ED Sciences Physiques et de l'Ingénieur
Numerical tools for the robust sensing of glucidic homeostasis using Multi-Organ-On-Chip technology.
by Roland GIRAUD (Laboratoire de l'Intégration du Matériau au Système)
The defense will take place at 9h30 - Amphitéatre Jean-Paul DOM Laboratoire IMS, Bâtiment A31, 351 Cours de la Libération, 33405 Talence Cedex, France
in front of the jury composed of
- Sylvie RENAUD - Professeure des universités - Bordeaux INP - Directeur de these
- Virginie HOEL - Professeure des universités - Université de Lilles - Rapporteur
- Laurent PEYRODIE - Maître de conférences - JUNIA - Rapporteur
- Hamida HALLIL ABBAS - Professeure des universités - Université de Bordeaux - Examinateur
- Eric LE CARPENTIER - Maître de conférences - Ecole Centrale de Nantes - Examinateur
Diabetes is a major public health challenge, characterized by a dysregulation of glucose homeostasis and an impaired ability of the body to control blood glucose levels. At the heart of this regulation, pancreatic islets play an essential role by adapting insulin secretion to changes in glucose concentration. A better understanding of their function, their response to stimulation, and their interactions with other metabolic tissues is therefore an important objective for diabetes research and for the development of new experimental tools. In this context, microphysiological systems, such as organ-on-chip and multi-organ-on-chip platforms, offer new possibilities. These platforms allow living tissues to be cultured in miniaturized, controlled, and dynamic environments, while reproducing certain conditions close to those encountered in the body. In the context of glucose homeostasis, they can integrate several biological tissues or micro-organs, in separate compartments, such as pancreatic islets, liver, muscle, or adipose tissue, in order to study their individual responses and their interactions under microfluidic perfusion. However, these devices also introduce important constraints for biological measurements. Fluid circulation, the use of pumps, valves, or collectors can generate perturbations that complicate the interpretation of recorded signals. In addition, the number of on-line sensors is often limited, especially when several tissues must be monitored simultaneously. The resulting measurements may therefore be noisy, variable, and sometimes incomplete. This requires the development of robust digital methods capable of extracting reliable information from complex experimental data. This thesis addresses this issue by analyzing the electrophysiological monitoring of pancreatic islets in a microphysiological environment. The collective electrical activity of beta cells, recorded using microelectrode arrays, provides valuable information about the functional state of the islet in response to glucose. The objective of this work is to propose tools to exploit these signals in a more robust and informative way, considering the experimental constraints imposed by microfluidic platforms. After a presentation of the context, the manuscript address the extraction of a simple and useful indicator of islet activity: the frequency of slow potentials. The objective is to evaluate different signal-processing strategies in order to identify those that remain reliable under real experimental conditions. The following part seeks to better understand what an electrode actually measures when recording islet activity. Using high-density recordings, the spatial organization of the electrophysiological response is studied and shows that islet activity is not homogeneous. This part therefore highlights the limitations of a description based only on a local measurement or on a single global indicator. The third and last part proposes a complementary approach to describe the islet activity in a richer way. Its objective is to move beyond the dominant frequency alone by constructing a multidimensional representation of the signal, and then using unsupervised classification methods to identify different activity states over time. Overall, this work proposes a progression from the robust measurement of a simple marker toward a more complete description of the bioelectrical behavior of pancreatic islets. It therefore contributes to the development of digital tools adapted to microphysiological platforms, with the ambition of making measurements more reliable, more interpretable, and better suited to the study of glucose homeostasis and diabetes.
ED Sciences et environnements
Evaluate the low latitude hydrological cycle in numerical climate models by constraining past ocean density and salinity changes.
by Héloïse BARATHIEU (Environnements et Paléoenvironnements Océaniques et Continentaux)
The defense will take place at 14h00 - Amphithéâtre B18 UMR CNRS 5805 EPOC – OASU Université de Bordeaux Site de Pessac – Bâtiment B18N Allée Geoffroy Saint-Hilaire CS 50023 33615 PESSAC CEDEX FRANCE
in front of the jury composed of
- Thibaut CALEY - Chargé de recherche - Université de Bordeaux - Directeur de these
- Didier PAILLARD - Directeur de recherche - LSCE - Rapporteur
- Stéphanie DUCHAMP-ALPHONSE - Professeure - GEOPS - UMR n°8148 - Rapporteur
- Thibault DE GARIDEL-THORON - Directeur de recherche - CEREGE - Rapporteur
- Masa KAGEYAMA - Directrice de recherche - LSCE - Examinateur
- Bruno MALAIZÉ - Professeur - UMR EPOC - Examinateur
The hydrological cycle plays an important role in Earth's climate and is vital for human populations. Significant uncertainties remain regarding projections of potential future changes in the hydrological cycle at low latitudes, particularly in monsoon regions, as highlighted in the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). In this context, sea surface salinity serves as a natural tracer of the hydrological cycle, as its distribution is closely linked to the evaporation–precipitation balance. Because it integrates the combined effects of temperature and salinity, surface density represents a particularly comprehensive indicator for studying interactions between freshwater fluxes, heat exchanges, and ocean circulation. This PhD thesis is part of the ANR HYDRATE project, which aims to better constrain hydrological cycle variability at low latitudes by combining paleoclimate reconstructions of ocean surface density and salinity with numerical climate modelling. The study of past climates makes it possible to explore climate responses to a variety of natural perturbations, to evaluate climate model performance beyond the instrumental period, and to potentially reduce uncertainties in future projections. As part of this PhD thesis, δ¹⁸O and Mg/Ca ratio measurements were performed on the planktonic foraminifera Globigerinoides ruber from several marine sediment cores in the Indian Ocean. δ²H measurements of alkenones were also carried out on core tops from this region. Bayesian age models (BACON) were developed for all cores based on radiocarbon dating, covering the past 23,000 years. In parallel with these paleo–observations, syntheses of climate model simulation results were produced using the databases of the Paleoclimate Modelling Intercomparison Project (PMIP 3 and 4). The main results obtained are as follows: 1) A Bayesian regression model enabling quantitative reconstruction of annual ocean surface paleodensity was developed and applied within the HYDRATE project, and used in this PhD thesis. 2) Reconstructions of ocean surface paleodensity were used, for the first time, to evaluate climate simulations of the Last Glacial Maximum (LGM, ~21,000 years BP). We show that, at the global scale, the simulations correctly reproduce the major features of density differences between the LGM and the preindustrial period, although limitations remain in the representation of regional processes, particularly those related to the hydrological cycle and ocean–atmosphere interactions. 3) The statistical method of observational constraints was used to reduce uncertainties, by more than 45%, in future projections of South Asian monsoon precipitation for the period 2070–2100. The specificity of this approach lies in the combined use of historical observations and paleoclimate data to constrain these projections. 4) Based on quantitative reconstructions of surface paleodensity, quantitative reconstructions of paleoprecipitation were produced for the Indian Ocean monsoon systems (South Asian and South African) over the past 23,000 years, using a new data–model integration approach. These results provide improved understanding of past monsoon dynamics and forcings. Finally, this PhD thesis opens perspectives for new methods and approaches for the quantitative reconstruction of past ocean surface salinity, which could strengthen model evaluation and improve understanding of the tropical hydrological cycle.