ED Sciences Chimiques
Total synthesis of Securinega alkaloids and study of their biological activities
by Adrien DENIZET (Institut des Sciences Moléculaires)
The defense will take place at 14h00 - Salle de conférence Centre d'Appui à la Recherche et de Formation (CARF),146 Rue Léo Saignat, 33000 Bordeaux
in front of the jury composed of
- Philippe PEIXOTO - Chargé de recherche - Institut des Sciences Moléculaires (ISM), UMR 5255, Université de Bordeaux - Directeur de these
- Thomas GRUTTER - Directeur de recherche - Chemo-Biologie Synthétique & Thérapeutiques (CBST), UMR 7199, Université de Strasbourg - Rapporteur
- Yung-Sing WONG - Directeur de recherche - Département de Pharmacochimie Moléculaire (DPM), UMR 5063, Université de Grenoble - Rapporteur
- Stéphanie CLUZET - Professeure des universités - Institut des Sciences de la Vigne et du Vin (ISVV), UMR 1366, Université de Bordeaux - Examinateur
- Antoine TALY - Directeur de recherche - Laboratoire de Biochimie Théorique (LBT), UPR 9080, Université de Paris - Examinateur
- Yannick LANDAIS - Professeur des universités - Institut des Sciences Moléculaires (ISM), UMR 5255, Université de Bordeaux - Examinateur
Securinega alkaloids constitute a family of tetracyclic metabolites isolated from subtropical plants of the Phyllanthaceae family. Since the identification of the first members of this class in the mid-XXth century, more than 170 compounds have been described to date. Over the years, these molecules have attracted increasing interest because of their diverse biological activities. Among them, the activity at GABAA receptors of (–)-securinine, the first isolated entity of this family, was demonstrated approximately forty years ago, paving the way for the study of these alkaloids in the context of central nervous system disorders. Despite these initial observations, the molecular mechanisms underlying their neurobiological effects remain largely unknown, leaving unresolved their pharmacological positioning among currently available GABAA receptor ligands. Limited access to these compounds from natural sources has long hindered their investigation. However, advances in organic synthesis now make it possible to explore them in greater depth using modern neurobiological tools. In this context, the main objective of this thesis was to characterize the activity of several Securinega alkaloids at GABAA receptors, and to determine which structural features of both the natural products and the receptor influence their activity. The first stage of this work consisted in producing several members of this family in sufficient quantities through total synthesis, relying on expertise developed and mastered within my research group. Electrophysiological studies performed on Xenopus laevis oocytes and mouse brain slices revealed structure–activity relationships within this series of alkaloids, as well as an influence of GABAA receptor subunit composition on their activity. These results were further supported by molecular modeling studies, providing a rationale for the observed selectivity profiles. Beyond the central nervous system, the potential application of these compounds was also explored in the field of phytopathology through the evaluation of their activity against grapevine downy mildew, a major socio-economic challenge in the Bordeaux region. Finally, a second component of this thesis, developed in parallel with the main project, led to the first asymmetric total synthesis of (+)- and (–)-myrioneurinols through the combination of two transformations developed within the laboratory: a chlorinative alkynyl-Prins reaction and a late-stage resolution strategy based on the Barton–McCombie reaction.
Block copolymer templating of nanostructured patches on nanoparticles
by Alexia FREIRE DE ARAUJO (Laboratoire de Chimie des Polymères Organiques)
The defense will take place at 14h00 - Amphithéâtre Centre de Recherche Paul Pascal - 115 Avenue du Dr Albert Schweitzer, 33600 Pessac
in front of the jury composed of
- Serge RAVAINE - Professeur des universités - Centre de Recherche Paul Pascal (CRPP) - Université de Bordeaux - Directeur de these
- Muriel LANSALOT - Directrice de recherche - Catalyse, Polymérisation, Procédés et Matériaux (CP2M) - Université Lyon 1 - Rapporteur
- André GRÖSCHEL - Professeur - University of Bayreuth - Rapporteur
- Guillaume FLEURY - Professeur des universités - Laboratoire de Chimie des Polymères Organiques (LCPO) - Université de Bordeaux - CoDirecteur de these
- Mona TRÉGUER-DELAPIERRE - Professeure des universités - Institut de Chimie de la Matière Condensée de Bordeaux (ICMCB) - Université de Bordeaux - Examinateur
- Raphaël FEOUGIER - Ingénieur de recherche - Laboratoire d'électronique des technologies de l'information (CEA-Leti) - Examinateur
Patchy nanoparticles (NPs) exhibit directional interactions and controllable surface valence, enabling the bottom-up assembly of complex colloidal structures. The formation of targeted colloidal architectures relies on precise control over the characteristics of the patches, including their size, shape, orientation and spatial arrangement. However, conventional nanofabrication strategies for patchy NPs face an inherent trade-off between resolution and patch geometrical complexity. Top-down lithographic approaches remain primarily confined to planar substrates and are limited by optical diffraction, whereas bottom-up chemical methods provide limited spatial control at length scales below 50 nm and often insufficient pattern fidelity on curved surfaces. To overcome these limitations, a versatile nanofabrication strategy was developed by combining the self-assembly properties of block copolymers (BCPs) and inorganic colloids. Monodisperse silica nanoparticles with diameters of 32, 96 and 545 nm were synthesized and subsequently organized either into close-packed monolayers using the Langmuir-Blodgett technique or into monolayers of isolated particles through electrostatic adsorption. In parallel, PS-b-PMMA thin films exhibiting various morphologies were self-assembled on a water-soluble sacrificial layer. Using this sacrificial membrane, the nanostructured PS-b-PMMA films were subsequently conformally transferred onto the surface of silica nanoparticle monolayers. Selective hybridization of the PMMA domains through the infiltration of metal oxide precursors using sequential infiltration synthesis, followed by oxygen plasma etching to selectively remove the polymer matrix, enabled the nanoscale patterning of the nanoparticle surfaces with alumina patches arranged according to a geometric configuration dictated by the morphology of the BCP film. By taking into account the geometrical commensurability between the nanoparticle diameter and the intrinsic period of the block copolymer, three different self-assembled BCP film morphologies enabled the fabrication of three distinct patch architectures on the NP surfaces: (i) hexagonal arrays of dots, (ii) hexagonal honeycomb arrays, and (iii) line/space patterns. The formation of these complex nanoparticle architectures featuring nanostructured patches was confirmed by GISAXS, AFM, SEM, TEM, and EDX. Finally, nanoparticle recovery and the selective functionalization of the patches were demonstrated, highlighting the potential of these building blocks for programmed self-assembly of complex colloidal structures.
ED Sciences Physiques et de l'Ingénieur
Model-based estimation of remaining useful life for predictive maintenance applied to lithium-ion batteries
by Antoine GAIGNARD (Laboratoire de l'Intégration du Matériau au Système)
The defense will take place at 9h30 - Amphi J.P.DOM A0.8 351 cours de la libération, 33405 Talence CEDEX, France Laboratoire IMS, bâtiment A31
in front of the jury composed of
- Franck CAZAURANG - Professeur des universités - Université de Bordeaux - Directeur de these
- Christophe BéRENGUER - Professeur des universités - Université Grenoble Alpes - Rapporteur
- Didier THEILLIOL - Professeur des universités - Université de Lorraine - Rapporteur
- Christophe FARGES - Maître de conférences - Université de Bordeaux - CoDirecteur de these
- Mitra FOULADIRAD - Professeure des universités - Aix Marseille Université - Examinateur
- Antoine PICOT - Maître de conférences - Université de Toulouse - Examinateur
The decarbonization of transportation and energy systems relies heavily on lithium-ion batteries, whose widespread use is accompanied by a growing need for state-of-health monitoring. The inevitable aging of batteries, which results in a progressive loss of capacity and power, along with the safety risks that may arise from it, makes it necessary to estimate the remaining useful life in order to anticipate maintenance and system replacement. This thesis proposes and compares two original approaches for estimating the remaining useful life of lithium-ion batteries, distinguished by their level of complexity, their data requirements, and their capacity for physical interpretation and generalization. The first approach is a so-called model-free method: the capacity measured over the cycles is approximated by a third-order polynomial, whose coefficients are estimated through least-squares optimization under linear constraints, updated at each cycle. Applied to experimental data from the database of the CALCE research team at the University of Maryland (USA), this method requires neither an electrochemical model nor an extensive aging history from other batteries, making it a solution with low computational and data costs, particularly well-suited to embedded implementation, although its accuracy remains limited in the very early stages of life, as does its adaptability to changing operating conditions. The second approach is based on an electrochemical model combining a charge-discharge model, describing the cell's behavior at the scale of a single cycle, with an aging model, accounting for the drift of parameters over the course of cycling. These two models are jointly initialized and identified using an extended Kalman filter based on the available measurements, and are then used in a prognostic step to predict the battery's future degradation and derive the RUL from it. Several charge-discharge and aging model formulations from the literature are compared, and the method is evaluated on experimental data from the Faraday Institution at multiple temperatures. The performance of both approaches is compared with each other as well as with several data-driven methods from the literature (recurrent neural networks, LSTM in particular), using standard prognostic metrics. The results show that the polynomial approach, despite its simplicity, achieves performance comparable to that of much more computationally expensive data-driven methods, while the electrochemical model-based approach offers better extrapolation capability and a physical interpretation of degradation mechanisms. This work highlights the complementarity of these two families of methods and proposes extensions to other datasets, other battery chemistries, as well as other types of modeling.