Biomedical Event Extraction with Machine Learning

Biomedical natural language processing (BioNLP) is a subfield of natural language processing, an area of computational linguistics concerned with developing programs that work with natural language: written texts and speech. Biomedical relation extraction concerns the detection of semantic relations such as protein-protein interactions (PPI) from scientific texts. The aim is to enhance information retrieval by detecting relations between concepts, not just individual concepts as with a keyword search. In recent years, events have been proposed as a more detailed alternative for simple pairwise PPI relations. Events provide a systematic, structural representation for annotating the content of natural language texts. Events are characterized by annotated trigger words, directed and typed arguments and the ability to nest other events. For example, the sentence “Protein A causes protein B to bind protein C” can be annotated with the nested event structure CAUSE(A, BIND(B, C)). Converted to such formal representations, the information of natural language texts can be used by computational applications. Biomedical event annotations were introduced by the BioInfer and GENIA corpora, and event extraction was popularized by the BioNLP'09 Shared Task on Event Extraction. In this thesis we present a method for automated event extraction, implemented as the Turku Event Extraction System (TEES). A unified graph format is defined for representing event annotations and the problem of extracting complex event structures is decomposed into a number of independent classification tasks. These classification tasks are solved using SVM and RLS classifiers, utilizing rich feature representations built from full dependency parsing. Building on earlier work on pairwise relation extraction and using a generalized graph representation, the resulting TEES system is capable of detecting binary relations as well as complex event structures. We show that this event extraction system has good performance, reaching the first place in the BioNLP'09 Shared Task on Event Extraction. Subsequently, TEES has achieved several first ranks in the BioNLP'11 and BioNLP'13 Shared Tasks, as well as shown competitive performance in the binary relation Drug-Drug Interaction Extraction 2011 and 2013 shared tasks. The Turku Event Extraction System is published as a freely available open-source project, documenting the research in detail as well as making the method available for practical applications. In particular, in this thesis we describe the application of the event extraction method to PubMed-scale text mining, showing how the developed approach not only shows good performance, but is generalizable and applicable to large-scale real-world text mining projects. Finally, we discuss related literature, summarize the contributions of the work and present some thoughts on future directions for biomedical event extraction. This thesis includes and builds on six original research publications. The first of these introduces the analysis of dependency parses that leads to development of TEES. The entries in the three BioNLP Shared Tasks, as well as in the DDIExtraction 2011 task are covered in four publications, and the sixth one demonstrates the application of the system to PubMed-scale text mining.

Morphological and functional characterization of placenta during gestation in bovine clones derived by somatic nuclear transfer

La technique de clonage par transfert nucléaire de cellules somatiques (SCNT) présente une page importante dans les annales scientifiques, mais son application pratique demeure incertaine dû à son faible taux de succès. Les anomalies placentaires et de développement fœtal se traduisent par des pertes importantes de gestation et des mortalités néonatales. Dans un premier temps, la présente étude a caractérisé les changements morphologiques des membranes fœtales durant la gestation clonée en les comparant à des gestations contrôles obtenues à partir de l’insémination artificielle. Les différentes anomalies morphologiques des placentomes telles que l’œdème chorioallantoique, la présence de zones hyperéchoiques et irrégulières dans la membrane amniotique et la présence de cellules inflammatoires dégénérées compromettent le développement fœtal normal de la gestation clonée. L’examen ultrasonographique représente une technique diagnostique importante pour faire le suivi d’une gestation et de caractériser les changements placentaires dans le cadre d’évaluation globale du bien-être fœtal. Le profil hormonal de trois stéroïdes (progestérone (P4), estrone sulfate (E1S), et œstradiol (E2)) et de la protéine B spécifique de gestation (PSPB) dans le sérum des vaches porteuses de clones SCNT a été déterminé et associé aux anomalies de gestations clonées. Une diminution de la P4 sérique au jour 80, une élévation du niveau de la concentration de la PSPB au jour 150, et une augmentation de la concentration d’E2 sérique durant le deuxième et troisième tiers de la gestation clonée coïncident avec les anomalies de gestation déjà reportées. Ces changements du profil hormonal associés aux anomalies phénotypiques du placenta compromettent le déroulement normal de la gestation clonée et gênent le développement et le bien-être fœtal. Sur la base des observations faites sur le placenta de gestation clonée, le mécanisme moléculaire pouvant expliquer la disparition de l’épithélium du placenta (l’interface entre le tissue maternel et le placenta) a été étudié. L’étude a identifié des changements dans l’expression de deux protéines d’adhérence (E-cadhérin et β-catenin) de cellules épithéliales pouvant être associées aux anomalies du placenta chez les gestations clonées. Le tissu de cotylédons provenant de gestations clonées et contrôles a été analysé par Western blot, RT-PCR quantitatif, et par immunohistochimie. Les résultats présentaient une diminution significative (p<0.05) de l’expression des dites protéines dans les cellules trophoblastiques chez les gestations clonées. Le RT-PCR quantitatif démontrait que les gènes CCND1, CLDN1 et MSX1 ciblés par la voie de signalisation de la Wnt/β-catenin étaient significativement sous exprimés. La diminution de l’expression des protéines E-cadherin et β-catenin avec une réduction de l’activation de la protéine β-catenin durant le période d’attachement de l’embryon peut potentiellement expliquer l’absence totale ou partielle de l’attachement des membranes fœtales au tissu maternel et éventuellement, l’insuffisance placentaire caractéristique des gestations clonées chez la vache. La caractérisation morphologique et fonctionnelle du placenta durant les gestations clonées à haut risque est essentielle pour évaluer le statut de la gestation. Les résultats de la présente étude permettront de prédire le développement et le bien-être fœtal de façon critique à travers un protocole standardisé et permettre des interventions médicales pour améliorer le taux de succès des gestations clonées chez les bovins.

Investigation of the structure and dynamics of the centromeric epigenetic mark

Le centromère est le site chromosomal où le kinetochore se forme, afin d’assurer une ségrégation fidèles des chromosomes et ainsi maintenir la ploïdie appropriée lors de la mitose. L’identité du centromere est héritée par un mécanisme épigénétique impliquant une variante de l’histone H3 nommée centromere protein-A (CENP-A), qui remplace l’histone H3 au niveau de la chromatine du centromère. Des erreurs de propagation de la chromatine du centromère peuvent mener à des problèmes de ségrégation des chromosomes, pouvant entraîner l’aneuploïdie, un phénomène fréquemment observé dans le cancer. De plus, une expression non-régulée de CENP-A a aussi été rapportée dans différentes tumeurs humaines. Ainsi, plusieurs études ont cherchées à élucider la structure et le rôle de la chromatine contenant CENP-A dans des cellules en prolifération. Toutefois, la nature moléculaire de CENP-A en tant que marqueur épigénétique ainsi que ces dynamiques à l'extérieur du cycle cellulaire demeurent des sujets débat. Dans cette thèse, une nouvelle méthode de comptage de molécules uniques à l'aide de la microscopie à réflexion totale interne de la fluorescence (TIRF) sera décrite, puis exploitée afin d'élucider la composition moléculaire des nucléosomes contenant CENP-A, extraits de cellules en prolifération. Nous démontrons que les nucléosomes contenant CENP-A marquent les centromères humains de façon épigénétique à travers le cycle cellulaire. De plus, nos données démontrent que la forme prénucléosomale de CENP-A, en association avec la protéine chaperon HJURP existe sous forme de monomère et de dimère, ce qui reflète une étape intermédiaire de l'assemblage de nucléosomes contenant CENP-A. Ensuite, des analyses quantitatives de centromères lors de différenciation myogénique, et dans différents tissus adultes révèlent des changements globaux qui maintiennent la marque épigénétique dans une forme inactive suite à la différentiation terminale. Ces changements incluent une réduction du nombre de points focaux de CENP-A, un réarrangement des points dans le noyau, ainsi qu'une réduction importante de la quantité de CENP-A. De plus, nous démontrons que lorsqu'une dédifférenciation cellulaire est induite puis le cycle cellulaire ré-entamé, le phénotype "différencié" décrit ci-haut est récupéré, et les centromères reprennent leur phénotype "prolifératif". En somme, cet oeuvre décrit la composition structurale sous-jacente à l'identité épigénétique des centromères de cellules humaines lors du cycle cellulaire, et met en lumière le rôle de CENP-A à l'extérieur du cycle cellulaire.

Downregulation of VAPB expression in motor neurons derived from induced pluripotent stem cells of ALS8 patients

Amyotrophic lateral sclerosis (ALS) is an incurable neuromuscular disease that leads to a profound loss of life quality and premature death. Around 10% of the cases are inherited and ALS8 is an autosomal dominant form of familial ALS caused by mutations in the vamp-associated protein B/C (VAPB) gene. The VAPB protein is involved in many cellular processes and it likely contributes to the pathogenesis of other forms of ALS besides ALS8. A number of successful drug tests in ALS animal models could not be translated to humans underscoring the need for novel approaches. The induced pluripotent stem cells (iPSC) technology brings new hope, since it can be used to model and investigate diseases in vitro. Here we present an additional tool to study ALS based on ALS8-iPSC. Fibroblasts from ALS8 patients and their non-carrier siblings were successfully reprogrammed to a pluripotent state and differentiated into motor neurons. We show for the first time that VAPB protein levels are reduced in ALS8-derived motor neurons but, in contrast to over-expression systems, cytoplasmic aggregates could not be identified. Our results suggest that optimal levels of VAPB may play a central role in the pathogenesis of ALS8, in agreement with the observed reduction of VAPB in sporadic ALS.

Hepatic denervation impairs the assembly and secretion of VLDL-TAG

VLDL secretion is a regulated process that depends on the availability of lipids, apoB and MTP. Our aim was to investigate the effect of liver denervation upon the secretion of VLDL and the expression of proteins involved in this process. Denervation was achieved by applying a 85% phenol solution onto the portal tract, while control animals were treated with 9% NaCl. VLDL secretion was evaluated by the Tyloxapol method. The hepatic concentration of TAG and cholesterol, and the plasma concentration of TAG, cholesterol, VLDL-TAG, VLDL-cholesterol and HDL-cholesterol were measured, as well as mRNA expression of proteins involved in the process of VLDL assembly. Hepatic acinar distribution of MTP and apoB was evaluated by immunohistochemistry. Denervation increased plasma concentration of cholesterol (125.3 +/- 10.1 vs. 67.1 +/- 4.9 mg dL(-1)) and VLDL-cholesterol (61.6 +/- 5.6 vs. 29.4 +/- 3.3 mg dL(-1)), but HDL-cholesterol was unchanged (45.5 +/- 6.1 vs. 36.9 +/- 3.9 mg dL(-1)). Secretion of VLDL-TAG (47.5 +/- 23.8 vs. 148.5 +/- 27.4 mg dL h(-1)) and mRNA expression of CPT I and apoB were reduced (p < 0.01) in the denervated animals. MTP and apoB acinar distribution was not altered in the denervated animals, but the intensity of the reaction was reduced in relation to controls. Copyright (C) 2008 John Wiley & Sons, Ltd.

Papel das Yops secretadas por Yersinia sobre a resposta imune do hospedeiro

The genus Yersinia contains three species pathogenic to humans: Y. pestis, Y. enterocolitica e Y. pseudotuberculosis. The pathogenicity of Yersinia is linked to the presence of a 70-kb virulence plasmid (pYV) that is common to the three species and codifies a type III secretion system and a set of virulence proteins, including those known as Yersinia outer proteins (Yops), that are exported by this system when the bacteria encounter host cells. Two Yops translocators (YopB and YopD) are inserted into the host plasma membrane and transport six effectors (YopO, YopH, YopM, YopJ and YopT) across the membrane into the cytosol of the host cell. The Yops effectors interfere with multiple signaling pathways of the infected cell, affecting both the innate and adaptive immune responses. This article focuses on the role of Yops in the modulation of the host immune response.

Influência dos níveis protéicos fornecidos na dieta sobre o sistema reprodutivo de carneiros

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Potential quimioprotetor do DIM (3, 3' - Di-indolil-metano) e da Genisteína em linhagens de células tumorais prostáticas humanas LNCaP e PC-3 expostas ao Bisfenol A

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Antitumor activity of irradiated riboflavin on human renal carcinoma cell line 786-O

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Analysis of Copy Number Variants identifies new candidate genes for Autism Spectrum Disorder and Intellectual Disability

Autism spectrum disorder (ASD) and Intellectual Disability (ID) are complex neuropsychiatric disorders characterized by extensive clinical and genetic heterogeneity and with overlapping risk factors. The aim of my project was to further investigate the role of Copy Numbers Variants (CNVs), identified through genome-wide studies performed by the Autism Geome Project (AGP) and the CHERISH consortium in large cohorts of ASD and ID cases, respectively. Specifically, I focused on four rare genic CNVs, selected on the basis of their impact on interesting ASD/ID candidate genes: a) a compound heterozygous deletion involving CTNNA3, predicted to cause the lack of functional protein; b) a 15q13.3 duplication containing CHRNA7; c) a 2q31.1 microdeletion encompassing KLHL23, SSB and METTL5; d) Lastly, I investigated the putative imprinting regulation of the CADPS2 gene, disrupted by a maternal deletion in two siblings with ASD and ID. This study provides further evidence for the role of CTNNA3, CHRNA7, KLHL23 and CADPS2 as ASD and/or ID susceptibility genes, and highlights that rare genetic variation contributes to disease risk in different ways: some rare mutations, such as those impacting CTNNA3, act in a recessive mode of inheritance, while other CNVs, such as those occurring in the 15q13.3 region, are implicated in multiple developmental and/or neurological disorders possibly interacting with other susceptibility variants elsewhere in the genome. On the other hand, the discovery of a tissue-specific monoallelic expression for the CADPS2 gene, implicates the involvement of epigenetic regulatory mechanisms as risk factors conferring susceptibility to ASD/ID.

Alteration of the pulmonary surfactant system in full-term infants with hereditary ABCA3 deficiency

RATIONALE: ABCA3 mutations are known to cause fatal surfactant deficiency. OBJECTIVE: We studied ABCA3 protein expression in full-term newborns with unexplained respiratory distress syndrome (URDS) as well as the relevance of ABCA3 mutations for surfactant homeostasis. METHODS: Lung tissue of infants with URDS was analyzed for the expression of ABCA3 in type II pneumocytes. Coding exons of the ABCA3 gene were sequenced. Surfactant protein expression was studied by immunohistochemistry, immunoelectron microscopy, and Western blotting. RESULTS: ABCA3 protein expression was found to be greatly reduced or absent in 10 of 14 infants with URDS. Direct sequencing revealed distinct ABCA3 mutations clustering within vulnerable domains of the ABCA3 protein. A strong expression of precursors of surfactant protein B (pro-SP-B) but only low levels and aggregates of mature surfactant protein B (SP-B) within electron-dense bodies in type II pneumocytes were found. Within the matrix of electron-dense bodies, we detected precursors of SP-C (pro-SP-C) and cathepsin D. SP-A was localized in small intracellular vesicles, but not in electron-dense bodies. SP-A and pro-SP-B were shown to accumulate in the intraalveolar space, whereas mature SP-B and SP-C were reduced or absent, respectively. CONCLUSION: Our data provide evidence that ABCA3 mutations are associated not only with a deficiency of ABCA3 but also with an abnormal processing and routing of SP-B and SP-C, leading to severe alterations of surfactant homeostasis and respiratory distress syndrome. To identify infants with hereditary ABCA3 deficiency, we suggest a combined diagnostic approach including immunohistochemical, ultrastructural, and mutation analysis.

Salivary antibodies directed against outer membrane proteins of Moraxella catarrhalis in healthy adults

Moraxella catarrhalis is a major mucosal pathogen of the human respiratory tract, but the mucosal immune response directed against surface components of this organism has not been characterized in detail. The aim of this study was to investigate the salivary immunoglobulin A (IgA) response toward outer membrane proteins (OMP) of M. catarrhalis in healthy adults, the group of individuals least likely to be colonized and thus most likely to display mucosal immunity. Unstimulated saliva samples collected from 14 healthy adult volunteers were subjected to IgA immunoblot analysis with OMP preparations of M. catarrhalis strain O35E. Immunoblot analysis revealed a consistent pattern of IgA reactivity, with the appearance of five major bands located at >250, 200, 120, 80, and 60 kDa. Eleven (79%) of 14 saliva samples elicited reactivity to all five bands. Immunoblot analysis with a set of isogenic knockout mutants lacking the expression of individual OMP was used to determine the identities of OMP giving rise to IgA bands. Human saliva was shown consistently to exhibit IgA-binding activity for oligomeric UspA2 (>250 kDa), hemagglutinin (200 kDa), monomeric UspA1 (120 kDa), transferrin-binding protein B (TbpB), monomeric UspA2, CopB, and presumably OMP CD. TbpB, oligomeric UspA2, and CopB formed a cluster of bands at about 80 kDa. These data indicate that the human salivary IgA response is directed consistently against a small number of major OMP, some of which are presently considered vaccine candidates. The functional properties of these mucosal antibodies remain to be elucidated.

Mechanisms of negative regulation of inflammasome activation

Thesis (Ph.D.)--University of Washington, 2016-06

Purification and characterization of a CENP-B homologue protein that binds to the centromeric K-type repeat DNA of Schizosaccharomyces pombe

We have purified and characterized a novel 60-kDa protein that binds to centromeric K-type repeat DNA from Schizosaccharomyces pombe. This protein was initially purified by its ability to bind to the autonomously replicating sequence 3002 DNA. Cloning of the gene encoding this protein revealed that it possesses significant homology to the mammalian centromere DNA-binding protein CENP-B and S. pombe Abp1, and this gene was designated as cbh+ (CENP-B homologue). Cbh protein specifically interacts in vitro with the K-type repeat DNA, which is essential for centromere function. The Cbh-binding consensus sequence was determined by DNase I footprinting assays as PyPuATATPyPuTA, featuring an inverted repeat of the first four nucleotides. Based on its binding activity to centromeric DNA and homology to centromere proteins, we suggest that this protein may be a functional homologue of the mammalian CENP-B in S. pombe.

Sulfonyl-hydrazones of cyclic imides derivatives as potent inhibitors of the Mycobacterium tuberculosis protein tyrosine phosphatase B (PtpB)

Searching lead compounds for new antituberculosis drugs, the activity of synthetic sulfonamides and sulfonyl-hydrazones were assayed for their potential inhibitory activity towards a protein tyrosine phosphatase from Mycobacterium tuberculosis - PtpB. Four sulfonyl-hydrazones N-phenylmaleimide derivatives were active (compounds 14, 15, 19 and 21), and the inhibition of PtpB was found to be competitive with respect to the substrate p-nitrophenyl phosphate. Structure-based molecular docking simulations were performed and indicated that the new inhibitor candidates showed similar binding modes, filling the hydrophobic pocket of the protein by the establishment of van der Waals contacts, thereby contributing significantly to the complex stability.