Biology subjects

Tsang, J. S.

Publications and source records attributed to Tsang, J. S..

11 recordsLinked to original sources

bioRxiv2026

MEA-LINK identifies the CCL4-CCR5 axis in neuronal hyperactivity control by human microglia

Microglia, the resident immune cells of the brain, act along a spectrum to maintain CNS homeostasis, respond to perturbations, and control neuronal activity. Disentangling the molecular mechanisms of human microglia-neuron crosstalk remains challenging due to the context-dependent, dynamic nature of their interaction. We introduce MEA-LINK, a systems-approach leveraging natural variation to screen for immune modulators of neuronal activity. This multi-modal platform integrates human induced pluripotent stem cell (hiPSC) technology with micro-electrode array (MEA) recordings and proteomic analyses of secreted immune factors, allowing for longitudinal samples and correlations across modalities. We applied MEA-LINK to explore microglia-neuron interactions during development and hyperactivity challenges. We show that human microglia accelerate neuronal network development and rescue hyperactive network phenotypes. Linking the secretome adaptations to neuronal network activity variations, we identified CCL4 as a top candidate in microglia-mediated hyperactivity control. Then, we functionally validated the context-dependent role of microglial CCL4 to neuronal CCR5 signaling in human neuronal networks. Our findings support a neuron-specific function of chemokines and their receptors in the brain and provide a new perspective for immune signaling in neuronal hyperactivity control. The MEA-LINK platform thus offers a foundation for comprehensive, systematic studies of human microglia-neuron interactions. HighlightsO_LIMEA-LINK integrates micro-electrode array recordings with proteomics of longitudinal samples to identify immune modulators of neuronal activity. C_LIO_LIHuman microglia rescue neuronal hyperactivity induced by pharmacological and genetic challenges. C_LIO_LIMicroglial CCL4 dampens neuronal activity via CCR5 signaling in a context-dependent manner. C_LI Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=147 SRC="FIGDIR/small/703799v1_ufig1.gif" ALT="Figure 1"> View larger version (39K): org.highwire.dtl.DTLVardef@18e44f7org.highwire.dtl.DTLVardef@151c450org.highwire.dtl.DTLVardef@12f7b89org.highwire.dtl.DTLVardef@57651b_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2025

Fetal and maternal immune cells exhibit tandem maturation across gestation and dysregulation in preterm birth

Appropriate regulation of maternal immune cells in the pregnant uterus is essential for successful pregnancy, yet the impact of immune cells in placental villi (PV) on placental health remain understudied. We enriched immune cells from term and preterm PV for transcriptomic and functional analyses. Maternal and fetal immune cells were abundant, with maternal chimerism increasing at term. Fetal lymphocytes displayed enhanced prostaglandin-D signatures, while maternal lymphocytes were enriched for antigen presentation pathways. Inflammatory cytokines and expanded maternal and fetal T cell clones peaked at term but did not induce cytotoxicity in placental organoids, whereas expanded T cells from idiopathic preterm cases impaired trophoblast growth and endocrine function ex vivo. These findings reveal significant immune chimerism within PV in both health and disease yet implicate PV T cells in idiopathic preterm birth. One Sentence SummaryFetal-maternal immune chimerism is characteristic of all placental villi, yet unchecked T cells and cytokines can hinder trophoblast growth and potentially precipitate preterm birth.

immunology↗

bioRxiv2025

Constraints and tunability of antigen-agnostic memory durability

Apart from antigen-specific immune memory, infection or vaccination can also induce antigen-agnostic memory including bystander T cell activation memory and trained innate immunity. Determinants of the durability of such memory remain unclear. We developed mathematical models to show that antigen-agnostic memory durability is constrained by immune cell turnover and cytokine dependence. Trained immunity durability is further constrained by the fidelity of epigenetic state transmission during self-renewal and differentiation. Using computer simulations and a machine learning-based parameter-phenotype mapping approach, we find that positive feedback mediated by immune cell-secreted cytokines, a motif seen across immune circuits, can lead to tunable memory involving excitable responses to acute challenges followed by slow relaxation to the pre-challenge baseline. We propose specific experiments to test predictions from our models. Our findings thus reveal a cell extrinsic mechanism that can overcome the constraints imposed by cell turnover and epigenetic state transmission to drive long-lasting, tunable antigen-agnostic memory.

immunology↗

bioRxiv2023

Non-coding sequence variation reveals fragility within interleukin 2 feedback circuitry and shapes autoimmune disease risk

Genetic variants associated with human autoimmune diseases commonly map to non-coding control regions, particularly enhancers that function selectively in immune cells and fine-tune gene expression within a relatively narrow range of values. How such modest, cell-type-selective changes can meaningfully shape organismal disease risk remains unclear. To explore this issue, we experimentally manipulated species-conserved enhancers within the disease-associated IL2RA locus and studied accompanying changes in the progression of autoimmunity. Perturbing distinct enhancers with restricted activity in conventional T cells (Tconvs) or regulatory T cells (Tregs)--two functionally antagonistic T cell subsets--caused only modest, cell-type-selective decreases in IL2ra expression parameters. However, these same perturbations had striking and opposing effects in vivo, completely preventing or severely accelerating disease in a murine model of type 1 diabetes. Quantitative tissue imaging and computational modelling revealed that each enhancer manipulation impinged on distinct IL-2-dependent feedback circuits. These imbalances altered the intracellular signaling and intercellular communication dynamics of activated Tregs and Tconvs, producing opposing spatial domains that amplified or constrained ongoing autoimmune responses. These findings demonstrate how subtle changes in gene regulation stemming from non-coding variation can propagate across biological scales due to non-linearities in intra- and intercellular feedback circuitry, dramatically shaping disease risk at the organismal level.

immunology↗

bioRxiv2022

Multiparameter stimulation mapping of signaling states in single pediatric immune cells reveals heightened tonic activation during puberty

Cellular stimulation via factors such as cytokines followed by multiparameter single-cell measurements is a powerful approach to interrogate cellular functions. However, transforming such high-dimensional data into biological insights presents unique challenges, particularly given the extensive response heterogeneity among single cells, such as the presence of bimodal responding versus non-responding subpopulations upon stimulation. Here we present an unsupervised high-dimensional approach for analyzing stimulation responses at the single cell level (HDStIM) and apply it to evaluate how pediatric development may shape peripheral immune cell signaling states and responsiveness to stimulations in 42 subjects (age: 2 - 16). We show that in comparison to the conventional approach of assessing one marker at a time by averaging across single cells, HDStIM can effectively learn, in an unsupervised fashion, the multi-parameter signature of responding versus non-responding cells to accurately quantify responses within cell populations. HDStIM reveals that the extent of pre-stimulation/baseline activation of interferon-related and TCR signaling molecules in myeloid and T cells, respectively, increases during puberty. This suggests that puberty is marked by a heightened "tonic" activation state in these cells, perhaps to strengthen defense against pathogens during this period of human development.

immunology↗

bioRxiv2022

Contrasting autoimmune and treatment effects reveals baseline set points of immune toxicity following checkpoint inhibitor treatment

Immune checkpoint inhibitors (ICIs) have changed the cancer treatment landscape, but severe immune-related adverse events (irAEs) can be life-threatening or fatal and may prohibit patients from receiving further ICI treatment. While the clinical features of irAEs are well documented, molecular signatures, predictive biomarkers, and mechanisms of impending irAEs are largely unknown. In addition, the markers and mechanisms of ICI-induced antitumor immunity often overlap with those for irAEs. It is thus critical to uncover signatures associated specifically with irAEs but not with antitumor immunity. To identify circulating immune cell states associated with irAEs, we applied multimodal single cell analysis (CITE-seq) to simultaneously measure the transcriptome and surface proteins from peripheral blood mononuclear cells (PBMCs) collected before and after treatment with an anti-PD-L1 antibody (avelumab) in patients with thymic cancers (thymic epithelial tumors). All patients had an antitumor response, yet a subset developed muscle autoimmunity (myositis), a potentially life-threatening irAE. Mixed-effect modeling disentangled cell type-specific transcriptional states associated with ICI treatment responses from those of irAEs to identify temporally stable pre-treatment immune set points associated with irAEs only. These pre-treatment baseline signatures of irAE developed post-avelumab irAEs reflect correlated transcriptional states of multiple innate and adaptive immune cell populations, including elevation of metabolic genes downstream of mTOR signaling in T-cell subsets. Together these findings suggest putative pre-treatment biomarkers for irAEs following ICI therapy in thymic cancer patients and raise the prospect of therapeutically dampening autoimmunity while sparing antitumor activity in cancer patients treated with ICIs. Together, pre-treatment biomarkers and interventional therapeutics could help mitigate treatment discontinuation and improve clinical outcomes.

immunology↗

bioRxiv2021

Pre-existing chromatin accessibility and gene expression differences among naïve CD4+ T cells influence effector potential

CD4+ T cells have a remarkable potential to differentiate into diverse effector lineages following activation. Here, we probed the heterogeneity present among naive CD4+ T cells before encountering their cognate antigen to ask whether their effector potential is modulated by pre-existing transcriptional and epigenetic differences. Using single-cell RNA sequencing, we showed that key drivers of variability are genes involved in T cell receptor (TCR) signaling. Using CD5 expression as a read-out of the strength of tonic TCR interactions with self-peptide MHC, and sorting on the ends of this self-reactivity spectrum, we find that pre-existing transcriptional differences among naive CD4+ T cells impact follicular helper cell (TFH) versus non-TFH effector lineage choice. Moreover, our data implicate TCR signal strength during thymic development in establishing differences in naive CD4 T cell chromatin landscapes that ultimately shape their effector potential.

immunology↗

bioRxiv2020

Blood gene expression-based prediction of lethality after respiratory infection by influenza A virus in mice

Lethality after respiratory infection with influenza A virus (IAV) is associated with potent immune activation and lung tissue damage. In a well-controlled animal model of infection, we sought to determine if one could predict lethality using transcriptional information obtained from whole blood early after influenza virus exposure. We started with publicly available transcriptomic data from the lung, which is the primary site of the infection and pathology, to derive a multigene transcriptional signature of death reflective of innate inflammation associated with tissue damage. We refined this affected tissue signature with data from infected mouse and human blood to develop and validate a machine learning model that can robustly predict survival in mice after IAV challenge using data obtained from as little as 10 l of blood from early time points post infection. Furthermore, in genetically identical, cohoused mice infected with the same viral bolus, the same model can predict the lethality of individual animals but, intriguingly, only within a specific time window that overlapped with the early effector phase of adaptive immunity. These findings raise the possibility of predicting disease outcome in respiratory virus infections with blood transcriptional data and pave the way for translating such approaches to humans.

immunology↗

bioRxiv2020

Normalizing and denoising protein expression data from droplet-based single cell profiling

Multimodal single-cell protein and transcriptomic profiling (e.g. CITE-seq) holds promise for comprehensive dissection of cellular heterogeneity, yet protein counts measured by oligo-conjugated-antibody can have substantial noise that masks biological variations. Here we integrated experiments and computational analysis to reveal two major noise sources: protein-specific noise from unbound antibodies and cell-specific noise captured by the shared variance of isotype controls and background protein counts. We provide an open source R package (dsb) to denoise and normalize CITE-seq data based on these findings. (https://cran.r-project.org/web/packages/dsb/index.html).

bioinformatics↗

bioRxiv2019

Machine learning of stochastic gene network phenotypes

A recurrent challenge in biology is the development of predictive quantitative models because most molecular and cellular parameters have unknown values and realistic models are analytically intractable. While the dynamics of the system can be analyzed via computer simulations, substantial computational resources are often required given uncertain parameter values resulting in large numbers of parameter combinations, especially when realistic biological features are included. Simulation alone also often does not yield the kinds of intuitive insights from analytical solutions. Here we introduce a general framework combining stochastic/mechanistic simulation of reaction systems and machine learning of the simulation data to generate computationally efficient predictive models and interpretable parameter-phenotype maps. We applied our approach to investigate stochastic gene expression propagation in biological networks, which is a contemporary challenge in the quantitative modeling of single-cell heterogeneity. We found that accurate, predictive machine-learning models of stochastic simulation results can be constructed. Even in the simplest networks existing analytical schemes generated significantly less accurate predictions than our approach, which revealed interesting insights when applied to more complex circuits, including the extensive tunability of information propagation enabled by feedforward circuits and how even single negative feedbacks can utilize stochastic fluctuations to generate robust oscillations. Our approach is applicable beyond biology and opens up a new avenue for exploring complex dynamical systems.

systems biology↗

bioRxiv2018

eNetXplorer: an R package for the quantitative exploration of elastic net families for generalized linear models

SummarySystems biology analysis often involves building predictive models by selecting informative features from a large number of measurements. The elastic net for generalized linear models is a popular regression and feature selection method, particularly useful when the number of features is greater than the sample size or when there exist many correlated predictor variables. This package provides a quantitative, cross-validation based toolkit to evaluate elastic net models and to uncover correlates contributing to prediction. Feature importance is evaluated by flexible criteria using out-of-bag prediction performance assessed via user-defined quality functions. Statistical significance is assigned to each model by comparison to null models generated by permutations of sample labels; analogous approaches are used to assess significance for the contribution of individual features to prediction. This package fits linear, binomial (logistic) and multinomial models, and provides a set of standard plots, summary statistics and output tables. eNetXplorer enables quantitative, exploratory analysis to generate hypotheses on which features may be associated with biological phenotypes of interest, such as in the identification of biomarkers for therapeutic responsiveness.\n\nAvailability and implementationThe eNetXplorer R package is available under GPLv3 license at https://CRAN.R-project.org/package=eNetXplorer

systems biology↗