Search

Abstract An unresolved issue in contemporary biomedicine is the overwhelming number and diversity of complex images that require annotation, analysis and interpretation. Recent advances in Deep Learning have revolutionized the field of computer vision, creating algorithms that compete with human experts in image segmentation tasks. However, these frameworks require large human-annotated datasets for training and the resulting “black box” models are difficult to interpret. In this study, we introduce Kartezio, a modular Cartesian Genetic Programming-based computational strategy that generates fully transparent and easily interpretable image processing pipelines by iteratively assembling and parameterizing computer vision functions. The pipelines thus generated exhibit comparable precision to state-of-the-art Deep Learning approaches on instance segmentation tasks, while requiring drastically smaller training datasets. This Few-Shot Learning method confers tremendous flexibility, speed, and functionality to this approach. We then deploy Kartezio to solve a series of semantic and instance segmentation problems, and demonstrate its utility across diverse images ranging from multiplexed tissue histopathology images to high resolution microscopy images. While the flexibility, robustness and practical utility of Kartezio make this fully explicable evolutionary designer a potential game-changer in the field of biomedical image processing, Kartezio remains complementary and potentially auxiliary to mainstream Deep Learning approaches.

Cytotoxic immune cells are endowed with a high degree of heterogeneity in their lytic function, but how this heterogeneity is generated is still an open question. We therefore investigated if human CD8+ T cells could segregate their lytic components during telophase, using imaging flow cytometry, confocal microscopy, and live-cell imaging. We show that CD107a+-intracellular vesicles, perforin, and granzyme B unevenly segregate in a constant fraction of telophasic cells during each division round. Mathematical modeling posits that unequal lytic molecule inheritance by daughter cells results from the random distribution of lytic granules on the two sides of the cleavage furrow. Finally, we establish that the level of lytic compartment in individual cytotoxic T lymphocyte (CTL) dictates CTL killing capacity.

Cytotoxic T lymphocytes recognise and eliminate tumour cells. Here, the authors show that on contact with these immune cells melanoma cells can resist T cell cytotoxicity by modulating the trafficking of their lysosomal compartment, this results in the degradation of the T cell protein perforin by the protease cathepsin B.

Interleukin-12 (IL-12), produced by dendritic cells in response to activation, is central to pathogen eradication and tumor rejection. The trafficking pathways controlling spatial distribution and intracellular transport of IL-12 vesicles to the cell surface are still unknown. Here, we show that intracellular IL-12 localizes in late endocytic vesicles marked by the SNARE VAMP7. Dendritic cells (DCs) from VAMP7-deficient mice are partially impaired in the multidirectional release of IL-12. Upon encounter with antigen-specific T cells, IL-12-containing vesicles rapidly redistribute at the immune synapse and release IL-12 in a process entirely dependent on VAMP7 expression. Consistently, acquisition of effector functions is reduced in T cells stimulated by VAMP7-null DCs. These results provide insights into IL-12 intracellular trafficking pathways and show that VAMP7-mediated release of IL-12 at the immune synapse is a mechanism to transmit innate signals to T cells.

The killing of antigen-bearing cells by clonal populations of cytotoxic T lymphocytes (CTLs) is thought to be a rapid phenomenon executed uniformly by individual CTLs. We combined bulk and single-CTL killing assays over a prolonged time period to provide the killing statistics of clonal human CTLs against an excess of target cells. Our data reveal efficiency in sustained killing at the population level, which relied on a highly heterogeneous multiple killing performance at the individual level. Although intraclonal functional heterogeneity was a stable trait in clonal populations, it was reset in the progeny of individual CTLs. In-depth mathematical analysis of individual CTL killing data revealed a substantial proportion of high-rate killer CTLs with burst killing activity. Importantly, such activity was delayed and required activation with strong antigenic stimulation. Our study implies that functional heterogeneity allows CTL populations to calibrate prolonged cytotoxic activity to the size of target cell populations.

The dynamics of the interaction between Cytotoxic T Lymphocytes (CTL) and tumor cells has been addressed in depth, in particular using numerical simulations. However, stochastic mathematical models that take into account the competitive interaction between CTL and tumors undergoing immunoediting, a process of tumor cell escape from immunesurveillance, are presently missing. Here, we introduce a stochastic dynamical particle interaction model based on experimentally measured parameters that allows to describe CTL function during immunoediting. The model describes the competitive interaction between CTL and melanoma cell nodules and allows temporal and two-dimensional spatial progression. The model is designed to provide probabilistic estimates of tumor eradication through numerical simulations in which tunable parameters influencing CTL efficacy against a tumor nodule undergoing immunoediting are tested. Our model shows that the rate of CTL/tumor nodule productive collisions during the initial time of interaction determines the success of CTL in tumor eradication. It allows efficient cytotoxic function before the tumor cells acquire a substantial resistance to CTL attack, due to mutations stochastically occurring during cell division. Interestingly, a bias in CTL motility inducing a progressive attraction towards a few scout CTL, which have detected the nodule enhances early productive collisions and tumor eradication. Taken together, our results are compatible with a biased competition theory of CTL function in which CTL efficacy against a tumor nodule undergoing immunoediting is strongly dependent on guidance of CTL trajectories by scout siblings. They highlight unprecedented aspects of immune cell behavior that might inspire new CTL-based therapeutic strategies against tumors.

Abstract Germinal centers (GCs) are the main site of T cell-dependent antibody responses. Upon antigen challenge, GCs comprise mostly B cells undergoing proliferation, somatic hypermutation and antigen-affinity selection. GC B cells down-modulate the expression of Bcl-2 protein and are highly sensitive to apoptosis to eliminate autoreactive or low-affinity cells. Bcl-2 is still expressed in a few GC cells, whose identity remains unclear. To address this issue, we examined by confocal microscopy the expression of Bcl-2 by different GC lymphocyte subsets in hyperplastic tonsils. We found that the vast majority of Bcl-2+ GC cells are T lymphocytes. Conversely, while in the mantle zone and in the interfollicular areas T cells are almost exclusively Bcl-2+, in the GC, most T lymphocytes are Bcl-2-. In addition, most of the CD4+ GC T cells are Bcl-2-, while nearly 100% of the CD8+ GC T cells are Bcl-2+. The Bcl-2 downregulation by both B and CD4+ T GC cells supports the concept that these two subsets may undergo a selection process in this microenvironment. [ABSTRACT FROM AUTHOR]