Search

Cellular morphology can be used to guide the de novo design of small molecules inducing a desired phenotype.

Morphological profiling is a powerful technology that enables unbiased characterization of cellular states through image-based screening. Inspired by recent progress in self-supervised learning (SSL), we sought to explore the potential benefits of using SSL in this domain and conducted a comprehensive benchmark study of recent SSL methods for learning representations from Cell Painting images without segmentation. We trained DINO, MAE, and SimCLR on subsets of the JUMP-CP consortium data, one of the largest publicly available Cell Painting image sets, and observed improved model performance with larger and more heterogeneous training sets. Our best model (DINO) surpassed the widely used profiling tool CellProfiler by 29% in mean average precision (mAP) on classifying chemical perturbations and significantly accelerated feature extraction by 50x, at a lower cost. Moreover, DINO outperformed CellProfiler in clustering gene families on an independent gene overexpression dataset. Our findings indicate that SSL methods can improve the efficiency and performance of morphological profiling, offering the potential to expedite drug discovery and reduce compute costs.

Image-based profiling has emerged as a powerful technology for various steps in basic biological and pharmaceutical discovery, but the community has lacked a large, public reference set of data from chemical and genetic perturbations. Here we present data generated by the Joint Undertaking for Morphological Profiling (JUMP)-Cell Painting Consortium, a collaboration between 10 pharmaceutical companies, six supporting technology companies, and two non-profit partners. When completed, the dataset will contain images and profiles from the Cell Painting assay for over 116,750 unique compounds, over-expression of 12,602 genes, and knockout of 7,975 genes using CRISPR-Cas9, all in human osteosarcoma cells (U2OS). The dataset is estimated to be 115 TB in size and capturing 1.6 billion cells and their single-cell profiles. File quality control and upload is underway and will be completed over the coming months at the Cell Painting Gallery:https://registry.opendata.aws/cellpainting-gallery. A portal to visualize a subset of the data is available athttps://phenaid.ardigen.com/jumpcpexplorer/.

Target residence time is emerging as an important optimization parameter in drug discovery, yet target and off-target engagement dynamics have not been clearly linked to the clinical performance of drugs. Here we developed high-throughput binding kinetics assays to characterize the interactions of 270 protein kinase inhibitors with 40 clinically relevant targets. Analysis of the results revealed that on-rates are better correlated with affinity than off-rates and that the fraction of slowly dissociating drug-target complexes increases from early/preclinical to late stage and FDA-approved compounds, suggesting distinct contributions by each parameter to clinical success. Combining binding parameters with PK/ADME properties, we illustrate in silico and in cells how kinetic selectivity could be exploited as an optimization strategy. Furthermore, using bio- and chemoinformatics we uncovered structural features influencing rate constants. Our results underscore the value of binding kinetics information in rational drug design and provide a resource for future studies on this subject.

Developing new small molecules that are bioactive is time-consuming, costly and rarely successful. As a mitigation strategy, we apply, for the first time, generative adversarial networks to de novo design of small molecules using a phenotype-based drug discovery approach. We trained our model on a set of 30,000 compounds and their respective morphological profiles extracted from high content images; no target information was used to train the model. Using this approach, we were able to automatically design agonist-like compounds of different molecular targets.

Mitochondria are semi-autonomous organelles that supply energy for cellular biochemistry through oxidative phosphorylation. Within a cell, hundreds of mobile mitochondria undergo fusion and fission events to form a dynamic network. These morphological and mobility dynamics are essential for maintaining mitochondrial functional homeostasis, and alterations both impact and reflect cellular stress states. Mitochondrial homeostasis is further dependent on production (biogenesis) and the removal of damaged mitochondria by selective autophagy (mitophagy). While mitochondrial function, dynamics, biogenesis and mitophagy are highly-integrated processes, it is not fully understood how systemic control in the cell is established to maintain homeostasis, or respond to bioenergetic demands. Here we used agent-based modeling (ABM) to integrate molecular and imaging knowledge sets, and simulate population dynamics of mitochondria and their response to environmental energy demand. Using high-dimensional parameter searches we integrated experimentally-measured rates of mitochondrial biogenesis and mitophagy, and using sensitivity analysis we identified parameter influences on population homeostasis. By studying the dynamics of cellular subpopulations with distinct mitochondrial masses, our approach uncovered system properties of mitochondrial populations: (1) mitochondrial fusion and fission activities rapidly establish mitochondrial sub-population homeostasis, and total cellular levels of mitochondria alter fusion and fission activities and subpopulation distributions; (2) restricting the directionality of mitochondrial mobility does not alter morphology subpopulation distributions, but increases network transmission dynamics; and (3) maintaining mitochondrial mass homeostasis and responding to bioenergetic stress requires the integration of mitochondrial dynamics with the cellular bioenergetic state. Finally, (4) our model suggests sources of, and stress conditions amplifying, cell-to-cell variability of mitochondrial morphology and energetic stress states. Overall, our modeling approach integrates biochemical and imaging knowledge, and presents a novel open-modeling approach to investigate how spatial and temporal mitochondrial dynamics contribute to functional homeostasis, and how subcellular organelle heterogeneity contributes to the emergence of cell heterogeneity.

In this work we present a mathematical model describing the coupled heat and moisture transport in paper. The model is solved numerically and the numerical solution is used to study the interdependency of the moisture and temperature distribution in paper. The results show that variation with temperature of the saturated water vapor concentration and the sorption isotherm parameters are both important for inducing moisture desorption. It is also found that for steep relative humidity ramps moisture sorption generates temperature increments that slow down the sorption process itself. The model is also used to study the moisture gradients in a paper sheet inside a printer from Oce Technologies, which contains a warm print surface. The results predict changes in moisture content of only 0.2%, which suggests that no deformations are induced on the printed sheet.

Background TFEB (transcription factor EB) regulates metabolic homeostasis through its activation of lysosomal biogenesis following its nuclear translocation. TFEB activity is inhibited by mTOR phosphorylation, which signals its cytoplasmic retention. To date, the temporal relationship between alterations to mTOR activity states and changes in TFEB subcellular localization and concentration has not been sufficiently addressed. Methods mTOR was activated by renewed addition of fully-supplemented medium, or inhibited by Torin1 or nutrient deprivation. Single-cell TFEB protein levels and subcellular localization in HeLa and MCF7 cells were measured over a time course of 15 hours by multispectral imaging cytometry. To extract single-cell level information on heterogeneous TFEB activity phenotypes, we developed a framework for identification of TFEB activity subpopulations. Through unsupervised clustering, cells were classified according to their TFEB nuclear concentration, which corresponded with downstream lysosomal responses. Results Bulk population results revealed that mTOR negatively regulates TFEB protein levels, concomitantly to the regulation of TFEB localization. Subpopulation analysis revealed maximal sensitivity of HeLa cells to mTOR activity stimulation, leading to inactivation of 100 % of the cell population within 0.5 hours, which contrasted with a lower sensitivity in MCF7 cells. Conversely, mTOR inhibition increased the fully active subpopulation only fractionally, and full activation of 100 % of the population required co-inhibition of mTOR and the proteasome. Importantly, mTOR inhibition activated TFEB for a limited duration of 1.5 hours, and thereafter the cell population was progressively re-inactivated, with distinct kinetics for Torin1 and nutrient deprivation treatments. Conclusion TFEB protein levels and subcellular localization are under control of a short-term rheostat, which is highly responsive to negative regulation by mTOR, but under conditions of mTOR inhibition, restricts TFEB activation in a manner dependent on the proteasome. We further identify a long-term, mTOR-independent homeostatic control negatively regulating TFEB upon prolonged mTOR inhibition. These findings are of relevance for developing strategies to target TFEB activity in disease treatment. Moreover, our quantitative approach to decipher phenotype heterogeneity in imaging datasets is of general interest, as shifts between subpopulations provide a quantitative description of single cell behaviour, indicating novel regulatory behaviors and revealing differences between cell types. Electronic supplementary material The online version of this article (doi:10.1186/s12885-016-2388-9) contains supplementary material, which is available to authorized users.