Biology subjects

Zhang, K.

Publications and source records attributed to Zhang, K..

At least 73 records · Page 4Linked to original sources

bioRxiv2026

Role of desolvation on biomolecular liquid-liquid phase separation

Biomolecular condensates play essential roles in cellular organization and are implicated in diverse pathological processes. Their formation is driven by liquid-liquid phase separation (LLPS), a process that requires coordinated multistep desolvation of biomolecular chains and multivalent inter-chain interactions. Although coarse-grained (CG) models with implicit solvent are widely used to probe LLPS thermodynamics and kinetics, they typically neglect water-mediated desolvation effects, limiting their accuracy and mechanistic interpretability. Here, guided by all-atom simulations and experimental measurements, we develop a desolvation-aware implicit-solvent CG model by incorporating residue-level desolvation terms directly into the pairwise energy function and apply it to investigate LLPS of intrinsically disordered proteins. Incorporating these desolvation interactions reshapes the phase diagram, alleviating dense-phase overcompaction. Notably, we observe an approximately linear correlation between the temperature gap (simulation temperature relative to the critical point) and the extent of conformational expansion accompanying the dilute-to-dense phase transition, a result further supported by theoretical analysis. We also find that desolvation barriers slow early density-fluctuation growth and shorten transient kinetic arrest, whereas solvent-separated contact interactions exert the opposite effects. Both terms further modulate chain mobility within mature condensates through competing packing and energy-landscape effects. Together, this framework enables an efficient representation of desolvation in CG simulations and reveals how desolvation energetics shape both the thermodynamic landscape and kinetic properties of biomolecular LLPS.

biophysics↗

bioRxiv2026

Functional Locality-Aligned Learning Reveals Structure-Function Causality in Enzyme Kinetics

Accurate estimation of enzyme kinetic parameters is essential for enzyme engineering and industrial biocatalysis, yet their experimental measurement remains labor-intensive and costly. Although machine learning offers an efficient alternative, existing methods still struggle to generalize to unseen enzymes and substrates. In particular, current three-dimensional (3D) structure-aware approaches rely on whole-enzyme 3D geometric structures and neglect substrate 3D geometry, often yielding limited or even degraded performance. We identify a fundamental limitation underlying these methods: a mismatch between the structural representation learning scale and the functional locality scale, which weakens structure-function causality in enzyme kinetics. To address this issue, we introduce EnzymePlex, a functional locality-aligned framework that aligns inductive biases with the localized structural determinants of enzyme function by prioritizing catalytic pockets, integrating substrate 3D geometry, and modeling nuanced enzyme-substrate interplay under the guidance of pocket-level structural priors. EnzymePlex achieves state-of-the-art performance across multiple benchmarks and substantially improves generalization under stringent out-of-distribution evaluations. Beyond predictive accuracy, EnzymePlex learns mechanistically aligned representations, with attention enriched at catalytic pocket residues and substrate reaction centers despite receiving no explicit super-vision for either. Moreover, when applied to recently reported wet-lab data, EnzymePlex effectively prioritizes high-activity enzyme variants and identifies potent inhibitors, highlighting its potential to accelerate enzyme engineering and drug discovery.

biochemistry↗

bioRxiv2026

Kindlin-2-Moesin interaction orchestrates sprouting angiogenesis via modulating endothelial membrane mechanics and VEGF signaling

Membrane mechanics play a crucial role in cellular signaling and fate determination, yet their impact on angiogenesis remains poorly understood. Here, we identify Kindlin-2 as a key regulator of sprouting angiogenesis via regulating endothelial membrane tension through its interaction with Moesin, a crucial linker protein between the cell membrane and the actin cortex. Mechanistically, Kindlin-2 binds to the N62 residue of Moesin, limiting its overactivation and maintaining proper membrane tension to facilitate VEGFR2 endocytosis and downstream signaling. Using both developmental and pathological models, we demonstrate that the interaction of Kindlin-2 and Moesin is enhanced in high angiogenic conditions, and endothelial Kindlin-2 deletion reduces angiogenesis. Furthermore, mutation of Moesin at N62 phenocopies the effects of Kindlin-2 loss. Together, these findings uncover a previously unrecognized mechanism linking membrane tension regulation to angiogenesis and provide new insights into targeting the Kindlin-2-Moesin axis for therapeutic intervention in neovascular diseases.

developmental biology↗

bioRxiv2026

Daraxonrasib (RMC-6236) is an effective targeted therapy for RAS-mutant neuroblastoma

Neuroblastoma (NB) is the most common extracranial solid tumor in children. Relapsed or refractory (R/R) high-risk (HR) NB tumors continue to exhibit poor outcomes despite intensive and protractive multimodal therapy. Activating mutations in the RAS- mitogen-activated protein kinase (MAPK) pathway are frequently observed in R/R HRNB. The early promise of ALK inhibitors to treat ALK-mutant NB underscores the ability of appropriate targeted therapies to improve outcomes for HRNB patients. While MAPK pathway activation is prominent in HRNB, FDA-approved MEK inhibitors and KRAS G12C inhibitors have failed to demonstrate significant preclinical single-agent activity. Daraxonrasib (RMC-6236), a potent and selective RAS(ON) inhibitor, has demonstrated activity in both preclinical models and early phase clinical trials of RAS-mutant adult cancers. A subset of R/R HRNB tumors is noteworthy for containing diverse RAS-mutations, providing rationale for RMC-6236 investigation. In this study, we evaluated the therapeutic efficacy and oncogenic signaling modulation of RMC-6236 across NB models harboring RAS pathway activation. RMC-6236 as a single-agent treatment led to a significant decrease in cell viability, suppression of downstream MAPK signaling, upregulation of the MAPK pathway effector protein BIM, and increased cell death in RAS-mutant NB models as well as in NF1-mutant NB models. In vivo studies evidenced that RMC-6236 had on-target activity that significantly reduced tumor growth and extended survival in RAS-mutant HRNB mouse models. Furthermore, RMC-6236-induced both BCL-2 and BIM upregulation and enhancement of BIM:BCL-2 complexes in RAS-mutant NB. As such, the BCL-2 inhibitor venetoclax further enhanced RMC-6236-mediated killing by disrupting RMC-6236 enhanced BIM:BCL-2 complexes. These findings demonstrate that RMC-6236 is a rationale targeted therapy for RAS-mutant NB, a subset of NB that is progressively understood as conferring particularly poor outcomes. RMC-6236 is a clinically relevant drug that can successfully target the MAPK pathway in these cancers. This study supports expanded clinical testing of this novel therapy to this important subset of neuroblastoma.

cancer biology↗

bioRxiv2026

Enhancing Pan-cancer Spatial Transcriptomics atSingle-cell Resolution with stPainter

Subcellular spatial transcriptomics technologies offer unprecedented views of tissue architecture but are fundamentally constrained by sparse gene panels and limited detection sensitivity. Current computational enhancement strategies typically rely on tissue-matched single-cell RNA sequencing (scRNA-seq) references and necessitate computationally intensive retraining for each dataset, impeding their scalability and clinical applicability. Here, we present STPAINTER, a conditional generative model that leverages a massive pretraining pan-cancer scRNA-seq atlas to universally enhance spatial transcriptomics data. Built upon a latent diffusion architecture with stochastic differential equation-guided generation, STPAINTER learns a universal manifold of cellular states to reconstruct genome-wide expression profiles from sparse spatial measurements. Uniquely, our pretraining paradigm enables zero-shot generalization and empowers downstream tasks by providing imputed transcriptomes and informative latent variables to enhance resolution at both the gene and cluster levels. Applied STPAINTER upon 6 spatial transcriptomics datasets of different cancer types, we demonstrate that our model empowers high-fidelity downstream analyses, including fine-grained subpopulation clustering and pathway enrichment. Furthermore, cross-validation with spatially resolved proteomics (CODEX) confirms the biological veracity of the imputed cellular landscapes. STPAINTER provides a robust, scalable framework for decoding complex tumor microenvironments without the need for auxiliary sequencing data.

bioinformatics↗

bioRxiv2026

Functional and Structural Basis of Omicron BA.3.2.1 Spike

SARS-CoV-2 Omicron descendant BA.3.2, derived from BA.3 and carrying 39 additional spike substitutions, prompts concerns about altered fitness and antigenicity. Using BA.3.2.1 as a representative sublineage, we have engineered live-attenuated mNeonGreen SARS-CoV-2 encoding spikes from BA.3.2.1 and the JN.1-derived LP.8.1 and XEC, and benchmarked them against BA.3 and KP.3 in Calu-3 cells and primary human airway epithelium. BA.3.2.1 outcompetes BA.3 yet replicates more slowly than JN.1-descendants. Despite increased RBD-hACE2 affinity, BA.3.2.1 spike binds hACE2 less efficiently than LP.8.1, and shows the greatest resistance to neutralization by sera from KP.2/KP.3 infections. Cryo-EM reveals a predominantly closed, compact, asymmetric trimeric spike with a remodeled N-terminal domain, a distinct fusion-peptide-proximal region and altered N-glycosylation. BA.3.2.1 spike adopts a fine-tuned receptor-binding interface and shifted monoclonal-antibody footprints. These findings identify a trade-off between immune evasion and replication fitness driven by closed spike conformation, rationalizing BA.3.2s limited prevalence and underscoring the necessity for continued variant surveillance.

microbiology↗

bioRxiv2026

Shaping hydrogel bioinks into 3D, multiscale, perfusable models using multimodal printing

Despite technological advances, the fabrication of multiscale, multi-material, and topologically complex 3D structures using soft hydrogel bioinks remains a challenge due to the inherent trade-offs between print size/resolution, bioink properties, and design complexity. In this work, we combine additive (macroscale) digital light projection (DLP) mode with subtractive (microscale) two-photon ablation (TPA) mode with multi-material exchange capability. We identify ideal hydrogel bioink formulations that are compatible with both DLP and TPA modes of processing. Technical challenges related to multimodal fabrication such as alignment of multiscale topologies to facilitate seamless media perfusion, soft-hard multi-material printing to facilitate handling of mechanically weak hydrogel constructs, and hydrogel swelling during printing, were resolved. To highlight the novelty of this hybrid platform, we fabricated centimeter-scale bioink constructs with embedded microscale perfusable topologies that cannot be achieved by isolated use of either DLP or TPA modes. This includes simpler microfluidic chips with independently perfusable microchannels to more complex 3D constructs with embedded, multiscale, perfusable dual-fluidic circuits that mimic the alveoli-capillary interface, or microfluidic chips with endothelialized microchannels. The unique ability of this multimodal platform to mimic in vivo-like multiscale complexities can be potentially used to develop next-generation organ-on-chips.

bioengineering↗

bioRxiv2026

An enzyme-level benchmark based on environmental bacterial laccases for predicting contaminant fate in water

Bacterial laccases are widespread multicopper oxidases whose roles in the fate of anthropogenic chemicals in aquatic environments remain poorly understood. Here, we integrate metagenomic analysis, a miniaturized high-throughput assay and machine learning to establish an enzyme-level benchmark for predicting biotransformation of wastewater-relevant trace organic contaminants by laccase-mediator systems. Using a laccase from an ammonia-oxidizing bacterium as a model enzyme, we screened 183 compounds and identified 38 that underwent significant removal. Following phylogenetic analysis of environmental homologs, we expressed and purified two additional laccases from the bacterial methanotrophic phylum Methylmirabilota and an archaeal phylum Thermoproteota, demonstrating the activity of this enzyme family across domains. Graph convolutional network models trained on the dataset achieved up to 78% accuracy in classifying degradable versus persistent chemicals, while quantum-chemical descriptors highlighted key electronic properties governing oxidation. This bottom-up approach to enzyme-chemical interactions establishes a trajectory towards predicting contaminant persistence in engineered and natural waters.

biochemistry↗

bioRxiv2026

A cross-species single-cell epigenome kidney atlas identifies epithelial cells as a driver of epigenetic aging

Epigenetic aging is a hallmark of chronic diseases, arising from sustained injuries and unresolved repairs. To investigate cell-type-specific epigenetic alterations, we built a cross-species single-cell multi-omics atlas of DNA methylomes, chromatin accessibilities, and transcriptomes on healthy, injured (human) and aging (mouse) kidneys. We identified accelerated epigenetic aging dominated by tubular epithelia in diseased kidneys. The pathological state mirrors transcriptional trajectories observed in normal aging, driven by the preferential dysregulation of lineage-specific genes lacking CpG islands. Spatially, these epigenetic changes mapped to pathological niches of failed repair. Co-profiling of single-cell DNA methylation and 3D genome architecture revealed that epithelial repair states in disease undergo significant higher-order genome reorganizations, activating genes associated with renal decline. Our findings demonstrate that epithelial aging is driven by a collapse of 3D chromatin structure and local methylome integrity, which silences cell identity and promotes a non-resolving repair state.

genomics↗

bioRxiv2026

A hypothalamo-septo-hippocampal circuit for REM sleep-dependent consolidation of social memory

REM sleep constitutes a critical window for memory consolidation, yet the brain circuits orchestrating this process remain incompletely defined. Here, we identify a lateral supramammillary nucleus (SuM)-medial septum (MS) projection as a REM sleep-specialized pathway essential for hippocampal memory consolidation. Fiber photometry and optrode recordings revealed that lateral SuM-MS projecting neurons were selectively active during REM sleep. REM-specific optogenetic silencing of this projection impaired consolidation of both social and contextual fear memories. Crucially, silencing of its downstream target, the MS-CA2 pathway, during REM sleep selectively disrupted social memory while sparing contextual fear memory. This functional dissection establishes a hypothalamo-septo-hippocampal circuit (lateral SuM-MS-CA2) dedicated to social memory processing, in parallel to the recently-described direct SuM-CA2 pathway. These results also position the SuM as a REM sleep-hub that routes information via parallel septal pathways to consolidate distinct memory modalities.

neuroscience↗

bioRxiv2026

Structural insights into the recognition and catalysis of tRNA by human NSUN2

The human RNA m5C methyltransferase NSUN2 catalyzes site-specific cytosine methylation across diverse RNA substrates and thereby regulates a wide range of biological and physiological processes. However, the molecular basis by which NSUN2 achieves broad substrate recognition while maintaining catalytic specificity has remained unclear. Here we determine structures of human NSUN2 in both substrate-free and substrate-bound states using X-ray crystallography and cryo-electron microscopy. Structures of NSUN2 in complex with multiple tRNA substrates reveal a structure-first, sequence-tolerant strategy in which NSUN2 actively remodels tRNA architecture, exposing the buried target cytosine and positioning it within the catalytic pocket for methyl transfer. This recognition strategy enables NSUN2 to accommodate diverse tRNA substrates through a largely conserved interaction interface. Guided by these structural insights, we identify a small-molecule inhibitor that suppresses NSUN2 activity and cancer cell proliferation. Together, our findings define the molecular principles underlying NSUN2-mediated RNA m5C modification and provide a structural foundation for targeting NSUN2 in cancers.

molecular biology↗

bioRxiv2026

p62 protein pleomorphism confers tissue-specific function

Protein structure dictate function, and proteins often perform the same function in different cell/tissue-type environments. Here, we report that in Drosophila melanogaster, the broadly expressed protein p62/SQSTM1 instead performs mutually exclusive functions across tissues - autophagy in midgut and mitochondria-associated proteostasis in muscle. These functional differences arise from tissue-specific secondary and tertiary structures adopted by the same p62 polypeptide, rather than protein abundance, isoforms, post-translational modification, or binding-partner availability. The tissue-specific structures are cell autonomous and resistant to environmental perturbation. Deletion of a short intrinsically disordered region converts muscle p62 to a midgut-like structural and functional state, but this forced switch causes muscle dysfunction, linking tissue-specific physiology to alternative structural states of the same polypeptide. We posit that Drosophila p62 exemplifies a mechanism for tissue-specificity in which intrinsic structural plasticity and the physiochemical/molecular environment of a tissue enable a broadly expressed protein to adopt tissue-specific structure and function.

cell biology↗

bioRxiv2026

Dynamic control of Raf-ERK signaling modulates neuronal activity across biological scales

Neuronal activity robustly engages the extracellular signal-regulated kinase (ERK) signaling pathway through Ca2+-dependent mechanisms; however, whether ERK can acutely and causally modulates ongoing neuronal activity remains unsolved due to complex upstream regulation and diverse subcellular functions. Here, we directly address this question using an optogenetic ERK activator, opto-miniRaf, that enables selective, rapid, graded, and reversible control of ERK signaling. Combining this AAV-compatible system with calcium imaging and electrophysiology, we interrogate ERK functions across biological scales, from cultured neurons, acute brain slices, and the intact brain. Acute optogenetic activation of ERK enhances synchronized network burst activity in cultured rat cortical neurons and increases calcium activity of cortical pyramidal neurons in awake and moving mice following non-invasive light stimulation. Together, these results establish ERK signaling as an acute modulator of neuronal and network activity, positioning opto-miniRaf as a generalizable platform for precise spatiotemporal control of intracellular kinase signaling in complex biological systems.

neuroscience↗

bioRxiv2025

Template-based RNA structure prediction advanced through a blind code competition

Automatically predicting RNA 3D structure from sequence remains an unsolved challenge in biology and biotechnology. Here, we describe a Kaggle code competition engaging over 1700 teams and 43 previously unreleased structures to tackle this challenge. The top three submitted algorithms achieved scores within statistical error of the winners of the recent CASP16 competition. Unexpectedly, the top Kaggle strategy involved a pipeline for discovering 3D templates, without the use of deep learning. We integrated this template-modeling pipeline and other Kaggle strategies to develop a single model RNAPro that retrospectively outperformed individual Kaggle models on the same test set. These results suggest a growing importance of template-based modeling in RNA structure prediction.

biophysics↗

bioRxiv2025

Primate-specific adaptation of Ku protects transcriptomic integrity by suppressing Alu-mediated alternative splicing

Accurate pre-mRNA splicing is essential for the transfer of genetic information but faces unique challenge in higher primates due to the massive expansion of intronic Alu elements1-3. While studying Ku, the Ku70/Ku80 heterodimer best known for initiating non-homologous end-joining (NHEJ) by encircling DNA ends4, we discovered that Ku expression increased markedly during primate evolution in parallel with Alu expansion5. Ku binds double-stranded RNA (dsRNA) stem-loops, including those at the antisense Alu (asAlu) elements within introns5. Here, we show that Ku-depletion in human cells has a broad impact on splicing largely independent of cell-cycle states, NHEJ, or innate immune signaling, significantly affecting ~8-10% of quantifiable alternative splicing events. Mechanistically, Ku directly binds exonic asAlu to prevent their aberrant inclusion and binds asAlu within inverted-repeat Alu (irAlu) pairs flanking canonical exons to prevent exon skipping6. Among human tissues, Ku expression in the brain is consistently ~50% lower, correlating with more permissive expression of Alu-derived splice variants, particularly those encoding mitochondrial proteins and RNA-binding factors. Correspondingly, heterozygous Ku loss in patient causes developmental delay, neurological dysfunction, and acidosis. Together our findings identified Ku as a critical suppressor of Alu-associated alternative splicing co-opted during evolution with implications for primate brain function and human disease.

molecular biology↗

bioRxiv2025

Functional dissection of Leishmania major membrane components in resistance to cholesterol-dependent cytolysins

Bacteria use cholesterol-dependent cytolysins (CDCs) to damage eukaryotes. While well-studied in mammals, the mechanisms by which CDCs bind to and kill protozoans remains unclear. CDCs bind to the human pathogen Leishmania major, but only kill in the absence of sphingolipids. The contribution of other leishmanial membrane components to CDC binding and cytotoxicity remains unknown. Here, we used genetic knockouts and inhibitors to determine the contribution of key membrane components to CDC binding and killing in L. major. We analyzed toxin binding and killing using flow cytometry and western blotting. Loss of the virulence factor GP63 enhanced toxicity of perfringolysin O, but not streptolysin O. Plasmenylethanolamine and lipophosphoglycan had minimal contributions to CDC binding and cytotoxicity. Removal of sterols protected cells from CDCs, yet failed to reduce binding. We used CDCs defective in engaging glycans or cholesterol to confirm that CDCs deficient in sterol binding, but not glycan binding, could bound to L. major. Thus, in non-mammalian systems, CDCs may rely on glycans for binding, while using sterols for pore-formation. This suggests that CDCs may not be sterol-specific probes in some non-mammalian systems. We conclude that early-branching eukaryotes use distinct mechanisms from mammals to limit CDC pore-formation and killing.

cell biology↗

bioRxiv2025

VIP-OT: Dissecting Single-Cell Biochemical State Dynamics under Perturbation via Vibrational Painting and Optimal Transport

Dissecting the heterogeneous response of individual cells towards genetic and chemical perturbations is central to understanding the dynamic functions of cells and multicellular systems. However, characterizing and modeling how individual cells transition between different states remains a major challenge. Vibrational imaging provides high-content, biochemically informative, label-free molecular fingerprints of single cells but it remains at its infancy for dynamic or predictive analysis of cell state transition. Here, we introduce Vibrational Painting-Optimal Transport (VIP-OT), an integrated experimental-computational framework that overcomes this fundamental limitation. VIP-OT couples multiplexed infrared (IR) and Raman imaging with optimal transport to computationally reconstruct single-cell perturbation trajectories from unpaired population snapshots. When applying to over 22,000 single-cell spectra profiles from human breast adenocarcinoma cells under 16 drug treatments, this framework can retrospectively trace drug response heterogeneity back to baseline metabolic states. We leverage the inferred cell pairings to develop a machine learning model that accurately predicts the full post-treatment metabolic state of individual cells from their pre-treatment spectra. Furthermore, by modeling transitions across dose gradients, we introduce the concept of Spectral Velocity to map dynamic response trajectories and resolve drug combination effects into distinct, path-dependent molecular routes. Together, VIP-OT opens a new direction for dissecting heterogeneous perturbation responses at single cell resolution and serves as a foundation for building virtual simulators of cells under perturbations through high-throughput, high-content, and low-cost vibrational imaging, as well as interpretable in silico modeling.

bioengineering↗

bioRxiv2025

RamanOmics Decodes Spatial Vibrational-Molecular Architecture and Rewiring in Aging and Repair

Aging and tissue repair involve multilayered and spatially heterogeneous remodeling across transcriptional, biochemical, and cellular dimensions, yet prevailing definitions rely on isolated molecular markers that obscure how biochemical and transcriptional states co-evolve in tissues. Here we present RamanOmics, a multimodal framework that integrates single-nucleus RNA sequencing (snRNA-seq), spatial transcriptomics, and label-free Raman imaging to map the spatial vibrational-biochemical and molecular architecture of aging and senescence directly in intact tissues. Applied to mouse lung and skin, RamanOmics generates spatially resolved biochemical-molecular maps revealing tissue-specific programs: lung senescent cells are enriched for extracellular matrix (ECM) remodeling and TGF-{beta} signaling (Serpine1, Dab2, Igfbp7), whereas skin senescence is dominated by keratinization and barrier homeostasis modules (Krt10, Lor, Sbsn). Across tissues, we identify a conserved branched-chain fatty-acid-linked biochemical profile and Raman signature (1131-1135 cm-{superscript 1}) that robustly marks p21 senescent cells. To unify these layers, we develop a machine learning derived "multimodal barcode" that quantitatively integrates biochemical and transcriptional features, enabling non-destructive identification of senescence in situ. In a wound-healing model, RamanOmics further reveals coordinated reactivation of barrier-repair programs in senescent cells, marked by upregulation of Krt10, Lor, Sbsn, Sfn, and Dmkn together with matching increases in lipid-associated Raman signatures, confirming biological generalizability beyond steady-state aging. By directly integrating gene programs to spatial vibrational-biochemical states, RamanOmics provides a general framework and resource for scalable, multimodal profiling of cellular states.

systems biology↗