Xiaoying Fan

Following zygotic genome activation (ZGA) in embryos, nuclear speckles, which are subnuclear domains where pre-mRNA undergoes splicing, are reassembled in a process called zygotic splicing activation (ZSA). ZSA is vital for early development, but its regulation is poorly understood. This study identifies the RNA-binding protein ZC3H14 as a key promoter of nuclear speckle formation during ZSA. Although viable, Zc3h14 knockout mice exhibit reduced female fertility. Most embryos from these females arrest during the 1-cell to 2-cell transition, and the remaining embryos arrest at the 2-cell stage, highlighting a critical role for maternally deposited ZC3H14. The protein localizes to nuclear speckles, and its phosphorylation by CDK12/13 regulates mRNA binding and initial speckles assembly. Blocking this phosphorylation disrupts proper splicing and leads to abnormal RNA accumulation. Consequently, phosphorylation-mimicking or phosphorylation-blocking ZC3H14 mutants expressed in zygotes positively and negatively affect development, respectively. Taken together, our results indicate that ZC3H14 phosphorylation by CDK12/13 maintains fertility in female mice by promoting nuclear speckle formation during ZSA.

Nanopore direct RNA sequencing (dRNA-Seq) simultaneously resolves RNA sequences, poly(A) tail lengths, and base modifications at the single-molecule level. Here, we developed RB-dRNAseq, a multiplexed dRNA-Seq method using RNA-barcoded reverse transcription adapters (RTAs) compatible with Oxford Nanopore Technologies (ONT) workflows, enabling highly efficient sample demultiplexing based solely on basecalled sequences. On the RNA004 platform, RB-dRNAseq achieved >99.5% demultiplexing accuracy, with 81% of reads assigned in 6-plex samples. This represents a 16%-24% relative increase in read recovery compared to conventional DNA barcoding methods that rely on basecalling-based demultiplexing. To evaluate the sensitivity of RB-dRNAseq, we performed sequencing with low-input samples, demonstrating that even with 1 ng of total RNA, gene expression (r=0.84) and m6A modification (r=0.91) profiles remained highly consistent with high-input benchmarks. We further explored the feasibility of RB-dRNAseq in ultra-low input scenarios by applying it to mouse zygotes and 2-cell stage embryos. The method successfully captured dynamic shifts in gene isoform expression and multiple RNA modifications (m6A, m5C, pseU, and Inosine) during zygotic genome activation (ZGA), uncovering site-specific regulatory patterns.

Chronic hepatitis B virus (HBV) persistence relies on the chromatin plasticity of covalently closed circular DNA (cccDNA), a viral minichromosome resistant to current therapies. Using proximity labeling (TurboID-dCas9), ChIP-seq and DNA pull-down assays, we identified SMARCC2—a BAF scaffolding subunit—bound to cccDNA enhancer-promoter regions (EnhⅠ/XP, CP/EnhII), where it sustains nucleosome-depleted regions (NDRs) and recruits RNA polymerase II. Genetic or pharmacological BAF inhibition compacted cccDNA chromatin, reduced histone acetylation (AcH3/AcH4), and enhanced SMC5/6-mediated silencing to suppress transcription, with the BAF ATPase inhibitor FHT-2344 reducing serum HBV DNA by 50% (P <.05) and intrahepatic HBV RNA by 70% (P <.01) without cccDNA loss, indicating epigenetic silencing. Mechanistically, BAF maintains NDRs by counteracting nucleosome retention and recruiting host transcription factors such as HNF4α. This work concludes that BAF safeguards cccDNA chromatin plasticity to enable viral persistence, and targeting BAF (e.g. FHT-2344) epigenetically silences cccDNA, offering a novel strategy for functional cure.

Background: DNA methylation and chromatin accessibility are pivotal epigenetic regulators of gene expression and cellular identity, with significant implications in tumorigenesis and progression. Current single‐cell multi‐omics methods are limited in throughput and sensitivity, hindering comprehensive biomarker discovery. Methods: We developed single‐cell split‐pool ligation‐based multi‐omics sequencing technology (SpliCOOL‐seq), a high‐throughput single‐cell sequencing technology that simultaneously profiles whole‐genome DNA methylation and chromatin accessibility in thousands of cells. By integrating in situ GpC methylation, universal Tn5 tagmentation, and split‐pool combinatorial barcoding, SpliCOOL‐seq achieves enhanced sensitivity and scalability. Results: SpliCOOL‐seq accurately distinguished lung cancer cell types based on genetic and multiple epigenetic modalities and revealed that the two DNA methyltransferase (DNMT) inhibitors, 5‐Azacitidine and Decitabine, both cause large‐scale demethylation but in distinct patterns. Applied to primary lung adenocarcinoma, SpliCOOL‐seq identified tumour subclones within the tumour lesion and uncovered novel DNA methylation biomarkers (e.g., FAM124B, SFN, OR7E47P ) associated with patient survival. Additionally, we demonstrated accelerated epigenetic ageing and mitotic activity in tumour subclones, providing new insights into tumorigenesis. Conclusion: SpliCOOL‐seq achieves parallel profiling of whole‐genome DNA methylation and chromatin accessibility in the same individual cells in a high‐throughput manner and is hopefully used to illustrate regulatory interactions under different cell states. SpliCOOL‐seq enables high‐resolution, multi‐modal epigenetic profiling at single‐cell resolution, offering a powerful platform for discovering cancer biomarkers. Its application reveals novel therapeutic targets and early‐diagnostic markers, underscoring its potential in precision oncology. Key points: SpliCOOL‐seq achieves high‐throughput single‐cell co‐profiling of DNA methylation and chromatin accessibility. DNMT inhibitors caused cancer cell demethylation with divergent patterns. SpliCOOL‐seq enables the discovery of genes related to LUAD tumorigenesis. Ageing and LUAD tumorigenesis may share similar epigenetic alterations.

DNA methylation shapes cellular identity and genome function, yet the sequence logic that governs its establishment remains largely unresolved. Here we present MethylAI, a cross-species–pretrained and human-specialized deep learning framework that predicts single-CpG methylation states directly from genomic sequence with high fidelity. Through quantitative attribution, MethylAI uncovers intrinsic cis-regulatory principles embedded in DNA, revealing conserved transcription factor (TF) motifs whose influence extends beyond CpG composition. Activated motifs demarcate gene-body CpG islands, encode lineage-defining methylation states, and align with TF occupancy and diverse histone modifications, thereby linking sequence syntax to local chromatin regulation. CTCF perturbation experiments validated MethylAI-predicted methylation shifts at activated motifs, showing strong concordance with in silico perturbations. Mapping genetic variants onto methylation-linked active motifs further uncovered widespread connections to zinc-finger TF families, providing mechanistic insight into how noncoding variation contributes to human traits and diseases. By transforming DNA sequence into interpretable methylation grammar, MethylAI establishes a generalizable framework for decoding the regulatory architecture of the human epigenome.

RNA splicing is pivotal in neural development, yet the role of isoform diversity across cell types remains unclear. Here, we combined metabolic RNA labeling and single-cell full-length transcriptome sequencing to capture transcriptional dynamics in developing mouse cortices. We observed predetermined cell states supported by nascent RNAs and characterized the driving isoforms of transcription factors that regulated the development of deep- and upper-layer neurons. Additionally, we investigated isoform regulation associated with autism spectrum disorder (ASD) during the embryonic development of BTBR T + Itpr3tf mice. Our findings indicated premature emergence of callosal projection neurons (CPNs) with an immature identity in ASD-affected cortices. These CPNs exhibited abnormal transcript usage, and the related RNA binding proteins included nearly 60% that have been reported to be ASD risk genes. We identified isoform switching events modulating neurogenesis and ASD development. Finally, we observed reduced isoform diversity in ASD potentially linked to dysregulated H3K27ac levels. Collectively, our study represents a significant advancement in understanding the molecular basis of cortical development and functions.

Hepatocellular carcinomas (HCC) with multinodular morphology were typically polyclonal and exhibited extensive heterogeneity. We explored the multiple-layer heterogeneity of HCC utilizing single-cell multiomics sequencing together with multiregional sampling. We found that Confluent Multi-Nodular samples exhibit more heterogeneous immune landscapes characterized by increased transcriptome heterogeneity and more complex immune-related interactions compared to Single Nodular (SN) samples. We identified differential DNA methylation patterns among distinct tumor foci in a subset of liver cancer patients. Global DNA hypomethylation in cancer cells predominantly occurs in partially methylated domains. We predicted and validated two genes, GADD45A and SNHG6, as potential drivers of DNA hypomethylation. Lastly, we demonstrated that DNA methylation distance serves as a more robust metric than gene expression distance for reconstructing tumor evolutionary trajectories.

Recent advancements in cancer immunotherapy have improved patient outcomes, yet responses to immunotherapy remain moderate. Immunosenescence has been shown to contribute to the development and progression of various diseases; however, its specific role in solid tumors has not been fully delineated. Here we conducted a phase 2 clinical trial involving 51 patients with cancer undergoing neoadjuvant chemoimmunotherapy and applied single-cell RNA as well as TCR and BCR sequencing on tumor and blood samples to elucidate the immune cell perturbations. Our findings associate poor response with reduced levels of CCR7⁺ CD4⁺ naive T cells and CD27⁺ memory B cells, as well as higher expression of immunosenescence-related genes in T and B cell subsets. Using naturally aged mice and Ercc1-deficient mice (premature aging), we found that senolytics enhance the therapeutic efficacy of immunotherapy in multiple solid tumors by mitigating immunosenescence. Notably, we launched a phase 2 clinical trial (COIS-01) investigating the combination of senolytics with anti-PD-1 therapy. The results showed that the combination therapy achieved a 33.3% (95% confidence interval 16.6–54.7%) major pathological response rate with a low incidence of grade 3–4 adverse events (4.2%). These findings underscore the pivotal role of immunosenescence characteristics in influencing the effectiveness of immunotherapy and suggest a promising therapeutic efficacy along with a favorable safety for the combination of senolytics with anti-PD-1 therapy. ClinicalTrials.gov Identifier: OOC-001(NCT04718415) and COIS-01(NCT05724329).

Enhancers, through the combinatorial action of transcription factors (TFs), dictate both the spatial specificity and the levels of gene expression, and their aberrations can result in diseases. While a HMX1 downstream enhancer is associated with ear malformations, the mechanisms underlying bilateral constricted ear (BCE) remain unclear. Here, we identify a copy number variation (CNV) containing three enhancers—collectively termed the positional identity hierarchical enhancer cluster (PI-HEC)—that drives BCE by coordinately regulating HMX1 expression. Each enhancer exhibits distinct activity-location-structure features, and the dominant enhancer with high mobility group (HMG)-box combined with Coordinator and homeodomain TF motifs modulating its activity and specificity, respectively. Mouse models demonstrate that neural crest-derived fibroblasts with aberrant Hmx1 expression in the basal pinna, along with ectopic distal pinna expression, disrupt outer ear development, affecting cartilage, muscle, and epidermis. Our findings elucidate mammalian ear morphogenesis and underscore the complexity of synergistic regulation among enhancers and between enhancers and transcription factors.