Biology subjects

Ma, J.

Publications and source records attributed to Ma, J..

At least 145 records · Page 8Linked to original sources

bioRxiv2025

Population-scale cellular GUIDE-seq-2 and biochemical CHANGE-seq-R profiles reveal human genetic variation frequently affects Cas9 off-target activity

Genome editing enzymes can introduce targeted changes to DNA in living cells1-4, transforming biological research and enabling the first approved gene editing therapy for sickle cell disease5. However, their activity can be altered by genetic variation at on- or off-target sites6-8, potentially impacting both their precision and therapeutic safety. Due to a lack of scalable methods to measure genome-wide editing activity in cells from large populations and diverse target libraries, the frequency and extent of these variant effects on editing remain unknown. Here, we present the first systematic, population-scale study of how genetic variation affects the cellular genome-wide activity of CRISPR-Cas9, enabled by a novel, sensitive, and unbiased cellular assay, GUIDE-seq-2, with improved scalability and accuracy compared to the original broadly adopted method9. Analyzing Cas9 genome-wide activity at 1,115 on- and off-target sites across six guide RNAs in cells from 95 individuals spanning four genetically diverse populations, we found that genetic variants frequently overlap off-target sites, with 14% significantly altering Cas9 editing activity. To understand the effect of mismatches in more diverse sequence contexts, we developed a novel method, combinatorial high-throughput analysis of nuclease cleavage effects (CHANCE-seq), the first massively parallel biochemical approach that can quantify Cas9 activity across millions of mismatched target sites. We leveraged this large-scale CHANCE-seq dataset to train a context-aware deep neural network model, CHANCE-net, to accurately predict and interpret the effects of single-nucleotide variants on off-targets with up to six mismatches. Our deep combinatorial profiling of Cas9 off-target activity revealed frequent, context-dependent synergistic effects of mismatches. Taken together, our findings illuminate an approach to accounting for genetic variation when designing genome-editing strategies for research and therapeutics.

genomics↗

bioRxiv2025

mRNA lipid nanoparticle-incorporated nanofiber-hydrogel composite generates a local immunostimulatory niche for cancer immunotherapy

Hydrogel materials have emerged as versatile platforms for various biomedical applications. Notably, the engineered nanofiber-hydrogel composite (NHC) has proven effective in mimicking the soft tissue extracellular matrix, facilitating substantial recruitment of host immune cells and the formation of a local immunostimulatory microenvironment. Leveraging this feature, here we report an mRNA lipid nanoparticle (LNP)-incorporated NHC microgel matrix, termed LiNx, by incorporating LNPs loaded with mRNA encoding tumour antigens. Harnessing the potent transfection efficiency of LNPs in antigen-presenting cells (APCs), LiNx demonstrates remarkable immune cell recruitment, antigen expression and presentation, and cellular interaction. These attributes collectively create an immunostimulating milieu and yield a potent immune response achievable with a single dose, comparable to the conventional three-dose LNP immunization regimen. Further investigations reveal that the LiNx not only generates heightened Th1 and Th2 responses but also elicits a distinctive Type 17 T helper cell-mediated response pivotal for bolstering antitumour efficacy. Our findings elucidate the mechanism underlying LiNxs role in potentiating antigen-specific immune responses, presenting a new strategy for cancer immunotherapy.

bioengineering↗

bioRxiv2025

fCUT&Tag-seq: An Optimized CUT&Tag-based Method for High-Resolution Histone Modification and Chromatin-Binding Protein Profiling in both Model and Plant Pathogenic Fungi

Histone modifications and chromatin-binding proteins play crucial roles in regulating gene expression in eukaryotes, with significant implications for fungal pathogenicity and development. However, profiling these modifications or proteins across the genome in fungi remains challenging due to the technical limitations of the traditional, widely used ChIP-Seq method. Here, we present an optimized CUT&Tag-Seq protocol (fCUT&Tag-Seq) specifically designed for filamentous fungi and dimorphic fungi. Our approach involves the preparation of protoplasts and nuclear extraction to enhance antibody accessibility, along with formaldehyde crosslinking to improve protein-DNA binding efficiency. We then successfully applied fCUT&Tag-Seq to accurately profile multiple histone modifications like H3K9me3, H3K27me3, H3K4me3, and H3K18ac, across different plant pathogenic or model fungal species, including Verticillium dahliae, Neurospora crassa, Fusarium graminearum, and Sporisorium scitamineum, showing good signal-to-noise ratios, reproducibility, and detection sensitivity. Furthermore, we extended this method to profile chromatin-binding proteins, such as the histone acetyltransferase Gcn5. This study establishes fCUT&Tag-Seq as a robust and useful tool for fungal epigenetic research, enabling detailed exploration of chromatin dynamics and advancing our understanding of fungal gene regulation, development, and pathogenicity. Impact StatementWe developed a faster, lowLJinput method to study how genes are turned on and off in fungi, even in tough species that are difficult to analyze with standard methods. Our new approach, named fCUT&Tag-Seq, requires only 10,000 cells and can be completed in just two days. It delivers clearer, more reliable results and has already been successfully applied to multiple fungi, including crop pathogens. By revealing the molecular "control switches" that govern fungal development and virulence, we expect it will accelerate basic research and help identify new targets for controlling destructive plant diseases.

microbiology↗

bioRxiv2025

Peptide Design through Binding Interface Mimicry

Peptides offer distinct advantages for targeted therapy, including oral bioavailability, cellular permeability, and high specificity, which set them apart from conventional small molecules and biologics. In this work, we developed an AI algorithm, named PepMimic, to transform a known protein receptor or an existing antibody of a target into a short peptide drug by mimicking the binding interfaces between targets and known binders. The structural root mean square deviation and interface DockQ with reference binders were 61% and 75% better than the best existing methods on the PepBench datasets. We then applied this novel peptide-design methodology to five drug targets: PD-L1, CD38, BCMA, HER2, and CD4. SPRi results show that 8% of the peptides exhibited dissociation constant (KD) values at the 10-8M level, and 26 peptides achieving KD values as low as 10-9M. This success rate was 20,000 times higher than that observed in a random library screening conducted under identical conditions. PepMimic was applied to target proteins lacking available binders by first utilizing AI algorithms to design protein binders, followed by the generation of peptides through simulation of these artificial interfaces. The top-ranked peptides underwent extensive cellular validation and in vivo testing through tail vein injections in breast, myeloma, and lung tumor mouse models. Experimental results demonstrated effective membrane binding and highlighted the strong potential of these peptides for clinical diagnostic imaging and targeted therapeutic applications.

biochemistry↗

bioRxiv2025

MiRNA Locker: A Modularized DNA Assembly As miRNA Inhibitors

MicroRNAs (miRNAs) are vital post-transcriptional regulators that govern key cellular processes such as proliferation, migration, and apoptosis. Current loss-of-function approaches, including chemically modified antisense oligonucleotides (ASOs), face significant challenges, including high costs, limited scalability, and off-target effects. To overcome these limitations, we developed "miRNA Locker", a novel miRNA inhibition platform created using the Overlapped Oligo Assembly (OOA) method. This innovative platform constructs highly stable dumbbell-shaped single-stranded DNA structures, offering improved target specificity, scalability, and cost-effectiveness. Using miR-214 as a proof-of-concept target, we demonstrated that miRNA Lockers effectively bind Argonaute-miRNA complexes, reduce miRNA levels, and induce downstream changes in gene expression and cellular phenotypes, surpassing the performance of commercial antagomirs. Furthermore, applying miRNA Lockers to miR-654 validated the regulatory role of miR-654 in modulating RNF8 expression and promoting epithelial-mesenchymal transition (EMT) in lung cancer cells. Our results highlight the potential of miRNA Lockers as a versatile tool for studying miRNA function and advancing miRNA-based therapies. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=114 SRC="FIGDIR/small/632138v1_ufig1.gif" ALT="Figure 1"> View larger version (26K): org.highwire.dtl.DTLVardef@c6ac1eorg.highwire.dtl.DTLVardef@14e2698org.highwire.dtl.DTLVardef@1c7d84aorg.highwire.dtl.DTLVardef@148f9c1_HPS_FORMAT_FIGEXP M_FIG C_FIG

synthetic biology↗

bioRxiv2025

DNALONGBENCH: A Benchmark Suite for Long-Range DNA Prediction Tasks

Modeling long-range DNA dependencies is crucial for understanding genome structure and function across a wide range of biological contexts. However, effectively capturing these extensive dependencies, which may span millions of base pairs in tasks such as three-dimensional (3D) chromatin folding prediction, remains a significant challenge. Furthermore, a comprehensive benchmark suite for evaluating tasks that rely on long-range dependencies is notably absent. To address this gap, we introduce DNALO_SCPLOWONGC_SCPLOWBO_SCPLOWENCHC_SCPLOW, a benchmark dataset encompassing five important genomics tasks that consider long-range dependencies up to 1 million base pairs: enhancer-target gene interaction, expression quantitative trait loci, 3D genome organization, regulatory sequence activity, and transcription initiation signals. To comprehensively assess DNALO_SCPLOWONGC_SCPLOWBO_SCPLOWENCHC_SCPLOW, we evaluate the performance of five methods: a task-specific expert model, a convolutional neural network (CNN)-based model, and three fine-tuned DNA foundation models - HyenaDNA, Caduceus-Ph, and Caduceus-PS. We envision DNALO_SCPLOWONGC_SCPLOWBO_SCPLOWENCHC_SCPLOW as a standardized resource with the potential to facilitate comprehensive comparisons and rigorous evaluations of emerging DNA sequence-based deep learning models that account for long-range dependencies.

bioinformatics↗

bioRxiv2024

Increased versatility and convenience: advances and strategy optimization of ReMOT-mediated genetic modification in insects

Genetic modification via gene editing has become a widely adopted and demonstrably effective method in functional gene research within entomology. However, the optimal efficiency and simplicity of delivering exogenous guide RNA-clustered regularly interspaced short palindromic repeats-associated protein 9 (gRNA-Cas9) complexes into target tissues are crucial for successful gene editing. The Receptor-Mediated Ovary Transduction of Cargo (ReMOT) strategy, which simplifies the delivery process, target site selection, technical requirements, and delivery cost compared with embryonic microinjection, enabling efficient editing at the germline level, is gaining increasing attention. Although the feasibility and advantages of this technique have been demonstrated in various insect species, further optimization of operational details and the overcoming of further bottlenecks are still required. This review focuses on advances in developing ReMOT as a valuable technology, exploring its applicability, rationale for selecting the ovary as a delivery target site, factors influencing its efficiency, and recommendations for improvement. The versatility and effectiveness of ReMOT make it a promising method for researchers looking to make precise genetic modifications with greater ease and efficiency.

genetics↗

bioRxiv2024

Dynamic p21-dependency during quiescence arrest unveiled by a rapid p21 depletion system

Stress- or developmentally-induced signals can trigger G0/G1 cell cycle arrest through the upregulation of p21, a key inhibitor that halts proliferation. While p21 is widely recognized as essential for establishing this arrest, its degradation is generally assumed to be a prerequisite for the exit from arrest and cell cycle re-entry. Using a rapid p21 degradation system that allows for both endogenous p21 tracking and controlled depletion, we uncovered an extended role for p21 during the exit from G0/G1 arrest. Our results showed that removal of the arrest inducing signals did not lead to an immediate decline of p21 during the exit process; instead, p21 levels continued to maintain for a period before its gradual down-regulation. Importantly, premature depletion of p21 during this process weakened the capacity of cell cycle re-entry, particularly in cells with high levels of pre-existent p21. We found that during the release from arrest, cells with high levels of pre-existent p21 but a premature p21 depletion exhibited reduced capacity to restore KRAS/ERK activity, and supplementing KRAS/ERK activity rescued the failure of cell cycle re-entry in these cells. These findings reveal a previously unappreciated function for p21 during the exit from G0/G1 arrest, aside and distinct from its role in initiating the arrest. This new paradigm extends across multiple cell lines and stress-induced arrest contexts, offering a new framework for understanding p21 function in response to anticancer therapies and in development.

cell biology↗

bioRxiv2024

OPUS-GO: An interpretable protein/RNA sequence annotation framework based on biological language model

Accurate annotation of protein and RNA sequences is essential for understanding their structural and functional attributes. However, due to the relative ease of obtaining whole sequence-level annotations compared to residue-level annotations, existing biological language model (BLM)-based methods often prioritize enhancing sequence-level classification accuracy while neglecting residue-level interpretability. To address this, we introduce OPUS-GO, which exclusively utilizes sequence-level annotations to provide both sequence-level and residue-level classification results. In other words, OPUS-GO not only provides the sequence-level annotations but also offers the rationale behind these predictions by pinpointing their corresponding most critical residues within the sequence. Our results show that, by leveraging features derived from BLMs and our modified Multiple Instance Learning (MIL) strategy, OPUS-GO exhibits superior sequence-level classification accuracy compared to baseline methods in most downstream tasks. Furthermore, OPUS-GO demonstrates robust interpretability by accurately identifying the residues associated with the corresponding labels. Additionally, the OPUS-GO framework can be seamlessly integrated into any language model, enhancing both accuracy and interpretability for their downstream tasks.

bioinformatics↗

bioRxiv2024

The first comorbidity networks in companion dogs in the Dog Aging Project

Comorbidity and its association with age are of great interest in geroscience. However, there are few model organisms that are well-suited to study comorbidities that will have high relevance to humans. In this light, we turn our attention to the companion dog. The companion dog shares many morbidities with humans. Thus, a better understanding of canine comorbidity relationships could benefit both humans and dogs. We present an analysis of canine comorbidity networks from the Dog Aging Project, a large epidemiological cohort study of companion dogs in the United States. We included owner-reported health conditions that occurred in at least 60 dogs (n=160) and included only dogs that had at least one of those health conditions (n=26,614). We constructed an undirected comorbidity network using a Poisson binomial test, adjusting for age, sex, sterilization status, breed background (i.e., purebred vs. mixed-breed), and weight. The comorbidity network reveals well-documented comorbidities, such as diabetes with cataracts and blindness, and hypertension with chronic kidney disease (CKD). In addition, this network also supports less well-studied comorbidity relationships, such as proteinuria with anemia. A directed comorbidity network accounting for time of reported condition onset suggests that diabetes precedes cataracts, elbow/hip dysplasia before osteoarthritis, and keratoconjunctivitis sicca before corneal ulcer, which are consistent with the canine literature. Analysis of age-stratified networks reveals that global centrality measures increase with age and are the highest in the Senior group compared to the Young Adult and Mature Adult groups. Only the Senior group identified the association between hypertension and CKD. Our results suggest that comorbidity network analysis is a promising method to enhance clinical knowledge and canine healthcare management. Author SummaryCompanion dogs age alongside humans and suffer many of the same diseases, making them an ideal "real-world" model for human health. Using owner-reported data from 26,614 dogs enrolled in the nationwide Dog Aging Project, we built the first large-scale maps--called comorbidity networks--that show which canine diseases tend to appear together and in what order. The networks correctly highlighted well-known pairings such as diabetes with cataracts and blindness, and hypertension with chronic kidney disease. They also revealed under-appreciated links--for example, protein loss in urine associated with anaemia--suggesting new avenues for veterinary research and care. By adding the reported date of diagnosis, we could infer likely sequences of the diseases: diabetes generally preceded cataracts, hip dysplasia came before osteoarthritis, and dry-eye disease often led to corneal ulcers. When we split the data by life stage, we saw disease webs become denser and more centred on a few key conditions as dogs grew older, echoing patterns seen in people. Together, these findings show that network analysis of large pet-health datasets can guide clinicians, inform breeding and prevention strategies, and ultimately improve the wellbeing of both dogs and humans.

physiology↗

bioRxiv2024

MorphoDiff: Cellular Morphology Painting with Diffusion Models

Understanding cellular responses to external stimuli is critical for parsing biological mechanisms and advancing therapeutic development. High-content image-based assays provide a cost-effective approach to examine cellular phenotypes induced by diverse interventions, which offers valuable insights into biological processes and cellular states. In this paper, we introduce MorphoDiff, a generative pipeline to predict high-resolution cell morphological responses under different conditions based on perturbation encoding. To the best of our knowledge, MorphoDiff is the first framework capable of producing guided, high-resolution predictions of cell morphology that generalize across both chemical and genetic interventions. The model integrates perturbation embeddings as guiding signals within a 2D latent diffusion model. The comprehensive computational, biological, and visual validations across three open-source Cell Painting datasets show that MorphoDiff can generate high-fidelity images and produce meaningful biology signals under various interventions. We envision the model will facilitate efficient in silico exploration of perturbational landscapes towards more effective drug discovery studies.

bioinformatics↗

bioRxiv2024

CD24-Fc resolves inflammation and rescues CD8 T cells with polyfunctionality during HIV-1 infection under cART

The persistence of HIV-1 reservoirs during combination anti-retroviral therapy (cART) leads to chronic immune activation and systemic inflammation in people with HIV (PWH), associating with a suboptimal immune reconstitution as well as an increased risk of non-AIDS events. This highlights the needs to develop novel therapy for HIV-1 related diseases in PWH. In this study, we assessed the therapeutic effect of CD24-Fc, a fusion protein with anti-inflammatory properties that interacts with danger-associated molecular patterns (DAMPs) and siglec-10, in chronic HIV-1 infection model using humanized mice undergoing suppressive cART. Our findings show that CD24-Fc treatment significantly reduced inflammation and immune hyperactivation in vivo when combined with cART. CD24-Fc mediated resolution of inflammation was associated with improved recovery of CD4 T cells, reduced immune activation, restored central memory T cells and reversal of immune cell exhaustion phenotype. Notably, CD24-Fc treatment rescued CXCR5+ CD8 central memory T cell (TCM) which correlated with increased polyfunctionality in HIV-specific T cells in humanized mice and in cultured peripheral blood mononuclear cells (PBMCs) from PWH. This restoration of CXCR5+ memory CD8 T cells was associated with HIV replication inhibition, delayed viral rebound and reduced HIV-1 pathogenesis upon cART cessation. This study suggests that CD24-Fc treatment could represent a promising new therapeutic strategy for managing chronic systemic inflammation and associated diseases in PWH. Author summaryCombination antiretroviral therapy (cART) cannot block viral gene expression from activated HIV proviral DNA in reservoir cells, contributing to chronic immune activation and inflammation associated diseases in people with HIV (PWH). The therapeutic treatment of anti-inflammatory fusion protein CD24-Fc in humanized mice during suppressive cART (i) resolves inflammation and chronic HIV-1 immune pathogenesis during suppressive cART, (ii) rescues CXCR5-expressing CD8 memory T cells and enhances antiviral response in humanized mice and PWH PBMCs, (iii) delays virus rebound and reduces viral pathogenesis after cART cessation. Thus, CD24-Fc could provide a novel therapeutic strategy for treating chronic systemic inflammation and associated diseases in PWH.

immunology↗

bioRxiv2024

L2G: Repurposing Language Models for Genomics Tasks

Pre-trained language models have transformed the field of natural language processing (NLP), and their success has inspired efforts in genomics to develop domain-specific foundation models (FMs). However, creating high-quality genomic FMs from scratch is resource-intensive, requiring significant computational power and high-quality pre-training data. The success of large language models (LLMs) in NLP has largely been driven by industrial-scale efforts leveraging vast, diverse corpora and massive computing infrastructure. In this work, we aim to bypass the data and computational bottlenecks of creating genomic FMs from scratch and instead propose repurposing existing LLMs for genomics tasks. Inspired by the recently observed cross-modal transfer phenomenon - where transformers pre-trained on natural language can generalize to other modalities - we introduce L2G, which adapts a pre-trained LLM architecture for genomics using neural architecture search (NAS) and a novel three-stage training procedure. Remarkably, without requiring extensive pre-training on DNA sequence data, L2G achieves superior performance to fine-tuned genomic FMs and task-specific models on more than half of tasks across multiple genomics benchmarks. In an enhancer activity prediction task, L2G further demonstrates its capacity to identify significant transcription factor motifs. Our work not only highlights the generalizability and efficacy of language models in out-of-domain tasks such as genomics, but also opens new avenues for more efficient and less resource-intensive methodologies in genomic research.

bioinformatics↗

bioRxiv2024

Exclusive cataloging of near-complete genomes unveils novel functional insights into the human gut microbiome

Understanding the human gut microbiome requires comprehensive genomic catalogues, yet many lack geographic diversity and contain medium-quality metagenome-assembled genomes (MAGs) missing up to 50% of genomic regions, potentially distorting functional insights. Here, we describe an enhanced Human Reference Gut Microbiome (HRGM2) resource, a catalogue of near-complete MAGs ([≥]90% completeness, [≤]5% contamination) and isolate genomes. HRGM2 comprises 155,211 non-redundant near-complete genomes from 4,824 prokaryotic species across 41 countries, representing a 66% increase in genome count and a 50% boost in species diversity compared to the Unified Human Gastrointestinal Genome catalogue. It enabled improved DNA-based species profiling, resolution of strain heterogeneity, and survey of human gut resistome. The exclusive use of these genomes improved metabolic capacity assessment enabling high-confidence, automated genome-scale metabolic models (GEMs) of the entire microbiota, revealing disease-associated microbial metabolic interactions. This resource will facilitate reliable functional insights into gut microbiomes.

microbiology↗

bioRxiv2024

Nonuple atg8 mutant provides genetic evidence for functional specialization of ATG8 isoforms in Arabidopsis thaliana

Autophagy sustains cellular health by recycling damaged or excess components through autophagosomes. It is mediated by conserved ATG proteins, which coordinate autophagosome biogenesis and selective cargo degradation. Among these, the ubiquitin-like ATG8 protein plays a central role by linking cargo to the growing autophagosomes through interacting with selective autophagy receptors. Unlike most ATG proteins, the ATG8 gene family is significantly expanded in vascular plants, but its functional specialization remains poorly understood. Using transcriptional and translational reporters in Arabidopsis thaliana, we revealed that ATG8 isoforms are differentially expressed across tissues and form distinct autophagosomes within the same cell. To explore ATG8 specialization, we generated the nonuple{Delta} atg8 mutant lacking all nine ATG8 isoforms. The mutant displayed hypersensitivity to carbon and nitrogen starvation, coupled with defects in bulk and selective autophagy as shown by biochemical and ultrastructural analyses. Complementation experiments demonstrated that ATG8A could rescue both carbon and nitrogen starvation phenotypes, whereas ATG8H could only complement carbon starvation. Proximity labeling proteomics further identified isoform-specific interactors under nitrogen starvation, underscoring their functional divergence. These findings provide genetic evidence for functional specialization of ATG8 isoforms in plants and lay the foundation for investigating their roles in diverse cell types and stress conditions.

plant biology↗

bioRxiv2024

ALDH3A1-mediated detoxification of reactive aldehydes contributes to distinct muscle responses to denervation and Amyotrophic Lateral Sclerosis progression

Different muscles exhibit varied susceptibility to degeneration in Amyotrophic Lateral Sclerosis (ALS), a fatal neuromuscular disorder. Extraocular muscles (EOMs) are particularly resistant to ALS progression, and exploring the underlying molecular nature may offer significant therapeutic value. Reactive aldehyde 4-hydroxynonenal (HNE) is implicated in ALS pathogenesis, and Aldh3a1 is an inactivation-resistant intracellular aldehyde dehydrogenase that detoxifies 4-HNE to protect eyes against UV-induced oxidative stress. We detected prominently higher levels of Aldh3a1 in mouse EOMs compared to other muscles under normal physiological conditions. In an ALS mouse model (hSOD1G93A) reaching end-stage, Aldh3a1 expression was maintained high in EOMs, substantially elevated in soleus and diaphragm, but only moderately increased in extensor digitorum longus (EDL) muscle, which endured the most severe pathological remodeling, as demonstrated by unparalleled upregulation of a denervation marker Ankrd1. Importantly, sciatic nerve transection in wildtype mice further confirmed induced Aldh3a1 and Ankrd1 expression in an inverse manner across muscle types in response to denervation. Mechanistically, whole-muscle RNA-Seq and pharmacological tests indicate that higher basal levels of lipid oxidation in soleus and diaphragm muscles may predispose them to stronger Nrf2 antioxidant responses under pathological stress compared to EDL, leading to more prominent Aldh3a1 upregulation. Additionally, the identification of the myoblast fusion marker Mymk as an EOM signature gene suggests that the spontaneous activation of satellite cells contributes to high levels of Aldh3a1 in EOMs. Functionally, adeno-associated virus-mediated overexpression of Aldh3a1 protected myotubes from 4-HNE-induced DNA fragmentation and plasma membrane leakage. It also restored MG53-mediated membrane repair, highlighting its potential for clinical applications.

cell biology↗

bioRxiv2024

Better Prior Distribution for Antibody Complementarity-Determining Regions Design

Designing antibodies with desired binding specificity and affinity is essential for pharmaceutical research. While diffusion-based models have advanced the co-design of the Complementarity-Determining Regions (CDRs) sequences and structures, challenges remain, including non-informative prior distribution, incompatibility with discrete amino acid types, and impractical computational cost in large-scale sampling. To address these, we proposed FlowDesign, a sequence-structure co-design approach based on Flow Matching, offering: (1) Flexible selection of prior distributions; (2) Direct matching of discrete distributions; (3) Enhanced computational efficiency for large-scale sampling. By leveraging various priors, data-driven structural models proved the most informative. FlowDesign outperformed baselines in Amino Acid Recovery (AAR), RMSD, and Rosetta energy. We also applied FlowDesign to design antibodies targeting the HIV-1 receptor CD4. FlowDesign yielded antibodies with improved binding affinity and neutralizing potency compared to the antibody Ibalizumab across multiple HIV mutants, validated by Biolayer Interferometry (BLI) and pseudovirus neutralization. This highlights FlowDesigns potential in antibody and protein design. A record of this papers Transparent Peer Review process is included in the Supplemental Information.

bioinformatics↗

bioRxiv2024

BMP4 is sufficient to induce a caudal organizer

The caudal part of a vertebrate embryo consists of somites, neural tube, lateral plate mesoderm derivatives and the tailbud. However, the minimal conditions and factors sufficient to induce the caudal region, particularly in humans, remain unresolved. Here, we show that BMP4 alone, when administered at appropriate dosage, is sufficient to induce the formation of an organizer for caudal induction. This organizer induces caudal cell fate specification and morphogenesis in zebrafish embryos. In 3D human pluripotent stem cells (hPSCs) aggregates, BMP4 can induce an elongated embryonic structure characterized by caudal fates. Importantly, hPSCs instructed by BMP4 are sufficient to induce a secondary caudal region when grafted into the animal pole of the zebrafish embryo. Our study thus uncovers BMP4 as the inducer in the formation of a caudal organizer in the vertebrate embryo. Significance statementWe demonstrate that BMP4 alone can induce the formation of a caudal organizer, a critical structure that guides the development of the caudal region in vertebrates. Using zebrafish embryos, embryonic explants, human pluripotent stem cell xenografts, and 3D human stem cell aggregates, we show that this organizer replicates morphogenesis and key differentiation pathways seen in natural development. Our findings uncover a novel role for BMP4 and define the minimal requirements for inducing this organizer, offering new insights into vertebrate development and potential applications in regenerative medicine.

developmental biology↗