Biology subjects

Hong, Y.

Publications and source records attributed to Hong, Y..

At least 19 recordsLinked to original sources

bioRxiv2026

Impacts of different SETD1A mutations and genetic backgrounds on human induced pluripotent stem cell-derived hippocampal neurons

Abstract Human induced pluripotent stem cells (hiPSCs) provide a translational platform to interrogate the impact of genetic risk factors for various disorders on human biology. It has been a common practice to engineer premature stop codons to model predicted loss-of-function mutations using isogenic hiPSCs, yet effects of mutation position for the same gene and genetic background have been rarely tested systematically and at multiple levels for comparison to those of patient-specific mutations. Ultra-rare, heterozygous, and predicted loss-of-function mutations in the SETD1A gene, which encodes a H3K4 methyltransferase, are highly penetrant genetic risk factors for schizophrenia. Here, we generated a large panel of isogenic heterozygous SETD1A mutant hiPSC lines, including premature stop-codon mutations in one early exon and one late exon in five genetic backgrounds and two patient-specific mutations, to gain insight into molecular, cellular, and functional changes in a disease-relevant and highly pure population of human hippocampal dentate granule cell-like neurons. Our multi-level analyses showed that different SETD1A mutations lead to divergent molecular and cellular phenotypes related to synapses, but convergent neuronal network deficits, while the genetic background influences gene expression with no substantial effects at cellular and network levels. Mechanistic analysis confirmed the early exon patient mutation as loss-of-function, but revealed that the predicted loss-of-function patient mutation in the late exon is likely gain-of-function. Consequently, while all mutations decreased global H3K4me3 levels, deficits were only partially rescued by inhibiting H3K4 demethylase LSD1 for early exon mutations, but not for late exon mutations. Together, these results suggest that highly penetrant mutations can lead to similar cellular and functional phenotypes across different genetic backgrounds, whereas different premature stop-codon mutations could have contrasting impacts due to distinct mechanisms. Our findings have practical implications for applying genome-editing in hiPSCs to model disorders, highlight the value of modeling patient-specific mutations and testing multiple mutations, and suggest the need for tailored mutation-dependent therapeutic strategies.

neuroscience↗

bioRxiv2026

CD24-dependent innate immune checkpoint targeting chimera disrupts TGF-β-driven immunosuppression to enhance anti-tumor therapy

The immunosuppressive tumor microenvironment (TME) poses formidable challenges for cancer immunotherapies. Among the complex TME landscape, the cytokine TGF-{beta}1 is an attractive soluble target for reinvigoration of anti-tumor immunity due to its pivotal function in immune cell suppression, extracellular matrix remodeling, and epithelial-to-mesenchymal transition (EMT) promotion, while immune checkpoint receptors exemplified by PD-1 and CD24 are promising surface targets. Though adaptive immune checkpoint inhibition is well established in clinical practice, TGF-{beta} blockade or innate immune checkpoint blockade is yet to be harnessed for tumor treatment. Here, leveraging CD24 overexpression on multiple tumor tissues and its unique recycling kinetics, we designed an innate Immune Checkpoint TArgeting Chimera (ICTAC) G01, a bifunctional CD24 antibody-TGF-{beta}1 receptor fusion protein that disrupts TGF-{beta}1 signaling by TGF-{beta}1 degradation instead of simple blockade and simultaneously inhibits the "don't eat me" CD24 signaling. We demonstrated that G01 effectively suppressed TGF-{beta}1-Smad signaling, reversed the EMT, and attenuated fibrotic collagen deposition. Additionally, it antagonized CD24 signaling to unleash robust antibody-dependent effector activity, further reprogramming the immunologically cold TME toward inflammatory phenotype. By selectively targeting malignant cells while sparing CD24-expressing B cells, G01 demonstrated a favorable safety profile. Notably, in patient-derived tumor fragments (PDTFs), G01 also inflamed the TME, underscoring its translational potential. Together, these results establish ICTAC as an effective strategy for tumor-specific receptor-mediated extracellular protein depletion, which supports its application for reversing TGF-{beta}1; and innate immune checkpoint-mediated tumor immunosuppression by concurrent targeting of both surface and soluble tumor-promoting factors.

immunology↗

bioRxiv2026

Allochthony in stream food webs decreases with temperature across a subcontinental scale

Riverine and riparian food webs are connected via 'allochthonous' resource flows from neighboring ecosystems. These resources, such as terrestrial leaf litter, contribute to benthic macroinvertebrate diets, supporting higher trophic levels. While allochthony occurs in streams globally, its sensitivity to broad environmental gradients remains poorly understood. We explored the drivers of allochthony across six river basins within the contiguous United States, including climate, stream size, and flow permanence. We used Bayesian mixing models to estimate macroinvertebrate dietary proportions in four functional feeding groups. Air temperature best predicted allochthony, with consumers relying less on allochthonous resources with increasing temperature. Our models indicate that macroinvertebrate diets become primarily autochthonous at approximately 15{degrees}C mean annual air temperature, with leaf litter-consuming shredders showing a slightly higher threshold for this shift. We also found increased allochthony in smaller, non-perennial streams. These findings highlight the complexity of stream allochthony and responses to multi-scale environmental factors.

ecology↗

bioRxiv2026

Comparing morphological and molecular methods for stream macroinvertebrate biodiversity assessment across seven watersheds in the southern United States

Freshwater biodiversity is declining at alarming rates globally, yet our capacity to monitor it depends heavily on the methods used for biodiversity assessment. Stream benthic macroinvertebrates are widely used as bioindicators, but traditional morphological identification is labor-intensive and limited by taxonomic expertise, while DNA metabarcoding offers a promising but incompletely evaluated alternative. Here, we compared biodiversity patterns detected by four macroinvertebrate sampling and identification methods (morphological identification of benthic samples, morphological identification of combined benthic and stream-edge samples, eDNA metabarcoding of bulk benthic samples, and eDNA metabarcoding of streamwater) across 62 sites in seven watersheds spanning the southern continental United States. We compared methods across three facets of biodiversity (gamma diversity, alpha diversity, and taxonomic composition) and three taxonomic levels (family, genus, and species). We found that which method detected more taxa shifted with taxonomic resolution: morphological methods detected more families and genera, while metabarcoding methods detected more species. Bulk benthic metabarcoding detected more families and genera than water eDNA metabarcoding, consistent with the "watered-down biodiversity" effect of DNA dilution in water samples, but the two metabarcoding methods did not differ significantly in species richness. All four methods detected largely distinct suites of taxa, with metabarcoding methods yielding more unique taxa than morphological methods. Notably, taxa uniquely detected by water eDNA were dominated by soft-bodied organisms (oligochaetes, leeches, earthworms) and water-column-associated taxa, while benthic eDNA uniquely detected hard-bodied EPT insects. We attribute this pattern to fundamental differences in eDNA shedding, transport, and deposition dynamics across organism types. These results demonstrate that no single method captures a complete picture of stream macroinvertebrate biodiversity, and that method choice should be guided by the taxonomic resolution and community components most relevant to the goals of the bioassessment program.

ecology↗

bioRxiv2026

Hygrochastic Seedpods: Engineering Microclimates for Forest Restoration

Forest restoration at scale depends on sowing seeds directly onto the land, yet seedling establishment is often poor and inconsistent. Seed enhancement technologies, such as priming, coating, and pelleting, improve germination under water, temperature, and salinity stress by supplying polymers, nutrients, or microorganisms at the seed surface. Being mainly conformal films, they act continuously once wetted and cannot gate when a seed is exposed to the soil. They also enclose no air volume, so the microenvironment they create is inseparable from the surrounding soil and has been inferred from plant performance rather than measured directly. Whether an engineered carrier can instead set both the state and the duration of the post-dispersal microenvironment, and how such an effect can be directly quantified, remains unexplored. Here we show that hygrochastic seedpods hold seeds in a timed, directly measurable microclimate and accelerate seedling emergence. The pods are programmable, responsive conical shells enclosing an air cavity around the seed. Two design solutions were investigated in parallel: wooden seedpods that crack open upon hydration, and poly(vinyl alcohol)/boric acid seedpods that gradually disintegrate after sustained wetting. In both, wax coatings act as a programmable diffusion-limited timer that sets activation from hours to days. The pod cavity permits direct sensing and monitoring, which no coating geometries allow. As directly measured during the initial phase after deployment, seedpod interiors ran 2 - 3 C warmer and drier than the surrounding soil, and these differences persisted after opening. Across indoor and field trials in several tree species, pods accelerated early seedling emergence without penalizing subsequent growth. These results identify artificial hygrochastic seedpods as a transient ecological mediator that regulates both the timing of seed exposure and the microclimate around the seed, offering a strategy for improving forest restoration under variable environmental conditions.

bioengineering↗

bioRxiv2026

When DL-Based Prescreening Meets Synthon-Based Docking: Target-Adapting PharmacoNet via MEL-Steered Correction

As chemical libraries expand into the trillions of molecules, Virtual SYNthon Hierarchical Enumeration Screening (V-SYNTHES) has emerged as a leading strategy for making gigascale virtual screening computationally tractable. In V-SYNTHES, a Minimal Enumeration Library (MEL) of chemical fragments is docked against a target first, and only the top-scoring fragments are expanded into full ligands for large-scale docking. However, among the large number of comparably well-docked fragments, only a small fraction can be expanded under a fixed docking budget, leaving most similarly promising fragments unexplored. General-purpose prescreening tools can be adopted to address this constraint, reallocating the same docking budget across a larger pool of fragments' enumerated full ligands by their proxy score. However, such tools are applied without accounting for target-specific pocket environments. One such method, PharmacoNet, predicts interaction hotspots from a protein structure and ranks candidates via graph matching against a fixed set of interaction-type weights. We recognize that V-SYNTHES's initial fragment-docking step, ordinarily used only for selection of best fragments for expansion, already reveals which of these hotspots and interaction types a given pocket actually favors, and we can recover this signal to fine-tune PharmacoNet accordingly. We introduce MEL-Steered PharmacoNet, a parameter-efficient adaptation framework that specializes PharmacoNet to a given target through two composable mechanisms: (i) empirical density-map steering of predicted pharmacophore hotspots, and (ii) empirical fine-tuning of interaction-type scoring weights. Across three structurally distinct GPCR targets (CB2, GPR91, 5-HT2AR), MEL-Steered PharmacoNet achieves substantial enrichment factor (EF100) gains over a random baseline, and improves EF100 over PharmacoNet by 8.94x, 6.87x, and 1.69x, respectively. The fitted per-target weights further reveal distinct, chemically interpretable interaction profiles that PharmacoNet's generic fixed weights fail to capture. These results show that fragment-docking data already generated by the standard V-SYNTHES pipeline can adapt a general-purpose pharmacophore prescreening method to an individual target, significantly improving its performance while retaining its ultra-fast screening ability, with no additional experimental data or model retraining.

bioinformatics↗

bioRxiv2026

Whole-genome resequencing-based comparative variant analysis identifies candidate genes associated with cross-beak phenotype in Huiyang Bearded chickens

Cross-beaks are deemed a threat to poultry health, productivity, and animal welfare. Nevertheless, due to sporadic cases, heterogeneity of gene loci and incomplete dominance, the molecular mechanism of cross-beak formation, especially the degree of cross, is not yet clear. Thus, we screen key genes and reveal the possible phenotypic formation mechanism of cross-beak by comparison with different degrees of deformity in Huiyang Bearded chickens by compare whole-genome resequencing-based variant analysis. Comparative analysis between cross-beak and normal-beaked chickens identified differential variants in several candidate genes, including CDH11, CTNNAL1, NRXN3, NRXN1, CDH5, SDC3, and DHFR. Genes harboring these variants were enriched in pathways related to cell adhesion molecules and metabolic processes, with functional annotations involving cell-cell adhesion and neural crest cell migration. Comparative analysis between chickens with severe and slight cross-beak deformities identified additional candidate genes, including MRPL21, NSUN2, DDX55, GNB3, and NFKB2. These genes were associated with enriched terms and pathways related to focal adhesion, amyotrophic lateral sclerosis, steroid 7 -hydroxylase activity, and skin-barrier establishment. These findings provide a preliminary catalogue of genetic variants and candidate genes for future functional studies of cross-beak development and severity in chickens.

genomics↗

bioRxiv2026

Genome-wide screen for genes required for smooth lipopolysaccharide production in Escherichia coli K-12

Lipopolysaccharide (LPS) is a major component of the outer membrane of Gram-negative bacteria, contributing to membrane integrity and environmental interactions. Genome-wide studies defining bacterial gene functions have been extensively performed in the model strain Escherichia coli K-12 which lacks O-antigen (OAg), and therefore does not produce smooth LPS (S-LPS). Consequently, the genetic requirements for S-LPS production in this model system remain incompletely defined. Here, a functional wbbL gene was introduced into the E. coli K-12 KEIO single-gene deletion mutant library to restore OAg synthesis, enabling genome-wide analysis of S-LPS production by screening with colicin E2 (ColE2) and validated with LPS silver staining. This identified 319 mutants with increased sensitivity to ColE2 in the presence of OAg, suggesting broader envelope-associated effects during screening. In addition, 27 mutants showed defects in S-LPS production, corresponding to genes involved in OAg biosynthesis, LPS core and sugar precursor synthesis, OAg ligation and regulation, and enterobacterial common antigen biosynthesis. A further 18 mutants initially appeared defective in S-LPS production but could not be validated upon reconstruction, and whole-genome sequencing revealed secondary mutations responsible for the observed phenotypes. This study provides a validated genetic framework for S-LPS production in E. coli K-12 and highlights the importance of rigorous validation in genome-wide screening approaches.

microbiology↗

bioRxiv2026

Synthesizing Mechanistic Hypotheses from Single-Cell Omics via Discretized Feature Attribution and Empirical Language Model Grounding

Single-cell multimodal omics offer unprecedented resolution of cellular networks, yet translating continuous computational attributions into structured, testable biological mechanisms remains a persistent bottleneck. To address this limitation, we introduce an analytical pipeline employing decision trees to discretize continuous neural network attributions into explicit regulatory thresholds. These boundaries then structurally constrain large language models, enabling them to integrate established literature with empirical data to synthesize context-specific hypotheses. Applying this continuous-to-discrete framework across sparse datasets yielded novel biological mechanisms. Specifically, the framework articulated a cytoskeletal gating hierarchy governing EGF-stimulated pathways, identified transcriptomic drivers of input resistance in cortical interneurons, and delineated translational logic predicting Ki-67 abundance within spatial transcriptomics. Retrospective benchmarking validated the frameworks capacity to autonomously reconstruct published regulatory logic. Supported by a locally deployable open-weight language model and a code-free interface, this approach establishes an auditable methodology to extract robust experimental hypotheses from high-dimensional single-cell data.

systems biology↗

bioRxiv2026

Agentic AI for Structural Elucidation and Discovery of Drug Metabolites from Mass Spectrometry Data

The majority of chemical signals detected in public metabolomics repositories remain structurally undefined. Large language models (LLMs) are probabilistic systems whose capacity to generate outputs beyond their training data, which can cause hallucinations, makes them also potentially suited to hypothesize structures for molecules that have never been described. We aimed to build a system that could harness this LLM generative capacity combined with domain specific tools/framework to constrain hallucination and produce validated discoveries. We developed a GNPS2 agentic AI system that interprets LC-MS/MS data by integrating spectral alignment, molecular formula inference, rule-based structural enumeration, machine learning-based spectrum prediction, and translates natural language hypotheses from domain experts into dynamically generated analytical workflows. We demonstrate the annotation of unknown drug metabolites from public data guided by chemical hypotheses. The agent predicted, and we experimentally confirmed, a phosphorylated hydroxyzine, an acetaminophen-p-coumaric acid ester, and identified two new oxidative ibuprofen-carnitine conjugates from public repositories. These results demonstrate that LLM-driven agentic reasoning, when combined with domain expertise, can indeed generate experimentally testable structural hypotheses for previously uncharacterized metabolites leveraging pan repository data.

bioinformatics↗

bioRxiv2026

Translation sustains productive Pol II elongation through maintenance of nuclear RNA surveillance

Although transcription and translation are spatially separated in eukaryotic cells, gene expression requires coordination between the nucleus and cytoplasm. Whether translation rapidly influences nuclear transcription remains unclear. Here we show that ongoing translation sustains productive RNA polymerase II (Pol II) elongation in mouse embryonic stem cells. Translation inhibition reduces nascent transcription within 15 minutes, impairs Pol II pause release, and preferentially represses long genes before substantial loss of chromatin-associated Pol II. Unexpectedly, despite reduced transcription, nuclear RNA transiently accumulates owing impaired in RNA turnover. Mechanistically, translational arrest redistributes RNA-processing and RNA-surveillance factors from the nucleus to stalled ribosome-associated complexes in the cytoplasm, reducing their nuclear availability. Acute depletion of RNA-surveillance components phenocopies elongation defects caused by translational inhibition, particularly in long genes. Together, our findings identify ongoing translation as a regulator of nuclear RNA surveillance capacity and reveal a feedback mechanism sustaining productive Pol II elongation and nuclear RNA homeostasis.

Molecular Biology↗

bioRxiv2026

Variation in infant subcortical brain development from 6 to 12 months in Down syndrome

IntroductionDown syndrome (DS), arising from Trisomy 21, is the most common genetic condition associated with intellectual disability. While smaller total brain volumes have been consistently observed in DS, no longitudinal neuroimaging studies have examined volumetric brain development in DS during infancy, a period of rapid neural growth when interventions may have the greatest impact. MethodHigh-resolution T1- and T2-weighted images were acquired during natural sleep in a multisite longitudinal cohort of 44 infants with DS and 39 control infants without DS at ages 6 and 12 months. Neuroimaging data were harmonized to reduce batch effects, and a novel deep-learning, repeated-measures segmentation approach was applied to optimize neuroanatomical segmentations. Total intracranial volume (ICV) and bilateral absolute subcortical volumes (amygdala, caudate, hippocampus, pallidum, putamen, thalamus) were first directly compared in infants with and without DS at 6 and 12 months. Hierarchical linear modeling (HLM) evaluated longitudinal group differences for each structure, accounting for sex, gestational age, and laterality. Subcortical group differences estimated by HLM were also compared to group differences in total ICV. ResultsICV in infants with DS was lower than controls at 6 months (12.6%; p<.001) and 12 months (16.3%; p<.001). Subcortical structures displayed a range of lower volumes (6.9%-13.1%; ps[≤].003) in infants with DS, although the caudate and putamen were exceptions. Caudate volumes were on average lower in DS but not significantly different from controls, while putamen volumes were on average higher in DS but not significantly different from controls, except for the right putamen, which was significantly larger (5.3%; p=.018) at 6 months. In HLM, ICV and all subcortical structures showed slower growth in DS from 6 to 12 months, except for the amygdala and putamen, which displayed similar growth rates to controls. DS-associated reductions in subcortical volumes were similar in magnitude to ICV, although 12-month caudate and 6- and 12-month putamen volumes were enlarged relative to ICV. ConclusionInfants with DS exhibited substantially reduced ICV and widespread reductions in subcortical volumes and growth from 6-12 months. Across a range of volumetric differences, findings were most distinct in the basal ganglia, for which volume reductions were attenuated in the caudate, while the putamen was uniquely enlarged with comparable growth to controls. These observations support early regional specificity in the neural impact of Trisomy 21 and underscore the utility of infant neuroimaging to inform biologically based interventions and clinical trial readiness in DS.

neuroscience↗

bioRxiv2026

Architecture of a portal complex embedded in the poxvirus core

Gene transcription within the viral core is a unique feature of poxvirus replication. Following entry into the cytoplasm, the poxvirus core dissociates from the lateral bodies and undergoes expansion, functioning as a compartment for early transcription. However, the mechanisms governing molecular exchange between the viral core and the host cytoplasm remain poorly understood. Here, we determine the structures of the portal complex and its pore on the poxvirus core at 7.1 [A] and 4.9 [A] resolution, respectively, using cryo-electron tomography and sub-tomogram averaging. The pore is assembled from three viral proteins, E8, E6, and L3, for which we constructed an atomic model. Structural and channel analyses reveal that the pore satisfies the geometric and electrostatic requirements for the transport of RNA and smaller negatively charged molecules, while excluding double-stranded DNA and cytosolic DNA sensors. Together, our findings establish a structural framework for understanding the assembly and function of the poxvirus portal complex and identify potential targets for antiviral intervention.

biophysics↗

bioRxiv2026

Geometry-enhanced protein language modeling enables discovery of novel antibiotic resistance genes

The global antibiotic resistome remains largely unexplored, not because antibiotic resistance genes (ARGs) are rare in the environment, but because many are evolutionarily distant from known ARGs. Current computational approaches primarily rely on sequence homology, and thus miss distant homologues. We develop GeoARG, a geometry-enhanced framework that integrates structural features with protein language models through knowledge distillation, enabling efficient large-scale screening using sequence input alone. Across multiple benchmarks, GeoARG substantially improves the detection of remotely homologous ARGs, particularly under low sequence identity and fragmented conditions. Large-scale metagenomic analysis uncovers 1,485 high-confidence ARG candidates that are highly divergent from known ARGs, expanding the phylogenetic and functional landscape of the resistome. Structural analyses further show that these candidates preserve active-site geometry and maintain stable ligand-binding configurations consistent with known resistance mechanisms. These results demonstrate that geometric constraints enable systematic expansion of the resistome and facilitate the discovery of evolutionarily distant yet functionally conserved genes. A public web server is available at https://ycclab.cuhk.edu.cn/GeoARG/.

bioinformatics↗

bioRxiv2026

Long-term Production and Recovery of Medium-Chain Carboxylates from Source-Separated Organics

Source-separated organics (SSO) are widely processed via anaerobic digestion to produce biogas, yet alternative conversion pathways could generate higher-value products. Here, we demonstrate long-term continuous production and recovery of medium-chain carboxylic acids (MCCAs) from SSO via microbial chain elongation using a bench-scale anaerobic bioreactor operated for 911 days. The bioreactor was fed with SSO samples collected from two full-scale municipal organics processing facilities in Toronto, Canada, capturing facility-specific and seasonal variability in SSO composition. MCCA production was closely associated with the availability of lactate as an electron donor, which varied with SSO preprocessing operations and outdoor collection temperatures. To mitigate product inhibition, an in-line extraction system using hollow-fiber polydimethylsiloxane (PDMS, also known as silicone polymer) membranes was integrated with the anaerobic membrane bioreactor, providing a robust and solvent-free alternative to solvent-based extraction methods. Selective extraction reduced MCCA product inhibition when lactate availability was not limiting, resulting in maximum MCCA yields and production rates of 0.31 g MCCA g-1 VSfeed and 0.84 g L-1 day-1, respectively, with notable octanoic acid production (up to 20% of total MCCA). Acidification of the alkaline extract produced a phase-separated MCCA-rich oil (0.88 g MCCA g-1 oil) without the addition of downstream separation steps. Microbial community analysis of the bioreactor revealed enrichment of putative chain-elongating bacteria, including Eubacterium and Pseudoramibacter species, while shifts in SSO feedstock microbiomes influenced substrate availability and product spectra. These results demonstrate the feasibility of sustained MCCA production from municipal organic waste streams and highlight opportunities to integrate chain elongation with existing anaerobic digestion infrastructure.

bioengineering↗

bioRxiv2026

Characterization of aliA and aliB deletion mutants reveals a dominant role of AliA in Haloferax volcanii lipoprotein lipidation

Protein lipidation is a widespread strategy for anchoring proteins to cellular membranes across all domains of life, yet the mechanisms underlying this process in archaea remain poorly understood. Recently, the first archaeal enzymes involved in lipobox-containing protein (lipoprotein) biogenesis, AliA and AliB, were identified and characterized in the model archaeon Haloferax volcanii. Although these paralogs share significant sequence similarity, distinct deletion phenotypes suggest differences in their substrate specificity and function. Here, we employed large-scale Triton X-114 fractionation followed by quantitative proteomics and lipid-specific mass spectrometry to systematically analyze AliA- and AliB-dependent lipoprotein lipidation. Deletion of aliA affected substantially more lipoproteins in Hfx. volcanii than deletion of aliB, markedly diminishing their TX-114 enrichment--indicating reduced hydrophobicity--and abolishing thioether-linked archaeol modification. This establishes AliA as the primary enzyme responsible for archaeal lipoprotein lipidation. In contrast, deletion of aliB affected only a small subset of lipoproteins and did not significantly reduce thioether-linked archaeol levels. In addition to defining distinct and non-redundant roles for AliA and AliB, this study provides the first large-scale experimental validation of predicted archaeal lipoproteins and identifies candidate components of the archaeal lipoprotein biogenesis pathway, substantially advancing mechanistic understanding and enabling improved lipoprotein prediction in this previously underexplored field.

molecular biology↗

bioRxiv2026

Reversible RNA Acylating Reagents with Nitro Reduction Strategy

We developed a series of nitro reduction-reversible acylating reagents. Following optimization of the acylation conditions, these reagents were tested for deacylation with sodium dithionite in vitro. We applied this reversible acylation to modulate RNAzyme-mediated pre-tRNA maturation, demonstrating its ability to regulate RNA-RNA interactions. Furthermore, the in vitro reversible acylation of EGFP mRNA indicated effective control of its translational activity. To explore cellular applications, we validated NQO1-mediated deacylation in vitro and then induced hypoxia in HepG2 cells using cobalt chloride, thereby reactivating the function of acylated EGFP mRNA via endogenous NQO1. Overall, this study highlights the potential for developing nitro reduction-reversible acylation as a new strategy for RNA functional control and RNA-based drug modification.

biochemistry↗

bioRxiv2026

Mechanistic Language Modeling and Oxygenated 3D Screening Reveal Berberine and Enzalutamide Synergy in Resistant Prostate Cancer

Resistance to androgen receptor inhibitors remains a primary challenge in prostate cancer treatment, yet identifying synergis-tic co-therapies is hindered by immense combinatorial search spaces and the limited interpretability of predictive computation models. Here, we developed an integrated discovery-validation axis coupling knowledge-augmented large language models with oxygen-supplemented 3D spheroid assays. By leveraging inherent model stochasticity, our framework measures the degree of consensus across independent predictions to establish a formal metric for predictive accuracy. This principle enables high-throughput assessment of complex signaling crosstalk, yielding mechanistic rationales for all predictions and defining a high-confidence zone that minimizes experimental attrition. Utilizing this approach to screen 3,592 natural products, we identified a previously unrecognized synergy between berberine and enzalutamide that re-sensitizes resistant cells. Validation confirms that berberine perturbs the PI3K/AKT/mTOR and AMPK axes, a finding consistent with the mechanistic rationales computationally derived by the framework. Integrating interpretable AI with physiologically relevant 3D screening provides a scalable methodology for the rational discovery of synergistic therapies. SignificanceIntegrating mechanistic AI with oxygenated 3D screening, we identify a novel berberine-enzalutamide synergy. This framework resolves complex signaling dependencies, providing a scalable, transparent methodology for the rational discovery of effective combination therapies.

bioengineering↗