acase@tulane.edu / Osteoporosis is a prevalent bone metabolic disease characterized by bone fragility. As a key pathophysiological mechanism, the disease is caused by excessive bone resorption (by osteoclasts) over bone formation (by osteoblasts). Peripheral blood monocytes (PBMs) represent a major systemic cell type for bone metabolism by serving as progenitors of osteoclasts and producing cytokines important for osteoclastogenesis. Our lab previously used microarray-based transcriptomics profiling to identify a list of novel genes for osteoporosis. My work is to further investigate the factors and regulatory network in osteoporosis, using microarray data of monocytes from subjects with extremely high/low hip bone mineral density. 1) We performed a pathway analysis and developed a novel approach to correct the “crosstalk” phenomenon which is caused by overlapping genes. 2) We analyzed the long non-coding RNA (lncRNA) profile by re-annotating exon array and predicted the regulatory mechanism of lncRNAs on protein coding genes in bone metabolism. 3) We identified the important potential transcription factors for osteoporosis and inferred the regulatory mechanism which exists between transcription factors and target genes in bone metabolism. My findings not only reported the key regulatory factors (lncRNAs and transcriptional factors) contributes to bone metabolism, but also explored the potential regulatory networks in osteoporosis. / 1 / Yu Zhou

<p>HIV-1 relies on both viral and cellular host factors for expression of its genome. Tat specific factor 1 (Tat-SF1) was identified as a cellular cofactor required for enhanced transcription of HIV-1 <italic>in vitro</italic>. Insight into the role of Tat-SF1 in the HIV-1 lifecycle has previously been limited to immunodepletions and <italic>in vitro</italic> analyses or transient overexpression experiments. Here, we present studies that utilize RNA interference (RNAi) to reevaluate Tat-SF1's role in Tat transactivation and HIV-1 replication <italic>in vivo</italic>. We report that although Tat-SF1 depletion reduces HIV-1 infectivity, it does not affect Tat transactivation <italic>in vivo</italic>. However, Tat-SF1 depletion changes the levels of unspliced and spliced RNAs. We propose that Tat-SF1 has a novel role of post-transcriptionally regulating HIV-1 gene expression, possibly through alternative splicing.</p><p>The functions of Tat-SF1 in cellular gene expression are not well understood, so we utilized the stable cell lines constructed for our HIV-1 studies to investigate the cellular functions of Tat-SF1. To identify target genes of Tat-SF1, we employed a combination of RNAi and human exon arrays. These arrays, which survey both transcript-level and exon-level changes genome-wide, revealed approximately 1,400 genes with alternative exon usage after Tat-SF1 depletion (p&le;0.01). In contrast, 500 genes showed significant transcript-level changes (p&le;0.01), all with minimal fold changes. Computational analyses showed that genes with alternative exon usage after Tat-SF1 depletion were over-represented in the insulin signaling and ubiquitin mediated proteolysis biological pathways. Furthermore, there was approximately 2-fold enrichment of Tat-SF1 target genes among previously reported HIV-1 dependency factors. The type of exon choice affected by Tat-SF1 depletion exhibited a strong 5&rsquo; bias. Finally, a novel Tat-SF1 binding motif, GACGGG, was found to be over-represented among target genes and may play a functional role in first exon choice. Together, these data are the strongest evidence to date of Tat-SF1 functioning in both transcription and splicing of cellular genes.</p> / Dissertation

Alternatives Spleißen ist ein wesentlicher Mechanismus, um Proteindiversität in Eukaryoten zu gewährleisten. Gewebespezifität sowie entwicklungsrelevante Prozesse werden unter anderem massgeblich davon beeinflusst. Aberrante (alternative) Spleißvorgänge können wiederum zu veränderten Proteinisoformen führen, die verschiedenste Krankheiten wie Krebs verursachen oder zu veränderter Medikamentenwirksamkeit beitragen können. In dieser Arbeit untersuchen wir differentielles Spleißen im Kontext von Krebserkrankungen. Dazu betrachten wir drei Aspekte, die uns wichtig erscheinen. Der erste Teil dieser Arbeit beschäftigt sich mit dem systematischen Vergleich verschiedener Methoden für die Detektion von differentiellem Spleißen in Exon-ArrayDaten. Anhand artifizieller und experimentell validierter Daten identifizieren wir Methoden, die über verschiedene Parameterszenarien hinweg robuste Ergebnisse liefern, und ermitteln bestimmte Datenparameter, die die Ergebnisgüte sowie die Qualität der angewandten Methoden beeinflussen. Im zweiten Teil identifizieren wir Spleiß-regulatorischer Proteine, die für die beobachteten Spleissveränderungen zwischen Krebs und einer Kontrolle verantwortlich sein könnten. Zu diesem Zweck stellen wir eine von uns entwickelte Methode basierend auf einem Netzwerkansatz vor. Hierbei werden Spleißfaktoren und differentiell gesplicete Exons in ein Netzwerk integriert und anschliessend anhand der Unterschiede in ihrer Zentralität geordnet. Im dritten Teil analysieren wir die Vergleichbarkeit zweier Datentypen, generiert durch unterschiedliche Technologien, in Bezug auf die Detektion von differentiellem Spleißen. Dazu beziehen wir mehrere Vergleichsebenen mit ein und wenden Methoden an, die für beide Technologien geeignet sind um eine methodenbasierte Beeinträchtigung der Vergleichbarkeit auszuschließen. Die Anwendung unseres Ansatzes auf zwei Datensätze identifiziert ähnliche Trends in der Vergleichbarkeit bei einer sich unterscheidenden Gesamtkonkordanz. / Alternative splicing is a crucial mechanism in eukaryotes, which provides an ample protein diversity that is necessary for maintaining an organism. In contrast, aberrant (alternative) splicing may lead to altered protein isoforms contributing to diseases such as cancer. In this thesis, we study differential splicing in cancer, i.e. splicing changes observed between cancerous and control tissues. We seek to identify methods best suited for the detection of differential splicing, we investigate regulatory factors potentially causal for the splicing changes observed, and we study the comparability of two data types obtained from different technologies with respect to differential splicing detection. The first part of the thesis assesses the performance of methods for detecting differential splicing from exon arrays as existing methods are often of low concordance. We examine global data parameters and their potential influence on results and method performance using artificial and validated experimental data. Overall, our evaluation indicates methods that perform robustly well across artificial and experimental data and identifies parameters impacting result performance. The second part aims at identifying regulatory factors responsible for splicing changes observed between cancer, and healthy tissue. Therefor, we develop a novel, network based approach which first integrates differentially spliced exons with splicing regulatory proteins (splicing factors), using transcriptomics data, and then ranks splicing factors according to their potential involvement in cancer. Third, we compare differential splicing detection based on RNA sequencing and exon array data by developing a multi-level comparison framework using two differential splicing detection methods applicable to both, RNA sequencing and exon array data, to avoid method inherent bias. We apply our multi-level framework to two data sets, leading, despite varying overall concordance, to similar trends in comparability.

Nous montrons l’utilisation de la puce exon d’Affymetrix pour l’analyse simultanée de l’expression des gènes et de la variation d’isoformes. Nous avons utilisé les échantillons d’ARN du cerveau et des tissus de référence qui ont été antérieurement utilisés dans l’étude du consortium MicroArray Quality Control (MAQC). Nous démontrons une forte concordance de la quantification de l’expression des gènes entre trois plateformes d’expression populaires à savoir la puce exon d’Affymetrix, la puce Illumina et la puce U133A d’Affymetrix. Plus intéressant nous montrons que la majorité des discordances entre les trois plateformes résulterait des positions différentes des sondes à travers les plateformes et que les variations d’isoforme exactes ne peuvent être identifiées que par la puce exon. Nous avons détecté avec succès, entre les tissus de référence et ceux du cerveau, une centaine de cas d’évènements d’épissage alternatif. La puce exon est requise dans l’analyse de l’épissage alternatif associé aux pathologies telles que les cancers et les troubles neurologiques. Comme application de cette technologie, nous avons analysé les variations d’épissage dans la métastase du cancer de sein développé dans le model de la souris. Nous avons utilisé une gamme bien définie de trois lignées de tumeur mammaire ayant différents potentiels métastatiques. Par des analyses statistiques, nous avons répertorié 2623 transcripts présentant des variations d’expression et d’isoformes entre les types de tumeur. Une analyse du réseau de gènes montre qu’environ la moitié d’entre eux est impliquée dans plusieurs activités cellulaires, ainsi que dans nombreux cancers et désordres génétiques. / We demonstrate how the Affymetrix Exon Array, can be used to simultaneously profile gene expression level, and detect variations at the isoform level. We use a well studied set of brain and reference RNA samples previously used by the MicroArray Quality Control (MAQC) consortium study. We demonstrate a high concordance of gene expression measurements among three popular expression platforms – Affymetrix Exon Array, Illumina, and Affymetrix 3’ targeted array (U133A). More interestingly, we show that in many cases of discordant results, the effect can be explained by differential probe placements across platforms, and that the exact isoform change can only be captured by the Exon Array. Finally, we are able to detect hundreds of cases of splicing, transcript initiation, and termination differences between the brain and reference tissue samples. We propose that the Exon Array is a highly effective tool for transcript isoform profiling, and that it should be used in a variety of systems where such changes are known to be associated with diseases, such as neurological disorders and cancer. As application, we used the Affymetrix Exon Array to identify metastatis-specific alternative splicing in mouse model of breast cancer at the whole genome level. We utilize a well characterized series of three mouse mammary tumor lines exhibiting varying levels of metastatic potential. We catalogued 2623 transcripts which exhibit splicing aberrations during the progression of cancer. A genetic pathway analysis shows the half of them implicated in several cell activities, cancers and genetic disorders.

Nous montrons l’utilisation de la puce exon d’Affymetrix pour l’analyse simultanée de l’expression des gènes et de la variation d’isoformes. Nous avons utilisé les échantillons d’ARN du cerveau et des tissus de référence qui ont été antérieurement utilisés dans l’étude du consortium MicroArray Quality Control (MAQC). Nous démontrons une forte concordance de la quantification de l’expression des gènes entre trois plateformes d’expression populaires à savoir la puce exon d’Affymetrix, la puce Illumina et la puce U133A d’Affymetrix. Plus intéressant nous montrons que la majorité des discordances entre les trois plateformes résulterait des positions différentes des sondes à travers les plateformes et que les variations d’isoforme exactes ne peuvent être identifiées que par la puce exon. Nous avons détecté avec succès, entre les tissus de référence et ceux du cerveau, une centaine de cas d’évènements d’épissage alternatif. La puce exon est requise dans l’analyse de l’épissage alternatif associé aux pathologies telles que les cancers et les troubles neurologiques. Comme application de cette technologie, nous avons analysé les variations d’épissage dans la métastase du cancer de sein développé dans le model de la souris. Nous avons utilisé une gamme bien définie de trois lignées de tumeur mammaire ayant différents potentiels métastatiques. Par des analyses statistiques, nous avons répertorié 2623 transcripts présentant des variations d’expression et d’isoformes entre les types de tumeur. Une analyse du réseau de gènes montre qu’environ la moitié d’entre eux est impliquée dans plusieurs activités cellulaires, ainsi que dans nombreux cancers et désordres génétiques. / We demonstrate how the Affymetrix Exon Array, can be used to simultaneously profile gene expression level, and detect variations at the isoform level. We use a well studied set of brain and reference RNA samples previously used by the MicroArray Quality Control (MAQC) consortium study. We demonstrate a high concordance of gene expression measurements among three popular expression platforms – Affymetrix Exon Array, Illumina, and Affymetrix 3’ targeted array (U133A). More interestingly, we show that in many cases of discordant results, the effect can be explained by differential probe placements across platforms, and that the exact isoform change can only be captured by the Exon Array. Finally, we are able to detect hundreds of cases of splicing, transcript initiation, and termination differences between the brain and reference tissue samples. We propose that the Exon Array is a highly effective tool for transcript isoform profiling, and that it should be used in a variety of systems where such changes are known to be associated with diseases, such as neurological disorders and cancer. As application, we used the Affymetrix Exon Array to identify metastatis-specific alternative splicing in mouse model of breast cancer at the whole genome level. We utilize a well characterized series of three mouse mammary tumor lines exhibiting varying levels of metastatic potential. We catalogued 2623 transcripts which exhibit splicing aberrations during the progression of cancer. A genetic pathway analysis shows the half of them implicated in several cell activities, cancers and genetic disorders.