NPJ Systems Biology and Applications最新文献

Production stability remains a major challenge in Chinese hamster ovary (CHO) cell-based therapeutic protein manufacturing, particularly during extended passaging where the underlying mechanisms of instability are not fully understood. Thus, in this study, we leveraged multivariate data analysis (MVDA) and flux balance analysis (FBA) with explainable AI (xAI) to mechanistically characterize the phenotypic differentiation between early (EP) and late passage (LP) of CHO cultures. Although EP and LP reached comparable peak viable cell densities, LP cultures exhibited a ~35% reduction in peak IgG titers and increased lactate and ammonia accumulation. Subsequent MVDA of temporal exometabolite profiles identified the exponential growth phase as the primary window of divergence, allowing us to interrogate metabolic rewiring via an FBA-xAI approach. This revealed that EP cells preferentially directed acetyl-CoA towards fatty acid biosynthesis to support proliferation. In contrast, LP prioritized oxidative stress mitigation by upregulating the trans-sulfuration pathway for de novo cysteine and glutathione synthesis while exhibiting heightened TCA cycle activity to maintain energy homeostasis. Overall, these mechanistic insights uncover a passage-associated shift from biosynthetic activity toward redox maintenance and identify the cysteine-glutathione axis as a critical metabolic lever for enhancing long-term stability and productivity in CHO cell culture.

Direct cellular reprogramming, converting one differentiated cell type directly into another, holds immense promise for regenerative medicine, developmental biology, and disease modeling. Identifying optimal transcription factor (TF) combinations to control this process remains complex and labor-intensive. Over the last decade, various computational tools emerged to infer TF sets for reprogramming. However, current methodologies possess critical limitations, and the absence of robust benchmarking standards makes it impossible to precisely validate and compare their performance. To address these challenges, we present a comprehensive analysis of existing computational methods for direct reprogramming and introduce a web application designed to support researchers in identifying and validating optimal TF sets. Our platform integrates predictions from established tools, incorporates a state-of-the-art Retrieval-Augmented Generation (RAG) system for efficient literature querying, and offers tools to further validate predictions. By providing a unified and interactive resource, our web application enhances the accessibility and efficiency of TF discovery for direct reprogramming. Furthermore, we discuss critical limitations shared by current methodologies and highlight the need for computational tools that can account for the complex regulatory dynamics of direct reprogramming. This work not only advances the toolkit available to researchers but also lays the groundwork for future innovations aimed at realizing the full potential of direct reprogramming.

Refeeding syndrome (RFS) is a life-threatening metabolic complication caused by refeeding in malnourished patients and is a barrier to early nutritional supplementation. Hypophosphatemia is a hallmark of RFS; however, its detailed pathogenic mechanism remains unclear. In this study, we aimed to unravel the underlying mechanism by generating a novel rat model that exhibits a decrease in plasma inorganic phosphorus levels comparable to that in patients with severe RFS and developing a mathematical model that reproduces the results. Although insulin is believed to be implicated in hypophosphatemia, we found a crucial role of amino acids. Furthermore, we inferred the dynamics of associated factors that play important roles in the regulation of hypophosphatemia, such as phosphate diuretic hormones, intracellular phosphorus content, and mTOR signaling, and experimentally confirmed them. Our findings provide new insights into the mechanisms underlying hypophosphatemia in RFS and should contribute to advancements in the prevention and intervention of RFS.

To assess cell migration in complex spatial environments, microfabricated chips, such as mazes and pillar forests, are routinely used to impose spatial and mechanical constraints, and cell trajectories are followed within these structures by advanced imaging techniques. In systems mechanobiology, computational models serve as essential tools to uncover how physical geometry influences intracellular dynamics; however, decoding such complex behaviors requires advanced inference techniques. Here, we integrated experimental observations of dendritic cell migration in a geometrically constrained microenvironment into a Cellular Potts model. We demonstrated that these spatial constraints modulate the motility dynamics, including speed and directional changes. We show that classical summary statistics, such as mean squared displacement and turning angle distributions, can resolve key mechanistic features but fail to extract richer spatiotemporal patterns, limiting accurate parameter inference. To solve this, we applied neural posterior estimation with in-the-loop learning of summary features. This learned summary representation of the data enables robust and flexible parameter inference, providing a data-driven framework for model calibration and advancing quantitative analysis of cell migration in structured microenvironments.

The interaction between a class I peptide-major histocompatibility complex (pMHC) and a T cell receptor (TCR) plays a central role in the elicitation of CD8⁺ T cell immune responses. As a result, considerable effort has been invested in understanding the structural, dynamic, and biophysical parameters that govern this recognition event, including designing altered peptide ligands (APLs) which seek to modulate the downstream signaling outcomes. However, dynamic links between modified peptide positions and distant residues have remained ill resolved until now. Using an integrative approach combining crystallographic ensemble and single models with atomistic molecular dynamics simulations and correlational analysis, we have established an approach that allows us to identify coupled dynamics between spatially distant residues at the pMHC interface. Furthermore, we constructed a network encoding the inter-residue couplings observed throughout the simulations. This computational workflow corroborates well with the functional and biophysical experimental data of our model system, and leads to novel insights regarding the differential immunogenicity of the closely related peptides analyzed in this study. Ultimately, we present an intuitive and comprehensive strategy for decoding the linked dynamics at the pMHC interface allowing for mechanistic insights into the biophysical bases governing immunogenicity.

To investigate how spatial constraints shape cancer metabolism, we devised the spatial Flux Balance Analysis (spFBA) framework for the enrichment of spatial transcriptomics data with relative estimates of metabolic fluxes. Applying spFBA to newly generated high-resolution datasets of paired primary colorectal tumors (CRC) and liver metastases revealed lactate consumption in both primary and metastatic regions. The presence of lactate-consuming niches was confirmed in an independent public dataset, suggesting this may be a recurrent metabolic feature of CRC. Importantly, application to public datasets of renal cancer showed widespread lactate production, consistent with a dominant but heterogeneous Warburg phenotype, ruling out general prediction biases or algorithmic artifacts. spFBA also consistently identified regions of increased proliferation across datasets, supporting the biological validity of its predictions. The framework is applicable to any sequencing-based spatial dataset to effectively uncover metabolic programs that remain invisible to gene expression analysis alone.

This perspective article discusses emerging advances at the interface of mechanistic modeling and data-driven machine learning, highlighting opportunities for AI to accelerate discovery, improve predictive modeling, and enhance clinical decision-making. We address critical limitations of current AI approaches and propose a perspective on a future where AI augments mechanistic rigor, clinical relevance, and human creativity under the umbrella of a redefined understanding of Mathematical Oncology.

Directed differentiation of human induced pluripotent stem cells (iPSCs) into anterior foregut endoderm (AFE) and lung progenitors (LPs) has wide-ranging implications for lung developmental biology, disease modeling, and regenerative medicine. We expand on a previously developed mathematical modeling framework and apply it to the directed differentiation of AFE into LPs. A model-based approach guides experimental design, followed by a multistage model inference process: maximum likelihood estimation based on in vitro data and identifiability analyses to eliminate unidentifiable candidates, thereby guiding model selection. To the authors' knowledge, this is the first mathematical model of the population dynamics of directed differentiation of AFE into LPs. The model suggests that the overall dynamics are primarily driven by AFE proliferation and differentiation into LPs. In silico experiments predict that daily media change nearly doubles LP yields compared to cultures without media replenishment. Moreover, the model suggests that higher split ratios on day 10 enhance yield per input cell, a measure of differentiation efficiency, by 26%. This work provides a blueprint for refining iPSC-based lung lineage differentiation protocols by combining empirical data and mathematical modeling.

Gene regulatory networks exhibit remarkable stability, maintaining functional phenotypes despite genetic and environmental perturbations. Discrete dynamical models, such as Boolean networks, provide systems biologists with a tractable framework to explore the mathematical underpinnings of this robustness. A key mechanism conferring stability is canalization. This perspective synthesizes historical insights, formal definitions of canalization in discrete dynamical models, quantitative measures of stability, and emerging challenges at the interface of theory and experiment.

Transcription factors play a central role in cancer growth, progression, and metastasis, and contribute to intratumor phenotypic plasticity that enable drug tolerance and cancer relapse. Changes in the regulatory activities of transcription factors in cancer may not always be detected from mutational signatures or differential expression of the transcription factors, as done in traditional analysis. In addition, past studies have focused on the activities of transcription factors in tumor as a whole and thus, have not fully captured the heterogeneity in gene regulation among different cell types within the tumor microenvironment. In this work, through an analysis of the transitions in regulatory network architecture and gene regulation dynamics, we identify the central transcription factors associated with lung adenocarcinoma progression. The gene NR2F1, associated with neurodevelopment and cancer dormancy, emerge as a key transcription factor in the progression of lung adenocarcinoma. We further identify transcription factors that are active in only cancer samples and uncover how changes in gene regulation dynamics influence intratumor heterogeneity. Taken together, our work elucidates the transitions in gene regulatory network during cancer progression, identifies central transcription factors in this process, and reveals the complex regulatory changes cooccurring in different cell types within the tumor microenvironment.