Biology subjects

Zou, A.

Publications and source records attributed to Zou, A..

13 recordsLinked to original sources

bioRxiv2026

RNA-guided contrastive learning enhances patient-level prediction from histology

Predicting molecular receptor status, including estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), directly from routine hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) could reduce reliance on costly immunohistochemistry, but image-only models have no access to the transcriptomic programs that define these subtypes. Here we show that coupling a whole-slide vision transformer (GigaPath) with bulk RNA-sequencing profiles through a training-time-only contrastive objective improves receptor-status classification from WSIs alone, without requiring any RNA-seq data at inference. Rather than training a dedicated bulk-transcriptomics encoder, we purpose the gene-sentence encoding and pretrained text encoder from BioBERT1 and apply them to bulk RNA-seq profiles from The Cancer Genome Atlas breast cancer cohort (TCGA-BRCA). Under five-fold cross-validation, this RNA-guided pre-training consistently improved AUROC for ER, PR, and HER2 relative to a contrastive-pre-training-free baseline and reproducibly introduced receptor-relevant structure into the learned slide representation. However, evaluation on independent external cohorts revealed a striking, receptor-dependent divergence: gains for ER and PR, both governed by broad, multi-gene luminal transcriptional programs, generalized robustly, whereas the apparent AUROC improvement for HER2, driven by focal ERBB2 amplification rather than a coordinated transcriptional signature, masked a severe loss of classification sensitivity at the operating threshold. These findings establish training-time transcriptomic supervision as a practical route to enhancing histology-only molecular subtyping, while showing that gene-selection strategy, and not model architecture alone, determines which molecular alterations such supervision can teach a vision model to recognize.

pathology↗

bioRxiv2026

TipQUANT: A robust algorithm for quantitative analysis of spatiotemporally dynamic activities in tip-growing cells

Cell polarity, essential for cell development and function, relies on dynamic subcellular distribution of structural and signaling molecules. Tip growth, an extreme form of polar growth, involves unidirectional expansion at the apical region of cells and requires precise spatiotemporal coordination to achieve periodic and directional growth. Understanding their spatiotemporal dynamics is critical for elucidating mechanisms and functions of cell polarity. However, manual quantification of such dynamics is extremely time-consuming, hindering advancements in the field. Current algorithms have limited power and flexibility in analyzing the distribution and dynamics of molecules and structures, particularly for tip-growing cells with oscillatory and dynamic behavior. To address this challenge, we present TipQuant, an automated analysis tool that robustly detects tips and analyzes spatiotemporal dynamics of fluorescently labeled molecules/structures on plasma membranes and in cytoplasm at apices of tip-growing cells, enabling quantitative understanding of signaling and structural components in these systems.

cell biology↗

bioRxiv2025

Deep Learning Enables Automated Segmentation and Quantification of Ultrastructure from Transmission Electron Microscopy Images

Transmission electron microscopy (TEM) has become an essential technique for observing subcellular ultrastructure, and is widely used in both clinical diagnosis and biomedical research. However, analysis of TEM data remains extremely labor-intensive and often inconsistent across operators due to the lack of dedicated computational methods. Here, we present TEAMKidney, a deep learning framework for accurate and scalable measurement of ultrastructures in TEM images across species, magnifications, and instrument platforms. We collected 12,991 TEM images from patients with multiple kidney diseases and from different animal models. By combining a self-training-based semantic segmentation stage with a TEM-tailored panoptic segmentation model, we address two major challenges in TEM data analysis: the lack of accurately labeled training data and the difficulty of achieving high segmentation accuracy for complex ultrastructure. Application of TEAMKidney to both human and animal images successfully reveals disease-associated changes in two critical glomerular ultrastructures: the glomerular basement membrane and podocyte foot processes. In addition to significantly outperforming existing tools, TEAMKidney shows close agreement with pathological expert measurements used in clinical assessment protocols. By reducing dependence on manual tracing while preserving expert-level accuracy, TEAMKidney demonstrates that deep learning can substantially reduce the burden of image analysis in both clinical pathology and biomedical research settings.

bioinformatics↗

bioRxiv2025

VaMiAnalyzer: An open source, python-based application for analysis of 3D in vitro vasculogenic mimicry assays

BackgroundVasculogenic mimicry (VM) is the phenomenon whereby non-vascular tumor cells develop vascular-like structures. VM is linked to more aggressive tumor phenotypes including higher rates of metastasis and invasion and is potentially resistant to anti-angiogenic cancer therapies. VM is investigated in vitro using 3D VM assays with microscopy images capturing the resulting VM structures. The standard method to quantify endpoint data is to count various structural features manually, which is time-consuming and open to bias. At present, no software solutions have been developed to specifically address the analysis and quantification of VM structures. ResultsTo address this limitation, we developed an open source, python-based application, VaMiAnalyzer, allowing straightforward quantification of several VM structural features. The application follows a two-step approach that optionally corrects and enhances the raw input images and then analyzes and quantifies the VM features. ConclusionsVaMiAnalyzer is stand-alone software that allows automated measurement of VM structural features from phase-contrast microscopy images. It produces results that are strongly consistent with manual counts but in a significantly shorter time, allowing speedy, non-biased analysis of VM from microscopy images.

bioinformatics↗

bioRxiv2025

FASN Inactivation-induced Progranulin (GRN) Expression Promotes Lysosome-dependent Cell Death to Suppress Leukemogenesis

Cancer cells rely on lipogenesis in addition to exogenous lipid uptake, and fatty acid synthase (FASN) is aberrantly overexpressed in myeloid leukemia. However, the precise role of FASN in leukemogenesis in vivo remains elusive. Here, we demonstrated that FASN is essential for leukemogenesis. Ablation of FASN impeded leukemic cell growth, survival, clonogenicity in vitro, and the leukemic burden in vivo. Conditional knockout of FASN barely affected hematopoiesis but significantly attenuated the leukemic progression of MLL-AF9 transplants. By screening a library of platensimycin derivatives, we identified compound MS-C19 as a potent FASN inhibitor. Pharmacological targeting of FASN by MS-C19 suppressed leukemic cell growth and clonogenicity in clinical AML blasts. Mechanistically, both MS-C19 treatment and FASN deficiency triggered the activation of lysosomal and inflammatory gene expression. Loss of FASN led to lysosomal membrane permeabilization and subsequent lysosome-associated cell death, but not lysosome biogenesis. We further identified that GRN, a lysosomal and neuroinflammatory gene, was potently transcribed by TFEB upon FASN loss. Depletion of GRN significantly reversed the inhibitory effects caused by FASN knockdown. Our work demonstrates that FASN is a therapeutic target for myeloid leukemia, and inactivation of FASN by the lead compound MS-C19 provides an alternative approach for leukemic intervention. Key PointsO_LIGenetic ablation of FASN induces apoptosis, impedes leukemia cell growth, and improves the survival of leukemic xenografts. C_LIO_LITargeting FASN by the compound MS-C19, a derivative of the natural microbe antibiotic platensimycin, reduces the cell survival of leukemic cell lines, the clinical leukemic blasts, and MLL-AF9 transformed murine leukemic cells. C_LIO_LIFASN inhibition induces lysosome-dependent cell death that relies on Progranulin (GRN) expression. C_LI

cancer biology↗

bioRxiv2024

Digital pathology assessment of kidney glomerular filtration barrier ultrastructure in an animal model of podocytopathy

BackgroundTransmission electron microscopy (TEM) images can visualize kidney glomerular filtration barrier ultrastructure, including the glomerular basement membrane (GBM) and podocyte foot processes (PFP). Podocytopathy is associated with glomerular filtration barrier morphological changes observed experimentally and clinically by measuring GBM or PFP width. However, these measurements are currently performed manually. This limits research on podocytopathy disease mechanisms and therapeutics due to labor intensiveness and inter-operator variability. MethodsWe developed a deep learning-based digital pathology computational method to measure GBM and PFP width in TEM images from the kidneys of Integrin-Linked Kinase (ILK) podocyte-specific conditional knockout (cKO) mouse, an animal model of podocytopathy, compared to wild-type (WT) control mouse. We obtained TEM images from WT and ILK cKO littermate mice at 4 weeks old. Our automated method was composed of two stages: a U-Net model for GBM segmentation, followed by an image processing algorithm for GBM and PFP width measurement. We evaluated its performance with a 4-fold cross-validation study on WT and ILK cKO mouse kidney pairs. ResultsMean (95% confidence interval) GBM segmentation accuracy, calculated as Jaccard index, was 0.73 (0.70-0.76) for WT and 0.85 (0.83-0.87) for ILK cKO TEM images. Automated and manual GBM width measurements were similar for both WT (p=0.49) and ILK cKO (p=0.06) specimens. While automated and manual PFP width measurements were similar for WT (p=0.89), they differed for ILK cKO (p<0.05) specimens. WT and ILK cKO specimens were morphologically distinguishable by manual GBM (p<0.05) and PFP (p<0.05) width measurements. This phenotypic difference was reflected in the automated GBM (p<0.05) more than PFP (p=0.06) widths. ConclusionsThese results suggest that certain automated measurements enabled via deep learning-based digital pathology tools could distinguish healthy kidneys from those with podocytopathy. Our proposed method provides high-throughput, objective morphological analysis and could facilitate podocytopathy research and translate into clinical diagnosis. Key pointsO_LIWe leveraged U-Net architecture in an algorithm to measure the widths of glomerular basement membrane and podocyte foot processes. C_LIO_LIDeep learning-based automated measurement of glomerular filtration barrier morphology has promise in podocytopathy research and diagnosis. C_LI

pathology↗

bioRxiv2024

scGND: Graph neural diffusion model enhances single-cell RNA-seq analysis

Single-cell sequencing technologies have played a pivotal role in advancing biomedical research over the last decade. With the evolution of deep learning, a variety of models based on deep neural networks have been developed to improve the precision of single-cell RNA sequencing (scRNA-seq) analysis from multiple angles. However, deep learning models currently used in scRNA-seq analysis frequently suffer from a lack of interpretability. In this study, we present a novel physics-informed graph generative model, termed Single Cell Graph Neural Diffusion (scGND). This model is founded on solid mathematical concepts and provides enhanced interpretability. Unlike methods that focus solely on gene expression in individual cells, scGND concentrates on the cell-cell interaction graph, incorporating two key physical concepts: local and global equilibrium. We show that achieving a balance between local and global equilibrium significantly improves the geometric properties of the graph, aiding in the extraction of inherent biological insights from the cell-cell interaction graph at multiple scales. The effectiveness of scGND has been proven through benchmark tests involving five independent scRNA-seq datasets from various tissues and species. scGND consistently achieves better or comparable results comparing with several established competitors in both clustering and trajectory analysis. scGND represents a comprehensive generative model based on cell graph diffusion, demonstrating considerable promise for both theoretical and practical applications in scRNA-seq data analysis.

bioinformatics↗

bioRxiv2024

Loss of STIM1 and STIM2 in salivary glands disrupts ANO1 function but does not induce Sjogren's disease

Sjogrens disease (SjD) is an autoimmune disease characterized by xerostomia (dry mouth), lymphocytic infiltration into salivary glands and the presence of SSA and SSB autoantibodies. Xerostomia is caused by hypofunction of the salivary glands and has been involved in the development of SjD. Saliva production is regulated by parasympathetic input into the glands initiating intracellular Ca2+ signals that activate the store operated Ca2+ entry (SOCE) pathway eliciting sustained Ca2+ influx. SOCE is mediated by the STIM1 and STIM2 proteins and the ORAI1 Ca2+ channel. However, there are no studies on the effects of lack of STIM1/2 function in salivary acini in animal models and its impact on SjD. Here we report that male and female mice lacking Stim1 and Stim2 (Stim1/2K14Cre) in salivary glands showed reduced intracellular Ca2+ levels via SOCE in parotid acini and hyposalivate upon pilocarpine stimulation. Bulk RNASeq of the parotid glands of Stim1/2K14Cre mice showed a decrease in the expression of Stim1/2 but no other Ca2+ associated genes mediating saliva fluid secretion. SOCE was however functionally required for the activation of the Ca2+ activated chloride channel ANO1. Despite hyposalivation, ageing Stim1/2K14Cre mice showed no evidence of lymphocytic infiltration in the glands or elevated levels of SSA or SSB autoantibodies in the serum, which may be linked to the downregulation of the toll-like receptor 8 (Tlr8). By contrast, salivary gland biopsies of SjD patients showed increased STIM1 and TLR8 expression, and induction of SOCE in a salivary gland cell line increased the expression of TLR8. Our data demonstrate that SOCE is an important activator of ANO1 function and saliva fluid secretion in salivary glands. They also provide a novel link between SOCE and TLR8 signaling which may explain why loss of SOCE does not result in SjD.

physiology↗

bioRxiv2022

Molecular Determinants for Differential Activation of the Vibrio parahaemolyticus Bile Acid Receptor

Bile acids are important for digestion of food and for antimicrobial activity. Pathogenic Vibrio parahaemolyticus senses bile acids via the co-component signal transduction system receptor VtrA/VtrC, an obligate membrane heterodimer. Intestinal bile acids bind to the periplasmic domain of the VtrA/VtrC complex, activating a DNA-binding domain in VtrA that induces expression of another membrane protein, VtrB. VtrB induces expression of the pathogenic Type III Secretion System 2. The bile acid taurodeoxycholate (TDC) activates VtrA/VtrC-induced VtrB expression, while others such as chenodeoxycholate (CDC) do not. This study demonstrates both CDC and TDC bind to the VtrA/VtrC periplasmic heterodimer using isothermal titration calorimetry (ITC). The crystal structure of the VtrA/VtrC heterodimer bound to CDC revealed it binds in the same hydrophobic pocket as TDC, but differently. Mutation of the binding pocket caused a decrease in bile acid binding affinity with exception of the S123A mutant, which bound with a similar affinity as the wild-type protein. The S123A mutant decreased TDC-induced T3SS2 activation, providing a molecular explanation for the selective activation of the T3SS2 by bile acids.

microbiology↗

bioRxiv2021

Plucking a string or playing a G? Degree of choice abstraction impacts human reinforcement learning

In reinforcement learning (RL) experiments, participants learn to make rewarding choices in response to different stimuli; RL models use outcomes to estimate stimulus-response values which change incrementally. RL models consider any response type indiscriminately, ranging from more concretely defined motor choices (e.g. pressing a key with the index finger), to more general choices that can be executed in a number of ways (e.g. getting dinner at the restaurant). But does the learning process vary as a function of the choice type? In Experiment 1, we show that it does: participants were slower and less accurate in learning correct choices of a general format compared to learning more concrete, motor actions. Using computational modeling, we show that two mechanisms contribute to this. First, there was evidence of irrelevant credit assignment: the values of motor actions interfered with the values of other choice dimensions, resulting in more incorrect choices when the correct response is not defined by a single motor action; second, information integration for relevant general choices was slower. In Experiment 2, we replicated and further extended the findings from Experiment 1, by showing that slowed learning was attributable to weaker working memory use, rather than slowed RL learning. In both experiments we ruled out the explanation that the difference in performance between two condition types was driven by difficulty/different levels of complexity. We conclude that defining a more abstract choice space used by multiple learning systems for credit assignment recruits executive resources, limiting how much such processes then contribute to fast learning.

animal behavior and cognition↗

bioRxiv2020

Mutations, Recombination and Insertion in the Evolution of 2019-nCoV

BackgroundThe 2019 novel coronavirus (2019-nCoV or SARS-CoV-2) has spread more rapidly than any other betacoronavirus including SARS-CoV and MERS-CoV. However, the mechanisms responsible for infection and molecular evolution of this virus remained unclear. MethodsWe collected and analyzed 120 genomic sequences of 2019-nCoV including 11 novel genomes from patients in China. Through comprehensive analysis of the available genome sequences of 2019-nCoV strains, we have tracked multiple inheritable SNPs and determined the evolution of 2019-nCoV relative to other coronaviruses. ResultsSystematic analysis of 120 genomic sequences of 2019-nCoV revealed co-circulation of two genetic subgroups with distinct SNPs markers, which can be used to trace the 2019-nCoV spreading pathways to different regions and countries. Although 2019-nCoV, human and bat SARS-CoV share high homologous in overall genome structures, they evolved into two distinct groups with different receptor entry specificities through potential recombination in the receptor binding regions. In addition, 2019-nCoV has a unique four amino acid insertion between S1 and S2 domains of the spike protein, which created a potential furin or TMPRSS2 cleavage site. ConclusionsOur studies provided comprehensive insights into the evolution and spread of the 2019-nCoV. Our results provided evidence suggesting that 2019-nCoV may increase its infectivity through the receptor binding domain recombination and a cleavage site insertion. One Sentence SummaryNovel 2019-nCoV sequences revealed the evolution and specificity of betacoronavirus with possible mechanisms of enhanced infectivity.

microbiology↗

bioRxiv2019

Accumulation of genetic variants associated with immunity in the selective breeding of broilers

To satisfy an increasing demand for dietary protein, the poultry industry has employed genetic selection to increase the growth rate of broilers by over 400% in the past 50 years. Although modern broilers reach a marketable weight of [~]2 kg in a short span of 35 days, a speed twice as fast as a broiler 50 years ago, the expedited growth has been associated with several negative detrimental consequences. Aside from heart and musculoskeletal problems, which are direct consequences of additional weight, the immune response is also thought to be altered in modern broilers. Given that identifying the underlying genetic basis responsible for a less sensitive innate immune response would be economically beneficial for poultry breeding, we decided to compare the genomes of two unselected meat control strains that are representative of broilers from 1957 and 1978, and a current commercial broiler line. Through analysis of genetic variants, we developed a custom prioritization strategy to identify genes and pathways that have accumulated genetic changes and are biologically relevant to immune response and growth performance. Our results highlight two genes, TLR3 and PLIN3, with genetic variants that are predicted to enhance growth performance at the expense of immune function. Placing these new genomes in the context of other chicken lines, reveal genetic changes that have specifically arisen in selective breeding programs that were implemented in the last 50 years.

genomics↗

bioRxiv2018

Reconstructing the human first trimester fetal-maternal interface using single cell transcriptomics

During the early weeks of human pregnancy, the fetal placenta implants into the uterine mucosa (decidua) where placental trophoblast cells intermingle and communicate with maternal cells. Here, we profile transcriptomes of [~]50,000 single cells from this unique microenvironment, sampling matched first trimester maternal blood and decidua, and fetal cells from the placenta itself. We define the cellular composition of human decidua, revealing five distinct subsets of decidual fibroblasts with differing growth factors and hormone production profiles, and show that fibroblast states define two distinct decidual layers. Among decidual NK cells, we resolve three subsets, each with a different immunomodulatory and chemokine profile. We develop a repository of ligand-receptor pairs (www.CellPhoneDB.org) and a statistical tool to predict the probability of cell-cell interactions via these pairs, highlighting specific interactions between decidual NK cells and invading fetal extravillous trophoblast cells, maternal immune and stromal cells. Our single cell atlas of the maternal-fetal interface reveals the cellular organization and interactions critical for placentation and reproductive success.

developmental biology↗