NPJ Systems Biology and Applications最新文献

Heterogeneity in breast cancer (BC) subtypes within a tumor contributes to therapy resistance and cancer recurrence. Subtype heterogeneity in tumors arises through a combination of stochastic genetic/epigenetic changes, phenotypic plasticity, and microenvironment-driven selection as the tumor evolves over time. Here, we sought to characterize how NF-κB epigenetic variability contributes to the progression of the HER2⁺ BC subtype. Initially, we used RNA to determine the expression levels of NF-κB, TWIST1, SIP1, and SLUG in two breast cancer (BC) cell lines, HCC-1954 and MDA-MB-231, classified as HER2⁺ and triple-negative breast cancer (TNBC) subtypes, respectively. Then, we built and calibrated a gene regulatory network (GRN) model that reproduces the transcriptional interactions between these genes. The model epigenetic landscape exhibits two attractor basins that reproduces the observed expression profiles of both HER2⁺ and TNBC subtypes, separated by an unstable stationary state. For validation, we used DHMEQ-treated cells, along with published patient data and in vitro results. Stochastic fluctuations in the NF-κB levels induce spontaneous irreversible transition from HER2⁺ to TNBC attractor basins at different times, contributing to subtype heterogeneity. The unstable state mediates this transition by providing a slow route between subtypes in the phase space that is susceptible to dynamic fluctuations. Mutations or drugs that change the availability of NF-κB alters the size of the subtype basins, changing the transition probabilities. Together, our findings enhance the established attractor landscape formulation and deepen understanding of BC heterogeneity, leading to more precise classification, prognosis, and targeted strategies for BC progression.

Glucagon-like peptide-1 (GLP-1) is a hormone known for its critical functions in managing blood sugar and offering cardiovascular benefits. Our study focuses on Glucagon Like Peptide 1 Receptor (GLP1R) agonists that act beyond glycemic control in cardiovascular and metabolic health. A comprehensive bioinformatic analysis was conducted, incorporating GLP1R, Gastric Inhibitory Polypeptide Receptor (GIPR), Gastric Inhibitory Polypeptide (GIP) and glucagon receptor (GCGR) to assess the effects of GLP1R agonists on gene and metabolite interactions. Interaction network analysis revealed 130 common genes among GLP1R, GLP1R/GIPR, GLP1R/GIP, and GLP1R/GIPR/GCGR associated with diabetes-related processes, including obesity and hyperglycemia. Enriched terms related to cardiovascular diseases, such as hypertension, calcium regulation in cardiac cells, and amino acid accumulation-induced mTOR activation. We also observed enrichment in gene sets linked to longevity and less recognized terms like fatty liver disease. In GLP1R/GIP, behavior-related terms and gastric acid secretion were identified; GLP1R/GIPR/GCGR linked to fibrosarcoma, thought/speech disturbances, and adipogenesis. The metabolite-gene layer revealed enrichment in galactose metabolism, platelet homeostasis, and nitric-oxide pathways. We found that GLP1R agonists network-level associations are stronger with heart diseases than sodium-glucose co-transporter 2 inhibitors, suggesting greater therapeutic benefits. Integrating networks with metabolites highlighted key interactors and clarified GLP1R agonists' mechanisms and therapeutic potential.

Complex biological systems often undergo a pre-transition phase prior to the onset of catastrophic event, during which a sharp and essential shift occurs. There is a pressing need to develop a swift and effective method for identifying such pre-transition phase or critical state, facilitating the timely implementation of tailored interventions to prevent irreversible and catastrophic transitions. Nonetheless, the identification of the pre-transition phase at the single-sample or single-cell level remains an exceedingly daunting task in modern clinical medicine, as reliance on small sample sizes often undermines the efficacy of traditional statistical methodologies. In this study, we propose a novel critical state algorithm based on individual sample data, termed single-sample network entropy (SNE), which effectively quantifies the disturbance caused by an individual sample relative to a set of reference samples, thereby revealing the pre-transition phases during biological development at the specific sample level. Our proposed method successfully identified pre-transition phases in both numerical simulations and eight real-world datasets, including an influenza infection dataset, three single-cell data (one associated with epithelial-mesenchymal transition (EMT) and two related to embryo development), and four tumor datasets: esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), and uterine corpus endometrial carcinoma (UCEC). In contrast to the existing single-sample approaches, our SNE method demonstrates enhanced effectiveness in detecting potential pre-transition phase. Moreover, it identifies two novel prognostic indicators: optimistic SNE (O-SNE) and pessimistic SNE (P-SNE) biomarkers for subsequent practical applications. Additionally, the reliability of computational findings is further strengthened by the functional roles of signaling biomarkers. Therefore, we present a novel computational approach that uncovers the pre-transition phases and signaling biomarkers of complex biological processes at the single sample or single-cell level, offering new insights and applications for early personalized biological analysis, including disease diagnosis and prognosis evaluation.

The BCL-2 protein family governs critical cell-fate decisions between survival, senescence, and apoptosis, yet the dynamical principles underlying these choices remain poorly understood. Here, we integrate mathematical modeling, bifurcation analysis, and stochastic simulations to dissect how BCL-2 network architecture encodes multistability and fate plasticity. Our coarse-grained model reveals tristable regimes requiring cooperative BH3-only and anti-apoptotic BCL-2 interactions, with stochastic fluctuations driving heterogeneous fate commitments in genetically identical cells. Comparative analysis of mechanistic models demonstrates that while bistability emerges from canonical BCL-2 interactions, robust tristability requires additional regulatory constraint, explaining the metastability of senescence in stress responses. Hybrid models further show that BH3-only binding cooperativity enables multistability, but physiological senescence likely depends on additional control mechanisms. These results establish a unified framework linking molecular interactions to cell-fate dynamics, with implications for targeting apoptosis resistance in disease.

Precision oncology relies on predictive biomarkers for selecting targeted cancer therapies. Network-based properties of proteins, together with structural features such as intrinsic disorder, are likely to shape their potential as biomarkers. We therefore designed a hypothesis-generating framework that integrates network motifs and protein disorder to explore their contribution to predictive biomarker discovery. This encouraged us to develop MarkerPredict by using literature evidence-based positive and negative training sets of 880 target-interacting protein pairs total with Random Forest and XGBoost machine learning models on three signalling networks. MarkerPredict classified 3670 target-neighbour pairs with 32 different models achieving a 0.7-0.96 LOOCV accuracy. We defined a Biomarker Probability Score (BPS) as a normalised summative rank of the models. The scores identified 2084 potential predictive biomarkers to targeted cancer therapeutics, 426 was classified as a biomarker by all 4 calculations. We detailed the biomarker potential of LCK and ERK1. This study encourages further validation of the high-ranked predictive biomarkers. The development of the MarkerPredict tool (which is available on GitHub) for predictive biomarker identification may have a significant impact on clinical decision-making in oncology.

Diabetic retinopathy (DR), a complex condition driven by inflammation, oxidative stress, and metabolic imbalances, calls for innovative treatment strategies. Engineered probiotics delivering angiotensin-converting enzyme 2 (ACE2) offer a promising strategy by leveraging gut microbiome-retina association. Advances in synthetic biology and computational techniques enable personalized, data-driven therapies. This review discusses computational approaches at multiple scales and presents an integrated framework for promoting personalized, systems-level DR management.

Alzheimer's disease (AD) is characterized by the accumulation of amyloid beta, which is strongly associated with disease progression and cognitive decline. Despite the approval of monoclonal antibodies targeting Aβ, optimizing treatment strategies while minimizing side effects remains a challenge. This study develops a mathematical framework to model Aβ aggregation dynamics, capturing the transition from monomers to higher-order aggregates, including protofibrils, toxic oligomers, and fibrils, using mass-action kinetics and coarse-grained modeling. Parameter estimation, sensitivity analysis, and data-driven calibration ensure model robustness. An optimal control framework is introduced to identify the optimal dose of the drug as a control function that reduces toxic oligomers and fibrils while minimizing adverse effects, such as amyloid-related imaging abnormalities (ARIA). The results indicate that Donanemab achieves the most significant reduction in fibrils. These findings provide a quantitative basis for optimizing AD treatments, providing valuable insight into the balance between therapeutic efficacy and safety.

Obesity impairs hepatic functions through abnormal functional protein expression, potentially through DNA methylation, which suppresses gene expression, and changes in transcription factors (TFs) expression. However, the specific protein expression changes associated with DNA methylation in the obese liver remain unclear. To dissect the relative association of DNA methylome and TF-binding with protein expression changes in the obese liver, we used a trans-omic integration approach combining DNA methylome, transcriptome, proteome, and TF-binding data for the livers of wild-type (WT) and obese (ob/ob) mice. We found that gene and protein expression changes were more strongly associated with TF expression changes than with changes in DNA methylation in promoter region. However, decreased protein expression of the complement and coagulation system in obesity was specifically associated with increased DNA methylation together with decreased expression of TF Hnf4a. Our study highlights abnormal protein expression specifically associated with DNA methylation and TF expression changes in obesity.

Boolean models are a powerful resource for studying dynamic processes of biological systems. However, their inherent discrete nature limits their ability to capture continuous aspects of signal transduction, such as signal strength or protein activation levels. Although existing tools provide some path exploration capabilities that can be used to explore signal transduction circuits, the computational workload often requires simplifying assumptions that compromise the accuracy of the analysis. Here, we introduce BooLEVARD, a Python package designed to efficiently quantify the number of paths leading either to node activation or repression in Boolean models, which offers a more detailed and quantitative perspective on how molecular signals propagate through signaling networks. By focusing on the collection of non-redundant paths directly influencing Boolean outcomes, BooLEVARD enhances the precision of signal strength representation. We showcase the application of BooLEVARD in studying cell-fate decisions using a Boolean model of cancer metastasis, demonstrating its ability to identify critical signaling events. In addition, through a second use case, we demonstrated BooLEVARD's capability to scale efficiently across increasingly large and connected Boolean models. Through these properties, BooLEVARD provides a distinctive tool for quantitative analysis of signaling dynamics within Boolean models, which can increase our understanding of disease development and drug responses. BooLEVARD is freely available at https://github.com/farinasm/boolevard .

Cancers are complex diseases characterized by dynamic perturbations of regulatory networks across multiple hierarchical levels, which cannot be fully captured by alterations in a small number of genes. To this end, based on the concept of Hallmarks of Cancer, a whole genomic data-driven approach is proposed to capture the dynamic variation from normal to cancerous cells. This framework focuses on the characteristic functional modules of cancer via hallmarks of cancer by constructing a coarse-grained gene regulatory network of hallmarks. Through this framework, with stochastic differential equations, macroscopic dynamic changes in tumorigenesis are simulated and further explored. The analysis results reveal that network topology undergoes significant reconfiguration before shifts in hallmark levels, serving as an early indicator of malignancy. A pan-cancer examination across 15 cancer types uncovers universal patterns, for example, the "Tissue Invasion and Metastasis" hallmark exhibits the most significant difference between normal and cancer states, while "Reprogramming Energy Metabolism" shows the least pronounced differences. These findings reinforce the systemic nature of cancer evolution, highlighting the potential of network-based systems biology methods for understanding critical transitions in tumorigenesis.