The main objective of the reported study was to investigate how biomedical ontologies, logically structured representations of various aspects of the biomedical reality, can help researchers in analyzing experimental data. The dissertation reports two attempts to construct tools for the analysis of high-throughput experimental results using explicit domain knowledge representations. Furthermore, integrative efforts made by the community of Open Biomedical Ontologies (OBO), in which the author has participated, are reported, and a framework for consistently connecting the Gene Ontology (GO) with the Taxonomy of Species is proposed and discussed.

The field of proteomics has rapidly expanded due to the completion of the human genome sequence. This thesis validates affinity-purified monospecific antibodies of polyclonal origin, for protein profiling in a broad spectrum of normal tissues and cells. Validation of antibodies is crucial for development of reliable binders for target proteins and this thesis evaluates the generation and application of large sets of msAbs in different settings. MsAbs were generated towards recombinant Protein Epitope Signature Tag (PrEST) antigens using a stringent affinity-purification strategy, presented in the first study. The specificity of msAbs was studied using reverse phase protein arrays and immunohistochemistry (IHC), and results presented over 90% success rate in the protein array analysis. In IHC, 81% of the msAbs displayed apparent specific staining in normal tissues. MsAbs were also compared with commercial analogs (cAbs) using IHC and Western blot. Results presented similar outcome between msAbs and cAbs in both applications, although interpretation suggested more extensive IHC staining patterns with msAbs than with monoclonal analogs. For antibody validation, an approach called paired antibodies was presented and involved the generation of two msAbs towards non-overlapping epitopes on the same protein. Similarities in protein detection between paired antibodies were studied using three different antibody-based methods. Similar results were observed in several applications, indicating that this strategy can be a useful tool for studying known and unknown proteins. Given the reliability of msAbs, they were also applied in a study investigating the impact of tissue fixatives on protein detection. The study showed that different fixation mechanisms appeared to affect protein recognition by indicating that aldehyde-based fixation, e.g. induced by neutral buffered formalin, was preferred for tissues used in IHC and non-aldehyde based fixation was applicable for tissues used in protein extraction analysis and Western blotting. Conclusively, validation results suggest that msAbs are reliable affinity binders that can be used as valuable tools for proteome-wide protein profiling in tissues and cells.

Proteins are essential building blocks in every living cell, and since the complete human genome was sequenced in 2004, researchers have attempted to map the human proteome, which is the functional representation of the genome. One such initiative is the Human Protein Atlas programme (HPA), which generates monospecific antibodies towards all human proteins and uses these for high-throughput tissue profiling on tissue microarrays (TMAs). The results are publically available at the website www.proteinatlas.org. In this thesis, TMAs were used for analysis of expression patterns in various research areas. Different search queries in the HPA were tested and evaluated, and a number of potential biomarkers were identified, e.g. proteins exclusively expressed in islets of Langerhans, but not in exocrine glandular cells or other abdominal organs close to pancreas. The identified candidates were further analyzed on TMAs with pancreatic tissues from normal and diabetic individuals, and colocalization studies with insulin and glucagon revealed that several of the investigated proteins (DGCR2, GBF1, GPR44 and SerpinB10) appeared to be beta cell specific. Moreover, a set of proteins differentially expressed in lung cancer stroma was further analyzed on a clinical lung cancer cohort in the TMA format, and one protein (CD99) was significantly associated with survival. In addition, TMAs with tissue samples from different species were generated, e.g. for mapping of influenza virus attachment in various human and avian tissues. The results showed that the gull influenza virus H16N3 attached to human respiratory tract and eye, suggesting possible transmission of the virus between gull and human. TMAs were also used for analysis of protein expression differences between humans and other primates, and two proteins (TCF3 and SATB2) proved to be significantly differentially expressed on the human lineage at both the protein level and the RNA level.   In conclusion, this thesis exemplifies the potential of the TMA technology, which can be used for analysis of expression patterns in a large variety of research fields, such as biomarker discovery, influenza virus research or further understanding of human evolution.

S'han estudiat els efectes de dos fàrmacs intercalants al DNA, Daunorubicina (usat com antitumoral) i Criptolepina (amb elevada citotoxicitat i usat com antimalàric), en el llevat "S.cerevisiae".Ambdues molècules s'uneixen al DNA però amb diferent preferència de seqüència: mentre que la Daunorubicina reconeix preferentment regions 5'(A/T)CG3', la Criptolepina s'uneix a 5'GG3'. S'han descrit nombroses regions reguladores riques en aquest tipu de seqüències, pel que ambdós fàrmacs podrien competir amb factors de transcripció nuclears, modificant l'expressió gènica.Els principals objectius del nostre estudi foren:1.Determinar l'efecte citotòxic dels dos fàrmacs en diferents soques de S.cerevisiae: BY4741, BY4741-delta-erg6 (soca amb major permeabilitat de la membrana plasmàtica als fàrmacs), BY4741-delta-rad52 (soca amb un gen de la maquinària de reparació delecionat):- En primer lloc, s'analitzaren els efectes sobre el creixement cel·lular de les soques tractades amb diferents concentracions de cada molècula, tant amb glucosa com galactosa com a fonts de carboni. Els nostres resultats. Els nostres resultats demostren que la permeabilitat de la membrana és necessària per aconseguir la concentració de Daunorubicina que desencadena el seu efecte citotòxic. En canvi, la maquinària de reparació del dany al DNA pot ser una diana potencial de la Criptolepina.- Mitjançant una anàlisi semiquantitativa de l'expressió d'alguns dels principals gens del metabolisme de la galactosa, es va observar un efecte diferencial dels fàrmacs, ja que el factor regulador dels gens de la galactosa presenta en la seva seqüència consens, la diana per a la Daunorubicina i no per a la Criptolepina. Aquest solapament de seqüència explica el patró repressor induït per la Daunorubicina, però no per a la Criptolepina, resultat que està d'acord amb el major efecte citotòxic de la Daunorubicina en cèl·lules crescudes en galactosa.2.Avaluar la resposta transcripcional a nivell global de cada intercalant en la soca BY4741-delta-erg6, utilitzant la glucosa com a font de carboni.- Mitjançant l'ús de microarrays, es va identificar els perfils d'expressió gènica diferencial per a cadascun dels tractaments (Daunorubicina o Criptolepina), en funció de la concentració i del temps. A partir d'aquests experiments es va confirmar que l'efecte de la Daunorubicina sobre la transcripció és pronunciat, mentre que els canvis observats per a la criptolepina són lleus.- A partir dels resultats obtinguts, es van classificar els gens en diferents categories funcionals i es varen validar els resultats obtinguts per RT-PCR a temps real. Es va observar que tot i que la Daunorubicina reprimeix l'expressió dels gens glicolítics, la criptolepina està associada principalment a la inducció dels gens de resposta a estrès. D'una manera menys significativa, la Criptolepina altera també la transcripció de gens implicats en el transport del ferro i de gens mitocondrials.- Mitjançant un estudi in silico, es va confirmar que les variacions en els perfils d'expressió gènica es deuen a un efecte directe dels fàrmacs sobre els factors de transcripció que regulen aquests gens, degut a que competeixen per les mateixes dianes d'unió al DNA.A partir de la identificació dels mecanismes d'acció de cada fàrmac en S.cerevisiae, els nostres resultats plantegen noves vies en el disseny de nous fàrmacs antitumorals i antiparasitaris. / The main objective of this thesis is to analyse the effects of two DNA interacting drugs, Daunorubicin (used as antitumoral) and Cryptolepine (with elevated cytotoxicity and used as an antimalarian) in Saccharomyces cerevisiae. Both drugs intercalate in CG rich regions, the preferent sequence for Daunorubicin being 5'(A/T)CG3' and for Cryptolepine being 5'GG3'. These sites are common in consensus sequences of transcription factors and the competition of drugs for those sites can alter gene expresion.First, we evaluated the effects in cellular proliferation in yeast strains of different genetic backgrounds (related to membrane permeability and DNA lesion repair), in glucose and galactose as carbon source. We observed that permeability of plasma membrane is necessary in order to reach a citotoxic concentration of Daunorubicin. On the other hand, the DNA repair machinery seems to be a potential target of Cryptolepine. By a semiquantitative analysis of expression of GAL genes, we observed a differential effect of both drugs. Daunorubicin caused a decrease in GAL genes expression while Cryptolepine did not. This can be due to the fact that the main regulatory protein of galactose metabolism has the preferent binding site for Daunorubicin in the consensus sequences.The effect of Daunorubicin and Cryptolepine on yeast transcriptome was studied in the most permeable yeast strain using glucose as carbon source. Using microarrays, we identified different gene expression patterns for each drug. Genome wide results were grouped into functional categories and confirmed by real time RT-PCR. We observed that Daunorubicin downregulates glycolytic genes and Cryptolepine induces a stress response and decreases the transcriptional levels of specific genes related to iron transport (siderophores) and mitochondrial genes. We used in silico assays to correlate the expression patterns induced by both drugs with a direct effect in the regulatory proteins involved in the transcription of these genes. Results confirm different action mechanisms for each intercalating drug tested. Daunorubicin may alter the regulatory complex Gcr1p-Gcr2p responsible of upregulation of glycolytic genes, and Cryptolepine may induce a pleiotropic effect.Overall, these results can raise new approaches in the development of new antitumour and antiparasit drugs.

La cirugía radical como único tratamiento en carcinomas escamosos de cabeza y cuello (CECC) ha sido substituida por protocolos que permiten preservar la función del órgano como la quimioradioterapia (QRT) concomitante y la quimioterapia de inducción (QTI) seguida de radioterapia (RT) / quimioradioterapia o cirugía. El objetivo general de este proyecto de tesis es identificar marcadores moleculares en biopsias pre-tratamiento de CECC localmente avanzado que permitan predecir la evolución clínica del paciente. En este sentido, hemos llevado a cabo dos estudios independientes. En el primero hemos analizado los niveles de expresión de los genes Ku70, K80 y DNA-PKcs del sistema de reparación no homóloga por unión de extremos en biopsias pre-tratamiento de pacientes con CECC tratados con quimioterapia de inducción seguida de RT/QRT o cirugía. En el segundo hemos realizado un estudio de microarrays de expresión para determinar que genes y procesos biológicos están implicados en la respuesta tumoral en pacientes localmente avanzados tratados con QTI seguida de RT/QRT o cirugía, o con quimioradioterapia desde un inicio. Los resultados del primer estudio muestran que los tumores con una respuesta a QTI superior al 50% presentan niveles de expresión de Ku70, tanto de mRNA como de proteína, mayores que los tumores con una respuesta inferior al 50%. Además los pacientes con tumores con niveles de expresión de Ku70 elevados tienen una supervivencia libre de recidiva local (SLRL) y una supervivencia global (SG) mayor que los pacientes con tumores que expresan bajos niveles de dicho marcador. En el segundo estudio hemos identificado tres subtipos de tumores con diferencias en la evolución clínica de los pacientes. Los pacientes con tumores del cluster 1 tienen mayor SLRL y SG que el resto de los pacientes. Su perfil de expresión muestra que presentan una mayor capacidad de migración e invasividad, características de transición epitelio mesénquima, sobre-activación de la vía secretora y un menor grado de diferenciación celular. Los pacientes con tumores del cluster 3 tienen una evolución clínica favorable, presentan una SG y SLRL mayor que el resto de los pacientes. Entre sus características destacan un menor grado de diferenciación celular y una sobre-expresión de genes localizados en las regiones cromosómica 1q21 y 19q13. Los pacientes con tumores del cluster 2 tienen una evolución clínica intermedia ya que presentan una SLRL similar a los del cluster 3 y una SG similar a los del cluster 1. Hemos identificado una serie de marcadores de mal pronóstico (tirosina sulfotransferasa, trombospondina 1, liprina-beta1, fibronectina IIIB y alfa1-prolil-4-hidroxilasa) cuyo mayor expresión aumenta el riego de recidiva local a los 2 años y el riesgo de muerte a los 3 años de seguimiento. También hemos identificado una serie de marcadores de buen pronóstico (Oxidasa dual 1, LYPD3 y GTPasa Rab25) cuya mayor expresión disminuye el riesgo de recidiva local y muerte de los pacientes. / Induction chemotherapy (IC), followed by radiotherapy (RT) / chemoradiotherapy (CRT) or surgery, and concomitant CRT are commonly used to treat locally advanced head and neck squamous cell carcinoma (HNSCC). The objective of this thesis-project is to identify molecular markers in pre-treatment HNSCC biopsies associated with clinical outcome in patients with advanced stage. We have conducted two independent studies. In the first study, we have evaluated the relationship between Ku80, Ku70 or DNA PKcs expression in pre-treatment tumor biopsies, and tumor response in patients treated with IC, followed by RT / CRT or surgery. In the second analysis, we have performed a microarray study to identify genes and biological processes associated with tumor response in patients treated with IC, followed by RT / CRT or surgery, or concomitant CRT.Results of the first study show that tumors with a response to IC higher than 50% have significantly higher mRNA and protein levels for Ku70 than tumors with a response to IC lower than 50%. Moreover, high tumor Ku70 expression was associated with significantly longer local recurrence-free survival (LRFS) and overall survival (OS).In the second microarray study we have identified three tumor subtypes with differences in patient outcome. Patients with tumor subtype1 have a shorter LRLS and OS. These tumors have a higher migration and invasiveness capacity, epithelial-mesenchymal transition, activation of the secretory pathway and lower differentiation grade. Patients with subtype 3 have a longer LRFS and OS. These tumors have a higher differentiation grade and overexpress genes located on chromosome 1q21 and 19q13.Moreover, we have identified a minimum set of genes (TPST1, THBS1, PPF1BP1, FNDC3B, P4HA1, DUOX1, LYPD3 y RAB25) associated with local recurrence and patient survival.

This century is surely the century of data (Donoho, 2000). Data analysis has been an emerging activity over the last few decades. High dimensional data is in particular more and more pervasive with the advance of massive data collection system, such as microarrays, satellite imagery, and financial data. However, analysis of high dimensional data is of challenge with the so called curse of dimensionality (Bellman 1961). This research dissertation presents several methodologies in the application of high dimensional data analysis. The first part discusses a joint analysis of multiple microarray gene expressions. Microarray analysis dates back to Golub et al. (1999). It draws much attention after that. One common goal of microarray analysis is to determine which genes are differentially expressed. These genes behave significantly differently between groups of individuals. However, in microarray analysis, there are thousands of genes but few arrays (samples, individuals) and thus relatively low reproducibility remains. It is natural to consider joint analyses that could combine microarrays from different experiments effectively in order to achieve improved accuracy. In particular, we present a model-based approach for better identification of differentially expressed genes by incorporating data from different studies. The model can accommodate in a seamless fashion a wide range of studies including those performed at different platforms, and/or under different but overlapping biological conditions. Model-based inferences can be done in an empirical Bayes fashion. Because of the information sharing among studies, the joint analysis dramatically improves inferences based on individual analysis. Simulation studies and real data examples are presented to demonstrate the effectiveness of the proposed approach under a variety of complications that often arise in practice. The second part is about covariance matrix estimation in high dimensional data. First, we propose a penalised likelihood estimator for high dimensional t-distribution. The student t-distribution is of increasing interest in mathematical finance, education and many other applications. However, the application in t-distribution is limited by the difficulty in the parameter estimation of the covariance matrix for high dimensional data. We show that by imposing LASSO penalty on the Cholesky factors of the covariance matrix, EM algorithm can efficiently compute the estimator and it performs much better than other popular estimators. Secondly, we propose an estimator for high dimensional Gaussian mixture models. Finite Gaussian mixture models are widely used in statistics thanks to its great flexibility. However, parameter estimation for Gaussian mixture models with high dimensionality can be rather challenging because of the huge number of parameters that need to be estimated. For such purposes, we propose a penalized likelihood estimator to specifically address such difficulties. The LASSO penalty we impose on the inverse covariance matrices encourages sparsity on its entries and therefore helps reducing the dimensionality of the problem. We show that the proposed estimator can be efficiently computed via an Expectation-Maximization algorithm. To illustrate the practical merits of the proposed method, we consider its application in model-based clustering and mixture discriminant analysis. Numerical experiments with both simulated and real data show that the new method is a valuable tool in handling high dimensional data. Finally, we present structured estimators for high dimensional Gaussian mixture models. The graphical representation of every cluster in Gaussian mixture models may have the same or similar structure, which is an important feature in many applications, such as image processing, speech recognition and gene network analysis. Failure to consider the sharing structure would deteriorate the estimation accuracy. To address such issues, we propose two structured estimators, hierarchical Lasso estimator and group Lasso estimator. An EM algorithm can be applied to conveniently solve the estimation problem. We show that when clusters share similar structures, the proposed estimator perform much better than the separate Lasso estimator.

<p>Renewable fibers produced by forest trees provide excellentraw material of high economic value for industrialapplications. Despite this, the genes and corresponding enzymesinvolved in wood fiber biosynthesis in trees are poorlycharacterized. This thesis describes a functional genomicsapproach for the identification of carbohydrate-active enzymesinvolved in secondary cell wall (wood) formation in hybridaspen.</p><p>First, a 3' target amplification method was developed toenable microarray-based gene expression analysis on minuteamounts of RNA. The amplification method was evaluated usingboth a smaller microarray containing 192 cDNA clones and alarger microarray containing 2995 cDNA clones that werehybridized with targets isolated from xylem and phloem.Moreover, a gene expression study of phloem differentiation wasperformed to show the usefulness of the amplificationmethod.</p><p>A microarray containing 2995 cDNA clones representing aunigene set of a cambial region EST library was used to studygene expression during wood formation. Transcript populationsfrom thin tissue sections representing different stages ofxylem development were hybridized onto the microarrays. It wasdemonstrated that genes encoding lignin and cellulosebiosynthetic enzymes, as well as a number of genes withoutassigned function, were differentially expressed across thedevelopmental gradient.</p><p>Microarrays were also used to track changes in geneexpression in the developing xylem of transgenic, GA-20 oxidaseoverexpressing hybrid aspens that had increased secondarygrowth. The study revealed that a number of genes encoding cellwall related enzymes were upregulated in the transgenic trees.Moreover, most genes with high transcript changes could beassigned a role in the early events of xylogenesis.</p><p>Ten genes encoding putative cellulose synthases (CesAs) wereidentified in our own<i>Populus</i>ESTdatabase. Full length cDNA sequences wereobtained for five of them. Expression analyses performed withreal-time PCR and microarrays in normal wood undergoingxylogenesis and in tension wood revealed xylem specificexpression of four putative CesA isoenzymes.</p><p>Finally, an approach combining expressionprofiling,bioinformatics as well as EST and full length sequencing wasadopted to identify secondary cell wall related genes encodingcarbohydrate-active enzymes, such as glycosyltransferases andglycoside hydrolases. As expected, glycosyltransferasesinvolved in the carbohydrate biosynthesis dominated thecollection of the secondary cell wall related enzymes that wereidentified.</p><p><b>Key words:</b>Populus, xylogenesis, secondary cell wall,cellulose, hemicellulose, microarrays, transcript profiling,carbohydrate-active enzyme, glycosyltransferase, glycosidehydrolase</p>

<p>Forest biotechnology is a fast emerging field of research. The application of biotechnological tools will enhance the quality of the forest products. The resultant value added and environmentally sustainable products are an absolute necessity in the future. The study of wood biosynthesis and degradation will result in enormous knowledge resources, which can be used for exploiting wood properties. This thesis addresses questions representing both wood degradation and biosynthesis.</p><p>The wood degrading fungus <i>Phanerochaete chrysosporium</i> is expression profiled with the microarray technology. The objective is to understand the expression pattern of the extracellular carbohydrate active enzymes (CAZymes) secreted by the organism. The data obtained increases our understanding of gene expression upon growth on cellulose.</p><p>Wood biosynthesis is studied with the model wood forming tree species, <i>Populus</i>. The plentiful data resources from the expression profiling during wood formation in Populus are used as the platform of this work. One of the wood specific genes, <i>PttMAP20</i>, previously with an unknown function is studied in this thesis. The immunolocalisation of PttMAP20 with specific antibodies is demonstrated. The putative microtubule-targeting domain of the protein is demonstrated microscopically and by using a biochemical binding assay. </p>

Source: Masters Abstracts International, Volume: 47-06, page: 3373. Adviser: Wenyaw Chan. Includes bibliographical references.

published_or_final_version / Pathology / Doctoral / Doctor of Philosophy