Biology subjects

ZHANG, Z.

Publications and source records attributed to ZHANG, Z..

8 recordsLinked to original sources

bioRxiv2025

STELLA: Self-Evolving LLM Agent for Biomedical Research

The staggering complexity of modern biomedical research has intensified the aspiration for a generalist "Biomedical World Model", yet current AI agents remain constrained by static capabilities and a lack of self-evolution mechanisms. To bridge this gap, we present STELLA, a self-evolving multimodal agent designed to progressively refine its computational reasoning and physical execution through interaction. STELLA operates via a collaborative multi-agent framework (comprising Manager, Developer, Critic, Critic, and Tool Creation agents) that continuously updates reasoning templates and autonomously expands a dynamic "Tool Ocean". We demonstrate STELLAs capabilities on the created Tool Creation Benchmark, where it attains a score of 4.01/5 with 100% task completion, significantly outperforming state-of-the-art models including GPT-5, Claude 4 Opus, and Biomni. Beyond computational metrics, STELLA drives experimentally validated scientific discovery. In oncology, the agent identified Butyrophilin Subfamily 3 Member A1 (BTN3A1) as a novel negative regulator of natural killer (NK) cell function in acute myeloid leukemia (AML), verified via CRISPR knockout studies. In protein engineering, STELLA orchestrated a complete directed evolution workflow for the enzyme strictosidine synthase, identifying variants, notably M276L, exhibiting more than a two-fold improvement in catalytic activity. Finally, the system extends to physical laboratory automation by training Vision-Language-Action (VLA) models through a Decompose-Monitor-Recover mechanism, which increased success rates from 17% to 82%. By integrating autonomous tool evolution, biological discovery, and robotic control, STELLA offers a blueprint for a self-evolving world model in the life sciences.

bioinformatics↗

bioRxiv2024

A genome-wide ATLAS of liver chromatin architecture reveals that sex dictates diet-induced nucleosome dynamics

The three-dimensional organization of the genome plays an important role in cellular function. Alterations between open and closed chromatin states contributes to DNA binding, collaborative transcriptional activities and informs post-transcriptional processing. The liver orchestrates systemic metabolic control and has the ability to mount a rapid adaptive response to environmental challenges. We interrogated the chromatin architecture in liver under different dietary cues. Using ATAC-seq, we mapped over 120,000 nucleosome peaks, revealing a remarkably preserved hepatic chromatin landscape across feeding conditions. Stringent analysis of nucleosome rearrangements in response to diet revealed that sex is the dominant factor segregating changes in chromatin accessibility. A lipid-rich diet led to a more accessible chromatin confirmation at promoter regions in female mice along with enrichment of promoter binding CCAAT-binding domain proteins. Male liver exhibited stronger binding for nutrient sensing nuclear receptors. Integrative analysis with gene expression corroborated a role for chromatin states in informing functional differences in metabolic traits. We distinguished the impact of gonadal sex and chromosomal sex as determinants of chromatin modulation by diet using the Four Core Genotypes mouse model. Our data provide mechanistic evidence underlying the regulation for the critical sex-dimorphic GWAS gene, Pnpla3. In summary, we provide a comprehensive epigenetic resource in murine liver that uncovers the complexity of chromatin dynamics in response to diet and sex. HighlightsATAC-Seq, RNA-Seq, and FCG model-integrated analysis unravel sex differences in chromatin accessibility and transcriptome responses to dietary challenges. Lipid-rich diet led to sex-biased chromatin confirmation at promoter regions. Gonadal sex emerged as the most prevalent determinant of the sex bias hepatic chromatin modulation by lipid-rich diets. The critical sex-dimorphic GWAS gene Pnpla3 is suppressed by testosterone, which underlies hepatic differences in expression between the sexes.

molecular biology↗

bioRxiv2024

Orchestrating the Acquisition of Oligodendrocyte Precursor Cell versus Olfactory Bulb Interneuron Fates through Olig1/2 during Mammalian Cortical Gliogenesis and Gliomagenesis

The hijacking of developmental gliogenesis programs is a hallmark of glioblastoma (GBM), in which glial precursor cells (GPCs) typically differentiate into both neurons and glial cells. In GBM, this process is disrupted, leading to the overproduction of proliferative glial-like cells. Our study demonstrates that the knockout of Olig1/2 in both normal development and gliomagenesis causes GPCs to shift from generating highly proliferative oligodendrocyte precursor cells to producing non-proliferative olfactory bulb interneurons. Mechanistically, Olig1/2 play dual roles by orchestrating distinct transcriptional programs in GPCs, particularly inhibiting the expression of Gsx2 through direct binding to its multiple enhancers. Additionally, we provide compelling evidence that human H3.3G34R/V-mutant tumors, a subtype of high-grade gliomas, originate from dorsal cortical-derived GPCs rather than from the previously assumed progenitors in the ventral basal ganglia. Collectively, our findings reveal a previously unrecognized role of Olig1/2 in both gliogenesis and gliomagenesis, offering deeper insights into the connections between normal neural development and tumorigenesis. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=159 HEIGHT=200 SRC="FIGDIR/small/623106v1_ufig1.gif" ALT="Figure 1"> View larger version (69K): org.highwire.dtl.DTLVardef@10ea118org.highwire.dtl.DTLVardef@1725006org.highwire.dtl.DTLVardef@1e3ce25org.highwire.dtl.DTLVardef@f300ee_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2024

FoldMark: Protecting Protein Generative Models with Watermarking

Proteins are the principal architects of life, fueling advances in bioengineering, drug discovery, and synthetic biology. The integration of generative AI with computational protein science has revolutionized protein design while also posing dual-use risks, such as enabling the creation of pandemic-capable proteins that require strong biosecurity safeguards. Here, we introduce FoldMark, a first-of-its-kind watermarking strategy leveraging distributional and evolutionary principles tailored for protein generative models, balancing watermark capacity and structural quality. FoldMark achieves over 95% watermark bit accuracy at 32 bits with minimal impact on structural quality (>0.9 scTM scores) for leading models including AlphaFold3, ESMFold, RFDiffusion, and RFDiffusionAA. For user tracing, FoldMark can successfully trace up to 1 million users. To validate FoldMark in wet lab, we applied it to structure-based design of EGFP and CRISPR-Cas13, showing wildtype-level function (98% fluorescence, 95% editing efficiency) and >90% watermark detection, demonstrating its practical utility for safeguarding AI-driven protein research.

bioinformatics↗

bioRxiv2023

Comprehensive Identification of Pathogenic Microbes and Antimicrobial Resistance Genes in Food Products Using Nanopore Sequencing-Based Metagenomics

Foodborne pathogens, particularly antimicrobial-resistant (AMR) bacteria, remain a significant threat to global health. Conventional culture-based approaches for detecting infectious agents are limited in scope and time-consuming. Metagenomic sequencing of food products offers a rapid and comprehensive approach to detect pathogenic microbes, including AMR bacteria. In this study, we used nanopore-based metagenomic sequencing to detect pathogenic microbes and antimicrobial resistance genes (ARGs) in 260 food products, including raw meat, sashimi, and ready-to-eat (RTE) vegetables. We identified Clostridium botulinum and Staphylococcus aureus as the predominant foodborne pathogens in the food samples, particularly prevalent in fresh, peeled, and minced foods. Importantly, RTE-vegetables, which harbored Acinetobacter baumannii and Toxoplasma gondii as the dominant foodborne pathogens, displayed the highest abundance of carbapenem resistance genes among the different food types. Exclusive blaCTX-Mgene-carrying plasmids were found in both RTE-vegetables and sashimi. Additionally, we assessed the impact of host DNA and sequencing depth on microbial profiling and ARG detection, highlighting the preference for nanopore sequencing over Illumina for ARG detection. A lower sequencing depth of around 25,000 is adequate for effectively profiling bacteria in food samples, whereas a higher sequencing depth of approximately 700,000 is required to detect ARGs. Our workflow provides insights into the development of food safety monitoring tools and can assess the potential risk to human health from foodborne pathogens and ARGs. This approach has the potential to revolutionize the screening of food products and enable more efficient and accurate detection of foodborne pathogens and ARGs, thereby reducing the risks of foodborne illness and improving public health.

microbiology↗

bioRxiv2023

UBAP2L drives scaffold assembly of nuclear pore complexes at the intact nuclear envelope

Assembly of macromolecular complexes at correct cellular sites is crucial for cell function. Nuclear pore complexes (NPCs) are large cylindrical assemblies with eightfold rotational symmetry, built through hierarchical binding of nucleoporins (Nups) forming distinct subcomplexes. Here, we uncover a direct role of ubiquitin-associated protein 2-like (UBAP2L) in the biogenesis of properly organized and functional NPCs at the intact nuclear envelope (NE) in human cells. UBAP2L localizes to the nuclear pores and drives the formation of the Y-complex, an essential scaffold component of the NPC, and its localization to the NE. UBAP2L facilitates the interaction of the Y-complex with POM121 and Nup153, the critical upstream factors in a well-defined sequential order of Nups assembly onto NE during interphase. Timely localization of the cytoplasmic Nup transport factor fragile X-related protein 1 (FXR1) to the NE and its interaction with the Y-complex are likewise dependent on UBAP2L. Thus, this NPC biogenesis mechanism integrates the cytoplasmic and the nuclear NPC assembly signals and ensures efficient nuclear transport, adaptation to nutrient stress and cellular proliferative capacity, highlighting the importance of NPC homeostasis at the intact nuclear envelope. TeaserLiao et al. show how UBAP2L drives the assembly of the scaffold elements into symmetrical and functional NPCs at the nuclear envelope in human cells.

cell biology↗

bioRxiv2023

KCNN4 links PIEZO-dependent mechanotransduction to NLRP3 inflammasome activation

Immune cells sense the microenvironment to fine-tune their inflammatory responses. Patients with cryopyrin associated periodic syndrome (CAPS), caused by mutations in the NLRP3 gene, develop auto-inflammation triggered by non-antigenic, e.g. environmental cues. However, the underlying mechanisms are poorly understood. Here, we uncover that KCNN4, a calcium-activated potassium channel, links PIEZO-mediated mechanotransduction to NLRP3 inflammasome activation. Yoda1, a PIEZO1 agonist, lowers the threshold for NLRP3 inflammasome activation. PIEZO-mediated sensing of stiffness and shear stress increases NLRP3-dependent inflammation. Myeloid-specific deletion of PIEZO1/2 protects mice from gouty arthritis. Activation of PIEZO1 triggers calcium influx, which activates KCNN4 to evoke potassium efflux promoting NLRP3 inflammasome activation. Activation of PIEZO signaling is sufficient to activate the inflammasome in cells expressing CAPS-causing NLRP3 mutants via KCNN4. Finally, pharmacologic inhibition of KCNN4 alleviates auto-inflammation in CAPS patient cells and in CAPS-mimicking mice. Thus, PIEZO-dependent mechanical inputs augment inflammation in NLRP3-dependent diseases including CAPS. One Sentence SummaryPIEZO-mediated mechanotransduction stimulates KCNN4-dependent potassium efflux to potentiate NLRP3 inflammasome activation.

immunology↗

bioRxiv2021

Defective endosome-TGN retrograde transport promotes NLRP3 inflammasome activation

Inflammasome complexes are pivotal in the innate immune response to pathogens and other danger signals1-4. The NLRP3 inflammasome is activated in response to a broad variety of cellular stressors. Most of the stimuli act in a potassium efflux-dependent manner but a primary and converging sensing mechanism by the NLRP3 receptor initiating inflammasome assembly remains ill-defined. Here we show that NLRP3 activators disrupt endosome-TGN retrograde transport (ETRT) and lead to localization of NLRP3 to endosomal vesicles. Genetic and pharmacologic perturbation of ETRT leads to accumulation of phosphoinositol-4-phosphate (PI4P) in endosomes to which NLRP3 is recruited. Disruption of ETRT potentiates NLRP3 inflammasome activation in murine and human macrophages in vitro. Mice with defects in ETRT in the myeloid compartment are more susceptible to LPS-induced sepsis showing enhanced mortality and IL-1{beta} serum levels as compared to control animals. Our study thus uncovers that changes in endocytic trafficking mediate NLRP3-dependent inflammatory responses.

immunology↗