Periodic Reporting for period 2 - CROSSINGSCALES (Reverse Scale-Crossing Effects In Biology)
Reporting period: 2022-03-01 to 2023-08-31
The Advanced Grant CROSSINGSCALES, had as aim to establish the importance of reverse causal effects in biological systems, which arise due to the fact that at different biological scales, new properties in form and function arise that have a superseding causal impact on the behaviour of the lower-scale components from which these new properties emerge. These properties relate to the physicochemical state of the cell and the multicellular context, the spatial organization of components across multiple scales, mechanical properties, the abundance and content of organelles, their interactions, etc. Their top-down or reverse scale-crossing effects must be taken into account in order to make predictions about spatiotemporally controlled single-cell fates, activities, levels of gene expression, or the functional outcome of cellular signalling. To achieve this, we need quantitative imaging methods that can reveal, within the same sample, multiple properties of the cellular state, of signaling activities, and of cellular decisions across multiple length scales, and we need computer vision approaches to extract information from these multiple scales in an unsupervised manner and in ways that allow us to compare conditions, such as different timepoints or different perturbations. The overall goals of the project are to develop such approaches and to apply them to cultured mammalian cells, multicellular systems grown in vitro (organoids) and early embryos (zebrafish) in order to test whether the data they generate allow predicting cellular activities and decision-making. Having such data is important for bioscience as a whole as contextual information provided by reverse causal effects may provide the missing link to predict the behavior of biological systems.
In the first part, we focused on whether reverse causal effects may have obscured our ability to quantify how reliable individual cells can make cellular decisions. To study the information processing capacity of human cells, we conducted multiplexed quantification of signaling responses and markers of the cellular state. We discovered that signaling nodes in a network displayed adaptive information processing, which led to heterogeneous growth factor responses and enabled nodes to capture partially non-redundant information about the cellular state. We could show that collectively, as a multimodal percept, this gives individual cells a large information processing capacity to accurately place growth factor concentration within the context of their cellular state and make cellular state–dependent decisions. This indicates that heterogeneity and complexity in signaling networks may have coevolved to enable specific and context-aware cellular decision-making in a multicellular setting. This work was published in Kramer at al., 2022 (PMID 35857483). We also developed new computational methods in order to better compare the collective spatial organization of a large number of different proteins between conditions, as this holds enormous promise for understanding how spatial context shapes the activity of the genome and its products at multiple length scales. We introduced the deep learning framework called CAMPA (Conditional Autoencoder for Multiplexed Pixel Analysis), which uses a conditional variational autoencoder to learn representations of molecular pixel profiles that are consistent across heterogeneous cell populations and experimental perturbations. Clustering these pixel-level representations identifies consistent subcellular landmarks, which
can be quantitatively compared in terms of their size, shape, molecular composition and relative spatial organization. Using high-resolution multiplexed immunofuorescence, this reveals how subcellular organization changes upon perturbation of RNA synthesis, RNA processing or cell size, and uncovers links between the molecular composition of membraneless organelles and cell-to-cell variability in bulk RNA synthesis rates. By capturing interpretable cellular phenotypes, we anticipate that CAMPA will greatly accelerate the systematic mapping of multiscale atlases of biological organization to identify the rules by which context shapes physiology and disease. This work was published in Spitzer et al., 2023 (PMID 37248388). In addition, we also started to apply these approaches to in vitro grown multicellular systems, or organoids, and in particular to retinal organoids that recapitulate the development of the human retina created from human induced pluripotent stem cells. To study how context across scales influences the development of these systems, we generated multiplexed protein maps over a retinal organoid time course and primary adult human retinal tissue. We developed a toolkit to visualize progenitor and neuron location, the spatial arrangements of extracellular and subcellular components and global patterning in each organoid and primary tissue. In addition, we generated a single-cell transcriptome and chromatin accessibility timecourse dataset and inferred a gene regulatory network underlying organoid development. We integrated genomic data with spatially segmented nuclei into a multimodal atlas to explore organoid patterning and retinal ganglion cell (RGC) spatial neighborhoods, highlighting pathways involved in RGC cell death and showing that mosaic genetic perturbations in retinal organoids provide insight into cell fate regulation. This work was published in Wahle et al., 2023 (PMID 37156914).
can be quantitatively compared in terms of their size, shape, molecular composition and relative spatial organization. Using high-resolution multiplexed immunofuorescence, this reveals how subcellular organization changes upon perturbation of RNA synthesis, RNA processing or cell size, and uncovers links between the molecular composition of membraneless organelles and cell-to-cell variability in bulk RNA synthesis rates. By capturing interpretable cellular phenotypes, we anticipate that CAMPA will greatly accelerate the systematic mapping of multiscale atlases of biological organization to identify the rules by which context shapes physiology and disease. This work was published in Spitzer et al., 2023 (PMID 37248388). In addition, we also started to apply these approaches to in vitro grown multicellular systems, or organoids, and in particular to retinal organoids that recapitulate the development of the human retina created from human induced pluripotent stem cells. To study how context across scales influences the development of these systems, we generated multiplexed protein maps over a retinal organoid time course and primary adult human retinal tissue. We developed a toolkit to visualize progenitor and neuron location, the spatial arrangements of extracellular and subcellular components and global patterning in each organoid and primary tissue. In addition, we generated a single-cell transcriptome and chromatin accessibility timecourse dataset and inferred a gene regulatory network underlying organoid development. We integrated genomic data with spatially segmented nuclei into a multimodal atlas to explore organoid patterning and retinal ganglion cell (RGC) spatial neighborhoods, highlighting pathways involved in RGC cell death and showing that mosaic genetic perturbations in retinal organoids provide insight into cell fate regulation. This work was published in Wahle et al., 2023 (PMID 37156914).
Although not anticipated explicitly at the start of the project, we did realize that eventually, the ability to measure multiple features of biological systems across length scales implies that samples need to be fixed for processing and multiplexed imaging. This makes the approach incompatible with timelapse imaging, and thus misses out on a crucial temporal aspect of information. To overcome this, we developed a method, termed CellOT, which leverages the theory of optimal transport and the recent advent of convex neural architectures. CellOT is a framework for learning the response of individual cells to a given perturbation by coupling unpaired distributions. CellOT outperforms current methods at predicting single-cell drug responses, as profiled by scRNA-seq and a multiplexed protein imaging technology. This work has recently been accepted for publication (Bunne et al., 2023). Currently we are applying the technologies developed thus far and the insights gained to early embryo development, and have successfully set up multiplexed quantitative imaging in zebrafish embryos and stem-cell derived early embryo-like models. Moreover, we have by now established the ratioFISH method to perform multiplex single-molecule FISH at high throughput and are applying that to ask whether heterogeneity in nuclear and chromatin states reflects heterogeneity in cellular states and whether this enables individual cells to couple their cellular state to chromatin remodelling during contextual decision making.