Biology subjects

Le, D.

Publications and source records attributed to Le, D..

At least 19 recordsLinked to original sources

bioRxiv2026

Long-Read Metagenomic Sequencing Reveals Persistent Gut Microbiome Alterations in Children with Celiac Disease in Remission

Background: Celiac Disease (CeD) is an immune-mediated enteropathy linked to gut microbiome dysbiosis driven by dietary gluten ingestion. Although most patients achieve clinical and histologic remission on a gluten-free diet (GFD), persistent symptoms and incomplete mucosal recovery are common. While gut microbiome alterations during active CeD are well documented, taxonomic and functional shifts during histologic remission remain considerably less characterized. Methods: We utilized PacBio's HiFi long-read metagenomic sequencing to profile the gut microbiome of 21 pediatric participants with histologically confirmed CeD remission on a GFD, active CeD, and non-CeD controls. Results: High-fidelity long reads enabled reconstruction of 857 high-quality metagenome-assembled genomes and high-resolution species- and pathway-level profiling. Microbial diversity analysis showed reduced alpha-diversity in remission patients compared with controls. Differential abundance analysis revealed persistent taxonomic and functional alterations in CeD remission, indicating incomplete restoration of microbiome homeostasis. Notably, remission patients showed depletion of commensals such as Alistipes finegoldii, Bifidobacterium longum, and Bacteroides uniformis, and persistence of some inflammation-associated taxa, alongside enrichment of beneficial species, including short-chain fatty acid producers, and suppression of other inflammation-associated taxa. At the functional level, pathways related to broader metabolic homeostasis were depleted in remission, while anti-inflammatory pathways such as propionate metabolism were enriched. Conclusions: Our findings suggest a distinct microbial ecosystem during histologic remission characterized by residual taxonomic dysbiosis and functional constraints, accompanied by adaptive remodeling that may promote mucosal healing. This work advances understanding of the gut microbiome in CeD after mucosal recovery and lays the groundwork for future studies of microbiome-based therapeutic interventions.

microbiology↗

bioRxiv2026

Direct RNA Sequencing reveals epitranscriptomic regulation of brain cells and Alzheimer's Disease pathology

Alternative mRNA splicing and post-transcriptional RNA modification are key mechanisms that regulate transcript function; however, their role in neuronal activity and neurodegenerative disease remains poorly defined. In this study, we evaluated two nanopore-based long-read sequencing (LR-seq) formats: cDNA-PCR sequencing (CPS) and direct RNA sequencing (DRS). We then applied DRS to profile both full-length isoforms and RNA modifications in major brain cell types derived from induced pluripotent stem cells (iPSCs) and post-mortem Alzheimers disease (AD) brains. Relative to CPS, DRS achieved higher accuracy and sensitivity for transcript quantification, de novo transcript model construction, and open reading frame (ORF) annotation across neuropathological gene sets. Focusing on iPSC-derived neurons, we built a multi-omic atlas to connect transcriptional output with translational engagement and protein abundance, by integrating DRS-based mRNA abundance, N6-methyladenosine (m6A) status and poly(A) tail length with ribosome profiling (Ribo-seq) and mass spectrometry (MS). The combination of DRS and Ribo-seq data demonstrated synergism in predicting protein abundance. This analysis also uncovered a significant inverse relationship between m6A modification and mRNA abundance, which was dependent on the engagement of the ribosomal A-site. Lastly, we applied DRS to the epitranscriptomic analysis of AD brain samples, demonstrating that m6A profiles can be used to distinguish early-versus late-stage disease.

neuroscience↗

bioRxiv2026

Phosphorylated ubiquitin is a secondary messenger and an epigenetic mark mediating mitochondria to nucleus signaling

AbstractParkinsons disease (PD) is commonly associated with dysfunctional mitochondrial homeostasis. PINK1, a S/T kinase mutated in early-onset PD, generates phosphoserine 65 ubiquitin (pS65Ub) on damaged mitochondria facilitating their removal. Here, we show that pS65Ub translocates into the nucleus after generation at damaged mitochondria and is directly attached to substrates by resident E3 ligases. Histone H2A is a major substrate and is modified at lysine 119 (H2AK119) by the polycomb silencer, E3 ligase RING1B. At nucleosomes, pS65Ub simultaneously suppresses RING1B and potentiates H2A deubiquitinases USP16 and USP21. Epigenetic profiling and RNA sequencing reveal that pS65Ub is enriched at the promoters of poorly expressed yet dynamically regulated genes and is associated with H2AK119ub depletion. Functionally, we show that pS65Ub enrichment drives polycomb target gene expression, which accelerates the maturation of dopaminergic neurons. Importantly, post-mortem PD brains exhibit elevated nuclear pS65Ub, potentially linking nuclear pS65Ub accumulation with disease pathogenesis. Together, these data indicate that pS65Ub generated at damaged mitochondria regulates fundamental cellular processes at distant sites.

cell biology↗

bioRxiv2026

2D spatial tissue analysis often misrepresents true biological patterns of 3D tissues

Spatial omics technologies and analysis methods that profile tissue sections are widespread in biomedical research, but how well these methods capture 3D tissue biology is unknown. Using 3D spatial simulations and data, coupled with virtual tissue sectioning, we found that spurious associations, misrepresentation of biological trends and false negative findings are common in 2D spatial analyses. Understanding the limitations of 2D spatial analyses is needed to avoid misinterpreting research findings.

bioinformatics↗

bioRxiv2025

Split trial analysis reveals the information capacity of neural population codes

Understanding how correlated neural noise affects neural population coding is a basic question in computational and systems neuroscience [1, 2, 3, 4]. Recent theoretical work suggests that shared noise along the stimulus encoding direction is the primary factor that limits information encoding (i.e., information-limiting noise) [5, 6]. Despite this theoretical insight, it has been difficult to test it experimentally due to the challenges in inferring information-limiting noise from neural data. To overcome this challenge, we have developed a method (i.e., split-trial analysis) to partition the noise in a neural population into information-limiting noise vs. non-information-limiting components. Our method is simple to implement, yet it is highly effective given a limited amount of data. Results from extensive numerical simulations show that split-trial analysis substantially outperforms existing methods in accuracy, efficiency, and robustness. Applications of split-trial analysis to a number of neurophysiological datasets reveals insights into the precision of the neural codes for several systems. First, it reveals a substantial amount of information-limiting noise in the mouse head direction system. Second, it uncovers a small yet positive information-limiting noise in the orientation code in mouse V1. Third, we discover that the information-limiting noise in the macaque pre-frontal cortex is highly consistent over time during a simple saccade task. Split-trial analysis is a general technique that should be widely applicable in analyzing the properties of neural population codes.

neuroscience↗

bioRxiv2025

Beyond Gatekeeping: Efflux Pumps Remotely Destabilize Cytoplasmic Drug-Target Interactions by Limiting Rebinding.

Bacterial efflux pumps are major contributors to multidrug resistance, classically described as "gatekeepers" that reduce drug entry. Here, we uncover an equally potent post-entry mechanism of efflux pumps, revealing that their function extends deep into intracellular drug-target interactions. Using quantitative live-cell fluorescence imaging, we monitored the activity of major efflux systems in Escherichia coli and Pseudomonas aeruginosa with Hoechst (HCT), a DNA-binding inhibitor. As expected, more HCT entered efflux-deficient cells than efflux-active cells. Unexpectedly, the intracellular HCT in efflux-deficient cells also bound DNA more stably, resulting in higher binding affinity. Statistical-physics-based modeling and experimental testing show that that unlike in dilute in vitro conditions, drug molecules that unbind from their targets in intracellular environments rapidly undergo successive rebinding, prolonging the lifetime of the drug-target complex. However, efflux pumps counteract this effect by suppressing rebinding, thereby kinetically destabilizing drug-target interactions and lowering the binding affinity. This kinetic destabilization acts synergistically with canonical gatekeeping to broaden and amplify drug resistance. Our findings reveal an underappreciated biophysical mechanism of efflux pumps, expanding the classical biochemical view of how efflux pumps confer multidrug resistance.

microbiology↗

bioRxiv2025

Zero-shot segmentation using embeddings from a protein language model identifies functional regions in the human proteome

The biological function of a protein is often determined by its distinct functional units, such as folded domains and intrinsically disordered regions. Identifying and categorizing these protein segments from sequence has been a major focus in computational biology which has enabled the automatic annotation of folded protein domains. Here we show that embeddings from the unsupervised protein language model ProtT5 can be used to identify and categorize protein segments without relying on conserved patterns in primary amino acid sequence. We present Zero-shot Protein Segmentation (ZPS), where we use embeddings from ProtT5 to predict the boundaries of protein segments without training or fine-tuning any parameters. We find that ZPS boundary predictions for the human proteome are more consistent with reviewed annotations from UniProt than established bioinformatics tools and ProtT5 embeddings of ZPS segments can categorize folded domains, sub-domains, and intrinsically disordered regions. To explore ZPS predictions, we introduce a new way to visualize protein embeddings that closely resembles diagrams of distinct functional units in protein biology. Since ZPS and segment embeddings can be used without training or fine-tuning, the approach is not biased towards known annotations and can used to identify and categorize unannotated protein segments. We used the segment embeddings to identify unannotated mitochondrion targeting signals and SYGQ-rich prion-like domains, which are functional regions within intrinsically disordered regions. We expect the protein segment organization revealed here to lead to valuable information about protein function, including about intrinsically disordered regions and other less understood protein regions.

bioinformatics↗

bioRxiv2025

Promoter-centric gene regulation in drug-resistant cancer

In eukaryotic cells, gene-distal regulatory elements (REs) facilitate long-range gene regulation, ensuring cell type-specific transcriptional programs. This mechanism is frequently disrupted in cancer, often driven by transcription factors (TFs) that serve as targets for cancer therapy. However, targeting these TFs can lead to acquired resistance mechanisms that are not fully understood. We demonstrate that mesothelioma cancer cells, dependent on the oncogenic driver TF family TEAD, develop resistance to a pan-TEAD inhibitor and revert to an evolutionarily ancient, promoter-centric gene regulatory mechanism to recover gene expression following TEAD inhibition. Base-pair-resolution 3D chromatin conformation mapping reveals that RE-promoter interactions are disrupted in resistant cells, despite epigenetic and transcriptomic recovery. Mechanistically, in resistant cells, TF complexes, including resistance-specific FOSL1 and KLF4, preferentially bind and enhance promoter activity to recover gene expression, rendering distal REs dispensable. Our findings highlight promoter elements and promoter-specific TFs as potential therapeutic targets using a model of drug-resistant cancer.

cancer biology↗

bioRxiv2024

Classification of Visual Imagery and Imagined Speech EEG based Brain Computer Interfaces using 1D Convolutional Neural Network

Non-invasive brain-computer interfaces (BCI) utilising electroencephalogram (EEG) signals are a current popular, affordable and accessible method for establishing communication paths between the mind and external devices. However, the challenges faced are inter-subject variability, BCI illiteracy and poor machine learning decoding performance. Two emerging intuitive mental paradigms, Visual Imagery (VI) and Imagined Speech (IS) show promise to optimise the development of non-invasive BCIs, which involves the extraction of corresponding neural patterns during the imagined tasks. This study took a comprehensive user-centric approach to build on the current foundation of knowledge on VI and IS EEG-BCIs utilising an adapted 1D-CNN to optimise the classification decoding performance. Twenty healthy participants were assessed for their ability to visualise imagery in their minds and performed the VI and IS mental paradigms in two class conditions "push" and "relax". It was shown that alpha and beta suppression was observed during the "push" condition of VI compared to the "relax" condition, and those that scored higher in the VVIQ had better VI classification accuracy than those who did not. The adapted 1D-CNN model performed well for classification between the two classes "push" and "relax" at 89.3% and 77.87% performance accuracy for VI and IS, respectively. These findings contribute to the current body of work on VI BCI, that it is a dynamic and plausible option compared to standard BCI paradigms, and VI BCI illiteracy could potentially be controlled by VVIQ. It also demonstrated the potential of the 1D-CNN model in classification of VI and IS EEG-BCIs.

neuroscience↗

bioRxiv2024

TEAD-targeting small molecules induce a cofactor switch to regulate the Hippo pathway

TEAD proteins are the main transcriptional effectors of the Hippo signaling pathway and a pharmacological target in oncology. Most TEAD-targeting small molecules act by disrupting interaction with the oncogenic transcriptional activators YAP and TAZ. Here, we describe an alternative mechanism for TEAD lipid pocket binding molecules. We report that select sulfonamide-containing compounds promote TEAD interaction with the transcriptional repressor VGLL4 to induce a small molecule-mediated cofactor switch from YAP to VGLL4. Chemically induced VGLL4-TEAD complexes counteract YAP activity at chromatin to repress pro-growth gene networks, including genes involved in cellular proliferation and mechanosignaling. VGLL4 is required for an anti-proliferative response to these select compounds, and genetic deletion of VGLL4 causes resistance to these molecules in vitro and in vivo. Our data reveal a category of molecules that facilitate the repressive VGLL4-TEAD interaction and open up new understandings for curbing the oncogenic activity of Hippo pathway deregulation.

cancer biology↗

bioRxiv2024

Clonal dynamics and somatic evolution of haematopoiesis in mouse

Haematopoietic stem cells maintain blood production throughout life. While extensively characterised using the laboratory mouse, little is known about how the population is sustained and evolves with age. We isolated stem cells and progenitors from young and old mice, identifying 221,890 somatic mutations genome-wide in 1845 single cell-derived colonies, and used phylogenetic analysis to infer the ontogeny and population dynamics of the stem cell pool. Mouse stem cells and progenitors accrue [~]45 somatic mutations per year, a rate only about 2-fold greater than human progenitors despite the vastly different organismal sizes and lifespans. Phylogenetic patterns reveal that stem and multipotent progenitor cell pools are both established during embryogenesis, after which they independently self-renew in parallel over life. The stem cell pool grows steadily over the mouse lifespan to approximately 70,000 cells, self-renewing about every six weeks. Aged mice did not display the profound loss of stem cell clonal diversity characteristic of human haematopoietic ageing. However, targeted sequencing revealed small, expanded clones in the context of murine ageing, which were larger and more numerous following haematological perturbations and exhibited a selection landscape similar to humans. Our data illustrate both conserved features of population dynamics of blood and distinct patterns of age-associated somatic evolution in the short-lived mouse.

genomics↗

bioRxiv2024

Comparative Analysis of Commercial Single-Cell RNA Sequencing Technologies

This study evaluates ten commercially available single-cell RNA sequencing (scRNA-seq) approaches across four technology groups: Emulsion-based kits from 10x Genomics and Fluent Biosciences; Microwell-based kits from Becton Dickinson, Honeycomb Technologies and Singerlon Technologies; Combinatorial-indexing kits from Parse Biosciences and Scale Biosciences; and a Matrigel-based kit from Scipio Biosciences. Peripheral blood mononuclear cells (PBMCs) from a single donor were used to assess analytical performance. Key features such as sample compatibility, cost, and experimental duration were also compared. Notably, superior analytical performance was demonstrated by the Chromium Fixed RNA Profiling kit from 10x Genomics, which uniquely features probe hybridization for transcript detection. Additionally, the Rhapsody WTA kit from Becton Dickinson provided a cost-effective balance of performance and expense per cell. With a rich dataset of 218,154 cells, this work provides a basis for differentiating commercial scRNA-seq technologies, which is intended to facilitate the effective application and further methodological development of single cell transcriptomics.

molecular biology↗

bioRxiv2024

Super-enhancer interactomes from single cells link clustering and transcription

Regulation of gene expression hinges on the interplay between enhancers and promoters, traditionally explored through pairwise analyses. Recent advancements in mapping genome folding, like GAM, SPRITE, and multi-contact Hi-C, have uncovered multi-way interactions among super-enhancers (SEs), spanning megabases, yet have not measured their frequency in single cells or the relationship between clustering and transcription. To close this gap, here we used multiplexed imaging to map the 3D positions of 376 SEs across thousands of mammalian nuclei. Notably, our single-cell images reveal that while SE-SE contacts are rare, SEs often form looser associations we termed "communities". These communities, averaging 4-5 SEs, assemble cooperatively under the combined effects of genomic tethers, Pol2 clustering, and nuclear compartmentalization. Larger communities are associated with more frequent and larger transcriptional bursts. Our work provides insights about the SE interactome in single cells that challenge existing hypotheses on SE clustering in the context of transcriptional regulation.

molecular biology↗

bioRxiv2023

Accurate long-read transcript discovery and quantification at single-cell resolution with Isosceles

Accurate detection and quantification of mRNA isoforms from nanopore long-read sequencing remains challenged by technical noise, particularly in single cells. To address this, we introduce Isosceles, a computational toolkit that outperforms other methods in isoform detection sensitivity and quantification accuracy across single-cell, pseudo-bulk and bulk resolution levels, as demonstrated using synthetic and biologically-derived datasets. Isosceles improves the fidelity of single-cell transcriptome quantification at the isoform-level, and enables flexible downstream analysis. As a case study, we apply Isosceles, uncovering coordinated splicing within and between neuronal differentiation lineages. Isosceles is suitable to be applied in diverse biological systems, facilitating studies of cellular heterogeneity across biomedical research applications.

bioinformatics↗

bioRxiv2023

A deep-learning-based screening platform in aging-relevant human motor neurons to identify therapeutic compounds for amyotrophic lateral sclerosis

Age is the primary risk factor for most neurodegenerative diseases including amyo-trophic lateral sclerosis (ALS). Despite the clear importance of age on the development and progression of ALS, age is rarely considered a factor in cellular or animal models of ALS. The recent advent of direct reprogramming methodologies, whereby somatic cells such as fibroblasts can be converted into neurons while retaining the age signature of the host, has facilitated the study of age-relevant human cells in vitro for the first time. Despite the promise of this technology, age is a complex, multidimensional, and dy-namic phenotype that makes the interpretation of the interplay between age and dis-ease a great challenge. To circumvent these challenges, we developed a screening platform for directly reprogrammed neurons in an age-relevant human cellular model to better model and identify disease-modifying targets and pathways for ALS. This plat-form uses deep learning image analysis to screen compounds for efficacy against ALS-associated cellular phenotypes and efficiently classifies ALS-like phenotypic signatures as well as predicts the age of the original fibroblast donor. Notably, our work identified NCB-0846, a TNIK/MAP4K7 inhibitor, as a novel compound with the ability to revert ALS-like phenotypes. Moreover, we show that chronic dosing of NCB-0846 in the SOD1G93A mouse model of ALS significantly reduced serum neurofilament-L levels. Our findings establish the power of a platform that combines direct reprogramming of human cells and deep learning as a powerful tool for combatting aging and age-related diseases.

neuroscience↗

bioRxiv2023

Single-cell chromatin state transitions during epigenetic memory formation

Repressive chromatin modifications are thought to compact chromatin to silence transcription. However, it is unclear how chromatin structure changes during silencing and epigenetic memory formation. We measured gene expression and chromatin structure in single cells after recruitment and release of repressors at a reporter gene. Chromatin structure is heterogeneous, with open and compact conformations present in both active and silent states. Recruitment of repressors associated with epigenetic memory produces chromatin compaction across 10-20 kilobases, while reversible silencing does not cause compaction at this scale. Chromatin compaction is inherited, but changes molecularly over time from histone methylation (H3K9me3) to DNA methylation. The level of compaction at the end of silencing quantitatively predicts epigenetic memory weeks later. Similarly, chromatin compaction at the Nanog locus predicts the degree of stem-cell fate commitment. These findings suggest that the chromatin state across tens of kilobases, beyond the gene itself, is important for epigenetic memory formation.

genetics↗

bioRxiv2023

Schizophrenia Risk Mapping and Functional Engineering of the 3D Genome in Three Neuronal Subtypes

Common variants associated with schizophrenia are concentrated in non-coding regulatory sequences, but their precise target genes are context-dependent and impacted by cell-type-specific three-dimensional spatial chromatin organization. Here, we map long-range chromosomal conformations in isogenic human dopaminergic, GABAergic, and glutamatergic neurons to track developmentally programmed shifts in the regulatory activity of schizophrenia risk loci. Massive repressive compartmentalization, concomitant with the emergence of hundreds of neuron-specific multi-valent chromatin architectural stripes, occurs during neuronal differentiation, with genes interconnected to genetic risk loci through these long-range chromatin structures differing in their biological roles from genes more proximal to sequences conferring heritable risk. Chemically induced CRISPR-guided chromosomal loop-engineering for the proximal risk gene SNAP91 and distal risk gene BHLHE22 profoundly alters synaptic development and functional activity. Our findings highlight the large-scale cell-type-specific reorganization of chromosomal conformations at schizophrenia risk loci during neurodevelopment and establish a causal link between risk-associated gene-regulatory loop structures and neuronal function.

genomics↗

bioRxiv2023

Single-cell long-read targeted sequencing reveals transcriptional variation in ovarian cancer

Single-cell RNA sequencing predominantly employs short-read sequencing to characterize cell types, states and dynamics; however, it is inadequate for comprehensive characterization of RNA isoforms. Long-read sequencing technologies enable single-cell RNA isoform detection but are hampered by lower throughput and unintended sequencing of artifacts. Here we developed Single-cell Targeted Isoform Long-Read Sequencing (scTaILoR-seq), a hybridization capture method which targets over a thousand genes of interest, improving the median number of unique transcripts per cell by 29-fold. We used scTaILoR-seq to identify and quantify RNA isoforms from ovarian cancer cell lines and primary tumors, yielding 10,796 single-cell transcriptomes. Using long-read variant calling we revealed associations of expressed single nucleotide variants (SNVs) with alternative transcript structures. In addition, phasing of SNVs across transcripts facilitated measurement of allelic imbalance within distinct cell populations. Overall, scTaILoR-seq is a long-read targeted RNA sequencing method and analytical framework for exploring transcriptional variation at single-cell resolution.

genomics↗