Biology subjects

Hong, L.

Publications and source records attributed to Hong, L..

At least 19 recordsLinked to original sources

bioRxiv2026

VenusGenome: Genome-Aware Masked Autoencoding for Microbial Genome Representation Learning

Genomic-context models have shown that contextualizing protein embeddings can improve predictions of microbial function. We observe that a gene's protein embedding is partially predictable from the embeddings of neighboring genes in the genome, motivating high-ratio masking that requires reconstruction from sparser genomic context. Genomic modeling must also distinguish biological relationships from assembly conventions, since the ordering of separate contigs does not imply adjacency and strand labels depend on the reference orientation. We therefore introduce VenusGenome, a gene-resolution masked autoencoder that combines visible-only encoding with genome-aware attention incorporating contig boundaries, relative strand orientations, and gene positions. We pretrain this model on a curated corpus of over 1.7 million microbial genome assemblies from isolate and metagenomic sources, using frozen protein embeddings as gene-level inputs. Across genome-level phenotype prediction and gene-level tasks, the resulting frozen representations achieve the highest mean performance among the evaluated models on nine of eleven primary metrics. An exploratory gene-input perturbation analysis further links representation sensitivity to experimentally measured fitness defects. Ablations support the contribution of genome-aware attention and show the benefits of masked autoencoding over BERT-style masked modeling, with further gains from higher masking ratios.

bioinformatics↗

bioRxiv2026

PepRePs: Peptide-Retargeted Phosphatases Designed with Generative Language Models

Phosphorylation is a fast, reversible layer of post-translational control that directly governs protein activity, signaling flux, and cell state. Kinases and phosphatases collectively establish these phosphorylation states, but their dysregulation drives cancer, neurodegenerative diseases, and developmental disorders. Selectively manipulating the phosphorylation of individual proteins in cells remains challenging, particularly for disordered substrates. Here, we introduce Peptide-Retargeted Phosphatases (PepRePs), a genetically encoded system that fuses a peptide binder to a serine/threonine phosphatase subunit to localize phosphatase activity to a target protein. We first establish the PepReP architecture using characterized tau-binding peptides, then apply protein language model-guided peptide design to target Rab8A, PTEN, and STAT3. Across endogenous and exogenous targets, PepRePs reduce phosphorylation, with downstream effects on tau aggregation, Rab8A-associated protein interactions, and cellular proliferation following PTEN and STAT3 targeting. Sequence-based peptide design enables targeting without structural inputs, supporting application to substrates with disordered regions that challenge conventional structure-based binder design. By coupling programmable peptide binders with modular phosphatase subunits, PepRePs enable targeted control of protein phosphorylation to probe cellular signaling and develop strategies for correcting disease-associated protein dysfunction.

synthetic biology↗

bioRxiv2026

ACAN regulates VSMC mitochondrial homeostasis through the YAP1/TAZ pathway and contributes to the onset and progression of aortic dissection

Objective: Aortic dissection (AD) is a serious, life-threatening cardiovascular crisis. While vascular proteoglycans normally buffer hemodynamic stress, their aberrant accumulation in the AD aorta predisposes to rupture. Aggrecan (ACAN) is among the most frequently elevated proteoglycans in this setting, yet its pathogenic relationship to AD remains undefined. This study aimed to investigate the role of ACAN in AD and identify potential therapeutic targets. Approach and Results: We employed human tissues, animal models, and cell cultures. Western blotting and immunohistochemistry (IHC) were performed on aortic samples from AD patients and controls to assess contractile markers, MMPs, ACAN, and SOX9. In vivo, AD was induced in mice using {beta}-aminoisobutyric acid (BAPN) and Ang II, with AAV-mediated ACAN knockdown in VSMCs. Histological staining confirmed model validity and enabled assessment of AD incidence, mortality, and protein expression changes. In vitro, human aortic vascular smooth muscle cells (HAVSMCs) were stimulated with Ang II, and SOX9 and ACAN were silenced via siRNA to evaluate the functional impact of ACAN downregulation. In human AD aortas, elevated ACAN and its upstream transcription factor SOX9 drove VSMC phenotypic switching and ECM degradation. In mice, ACAN knockdown corrected AD-induced medial structural disruption, elastin fragmentation, and collagen deposition, markedly reducing incidence and mortality while curbing excessive mitochondrial fission and rescuing functional integrity. ACAN silencing attenuated Ang II-induced mitochondrial damage in HAVSMCs via the RNA-seq-identified YAP1/TAZ pathway, and Verteporfin restored homeostasis. Conclusion: ACAN is aberrantly expressed in AD and murine aortas. ACAN downregulation delays disease progression by preserving mitochondrial homeostasis and function via the YAP1/TAZ pathway.

pharmacology and toxicology↗

bioRxiv2026

DLL3 expression predicts response and long-term outcome in small cell lung cancer patients treated with tarlatamab

Background: Small cell lung cancer (SCLC) is an aggressive, high-grade neuroendocrine carcinoma (hgNEC) with poor prognosis. Tarlatamab, a DLL3-targeting T cell engager was approved in 2024 for relapsed SCLC, with almost doubled overall survival benefit compared to chemotherapy. However, over half of patients do not respond to tarlatamab and others develop resistance within months. With multiple DLL3-targeting therapies in clinical trials, there are currently no validated predictive biomarkers to identify patients most likely to benefit from tarlatamab. In this study, we evaluate baseline DLL3 IHC level in patients as a predictive biomarker to tarlatamab clinical outcome. Methods: We assembled a cohort of 138 patients within IRB-approved MD Anderson GEMINI database (PA13-0589) with DLL3 expression by CLIA-validated immunohistochemistry (IHC), and those treated with tarlatamab from 7/1/2024 to 3/30/2026. Only those with DLL3 IHC results and treated with tarlatamab were included in this predictive biomarker study. All DLL3 levels were reported as percentage cells with cytoplasmic or surface DLL3 expression, including 24 with additional intensity score and H-score calculation. demographic, clinical, and outcome data were collected. Systemic and intracranial response data were assessed by RECIST and mRANO BM criteria respectively. Longitudinal liquid biopsies were taken from a subset of patients for circulating tumor DNA (ctDNA) and (Cytometry by Time-of-Flight) CyTOF analyses. Results: In the predictive biomarker cohort of 54 patients with SCLC treated with tarlatamab, Median DLL3% level in tarlatamab treated cohort was 75%. Median follow-up time was 9.9 months, median time on tarlatamab treatment (ToT) was 5.8 months. At the data cut-off, 61% (33 out of 54) patients were alive. Kaplan-Meier curve stratified by DLL3 % of <=75% vs. >75 showed significant longer ToT and a trend toward longer median OS (mOS). In DLL3>75% compared to DLL3 <=75%: mToT was 10.1 months vs. 2.9 months (p=0.036), mOS: not reached vs. 9.5 months (p= 0.066). Among 41 patients who completed one cycle of tarlatamab treatment and were eligible for systemic (extra-cranial) response per RECIST criteria; we observed partial response (PR) in 51% (21) patients, stable disease (SD) in 22% (9) patients; and progressive disease (PD) in 27% (11) patients. The median DLL3 % in those with PR vs. PD were 80% vs. 40% (p=0.006); median DLL3% comparing PR and SD were 80% vs. 60% (p=0.051). Among 35 patients evaluable for intracranial response, best objective response prior to any brain radiation were 5 (14%) CR,11(31%) PR, 8 (23%) SD and 11 (31%) PD. 9 received brain radiation (8 SRS, 1 WBRT) during tarlatamab to achieve better intracranial control while systemic control is maintained. DLL3% does not appear to be correlated with intracranial response. Longitudinal ctDNA and CyTOF analyses showed persistent ctDNA positivity, as well as increasing NEUROD1 subtype cells in circulating tumor cells in post-tarlatamab progression samples. We additionally profiled DLL3 expression by IHC in 138 patients with SCLC and found heterogenous DLL3 expression, with Median DLL3% of 70% across different biopsy sites. Conclusions: Our data suggests DLL3 expression in patients with SCLC is heterogeneous and not ubiquitous, with median DLL3 IHC of 70%. DLL3 IHC predicts objective response and long-term outcome with tarlatamab. In addition, dynamic monitoring of ctDNA level and CTCs by CyTOF are suggestive of development of tarlatamab resistance, however further studies with larger cohort of paired longitudinal samples are needed to validate the novel blood-based biomarkers. To our knowledge, this study is the first to validate DLL3 IHC as a predictive biomarker for tarlatamab in SCLC, which leads the way for optimizing treatment selection and combinatorial therapies for the subset of patients less likely to respond to tarlatamab.

cancer biology↗

bioRxiv2026

A Mammalian High-Throughput Screen for AI-Designed Peptide-Guided Protein Degraders

Targeted protein degradation (TPD) offers a route to eliminate disease-driving proteins that remain inaccessible to conventional inhibitors. However, degrader discovery remains low-throughput, labor-intensive, and dependent on randomized libraries or non-human display systems, limiting functional selection in mammalian cells. Here, we present a high-throughput, human cell-based platform for screening peptide-guided ubiquibodies (uAbs). These genetically encodable, doxycycline-inducible degraders fuse peptide guides generated by protein language models to the CHIP{Delta}TPR E3 ligase domain, creating a modular, CRISPR-like system for programmable TPD. For each target, we introduce a pooled uAb library into the corresponding fluorescent reporter cell line, isolate cells with reduced target abundance by FACS, and recover enriched peptide guides by sequencing. For {beta}-catenin, enriched uAbs reduced endogenous {beta}-catenin abundance and Wnt signaling in DLD1 cells. GFAP-directed uAbs reduced endogenous GFAP abundance and cell viability in U251 glioblastoma cells, while EWS::FLI1-directed uAbs reduced fusion oncoprotein abundance, suppressed EWSAT1 expression, and increased apoptosis in Ewing sarcoma models. Finally, a screen using endogenously tagged GATA2 further identified uAbs that reduced GATA2 under native genomic regulation. Overall, our platform connects generative peptide design to functional mammalian selection and establishes a scalable strategy for CRISPR-like proteome perturbation.

synthetic biology↗

bioRxiv2026

MAXWELL: Calibrating the probabilistic outputs of protein language models to the mutation-induced stability change landscape

Designing mutations that enhance protein stability is a central goal in protein engineering. However, experimentally screening large numbers of candidate mutations is costly and time-consuming, creating a strong need for computational methods that can identify potentially stabilizing mutations. Among these approaches, protein language models are particularly promising because they learn context-dependent amino acid preferences from large-scale sequence and structure datasets. Nevertheless, most existing stability prediction methods use these models primarily as feature extractors and do not fully exploit the amino acid probability distributions they encode. Here, we introduce MAXWELL (Matrix-wise Landscape Learning), a novel post-training method that calibrates the probabilistic outputs learned by protein language models during pretraining to generate mutational landscapes that quantify the effects of individual amino acid substitutions on protein stability. When applied to ProteinMPNN, MAXWELL yields a state-of-the-art predictor of the effects of protein mutations on stability, outperforming ThermoMPNN and other representative methods on a curated benchmark of experimentally measured stability changes. We next applied MAXWELL to the design of ten single-point mutations in the DhaA dehalogenase, seven of which (70%) increased thermal stability. Among them, G171W showed the largest improvement, with a measured {Delta}Tm of 4.91 {degrees}C. These experimental results establish MAXWELL as a novel post-training strategy for protein language models and a practical framework for designing stabilizing mutations. Repositoryhttps://github.com/ai4protein/Venus-MAXWELL

bioinformatics↗

bioRxiv2026

HER2 mutation-derived neoantigens in NSCLC as actionable targets for TCR therapy

HER2 mutations are oncogenic drivers in 1-6% of non-small cell lung cancers (NSCLC), but therapeutic resistance limits the durability of current HER2-targeted treatments. Here, we identify T-cell receptors (TCRs) targeting recurrent HER2 hotspot mutations as a potential immunotherapeutic strategy for HER2-mutant NSCLC. Using neoepitope prediction and antigen-specific T-cell enrichment, we isolated HLA-A*02:01restricted TCRs recognizing HER2 A775insYVMA, S310F, and G776delinsVC mutations, collectively covering approximately 60% of HER2-mutant NSCLC. These TCRs selectively recognized mutant HER2 epitopes without detectable wild-type reactivity and some displayed cross-recognition of related hotspot variants, expanding the spectrum of targetable tumors. The G776delinsVC-specific TCR also exhibited co-receptorindependent activity showcased by its ability to activate CD4+ T cells. Importantly, timelapse single-cell flow cytometry analyses demonstrated that TCR-engineered T cells repeatedly reacquired activated polyfunctional states following serial antigen stimulation, while serial tumor rechallenge assays confirmed sustained cytotoxic activity across multiple rounds of tumor killing. These findings identify recurrent HER2 mutations as shared immunotherapeutic targets and provide a foundation for the development of TCR-based therapies for HER2-mutant NSCLC.

immunology↗

bioRxiv2026

Deep learning-guided design of hydrolases for crystalline PET depolymerization

Poly(ethylene terephthalate) (PET), a ubiquitous polyester used in packaging and textiles, persists in the environment due to its high crystallinity and stability, contributing substantially to global plastic pollution. Enzymatic depolymerization by PET hydrolase (PETase) offers a chemically precise and environmentally sustainable route to convert PET into its monomeric building blocks, enabling recycling. However, practical implementation remains hindered by the rigid, crystalline architecture of PET, which restricts enzyme access and necessitates energy-intensive pretreatment to enable efficient depolymerization. In addition, most PETases achieve only partial depolymerization with low terephthalic acid (TPA) yields and accumulate inhibitory intermediates, while limited thermostability and slow surface kinetics further restrict efficiency. To overcome these barriers, we report VenusPETase, an engineered variant of KbPETase designed using PET-Flow, a state-of-the-art computational framework for PETase engineering. Compared with KbPETase, VenusPETase exhibits a 3.4-fold increase in hydrolytic activity, up to a 27-fold improvement after heat treatment of protein, and a 12 {degrees}C enhancement in thermostability. VenusPETase exhibits rapid degradation across a wide crystallinity range (8%-50%) at 50 {degrees}C, effectively spanning the entire spectrum of commercial PET products. Moreover, VenusPETase outperformed nine high-performance PETases under their respective optimal conditions and degraded untreated PET substrates across 8%-50% crystallinity, producing TPA as over 95% of the released products with minimal intermediate accumulation. X-ray crystallography and molecular dynamics simulations suggest that dynamic modulation, elevated surface electrostatic potential enhance the interaction of VenusPETase with crystalline PET, thereby lowering the hydrolytic energy barrier and improving catalytic performance. We also demonstrate that untreated, postconsumer-PET from nine different products can all be degraded by VenusPETase. The recovered monomers can be directly repolymerized into virgin-quality PET, demonstrating a closed-loop enzymatic recycling process. In a 100 L bioreactor, VenusPETase completely depolymerizes post-consumer crystalline PET (28% crystallinity) within 24 h under 50 {degrees}C. These results establish VenusPETase as a robust biocatalyst that enables efficient, closed-loop recycling of crystalline PET under mild conditions.

biochemistry↗

bioRxiv2026

Phenotypic Bioactivity Prediction as Open-set Biological Assay Querying

The traditional drug discovery pipeline is severely bottlenecked by the need to design and execute bespoke biological assays for every new target and compound--a process that is both time-consuming and prohibitively expensive. While machine learning has accelerated virtual screening, current models remain confined to "closed-set" paradigms, unable to generalize to entirely novel biological assays without target-specific experimental data. Here, we present OpenPheno, a groundbreaking multimodal foundation model that fundamentally redefines bioactivity prediction as an open-set, visual-language question-answering (QA) task. By integrating chemical structures (SMILES), universal phenotypic profiles (Cell Painting images), and natural language descriptions of biological assays, OpenPheno unlocks the highly coveted "profile once, predict many" paradigm. Instead of conducting countless target-specific wet-lab experiments, researchers only need to capture a single, low-cost Cell Painting image of a novel compound. OpenPheno then evaluates this universal phenotypic "fingerprint" against the text-based description of any unseen assay, predicting bioactivity in a zero-shot manner. On 54 entirely unseen assays, it achieves strong zero-shot performance (mean AUROC 0.75), exceeding supervised baselines trained with full labeled data, and few-shot adaptation further improves predictions. In the most stringent setting where both compounds and assays are novel, OpenPheno maintains robust generalization (mean AUROC 0.66), opening up a new paradigm for a highly scalable, cost-effective, and universal engine for next-generation drug discovery.

bioinformatics↗

bioRxiv2026

Spatial multi-omics identify an immunosuppressive lipid-laden macrophage niche in primary CNS lymphoma

Primary central nervous system lymphoma (PCNSL) is histologically a subtype of diffuse large B-cell lymphoma (DLBCL), sharing genetic and transcriptomic similarity, but with distinct clinical features, particularly its confinement to the CNS and higher relapse risk. The microenvironmental basis for its divergence from systemic DLBCL remains unclear. Using spatial transcriptomic approaches (Xenium/GeoMx digital spatial profiling) in a comparative study of PCNSL (n=17) and DLBCL (n=76), we found that PCNSL, unlike systemic DLBCL, is dominated by an immunosuppressive macrophage compartment enriched for cholesterol-metabolism programs. In independent cohorts of PCNSL profiled by single-cell RNA sequencing, we validated the presence of a recurrent population of lipid-laden macrophages (LLMs): TREM2/GPNMB-expressing, lipid-remodeled cells transcriptionally distinct from resident microglia and consistent with an infiltrating monocyte origin, not previously characterized in CNS lymphoma. LLMs formed immunosuppressive niches with regulatory T cells, and using Cellscape hyperplex proteomic imaging we demonstrate that LLM-Treg spatial interactions are associated with chemotherapy response. To test whether LLMs are lymphoma-driven and functionally important, we developed an immunocompetent syngeneic PCNSL mouse model, driven by Myd88L252P and Cd79b mutations with Bcl2 overexpression. Monocyte-derived macrophages in lymphoma-bearing brain regions acquired an LLM-like state, not seen in lymphoma-free brain regions or in splenic tumors driven by the same oncogenic lesions. TREM2-SYK signaling sustained this state, and SYK inhibition reversed its tumor-supportive activity ex vivo. These findings identify the LLM program as a targetable immunosuppressive myeloid state in CNS lymphoma.

cancer biology↗

bioRxiv2026

Spatiotemporal Mapping of Tertiary Lymphoid Structure Heterogeneity Shapes Immune Niches and Clinical Outcomes in Intrahepatic Cholangiocarcinoma

Intrahepatic cholangiocarcinoma (iCCA) is a highly lethal malignancy with limited therapeutic options. The spatial architecture and functional diversity of tertiary lymphoid structures (TLSs) in iCCA remain unclear. Here, we present a multimodal spatial atlas of TLSs and identified Intra-tumoral TLSs (iTLSs) as independent prognostic markers. Bulk proteomic profiling of 214 discovery and 155 validation cases identified a four-tier TLS-based TME classification system and supported development of a TLS-predictive random forest classifier. Imaging mass cytometry revealed that iTLS tumors harbor structured immune architectures, where M1-like tissue-resident macrophages (RTMs), dendritic cells, and CXCL13 CD4 T cells co-localize to form antigen-presenting neighborhoods (apc-CNs) spatially coupled to TLS core regions (TLScore-CNs). Single-cell spatial transcriptomics further resolved 61 TLSs into 14 spatial niches and defined a pseudo-temporal maturation continuum: aggregated, activated, and post-activated. Intra-niche communication, primarily mediated by ifnCAFs, iCAFs, and CXCL12 macrophages, evolved dynamically with maturation. Single-nucleus RNA-sequencing combined with Tangram-based spatial mapping revealed CXCL12 macrophages and iCAFs forming a peripheral band in aggregated TLSs, while ifnCAFs infiltrated TLS interiors during activation. These findings define TLS heterogeneity and provide insights for stroma-directed immunotherapy. TeaserAn atlas of tertiary lymphoid structures reveals immune-stroma interactions in Intrahepatic cholangiocarcinoma.

immunology↗

bioRxiv2025

Multi-stem-loop constrained library enables accelerated SELEX for aptamers with superior kinetics and affinity

AbstractAptamers are single-stranded nucleic acids with ligand-binding capacity as cost-effective alternatives to antibodies. However, their utility is often undermined by the inherent conformational instability of single-stranded nucleic acids, which can compromise binding kinetics and final affinity. To overcome this critical limitation, we synthesized a novel primary library incorporating multiple inlaid stem-loop structures to enhance molecular stability and rigidity. This structural constraint dramatically accelerated the selection process, allowing us to isolate aptamer candidates with superior affinity in just three rounds of SELEX (Systematic Evolution of Ligands by EXponential enrichment), a significant reduction compared to conventional methodologies. Critically, the resulting aptamer demonstrated higher affinity and faster binding kinetics compared to previously reported sequences, which directly validate our hypothesis that structural stabilization leads to high-performance aptamers. This methodology, which integrated rational conformational constraint with a high-throughput selection process, offered a generalizable strategy to efficiently select aptamers with excellent kinetic and thermodynamic performance for demanding applications such as continuous, real-time ligand detection.

biochemistry↗

bioRxiv2025

Changes in endophytic microbial diversity and function in the root system affect the growth of continuously planted Casuarina equisetifolia

Withdrawal StatementThe authors have withdrawn this manuscript because additional experimental data are necessary to strengthen the conclusions and improve the overall quality of the study. Therefore, the authors do not wish this work to be cited as reference for the project. If you have any questions, please contact the corresponding author.

microbiology↗

bioRxiv2025

Endogenous Nitroalkene Exploits Dependence on Autophagy-Lysosome Pathway in PARPi-Resistant TNBC

Lack of DNA double-strand break repair efficiency exquisitely sensitizes cancers to poly-ADP ribose polymerase inhibitors (PARPi). Unfortunately, resistance to PARPi poses an insurmountable challenge for patients. Mechanisms that confer insensitivity to PARPi therapy include enhanced DNA damage repair and autophagy. Natural and non-natural unsaturated fatty acid nitroalkene derivatives (NFA) show anticancer actions that sensitize TNBC cells to PARPi and other DNA-damaging treatments. We reveal that nitro-oleic acid (OA-NO2) re-sensitizes PARPi-resistant TNBC cells to PARPi. RNA-seq analysis of clinically relevant mutBRCA1 PARPi-resistant TNBC cell lines exhibited upregulation in autophagy and lysosomal pathways. Bio-orthogonal analysis identified the autophagy regulator SQSTM1/p62 as a novel OA-NO2 target, alkylating two redox-sensitive Cys residues of p62 (Cys105 and Cys113). These Cys are essential for p62 regulation of autophagy and mimicked the effects of p62 Cys105 and Cys113Ala mutants and when alkylated by OA-NO2 showed impaired p62 oligomerization, degradation, and inhibition of autophagy. Combination treatment of PARPi-resistant TNBC with a PARPi and OA-NO2 synergistically inhibited p62-associated autophagy and lysosome function. These data emphasize the clinical potential of OA-NO2 for treating PARPi-resistant TNBC patients.

pharmacology and toxicology↗

bioRxiv2025

Potent Efficacy of Computer-Aided Designed Peptide Degrader Drug on PCSK9-Mediated Hypercholesterolemia

Proprotein convertase subtilisin/kexin type 9 (PCSK9) is a key regulator of low-density lipoprotein receptor (LDLR) degradation, which leads to increased levels of low-density lipoprotein cholesterol (LDL-C) and a heightened risk of hypercholesterolemia. Existing therapeutic strategies--such as monoclonal antibodies and gene-editing approaches--are limited by high production costs, potential safety concerns, and limited effectiveness in targeting intracellular PCSK9. Targeted protein degradation (TPD) represents a promising alternative capable of overcoming these limitations. In this study, we utilized computer-aided drug design (CADD) to develop a peptide-based degrader, Cadd4, aimed at selectively reducing PCSK9 levels. Through molecular docking and structural refinement, a peptide sequence with strong binding affinity to PCSK9 was identified. Functional evaluation was conducted both in vitro using human LX-2 cells and in vivo in high-fat diet (HFD)-induced hypercholesterolemic mouse models. Cellular uptake, PCSK9 degradation, and LDLR restoration were assessed via confocal microscopy, western blotting, and biochemical analysis. Safety was evaluated by monitoring liver enzyme levels and general physiological parameters. Cadd4 exhibited efficient intracellular delivery and selective PCSK9 degradation in LX-2 cell. In HFD-fed mice, hepatic PCSK9 expression was reduced by 38%, accompanied by a marked increase in LDLR levels. These changes translated into a 25% reduction in total plasma cholesterol and a 29% decrease in LDL-C concentrations. Biodistribution studies indicated liver-specific accumulation of Cadd4, consistent with its intended site of action. Experiments using human liver tissue further confirmed its ability to decrease PCSK9 and enhance LDLR levels. No signs of liver toxicity or adverse systemic effects were observed throughout the study. This work introduces Cadd4 as a CADD-designed peptide-based degrader capable of lowering plasma cholesterol by promoting LDLR recovery via PCSK9 degradation. Its intracellular mechanism of action addresses key gaps in current PCSK9-targeted therapies and supports the broader utility of TPD platforms in managing hypercholesterolemia. These findings offer a foundation for developing cost-effective, scalable therapies within the context of precision cardiovascular medicine.

pharmacology and toxicology↗

bioRxiv2025

Venus-MAXWELL: Efficient Learning of Protein-Mutation Stability Landscapes using Protein Language Models

In-silico prediction of protein mutant stability, measured by the difference in Gibbs free energy change ({Delta}{Delta}G), is fundamental for protein engineering. Current sequence-to-label methods typically employ the two-stage pipeline: (i) encoding mutant sequences using neural networks (e.g., transformers), followed by (ii) the {Delta}{Delta}G regression from the latent representations. Although these methods have demonstrated promising performance, their dependence on specialized neural network encoders significantly increases the complexity. Additionally, the requirement to individually compute latent representations for each mutant site negatively impacts computational efficiency and poses the risk of overfitting. This work proposes the Venus-MO_SCPLOWAXWELLC_SCPLOW framework, which reformulates mutation {Delta}{Delta}G prediction as a sequence-to-landscape task. In Venus-MO_SCPLOWAXWELLC_SCPLOW, mutations of a protein and their corresponding {Delta}{Delta}G values are organized into a landscape matrix, allowing our framework to learn the {Delta}{Delta}G landscape of a protein with a single forward and backward pass during training. Besides, to facilitate future works, we also curated a large-scale {Delta}{Delta}G dataset with strict controls on data leakage and redundancy to ensure robust evaluation. Venus-MO_SCPLOWAXWELLC_SCPLOW is compatible with multiple protein language models and enables these models for accurate and efficient {Delta}{Delta}G prediction. For example, when integrated with the ESM-IF, Venus-MO_SCPLOWAXWELLC_SCPLOW achieves higher accuracy than ThermoMPNN with 10x faster in inference speed (despite having 50x more parameters than ThermoMPNN). The training codes, model weights, and datasets are publicly available at https://github.com/ai4protein/Venus-MAXWELL.

bioinformatics↗

bioRxiv2025

ROS inhibits microtubule dynamics and cell growth heterogeneity during Arabidopsis sepal morphogenesis

Developing organs grow to reproducible sizes and shapes yet the growth of their constituent cells can be highly heterogeneous and fluctuating. During wild-type Arabidopsis thaliana sepal development, the fluctuations in cell growth average such that the sepals grow to uniform sizes and shapes. The uniform size and shape of the sepals allow the flower bud to stay closed and protected until the floral organs are mature. In contrast, cell growth averaging is reduced in the ftsh4-5 mutant, and the sepals develop to variable sizes and shapes. FTSH4 encodes a mitochondrial i-AAA protease localized to the mitochondria. Reactive oxygen species (ROS) accumulate in ftsh4-5 mutants, and lowering ROS levels rescues the sepal size and shape variability. Here, we investigate the effects of ROS on cell growth heterogeneity and cortical microtubule dynamics. We find that elevated ROS suppresses cell growth heterogeneity and averaging. We also find that elevated ROS causes cortical microtubules to become more crisscrossed, as well as more stable. The growth of the cells with crisscrossed microtubules changes less in time, which impairs cell growth averaging. However, depolymerizing microtubules is insufficient to restore normal growth fluctuations. Altogether, our results suggest that ROS affects microtubule dynamics and cell growth fluctuations which are necessary for robust morphogenesis.

developmental biology↗

bioRxiv2025

HISTONE DEACETYLASE COMPLEX 1 modulates sepal length through the ethylene-ROS module

Organ size is precisely regulated during development to ensure proper function. The sepal must maintain an appropriate size throughout its growth to protect the developing flower bud until blooming. Despite its significance, the mechanisms regulating sepal size during its maturation remain poorly understood. Here we identify the HISTONE DEACETYLASE COMPLEX 1 (HDC1) as a positive regulator of sepal size during maturation in Arabidopsis thaliana. hdc1 mutation prolongs sepal elongation due to delayed sepal maturation, thus hindering flowers from opening timely. Transcriptomic and proteomic analyses reveal that both plant hormone ethylene and reactive oxygen species (ROS) signaling are involved in the HDC1s modulation of sepal size during maturation. Further genetic and chemical data show that during sepal maturation, HDC1 upregulates ethylene levels to promote ROS accumulation, thereby regulating the cessation of sepal growth. Collectively, our study revealed that HDC1 integrates ethylene and ROS signaling to regulate sepal maturation, providing new insights into the molecular mechanisms that govern organ size during maturation in plants.

plant biology↗