Living Systems (United States)

The human heart, originating from the splanchnic mesoderm, is the first functional organ to develop, co-evolving with the foregut endoderm through reciprocal signaling. Previously, cardioid models offered new insights on cardiovascular cell lineages and tissue morphogenesis during heart development, while mesoderm-endoderm crosstalk remain incompletely understood. Here, we integrated micropatterned cardioids, CRISPR-engineered reporter hiPSCs, deep-tissue imaging, and single-cell RNA sequencing (scRNA-seq) to explore synergistic mesoderm-endoderm co-development. scRNA-seq with PHATE trajectory mapping reconstructed lineage bifurcations of mesoderm-heart and endoderm-foregut lineages, identifying key cell types in cardiac and hepatic development. Ligand-receptor interaction analysis highlighted mesodermal cells enriched in non-canonical WNT, NRG, and TGF-β signaling, while endodermal cells exhibited VEGF and Hedgehog activity. We found that micropattern sizes influenced cellular composition, cardioid cavitation, contractile functions, and mesoderm-endoderm signaling crosstalk. The cardioids generated from 600µm diameter circle patterns showed larger cavity formation resembling early heart chamber formation. Our findings establish micropatterned cardioids as a model for mesoderm-endoderm co-development, enhancing our understanding of heart-foregut synergy during early embryogenesis.

As AI systems become more capable, it is important that their decisions are understandable and aligned with human expectations. A key challenge is the lack of interpretability in deep models. Existing methods such as GradCAM generate heatmaps but provide limited conceptual insight, while prototype-based approaches offer example-based explanations but often rely on rigid region selection and lack semantic consistency. To address these limitations, we propose PCMNet, a Part-Prototypical Concept Mining Network that learns human-comprehensible prototypes from meaningful regions without extra supervision. By clustering these into concept groups and extracting concept activation vectors, PCMNet provides structured, concept-level explanations and enhances robustness under occlusion and adversarial conditions, which are both critical for building reliable and aligned AI systems. Experiments across multiple benchmarks show that PCMNet outperforms state-of-the-art methods in interpretability, stability, and robustness. This work contributes to AI alignment by enhancing transparency, controllability, and trustworthiness in modern AI systems.

Patient-centered outcomes are increasingly recognized by the transplant community as a key component to improving knowledge and patient counseling, and long-term outcomes after living kidney or liver donation are not well understood. With efforts underway to follow living donors and donor candidates long term, it is critical that living donors and candidates drive the process of patient-centered data collection. Living donors on the Scientific Registry of Transplant Recipients' Living Donor Steering Committee were engaged to brainstorm and prioritize which data elements should be captured. Committee members identified long-term lab values, difficulty with normal activities, and diagnosis of chronic kidney or liver disease as critical long-term outcomes. Additional high-priority elements included short-term lab values, deaths related to donation, mortality among donors compared to nondonors, hospital readmissions, costs and physical limitations related to donation, and whether a donor would donate again. Finally, living donors wanted to know the risk of organ failure among donors compared to nondonors. This prioritization exercise provided valuable insights from living donors to help determine key themes and discrete data elements that should be prioritized for long-term collection and reporting to living donors and living donor candidates.

Essential nuclear processes require pairs of chromosomal loci to find each other in three-dimensional space. Polymer models of chromosome dynamics typically assume that the stochastic forces driving such locus motion are spatially uncorrelated, implying that relative diffusion follows directly from single-locus dynamics. Here we show that this assumption fails in living cells. Using live-cell imaging in fly embryos and mouse embryonic stem cells, we find that pairwise locus distances diffuse markedly slower than predicted for independent fluctuations. Combining stochastic trajectory analysis with polymer simulations, we demonstrate that this slowdown arises from non-equilibrium spatially correlated fluctuations (SCFs) in the nucleoplasm, which cause nearby loci to move coherently. We establish three experimentally testable signatures of SCFs: fluctuation amplitudes plateau at large distances, are independent of genomic separation, and show an anomalous temporal scaling. All three predictions are confirmed experimentally, including for loci on separate chromosomes. ATP depletion and disruption of cohesin-mediated loop extrusion reveal that both active processes and crosslinking contribute to correlation magnitudes. Because SCFs slow relative motion preferentially at short distances, they reduce encounter frequencies while prolonging encounter durations, generating a trade-off with direct implications for gene regulation. Our results identify spatially correlated fluctuations as a fundamental determinant of relative motion in confined active polymers.

Phenotypic heterogeneity in microbes can be a double-edged sword — boosting individual survival in uncertain environments, but potentially compromising beneficial population coordination. Dissecting how microbes resolve this tradeoff is challenging because interactions span multiple scales—from molecular interactions to single-cell behavior, population dynamics, and environmental feedback. Here, we address this question for the Escherichia coli chemotaxis system, which implements both individual motile explorations and collective resource exploitations using the same cellular machinery. We quantify the heterogeneity of individual sensory and behavioral phenotypes, as well as the abundance of key signaling proteins during growth in various environments, and test their impact on population-scale collective migration. We identify growth rate as a key environment-dependent physiological variable governing not only the mean but also the variance of sensory phenotypes. Remarkably, rather than hindering population coordination, we find that sensory diversity benefits collective chemotactic navigation in uncertain environments. Strong heterogeneity of expressed phenotypes enhances readiness to multiple sensory cues, and the required population coordination is achieved by phenotypic filtering of that diversity by the collective behavior itself. These results reveal a sensory bet-hedging strategy for collective navigation during population growth: diversity in sensitivity to nutrients currently being consumed is reduced to promote focused exploitation, while it is increased for nutrients not yet encountered to enhance exploration for new growth opportunities.

Abstract A dynamic actin cytoskeleton, characterized by rapid filament turnover, is essential for driving intracellular transport and cellular movement. Despite extensive study, the contributions of many actin-binding proteins (ABPs) that regulate the different steps of the actin turnover cycle (filament assembly, disassembly and recycling into polymerizable actin monomers) remain ill-defined. Here, we introduce novel sensitive in vitro assays to quantitatively assess how ABPs catalyze actin turnover. By accurately measuring nucleotide exchange dynamics and ATP consumption, these assays enable robust characterization of ABP activity across broad concentration ranges. Using these methods and modeling of these reactions, we systematically examined the contributions of five conserved regulators, both individually and in combination, and identified conditions that maximize turnover efficiency. We also determined that increasing F-actin concentration to cellular levels affects ABP activity. Finally, we demonstrated that rapid actin turnover is preserved during encapsulation in cell-sized vesicles using the cDICE method. Together, these advances provide versatile tools and new insights into actin cytoskeletal dynamics under physiologically relevant conditions.

Molecular dynamics simulation is a powerful computational technique used for predicting and understanding the dynamic behavior of biomolecular systems. Steered molecular dynamics (SMD) simulations enable the study of force-induced processes in biomolecules, effectively mimicking single-molecule force spectroscopy experiments probing protein unfolding and receptor-ligand unbinding. Given the stochastic nature of these mechanical events, accurately exploring the dynamic behavior of biomolecules and extracting accurate physical information requires severalin-silicoexperiments. This includes performing many pulling simulations at different velocities or force loading rates. The large amount of data obtained from these simulation sets requires efficient automated data processing tools. We present PySteMoDA, a novel Python package with a user-friendly graphical interface specifically designed for constant-velocity SMD data analysis. The automated force peak detection methods reduce user bias, improve accuracy, and accelerate data analysis. The package also allows identification of residues involved in mechanical events through computation of the time-dependent mechanical work and correlation factors between residue pairs. This package not only addresses automated data processing in SMD simulations and accurate parameter extraction, but also significantly enhances accessibility and usability. Through PySteMoDA, users can efficiently analyze simulation data without the barrier of coding, facilitating a wider range of investigations and insights in the field of computational biochemistry and biophysics.

Abstract The protection of the respiratory tract relies on a layer of mucus that is transported along the epithelial surface by the beating of millions of microscopic cilia. This mechanism, called mucociliary clearance, is defective in chronic respiratory diseases. Since these pathologies alter mucus rheology, mucus transport efficiency is hypothesized to rely on its mechanical properties. Yet, this link is difficult to test due to limited experimental models. Here, we introduce an experimental framework that enables us to identify conditions associated with efficient mucus transport or arrest. Strikingly, we show that cilia are able to efficiently propel mucus with properties ranging from a viscoelastic fluid to an elastic solid, revealing that bulk mucus rheology is not the critical determinant of clearance failure. Instead, we demonstrate that the efficiency of mucociliary clearance is governed by the hydration of a thin fluid layer at the cilia-mucus interface. Finally, we use a hydrodynamic model informed by measurements on ciliary beat patterns to infer the properties of this critical layer under both transported and arrested mucus conditions. This work not only offers a novel understanding of the fundamental physical mechanism of mucus transport, but also provides a well-defined and quantitative assay to test effects of mucolytic agents or drugs on mucus clearance.

Neural dynamics change over time and with physiologic state. Modeling of neural dynamics can thus be understood at two levels: (i) identifying the latent process that governs how and when states change, and (ii) identifying the generative circuit mechanisms within each state. Here, we develop a data-driven modeling method that tackles these two levels simultaneously. We formulate a parametric network model of neural dynamics that embeds state-dependent modulation. The modulation itself is controlled by a latent switching process, modeled as a Hidden Markov Model (HMM). A key challenge is that the model itself has internal states that must be estimated from observed data. This leads to a triune optimization problem, consisting of model parameterization of the HMM and neural dynamics, alongside state estimation. Our method brings together several optimization frameworks alongside estimation-theoretic constructs to solve this problem efficiently, enabling blind identification of state transitions and neural dynamics. We validate this methodology on ground-truth data with known parameters, and find that it accurately infers the transitions in latent state and the dynamics of each state. We demonstrate its capability of inferring changes in brain dynamics from electrophysiological data by testing it on electroencephalography recordings with labeled state transitions. While similar methods exist to infer switches and dynamics on the level of individual neurons, there is no directly comparable method available for mesoscale modeling of neural circuits. Our methodology enables blind modeling of changing neural dynamics allowing for inference of modulatory circuit mechanisms.