Biology subjects

Müller, S.

Publications and source records attributed to Müller, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Sustained antigenic engagement programs multidimensional CTL states with distinct cytotoxic capacities

Cytotoxic T lymphocytes (CTLs) eliminate infected or malignant cells through repeated target-cell encounters, yet how they sustain cytotoxic activity during extended antigenic stimulation remains poorly understood. In particular, it remains unclear whether preserved population-level killing reflects maintenance of a uniform effector phenotype or diversification into functionally distinct CTL populations. Here, we combined live-cell imaging, high-dimensional cytometry, single-cell RNA sequencing, phenotypic trajectory analysis, population sorting, and rechallenge killing assays to define how human clonal CTLs remodel during sustained antigenic engagement. Prior antigen encounter altered the kinetics and pattern of target-cell membrane permeabilization but did not compromise overall subsequent killing upon rechallenge. Single-cell transcriptomic profiling after prolonged engagement revealed broad remodeling of antigen-experienced CTLs and heterogeneous cytotoxic transcriptional programs. Time-resolved scRNA-seq further showed that these functional dimensions remodeled with distinct kinetics, with rapid induction of activation- and cytotoxicity-associated programs, later remodeling of vesicular-trafficking programs, and progressive reduction of adhesion/ contact interaction -associated programs. High-dimensional cytometry showed that this temporal remodeling was accompanied by progressive redistribution of CTLs across a multidimensional protein-expression landscape defined by distinct combinations of cytotoxic, vesicular/degranulation, activation, cytokine, adhesion, and regulatory features. Phenotypic trajectory analysis organized this protein landscape along a temporal remodeling axis, with early lytic-granule-rich profiles giving way to multiple late-engaged phenotypes with distinct combinations of functional protein features. Surface-marker-defined late CTL populations displayed distinct killing capacities upon rechallenge, demonstrating that preserved population-level cytotoxicity can coexist with substantial functional heterogeneity. Finally, CAR T cells exposed to CD19+ targets retained cytotoxic activity upon rechallenge and acquired multidimensional phenotypic features that partially overlapped with the remodeling framework identified in clonal CTLs. Reanalysis of longitudinal patient-derived CAR T-cell transcriptomes further showed that related functional programs were dynamically organized after infusion, with the features distinguishing responder and non-responder tisa-cel samples changing between the infusion product and day 7. Together, these findings indicate that sustained CTL killing is accompanied by progressive multidimensional remodeling and diversification rather than maintenance of a fixed effector phenotype. Our findings support a model in which population-level cytotoxic activity can be maintained while the CTLs composing that population progressively diversify into functionally distinct states during sustained antigenic engagement.

immunology↗

bioRxiv2026

Water Kefir as a Model for Aggregative Development of Microbial Multicellularity within a Multispecies Biofilm

Water kefir is a traditional fermented beverage and a non-dairy alternative to milk kefir. Water kefir grains are cauliflower-like aggregates, ranging in size from 1-20 mm. They are composed of an exopolysaccharide matrix derived from the fermentation substrate sucrose. The microbial community on and within the grains is a complex, highly stable multicellular aggregate typically consisting of 10-15 core species that primarily belong to three major groups: lactic acid bacteria, acetic acid bacteria, and yeasts. Remarkably, the structural organisation, stability, and growth of kefir grains cannot be readily explained by their constituent species alone, and attempts to reconstitute them from the isolated species have so far been unsuccessful. Here, we combined a comprehensive structural investigation of the macro- and micromorphology of water kefir grains with culture-dependent and -independent analyses of its microbial biodiversity. The dynamics and interactions within this complex microbial community were studied by combining flow cytometry, fluorescence-activated cell sorting, and amplicon sequencing of selected subpopulations. Furthermore, the growth of water kefir grains was spatiotemporally resolved and quantitatively characterised, thereby unravelling an emergent multicellular growth function, documented for the first time. Our combined evidence provides an unprecedented understanding of a highly structured and metabolically synergistic multicellular microbial aggregate and lets us propose water kefir as a model system for the aggregative development of microbial multicellularity from multiple species. It should lay the foundation for future mechanistic studies aimed at understanding how multispecies microbial communities self-organise into stable, spatially structured multicellular aggregates.

microbiology↗

bioRxiv2026

Deep learning design and in vivo validation of Müller glia-specific cis-regulatory elements

Effective recombinant adeno-associated virus gene therapies require promoters that are compact and cell-type-specific. Ideally, promoters should also exhibit functional conservation when tested in model organisms to ensure that preclinical findings translate reliably to human patients. Here, we introduce a deep learning framework for designing cis-regulatory elements (CREs) meeting these criteria, applied to retinal Muller glia (MG). Using single-cell chromatin accessibility data from human and mouse retinas, we trained species-specific models to predict cell-type accessibility, and designed compact CREs using two complementary strategies. In silico validation predicted that the designed CREs exhibit high MG-specific accessibility (on-target) in both species with minimal off-target accessibility across hundreds of human cell types and tissues. Mechanistic analysis revealed that the predicted MG accessibility is driven by the creation of LHX2 motifs. In vivo validation confirmed that the designed CREs successfully restrict reporter expression to MG in the murine retina. Our deep learning framework is highly generalizable and enables the rapid design of compact, on-target, and species-conserved CREs for precision gene therapy.

genomics↗

bioRxiv2025

The geometry of cooperation: decoding microbial interactions

Bridging the gap between mechanistic models of metabolism and ecological theory remains a key challenge in understanding interactions within microbial communities. We propose a geometric framework for analyzing community metabolism, based on constraint-based modeling. By extending pathway analysis methods from single organisms to multi-species systems, we define the community metabolic space as the set of all feasible fluxes between species and their environment, conditional on growth rate and medium composition. Embedded within a nonlinear geometry, this space forms a polytope whose vertices represent the minimal building blocks of community metabolism from which every feasible solution can be constructed. Strikingly, these elementary community flux modes invite direct ecological interpretation - as specialist, commensalist, or mutualist modes of cooperation. Furthermore, we find that mutualism is either isotypic (arising from a minimal mutualistic behavior) or anisotypic (emerging from a combination of reciprocal commensalists). This distinction demonstrates that bidirectional cross-feeding alone is insufficient to determine the ecological interaction type. Our framework also offers significant potential for applications. Because it does not rely on optimization, powerful unbiased tools from metabolic engineering, such as production envelopes and minimal cut sets, may be extended to microbial communities. Taken together, this perspective aims to unify ecological and metabolic viewpoints by linking interaction types to the geometric structure of the community metabolic space, thereby laying the foundation for a deeper understanding of community structure, function, and design.

systems biology↗

bioRxiv2025

Transcriptome-wide profiling of alternative splicing regulators with CRISPore-seq

Alternative splicing creates diverse RNA isoforms from individual genes, yet single-cell CRISPR screens are limited to gene-level quantification and cannot detect changes in alternative splicing and transcript isoforms. To overcome this limitation, we develop CRISPore-seq, which couples massively-parallel CRISPR perturbations with joint short- and long-read transcriptomics. CRISPore-seq simultaneously captures genetic perturbations and expression of genes, full-length transcripts and surface proteins in single cells. CRISPore-seq long reads identify 80% more transcript isoforms than short reads. Nearly all long reads map to unique transcript isoforms -- in contrast to existing single-cell perturbation methods, which rarely distinguish specific isoforms. Using CRISPore-seq, we knock-down 15 different RNA-binding proteins (RBPs) and identify thousands of perturbation-driven alternative splicing events (ASEs). We find that exon skipping is the most common ASE observed and that skipped exons are enriched for binding sites of perturbed RBPs. Loss of the Nager syndrome-associated spliceosomal factor SF3B4 triggers skipping of exon 2 in the cell-cycle regulator CCND1, preventing formation of a complex with CDK6 and blocking the G1-S transition. After rescue with a CCND1 isoform containing the skipped exon, both holoenzyme complex formation and cell proliferation are restored. By linking genes to transcriptional phenotypes with isoform-level resolution, CRISPore-seq is a highly scalable tool for understanding the impact of genetic perturbations on the human transcriptome.

genomics↗

bioRxiv2025

Sodium Aurothiomalate Induces Ferroptosis by Targeting GPX4 via Gold-Dependent Thiomalate Covalent Modification

Ferroptosis, an iron-dependent form of oxidative cell death, is predominantly regulated by glutathione peroxidase 4 (GPX4), making it a promising target for cancer therapy. However, the majority of GPX4 inhibitors, most of which contain a chloroacetamide moiety such as RSL3, are limited by poor pharmacokinetic properties and off-target effects, hindering their preclinical translation. Utilizing a range of interdisciplinary methodologies, we show that sodium aurothiomalate (ATM), a drug approved by many agencies, induces ferroptosis by covalently targeting GPX4 via formation of a selenenylsulfide bond. In preclinical models of neuroblastoma and acute myeloid leukemia (AML), ATM combined with ferric ammonium citrate (FAC) yields a synergistic effect, resulting in a significant reduction in tumor growth. Mechanistically, ATM disrupts GPX4 activity by covalently binding thiomalate to the active site selenocysteine, while modification of specific cysteine residues leads to destabilization of the protein and impaired binding to phospholipids. We propose that these covalent modifications are achieved through a unique reaction mechanism, in which the gold component of ATM acts as a thiol-masking carrier and is only transiently present, being subsequently displaced and allowing the reaction of the thiomalate moiety with the target selenocysteine or cysteine. Our data lay the foundation for development of novel, drug-like thiol-based GPX4 inhibitors.

biochemistry↗

bioRxiv2025

RNA-seq derived sequence variations are excellent features for cell line identification

1Cell lines are indispensable models for analyzing molecular mechanisms underlying human diseases. However, incorrect annotation and cross-contamination can introduce severe bias in respective studies. Accordingly, various publishers request authentication of cell lines before publication. Short tandem repeat profiling is commonly used to verify cell line identity and purity but does not guarantee that published results are based on the samples tested by this method. In this study, we demonstrate that RNA-seq-derived sequence variation information is eligible for unambiguous cell line-specific clustering. Based on this finding, we propose methods for reliable cell line identification from RNA-seq data using supervised machine learning methods. In addition, we demonstrate the ability to detect cross-contamination of human cell lines. The presented methods are insensitive to different data pre-processing steps and quality measures. The proposed TopFracCCLE algorithm for cell line identification and detection of cross-contamination is available as R-script at https://github.com/HuettelmaierLab/topFracCCLE.

genomics↗

bioRxiv2025

Elementary vectors reveal minimal interactions in microbial communities

Understanding microbial communities is essential for progress in ecology, biotechnology, and human health. In the last decade, constraint-based metabolic models of individual organisms have been combined to study microbial consortia. In this work, we present a geometric framework for characterizing all feasible microbial interactions. We project community models onto the key variables of interaction: exchange fluxes and community compositions. Based on this projection, we compute elementary composition/exchange fluxes (ECXs), extending the concept of minimal metabolic pathways from individual species to entire communities. Every feasible metabolic state of a community can be expressed as a combination of these elementary vectors. Notably, each ECX corresponds to a distinct ecological interaction type, such as specialization, commensalism, or mutualism. Finally, our geometric formulation enables the direct application of existing constraint-based methods, such as flux variability analysis and minimal cut sets, to microbial communities, providing a foundation for rational community design.

systems biology↗

bioRxiv2025

Polyamines buffer labile iron to suppress ferroptosis

Polyamines are essential and evolutionarily conserved metabolites present at millimolar concentrations in mammalian cells. Cells tightly regulate polyamine homeostasis through complex feedback mechanisms, yet the precise role necessitating this regulation remains unclear. Here, we show that polyamines contribute to endogenous buffering of redox-active iron, providing a molecular link between polyamine metabolism and ferroptosis. Using a genome-wide CRISPR screen, we identified a synthetic lethal dependency between polyamine depletion and the key ferroptosis suppressor, GPX4. Mechanistically, we show that polyamine deficiency triggers a redistribution of cellular iron, increasing the labile iron pool and upregulating ferritin. To directly visualize this iron buffering in living cells, we developed a genetically encoded fluorescent reporter for redox-active iron. Live-cell analysis revealed a striking inverse correlation between intracellular polyamine levels and redox-active iron at single-cell resolution. These findings reposition polyamines as key regulators of iron homeostasis, with implications for ferroptosis-linked disease states and cellular redox balance.

cell biology↗

bioRxiv2024

Bacterial colony biopsies: spatial discrimination of heterogeneous cell types by cytometric fingerprinting

Colonies of pure bacterial strains are highly dense cell structures that are organized in distinct and typical arrangements. The size, shape and variability of bacterial colonies are strongly dependent on the species and also influenced by environmental conditions. However, the spatial organization of individual cells is unknown for most strains. How specific heterogeneous cell types at different locations in a colony contribute to the overall structure and how they may influence the overall function of a colony is largely unknown. The study aims to investigate the local diversification of bacterial colony structures by introducing a local biopsy technique. The biopsied cells were analyzed by microbial flow cytometry and cytometric fingerprinting, which diversified the biopsied samples into many heterogeneous cell states. This two-stage resolution insight into colony structure was performed on five bacterial strains: Bacillus subtilis, Paenibacillus polymyxa, Kocuria rhizophila, Stenotrophomonas rhizophila, and Pseudomonas citronellolis. The effects of biopsy tool size (27G needle and 10 {micro}L, 200 {micro}L, 1000 {micro}L pipette tips) and sampling location on the precision of the technique were tested by using both gates setting along Gaussian distributions of subpopulations and a grid gating tool as well as the t-distributed Stochastic Neighbor Embedding (t-SNE) method. The biopsy technique uncovered significant heterogeneity among the cells within bacterial colonies, identifying differences in cell cycle stages, the proportion of living and dead cells, and the abundance of spore types. Cells from different biopsy sites displayed distinct physiological states, revealing that colony structure is far more complex than previously understood. The techniques precision depends on the biopsy tool size, dye equilibration, and cell handling, underscoring the importance of method calibration. The biopsy method, combined with cytometric fingerprinting, provided insights within only 15 to 45 minutes and is universally applicable. The study provides a high-resolution biopsy technique that explores the spatial distribution of cell types and their heterogeneous physiological cell states, allowing conclusions to be drawn from biopsy composition at different locations to overarching functions of the entire bacterial colony. This method also facilitates downstream analysis through further cell sorting, offering a powerful approach for future functional investigations.

microbiology↗

bioRxiv2024

Dermatopontin-expressing fibroblasts mediate an essential skin macrophage niche

Fibroblasts are present in all tissues and are crucial for maintaining tissue homeostasis. We previously identified fibroblasts marked by Dermatopontin (Dpt) but their role in supporting macrophage homeostasis remains unclear. Here, we generated novel mesenchymal lineage-restricted genetic tools to target Dpt expressing fibroblasts and elucidate their role in supporting skin macrophages. Transcriptional profiling, flow cytometry, and in situ hybridization uncovered two broad populations of F4/80-expressing skin macrophages, denoted by high expression of CD206 and CD64 (CD206hiCD64+), or CD11c. Targeted depletion of Dpt+ fibroblasts resulted in a profound loss of both macrophage populations. Conditional deletion of colony-stimulating factor-1 (Csf1) in Dpt+ fibroblasts revealed that CD206hiCD64+, and not CD11c+, macrophages are acutely dependent on fibroblast-derived Csf1, consistent with their higher expression of the Csf1 receptor. Following Csf1 deletion in Dpt+ fibroblasts, loss of CD206hiCD64+ macrophages were observed across the dermis, dermal white adipose tissue (dWAT), and adventitia, accompanied by a modest upregulation of fibroblast-related and extracellular matrix (ECM) genes and structural changes to the skin. Alterations to the skin network upon loss of fibroblast-derived Csf1 and CD206hiCD64+ macrophages led to a significant delay in wound healing. We also demonstrate the CSF1-CSF1R signaling pathway is functionally relevant in human systemic sclerosis, or scleroderma, as elevated levels of CSF1 produced by fibroblasts and an increased abundance of macrophages both correlate with disease severity. Our findings demonstrate the role of Dpt+ fibroblasts in regulating a Csf1-dependent macrophage niche in skin and orchestrating responses in injury and disease.

immunology↗

bioRxiv2024

Automated and closed clinical-grade manufacturing protocol produces potent NK cells against neuroblastoma cells and AML blasts

Natural killer (NK) cells have great potential as allogeneic immune cell therapy due to their natural ability to recognize and kill tumor cells, and due to their apparent safety. This study describes the development of an immunotherapy option tailored for high-risk acute myeloid leukemia (AML) in adults and neuroblastoma in children. A GMP-compliant manufacturing protocol for the local production of functionally potent NK cells is detailed in the study, including a comprehensive description of the quality control strategy and considerations for product batch specifications in early clinical development. The protocol is based on the closed, automated CliniMACS Prodigy(R) platform (Miltenyi Biotec) and a modified Natural Killer Cell Transduction (NKCT) process without transduction and expansion. NK cells are isolated from leukapheresis through CD3 depletion and CD56 enrichment, followed by a 12-hour activation with cytokines (500 IU/ml IL-2, 140 IU/ml IL-15). Three CliniMACS Prodigy(R) NKCT processes were executed, demonstrating the feasibility and consistency of the modified NKCT process. A three-step process without expansion, however, compromised the NK cell yield. T cells were depleted effectively, indicating excellent safety of the product for allogeneic use. Phenotypic and functional characterization of the NK cells before and after cytokine activation revealed a notable increase in the expression of activation markers, particularly CD69, consistent with enhanced functionality. Intriguingly, even following a brief 12-hour activation period, the NK cells exhibited increased killing efficacy against CD33+ AML blasts isolated from patients and against SH-SY5Y neuroblastoma (NBL) target cells in vitro, suggesting a potential therapeutic benefit for AML and NBL patients.

immunology↗

bioRxiv2024

Ether lipids influence cancer cell fate by modulating iron uptake

Cancer cell fate has been widely ascribed to mutational changes within proteincoding genes associated with tumor suppressors and oncogenes. In contrast, the mechanisms through which the biophysical properties of membrane lipids influence cancer cell survival, dedifferentiation and metastasis have received little scrutiny. Here, we report that cancer cells endowed with high metastatic ability and cancer stem celllike traits employ ether lipids to maintain low membrane tension and high membrane fluidity. Using genetic approaches and lipid reconstitution assays, we show that these ether lipid-regulated biophysical properties permit non-clathrin-mediated iron endocytosis via CD44, resulting in significant increases in intracellular redox-active iron and enhanced ferroptosis susceptibility. Using a combination of in vitro threedimensional microvascular network systems and in vivo animal models, we show that loss of ether lipids from plasma membranes also strongly attenuates extravasation, metastatic burden and cancer stemness. These findings illuminate a mechanism whereby ether lipids in carcinoma cells serve as key regulators of malignant progression while conferring a unique vulnerability that can be exploited for therapeutic intervention.

cancer biology↗

bioRxiv2024

ERLIN1/2 scaffolds bridge TMUB1 and RNF170 and restrict cholesterol esterification to regulate the secretory pathway

Complexes of ERLIN1 and ERLIN2 form large ring-like cup-shaped structures on the endoplasmic reticulum (ER) membrane and serve as platforms to bind cholesterol and E3-ubiquitin ligases, potentially defining functional nanodomains. Here, we show that ERLIN scaffolds mediate the interaction between the full-length isoform of TMUB1 and RNF170. We identify a luminal N-terminal conserved region in TMUB1 and RNF170 required for this interaction. Three-dimensional modelling shows that this conserved motif binds the SPFH domain of two adjacent ERLIN subunits at different interfaces. Protein variants that preclude these interactions have been previously linked to hereditary spastic paraplegia (HSP). By using omics approaches in combination with phenotypic characterisation of cells lacking both ERLINs, we demonstrate a role for ERLIN scaffolds in maintaining cholesterol levels in the ER by favouring transport to the Golgi over esterification, thereby regulating Golgi morphology and the secretory pathway.

cell biology↗

bioRxiv2023

Does ribosome composition maximize growth rate? The role of RNA instability

Ribosomes are protein synthesis machines that are central to cellular self-fabrication, and the synthesis time of a ribosome places an upper bound on growth rate. While most cellular enzymes are proteins, ribosomes consist of 1/3 protein and 2/3 RNA (in E. coli). Recent research suggests that ribosome composition arises from a trade-off between two "autocatalytic loops", ribosomal protein and RNA polymerase synthesis, respectively. In this study, we develop a (coarse-grained) mechanistic model of a self-fabricating cell and validate it under various growth conditions. Using resource balance analysis (RBA), we examine how the maximum growth rate varies with ribosome composition, assuming that all kinetic parameters remain independent of ribosome composition. Our model highlights the importance of RNA instability. If we neglect it, RNA synthesis is always "cheaper" than protein synthesis, leading to an RNA-only ribosome at maximum growth rate. However, when we account for RNA turnover, we find that a mixed ribosome composed of RNA and proteins maximizes growth rate. To account for RNA turnover, we explore two scenarios regarding the activity of RNases. In (a) degradation is proportional to RNA content, whereas in (b) ribosomal proteins cooperatively mitigate RNA instability by protecting it from misfolding and subsequent degradation. In both cases, an increase in ribosomal protein content raises protein synthesis costs while reducing RNA turnover costs. This leads to a mixed ribosome composed of RNA and proteins. However, only in scenario (b), where we consider the cooperative protection of ribosomal RNA by proteins, our model predictions are in qualitative agreement with experimental data under different growth conditions. Our research offers new mechanistic insights into ribosome biogenesis and evolution. Furthermore, it paves the way for understanding the protein-rich ribosome composition found in archaea and mitochondria.

systems biology↗

bioRxiv2022

Lymph node expansion predicts magnitude of vaccine immune response

Lymph nodes (LNs) dynamically expand in response to immunization, but the relationship between LN expansion and the accompanying adaptive immune response is unclear. Here, we first characterized the LN response across time and length scales to vaccines of distinct strengths. High-frequency ultrasound revealed that a bolus weak vaccine induced a short-lived, 2-fold volume expansion, while a biomaterial-based strong vaccine elicited an [~]7-fold LN expansion, which was maintained several weeks after vaccination. This latter expansion was associated with altered matrix and mechanical properties of the LN microarchitecture. Strong vaccination resulted in massive immune and stromal cell engagement, dependent on antigen presence in the vaccine, and conventional dendritic cells and inflammatory monocytes upregulated genes involved in antigen presentation and LN enlargement. The degree of LN expansion following therapeutic cancer vaccination strongly correlated with vaccine efficacy, even 100 days post-vaccination, and direct manipulation of LN expansion demonstrated a causative role in immunization outcomes.

immunology↗

bioRxiv2022

Individualized Gaussian Process-based Prediction of Memory Performance and Biomarker Status in Ageing and Alzheimer's disease

Neuroimaging markers based on Magnetic Resonance Imaging (MRI) combined with various other measures (such as informative covariates, vascular risks, brain activity, neuropsychological test etc.,) might provide useful predictions of clinical outcomes during progression towards Alzheimers disease (AD). The Bayesian approach aims to provide a trade-off by employing relevant features combinations to build decision support systems in clinical settings where uncertainties are relevant. We tested the approach in the MRI data across 959 subjects, aged 59-89 years and 453 subjects with available neuropsychological test scores and CSF biomarker status (amyloid-beta (A{beta})42/40 & and phosphorylated tau (pTau)) from a large sample multi-centric observational cohort (DELCODE). In order to explore the beneficial combinations of information from different sources, we presented a MRI-based predictive modelling of memory performance and CSF biomarker status (positive or negative) in the healthy ageing group as well as subjects at risk of Alzheimers disease using a Gaussian process multikernel framework. Furthermore, we systematically evaluated predictive combinations of input feature sets and their model variations, i.e. (A) combinations of brain tissue classes and feature type (modulated vs. unmodulated), choices of filter size of smoothing (ranging from 0 to 15 mm full width at half maximum), and image resolution (1mm, 2mm, 4mm and 8mm); (B) incorporating demography and covariates (C) the impact of the size of the training data set (i.e., number of subjects); (D) the influence of reducing the dimensions of data and (E) choice of kernel types. Finally, the approach was tested to reveal individual cognitive scores at follow-up (up to 4 years) using the baseline features. The highest accuracy for memory performance prediction was obtained for a combination of neuroimaging markers, demographics, genetic information (ApoE4) and CSF-biomarkers explaining 57% of outcome variance in out of sample predictions. The best accuracy for A{beta}42/40 status classification was achieved for combination demographics, ApoE4 and memory score while usage of structural MRI improved the classification of individual patients pTau status.

neuroscience↗

bioRxiv2022

Spatial positioning and matrix programs of cancer-associated fibroblasts promote T cell exclusion in human lung tumors

It is currently accepted that activated cancer-associated fibroblasts (CAF) participate in T cell exclusion from tumor nests, but it remains unclear how they promote barrier phenotypes, and whether specific subsets are involved. Here, using single-cell RNA sequencing coupled with multiplex imaging on a large cohort of lung tumors, we identify four main CAF populations, of which only two are associated with T cell exclusion: (i) MYH11+SMA+ CAF, which are present in early-stage tumors and form a single-cell layer lining cancer aggregates, and (ii) FAP+SMA+ CAF, which appear in more advanced tumors and organize in patches within the stroma or in multiple layers around tumor nests. Both CAF populations show a contractility phenotype together with dense and aligned matrix fiber deposition compared to the T cell-permissive CAF. Yet they express distinct matrix genes, including COL4A1/COL9A1 (MYH11+SMA+ CAF) and COL11A1/COL12A1 (FAP+SMA+ CAF). Hereby, we uncovered unique molecular programs of CAF driving T cell marginalization, whose targeting should increase immunotherapy efficacy in patients bearing T cell-excluded tumors. SIGNIFICANCEThe cellular and molecular programs driving T cell marginalization in solid tumors remain unclear. Here, we describe two CAF populations associated with T cell exclusion in human lung tumors. We demonstrate the importance of pairing molecular and spatial analysis of the tumor microenvironment, a prerequisite to develop new strategies targeting T cell-excluding CAF.

cancer biology↗