Understanding the mechanisms of transcriptional regulation and the constraints that operate in gene promoter sequences is the key step in understanding the evolutionary conservation of transcriptional regulation. It is well known that regulatory regions with the same expression outputs do not have to share the sequence similarity. The most important elements in regulatory sequences are transcription factor binding sites and their position relocation does not usually influence the expression output. The least complex transcriptional regulation is characteristic for housekeeping genes. For their expression they require only basal core promoter elements (sometimes only CpG islands are sufficient) and general transcription factors, so they can be transcribed easily and immediately whenever they are needed. In this study we focused on transcriptional regulation of invertebrate amphioxus (Branchiostoma floridae) housekeeping genes in vertebrate systems. We prepared a set of constructs with amphioxus regulatory regions for testing their activity in different mammalian cell lines and a set of constructs with the same amphioxus regulatory regions for observing their spatial recognition in developing medaka fish embryo. We found that half of investigated amphioxus regulatory regions are recognized by...

Despite the rapid advancement of sequencing technologies and although the wide diffusion of Whole Genome Sequencing (WGS) and Whole Exome Sequencing (WES) led to an increase in the diagnoses of diseases (A. C. Lionel, et al. 2018; D. J. Stavropoulos, et al. 2016; J. C. Taylor, et al. 2015) most genetic variants remain without a clear interpretation. One of the main difficulty related with the assessment of sequencing results is the abundance of Single Nucleotide Variant (SNV), around 4 million, that each healthy individual carries. Nearly all of these mutations will not produce any phenotype, that is equal to say that they have a benign or neutral effect. Only handful of those variants are potentially pathogenic, namely disease-causing. That is why computational Variant Effect Predictor (VEP) tools are used to prioritize variants worth investigating for medical consideration. Furthermore, the evidence of computational tools is considered among the different sources for variant effect assessment according to the American College of Medical Genetics and Genomics (ACMG) and Association for Molecular Pathology (AMP) guidelines. In addition, those software tools can be recognized as medical devices according to the second article of the Medical Device Regulation (MDR) of the European Union (Regulation (EU) 2017/745). That is why building a computational tool that predicts with high accuracy variant pathogenicity might have a direct impact on the healthcare system. Since 2001 more than 100 VEPs tools have been developed. Yet, their thresholds to classify a variant as pathogenic are often set for high sensitivity, that results in high false positive rate, namely misclassification of benign variants (C. Cubuk, et al. 2021). During my PhD, I developed Deciphering Mutations in Actionable Genes (DeMAG), a supervised classifier for interpreting missense mutations, namely SNVs that alter the protein sequence, in a list of 59 actionable genes as identified by the ACMG Secondary Findings (SF) v2.0 list (S. S. Kalia, et al. 2017). DeMAG is a supervised classifier trained with a Gradient boosting machine (GBM) model that employs only 13 conservation-based and structural features derived from AlphaFold 3D models and manually curated Multiple Sequence Alignment (MSA). DeMAG yields the best performance on clinical data among other popular VEP tools, balancing sensitivity and specificity, reaching the highest Matthews Correlation Coefficient (MCC). The advancement of DeMAG is due to the assembling of a balanced and high-quality training set and to the design of the partners score, a feature that captures epistasis, both in the sequence and in the 3D space of the protein. Here, epistasis refers to residues co-evolution in the sequence and residues spatial proximity in the 3D structure of the protein. The feature is a probabilistic score obtained with a mixture discriminant analysis that predicts pathogenicity based on the phenotypic effect of co-evolving and spatially close residues. The partners score feature is a general framework to study genotype and phenotype interactions. For example, those interactions might be between hetero or homoproteins forming a complex as tertiary structure and genetic variants occurring at interfaces, already known to be disease-causing, might be enriched for the same phenotypic effect. The framework of the partners score might not be limited to protein sequence, for example, interactions in the 3D genome might reveal regions enriched with the same phenotypic effect. DeMAG has been trained only on a small set of genes and yet, without further training, it generalizes well to additional 257 genes that have enough clinical data. Because for those new genes I did not manually curate MSA, I noted that the partners score from protein 3D models seems necessary for reaching high performance, while the contribution of the partners score obtained from long-range interactions, as derived from the co-evolution analysis, does not seem crucial for variant effect predictions. DeMAG is a supervised method especially designed for clinical translation purposes. That is why it focuses on clinically actionable genes and it balances its performance between the accuracy of the pathogenic and the benign class, acknowledging the importance of minimizing both the false negatives and false positives to avoid under and over diagnosis, critical to reduce health costs and patients psychological burden. Unsupervised general VEPs are powerful tools to investigate the functional effect of genetic variants as demonstrated by their higher correlation, over supervised tools, with data from Multiplexed Assay of Variant Effect (MAVE) and Deep Mutational Scanning (DMS) experiments. Nevertheless, for targeted clinical applications, I endorse the development of specialized tools that can leverage the existing wealth of data and knowledge available to minimize predictions errors. In order to make DeMAG readily available, I developed a web application available at https://demag.org/demag_app/ that provides predictions for all amino acids substitutions in the 59 and additional 257 genes together with training and testing datasets. Moreover, the app displays all the features of DeMAG highlighting the specific value annotated for the query mutation in relation to the distribution of the features for the pathogenic and benign mutations in the training set. This provides more insights than the minimalistic prediction label.

Humans have two plasminogen activators (PAs), tissue-type plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA), which generate plasmin to breakdown fibrin and other barriers to cell migration. Both PAs are used as pharmaceuticals but their efficacies are limited by their rapid clearance from the circulation, predominantly by parenchymal cells of the liver. At the commencement of the work presented here, the hepatic receptors responsible for mediating the catabolism of the PAs were little understood. tPA degradation by hepatic cell lines was known to depend on the formation of binary complexes with the major PA inhibitor, plasminogen activator inhibitor type-1 (PAI-1). Initial studies presented here established that uPA was catabolised in a fashion similar to tPA by the hepatoma cell line, HepG2. Other laboratories around this time found that the major receptor mediating the binding and endocytosis of the PAs is Low Density Lipoprotein Receptor-related Protein (LRP1). LRP1 is a giant 600 kDa protein that binds a range of structurally and functionally diverse ligands including, activated α2 macroglobulin, apolipoproteins, β amyloid precursor protein, and a number of serpin-enzymes complexes, including PA??PAI-1 complexes. Further studies for the work presented here centred on this receptor. By using radiolabelled binding assays, ligand blots, and Western blots on cultured cells, the major findings are that: (1) basal LRP1 expression on HepG2 is low compared to a clone termed, HepG2a16, but appears to increase in long term culture; (2) a soluble form of LRP1, which retains ligand-binding capacity, is present in human circulation; (3) soluble LRP1 is also present in cerebral spinal fluid where its role in neurological disorders such as Alzheimer??s disease is a developing area of interest; and (4) the release of LRP1 is a mechanism conserved in evolution, possibly as distantly as molluscs. The discovery, identification, and characterisation of soluble LRP1 introduces this protein in the human circulation, and presents a possible further level of regulation for its associated receptor system.

La localisation subcellulaire de l’ARN permet un déploiement prompt et spatialement restreint autant des activités protéiques que des ARN noncodant. Le trafic d’ARN est dirigé par des éléments de séquences (sous-séquences primaires, structures secondaires), aussi appelés motifs de régulation, présents en cis à même la molécule d’ARN. Ces motifs sont reconnus par des protéines de liaisons aux ARN qui médient l’acheminement des transcrits vers des sites précis dans la cellule. Des études récentes, chez l’embryon de Drosophile, indiquent que la majorité des ARN ont une localisation subcellulaire asymétrique, suggérant l’existence d’un « code de localisation » complexe. Cependant, ceci peut représenter un exemple exceptionnel et la question demeurait, jusqu’ici, si une prévalence comparable de localisation d’ARN est observable chez des cellules standards développées en culture. De plus, des informations facilement disponibles à propos des caractéristiques de distribution topologique d’instances de motifs à travers des transcriptomes complets étaient jusqu’à présent manquantes. Afin d’avoir un aperçu de l’étendue et des propriétés impliquées dans la localisation des ARN, nous avons soumis des cellules de Drosophile (D17) et de l’humain (HepG2) à un fractionnement biochimique afin d’isoler les fractions nucléaire, cytosolique, membranaire et insoluble. Nous avons ensuite séquencé en profondeur l’ARN extrait et analysé par spectrométrie de masse les protéines extraites de ces fractions. Nous avons nommé cette méthode CeFra-Seq. Par des analyses bio-informatiques, j’ai ensuite cartographié l’enrichissement de divers biotypes d’ARN (p. ex. ARN messager, ARN long non codant, ARN circulaire) et protéines au sein des fractions subcellulaires. Ceci a révélé que la distribution d’un large éventail d’espèces d’ARN codants et non codants est asymétrique. Une analyse des gènes orthologues entre mouche et humain a aussi démontré de fortes similitudes, suggérant que le processus de localisation est évolutivement conservé. De plus, j’ai observé des attributs (p. ex. la taille des transcrits) distincts parmi les populations d’ARN messagers spécifiques à une fraction. Finalement, j’ai observé des corrélations et anti-corrélations spécifiques entre certains groupes d’ARN messagers et leurs protéines. Pour permettre l’étude de la topologie de motifs et de leurs conservations, j’ai créé oRNAment, une base de données d’instances présumée de sites de liaison de protéines chez des ARN codants et non codants. À partir de données de motifs de liaison protéique par RNAcompete et par RNA Bind-n-Seq, j’ai développé un algorithme permettant l’identification rapide d’instances potentielles de ces motifs dans un transcriptome complet. J’ai pu ainsi cataloguer les instances de 453 motifs provenant de 223 protéines liant l’ARN pour 525 718 transcrits chez cinq espèces. Les résultats obtenus ont été validés en les comparant à des données publiques de eCLIP. J’ai, par la suite, utilisé oRNAment pour analyser en détail les aspects topologiques des instances présumées de ces motifs et leurs conservations évolutives relatives. Ceci a permis de démontrer que la plupart des motifs sont distribués de façon similaire entre espèces. De plus, j’ai discerné des points communs entre les sous-groupes de protéines liant des biotypes distincts ou des régions d’ARN spécifiques. La présence de tels patrons, similaires ou non, entre espèces est susceptible de refléter l’importance de leurs fonctions. D’ailleurs, l’analyse plus détaillée du positionnement d’un motif entre régions transcriptomiques comparables chez les vertébrés suggère une conservation synténique de ceux-ci, à divers degrés, pour tous les biotypes d’ARN. La topologie régionale de certaines instances de motifs répétées apparaît aussi comme évolutivement conservée et peut être importante afin de permettre une liaison adéquate de la protéine. Finalement, les résultats compilés avec oRNAment ont permis de postuler sur un nouveau rôle potentiel pour l’ARN long non codant HELLPAR comme éponge de protéines liant l’ARN. La caractérisation systématique d’ARN localisés et de motifs de régulation en cis présentée dans cette thèse démontre comment l’intégration d’information à l’échelle transcriptomique permet d’évaluer la prévalence de l’asymétrie, les caractéristiques distinctes et la conservation évolutive de collections d’ARN. / The subcellular localization of RNA allows a rapid and spatially restricted deployment of protein and noncoding RNA activities. The trafficking of RNA is directed by sequence elements (primary subsequences, secondary structures), also called regulatory motifs, present in cis within the RNA molecule. These motifs are recognized by RNA-binding proteins that mediate the transport of transcripts to specific sites in the cell. Recent studies in the Drosophila embryo indicate that the majority of RNAs display an asymmetric subcellular localization, suggesting the existence of a complex "localization code". However, this may represent an exceptional example and the question remained, until now, whether a comparable prevalence of RNA localization is observable in standard cells grown in culture. In addition, readily available information about the topological distribution of pattern instances across full transcriptomes has been hitherto lacking. In order to have a broad overview of the extent and properties involved in RNA localization, we subjected Drosophila (D17) and human (HepG2) cells to biochemical fractionation to isolate the nuclear, cytosolic, membrane and insoluble fractions. We then performed deep sequencing on the extracted RNA and analyzed through mass spectrometry the proteins extracted from these fractions. We named this method CeFra-Seq. Through bioinformatics analyses, I then profiled the enrichment of various RNA biotypes (e.g. messenger RNA, long noncoding RNA, circular RNA) and proteins within the subcellular fractions. This revealed the high prevalence of asymmetric distribution of both coding and noncoding RNA species. An analysis of orthologous genes between fly and human has also shown strong similarities, suggesting that the localization process is evolutionarily conserved. In addition, I have observed distinct attributes (e.g. transcript size) among fraction-specific messenger RNA populations. Finally, I observed specific correlations and anti-correlations between defined groups of messenger RNAs and the proteins they encode. To study motifs topology and their conservation, I created oRNAment, a database of putative RNA-binding protein binding sites instances in coding and noncoding RNAs. Using data from protein binding motifs assessed by RNAcompete and by RNA Bind-n-Seq experiments, I have developed an algorithm allowing their rapid identification in a complete transcriptome. I was able to catalog the instances of 453 motifs from 223 RNA-binding proteins for 525,718 transcripts in five species. The results obtained were validated by comparing them with public data from eCLIP. I then used oRNAment to further analyze the topological aspects of these motifs’ instances and their relative evolutionary conservation. This showed that most motifs are distributed in a similar fashion between species. In addition, I have detected commonalities between the subgroups of proteins linking preferentially distinct biotypes or specific RNA regions. The presence or absence of such pattern between species is likely a reflection of the importance of their functions. Moreover, a more precise analysis of the position of a motif among comparable transcriptomic regions in vertebrates suggests a syntenic conservation, to varying degrees, in all RNA biotypes. The regional topology of certain motifs as repeated instances also appears to be evolutionarily conserved and may be important in order to allow adequate binding of the protein. Finally, the results compiled with oRNAment allowed to postulate on a potential new role for the long noncoding RNA HELLPAR as an RNA-binding protein sponge. The systematic characterization of RNA localization and cis regulatory motifs presented in this thesis demonstrates how the integration of information at a transcriptomic scale enables the assessment of the prevalence of asymmetry, the distinct characteristics and the evolutionary conservation of RNA clusters.