Biology subjects

Xu, L.

Publications and source records attributed to Xu, L..

At least 37 records · Page 2Linked to original sources

bioRxiv2026

ApoE4 Promotes Thrombosis via Endothelial Cell ApoER2 and PP2A Activation

Background: APOE4, the variant of apolipoprotein E carried by 25% of individuals, is a common genetic risk factor for cardiovascular disease (CVD). Although ApoE classically participates in lipid transport, APOE4-associated risk goes beyond impact on circulating lipids. Life-threatening CVD events including myocardial infarction and stroke are driven by atherogenesis and thrombosis. In mice ApoE4 increases atherosclerosis severity, but whether other major drivers of CVD events are influenced by ApoE4 is unknown. Methods: GWAS data for venous thromboembolism (VTE) were analyzed. In humanized APOE3 (hE3) and APOE4 (hE4) mice, thrombosis was assessed by intravital microscopy (IVM) in the mesenteric microcirculation and by inferior vena cava (IVC) partial ligation. Actions of ApoE3 versus ApoE4 on endothelial cells (EC) and their underpinnings were studied in cultured human and mouse aortic EC, interrogating interactomes with immunoprecipitation-mass spectrometry and quantifying the secretion of Von Willebrand Factor (vWF), a critical initiator of thrombosis. Single cell transcriptomics datasets were queried do localize endothelial cell gene expression. Results: GWAS showed that APOE4 is associated with increased VTE risk, and whereas plasma lipids were similar, both microvascular and venous thrombosis were markedly increased in hE4 compared to hE3 mice. In cultured EC, whereas ApoE3 attenuated vWF secretion, it was enhanced by ApoE4, and both processes were mediated by ApoE receptor 2 (ApoER2). ApoE4, but not ApoE3, suppressed VEGF eNOS activation and NO production by causing the recruitment of the protein phosphatase 2A (PP2A) catalytic subunit to ApoER2 and the activation of PP2A. PP2A deletion prevented ApoE4-induced eNOS antagonism and vWF secretion by preserving Akt activation, and the NO donor spermine NONOate negated apoE4 stimulation of vWF secretion. PP2A activity was increased in hE4 aortas and IVC, and EC ApoER2 deletion or pharmacologic PP2A inhibition fully prevented exaggerated thrombosis in hE4 mice. In human great saphenous vein ApoER2 is primarily expressed in valvular endothelium. Conclusions: APOE4 is a risk allele for thrombosis, and ApoE4 is prothrombotic in microvasculature and veins in mice. Mechanistically, the ApoE4-EC ApoER2 tandem enhances vWF secretion by recruiting and activating PP2A and antagonizing eNOS, resulting in exaggerated thrombosis. In human veins ApoER2 is expressed in valvular endothelium, which is the most common site of initiation of venous thrombosis. Targeting these processes may afford protection from both primary thrombotic disorders like VTE and acute CVD events such as myocardial infarction and stroke in 25% of the population.

pathology↗

bioRxiv2026

Heteromeric LGP2-MDA5 receptor complexes drive antiviral signaling

The innate immune receptor MDA5 serves as a first line of defense against RNA viruses by recognizing viral double-stranded RNA (dsRNA) and initiating antiviral signaling. MDA5 is upregulated by another RLR family member known as LGP2, which has been proposed to enhance MDA5 activity by recruiting MDA5 to dsRNA. However, it has remained unclear whether LGP2 functions solely as a transient recruitment factor or actively participates in signaling-competent MDA5 filaments. Here, we demonstrate that LGP2 does more than recruiting MDA5 to dsRNA, and that LGP2 becomes an integral component of MDA5 filaments. Mechanistically, LGP2 stabilizes MDA5 filaments and reduces the number of MDA5 molecules required to nucleate stable signaling-competent filaments. This interplay between the two RLR proteins is mediated by specific dsRNA-dependent interactions between the C-terminal tail of LGP2 and MDA5, resulting in a network of contacts that is essential for LGP2-mediated enhancement of MDA5 signaling. Together, these findings provide new mechanistic insight into how LGP2 regulates the assembly and persistence of signaling-competent MDA5 filaments, establishing filament architecture and stability as key determinants of MDA5 signaling. Collectively, this work provides a mechanistic framework for understanding LGP2-mediated regulation of MDA5 signaling and advances our knowledge of the molecular mechanisms governing antiviral immunity.

biochemistry↗

bioRxiv2026

Inflammatory proteolysis generates pathogenic APOL1 fragments with distinct intracellular toxicities in podocytes derived from children with HIV associated nephropathy

APOL1 risk variants are the strongest genetic determinants of HIV-associated nephropathy (HIVAN), yet the mechanisms linking inflammation to APOL1-mediated podocyte injury remain poorly understood because authentic patient-derived human disease models are lacking. Using urine-derived podocytes established from children with HIVAN and endogenous APOL1 reporter cell lines derived from these cells, we identified a previously unrecognized pathway of inflammatory, cathepsin-dependent APOL1 proteolysis. Endogenous APOL1 cleavage was detected in patient-derived podocytes, whereas reporter cell lines enabled the identification and functional characterization of N-terminal and C-terminal APOL1 fragments with distinct intracellular localization and pathogenic functions. The nuclear N-terminal fragment activated inflammatory transcriptional programs and promoted podocyte injury, whereas the membrane-associated C-terminal fragment mediated membrane toxicity and remained susceptible to pharmacologic inhibition by inaxaplin. Cathepsin S directly cleaved APOL1 in vitro, linking inflammatory signaling to APOL1 fragmentation. These findings identify inflammatory APOL1 proteolysis as a mechanism that partitions APOL1 toxicity into distinct pathogenic programs and nominate APOL1 processing as a therapeutic target for HIV-associated and other APOL1-mediated kidney diseases.

cell biology↗

bioRxiv2026

Targeting Tumor-derived Sphingosine Kinase 2 Unleashes Antitumor Immunity and Improves Survival of Mice with Group 3 Medulloblastoma

Group 3 medulloblastomas (G3MB) carry the worst prognosis among medulloblastoma subtypes, yet molecularly targeted therapies remain elusive. Standard treatments cause severe long-term morbidity in survivors. Here, we identify tumor-derived sphingosine kinase 2 (SPHK2) as an essential driver of G3MB initiation and progression. SPHK2 exacerbates local immunosuppression by suppressing cytotoxic T-cell and NK-cell activity while promoting regulatory T-cell infiltration. Genetic or pharmacologic SPHK2 inhibition using Opaganib attenuates pro-survival tumor signaling and restores anti-tumor immunity, significantly improving survival in syngeneic G3MB mouse models. Combining Opaganib with fractionated low-dose radiation (f-LDRT) further enhances antigen presentation and reprograms tumor-associated myeloid cells toward an anti-tumor phenotype. This combination therapy markedly prolongs survival without inducing significant toxicity. Overall, our study establishes SPHK2 as a previously unrecognized therapeutic target and presents a safe, effective, microenvironment-reprogramming regimen for G3MB. One Sentence SummaryDirect inhibition of tumor-derived SPHK2 overcomes local immunosuppression and downregulates pro-survival signaling in Group 3 medulloblastoma, while combination with fractionated low-dose radiation further enhances anti-tumor immunity and significantly improves survival.

cancer biology↗

bioRxiv2026

Design of high-affinity miniprotein antagonist targeting IL-4Ra for IL-4/IL-13 signal blockade

IL-4Ra is a key regulatory receptor for type 2 inflammatory responses, signal transduce from IL-4 and IL-13 through binding with IL-13Ra or the gamma c chain to activate the downstream JAK1-STAT6 pathway. IL-4Ra is currently the most successful "golden target" in the field of allergic disease therapeutics. Its representative monoclonal antibody drug, dupilumab, through the dual blockade mechanism of IL-4/IL-13 has pioneered a new era of precision therapy for type 2 inflammation. In our manuscript, we employed large-scale deep learning-based computational design methods to de novo design mini-protein antagonists specific for both human and mouse IL-4Ra. The binding affinity was improved from 22.1 nM to 569 pM through partial diffusion. The design accuracy and binding specificity were verified through X-ray crystallography and biochemical studies. In vitro IL4/IL13 signal blockade assays revealed that de novo designed monomeric mini-protein antagonist exhibited comparable blockade ability to bivalent dupilumab. In vivo pharmacokinetic half-life studies demonstrated that fusion to an HSA-binding domain extended the half-life of the mini-protein antagonist from 2.7 hours to 60.6 hours. The IL-4Ra mini-protein antagonist had excellent expression levels, solubility and thermal stability. The IL4/IL13 signal blockade ability remained unchanged even after being heating to 95 degrees. In conclusion, through large-scale cluster computing and deep learning-based de novo design, we developed well-performed IL-4Ra mini-protein antagonist, and demonstrates certain potential for drug development.

synthetic biology↗

bioRxiv2026

Programmable Allosteric DNAzyme Coupled with CRISPR/Cas12a System for Multiplexed and Sensitive Detection of Extracellular Vesicle Derived MicroRNAs

Extracellular vesicle (EV)-derived microRNAs serve as important biomarkers for cancer diagnosis, yet their accurate detection remains limited by insufficient control of nucleic acid recognition and signal activation. Here, we identified a previously unrecognized feature of CRISPR/Cas12a, in which incorporation of ribonucleotides into single stranded DNA targets modulates Cas12a activation efficiency, revealing a hybrid DNA/RNA-dependent regulation of Cas12a activity. Leveraging this mechanism, we established a programmable detection strategy that enables sequence dependent tuning of Cas12a activation without the need for target amplification. By coupling DNAzyme mediated cleavage with Cas12a trans-cleavage, a cascade signal amplification system was established, enabling highly sensitive and selective detection of miRNAs. To facilitate clinical applications, an EV-based sample processing strategy was integrated to simplify isolation of EV associated miRNAs and allow direct miRNA detection without conventional RNA extraction. The resulting platform demonstrated robust discrimination of multiple miRNA targets in clinical cohorts and supported accurate classification of cancer subtypes according to expression signatures. By integrating machine learning analysis, the system accurate distinguished breast cancer (BC) patients from healthy donors (HD), as well as triple-negative breast cancer (TNBC) from BC. This study provides a mechanism-guided strategy for programmable CRISPR-based nucleic acid detection in complex biological samples.

bioengineering↗

bioRxiv2026

Coronary Artery Disease Transcriptomics Reveals Two Drivers of the Endothelial Cell SR-BI Expression and LDL Transport that Underlie Atherosclerosis

Atherosclerosis is initiated by circulating low-density lipoprotein (LDL) cholesterol transfer into the artery wall, which is mediated by scavenger receptor class B, type I (SR-BI) in endothelial cells(1). Employing single-cell RNA sequencing in human coronary artery disease (CAD) samples, here we show that endothelial SR-BI expression is increased in atheroma, and in endothelial cells with a transcript signature indicative of responding to disturbed blood flow. In vivo in mice hypercholesterolemia and disturbed blood flow independently upregulate endothelial SR-BI; the flow-related upregulation initiates endothelial cell LDL uptake and atherogenesis. Guided by transcription factor networks, it is revealed that HIF-1 binding to human Scarb1 Intron 1 governs endothelial SR-BI transcription, and in mice HIF-1 drives hypercholesterolemia-related SR-BI upregulation and artery LDL uptake. Thus, the two major instigators of atherosclerotic lesion formation, hypercholesterolemia and disturbed blood flow, both upregulate endothelial SR-BI to drive the LDL transport that underlies the disorder. Targeting the processes regulating endothelial SR-BI potentially represents a new therapeutic strategy against CAD.

cell biology↗

bioRxiv2026

Repurposing the TIGR-Tas system for programmable transcription activation

Programmable transcription activators are central to functional genomics, gene therapy, and synthetic biology. However, application of the most versatile CRISPR activators (CRISPRa) is hampered by the bulky size and compact CRISPR systems are constrained by the requirement of long protospacer adjacent motifs (PAMs). To address these limitations, we repurposed the TIGR-Tas RNA-guided DNA-targeting system as a versatile platform for gene activation. By systematically optimizing the fusion architecture of a nuclease-inactivated TasR with gene activation domains, we achieved robust, specific, and multiplexable transcriptional activation of endogenous human genes. This TIGR activation (TIGRa) system could be a superior alternative to CRISPRa systems due to its extreme compactness and potentially PAM-independent targeting capability.

genetics↗

bioRxiv2026

Multiplex genome engineering in yeast using the TIGR-Tas system

Tandem Interspaced Guide RNA (TIGR)-TIGR-associated (Tas) systems are a newly discovered family of ultracompact, modular RNA-guided DNA-targeting proteins that function without a protospacer adjacent motif (PAM) requirement. Their utility as genome engineering tools in microbes remains unexplored. Here, we report the first functional implementation of TIGR-Tas in Saccharomyces cerevisiae for genome engineering. We show that TasR from Parcubacteria (ParTasR) can be programmed by user-defined tigRNAs to generate targeted DNA double-strand breaks at yeast endogenous loci. By co-delivering ParTasR with customized tigRNAs and donor templates, we achieved precise gene fragment deletion and targeted codon substitutions at multiple genomic loci. The multiplex genome engineering capability of this TIGR-Tas system was demonstrated through high-efficiency multiplex gene disruption and chromosomal assembly of a lycopene biosynthesis pathway while inactivating an endogenous gene. This work establishes TIGR-Tas as a valuable addition to the yeast genome engineering toolbox, particularly for applications requiring PAM-independent targeting or compact delivery.

synthetic biology↗

bioRxiv2026

Genomic alterations enable BRCA1 methylation loss and promoter bypass to drive resistance in high-grade serous ovarian cancer

BRCA1 promoter methylation predicts sensitivity to PARP inhibitors in high-grade serous ovarian cancer, yet therapeutic resistance is common and mechanistically unresolved. Using long-read direct DNA sequencing of patient-derived xenografts and cell lines, we resolved BRCA1 methylation at single-molecule resolution with structural and transcriptomic analyses. We revealed two convergent PARP inhibitor and platinum resistance mechanisms, validated in patient tumors. First, focal, allele-specific loss of BRCA1 methylation arose through local cis-acting genomic alterations, instead of global epigenetic reprogramming. Engineered in cis sequence alterations near the methylated BRCA1 promoter were sufficient to induce methylation loss, restore homologous recombination, and confer resistance. Similar associations were observed across the genome, suggesting this mechanism extends beyond BRCA1. Second, BRCA1 expression was restored despite intact promoter methylation via structural variant-mediated promoter bypass or alternative transcription initiation. Together, these findings redefine BRCA1 methylation loss as a locus-restricted process and reveal multiple routes by which tumors escape PARP inhibitor therapy. Statement of SignificanceWe show that high-grade serous ovarian cancers can restore BRCA1 expression after therapy through multiple genomic mechanisms, including local methylation loss and promoter bypass, thereby re-establishing homologous recombination and driving PARP inhibitor resistance. These findings challenge reliance on BRCA1 methylation alone as a predictive biomarker and support rational combination therapies for more durable responses.

cancer biology↗

bioRxiv2026

Beyond Generic Signal Peptides: ApexSP Enables Cargo-Specific Secretion Design

Recombinant proteins are widely used in biopharmaceuticals, industrial manufacturing, and molecular diagnostics; however, efficient secretion remains a major bottleneck limiting their large scale production. As core elements controlling protein entry into secretion pathways, signal peptides do not function solely based on their own sequences, but rather depend on coordinated compatibility among cargo protein properties, secretion pathways, and host backgrounds. Current signal peptide engineering mainly relies on a limited number of commonly used signal peptides, empirical selection, and individual experimental screening, making it difficult to design efficient secretion elements tailored to specific expression systems. Here, we present ApexSP, a signal peptide design framework for secretion engineering tailored to individual cargo proteins. ApexSP is built upon a large scale, high quality signal peptide knowledge base and integrates a discrete diffusion generative model constrained by evolutionary information, a multitask biological filter incorporating topology information, and a signal peptide and mature protein compatibility ranking model to enable integrated design of signal peptides for specific expression systems. ApexSP achieved high accuracy in multiple attribute prediction tasks, including pathway classification and cleavage site prediction, reaching or exceeding the performance of existing signal peptide prediction tools. Moreover, the cargo protein aware ranking module improved the enrichment of candidates with high secretion performance. Experimental validation across multiple eukaryotic and prokaryotic expression systems demonstrated that screening only 10 candidate sequences generated by ApexSP yielded multiple designed signal peptides outperforming reference signal peptides, with the best performing design in the ApGA expressed Pichia pastoris system achieving a 1.71 fold increase in secretion performance compared with the factor signal peptide. Overall, ApexSP advances signal peptide research from sequence prediction toward secretion element design, demonstrating that a single round of designed signal peptides can achieve high secretion performance suitable for further engineering optimization.

synthetic biology↗

bioRxiv2026

A hierarchical activation threshold shapes RSV F antibody repertoires from toddler vulnerability to adult protection

Respiratory syncytial virus (RSV) causes severe infant morbidity, yet the mechanisms underlying their suboptimal immunity and the failure of prefusion F-stabilized adult vaccines in infants remain unclear. Here we integrate deep mutational scanning of 761 preF-binding antibodies, repertoire profiling of 102 repeatedly exposed pediatricians and 61 RSV-experienced toddlers, structural analysis, and >40,000 viral genomes to decode RSV F immunity. We resolved 12 immunologically distinct functional antibody subclasses and uncovered a hierarchical activation threshold: apical epitope-targeting antibodies with superior neutralizing potency demand extensive somatic hypermutation and cooperative CDR networks, whereas central-to-basal epitope-directed, marginal or non-neutralizing antibodies engage germline-encoded antibodies through minimal mutations (e.g., S31G). Toddler repertoires are confined to low-threshold, non-productive sites; adult repertoires enrich for apical elite neutralizers whose immune pressure drives contemporary RSV-B evolution. These reframe pediatric RSV vulnerability as a threshold-gated repertoire deficit and prescribe vaccine strategies that actively redirect immunodominance from permissive epitopes toward high-barrier apical targets.

immunology↗

bioRxiv2026

Medulloblastoma Forms Symbiotic Metabolic Partnerships with Macrophages to Establish Leptomeningeal Metastases

Leptomeningeal metastases are the primary source of morbidity and mortality for pediatric medulloblastoma patients. Due to limited surgical sampling of metastases in patients, little is understood of the mechanisms of metastasis. Here, we identify biologically distinct quiescent small metastases (designated as micrometastases) and mitotically active larger metastases (macrometastases). Macrometastases are more metabolically active than micrometastases and contain higher levels of lipids, particularly cholesterol. Macrometastases secrete CXCL12, which attracts lipid-laden macrophages into the tumor. Lipid-laden macrophages upregulate the cholesterol transporter ABCG1, promoting the efflux of free cholesterol, which is then taken up by tumor cells via the HDL receptor SCARB1. CXCL12-driven macrophage recruitment and exogenous cholesterol are sufficient and necessary to drive progression of medulloblastoma leptomeningeal metastases in vivo. High fat diets drive metastatic progression in vivo. Dietary or pharmacological interventions targeting the CXCL12-SCARB1-cholesterol axis represent therapeutic strategies to either prevent or treat medulloblastoma leptomeningeal metastases.

cancer biology↗

bioRxiv2026

PLANCK: super-multiplex optical imaging without labeling

Molecular information is vital for imaging technology. Optical imaging acquires molecular specificity almost exclusively via labeling strategy, which is fundamentally constrained by limited multiplexing capacity, high running costs, and experimental complexity. Conversely, label-free optical imaging offers substantial technical simplicity but is believed to have little true molecular specificity. Contrary to common belief, here we introduce super-multiplex optical imaging without labeling. By systematically studying paired vibrational spectroscopic imaging and mass spectrometry imaging, we discovered a surprisingly strong (more than 0.9) correlation between their latent space representations, supported by both experiments and theory. This insight prompts us to build supervised learning models to successfully predict spatial distribution of 100 molecular species directly from label-free vibrational images across diverse tissue systems. We developed this technology, named Prediction through Learning with AdvaNced Chemical Kaleidoscope (PLANCK), and demonstrated it with both infrared-based vibrational imaging of organ-scale tissues and Raman-based vibrational imaging of live tissues. Powered by AI, PLANCK decodes the exquisitely rich but otherwise hidden vibrational information into a surprisingly large number of ( [≥] 100) specific molecular species, providing a cost-effective and scalable solution for basic research and translation, including applications in live imaging.

biophysics↗

bioRxiv2026

Transfer learning reveals species-specific olfactory preferences in Diptera and informs pest management strategies

The interaction between volatile organic compounds (VOCs) and odorant receptors (ORs), termed VOI, constitutes the molecular basis for species to recognize chemical information in the environment. Characterization of subtle differences in VOI relationships among closely related species is crucial for understanding the rapid evolution of ORs and for identifying species-specific chemical cues relevant to pest management. However, progress in this field has been constrained by both the scarcity of experimental data and limitations in computational prediction accuracy. To address this, we developed a virtual screening-enhanced transfer learning strategy that integrates large-scale molecular docking data with sparse functional experimental data. The resulting VOI prediction model was validated through functional experiments on the pest species Bactrocera dorsalis, demonstrating its cross-species predictive capacity. Using this model, we investigated the relationship between ecological niches and olfactory sensitivity in Diptera insects from multiple perspectives, with an emphasis on patterns that may inform pest behavioral research. The models predictive reliability was further supported by its consistency with known olfactory trends: it recapitulated the preferential responses of fruit flies and mosquitoes to esters and aromatics, respectively, and reproduced the previously reported negative correlation between olfactory and visual capacities in Drosophila. As a proof-of-concept application, we compared hematophagous and non-hematophagous mosquitoes, revealing overlapping chemical spaces but distinct patterns in their specifically recognized compounds. This study provides a reliable VOI prediction framework that reveals species-level olfactory preference patterns in Diptera, offering a computational pipeline to accelerate the discovery of behaviorally active compounds for species-specific pest monitoring and control.

bioinformatics↗

bioRxiv2026

BOSE: A Bayesian Order Statistics-Based Estimator for Recovering the Sample Mean and Standard Deviation

In meta-analyses of continuous outcomes, the sample mean and standard deviation (SD) are essential for synthesizing effect sizes across studies. However, clinical studies frequently report alternative summary statistics, such as the median, quartiles, and range. To enable inclusion of such studies, various methods have been proposed to estimate the sample mean and SD from these reported summaries. We propose the Bayesian Order Statistics-based Estimator (BOSE), which leverages the joint likelihood of observed order statistics together with weakly informative priors to obtain the full posterior distribution for the mean and SD without relying on computationally intensive iterative procedures such as Markov chain Monte Carlo algorithms. Our numerical studies demonstrate that BOSE performs competitively with existing approaches in estimating the mean, while achieving superior performance for estimating the SD across all evaluated scenarios, particularly in small-sample settings. Under non-normal distributions including skewed, heavy-tailed, and bimodal settings with mild or moderate deviations from normality, BOSE remains robust and stable, whereas methods specifically designed for skewed distributions may become unstable or even inapplicable. Beyond point estimation, BOSE naturally provides empirically validated posterior credible intervals, enabling researchers to formally quantify uncertainty for study-level estimates and make reliable, evidence-based decisions in meta-analytic research synthesis. A publicly accessible web application implementing BOSE and competing methods is also provided to facilitate practical use in meta-analytic research.

bioinformatics↗

bioRxiv2026

Mesothelial Plasticity Specifies a Pleural Immune Circuit that Orchestrates Lung Regeneration

Following injury, predisposition towards regenerative repair and away from degenerative remodeling is central to organismal health, yet the upstream determinants that instruct this fate choice remain poorly understood. Here, we identify mesothelial cell plasticity as a central determinant between regeneration and degeneration by comparing mouse models of pneumonectomy (PNX) versus chronic lung allograft dysfunction (CLAD). While mesothelial cells expand in both settings, following PNX, these cells undergo differentiation through multiple transitional states, culminating in an inflammatory population that orchestrates monocyte recruitment and subsequent tissue regeneration. In contrast, in CLAD, mesothelial cells enrich in an extracellular matrix-enriched state that lacks pro-regenerative signaling capacity. Using single cell epigenomic profiling and in vivo genetics, we define a mesothelium-specific TWIST1-CCL2 axis that governs mesothelium plasticity and signaling, monocyte recruitment and lung regrowth. Together, these findings demonstrate that context-dependent reprogramming of a pleural population, known to protect organs, can be leveraged to drive regeneration.

Developmental Biology↗

bioRxiv2026

DNA-binding specificity recognition from predicted homologous protein-DNA structures

Predicting protein DNA-binding specificity is essential for understanding gene regulation and disease mechanisms. Existing deep learning methods typically infer specificity from a single protein-DNA complex structure, which limits their ability to capture the diverse geometric patterns underlying protein-DNA recognition. Homologous protein-DNA interfaces provide complementary structural evidence and richer geometric features related to interatomic interactions. To address the limited diversity and coverage of experimentally determined complexes, we constructed a large-scale library of predicted homologous protein-DNA complex structures. Building on this resource, we propose HomoDSP, a template-retrieval-based framework for accurate DNA-binding specificity prediction. Benchmark evaluations and validation on newly released JASPAR 2026 samples indicate that HomoDSP outperforms existing methods in both accuracy and generalization, with particularly substantial gains on high-error samples. Moreover, this performance is largely retained when AlphaFold3-predicted complex structures are used as input. Template- and residue-level interpretability analyses suggest that HomoDSP improves prediction by focusing on DNA-affinity residues across multiple homologous templates. Finally, universal Protein Binding Microarrays evaluations on AI-designed DNA-binding proteins show that HomoDSP rescues a baseline failure mode in which the baseline method produces incorrect predictions because of training-set bias. Together, these results support the use of homologous template interfaces as informative structural priors for decoding protein DNA-binding specificity.

bioinformatics↗