Biology subjects

Pandey, P.

Publications and source records attributed to Pandey, P..

At least 19 recordsLinked to original sources

bioRxiv2026

A gut-on-chip model reveals architecture-dependent epithelial responses to microbial challenge

Advanced in vitro models that better recapitulate human intestinal physiology are needed to improve translational relevance and reduce reliance on animal experimentation. Microfluidic gut-on-chip systems provide controlled architectural and mechanical microenvironments that more closely resemble native tissue than conventional static cultures. However, how these features influence epithelial responses to microbial stimuli remains incompletely understood. In this study, we compared conventional two-dimensional (2D) Caco-2 monolayers with a perfused three-dimensional (3D) gut-on-chip model exhibiting crypt-like epithelial organization. Microbial challenge was employed as a biologically relevant perturbation to investigate barrier-associated responses under distinct culture conditions. Following exposure to the probiotic Escherichia coli Nissle 1917 (EcN), epithelial morphology, barrier-associated gene expression, inflammatory signaling, and temporal response patterns were assessed over time. In a complementary approach, susceptibility to bacterial stress was examined using increasing doses of Citrobacter rodentium (C. rodentium). Both culture systems responded dynamically to microbial exposure; however, differences in response patterns were observed between static and perfused cultures. EcN induced transient alterations in epithelial organization followed by recovery-associated responses, while C. rodentium caused dose-dependent structural and transcriptional changes. Notably, epithelial damage became apparent at lower bacterial loads in conventional 2D cultures, whereas the chip-based model maintained tissue integrity under comparable conditions. Together, these findings indicate that epithelial architecture and the surrounding culture environment influence barrier-associated responses to microbial challenge. The study highlights the utility of gut-on-chip systems for investigating host-microbe interactions and provides a framework for functional comparison of advanced intestinal in vitro models.

cell biology↗

bioRxiv2026

Holocene North Asian gene flow and cryptic substructure in the genome history of Nilgiri tribes

The Nilgiri Hills of the Western Ghats harbour several historically isolated tribal populations whose genomic history remains incompletely resolved. Here, we present a comprehensive analysis of whole genome and complete mitochondrial data from three major Nilgiri groups i.,e, Toda, Kota, and Kurumba, in the context of a broad Eurasian reference panel. Autosomal analyses reveal previously unrecognised fine scale substructure, with both Kota and Kurumba resolving into two genetically distinct clusters (Kota1/Kota2 and Kurumba1/Kurumba2). Principal component, ADMIXTURE, and fineSTRUCTURE analyses identify Toda as the most genetically drifted Nilgiri group, whereas Kota2 exhibits elevated Austroasiatic related ancestry. qpWave analysis indicates that the five genetic clusters require at least three independent ancestry streams, while qpAdm further differentiates the groups based on Onge, Indus Periphery, and Western SteppeMLBA related ancestry. Toda shows the highest Indus Periphery related and lowest Onge related ancestry among the Nilgiri groups. A distinctive ancestry component shared by Kurumba2, Kota1, Gaud and the Sri Lankan Vedda suggests deeper ancestral connections across the southern Indian Ocean region. Mitochondrial genomes reveal strong founder effects and lineage specific patterns, including R5a2b in Toda, M3a1 in Kota, and multiple M subclades together with the rare C4a2c1 lineage in Kurumba. Notably, C4a2c1, previously understudied in South India, shows phylogenetic and ancient DNA continuity with North Asian and Siberian lineages. Bayesian dating suggests its arrival in the Indian subcontinent during the early mid Holocene (~7 ka), followed by local diversification after ~5 ka. Together, these findings reveal a complex history of long range Holocene maternal gene flow and subsequent isolation driven differentiation, providing new insights into the genomic history of South Indian tribal populations.

evolutionary biology↗

bioRxiv2026

ENAH Binding to the Huntingtin Proline-Rich Domain Modulates HTT1a Assembly

The proline-rich domain (PRD) of huntingtin (HTT), located C-terminal to the polyglutamine tract within the exon-1 region, plays a critical role in modulating the aggregation and toxicity of mutant HTT1a. To identify PRD-interacting proteins, we performed mass spectrometry analysis of PRD peptide pulldowns from human neurons. ENAH, a member of the Ena/VASP family of actin regulatory proteins, emerged as the top enriched interactor from neuronal membrane fractions. In neurons, ENAH colocalized with HTT1a species, suggesting a biologically relevant interaction. Because the EVH1 domain of ENAH binds polyproline motifs with high affinity, we examined its interaction with HTT1a using recombinant proteins. EVH1 directly bound HTT1a and unexpectedly formed condensate-like assemblies capable of recruiting HTT1a monomers. Remarkably, EVH1 robustly suppressed HTT1a oligomerization and promoted remodeling of preformed HTT1a fibrils in vitro. These findings demonstrate that ENAH-PRD interactions modulate HTT1a aggregation dynamics and identify EVH1 assembly as a potential proteostatic mechanism regulating mutant HTT1a states. More broadly, this work establishes a framework for investigating how EVH1-containing proteins may influence HTT1a proteostasis and neurotoxicity in Huntington's disease.

neuroscience↗

bioRxiv2026

COP9 complex maintains neuroblast growth and proliferation by regulating Akt/mTor pathway

The nervous system consists of the brain and associated structures that are replete with highly specialised neurons and glia. Central to the development of the brain are Neural Stem Cells, which self-renew to maintain their number and also give rise to differentiated progeny. To identify genes involved in the maintenance of Drosophila neural stem cells, neuroblasts, we performed a protein expression screen followed by a protein knockdown screen using deGradFP system. Through this, we identified CSN7, a COP9 signalosome (CSN) subunit, which is enriched in neuroblasts and essential for neural development. CSN is a highly conserved multi-protein complex that regulates proteasome-mediated protein degradation via modulation of Cullin-RING E3 ligases. We found that loss of CSN7 and CSN1b lead to a decrease in neuroblast size and a reduced mitotic index. Our results show that CSN7/CSN1b regulates Akt-TOR signalling in the developing larval brain. Furthermore, we found that this regulation is mediated by Cul1. Overall, our work describes a hitherto undescribed role for the components of the CSN complex in neural development.

developmental biology↗

bioRxiv2026

LOCALE: Local-Alignment Embeddings for Noise-Robust DNA Search at SRA Scale

Searching petabase-scale repositories of raw sequencing data such as the NIH Sequence Read Archive (SRA) could transform biological discovery, but existing methods either do not scale well or rely on exact k-mer matching that is brittle to sequencing errors and biological divergence. We recast sequence search as dense retrieval: we learn vector embeddings whose inner-product similarity ranks locally aligned sequences above unaligned ones. Our key observation is that effective retrieval does not require accurate regression of global edit distance--it only requires that sequences with better local alignments score higher than sequences with worse ones. We train a DNABERT-2 encoder with an InfoNCE objective on biologically informed augmentations: overlapping crops of parent sequences corrupted with substitutions, insertions, and deletions. On a 50-accession SRA benchmark, LOCALE maintains 62.4% average Recall@Rq at a 10% mutation rate, while every baseline we evaluated falls below 60% Recall@Rq in the noisy-query setting. The advantage holds at scale: on a 500-accession, 15-Gbp benchmark, LOCALE achieves AUPRC 0.508 at 10% mutation versus 0.129 for MetaGraph.

bioinformatics↗

bioRxiv2026

Systems-level analysis of RDK1 reveals compartment-specific kinase activity and a function in the maintenance of the mitochondrial proteome in Trypanosoma brucei

Repressor-of-differentiation kinase 1 (RDK1) is one of two kinases expressed in bloodstream form Trypanosoma brucei parasites that were found to repress premature and spontaneous differentiation into the insect procyclic form. However, the effect of RDK1 RNAi was previously limited to the expression of a single surface coat protein, EP1 procyclin. Thus, there remains a significant gap in knowledge on the impact of RDK1 expression in bloodstream form T. brucei parasites. Here, we employ a systems biology approach and performed several proteomics analyses to identify RDK1 protein interactions and to determine the impact of loss of RDK1 expression on the bloodstream form proteome and phosphoproteome to uncover clues about potential mechanisms for RDK1 function. We found that RDK1 is dual localized to the cell membrane and the mitochondrial inner membrane with the kinase domain oriented towards the cytoplasm and mitochondrial inner membrane. Unexpectedly, the most enriched RDK1-proximal proteins were mitochondrial proteins. Furthermore, RDK1 depletion causes bloodstream form parasites to significantly upregulate many mitochondrial proteins and glycosomal proteins, several of which are upregulated in procyclic form parasites. Surprisingly, the mitochondrial phosphoproteome is largely unaffected by RDK1 depletion, while RDK1-dependent phosphoregulation is restricted to the cell membrane localization of RDK1. Lastly, we determined that RDK1 does not possess adenyl cyclase activity or alter intracellular cAMP levels; however, the dysregulated phosphoproteins correlate with functions in cyclic nucleotide signaling. In conclusion, RDK1 exhibits localization-specific kinase activity to regulate cyclic nucleotide signaling and mitochondrial proteomic maintenance in bloodstream form parasites. IMPORTANCETrypanosoma brucei is the unicellular parasite that causes African sleeping sickness and nagana disease in livestock across 36 sub-Saharan African countries. The parasite encounters different environmental niches as it is transmitted from an infected human to the tsetse fly vector as the fly takes a blood meal. T. brucei must sense environmental cues to initiate intracellular signaling pathways to promote effective differentiation and cellular remodeling from the mammalian bloodstream forms to the insect procyclic form. RDK1 is one of two kinases shown to repress premature differentiation to procyclic form, which would be detrimental for parasite survival in the human host. Therefore, it is essential to uncover mechanisms of RDK1 function to better understand how T. brucei maintains homeostasis in the human host and signals for effective cellular remodeling during parasite transmission.

microbiology↗

bioRxiv2026

South Asian Maternal Lineage haplogroup R30 Provides Phylogenetic Evidence of human dispersal across South Asia

South Asia is central to debates on early human dispersals, particularly the Out of Africa model and Eurasian colonization. Studies of M haplogroups have been used to support both Northern and Southern route hypotheses, but current archaeological and genetic evidence in the region remains contradictory. In the present work, we find that in addition to haplogroup M lineages, a few R lineages exhibit ancient, locally rooted variation, with R30 being one of the widespread haplogroup of R lineages across South Asia. To better understand South Asian demographic history, we investigated the phylogeographic distribution of haplogroup R30, an indigenous lineage. We used 190 complete modern and ancient sequences from diverse mainland and island populations including incorporation of 44 newly generated sequences which enabled the refinement of the R30 phylogeny and the identification of a novel basal lineage, R30c. Bayesian and {rho}-based age estimates suggest that R30 originated in the Indian subcontinent ~50 kya. Early diversification likely occurred in Northern India, giving rise to R30b (~44 kya), while R30a and R30c differentiated primarily in Southern India. Several subclades of haplogroup R30 exhibit strong signatures of founder effects, particularly among the language isolate Vedda of Sri Lanka, Uru Kurumban of Southern India, and the populations of the Lakshadweep archipelago. Bayesian skyline analyses indicate long-term demographic stability followed by rapid lineage expansion ~20 kya and more recent declines consistent with localised drift and relatively recent founder events. The presence of early-diverging R30 lineages in Thailand and Indonesia further supports long-term connections between South and Southeast Asia. Overall, archaeological and genetic evidence point towards the multiple migrations for South Asia colonizations.

evolutionary biology↗

bioRxiv2026

Evaluating 6- and 18-hour stimulation durations for natural killer cell degranulation (CD107a assay) to optimize workflow efficiency in a clinical immunology laboratory

BackgroundNatural killer (NK) cell degranulation is a key immune defense mechanism where exposure to tumor or virus-infected cells triggers the fusion of cytoplasmic granules containing apoptotic proteins, perforin, and granzyme with the cell membrane. This process transiently expresses CD107a on the NK cell surface, and measuring CD107a is a standard method to assess NK cell activity. MethodsWe compared two stimulation protocols differing only in duration (6-hour vs. 18-hour) using K562 target cells to induce NK cell degranulation. Isolated PBMCs without stimulation served as controls to assess spontaneous degranulation. Anti-CD107a-PE antibody was present throughout stimulation in both test and control samples. After stimulation, cells were stained with anti-CD45, anti-CD3, and anti-CD56 and analyzed by flow cytometry. ResultsFor 6 of 7 healthy controls, results from both methods fell within 2 standard deviations. Notably, longer (18-hour) stimulation resulted in lower CD107a expression than the 6-hour assay. Interlaboratory comparisons of two samples showed no significant difference (p>0.05). In a suspected hemophagocytic lymphohistiocytosis (HLH) case, two labs reported similarly reduced CD107a expression (9% and 7%). Inter-day variability was observed in a donor across both time points. The 6-hour assay showed higher sensitivity and specificity than the 18-hour assay. A resting period before ex vivo PBMC assays was found necessary. ConclusionStimulation periods beyond 6 hours are unsuitable for clinical NK degranulation assays. Screening for HLH should include multiple stimulants to improve assay reliability.

immunology↗

bioRxiv2026

kache-hash: A dynamic, concurrent, and cache-efficient hash table for streaming k-mer operations

MotivationHash tables are fundamental to computational genomics, where keys are often k-mers--fixed-length substrings that exhibit a "streaming" property: consecutive k-mers share k-1 nucleotides and are processed in order. Existing static data structures exploit this locality but cannot support dynamic updates, while state-of-the-art concurrent hash tables support dynamic operations but ignore k-mer locality. ResultsWe introduce kache-hash, the first dynamic, concurrent, and resizable hash table that exploits k-mer locality. kache-hash builds on Iceberg hashing--a multi-level design achieving stability and low associativity--but replaces generic hashing with minimizer-based hashing, ensuring that consecutive k-mers map to the same buckets. This keeps frequently accessed buckets cache-resident during streaming operations. On the human genome, kache-hash achieves 1.58-2.62x higher insertion throughput than IcebergHT and up to 6.1x higher query throughput, while incurring 7.39x fewer cache misses. kache-hash scales near-linearly to 16 threads and supports dynamic resizing without sacrificing locality. Our theoretical analysis proves that streaming k-mer operations achieve [O](1/r) amortized cache misses per operation, where r is the minimizer run length, explaining the substantial performance gains over general-purpose hash tables. Availabilitykache-hash is implemented in C++20 and is available at https://github.com/jamshed/kache-hash. Contactp.pandey@northeastern.edu Supplementary informationSupplementary material are available for this manuscript.

bioinformatics↗

bioRxiv2025

Interplay of Spoonbill, Larp7 and survival motor neuron, in Drosophila model of Spinocerebellar Ataxia 8 non-coding RNA associated neurodegeneration

A deeper understanding of neurodegenerative disorders at the level of genetic or environmental risk factors as well as contributing genes, pathways, and networks suggests the presence of shared molecular mechanisms. Previously, we have reported that the KH domain of the Spoonbill protein alone can suppress non-coding Spinocerebellar Ataxia 8 (SCA8) associated neurodegeneration. In the current study, we have identified dLarp7 as a novel interacting partner of Spoonbill in Drosophila. Mammalian Larp7 is associated with a neurodevelopmental disorder, Alazami Syndrome (AS). In this study, we report that dLarp7 protein is recruited into the pathogenic SCA8 RNA foci, which leads to depletion of its downstream target, 7SKsnRNA. Soaking away of Larp7 into toxic RNA foci results in its depletion from the physiological pool resulting in destabilization and depletion of 7SKsnRNP. Hence, increasing the dose of dLarp7 suppressed SCA8 associated neurodegeneration, by restoring the physiological levels of dLarp7 and 7SKsnRNP. In addition, it was observed that dLarp7 interacts with the somatic motor neuron (SMN) protein which is associated with spinal muscular atrophy (SMA). This observation led us to explore the interaction of Drosophila SMN1 orthologue with pathogenic SCA8 associated neurodegeneration. Intriguingly, SMN protein modulated molecular neuropathogenesis associated with SCA8. The presence of orthologues of Drosophila RNA binding proteins, dLarp7 and SMN, with mammalian counterparts underlines the translational importance of our findings. Our study also hints at the shared molecular mechanisms that underlies multiple neurodegenerative diseases. A novel association of Drosophila homologs of AS-linked Larp7 and SMA-causing SMN1 with SCA8 associated neurodegeneration suggests an overlap of molecular threads underlying the pathogenesis of neurodegenerative disorders.

neuroscience↗

bioRxiv2025

Repurposing Melatonin in dual-mode for Wilson disease therapy as a Copper Chelator and an antioxidant agent

Loss-of-function mutations in copper-ATPase ATP7B underlie Wilson disease (WD), a disorder characterized by hepatic copper accumulation and severe hepato-neuropathology. Existing chelation therapeutics remove excess copper but lack intrinsic antioxidant capacity and frequently cause systemic toxicity. Here we evaluate melatonin, an FDA-approved indoleamine with antioxidant and putative metal-chelating activity, as a candidate therapeutic for WD. In ATP7B-/- hepatocytes, melatonin restored copper-induced reactive oxygen species (ROS) to basal levels, reduced apoptosis twofold, and attenuated Nrf2 nuclear translocation leading to reduction of Hemoxygenase-1 abundance. Live-cell ratiometric analysis of GSSG/GSH using GRX1-roGFP2 expressed in melatonin-treated ATP7B-/- hepatocytes revealed a significant reduction in intensity-ratio, indicating an effective mitigation of copper-induced glutathione oxidation. Isothermal calorimetric titration revealed moderate Cu2+ affinity (K ITC=4.54 x 103 M-{superscript 1}), rationalized by MD-simulations showing an interaction energy of 18.5 x 10-3 kcal{middle dot}mol-{superscript 1} via amide-Cu{superscript 2} coordination. In-cellulo studies also revealed that copper-induced vesicularized ATP7B reinstates to Golgi in melatonin-treated hepatocytes. In vivo, melatonin treatment reduced copper-induced oxidative stress in zebrafish embryos and lowered copper burden in Caenorhabditis elegans WD model. Our studies revealed that encapsulation of melatonin within an engineered polymeric nanocapsules having dithiol linkers, susceptible to cleavage by GSH, extended melatonins circulatory half-life ten-fold and enhanced its ROS-scavenging efficacy three-fold relative to free melatonin. This work introduces a unique dual-function therapeutic strategy that integrates antioxidant activity with copper chelation, simultaneously addressing copper overload and redox imbalance. Repurposing melatonin, with its established clinical safety, offers rapid and cost-effective translational pathway toward WD-therapy while providing a generalizable platform for redox- and metal-associated disorders.

cell biology↗

bioRxiv2025

Systematic comparison of Generative AI-Protein Models reveals fundamental differences between structural and sequence-based approaches.

Recent advances in artificial intelligence have led to the development of generative models for de novo protein design. We compared 13 state-of-the-art generative protein models, assessing their ability to produce feasible, diverse, and novel protein monomers. Structural diffusion models generally create designs with higher confidence in predicted structures and more biologically plausible energy distributions, but exhibit limited diversity and strong sequence biases. Conversely, protein language models generate more diverse and novel designs but with lower structural confidence. We also evaluated these models ability to generate unique proteins, conditionally based on the Tobacco Etch Virus (TEV) protease. Generative models were successful in producing functional enzymes, albeit with diminished activity compared to the wildtype TEV. Our systematic benchmarking provides a foundation for evaluating and selecting generative protein models, while highlighting the complementary strengths of different generative paradigms. This framework will facilitate an informed application of these tools for bio-medical engineering and design.

synthetic biology↗

bioRxiv2025

Predictive Modeling of acute Graft-versus-Host-Disease using Machine Learning on Immune Cell and Cytokine Profiles at Engraftment

BackgroundAcute Graft-versus-Host-Disease (aGvHD) is a major immune complication following allogenic hematopoietic stem cell transplantation (Allo-HSCT), initiated by conditioning regimen-associated tissue damage. It involves the complex interplay of immune cells and cytokines. Our study aims to leverage machine learning (ML) algorithms on the immune and cytokine profile of Allo-HSCT recipients to develop biomarker-based classification models to predict the onset of aGvHD at the time of engraftment. Materials and MethodsSeventy patients diagnosed with hematological disorders who had undergone Ist Allo-HSCT were recruited from All India Institute of Medical Sciences, New Delhi, India. Peripheral blood (PB) was collected from the patients at the time of engraftment, and the immune cell subtypes and cytokine profiles were analyzed using flow cytometry and ELISA respectively. The individual cell counts were then processed using basic ML models, including support vector classifier with RBF kernel, Decision Tree, and Random Forest, chosen for their mathematical simplicity and feature importance advantage of Decision Trees and Random Forests. Various data settings were utilized in the study: combined immune and cytokine counts, immune cell counts only, cytokine counts only, T-cell counts only, NK cell counts only, dendritic cell counts only, and B-cell counts only. These configurations were selected to investigate how different data sets impact the prediction of aGvHD before its onset. ResultsAt the engraftment flow cytometric analysis of reconstituted lymphocytes in patients who developed aGvHD revealed that there was a remarkable decrease in the ratio of CD4+/CD8+ T-cell and Tregs, with an increase in the cytotoxic regulatory NK-cell, dendritic cells, and B-cell. The levels of pro-inflammatory cytokines (IFN-{gamma}, IL-1{beta}, IP-10, TNF-, IL-17, IL-12p70, MIP-1, MIP-1{beta}, RANTES), and Th17-and Th1-cells were elevated with consequent decline of the levels of anti-inflammatory cytokine IL-10, IL-2, IL-4 and Th2-, Th9-cells. Machine learning based on 48 parameters [all immune cell subsets n=34 and all cytokines (n=14)]. The correlation heat map shows a higher correlation of aGvHD with the cytokine profile with or without immune cells (accuracy: 1), T-cell alone (accuracy: 0.96); NK-cell alone (accuracy: 0.93); dendritic cells alone (accuracy: 0.90), B-cell alone (accuracy: 0.86). ConclusionThe current models classify perfectly, indicating the potential for a ML algorithm in predicting the onset of aGvHD. However, a study with a larger sample size is required to validate these classification models and mitigate the risk of overfitting observed due to the consistently high performance. The study also highlights the potential of cytokine profiles as a viable alternative to T-cell counts, as evidenced by the correlation heat map and classifier models. These findings provide valuable insights into dataset requirements and future directions for integrating ML models into aGvHD prediction.

immunology↗

bioRxiv2024

EEG data quality in large scale field studies in India and Tanzania

There is a growing imperative to understand the neurophysiological impact of our rapidly changing and diverse technological, social, chemical, and physical environments. To untangle the multidimensional and interacting effects requires data at scale across diverse populations, taking measurement out of a controlled lab environment and into the field. Electroencephalography (EEG), which has correlates with various environmental factors as well as cognitive and mental health outcomes, has the advantage of both portability and cost-effectiveness for this purpose. However, with numerous field researchers spread across diverse locations, data quality issues and researcher idle time due to insufficient participants can quickly become unmanageable and expensive problems. In programs we have established in India and Tanzania, we demonstrate that with appropriate training, structured teams, and daily automated analysis and feedback on data quality, non-specialists can reliably collect EEG data alongside various survey and assessments with consistently high throughput and quality. Over a 30-week period, research teams were able to maintain an average of 25.6 subjects per week, collecting data from a diverse sample of 7,933 participants ranging from Hadzabe hunter-gatherers to office workers. Furthermore, data quality, computed on the first 2,400 records using two common methods, PREP and FASTER, was comparable to benchmark datasets from controlled lab conditions. Altogether this resulted in a cost per subject of under $50, a fraction of the cost typical of such data collection, opening up the possibility for large-scale programs particularly in low- and middle-income countries. Significance StatementWith wide human diversity, a rapidly changing environment and growing rates of neurological and mental health disorders, there is an imperative for large scale neuroimaging studies across diverse populations that can deliver high quality data and be affordably sustained. Here we demonstrate, across two large-scale field data acquisition programs operating in India and Tanzania, that with appropriate systems it is possible to generate high throughput EEG data of quality comparable to controlled lab settings. With effective costs of under $50 per subject, this opens new possibilities for low- and middle-income countries to implement large-scale programs, and to do so at scales that previously could not be considered.

neuroscience↗

bioRxiv2024

Spatiotemporal coordination of actin regulators generates invasive protrusions in cell-cell fusion

Invasive membrane protrusions play a central role in a variety of cellular processes. Unlike filopodia, invasive protrusions are mechanically stiff and propelled by branched actin polymerization. However, how branched actin filaments are organized to create finger-like invasive protrusions is unclear. Here, by examining the mammalian fusogenic synapse, where invasive protrusions are generated to promote cell membrane juxtaposition and fusion, we have uncovered the mechanism underlying invasive protrusion formation. We show that two nucleation promoting factors (NPFs) for the Arp2/3 complex, WAVE and N-WASP, exhibit different localization patterns in the protrusions. While WAVE is closely associated with the plasma membrane at the leading edge of the protrusive structures, N-WASP is enriched with WIP along the actin bundles in the shafts of the protrusions. During protrusion initiation and growth, the Arp2/3 complex nucleates branched actin filaments to generate low density actin clouds, in which the large GTPase dynamin organizes the new branched actin filaments into bundles, followed by actin-bundle stabilization by WIP, the latter functioning as a novel actin-bundling protein. Disrupting any of these components results in defective protrusions and failed myoblast fusion in cultured cells and mouse embryos. Taken together, our study has revealed the intricate spatiotemporal coordination between two NPFs and two actin-bundling proteins in building invasive protrusions at the mammalian fusogenic synapse, and has general implications in understanding invasive protrusion formation in cellular processes beyond cell-cell fusion.

cell biology↗

bioRxiv2024

Branched actin polymerization drives invasive protrusion formation to promote myoblast fusion during skeletal muscle regeneration

Skeletal muscle regeneration is a multistep process involving the activation, proliferation, differentiation, and fusion of muscle stem cells, known as satellite cells. The fusion of satellite cell-derived mononucleated muscle cells (SCMs) is indispensable for the generation of multinucleated, contractile myofibers during muscle repair. However, the molecular and cellular mechanisms underlying SCM fusion during muscle regeneration remain poorly understood. In this study, we uncovered an essential role for branched actin polymerization in SCM fusion. Using conditional knockouts of the Arp2/3 complex and its actin nucleation-promoting factors, N-WASP and WAVE, we demonstrated that branched actin polymerization is required for the SCM fusion, but not for satellite cell proliferation, differentiation, and migration. We showed that the N-WASP and WAVE complexes have partially redundant functions in regulating SCM fusion. Furthermore, we revealed that branched actin polymerization is essential for generating invasive protrusions at the fusogenic synapses in SCMs. Taken together, our study has identified new components of the myoblast fusion machinery in skeletal muscle regeneration and demonstrated a critical role for branched actin-propelled invasive protrusions in this process.

developmental biology↗

bioRxiv2024

ChatNT: A Multimodal Conversational Agent for DNA, RNA and Protein Tasks

Language models are thriving, powering conversational agents that assist and empower humans to solve a number of tasks. Recently, these models were extended to support additional modalities including vision, audio and video, demonstrating impressive capabilities across multiple domains including healthcare. Still, conversational agents remain limited in biology as they cannot yet fully comprehend biological sequences. On the other hand, high-performance foundation models for biological sequences have been built through self-supervision over sequencing data, but these need to be fine-tuned for each specific application, preventing transfer and generalization between tasks. In addition, these models are not conversational which limits their utility to users with coding capabilities. In this paper, we propose to bridge the gap between biology foundation models and conversational agents by introducing ChatNT, the first multimodal conversational agent with an advanced understanding of biological sequences. ChatNT achieves new state-of-the-art results on the Nucleotide Transformer benchmark while being able to solve all tasks at once, in English, and to generalize to unseen questions. In addition, we have curated a new set of more biologically relevant instructions tasks from DNA, RNA and proteins, spanning multiple species, tissues and biological processes. ChatNT reaches performance on par with state-of-the-art specialized methods on those tasks. We also present a novel perplexity-based technique to help calibrate the confidence of our model predictions. Our framework for genomics instruction-tuning can be easily extended to more tasks and biological data modalities (e.g. structure, imaging), making it a widely applicable tool for biology. ChatNT is the first model of its kind and constitutes an initial step towards building generally capable agents that understand biology from first principles while being accessible to users with no coding background.

genomics↗

bioRxiv2024

SegmentNT: annotating the genome at single-nucleotide resolution with DNA foundation models

Genome annotation models that directly analyze DNA sequences are indispensable for modern biological research, enabling rapid and accurate identification of genes and other functional elements. This capability is paramount as the volume of sequenced genomes rapidly expands, making the need for efficient and accurate annotation methods increasingly critical, particularly in the context of genetic variant prediction and in-silico sequence design. Current annotation tools are typically developed for specific element classes and trained from scratch using supervised learning on datasets that are often limited in size. This approach constrains their performance and ability to generalize to new genomes. Here, we frame the genome annotation problem as instance segmentation and introduce a novel methodology for fine-tuning pre-trained DNA foundation models to segment 14 different genic and regulatory elements at single-nucleotide resolution. We leverage the self-supervised pre-trained model Nucleotide Transformer (NT) to develop a general segmentation model, SegmentNT, capable of processing DNA sequences up to 50kb long. By utilizing pre-trained weights from NT, SegmentNT surpasses the performance of several ablation models and baselines, including convolutional networks with one-hot encoded nucleotide sequences and large models trained from scratch. We demonstrate state-of-the-art performance on gene annotation, splice site and regulatory elements detection throughout the genome. We also leveraged our framework to accommodate two extra DNA foundation models, Enformer and Borzoi, extending the sequence context up to 500kb and enhancing performance on regulatory elements. Finally, we show that a SegmentNT model trained on human genomic elements generalizes to elements of different species, and a multi-species SegmentNT model achieves strong generalization across unseen species. Our approach is readily extensible to additional genomic elements and species. We have made our SegmentNT human and multi-species models, as well as the SegmentEnformer and SegmentBorzoi models, available on our github repository in Jax and HuggingFace space in Pytorch.

genomics↗