Research Radar — 2026-05-24
Methods & AI
Computational
PRIME: Atlas-Level Single-Cell and Spatial Transcriptomics Data Integration via Projection-Based Robust Manifold Embedding
bioRxiv (Bioinformatics) Published 2026-05-23 preprint DOI: 10.64898/2026.05.20.726698
spatial transcriptomics single-cell data integration batch correction manifold learning atlas-level computational method
Summary: Introduces PRIME (Projection-based Robust Integration via Manifold Embedding), an ensemble integration framework for atlas-level single-cell RNA-seq and spatial transcriptomics data integration. PRIME uses random-projection-based consensus anchoring to identify robust cell pairs across datasets, then applies graph-Laplacian correction with optional spatial-neighborhood regularization for spatial transcriptomics. Unlike pairwise batch correction methods, PRIME is designed for the hierarchical, imbalanced batch effects characteristic of atlas-scale data encompassing millions of cells. Benchmarking demonstrates consistent outperformance over state-of-the-art methods across scRNA-seq and spatial transcriptomics integration, trajectory inference, spatial-domain preservation, and perturbation-response analysis. Notably, PRIME preserves developmental trajectories in human hematopoiesis (33,000 cells, 8 donors), maintains cortical laminar architecture in spatial brain data, and recovers drug-target relationships in a 1-million-cell perturbation atlas while suppressing batch confounders.
Why it matters: As consortium efforts assemble single-cell and spatial atlases across tissues, donors, and technologies, atlas-level integration has become a critical bottleneck. PRIME is the first framework specifically designed for this scale — combining expression-based anchoring with spatial-neighborhood regularization in a unified closed-form solution. It directly addresses the practical challenge of integrating data from multiple spatial transcriptomics platforms and scRNA-seq references, which is essential for building comprehensive tissue atlases for cancer and developmental biology.
Why for Yiru: Spatial transcriptomics integration is directly at the core of the user's research interests in spatial multi-omics. PRIME's ability to jointly integrate scRNA-seq and spatial transcriptomics across platforms provides a computational foundation for TME studies where multiple spatial technologies (Visium HD, MERFISH, Xenium) and scRNA-seq references must be harmonized to understand tumour-immune-stromal spatial organization.
SpatialCCCbench: Standardized Metrics for the Systematic Evaluation of Spatial Cell-Cell Communication Methods
bioRxiv (Bioinformatics) Published 2026-05-22 preprint DOI: 10.64898/2026.05.19.724475
spatial transcriptomics cell-cell communication benchmarking ligand-receptor computational method evaluation framework
Summary: Presents SpatialCCCbench, a comprehensive benchmarking framework with standardized metrics for systematically evaluating spatial cell-cell communication (CCC) inference methods. As the field of spatial transcriptomics matures, dozens of methods have emerged to infer ligand-receptor interactions between spatially proximal cell types, yet comparative evaluation has been inconsistent and metric-dependent. SpatialCCCbench establishes a unified evaluation protocol with ground-truth benchmarks, simulated spatial data with known communication patterns, and real spatial transcriptomics datasets, assessing methods across accuracy, robustness, scalability, and spatial specificity. The framework reveals substantial performance variability across methods and identifies key factors — spatial resolution, cell-type granularity, and communication distance thresholds — that critically influence inference quality.
Why it matters: Cell-cell communication inference is one of the most widely used downstream analyses in spatial transcriptomics, directly informing hypotheses about tissue coordination mechanisms. Without standardized evaluation, the field risks building biological conclusions on method-specific artifacts. SpatialCCCbench provides the community with the first dedicated benchmarking resource for spatial CCC, analogous to what CASP did for protein structure prediction.
Why for Yiru: Spatial cell-cell communication in the TME — between tumour cells, immune cells, and stromal fibroblasts — is central to understanding immunosuppression and therapy resistance. A rigorous benchmark for choosing and interpreting CCC methods directly impacts the quality of TME spatial analyses and the reliability of ligand-receptor hypotheses generated from spatial data.
StringTie3 Improves Total RNA-seq Assembly by Resolving Nascent and Mature Transcripts
Nature Methods Published 2026-05-19 research article DOI: 10.1038/s41592-026-03080-3
RNA-seq transcript assembly StringTie bioinformatics intron retention transcriptomics computational method
Summary: Introduces StringTie3, a major update to the widely used transcript assembly tool that, for the first time, explicitly resolves nascent (intron-containing) and mature (spliced) transcripts from total RNA-seq data. StringTie3 leverages the information contained in intronic reads — previously a source of assembly noise — to reconstruct both pre-mRNA and mRNA isoforms simultaneously, improving transcriptome completeness and accuracy. The method incorporates a new model for distinguishing transcriptional noise from genuine nascent transcription, and demonstrates improved performance on both short-read and long-read RNA-seq data. Benchmarking on human and mouse datasets shows StringTie3 recovers more full-length transcripts with fewer false chimeras compared to its predecessor and competing assemblers.
Why it matters: StringTie is one of the most cited bioinformatics tools for transcriptome assembly, with StringTie2 accumulating thousands of citations. The ability to simultaneously resolve nascent and mature transcripts from the same RNA-seq experiment doubles the information extracted — enabling co-analysis of transcriptional regulation (from intronic reads) and mature isoform expression without additional experiments. This is particularly valuable for cancer studies where both transcriptional dysregulation and alternative splicing are widespread.
Why for Yiru: Transcript assembly is fundamental to any RNA-seq analysis pipeline. StringTie3's ability to resolve nascent transcription directly from total RNA-seq enables deeper interrogation of gene regulation in TME cell types — distinguishing transcriptional activation from post-transcriptional isoform changes in tumour, immune, and stromal compartments.
A Community Machine Learning Challenge to Predict the Effects of Gene Perturbations on T Cell Differentiation for Cancer Immunotherapy
bioRxiv (Bioinformatics) Published 2026-05-22 preprint DOI: 10.64898/2026.05.21.726863
machine learning T cell immunotherapy gene perturbation computational immunology benchmark community challenge
Summary: Describes a community-organized machine learning challenge to predict how CRISPR-mediated gene perturbations affect T cell differentiation trajectories relevant to cancer immunotherapy. The challenge provides a curated multi-modal dataset combining single-cell CRISPR screens, transcriptomic profiles, and functional T cell assays, asking participants to predict which genetic perturbations enhance effector T cell differentiation and anti-tumour function. The paper reports the challenge design, baseline models, and key insights from top-performing solutions — revealing that models incorporating gene regulatory network priors and protein-protein interaction information outperform purely data-driven approaches. The challenge establishes a benchmark for computational prediction of T cell engineering targets.
Why it matters: Engineering T cells for cancer immunotherapy — through CAR-T, TCR-T, or gene-edited tumour-infiltrating lymphocytes — requires knowing which genetic perturbations enhance anti-tumour function. A community benchmark for predicting perturbation effects accelerates target discovery and establishes standardized evaluation for computational immunology methods, analogous to DREAM challenges in systems biology.
Why for Yiru: Computational prediction of T cell engineering targets directly bridges the user's interests in computational immunology and immunotherapy. Understanding which gene perturbations enhance anti-tumour T cell function through community-benchmarked ML methods provides a rational framework for designing next-generation T cell therapies.
HELIX: A Scalable Model for Predicting Context-Dependent Regulation of RNA Splicing and Isoform Usage
Nature Computational Science Published 2026-05-19 research article DOI: 10.1038/s43588-026-00988-w
RNA splicing deep learning isoform transcriptomics gene regulation computational method context-dependent
Summary: Presents HELIX, a scalable deep learning model for predicting context-dependent regulation of RNA splicing and isoform usage across diverse cellular contexts. Unlike existing splicing code models that predict splicing from local sequence features alone, HELIX integrates sequence, RNA-binding protein expression, chromatin state, and cell-type context to predict isoform usage in a context-dependent manner. The model captures how the same splicing regulatory sequence can produce different outcomes depending on the cellular environment — a phenomenon central to tissue-specific and disease-associated alternative splicing. HELIX demonstrates strong generalization to unseen cell types and accurately predicts splicing alterations in cancer.
Why it matters: Alternative splicing generates proteomic diversity and is frequently dysregulated in cancer, yet predicting how splicing responds to cellular context has been a fundamental challenge. HELIX's context-dependent modeling represents a conceptual advance over purely sequence-based splicing predictors, enabling in silico prediction of how splicing changes when regulatory factor expression is altered — relevant to understanding splicing alterations in the TME.
Why for Yiru: Alternative splicing is a layer of gene regulation relevant to immune cell differentiation, tumour neoantigen generation, and drug resistance. HELIX provides a tool for predicting how splicing programs differ between TME cell types and how they respond to perturbations, complementing transcript-level analyses in TME studies.
Immunotherapy Drug Target Identification Using Machine Learning and Patient-Derived Tumour Explant Validation
Nature Machine Intelligence Published 2026-05-18 research article DOI: 10.1038/s42256-026-01201-3
machine learning immunotherapy drug target patient-derived explant cancer target identification translational research
Summary: Reports a machine learning framework for identifying immunotherapy drug targets, validated using patient-derived tumour explant cultures — a physiologically relevant ex vivo system that preserves the native TME architecture and immune composition. The framework integrates multi-omic data (genomics, transcriptomics, and immunophenotyping) from patient tumours to prioritize targets predicted to enhance anti-tumour immunity when modulated. Top-ranked targets are functionally validated in patient-derived tumour fragments, demonstrating that ML-predicted targets, when inhibited or activated pharmacologically, enhance T cell-mediated tumour killing in a patient-specific manner. The study establishes an end-to-end pipeline from computational prediction to functional validation in a clinically relevant model system.
Why it matters: Immunotherapy target discovery typically follows either target-agnostic functional screens or hypothesis-driven single-target studies. This work bridges the gap between computational prediction and functional validation using patient-derived models that preserve TME complexity — a crucial advance for translating ML predictions into clinically relevant target nominations. The patient-derived explant validation provides a higher level of evidence than cell-line-based screening.
Why for Yiru: This paper exemplifies the translational potential of computational immunology — ML-driven target discovery validated in patient TME models. The end-to-end pipeline from multi-omic integration to functional validation is directly relevant to computational TME research aimed at identifying novel immunotherapy targets.
Biomedical discoveries
Biomedicine
ZNF274 Constrains Lineage Plasticity and Drives Intrinsic Resistance to CDK7 Inhibitors in Pancreatic Cancer
Nature Communications Published 2026-05-23 research article DOI: 10.1038/s41467-026-73380-x
pancreatic cancer ZNF274 lineage plasticity CDK7 inhibitor drug resistance epigenetics transcription factor
Summary: Identifies ZNF274, a KRAB zinc-finger transcription factor, as a critical constraint on lineage plasticity and a driver of intrinsic resistance to CDK7 inhibitors in pancreatic ductal adenocarcinoma (PDAC). Using genome-wide CRISPR screens in PDAC models treated with CDK7 inhibitors, the authors find that ZNF274 depletion sensitizes resistant cells by relieving epigenetic silencing of lineage-inappropriate genes, forcing cells into a more differentiated and drug-sensitive state. Mechanistically, ZNF274 recruits the HUSH complex and SETDB1 to deposit H3K9me3 at target loci, maintaining a repressive chromatin environment that locks PDAC cells in a progenitor-like, plastic state. ZNF274 loss derepresses gene programs that push cells toward acinar differentiation, simultaneously reducing the transcriptional addiction that underlies CDK7 inhibitor sensitivity in the progenitor state — paradoxically increasing drug sensitivity by reducing the pool of drug-tolerant persister cells.
Why it matters: Lineage plasticity is increasingly recognized as a mechanism of drug resistance across multiple cancer types, enabling tumour cells to escape targeted therapies by shifting their differentiation state. ZNF274 represents a novel druggable node controlling this plasticity in PDAC — a cancer with dismal prognosis and few effective targeted therapies. The finding that disrupting plasticity via ZNF274 inhibition can enhance CDK7 inhibitor efficacy provides a new therapeutic strategy for one of the deadliest cancers.
Why for Yiru: Lineage plasticity and epigenetically controlled drug resistance are frontier concepts in cancer biology with direct relevance to TME studies — plastic tumour cells may interact differently with immune and stromal components. Understanding the chromatin-level mechanisms that lock cells into drug-tolerant states is valuable for designing combination therapies that address both tumour cell-intrinsic and microenvironmental resistance.
Mirror-Image mRNA Display Uncovers Isoform-Selective D-Peptide Macrocycles Targeting a Cryptic KRAS Pocket
bioRxiv (Cancer Biology) Published 2026-05-22 preprint DOI: 10.64898/2026.05.20.726527
KRAS D-peptide mirror-image display macrocycle cancer undruggable chemical biology RAS
Summary: Reports the development of mirror-image mRNA display technology to discover D-peptide macrocycles that selectively target a cryptic pocket on KRAS — achieving a feat that has challenged conventional drug discovery for decades. Using chemically synthesized mirror-image mRNA display libraries, the authors screen for D-peptide binders against the D-enantiomer of KRAS protein, then synthesize the L-enantiomer peptide (which, by symmetry, binds native L-KRAS). This approach identifies isoform-selective macrocycles that bind a cryptic pocket distinct from the switch-II pocket targeted by approved KRAS(G12C) inhibitors, enabling targeting of KRAS mutants beyond G12C. The D-peptide macrocycles show cellular permeability, target engagement, and selective anti-proliferative activity in KRAS-mutant cancer cells, demonstrating a new chemical modality for RAS targeting.
Why it matters: KRAS mutations drive approximately 25% of all human cancers, yet approved covalent inhibitors only target KRAS(G12C), leaving the majority of KRAS-mutant patients without targeted therapy options. Mirror-image mRNA display represents a fundamentally new screening paradigm that enables discovery of proteolytically stable, cell-permeable D-peptide ligands against historically undruggable targets. Targeting a cryptic pocket distinct from switch-II opens the possibility of pan-KRAS or multi-mutant targeting with a single agent.
Why for Yiru: KRAS is one of the most important oncogenes in human cancer, and therapeutic targeting of RAS has been a holy grail for decades. New chemical modalities for targeting KRAS — and the cryptic pockets they reveal — are directly relevant to understanding how drugging this oncogene alters the TME, including effects on tumour immunogenicity, stromal signalling, and immune cell recruitment.
Fibroblast TGF-β3 Promotes Tissue-Residency and Survival of CD8 T Cells in Barrier Tissues and Tumours
bioRxiv (Immunology) Published 2026-05-22 preprint DOI: 10.64898/2026.05.20.726599
tissue-resident memory T cell fibroblast TGF-β3 CD8 T cell tumour microenvironment stromal niche immunosurveillance
Summary: Identifies fibroblast-derived TGF-β3 as a conserved stromal niche factor that promotes the tissue residency and survival of CD8+ tissue-resident memory T cells (TRM) in barrier tissues and tumours. While TGF-β is broadly implicated in TRM biology, the specific TGF-β isoform and cellular source responsible for TRM maintenance in non-lymphoid tissues have been unclear. Using fibroblast-specific conditional knockout models and spatial transcriptomics, the authors demonstrate that TGF-β3 — not the more widely studied TGF-β1 — is the critical fibroblast-derived factor sustaining TRM persistence. In tumours, fibroblast TGF-β3 maintains TRM numbers and effector function, and its loss accelerates TRM attrition and impairs tumour control. The study establishes a direct mechanistic link between the stromal fibroblast niche and CD8 T cell-mediated tumour immunosurveillance.
Why it matters: TRM cells are emerging as critical mediators of anti-tumour immunity, particularly for preventing local recurrence. Identifying fibroblast-derived TGF-β3 as the specific niche factor sustaining TRM in tumours provides a molecular target for enhancing TRM persistence therapeutically — a strategy orthogonal to checkpoint blockade that could improve durable tumour control. It also reframes cancer-associated fibroblasts not merely as immunosuppressive barriers but as potential supporters of protective T cell responses, depending on context.
Why for Yiru: CAF-T cell interactions in the TME are central to understanding why some tumours are immunologically 'hot' while others are 'cold.' The discovery that TGF-β3 from fibroblasts sustains anti-tumour TRM provides a mechanistic framework for studying how stromal composition shapes T cell-mediated tumour control — directly relevant to spatial TME studies combining fibroblast and T cell analysis.
PD-1 Remodels SHP2 Dynamics and Drives the Non-Catalytic Inhibition of T Cell Activation
bioRxiv (Immunology) Published 2026-05-22 preprint DOI: 10.64898/2026.05.20.725130
PD-1 SHP2 T cell immune checkpoint signalling phosphatase immunotherapy mechanism
Summary: Reveals a previously unappreciated mechanism by which PD-1 inhibits T cell activation: through remodelling of SHP2 dynamics rather than simply recruiting it to the immunological synapse. Using quantitative biochemical reconstitution, single-molecule imaging, and functional T cell assays, the authors show that PD-1 engagement reorganizes the spatial and temporal dynamics of the phosphatase SHP2, shifting it from a transient, scanning interaction mode to a stabilized, processive dephosphorylation mode at the T cell receptor signalling complex. This SHP2 remodelling does not require increased SHP2 recruitment — the total amount of SHP2 at the synapse remains similar — but rather changes how SHP2 engages its substrates, driving non-catalytic inhibition by physically occluding kinase docking sites in addition to catalytic dephosphorylation. The findings explain why SHP2 catalytic inhibitors alone incompletely reverse PD-1-mediated suppression.
Why it matters: PD-1 blockade has transformed cancer immunotherapy, yet the precise molecular mechanism by which PD-1 suppresses T cell activation has remained surprisingly controversial. This study provides a unifying model — PD-1 remodels SHP2 dynamics to achieve both catalytic and non-catalytic (steric) inhibition — that reconciles conflicting observations in the field. The finding that SHP2's scaffolding function is as important as its catalytic activity has direct implications for designing next-generation checkpoint inhibitors and SHP2-targeted therapies.
Why for Yiru: Understanding PD-1 signalling at the molecular level is essential for rational immunotherapy design. The concept that immune checkpoints operate through both catalytic and non-catalytic mechanisms expands the repertoire of potential intervention points — and may explain why certain combination immunotherapy strategies succeed or fail in the TME context.
Ferrous Iron Accumulation Is a Hallmark and Therapeutic Vulnerability of Therapy-Induced Senescence
bioRxiv (Cancer Biology) Published 2026-05-23 preprint DOI: 10.64898/2026.05.20.726695
therapy-induced senescence ferroptosis iron cancer chemotherapy tumour recurrence senescence vulnerability
Summary: Identifies lysosomal ferrous iron (Fe²⁺) accumulation as a conserved hallmark and actionable therapeutic vulnerability of therapy-induced senescence (TIS) in tumour cells. Chemotherapy and radiation eliminate most cancer cells but leave behind a reservoir of senescent cells that can drive tumour recurrence through the senescence-associated secretory phenotype (SASP). Using iron-specific fluorescent probes and quantitative metallomics, the authors discover that senescent tumour cells accumulate high levels of lysosomal ferrous iron across diverse cancer types and senescence-inducing treatments. This iron accumulation renders TIS cells exquisitely sensitive to ferroptosis induction — a form of regulated cell death driven by iron-dependent lipid peroxidation. Treatment with ferroptosis inducers selectively eliminates senescent tumour cells in vitro and in vivo, reducing tumour recurrence in preclinical models without affecting non-senescent proliferating cells.
Why it matters: Therapy-induced senescence is a double-edged sword: senescent cells are non-proliferative but secrete factors that promote tumour recurrence, metastasis, and therapy resistance. Selectively eliminating senescent tumour cells (a 'senolytic' approach) after chemotherapy could dramatically reduce recurrence rates. This study identifies ferroptosis induction as a senolytic strategy grounded in a fundamental metabolic feature of senescence (iron accumulation), providing both a mechanistic rationale and a clinically translatable therapeutic approach.
Why for Yiru: Tumour recurrence after therapy is a major clinical challenge, and senescent cells in the TME may contribute to an immunosuppressive milieu that facilitates regrowth. Understanding how senescence creates specific metabolic vulnerabilities — and how targeting these vulnerabilities reshapes the TME — is relevant to designing post-chemotherapy combination strategies that prevent immune escape and recurrence.
An Opioid/Cancer-Associated Fibroblast Axis Drives Extracellular Matrix Remodeling and Tumour Aggressiveness in Pancreatic Cancer
bioRxiv (Cancer Biology) Published 2026-05-21 preprint DOI: 10.64898/2026.05.20.726559
pancreatic cancer cancer-associated fibroblast opioid ECM remodeling tumour microenvironment neural signalling
Summary: Identifies a targetable opioid signalling axis through which cancer-associated fibroblasts (CAFs) drive extracellular matrix (ECM) remodelling and tumour aggressiveness in pancreatic ductal adenocarcinoma. The study reveals that PDAC tumours and their associated nerves produce endogenous opioids that act on opioid receptors expressed by CAFs. Opioid-stimulated CAFs increase ECM deposition, crosslinking, and tissue stiffness, which in turn promotes tumour cell invasion, metastasis, and therapy resistance. Pharmacological blockade of CAF opioid receptors with clinically available opioid antagonists (naltrexone) reduces ECM remodelling, decreases tumour stiffness, and impairs metastatic progression in preclinical PDAC models. The study uncovers a neuro-CAF-ECM signalling axis that connects neural inputs to the physical properties of the TME.
Why it matters: PDAC is characterized by extreme desmoplasia — an abundant, stiff ECM that compresses blood vessels, limits drug delivery, and promotes invasion. The finding that opioid signalling from nerves to CAFs drives this desmoplasia reveals a neuro-immune-stromal circuit that can be pharmacologically interrupted with existing drugs. Given that opioid antagonists are already FDA-approved, this discovery has a rapid translational path for improving PDAC therapy.
Why for Yiru: The neuro-CAF-ECM axis adds a new dimension to TME biology — neural regulation of stromal ECM remodelling. This connects to spatial TME studies where nerve fibres, CAFs, and ECM architecture can now be jointly analyzed to understand how neural inputs shape tumour aggressiveness through stromal intermediaries.
Cross-disciplinary watchlist
Other Fields
Cryo-EM Structure of the TRPC1/5 Heteromer Enables Design of Antidepressant and Anxiolytic Drug with Reduced Side Effects
Nature Communications Published 2026-05-23 research article DOI: 10.1038/s41467-026-73409-1
cryo-EM structural biology TRP channel drug design antidepressant anxiolytic ion channel structure-based drug discovery
Summary: Presents the high-resolution cryo-EM structure of the human TRPC1/5 heteromeric ion channel, a calcium-permeable channel implicated in anxiety, depression, and mood disorders. TRPC1/5 heteromers represent the native channel composition in brain regions associated with emotional regulation, yet previous drug development targeting TRPC channels has been hampered by lack of structural information and off-target effects on related TRP family members. The structure reveals the heteromer assembly mechanism, the ligand binding pocket architecture, and key structural determinants distinguishing TRPC1/5 from homomeric TRPC4 and TRPC5 channels. Guided by this structure, the authors design subtype-selective small molecules that preferentially target TRPC1/5 heteromers over related channels, demonstrating antidepressant and anxiolytic efficacy in mouse behavioural models with significantly reduced cardiovascular and motor side effects compared to earlier, less selective TRPC modulators.
Why it matters: TRPC channels are emerging drug targets for neuropsychiatric disorders, but achieving subtype selectivity has been a major obstacle. This study demonstrates the power of cryo-EM structural information for rational, structure-based design of ion channel modulators with improved therapeutic windows — a template for drug discovery against challenging membrane protein targets.
Why for Yiru: Structure-based drug design enabled by cryo-EM is a cross-disciplinary paradigm with applications well beyond neuroscience — including structure-guided design of immunomodulatory compounds targeting ion channels and receptors in immune cells within the TME. The principles of achieving subtype selectivity through structural understanding are broadly transferable.
Combined Biosynthesis and Site-Specific Incorporation of Phenylalanine Derivatives from Aryl Aldehydes or Carboxylic Acids
Nature Communications Published 2026-05-23 research article DOI: 10.1038/s41467-026-73618-8
synthetic biology genetic code expansion unnatural amino acid biosynthesis protein engineering biocatalysis chemical biology
Summary: Reports a combined biosynthesis and genetic code expansion system for the site-specific incorporation of diverse phenylalanine derivatives into proteins, starting from simple aryl aldehyde or carboxylic acid precursors. The system couples a biosynthetic pathway — converting inexpensive chemical feedstocks into non-canonical amino acids (ncAAs) in vivo — with an engineered orthogonal aminoacyl-tRNA synthetase/tRNA pair that specifically charges the biosynthesized ncAAs. This eliminates the need for chemical synthesis and exogenous addition of expensive ncAAs, making site-specific protein modification dramatically more accessible. The system is demonstrated for incorporating photocaged, fluorinated, and bioorthogonal-handle-containing phenylalanine derivatives at programmed sites in proteins expressed in E. coli and mammalian cells.
Why it matters: Genetic code expansion enables site-specific installation of chemical probes, post-translational modification mimics, and biophysical reporters into proteins — a technology with broad applications from basic biology to therapeutic protein engineering. The combined biosynthesis approach removes the cost and accessibility barriers that have limited adoption, potentially democratizing unnatural amino acid mutagenesis for the broader research community.
Why for Yiru: Site-specific protein modification is relevant to engineering therapeutic proteins, developing biosensors for TME metabolites, and studying protein function with chemical precision. The ability to incorporate probes into proteins expressed directly in mammalian cells — without expensive external ncAA addition — could enable new experiments probing protein dynamics in TME-relevant cell types.
Long-Term Comparative Analysis of AAV9-Mediated Gene Replacement Therapies for Spinal Muscular Atrophy in Mice
Nature Communications Published 2026-05-23 research article DOI: 10.1038/s41467-026-73545-8
gene therapy AAV9 spinal muscular atrophy SMA SMN comparative analysis rare disease long-term efficacy
Summary: Reports a comprehensive long-term comparative analysis of multiple AAV9-mediated gene replacement therapy strategies for spinal muscular atrophy (SMA) in a mouse model, tracking outcomes over extended periods beyond typical study durations. The study directly compares different SMN transgene designs — including codon-optimized, self-complementary, and regulatory-element variants — for their long-term efficacy, durability, and safety profiles. Key findings include evidence that certain transgene designs confer superior motor neuron protection and survival beyond two years post-treatment but also reveal late-emerging phenotypes (cardiac and hepatic) that were not apparent in shorter-term studies. The work provides critical data for optimizing the next generation of SMA gene therapies and highlights the importance of long-term follow-up in gene therapy preclinical development.
Why it matters: AAV9-mediated SMN gene replacement (Zolgensma) is one of the most expensive drugs in the world and a landmark for gene therapy. However, questions remain about the durability of effect, optimal transgene design, and long-term safety. This study provides head-to-head comparison data essential for designing improved gene therapies — not just for SMA but for any AAV-mediated neurological gene therapy.
Why for Yiru: Gene therapy and AAV vector engineering represent one of the most exciting translational frontiers in biomedicine. Understanding the principles of transgene design, durability, and tissue-specific expression has conceptual relevance to engineering gene therapies for cancer — including AAV-delivered immunomodulators for the TME or gene-editing vectors for tumour-specific targeting.
A Cross-Tissue POSTN+ Fibroblast Atlas Links Periodontal, Tumour, and Fibrotic Stromal Niches
bioRxiv (Bioinformatics) Published 2026-05-21 preprint DOI: 10.64898/2026.05.20.726414
fibroblast POSTN atlas cross-tissue tumour microenvironment fibrosis periodontal stromal biology
Summary: Constructs a cross-tissue single-cell atlas of periostin-expressing (POSTN+) fibroblasts, integrating data from periodontal tissue, solid tumours, and fibrotic organs to identify shared and tissue-specific features of this fibroblast subpopulation. POSTN+ fibroblasts have been independently implicated in tissue remodelling in each of these contexts, but their cross-tissue transcriptional conservation has not been systematically assessed. The atlas reveals a core POSTN+ fibroblast gene program — including ECM organization, TGF-β responsiveness, and immune-modulatory pathways — that is conserved across tissues, alongside tissue-specific modules driven by the local microenvironment. In tumours, POSTN+ fibroblasts express additional pro-invasive and immunosuppressive programs not seen in non-malignant tissues, suggesting that the tumour context co-opts and amplifies an evolutionarily conserved remodelling program.
Why it matters: Fibroblasts are increasingly recognized as functionally heterogeneous, yet most studies examine them within a single tissue context. A cross-tissue atlas reveals both conserved core programs and context-specific adaptations — essential for understanding which fibroblast functions are targetable across diseases and which require tissue-specific strategies. For cancer, the finding that tumours amplify a conserved remodelling program suggests that POSTN+ fibroblast biology is fundamentally linked to tissue repair mechanisms hijacked by malignancy.
Why for Yiru: Fibroblast heterogeneity in the TME is a frontier topic in spatial multi-omics. Understanding which fibroblast programs are conserved and which are tumour-specific directly informs how CAF subpopulations should be analyzed and targeted — and whether lessons from fibrosis biology can be applied to TME stromal targeting.
Immune Aging Captures Complementary Aging Biology Beyond Epigenetic Clocks
bioRxiv (Immunology) Published 2026-05-21 preprint DOI: 10.64898/2026.05.19.726183
immune aging epigenetic clock biological age immunosenescence aging biomarker multi-omic
Summary: Demonstrates that immune system aging captures complementary dimensions of biological aging beyond what is measured by conventional epigenetic clocks. Using multi-omic profiling (transcriptomics, proteomics, and immune repertoire sequencing) across a large human cohort spanning ages 20-90, the authors construct an immune aging clock that quantifies age-related changes in immune cell composition, function, and repertoire diversity. Critically, immune age and epigenetic age are only modestly correlated, and each independently predicts all-cause mortality and age-related disease risk — with immune age providing unique predictive power for infection susceptibility, vaccine response, and cancer incidence. The study establishes immune aging as a distinct dimension of the aging process with clinical relevance.
Why it matters: Epigenetic clocks have become the dominant framework for measuring biological aging, but they primarily capture mitotic history and chromatin state. This study provides strong evidence that immune system aging represents a largely independent aging axis — with direct implications for age-related cancer risk, immunotherapy response prediction, and vaccine strategies in elderly populations. It argues for incorporating immune metrics alongside epigenetic clocks in aging research and clinical geriatric assessment.
Why for Yiru: The intersection of aging, immunity, and cancer is directly relevant to understanding why cancer incidence increases with age and why immunotherapy responses differ between young and elderly patients. An immune-specific aging clock could help predict immunotherapy outcomes and inform age-adapted treatment strategies — relevant to TME immunology across the lifespan.
Somatic Mutations Reveal the Ontogeny of Human Microglia
bioRxiv (Immunology) Published 2026-05-20 preprint DOI: 10.64898/2026.05.19.726366
microglia somatic mutation ontogeny neuroimmunology lineage tracing brain single-cell
Summary: Uses naturally occurring somatic mutations as lineage barcodes to trace the developmental ontogeny of human microglia — the brain's resident immune cells. By performing ultra-deep whole-genome sequencing of microglia isolated from post-mortem human brain samples, the authors identify somatic mutations shared across microglial cells, which serve as indelible marks of shared ancestry. The mutational patterns reveal that human microglia arise from yolk-sac-derived progenitors that seed the brain during early embryogenesis and are maintained through local self-renewal throughout life, with minimal contribution from bone marrow-derived monocytes under non-pathological conditions. The study provides the first direct evidence from human tissue for the embryonic origin and lifelong persistence of microglia, resolving a long-standing debate about microglial turnover and replacement.
Why it matters: Microglia are key mediators of neuroinflammation, neurodegeneration, and brain tumour immunology (glioma-associated microglia/macrophages). Understanding their developmental origin and turnover dynamics is essential for interpreting microglial heterogeneity and for designing therapies that target microglia in disease — including whether peripheral monocytes can functionally replace microglia or whether embryonic microglia have unique properties that must be preserved.
Why for Yiru: Tissue-resident macrophages — including microglia in the brain and tumour-associated macrophages in the TME — share developmental origins and self-renewal properties. Understanding the ontogeny of tissue-resident immune cells is directly relevant to determining whether TAMs in tumours are locally maintained or continuously replenished from circulating monocytes, which has therapeutic implications for macrophage-targeted therapies.