Revealing cell-cell communication pathways with their spatially coupled gene programs.

Inference of cell-cell communication (CCC) provides valuable information in understanding the mechanisms of many important life processes. With the rise of spatial transcriptomics in recent years, many methods have emerged to predict CCCs using spatial information of cells. However, most existing methods only describe CCCs based on ligand-receptor interactions, but lack the exploration of their upstream/downstream pathways. In this paper, we proposed a new method to infer CCCs, called Intercellular Gene Association Network (IGAN). Specifically, it is for the first time that we can estimate the gene associations/network between two specific single spatially adjacent cells. By using the IGAN method, we can not only infer CCCs in an accurate manner, but also explore the upstream/downstream pathways of ligands/receptors from the network perspective, which are actually exhibited as a new panoramic cell-interaction-pathway graph, and thus provide extensive information for the regulatory mechanisms behind CCCs. In addition, IGAN can measure the CCC activity at single cell/spot resolution, and help to discover the CCC spatial heterogeneity. Interestingly, we found that CCC patterns from IGAN are highly consistent with the spatial microenvironment patterns for each cell type, which further indicated the accuracy of our method. Analyses on several public datasets validated the advantages of IGAN.

Brain-inspired chaotic spiking backpropagation.

Spiking neural networks (SNNs) have superior energy efficiency due to their spiking signal transmission, which mimics biological nervous systems, but they are difficult to train effectively. Although surrogate gradient-based methods offer a workable solution, trained SNNs frequently fall into local minima because they are still primarily based on gradient dynamics. Inspired by the chaotic dynamics in animal brain learning, we propose a chaotic spiking backpropagation (CSBP) method that introduces a loss function to generate brain-like chaotic dynamics and further takes advantage of the ergodic and pseudo-random nature to make SNN learning effective and robust. From a computational viewpoint, we found that CSBP significantly outperforms current state-of-the-art methods on both neuromorphic data sets (e.g. DVS-CIFAR10 and DVS-Gesture) and large-scale static data sets (e.g. CIFAR100 and ImageNet) in terms of accuracy and robustness. From a theoretical viewpoint, we show that the learning process of CSBP is initially chaotic, then subject to various bifurcations and eventually converges to gradient dynamics, consistently with the observation of animal brain activity. Our work provides a superior core tool for direct SNN training and offers new insights into understanding the learning process of a biological brain.

Sex differences orchestrated by androgens at single-cell resolution.

Sex differences in mammalian complex traits are prevalent and are intimately associated with androgens1-7. However, a molecular and cellular profile of sex differences and their modulation by androgens is still lacking. Here we constructed a high-dimensional single-cell transcriptomic atlas comprising over 2.3 million cells from 17 tissues in Mus musculus and explored the effects of sex and androgens on the molecular programs and cellular populations. In particular, we found that sex-biased immune gene expression and immune cell populations, such as group 2 innate lymphoid cells, were modulated by androgens. Integration with the UK Biobank dataset revealed potential cellular targets and risk gene enrichment in antigen presentation for sex-biased diseases. This study lays the groundwork for understanding the sex differences orchestrated by androgens and provides important evidence for targeting the androgen pathway as a broad therapeutic strategy for sex-biased diseases.

Multicellular ecotypes shape progression of lung adenocarcinoma from ground-glass opacity toward advanced stages.

Lung adenocarcinoma is a type of cancer that exhibits a wide range of clinical radiological manifestations, from ground-glass opacity (GGO) to pure solid nodules, which vary greatly in terms of their biological characteristics. Our current understanding of this heterogeneity is limited. To address this gap, we analyze 58 lung adenocarcinoma patients via machine learning, single-cell RNA sequencing (scRNA-seq), and whole-exome sequencing, and we identify six lung multicellular ecotypes (LMEs) correlating with distinct radiological patterns and cancer cell states. Notably, GGO-associated neoantigens in early-stage cancers are recognized by CD8+ T cells, indicating an immune-active environment, while solid nodules feature an immune-suppressive LME with exhausted CD8+ T cells, driven by specific stromal cells such as CTHCR1+ fibroblasts. This study also highlights EGFR(L858R) neoantigens in GGO samples, suggesting potential CD8+ T cell activation. Our findings offer valuable insights into lung adenocarcinoma heterogeneity, suggesting avenues for targeted therapies in early-stage disease.

Multi-modal molecular determinants of clinically relevant osteoporosis subtypes.

Due to a rapidly aging global population, osteoporosis and the associated risk of bone fractures have become a wide-spread public health problem. However, osteoporosis is very heterogeneous, and the existing standard diagnostic measure is not sufficient to accurately identify all patients at risk of osteoporotic fractures and to guide therapy. Here, we constructed the first prospective multi-omics atlas of the largest osteoporosis cohort to date (longitudinal data from 366 participants at three time points), and also implemented an explainable data-intensive analysis framework (DLSF: Deep Latent Space Fusion) for an omnigenic model based on a multi-modal approach that can capture the multi-modal molecular signatures (M3S) as explicit functional representations of hidden genotypes. Accordingly, through DLSF, we identified two subtypes of the osteoporosis population in Chinese individuals with corresponding molecular phenotypes, i.e., clinical intervention relevant subtypes (CISs), in which bone mineral density benefits response to calcium supplements in 2-year follow-up samples. Many snpGenes associated with these molecular phenotypes reveal diverse candidate biological mechanisms underlying osteoporosis, with xQTL preferences of osteoporosis and its subtypes indicating an omnigenic effect on different biological domains. Finally, these two subtypes were found to have different relevance to prior fracture and different fracture risk according to 4-year follow-up data. Thus, in clinical application, M3S could help us further develop improved diagnostic and treatment strategies for osteoporosis and identify a new composite index for fracture prediction, which were remarkably validated in an independent cohort (166 participants).

mNFE: microbiome network flow entropy for detecting pre-disease states of type 1 diabetes.

In the development of Type 1 diabetes (T1D), there are critical states just before drastic changes, and identifying these pre-disease states may predict T1D or provide crucial early-warning signals. Unlike gene expression data, gut microbiome data can be collected noninvasively from stool samples. Gut microbiome sequencing data contain different levels of phylogenetic information that can be utilized to detect the tipping point or critical state in a reliable manner, thereby providing accurate and effective early-warning signals. However, it is still difficult to detect the critical state of T1D based on gut microbiome data due to generally non-significant differences between healthy and critical states. To address this problem, we proposed a new method - microbiome network flow entropy (mNFE) based on a single sample from each individual - for detecting the critical state before seroconversion and abrupt transitions of T1D at various taxonomic levels. The numerical simulation validated the robustness of mNFE under different noise levels. Furthermore, based on real datasets, mNFE successfully identified the critical states and their dynamic network biomarkers (DNBs) at different taxonomic levels. In addition, we found some high-frequency species, which are closely related to the unique clinical characteristics of autoantibodies at the four levels, and identified some non-differential 'dark species' play important roles during the T1D progression. mNFE can robustly and effectively detect the pre-disease states at various taxonomic levels and identify the corresponding DNBs with only a single sample for each individual. Therefore, our mNFE method provides a new approach not only for T1D pre-disease diagnosis or preventative treatment but also for preventative medicine of other diseases by gut microbiome.

Mutant U2AF1-Induced Mis-Splicing of mRNA Translation Genes Confers Resistance to Chemotherapy in Acute Myeloid Leukemia.

Patients with primary refractory acute myeloid leukemia (AML) have a dismal long-term prognosis. Elucidating the resistance mechanisms to induction chemotherapy could help identify strategies to improve AML patient outcomes. Herein, we retrospectively analyzed the multiomics data of more than 1,500 AML cases and found that patients with spliceosome mutations had a higher risk of developing refractory disease. RNA splicing analysis revealed that the mis-spliced genes in refractory patients converged on translation-associated pathways, promoted mainly by U2AF1 mutations. Integrative analyses of binding and splicing in AML cell lines substantiated that the splicing perturbations of mRNA translation genes originated from both the loss and gain of mutant U2AF1 binding. In particular, the U2AF1S34F and U2AF1Q157R mutants orchestrated the inclusion of exon 11 (encoding a premature termination codon) in the eukaryotic translation initiation factor 4A2 (EIF4A2). This aberrant inclusion led to reduced eIF4A2 protein expression via nonsense-mediated mRNA decay. Consequently, U2AF1 mutations caused a net decrease in global mRNA translation that induced the integrated stress response (ISR) in AML cells, which was confirmed by single-cell RNA sequencing. The induction of ISR enhanced the ability of AML cells to respond and adapt to stress, contributing to chemoresistance. A pharmacologic inhibitor of ISR, ISRIB, sensitized U2AF1 mutant cells to chemotherapy. These findings highlight a resistance mechanism by which U2AF1 mutations drive chemoresistance and provide a therapeutic approach for AML through targeting the ISR pathway. SIGNIFICANCE: U2AF1 mutations induce the integrated stress response by disrupting splicing of mRNA translation genes that improves AML cell fitness to enable resistance to chemotherapy, which can be targeted to improve AML treatment.

Adeno-to-squamous transition drives resistance to KRAS inhibition in LKB1 mutant lung cancer.

KRASG12C inhibitors (adagrasib and sotorasib) have shown clinical promise in targeting KRASG12C-mutated lung cancers; however, most patients eventually develop resistance. In lung patients with adenocarcinoma with KRASG12C and STK11/LKB1 co-mutations, we find an enrichment of the squamous cell carcinoma gene signature in pre-treatment biopsies correlates with a poor response to adagrasib. Studies of Lkb1-deficient KRASG12C and KrasG12D lung cancer mouse models and organoids treated with KRAS inhibitors reveal tumors invoke a lineage plasticity program, adeno-to-squamous transition (AST), that enables resistance to KRAS inhibition. Transcriptomic and epigenomic analyses reveal ΔNp63 drives AST and modulates response to KRAS inhibition. We identify an intermediate high-plastic cell state marked by expression of an AST plasticity signature and Krt6a. Notably, expression of the AST plasticity signature and KRT6A at baseline correlates with poor adagrasib responses. These data indicate the role of AST in KRAS inhibitor resistance and provide predictive biomarkers for KRAS-targeted therapies in lung cancer.

EML4-ALK fusions drive lung adeno-to-squamous transition through JAK-STAT activation.

Human lung adenosquamous cell carcinoma (LUAS), containing both adenomatous and squamous pathologies, exhibits strong cancer plasticity. We find that ALK rearrangement is detectable in 5.1-7.5% of human LUAS, and transgenic expression of EML4-ALK drives lung adenocarcinoma (LUAD) formation initially and squamous transition at late stage. We identify club cells as the main cell-of-origin for squamous transition. Through recapitulating lineage transition in organoid system, we identify JAK-STAT signaling, activated by EML4-ALK phase separation, significantly promotes squamous transition. Integrative study with scRNA-seq and immunostaining identify a plastic cell subpopulation in ALK-rearranged human LUAD showing squamous biomarker expression. Moreover, those relapsed ALK-rearranged LUAD show notable upregulation of squamous biomarkers. Consistently, mouse squamous tumors or LUAD with squamous signature display certain resistance to ALK inhibitor, which can be overcome by combined JAK1/2 inhibitor treatment. This study uncovers strong plasticity of ALK-rearranged tumors in orchestrating phenotypic transition and drug resistance and proposes a potentially effective therapeutic strategy.

Cryoablation triggers type I interferon-dependent antitumor immunity and potentiates immunotherapy efficacy in lung cancer.

BACKGROUND: Cryoablation is a minimally invasive option for patients with medically inoperable non-small cell lung cancer (NSCLC) and can trigger abscopal immune-regulatory effects. However, it remains unclear how cryoablation affects the host-level immune response in NSCLC. In this study, we investigated the local and systemic immunological effects of cryoablation and the potential of combining cryoablation with programmed cell death protein 1 (PD-1) blockade to boost immunotherapy efficacy in NSCLC. METHODS: We first investigated systemic immunological effects induced by cryoablation in patients with early-stage NSCLC. Subsequently, we explored cryoablation-induced antitumor immunity and the underlying biological mechanisms using KP (Kras G12D/+, Tp53 -/-) mutant lung cancer cell allograft mouse models. Moreover, the synergistic efficacy of cryoablation and PD-1 blockade was explored in both mouse models and patients with unresectable NSCLC. RESULTS: We found that cryoablation significantly increased circulating CD8+ T cell subpopulations and proinflammatory cytokines in patients with early-stage NSCLC. In lung cancer cell allograft mouse models, we demonstrated that cryoablation resulted in abscopal growth inhibition of contralateral, non-ablated tumors. Integrated analysis of bulk, single-cell RNA and T cell receptor (TCR) sequencing data revealed that cryoablation reprogrammed the intratumoral immune microenvironment and increased CD8+ T cell infiltration with higher effector signature, interferon (IFN) response, and cytolytic activity. Mechanistically, cryoablation promoted antitumor effect through the STING-dependent type I IFN signaling pathway, and type I IFN signaling blockade attenuated this antitumor effect. We also found that the combination of PD-1 blockade with cryoablation further inhibited tumor growth compared with either treatment alone in an allograft mouse model. Moreover, the combination therapy induced notable tumor suppression and CD8+ T cell infiltration in patients with unresectable NSCLC. CONCLUSIONS: Our results provide mechanistic insights into how cryoablation triggers the antitumor immune effect in lung cancer, thereby potentiating programmed cell death ligand 1 (PD-L1)/PD-1 blockade efficacy in the clinical treatment of NSCLC.

Integrated characterization of hepatobiliary tumor organoids provides a potential landscape of pharmacogenomic interactions.

Despite considerable efforts to identify human liver cancer genomic alterations that might unveil druggable targets, the systematic translation of multiomics data remains challenging. Here, we report success in long-term culture of 64 patient-derived hepatobiliary tumor organoids (PDHOs) from a Chinese population. A divergent response to 265 metabolism- and epigenetics-related chemicals and 36 anti-cancer drugs is observed. Integration of the whole genome, transcriptome, chromatin accessibility profiles, and drug sensitivity results of 64 clinically relevant drugs defines over 32,000 genome-drug interactions. RUNX1 promoter mutation is associated with an increase in chromatin accessibility and a concomitant gene expression increase, promoting a cluster of drugs preferentially sensitive in hepatobiliary tumors. These results not only provide an annotated PDHO biobank of human liver cancer but also suggest a systematic approach for obtaining a comprehensive understanding of the gene-regulatory network of liver cancer, advancing the applications of potential personalized medicine.

Identification of eight genomic protective alleles for mitochondrial diabetes by Kinship-graph convolutional network.

AIMS: Nearly 85% of maternally inherited diabetes and deafness (MIDD) are caused by the m.3243A>G mutation in the mitochondrial DNA. However, the clinical phenotypes of MIDD may also be influenced by the nuclear genome, this study aimed to investigate nuclear genome variants that influence clinical phenotypes associated with m.3243A>G mutation in MIDD based on whole-genome sequencing of the patients belonging to pedigrees. MATERIALS AND METHODS: We analyzed a whole-genome sequencing (WGS) dataset from blood samples of 38 MIDD patients with the m.3243A > G mutation belonging to 10 pedigrees, by developing a Kinship-graph convolutional network approach, called Ki-GCN, integrated with the conventional genome-wide association study (GWAS) methods. RESULTS: We identified eight protective alleles in the nuclear genome that have protective effects against the onset of MIDD, related deafness, and also type 2 diabetes. Based on these eight protective alleles, we constructed an effective logistic regression model to predict the early or late onset of MIDD patients. CONCLUSIONS: There are protective alleles in the nuclear genome that are associated with the onset-age of MIDD patients and might also provide protective effects on the deafness derived from MIDD patients.

Correction to: Identification of TAZ as the essential molecular switch in orchestrating SCLC phenotypic transition and metastasis.

[This corrects the article DOI: 10.1093/nsr/nwab232.].

STAMarker: determining spatial domain-specific variable genes with saliency maps in deep learning.

Spatial transcriptomics characterizes gene expression profiles while retaining the information of the spatial context, providing an unprecedented opportunity to understand cellular systems. One of the essential tasks in such data analysis is to determine spatially variable genes (SVGs), which demonstrate spatial expression patterns. Existing methods only consider genes individually and fail to model the inter-dependence of genes. To this end, we present an analytic tool STAMarker for robustly determining spatial domain-specific SVGs with saliency maps in deep learning. STAMarker is a three-stage ensemble framework consisting of graph-attention autoencoders, multilayer perceptron (MLP) classifiers, and saliency map computation by the backpropagated gradient. We illustrate the effectiveness of STAMarker and compare it with serveral commonly used competing methods on various spatial transcriptomic data generated by different platforms. STAMarker considers all genes at once and is more robust when the dataset is very sparse. STAMarker could identify spatial domain-specific SVGs for characterizing spatial domains and enable in-depth analysis of the region of interest in the tissue section.

Potential network markers and signaling pathways for B cells of COVID-19 based on single-cell condition-specific networks.

To explore the potential network markers and related signaling pathways of human B cells infected by COVID-19, we performed standardized integration and analysis of single-cell sequencing data to construct conditional cell-specific networks (CCSN) for each cell. Then the peripheral blood cells were clustered and annotated based on the conditional network degree matrix (CNDM) and gene expression matrix (GEM), respectively, and B cells were selected for further analysis. Besides, based on the CNDM of B cells, the hub genes and 'dark' genes (a gene has a significant difference between case and control samples not in a gene expression level but in a conditional network degree level) closely related to COVID-19 were revealed. Interestingly, some of the 'dark' genes and differential degree genes (DDGs) encoded key proteins in the JAK-STAT pathway, which had antiviral effects. The protein p21 encoded by the 'dark' gene CDKN1A was a key regulator for the COVID-19 infection-related signaling pathway. Elevated levels of proteins encoded by some DDGs were directly related to disease severity of patients with COVID-19. In short, the proteins encoded by 'dark' genes complement some missing links in COVID-19 and these signaling pathways played an important role in the growth and activation of B cells.

Identification and Characterization of CD8+ CD27+ CXCR3- T Cell Dysregulation and Progression-Associated Biomarkers in Systemic Lupus Erythematosus.

Systemic Lupus Erythematosus (SLE) etiopathogenesis highlights the contributions of overproduction of CD4+ T cells and loss of immune tolerance. However, the involvement of CD8+ T cells in SLE pathology and disease progression remains unclear. Here, the comprehensive immune cell dysregulation in total 263 clinical peripheral blood samples composed of active SLE (aSLE), remission SLE (rSLE) and healthy controls (HCs) is investigated via mass cytometry, flow cytometry and single-cell RNA sequencing. This is observed that CD8+ CD27+ CXCR3- T cells are increased in rSLE compare to aSLE. Meanwhile, the effector function of CD8+ CD27+ CXCR3- T cells are overactive in aSLE compare to HCs and rSLE, and are positively associated with clinical SLE activity. In addition, the response of peripheral blood mononuclear cells (PBMCs) is monitored to interleukin-2 stimulation in aSLE and rSLE to construct dynamic network biomarker (DNB) model. It is demonstrated that DNB score-related parameters can faithfully predict the remission of aSLE and the flares of rSLE. The abundance and functional dysregulation of CD8+ CD27+ CXCR3- T cells can be potential biomarkers for SLE prognosis and concomitant diagnosis. The DNB score with accurate prediction to SLE disease progression can provide clinical treatment suggestions especially for drug dosage determination.

Revealing Tissue Heterogeneity and Spatial Dark Genes from Spatially Resolved Transcriptomics by Multiview Graph Networks.

Spatially resolved transcriptomics (SRT) is capable of comprehensively characterizing gene expression patterns and providing an unbiased image of spatial composition. To fully understand the organizational complexity and tumor immune escape mechanism, we propose stMGATF, a multiview graph attention fusion model that integrates gene expression, histological images, spatial location, and gene association. To better extract information, stMGATF exploits SimCLRv2 for visual feature exaction and employs edge feature enhanced graph attention networks for the learning potential embedding of each view. A global attention mechanism is used to adaptively integrate 3 views to obtain low-dimensional representation. Applied to diverse SRT datasets, stMGATF is robust and outperforms other methods in detecting spatial domains and denoising data even with different resolutions and platforms. In particular, stMGATF contributes to the elucidation of tissue heterogeneity and extraction of 3-dimensional expression domains. Importantly, considering the associations between genes in tumors, stMGATF can identify the spatial dark genes ignored by traditional methods, which can be used to predict tumor-driving transcription factors and reveal tumor immune escape mechanisms, providing theoretical evidence for the development of new immunotherapeutic strategies.

Earthquake alerting based on spatial geodetic data by spatiotemporal information transformation learning.

Alerting for imminent earthquakes is particularly challenging due to the high nonlinearity and nonstationarity of geodynamical phenomena. In this study, based on spatiotemporal information (STI) transformation for high-dimensional real-time data, we developed a model-free framework, i.e., real-time spatiotemporal information transformation learning (RSIT), for extending the nonlinear and nonstationary time series. Specifically, by transforming high-dimensional information of the global navigation satellite system into one-dimensional dynamics via the STI strategy, RSIT efficiently utilizes two criteria of the transformed one-dimensional dynamics, i.e., unpredictability and instability. Such two criteria contemporaneously signal a potential critical transition of the geodynamical system, thereby providing early-warning signals of possible upcoming earthquakes. RSIT explores both the spatial and temporal dynamics of real-world data on the basis of a solid theoretical background in nonlinear dynamics and delay-embedding theory. The effectiveness of RSIT was demonstrated on geodynamical data of recent earthquakes from a number of regions across at least 4 y and through further comparison with existing methods.

A single-cell transcriptional landscape of immune cells shows disease-specific changes of T cell and macrophage populations in human achalasia.

Achalasia is a rare motility disorder of the esophagus caused by the gradual degeneration of myenteric neurons. Immune-mediated ganglionitis has been proposed to underlie the loss of myenteric neurons. Here, we measure the immune cell transcriptional profile of paired lower esophageal sphincter (LES) tissue and blood samples in achalasia and controls using single-cell RNA sequencing (scRNA-seq). In achalasia, we identify a pattern of expanded immune cells and a specific transcriptional phenotype, especially in LES tissue. We show C1QC+ macrophages and tissue-resident memory T cells (TRM), especially ZNF683+ CD8+ TRM and XCL1+ CD4+ TRM, are significantly expanded and localized surrounding the myenteric plexus in the LES tissue of achalasia. C1QC+ macrophages are transcriptionally similar to microglia of the central nervous system and have a neurodegenerative dysfunctional phenotype in achalasia. TRM also expresses transcripts of dysregulated immune responses in achalasia. Moreover, inflammation increases with disease progression since immune cells are more activated in type I compared with type II achalasia. Thus, we profile the immune cell transcriptional landscape and identify C1QC+ macrophages and TRM as disease-associated immune cell subsets in achalasia.

Single-cell causal network inferred by cross-mapping entropy.

Gene regulatory networks (GRNs) reveal the complex molecular interactions that govern cell state. However, it is challenging for identifying causal relations among genes due to noisy data and molecular nonlinearity. Here, we propose a novel causal criterion, neighbor cross-mapping entropy (NME), for inferring GRNs from both steady data and time-series data. NME is designed to quantify 'continuous causality' or functional dependency from one variable to another based on their function continuity with varying neighbor sizes. NME shows superior performance on benchmark datasets, comparing with existing methods. By applying to scRNA-seq datasets, NME not only reliably inferred GRNs for cell types but also identified cell states. Based on the inferred GRNs and further their activity matrices, NME showed better performance in single-cell clustering and downstream analyses. In summary, based on continuous causality, NME provides a powerful tool in inferring causal regulations of GRNs between genes from scRNA-seq data, which is further exploited to identify novel cell types/states and predict cell type-specific network modules.