitation. A solution-based SILAC-SPROX protocol is described in Part III and a SILAC-SPROX protocol involving the use of cyanogen bromide and a gel-based fractionation step is described in Part IV. The SILAC-SPROX-Cyanogen bromide (SILAC-SPROX-CnBr) protocol is demonstrated to significantly improve the peptide and protein coverage in proteome-wide SPROX experiments. Both the SILAC-SPROX and SILAC-SPROX-CnBr porotocols were used to characterize the ATP binding properties of yeast proteins. Ultimately, the two protocols enabled 526 yeast proteins to be assayed for binding to AMP-PNP, an ATP mimic. A total of 140 proteins, including 37 known ATP-binding proteins, were found to have ATP binding interactions.

Protein stability profiling techniques, including stability of proteins from rates of oxidation (SPROX), thermal proteome profiling (TPP), and limited proteolysis (LiP), have created a new toolbox for making protein folding stability measurements on the proteomic scale. Despite the broad usage of these techniques in an increasingly large number of applications, few studies have been reported in which more than one of these techniques have been used for the same application. Therefore, detailed information about the complementarity of these different experimental approaches is lacking. A major focus of the work described in this thesis is the comparative analysis of these different experimental strategies in different application areas. Described in Chapter 2 of this dissertation is a new experimental workflow for multiplexing TPP analyses that is also amenable to SPROX analyses. In a proof-of-principle study, this new workflow (termed, "one-pot" workflow) was demonstrated to significantly reduce the instrument time and cost of materials by 10-fold, which consequentially enabled enhanced multiplexing compared to traditional protein stability profiling analysis. This "one-pot" workflow, which is also amenable to SPROX analysis, makes feasible the acquisition of protein stability profiling data from multiple biological replicates and even multiple techniques (e.g., SPROX and TPP) in a given application.

Described in Chapters 3, 4, and 5 of this dissertation are applications of SPROX, TPP, and LiP to a series of protein target discovery projects involving several different small molecule drugs with known biological activities but unknown modes of action. These applications included the use of multiple energetics-based techniques and in some cases included the "one-pot" workflow, to identify the protein targets of the bioactive small molecules. These results of these studies included: i) identification of TCP-1 as the anti-malarial target of clemastine in a Plasmodium falciparum (Pf) cell lysate; ii) identification of potential Pf protein targets of several additional anti-malarial drugs; iii) identification of Rab-1A as the anti-Parkinson target of NAB1; iv) identification of potential protein targets of anti-salmonella drugs in a complex host-pathogen system; and v) identification of the interactome of CoA and de-phosphorylated CoA in mammalian cells. The results of these protein target discovery studies not only helped elucidate the molecular basis of the biological activities of the above small molecules, but the results also demonstrated the power of using multiple protein stability profiling techniques to deconvolute protein targets and identify small molecule interactomes.

The final chapter of this dissertation describes the use of protein stability profiling methods, including the "one-pot" SPROX and TPP workflows as well as LiP, to identify potential prognostic biomarkers of oxaliplatin resistance in colorectal cancer. The goal of this work was to identify the overlapping protein hits identified in SPROX, TPP, and LiP analyses of oxaliplatin resistant and sensitive colorectal cancer cell lines (n=3 cell line comparisons) and patient-derived xenograft mouse models of oxaliplatin resistance and sensitivity (n= 1 mouse model comparison). Ultimately, 23 differentially stabilized proteins were identified with differential stability in at least 2 techniques and consistent results in both cell culture models and PDX mouse models. These 23 proteins included 9 proteins being previously connected to cancer chemo-resistance. One of these 9 differentially stabilized proteins, major vault protein (MVP), was further investigated to elucidate its role on colorectal cancer chemoresistance. This research demonstrated that complementary protein stability profiling is a complementary approach to identify potential prognostic biomarkers and shed light on understanding the molecular mechanism of colorectal cancer chemoresistance.

In conclusion, the work in this dissertation described an integrated workflow for energetics-based proteomic techniques that can be used for target deconvolution and biomarker identification in complex biological systems.

The characterization of protein stability changes and protein-ligand interactions on the proteomic scale is important for understanding the biology of cellular processes. The identification and quantification of protein-ligand binding affinities is critical for disease state analyses and drug discovery. A mass spectrometry-based technique, Stability of Proteins from Rates of Oxidation (SPROX), has been established for the thermodynamic analysis of protein stability and protein-ligand interactions. In the first part of this dissertation, a previously published iTRAQ-SPROX protocol is improved by incorporating a filter assisted sample preparation (FASP) protocol to significantly reduce sample loss during the experiment. Also, in order to eliminate methionine as a potential contaminant that can cause signal suppression during LC-MS/MS analysis, TCEP•HCl is used to quench the H2O2 oxidation instead of methionine. This avoids the potential reaction between the free methionine and the iTRAQ reagents. The improved protocol, referred to hereafter as the iTRAQ-FASP-SPROX protocol, is shown to increase the peptide/protein coverages for less concentrated cell lysate samples, and it is applied here to study the protein-ligand interaction networks between human ARPE-19 cells lysis and two different iron chelators (HAPI and Exjade). Information on potential protein targets of these two iron chelators are reported.

In the second part of this dissertation, a targeted MS-based approach for protein-ligand binding analysis is developed to analyze targeted subsets of proteins in a proteome. The so-called PAB-SPROX protocol is demonstrated to be applicable for the detection and relative quantitation of targeted methionine-containing peptides in +/- ligand samples by using isotopically labeled light and heavy PAB (i.e. 12C6-PAB and 13C6-PAB). Multiple reaction monitoring (MRM) and parallel reaction monitoring (PRM) methods are demonstrated to be amenable to PAB-SPROX analyses. In addition to proof-of-principle studies involving the cylclophilin A-cyclosporine A binding interaction, the PAB-SPROX protocol was used to validate the direct interaction between YBX1 protein and tamoxifen using very limited amount of purified YBX1 protein. Applications of PAB-SPROX protocol have also included the validation of potential binding targets of Staurosporine, Manassatin A and Tamoxifen. The PAB-SPROX studies with these latter ligands facilitated the identification of false positives in previous proteome-wide SRPOX studies.

Protein-ligand interactions can be detected and quantified using protein folding stability measurements. Thus, protein folding stability changes are closely linked to protein function and are an important biophysical measurement. Stability of Proteins from Rates of Oxidation (SPROX) is one approach for making proteome wide stability measurements on proteins. Previous SPROX data analysis strategies relied heavily on visual inspection of the data to identify protein stability changes. This created a bottleneck in the analysis and chances for human error. As part of this work, several new data analysis strategies were evaluated using data from previously reported ligand-binding studies. One strategy, the so-called difference analysis, was determined to be the data analysis strategy of choice as it minimized false positive and negatives. The difference analysis strategy was applied to identify protein targets of the breast cancer therapeutic tamoxifen (TAM) and its active metabolite 4-hydroxytamoxifen (4OHT) in yeast and protein targets of TAM and its most abundant metabolite n-desmethyl tamoxifen (NDT) in MCF-7 cell lysate. Several yeast protein targets of TAM and 4OHT were identified using SPROX, and they were subsequently validated using a pulse proteolysis strategy. The proteins in an MCF-7 cell line were probed for TAM- and NDT-induced stability changes using the SPROX in combination with two different quantitative proteomics strategies. Together the SPROX experiments on the proteins in an MCF-7 cell lysate enabled over 1000 proteins to be assayed for TAM- and NDT- induced protein stability changes. Ultimately, a total of 163 and 200 proteins with TAM- and NDT-induced stability changes were identified, respectively. A subset of 33 high confidence hits, including those identified using both proteomics strategies or those identified with multiples peptide probes, were assessed for experimental links to the ER using a STRING analysis. One high confidence protein hit, Y-box binding protein 1 (YBX1), was recently shown to bind the estrogen receptor, which is the known target of TAM. Preliminary results generated here using pulse proteolysis and a purified recombinant YBX1 protein construct suggest that YBX1 is a direct protein target of TAM. Proteins with altered expression levels with TAM and NDT treatment were also identified here. In total, 799 and 671 proteins were probed for TAM- and NDT- induced expression changes, respectively, and 49 and 30 proteins had altered expression. Out of the 49 and 30 proteins with TAM- and NDT- induced expression changes, 14 and 4 proteins had TAM- and NDT- induced altered stability, respectively. In addition to the above ligand-binding studies, SPROX was utilized to characterize the stability of the allergen containing proteomes of the European house dust mite, timothy grass pollen, and ragweed pollen. It was determined that the protein allergens in these proteomes were more stable and more abundant (based on transcriptomic data), than non-allergen proteins from these sources.

Articular cartilage is a highly specialized avascular tissue and consists of chondrocytes and two major components, a collagen-rich framework and highly abundant proteoglycans. The chondrocyte morphology and extracellular matrix properties vary with the depth of cartilage. Some past studies have defined the zonal distribution of a broad range of cartilage proteins in different layers. Based on the variations within each layer, the extracellular matrix can be further distinguished to pericellular, territorial and interterritorial regions. However, most of these studies used guanidine-HCl extraction that leaves an unextracted residual with a substantial amount of collagen. The high abundance of anionic polysaccharide molecules from cartilage adversely affects the chromatographic separation. Scatter oriented chondrocytes only account for the small proportion of the whole tissue protein extraction. However, the density of the cell varies with depth of cartilage as well. Moreover, the physiological status may also altered the extracellular matrix properties. Therefore, a comprehensive strategy to solve all these difficulties are necessary to elucidate the molecular structure of cartilage.

In this study, we used quantitative and qualitative proteomic analysis to investigate various cartilage tissue processing protocols. We established a method for removing chondrocytes from cartilage sections that minimized matrix protein loss. Quantitative and qualitative proteomic analyses were used to evaluate different cartilage extraction methodologies. The addition of surfactant to guanidine-HCl extraction buffer improved protein solubility. Ultrafiltration removed interference from polysaccharides and salts. The different extraction methods yielded different protein profiles. For instance, an overwhelming number of collagen peptides were extracted by the in situ trypsin digestion method. However, as expected, proteoglycans were more abundant within the guanidine-HCl extraction.

Subsequently we applied these methods to extract cartilage sections from different cartilage layers (superficial, intermediate and deep), joint types (knee and hip), and disease states (healthy and osteoarthritic). We also utilized lase capture microscopy (LCM) to harvest cartilage sample from individual subregions (territorial and interterritorial regions). The results suggested that there is more unique proteins existed in the superficial layer. By removing the chondrocytes, we were able to identify more extracellular matrix proteins. The phenotyping of cartilage subregions provided the chance to precisely localize the protein distribution, such as clusterin protein. We observed that the guanidine-HCl extractability (guanidine-HCl/ guanidine-HCl + in situ digestion extracts) of cartilage proteins. Proteoglycans showed high extractability while collagen and non-collagenous proteins had lower extractability. We also observed that the extractability might differ with depth of cartilage and also disease states might alter the characters as well.

Laser capture microscopy provides us the access to the cartilage subregions in which only few studies have investigated because of the difficulties to separate them. We established the proteomic analysis compatible-protocol to prepare the cartilage section for LCM application. The results showed that most of the proteoglycans and other proteins were enriched in the interterritorial regions. Type III and VI collagens, and fibrillin-1 were enriched in the territorial regions. We demonstrated that this distribution difference also varied with depth of cartilage. The difference of protein abundance between subregions might be altered because of disease states.

Last we were looking for the post-transliational modification existed in these subregions of cartilage. Deamidation is one of the modification without the enzyme involved. Previous studies have showed that deamidation may accumulated in the tissue with low turnover rate. Our proteomic analysis results suggests that abundance of deamidated peptides also varied in different layers and subregions of cartilage.

We have developed the monoclonal antibody based immunoassay to quantify the deamidated cartilage oligomeric matrix protein within cartilage tissue from different joints (hip and knee) and disease states (healthy, para-lesion, and remote lesion). The results suggests that the highest concentration of deamidated COMP was identified in arthritic hip cartilage.

The results of this study generated several reliable protocols to perform cartilage matrix proteomic analysis and provided data on the cartilage matrix proteome, without confounding by intracellular proteins and an overwhelming abundance of collagens. The discovery results elucidated the molecular architecture of cartilage tissue at different joint sites and disease states. The similarities among these cartilages suggested a constitutive role of some proteins such as collagen, prolargin, biglycan and decorin. Differences in abundance or distribution patterns, for other proteins such as for cartilage oligomaric matrix protein, aggrecan and hyaluronan and proteoglycan link protein, point to intriguing biological difference by joint site and disease state. Decellularization and a combination of extraction methodologies provides a holistic approach in characterizing the cartilage extracellular matrix. Guanidine-HCl extractability is an important marker to characterize the statue of cartilage; however it has not been fully understand. The protein distributions in matrix subregions may also serve as an index to characterize the metabolic status of cartilage in different disease states. A large sample cohort will be necessary to elucidate these characters.