T cell acute lymphoblastic leukaemia (T-ALL) is a rare disease that affects around 10-13 individuals in one million and affects pediatric and adult patients. High-dose chemotherapy is an adequate therapeutic treatment for most patients, however 10-20% of patients relapse and will require a bone marrow transplantation as a unique therapeutic window. T-ALL is a very aggressive disease and the rapidness in the response is crucial, however, one problem is that we cannot predict which patients will relapse. Several studies have been done to understand the origin of relapse, and there is evidence for several strategies for leukemia evolution: a) the relapse clone already exists at the time of diagnosis and is selected through treatment or b) treatment provides new mutations that will generate the relapse clone. Predicting the evolution of leukemic clones after chemotherapy pressure and understanding the specific mechanisms contributing to chemotherapy-resistant cell populations' appearance is necessary to improve cancer treatment. In our proposal, we wish to find feasible and clinically applicable ways to detect the earlier-stage clones that become resistant to chemotherapy. We wish to apply multiple approaches to better understand the specific clones like single-cell sequencing of primary T-ALL samples at diagnosis and relapse, and most importantly, compare their evolution with or without chemotherapy treatment through 2D, 3D culture and the current pdx in vivo model of leukaemia. Application of the 3D cell culture technique to T-ALL clonal evolution may represent a major scientific discovery to be used for different purposes focusing on disease modelling and relapse clone study. Our research will be pursued to optimize a method mimicking T-ALL in vitro, characterize the cells (and factors) causing the relapse, and use this information to identify new therapeutic options.

Animals and humans adapt to changes in the environment through the encoding and storage of previous experiences. Although associative learning involving a reinforcer has been the major focus in the field of cognition, other forms of learning are gaining popularity as they are likely more relevant and frequent in human daily choices. Indeed, associations between non-reinforcing stimuli represent the most evolutionarily advanced way to increase the chances of predicting future events and adapting individuals’ behavior. Animals are also able to form these higher-order conditioning processes, but more research is needed to understand how the brain encode and store these complex cognitive processes. In this project, I propose to study the role of hippocampo-cortical circuits in higher-order conditioning processes. These processes explain why subjects are often repulsed or attracted by stimuli, which do not have intrinsic repellent or appealing value and they were never explicitly paired with negative or positive outcomes. A proposed explanation of these “ungrounded” aversion or attraction is that these stimuli were incidentally associated with other cues directly reinforced, through a process called mediated learning (ML). However, with increased incidental associations, the subjects acquire more information, allowing them to separate the real saliences of the different stimuli. Therefore, ML evolves into “reality testing”(RT), a behavioral process that has been even less studied. These processes involve multiple brain regions and are characterized by accessible phases, making them perfect models to study the circuit-level regulation of complex behavior. By using genetic, pharmacological, imaging and mouse behavioral approaches (sensory preconditioning), HighMemory proposes to characterize at macro- (brain regions), meso- (cell-types) and micro-scale (activity changes), the causal involvement of hippocampo-cortical projections in higher-order conditioning processes.

Genes are fundamental units of life and their origin has fascinated researchers since the beginning of the molecular era. Many of the studies on the formation of new genes in genomes have focused on gene duplication and subsequent divergence of the two gene copies. But, in recent years, we have learnt that genes can also arise de novo from previously non-genic sequences. The discovery of de novo genes has become possible by the sequencing of complete genomes and the comparison of gene sets between closely related species. Here we wish to test a novel hypothesis, we propose that de novo gene formation dynamics in populations results in substantial differences in gene content between individuals. If they exist, these differences would be not be visible by the current methods to study gene variation, which are based on the comparison of the sequences of each individual to a common set of reference genes. To test our hypothesis, we will need to develop novel computational approaches to first obtain an accurate representation of all transcripts and translated open reading frames in each individual, and then integrate the information at the population level. We propose to apply these methods to two very distinct biological systems, a large collection of Saccharomyces cerevisiae world isolates and a human lymphoblastoid cell line (LCL) panel. For this, we will collect and generate RNA (RNA-Seq) and ribosome profiling (Ribo-Seq) sequencing data. In order to identify de novo originated events occurred within populations, as opposed to phylogenetically conserved genes that have been lost in some individuals, we will also generate similar data from a set of closely related species in each of the two systems. Combined with genomics data, we will identify the spectrum of mutations associated with de novo gene birth with an unprecedented level of detail and uncover footprints of adaptation linked to the birth of new genes.

Pregnancy involves biological adaptations that are necessary for the onset, maintenance and regulation of maternal behavior. We were the first group to find (1, 2) that pregnancy is associated with consistent, pronounced and long-lasting reductions in cerebral gray matter (GM) volume in areas of the social-cognition network. The aim of BEMOTHER is to develop an integrative model of the adaptations for motherhood that occur during pregnancy and the postpartum period by: i) establishing when the brain of pregnant women begins to change and how it evolves; ii) characterizing the dynamics of cognitive performance, theory-of-mind, maternal-infant bonding and psychiatric measures; iii) assessing the effect of environmental and/or psychological factors in the maternal adaptations, iv) identifying the metabolomics biomarkers associated with maternal adaptations, and v) integrating the previous findings within the Research Domain Criteria framework (RDoC) (3). We will use a prospective longitudinal design at 5 time points (1 pre-pregnancy session, 2 intra-pregnancy sessions and 2 postpartum sessions) during which neuroimaging, psychological, behavioral and metabolomics data will be acquired in 3 groups of women: a group of nulliparous women who will be undergoing a full-term pregnancy, another group of nulliparous women whose same-sex partners will undergo a full-term pregnancy, and a group of control nulliparous women. We will provide the longitudinal RDoC-based model at the end of the study, but we will also deliver intermediate longitudinal evaluations after the postpartum session, as well as cross-sectional analyses after the first intra-pregnancy session and the postpartum session. BEMOTHER is timely and innovative. It adopts the translational RDoC framework in order to provide a pioneering, comprehensive and dynamic characterization of the adaptations for motherhood, addressing the interaction among different functional domains at different levels of analysis.

Metastasis is the major cause of death in cancer patients due to cancer cell dissemination to distant organs. Cell plasticity is a core characteristic of metastatic cells and confers cellular adaptation capabilities to variable mechano-chemical tissue microenvironments. However, to date, specific quantitative measures of cancer cell plasticity associated with tumour aggressiveness and therapy resistance have remained difficult to establish. A major limitation is the availability of highthroughput multiplexed assays that can capture phenotypic heterogeneity and morphodynamic plasticity at the single cell level in standardized 3D culture conditions reflecting in vivo tissue microenvironments. The PLAST_CELL interdisciplinary consortium will pioneer the development of a microfluidics-based imaging platform to categorize and score cancer cell plasticity within diverse physiologically relevant 3D biomimetic culture conditions. The platform will enable to perform single cell multi-scale morphometric and molecular live cell data collection (PLAST_DATA) with minimal sample size (<10k cells). Data will be generated based on single molecule-sensitivity marker detection and cellular/subcellular morphodynamic feature recognition via minimal-invasive long-term super-resolution microscopy and parallel morphodynamic imaging of cellular behaviour. Computational integration of PLAST_DATA using preclinical models and patient samples will enable to develop a quantitative classification of tumour cell plasticity and predictive scoring of cancer aggressiveness, metastasis and drug resistance (PLAST_SCORE). The ability to assess cell plasticity based on cellular behaviours is beyond current clinical parameters and will strongly impact diagnosis, prognosis and treatments. The PLAST_CELL platform will be a technology breakthrough to establish new quantitative standards to evaluate cell plasticity and mechanisms of tumour malignancy for a new era of basic research and personalized medicine.